Sajjad ahmad bioinformatics 2020 qau isb prr.pdf

Deciphering the Dynamics of Therapeutic

Proteins from Nosocomial Pathogens

By

SAJJAD AHMAD

National Center for Bioinformatics

Faculty of Biological Sciences

Quaid-I-Azam University, Islamabad

2020

Deciphering the Dynamics of Therapeutic

Proteins from Nosocomial Pathogens

A dissertation submitted in partial fulfillment of the

requirements for the degree of Doctor of Philosophy in

Bioinformatics

By

SAJJAD AHMAD

National Center for Bioinformatics

Faculty of Biological Sciences

Quaid-I-Azam University, Islamabad

2020

Plagiarism Undertaking

I solemnly declare that research work presented in this dissertation entitled as

“Deciphering the Dynamics of Therapeutic Proteins from Nosocomial Pathogens”

is solely my research work with no significant contributions from any other person.

Small contribution/help wherever taken has been duly acknowledged and that

dissertation has been written by me.

I understand the zero tolerance policy of the HEC and Quaid-i-Azam University

towards plagiarism. Therefore, I as an author of the above titled dissertation declare

that no portion of my thesis has been plagiarized and any material used as reference

is properly referred/cited.

I undertake that If I found guilty of any formal plagiarism in the above titled thesis

ever after award of PhD degree, the University reserves the rights to

withdraw/revoke my PhD degree and that HEC and the University has the right to

publish my name on the HEC/University website on which names of students are

placed who submitted plagiarized thesis.

Date: 10 June, 2020 Sajjad Ahmad

Dedication

I dedicate this humble effort of mine to humanity and to

the land I was born and will always belong

Contents

Acknowledgments………………………………………………………………………….. i-iii

List of Figures……………………………………………………………………………… iv-vii

List of Tables……………………………………………………………………………….. viii-ix

Abbreviations………………………………………………………………………………. x-xii

Summary……………………………………………………………………………………. xvi-xvii

Chapter # 1

Introduction to Nosocomial Infections and Nosocomial Pathogens……………………... 1-40

1.1. Nosocomial Infections…………………………………………………………………... 1

1.1.1. Urinary Tract Infections (UTIs)…………………………………………………….. 2

1.1.2. Nosocomial Respiratory Tract Infections (NRTIs)…………………………………. 3

1.1.3. Nosocomial Bloodstream Infections (NBSIs)……………………………………… 4

1.1.4. Nosocomial Central Nervous System Infections (NCNSIs)………………………... 4

1.1.5. Nosocomial Skin and Soft Tissue Infections (NSSTIs)…………………………….. 5

1.2. Nosocomial Bacterial Pathogens………………………………………………………... 6

1.2.1. A. baumannii Infections…………………………………………………………….. 8

1.2.2. Natural Habitats…………………………………………………………………….. 10

1.2.3. Adherence, Biofilm Formation and Pathogenicity………………………………….. 10

1.2.4. Iron Acquisition…………………………………………………………………….. 12

1.2.5. Virulence Factors…………………………………………………………………... 12

1.2.6. Epidemiology………………………………………………………………………. 13

1.2.7. Antibiotic Resistance……………………………………………………………….. 16

1.3. References………………………………………………………………………………. 20

Chapter # 2

Introduction to Computer Aided Vaccine and Drug Designing…………………………. 41-62

2.1. Computer Aided Vaccine Designing (Reverse Vaccinology)…………………………… 41

2.2. Computer Aided Drug Designing……………………………………………………….. 42

2.2.1. Subtractive Proteomics …………………………………………………………….. 42

2.2.2. Modeling of Proteins Structure…………………………………………………….. 43

2.2.3. Protein Structure Validation……………………………………………………….. 45

2.2.4. Molecular Docking Theory………………………………………………………… 46

2.2.4.1. Sampling Algorithms………………………………………………………… 46

2.2.4.2. Scoring Functions…………………………………………………………….. 47

2.2.5. Docking Methodologies……………………………………………………………. 48

2.2.6. MD Simulations……………………………………………………………………. 49

2.2.7. Force Fields………………………………………………………………………… 52

2.2.8. Periodic Boundary Conditions……………………………………………………... 52

2.2.9. Binding Free Energy Calculations………………………………………………….. 53

2.3. References………………………………………………………………………………. 53

Chapter # 3

Combating Tigecycline Resistant Acinetobacter baumannii: A Leap Forward towards

Multi-epitope based Vaccine Discovery...............................................................................

63-112

3.1. Abstract………………………………………………………………………………... 63

3.2. Introduction……………………………………………………………………………... 64

3.3. Materials and Methods………………………………………………………………….. 67

3.3.1. Proteome Retrieval and Subtractive Proteomics…………………………………… 67

3.3.2. Exo-proteome and Secretome Prediction………………………………………… 67

3.3.3. Virulent Proteins Evaluation……………………………………………………... 68

3.3.4. Screening of Non-probiotic Proteins……………………………………………….. 68

3.3.5. Screening of Non-similar Mouse Proteins………………………………………... 68

3.3.6. Physicochemical Characterization……………………………………………….. 69

3.3.7. Antigenicity Prediction…………………………………………………………….. 69

3.3.8. Prediction of B-cell derived T-cell Epitopes………………………………………... 69

3.3.9. Targeted Proteins Structure Prediction……………………………………………... 70

3.3.10. Predicting Proteins with a Strong Interactome…………………………………….. 70

3.3.11. Multi-epitope Vaccine Sequence Construction…………………………………… 72

3.3.12. Molecular Docking………………………………………………………………... 72

3.3.13. MD Simulations and Binding Free Energy Calculations………………………….. 72

3.4. Results and Discussion………………………………………………………………….. 74

3.4.1. Proteome Retrieval and Subtractive Proteomics……………………………………. 74

3.4.2. TRAB Exo-proteome and Secretome……………………………………………... 76

3.4.3. TRAB Virulent Proteins…………………………………………………………... 79

3.4.4. Screening of Non-probiotic Bacterial Proteins……………………………………... 80

3.4.5. Screening of Non-similar Mouse Proteins………………………………………... 81

3.4.6. Physiochemical Prioritization of Vaccine Proteins………………………………. 81

3.4.7. Antigenicity Prediction…………………………………………………………… 82

3.4.8. Prediction of B-cell derived T-cell Epitopes……………………………………... 82

3.4.9. Structure Prediction and Evaluation of Shortlisted Epitopes Proteins…………… 84

3.4.10. Interacting Network Analysis……………………………………………………... 85

3.4.11. Secondary and Tertiary Structure of Construct……………………………………. 86

3.4.12. Intrinsic Disorder Regions Prediction……………………………………………... 90

3.4.13. Tertiary Structure Refinement and Disulfide Engineering………………………... 90

3.4.14. Codon Optimization of the Vaccine-construct…………………………………….. 91

3.4.15. Molecular Docking of the Vaccine-construct with the TLR4 Immune Receptor….. 91

3.4.16. MD Simulations and MM/GBSA based Binding Free Energy Calculations………. 96

3.5. Conclusions……………………………………………………………………………. 99

3.6. Supplementary Files…………………………………………………………………... 99

3.7. References………………………………………………………………………………. 100

Chapter # 4

A Novel Approach of Virulome Based Reverse Vaccinology For Exploring and

Validating Peptide-Based Vaccine Candidates against the Most Troublesome

Nosocomial Pathogen: Acinetobacter baumannii………………………………………….

113-140

4.1. Abstract…………………………………………………………………………………. 113

4.2. Introduction……………………………………………………………………………... 114

4.3. Material and Methods…………………………………………………………………… 115

4.3.1. Virulome Retrieval and Functional Categorization………………………………… 115

4.3.2. Physicochemical Prioritization…………………………………………………….. 115

4.3.3. Epitopes Mapping………………………………………………………………….. 116

4.3.4. Interactome Evaluation…………………………………………………………….. 116

4.3.5. Structure Prediction and Evaluation………………………………………………... 118

4.3.6. Epitopes Exo-membrane Topology Evaluation…………………………………….. 118

4.3.7. Epitopes Molecular Docking……………………………………………………….. 118

4.3.8. Methodology Validation through a Negative Control………………………………. 118


4.4.1. A. baumannii Virulome Assembly and Evaluation…………………………………. 119

4.4.2. Exo-proteome and Secretome Exploration…………………………………………. 120

4.4.3. Human Non-Homologous Proteins………………………………………………… 121

4.4.4. Physicochemical Characterization…………………………………………………. 122

4.4.5. B-cell Epitope Derived T-cell Epitope Mapping…………………………………… 123

4.4.6. Epitopes Allergenicity and Conservation…………………………………………... 124

4.4.7. Cellular Interactome of CsuB and EpsA……………………………………………. 124

4.4.8. Structure Prediction and Evaluation………………………………………………... 125

4.4.9. Pepitope Analysis…………………………………………………………………... 127

4.4.10. Epitopes Binding Mode and Interactions Analysis………………………………... 127

4.4.11. Methodology Validation through a Negative Control……………………………... 128

4.4. Conclusions……………………………………………………………………………... 131

4.5. Supplementary Files…………………………………………………………………….. 131

4.6. References………………………………………………………………………………. 132

Chapter # 5

Comparative Subtractive Proteomics Based Ranking for Antibiotic Targets against the

Dirtiest Superbug: Acinetobacter baumannii……………………………………………... 141-189

5.1. Abstract…………………………………………………………………………………. 141

5.2. Introduction……………………………………………………………………………... 142


5.3.1. Proteome Subtraction………………………………………………………………. 146

5.3.2. Drug Target Prioritization and Selection…………………………………………… 146

5.3.3. Comparative Structure Modeling and Evaluation…………………………………... 147

5.3.4. KdsA Intrinsic Disorder Regions Prediction……………………………………….. 148

5.3.5. Comparative Molecular Docking…………………………………………………... 148

5.3.6. MD Simulations..…………………………………………………………………... 150



5.4.1. Drug Candidate’s Prioritization…………………………………………………….. 152

5.4.2. Virulence Proteins Analysis………………………………………………………... 154

5.4.3. Physicochemical Characterization…………………………………………………. 156

5.4.5. Interacting Networks of Targeted Proteins…………………………………………. 157

5.4.6. Drug Target Selection………………………………………………………………. 159

5.4.7. Comparative Structure Modelling………………………………………………….. 160

5.4.8. KdsA Intrinsic Disorder Regions Analysis…………………………………………. 163

5.4.9. Protein Active Site Prediction………………………………………………………. 163

5.4.10. Comparative Molecular Docking…………………………………………………. 164

5.4.11. MD Simulations..…………………………………………………………………. 172

5.4.12. Estimation of Binding Free energy………………………………………………... 175

5.4.13. Free Energy Decomposition………………………………………………………. 176

5.5. Conclusions……………………………………………………………………………... 179


5.7. References………………………………………………………………………………. 180

Chapter # 6

Toward Novel Inhibitors against KdsB: A Highly Specific and Selective Broad-

Spectrum Bacterial Enzyme………………………………………………………………..

190-228

6.1. Abstract…………………………………………………………………………………. 190

6.2. Introduction……………………………………………………………………………... 191


6.3.1. Molecular Docking…………………………………………………………………. 193

6.3.2. Druglikeness and Computational Pharmacokinetics……………………………….. 194

6.3.3. MD Simulations………………………………………………..…………………... 195

6.3.4. Radial Distribution Function (RDF)………………………………………………... 196




6.4.2. Trajectories Analysis……………………………………………………………….. 206

6.4.3. Binding Pattern Analysis…………………………………………………………… 210

6.4.4. RDF Analysis………………………………………………………………………. 212


6.5. Conclusions……………………………………………………………………………... 221


6.7. References………………………………………………………………………………. 221

Chapter # 7

Identification of natural inhibitors against Acinetobacter baumannii D-alanine-D-

alanine ligase enzyme: A multi-spectrum in silico approach..............................................

229-269

7.1. Abstract…………………………………………………………………………………. 229

7.2. Introduction……………………………………………………………………………... 230


7.3.1. Protein and Inhibitors Preparation………………………………………………….. 232

7.2.1. 7.3.2. Binding Cavity Prediction………………………………………………………….. 232


7.3.4. MD Simulations………………………………………………………………..…... 233

7.3.5. RDF and AFD………………………………………………………………………. 234



7.4.1. Active Site Prediction………………………………………………………………. 236


7.4.3. SwissADME and PreADMET Analysis……………………………………………. 246

7.4.4. MD Simulations…………………………………………………………..………... 248

7.4.5. RDF and AFD Analysis…………………………………………………………….. 250

7.4.6. MM/GBSA Based Energy Calculations……………………………………………. 255

7.4.7. WaterSwap Based Energy Calculations……………………………………………. 257

7.3. 7.5. Conclusions……………………………………………………………………………... 260

7.4. 7.6. Supplementary Files…………………………………………………………………….. 260

7.5. 7.7. References………………………………………………………………………………. 261

Chapter # 8

Blocking the Catalytic Mechanism of MurC Ligase Enzyme from Acinetobacter

baumannii: An in Silico Guided Study towards the Discovery of Natural Antibiotics…..

270-320

8.1. Abstract…………………………………………………………………………………. 270

7.2. Introduction……………………………………………………………………………... 271


8.3.1. Receptor Protein Structure Modelling and Evaluation……………………………... 273

8.3.2. Inhibitors Dataset Preparation……………………………………………………… 273

8.3.3. Structure based Virtual Screening using GOLD……………………………………. 274

8.3.4. Computational Pharmacokinetics…………………………………………………... 274

8.3.5. MD Simulations Setup…………………………..…………………………………. 274

8.3.6. Binding Energies Calculation………………………………………………………. 277

8.3.7. WaterSwap based Energy Calculations…………………………………………….. 278


8.4.1. MurC Structure Modelling…………………………………………………………. 279

8.4.2. Inhibitor Library Preparation……………………………………………………….. 280


8.4.4. Computational Pharmacokinetics of the Top Most Inhibitor……………………….. 285

8.4.4.1. Absorption……………………………………………………………………. 285

8.4.4.2. Distribution…………………………………………………………………... 286

8.4.4.3. Metabolism…………………………………………………………………… 289

8.4.4.4. Excretion……………………………………………………………………... 289

8.4.4.5. Toxicity………………………………………………………………………. 289

8.4.5. MD Simulation of the Complex…………………………………………………….. 290

8.4.5.1. RMSD Analysis………………………………………………………………. 290

8.4.5.2. RMSF Analysis………………………………………………………………. 300

8.4.5.3. Rg Analysis…………………………………………………………………... 300

8.4.5.4. β-factor Analysis……………………………………………………………... 300

8.4.6. RDF and AFD Analysis…………………………………………………………….. 301

8.4.7. Binding Free Energies Calculation…………………………………………………. 303

8.4.8. WaterSwap based Binding Free Energy Calculations………………………………. 308

8.3. 8.5. Conclusions……………………………………………………………………………... 309

8.4. 8.6. Supplementary Files…………………………………………………………………….. 310

8.5. 8.7. References………………………………………………………………………………. 310

Chapter # 9

Binding Mode Analysis, Dynamic Simulation and Binding Free Energy Calculations

of the Murf Ligase from Acinetobacter baumannii………………………………………. 321-353

9.1. Abstract………………………………………………………………………………... 321

9.2. Introduction……………………………………………………………………………. 322

9.3. Materials and Methods………………………………………………………………… 324

9.3.1. Receptor Protein Preparation……………………………………………………… 324

9.3.2. Ligands Search and Preparation…………………………………………………... 324

9.3.3. Molecular Docking………………………………………………………………... 324

9.3.4. Computational Pharmacokinetics Evaluation……………………………………... 325

9.3.5. MD Simulations…………………………………………………………………... 325

9.3.6. AFD……………………………………………………………………………….. 326

9.3.7. MM(PB/GB)SA Analysis………………………………………………………… 326

9.4. Results and Discussion………………………………………………………………… 328

9.4.1. Comparative Molecular Docking Analysis………………………………………... 328

9.4.2. Computational Pharmacokinetics…………………………………………………. 333

9.4.3. MD Simulations…………………………………………………………………... 334

9.4.4. RDF and AFD Analysis…………………………………………………………… 336

9.4.5. Binding Free Energy Calculations………………………………………………… 341

9.5. Conclusions……………………………………………………………………………. 347

9.6. Supplementary Files…………………………………………………………………… 347

9.7. References……………………………………………………………………………... 347

Chapter # 10

Moleculer Dynamics Simulation Revealed Reciever Domain of Acinetobacter

baumannii BfmR Enzyme as the Hot Spot for future Antibiotics Designing.....................

354-389

10.1. Abstract………………………………………………………………………………... 354

10.2. Introduction……………………………………………………………………………. 355

10.3. Materials and Methods………………………………………………………………… 357

10.3.1. BfmR Enzyme Retrieval and Minimization……………………………………….. 357

10.3.2. Inhibitors Preparation……………………………………………………………... 357

10.3.3. Molecular Docking of Lead-like Inhibitors……………………………………….. 358

10.3.4. Computational Pharmacokinetics…………………………………………………. 361

10.3.5. MD Simulations……………………………………..……………………………. 361

10.3.6. MMGB\PBSA Analysis…………………………………………………………... 362

10.3.7. WaterSwap Analysis……………………………………………………………… 363

10.4. Results and Discussion………………………………………………………………… 365

10.4.1. Molecular Docking of Lead-like Inhibitors……………………………………….. 365

10.4.2. SwissADME and preADMET Analysis…………………………………………... 371

10.4.3. MD Simulations…………………………………………………………………... 372

10.4.4. Binding Free Energies Calculation………………………………………………... 378

10.5. Conclusions……………………………………………………………………………. 382

10.6. Supplementary Files…………………………………………………………………… 382

10.7. References……………………………………………………………………………... 382

Chapter # 11

Supplementary Data………………………………………………………………………..

390

i

Acknowledgments

All praises and gratitude to the Almighty Allah, the Most Beneficent and the Most Merciful, our

soul provider and nourisher, the creator of earth and all heavens, the Lord of the 'Alamin

(mankind, jinns and all that exists) -nothing could be done without his will. Salutations and peace

be upon the Holy Prophet Muhammad SAW., the last messenger of Allah and greatest of all

humans ever lived or would ever live on earth.

Having said that, it is my privilege and honor to work under the supervision of Dr. Syed Sikander

Azam. I wish to express my deepest gratitude for his expert guidance, appreciation, sincere

advice, and support during the period of my research work. His endless encouragement and

familiar deeds have been the major driving force throughout my research work. It has been a

lifetime experience with him and he is a source of constant support at each step of hurdles that I

faced during this period.

I also owe a deep sense of gratitude to Professor Dr. Laurence Rahme, Department of Molecular

Surgery, Massachusetts General Hospital/ Harvard Medical School for accepting me as a

visiting scholar at her laboratory for a period of six months. Her welcoming behavior and

kindness will be fondly remembered forever. Her training in anti-infective research not only

refined my knowledge of things but also at the same time introduced me to lots of other

fascinating research areas. Her valuable experience in the field equipped me with the technical

skills needed to combat the antibiotic resistance of nosocomial pathogens.

I am also very thankful to Dr. Amir Ali Abassi, Chairperson, National Center for Bioinformatics,

Quaid-i-Azam University, Islamabad, for his continuous support to the cause of promoting

research in general. I would also like to acknowledge Professor Dr. Muhammad Shahab, Dean

Faculty of Biological Sciences for the academic support towards my Doctorate affairs.

It is my privilege to thank Dr. Abbas Hassan and his team, Department of Chemistry, Faculty of

Physical Sciences, Quaid-i-Azam University, Islamabad, Pakistan for synthesizing derivatives of

computationally shortlisted inhibitors in our study. I am also highly thankful to Dr. Thanyada

Rungrotmongkol, Assistant Professor, Chulalongkorn University, Thailand, Dr. Kara E.

Ranaghan, University of Bristol, UK, Dr. Klaus R. Liedl, Professor, University Innsbruck and

Dr. Riaz Uddin, Karachi University for their support in computational molecular dynamics

simulations studies.

I also feel a deep sense of gratitude to Dr. Saad Raza, for his kind help, useful suggestions,

encouragement, and friendly behavior that enabled me to complete my work successfully.

I thank profusely the International Research Support Initiative Program (IRSIP) sponsored by

the Higher Education Commission (HEC), Pakistan for providing me the opportunity to work in

the laboratory of Professor Dr. Laurence Rahme, Massachusetts General Hospital/ Harvard

http://www.uibk.ac.at/

ii

Medical School, United States of America for a period of six months. It is through this program

I got the opportunity to interact with the vibrant scientific community.

I am extremely thankful to Pakistan-United States Science and Technology Cooperation Program

(Grant No. Pak-US/2017/360), Higher Education Commission (HEC) and International

Foundation for Science (IFS) for granting financial assistance during my research pursuit.

I highly appreciate and obliged to the help of Dr. Sumra Wajid Abbasi and Dr. Asma Abro for

their valuable scientific inputs and guidance. I could never forget the continuous support and

inspiring encouragement of Dr. Sumra Wajid Abbasi during difficult phases of my Ph.D.

I still remembered that cup of tea session with my lab brothers, Mr. Farhan Ul Haq and Ghulam

Abbas during the early times of my Ph.D. I also sincerely thanks to Hira Jabeen, Mawra, Gul

Sanober, Noor Ul Ain Sajid, Iqra Ahmed, and Sundus Iqbal for introducing me to the world of

Microbial Informatics and from there I fall in love with this area of biological research.

I take this opportunity to express my deep sense of gratitude to the group of five: Sheneela Baseer,

Yelda Asad, Qurat ul Ain, Zunera Khalid, and Nosheen Ehsan for their co-operation, constant

encouragement and understanding that are indeed the sustaining factors in accomplishing this

work. I owe thanks to Faiza Saddique for her gesture and good will.

I would like to thank Naima Javed, Komal Aslam, and Tayyaba for the lovely time we spend

together in the fight against bacterial pathogens.

I sincerely thank my Ph.D. fellow Afifa Naveed who made the time spend at lab memorable and

joyful. I will prefer to call her “A family lady”.

Lovely thanks go to my lab members: Rozina, Rabia Farid, Naila Zaman, Nosheen Parvaiz, Rida

Sajjad, Anita Zaib, Iqra Zafar, Syeda Dure Zahra, Bilal Shakir (Gym boy), Uzair Ali Murtaza,

Azka, Maham, Zahra (Junior), Zartasha, Kinza, Laila, Tayeba, Saba, Ammara, and Zahra.

I would like to acknowledge the support offered by Chaudhary Muhammad Yasir during my stay

in the lab.

My acknowledgment also goes to the all non-academic and technical staff of National Center for

Bioinformatics especially Mr. Naseer Ahmed (senior), Mr. Naseer Ahmed (junior), Mr. Yasir

Mehmood Abbasi (computer programmer), Mr. Ali, Mr. Robin, and Mr. Mansoor.

I am greatly indebted to my loyal friends Salman Khan, Faisal Ahmad, Sabir Nawaz, Muhammad

Arshad, Muhammad Talal Amin, and Bilal Shakir who are like a source of oxygen in my dull life.

The time spends with them keeps me energetic and moving.

iii

My expressions are still begging the words to pay gratitude to my beloved father (Haji Said

Hassan), mother, brothers (Majid Khan and Kamran Hassan), Sisters, Fatima (late), Adaan

(late), Emaan (late), Gul Hassan (late), Moor (late) and all members of my family who softened

my heart forever and filled my heart with love for this beautiful creation of Allah-the humans. I

would like to separately mention my most beloved cousin and friend Mr. Taj Muhammad (Audit

Officer) for continuous backing and providing me the maximum comfort through his presence.

Sajjad Ahmad

iv

List of Figures

Fig.1.1 The major nosocomial infections due to A. baumannii……………………………..9

Fig.1.2 Resistance mechanisms in A. baumannii: (I) β-lactams; (II) aminoglycosides; (III)

quinolones; (IV) colistin. AME, aminoglycoside modifying enzyme; LPS,

lipopolysaccharide; OMP, outer membrane porin; PBP, penicillin-binding

protein………………………………………………………………………...........17

Fig.2.1 A schematic view of genomic based RV approach…………………………………42

Fig.3.1 The complete hierarchy of steps applied in the current study…………………........71

Fig.3.2

The shortlisted 14 proteins for virulent protein analysis. C.P., Complete proteome,

N.R.P., Non-redundant proteome, N.H.P., Non-homologous proteome, E.P.,

Essential proteome, O.M., Outer membrane, E.C.,

Extracellular……………………………………………………………………...77

Fig.3.3 Tertiary structure of the shortlisted potential vaccine proteins. From left to right,

BamA, FimD and Rhs……………………………………………………………84

Fig.3.4 Exo-membrane topology of the shortlisted antigenic epitopes on their respective

protein surface…………………………………………………………………....85

Fig.3.5 Interacting network of (A) BamA and (B) FimD. Empty balls are proteins with

unknown 3D structure while filled balls are proteins with known 3D structure….....86

Fig.3.6 Secondary structure of the multi-epitope peptide vaccine…………………………87

Fig.3.7 (A) Tertiary structure (B) ProsA Z-score and (C) Ramachandran plot of the multi-

epitope peptide vaccine…………………………………………………………….89

Fig.3.8 Intrinsic disorder graph predicting that the majority of the multi-epitope peptide

regions are below the threshold value………………………………………………90

Fig.3.9 In silico prediction of the cloning of the final multi-epitope peptide vaccine-construct

(red) into pET28a expression vector…………………………………………….…92

Fig.3.10 The best predicted binding side of the TLR4 receptor protein. The red spheres

indicate the active site region of the protein………………………………………..92

Fig.3.11 Binding conformation of the multi-epitope peptide construct at the binding site of

Chain A and C (A), Surface view of the binding pose (B), Closer view of the binding

pose, (D) Binding interactions between multi-epitope peptide construct and

TLR4……………………………………………………….………………………95

Fig.3.12 Vaccine-construct adjustments at the active site of TLR4 receptor during the

simulation period. The multi-epitope peptide is in gold, while the TLR4 receptor is

in dark magenta…………………………………………………………………….97

Fig.4.1 Schematic representation of in silico framework for identification of putative vaccine

candidates against A. baumannii……………………………….…………………117

Fig.4.2 Subcellular localization analysis of virulent proteins……………………….…….121

Fig.4.3 Minimized 3D structure and pepitope analysis revealing expose topology of the

epitopes for CsuB (A) and EpsA (B) proteins………………….…………………129

Fig.4.4 The binding pose of CsuB (left) and EpsA (right) epitope in the binding pocket of

DRB1*0101 allele. Receptor protein is in light gray while epitopes are in cyan color.

v

Hydrogen bonds between receptor and epitopes with distances are also

shown……………………………………….…………………………………….130

Fig.5.1 Biosynthetic pathway for KDO in four sequential steps. The four Kds enzymes

involved in the pathway are also shown. KDO is a part of lipid A molecule of Gram-

negative bacteria outer membrane lipopolysaccharide……………………….…..144

Fig.5.2 Complete step by step flow of the methodology employed in the current study…..145

Fig.5.3 Interacting networks for prioritized 10 drug candidates…………….……………159

Fig.5.4 (a) Superimposing the best modeled KdsA structure (blue) over 4luo template (Red).

(b) Secondary structure of the best modelled KdsA structure……………………161

Fig.5.5 KdsA enzyme intrinsic disorder plot. The threshold value was set to 0.5. The residues

making disorder region of enzyme are plotted over the red line…………………163

Fig.5.6 Correlation coefficient between GOLD fitness score and Autodock Vina binding

energy for 10 best inhibitors shortlisted in the current study………………….…165

Fig.5.7 Binding mode and interactions of inhibitor-4636 in the binding pocket of KdsA

enzyme……………….…………………………………………………………...166

Fig.5.8 (A) Binding mode of the best-characterized inhibitor in KdsA enzyme cavity. (B)

Closer view…………………………………………….…………………………167

Fig.5.9 Statistical properties of KdsA enzyme and KdsA-inhibitor 4636 complex………173

Fig.5.10 (A) Superimposition of initial conformation of KdsA (dark khaki) over that obtained

after 2 ns (purple). (B) Conversion of N-terminal loop into helix of KdsA at 2

ns…………………………………….……………………………………………174

Fig.5.11 MM/GBSA based free energy decomposition into KdsA residues and inhibitor…178

Fig.6.1 Multiple sequence alignment of KdsB sequences from different bacterial species,

4FCU (A. baumannii), 3K8D (E.coli), 4XWI (P. aeruginosa), 3TQD (C. Burnetii),

3QAM (V. Cholerae), 3JTJ (Y. Pestis). Sequences in green box represent the major

conserved catalytic pocket of KdsB enzyme……………………………………..202

Fig.6.2 Binding interactions of ligand into protein active pocket from GOLD analysis (left)

and AD-Vina (right)………………………………………………………………204

Fig.6.3 The back and forth rotating movement of inhibitor in KdsB enzyme. The coils, helix,

strand, and inhibitor are colored green, cyan, and magenta, respectively………..208

Fig.6.4 Statistical parameters for analyzing docked enzyme-inhibitor complex through

RMSD (A), RMSF (B), β-factor (C) and Rg (D)………………………………… 209

Fig.6.5 Binding interactions between ligand and enzyme active site residues, during different

time scale of simulation (a) at 50 ns, (b) 100 ns, (c) at 150 ns, and (d) at 200

ns………………………………………………………………………………….211

Fig.6.6 RDF graphs for hotspot amino acids involved in stability of enzyme-inhibitor

complex stability towards the last 10 ns of simulation period……………………213

Fig.6.7 Total binding free energy decomposition per residue of the receptor enzyme based

on MM/GBSA method for KdsB-226 complex…………………………………..216

Fig.6.8 Total binding free energy for 1000 frame extracted from 200 ns of simulation

trajectories……………………………………………………………………….. 219

Fig.6.9 The binding energy of KdsB residues that contribute significantly to the overall

binding affinity of the complex………………………………………………….. 220

Fig.7.1 The three dimensional structure of Ddl monomer. The C-terminal, Central and N-

terminal domains are shown in orange, purple and cyan, respectively……………237

vi

Fig.7.2 Multiple sequence alignment of Ddl enzymes from different bacterial enzymes.

(2187, S. aureus), (3E5N, X. oryzae), (3LWB, M. tuberculosis), (3R23, B.

anthracis),(4FUO, E. faecalis), (5D8D, A. baumannii), (3V4Z, Y. pestis), (4DGJ,B.

xenovorans), (4EGO, B. ambifaria), (4EGQ, B. pseudomallei)…………………238

Fig.7.3 Correlation coefficient between GOLD fitness score and AutoDock Vina binding

energy for the top ten best inhibitors…………………………………………….. 240

Fig.7.4 2D representation of binding mode and interactions of the best characterized drug-

like inhibitor in the active pocket of Ddl enzyme…………………………………241

Fig.7.5 (A) Binding mode of compound-331 in active pocket of KdsA enzyme. (B) Closer

view……………………………………………………………………………… 242

Fig.7.6 RMSD (A), RMSF (B), β-factor (C), and Rg (D) for enzyme and enzyme-inhibitor

complex………………………………………………………………………….. 252

Fig.7.7 RDF graph for: A. Lys-176-O-Lig-H13, B. Lys176-O-Lig-H14, C. Lys176-O-Lig-

N3, D. Trp177-N-Lig-H13………………………………………………………..253

Fig.7.8 AFD:A. Lys-176-O-Lig-H13, B. Lys176-O-Lig-H14, C. Lys176-O-Lig-N3, D.

Trp177-N-Lig-H13……………………………………………………………….254

Fig.7.9 Inhibitor movement from 0-ns to 100-ns………………………………………….255

Fig.7.10 MM/GBSA based binding free energy decomposition into each residue of the

enzyme. Amino acids are reppresented by a single lettere code………………….259

Fig.8.1 MurC-inhibitor complex in the TIP3P water box. The MurC enzyme is shown in

yellow cartoon while the inhibitor is in red CPK………………………………….276

Fig.8.2 A. Superimposition of Modeled MurC structure (Red) over 4HVA template (Blue),

B. 3D structure of the modeled MurC protein three different domains are shown..282

Fig.8.3 Correlation coefficient among scoring functions…………………………………284

Fig.8.4 Binding mode of the inhibitor in GOLD (A) and AutoDock/Vina (B)……………287

Fig.8.5 Binding interactions of the inhibitor in GOLD (A) and AutoDock/Vina (B)…….288

Fig.8.6 MD simulation trajectories analysis. A. RMSD, B. RMSF, C. Rg, D. β-factor…..291

Fig.8.7 A.Superimposed protein at 10th-ns (Dark Khaki) over 0-ns (Coral). B. Superimposed

inhibitor at 10th-ns (Dark Khaki) over 0-ns (Coral)………………………………292

Fig.8.8 The superimposed complex of 20th-ns (Dark Khaki) over 10th-ns (Coral)……… 293







Fig.8.15 A.Asp334-OD1-Lig-HN, B. Asp334-OD2-Lig-HN……………………………...302

Fig.8.16 AFD for Asp334-OD1 and inhibitor HN atom……………………………………304

Fig.8.17 AFD for Asp334-OD2 and inhibitor HN atom……………………………………305

Fig.9.1 The schematic workflow illustrating complete hierarchy of docked protein

analysis…………………………………………………………………………... 327

Fig.9.2 Correlation coefficient between GOLD fitness scores and AD-Vina binding

energies…………………………………………………………………………...329

Fig.9.3 2D depiction of docked compound 114 into the active site of AbMurF………….331

Fig.9.4 RMSD (A), RMSF (B), β-factor (C) and Rg (D) plot for the Gold docked

complex………………………………………………………………………….. 336

vii

Fig.9.5 A. RDF graph of Thr42 (hydrogen) and ligand (oxygen).B. RDF graph of Thr42

(nitrogen) and ligand (oxygen).C. RDF graph of Thr42 (oxygen) and ligand

(oxygen). D. RDF graph of Asp43 (hydrogen) and ligand (oxygen)……………..339

Fig.9.6 AFD graphs for Thr42 and Asp43 before simulation (A) and after simulation

(B)………………………………………………………………………………...340

Fig.9.7 Energy values vs number of frames from MM/GBSA calculations………………343

Fig.9.8 Energy values vs number of frames from MM/PBSA calculations………………344

Fig.9.9 Decomposition of MM/GBSA free energy per residue of the protein……………346

Fig.10.1 Step-wise flow of the methodology used in the current study……………………359

Fig.10.2 Active and inactive homodimer of A. baumannii BfmR enzyme…………………360

Fig.10.3 Correlation coefficient among different docking scores used in the study……….367

Fig.10.4 Binding interactions of the best screened inhibitor at BfmR docked site…………368

Fig.10.5 MD simulation analysis for BfmR-inhibitor complex. A. RMSD, B. RMSF, C. Rg,

D. β-factor………………………………………………………………………...374

Fig.10.6 Hydrogen bonds analysis of BfmR-inhibitor complex over the course of

simulation………………………………………………………………………...375

Fig.10.7 Inhibitor movement from phosphorylation site to α4-β5-α5 face of the enzyme

receiver domain during simulation……………………………………………….377

Fig.10.8 MM/GBSA based decomposition of binding free energies into enzyme residues. The

number around the circle represents the residue number of enzyme………………380

Fig.10.9 MM/PBSA based decomposition of binding free energies into enzyme residues. The

number around the circle represents the residue number of enzyme………………381

viii

List of Tables

Table 1.1 A. baumannii virulence factors………………………………………….. 14

Table 3.1 Fourteen proteins with subcellular localization of outer membrane and

extracellular matrix……………………………………………………..

78

Table 3.2 Predicted B-cell derived T-cell epitopes for potential three vaccine

proteins…………………………………………………………………..

83

Table 3.3 Structural evaluation of predicted five models of the multi-epitope

peptide vaccine……………………………………………………………

88

Table 3.4 Top 10 generated models by PatchDocK…………………………………. 93

Table 3.5 FireDock refinement of PatchDocK models……………………………... 94

Table 4.1 Prioritized epitopes that can elicit both humoral and Cell mediated

immunity………………………………………………………………….

126

Table 5.1 Structure evaluation of the predicted structures for KdsA enzyme using

different tools. R.M.F.R, Residues in Most favorable region, R.A.A.R,

Residues in additionally allowed region, R.G.A.R, Residues in

generously allowed regions, R.D.R, Residues in Disallowed

regions………............................................................................................

162

Table 5.2 Top 10 inhibitors screened in the current study along with GOLD fitness

score, Autodock binding energy, and

druglikeness………………………............................................................

168

Table 5.3 Binding free energies for Kds-inhibitor 4636

complex…………………………………………………………………..

176

Table 5.4 Decomposition of free energy into components for active residues………. 177

Table 6.1 Top five best characterized natural compounds A. baumannii KdsB

enzyme……………………………………………………………………

203

Table 6.2 Binding free energy values for the top five docked

complexes………………………………………………………………...

214

Table 6.3 Binding free energy from Monte Carlo calculations in Waterswap……… 218

Table 7.1 Top ten best natural inhibitors shortlisted in the current

study……………........................................................................................

243

Table 7.2 Estimated MM/GBSA based binding free energy values for enzyme-

inhibitor complex…………………………………………………….......

256

Table 7.3 WaterSwap based absolute binding free energy calculation for enzyme-

inhibitor complex…………………………………………………………

258 Table 8.1 The contribution of different energies involved in complex formation

between MurC and the top inhibitor………………………………………

307 Table 8.2 WaterSwap estimation of absolute binding free energy for the MurC-

inhibitor complex………………………………………………………....

309

Table 9.1 Top 5 best ranked and active compounds against AbMurF……………….. 330

Table 9.2 Residues of target protein involved in hydrogen, hydrophobic and ionic

interactions with the ligand……………………………………………….

332

Table 9.3 Binding energy values for the complex………………………………….. 341

Table 10.1 Docking scores of shortlisted ten inhibitors………………………………. 369

ix

Table 10.2 SwissADME and preADMET analysis of top ten hits screened in this

study……………………………………………………………................

376

Table 10.3 Binding free energies for the complex……………………………………. 378

Table.10.4

.

WaterSwap calculations for the complex……............................................ 379

x

Abbreviations 2D Two Dimensional

3D Three Dimensional

Å Angstrom

A. baumannii Acinetobacter baumannii

ADMET Absorption, Distribution, Metabolism, and Excretion-Toxicity

AP Alchemical Perturbation

AFD

B. anthracis

BBB

Axial Frequency Distribution

Bacillus anthracis

Blood-Brain Barrier

β-factor

BLAST

Beta Factor

Protein Basic Local Alignment Search Tool

BSIs Bloodstream Infections

B. ambifaria

B. pseudomallei

X. xenovorans

CADD

Burkholderia ambifaria

Burkholderia pseudomallei

Burkholderia xenovorans

Computer-Aided Drug Designing

CAI Codon Adaptation Index

CARD Comprehensive Antibiotic Resistance Database

CDC Center for Disease Control and Prevention

CD-HIT Cluster Database At High Identity With Tolerance

CNS Central Nervous System

COG Cluster of Orthologous Genes

CRAB Carbapenem-Resistant Acinetobacter baumannii

CRV Comparative Reverse Vaccinology

Ddl D-Alanine-D-Alanine Ligase

DEG Database of Essential Genes

E. coli

E. faecalis

ESBLs

Escherichia coli

Enterococcus faecalis

Extended-Spectrum β-Lactamases

ESKAPE FEP

E. faecium, S. aureus, K. pneumonia, P. aeruginosa,and Enterobacter Free Energy Perturbation

GAFF General Amber Force Field

GOLD Genetic Optimization For Ligand Docking

HAP Hospital-Acquired Pneumonia

HTS High-Throughput Screening

IC50 Half Maximal Inhibitory Concentration

ICU Intensive Care Unit

LCPO Linear Combinations Of Pairwise Overlaps

LD50 Lethal Dosage 50

LPS Lipopolysaccharide

MBLs Metallo-Beta-Lactamases

MC Monte Carlo

MD Molecular Dynamics

MDAB Multi-Drug Resistant Acinetobacter baumannii

xi

MDR Multi-Drug Resistant

MHC Major Histocompatibility Complex

MIC Minimum Inhibitory Concentration

MM/GBSA Molecular Mechanics Generalized Born Surface Area Continuum Solvation

MM/PBSA Molecular Mechanics Poisson–Boltzmann Surface Area Continuum Solvation

MOE Molecular Operating Environment

MSA Multiple Sequence Alignment

M. tuberculosis

Ns

NBSIs

Mycobacterium tuberculosis nanoseconds

Nosocomial Bloodstream Infections

NIs Nosocomial Infections

NMR Nuclear Magnetic Resonance

NPs Nosocomial Pathogens

NRTIs Nosocomial Respiratory Tract Infections

NSSTIs Nosocomial Skin and Soft Tissue Infections

OMPs Outer Membrane Proteins

PATRIC Pathosystems Resource Integration System

PDB Protein Data Bank

PG Peptidoglycan

PGRV Pan Genomic Reverse Vaccinology

PNAG Ex-Poly-B-1-6-N-Acetylglucosamine

PPI Protein-Protein Interactions

P. aeruginosa

PrDOS

Pseudomonas aeruginosa Protein Disorder Prediction System

RDF

Rg

Radial Distribution Function

Radius of Gyration

RMSD Root Mean Square Deviation

RMSF Root Mean Square Fluctuation

RO3 Rule of Three

RO5 Rule of Five

RV Reverse Vaccinology

SP Subtractive Proteomics

SSIs Surgical Site Infections

SSTIs Skin and Soft Tissue Infections

S. aureus

STRING

Staphylococcus aureus

Search Tool For The Retrieval Of Interacting Genes/Proteins

TFF Tripos Force Field

TI Thermodynamic Integration

TMHMM Transmembrane Helices; Hidden Markov Model

TPSA Topological Polar Surface Area

TRAB Tigecycline Resistant Acinetobacter baumannii

UCSF University of California At San Francisco Chimera

UTIs Urinary Tract Infections

VAP Ventilator-Associated Pneumonia

VD Volume of Distribution

VFDB Virulent Factors Data Base

https://www.chemcomp.com/MOE-Molecular_Operating_Environment.htm

xii

VMD Visual Molecular Dynamics

VVA Velocity-Verlet Algorithm

WHO World Health Organization

WSRC Waterswap Reaction Coordination

X. oryzae

Y. pestis

Xanthomonas oryzae

Yersinia pestis

xiii

Research Articles Part of the Dissertation

1. Ahmad, S. Ranaghan, K.E., Azam, S.S. 2019. Combating tigecycline resistant

Acinetobacter baumannii: A leap forward towards multi-epitope based vaccine

discovery. European Journal of Pharmaceutical Sciences. 132, pp, 1-17.

2. Ahmad, S. and Azam, S.S., 2018. A novel approach of virulome based reverse

vaccinology for exploring and validating peptide-based vaccine candidates against

the most troublesome nosocomial pathogen: Acinetobacter baumannii. Journal of

Molecular Graphics and Modelling, 83, pp.1-11.

3. Ahmad, S., Raza, S., Uddin, R. and Azam, S.S., 2018. Comparative subtractive

proteomics based ranking for antibiotic targets against the dirtiest superbug:

Acinetobacter baumannii. Journal of Molecular Graphics and Modelling, 82, pp.74-92.

4. Ahmad, S., Raza, S., Abro, A., Liedl, K.R. and Azam, S.S., 2018. Toward novel

inhibitors against KdsB: a highly specific and selective broad-spectrum bacterial

enzyme. Journal of Biomolecular Structure and Dynamics, pp.1-20.

5. Ahmad, S., Raza, S., Abbasi, S.W. and Azam, S.S., 2018. Identification of natural

inhibitors against Acinetobacter baumannii D-alanine-D-alanine ligase enzyme: A

multi-spectrum in silico approach. Journal of Molecular Liquids, 262, pp.460-475.

6. Ahmad, S., Murtaza, U.A., Raza, S., Azam, S.S., 2019. Blocking the Catalytic

Mechanism of MurC Ligase Enzyme from Acinetobacter baumannii: An in Silico

Guided Study towards the Discovery of Natural Antibiotics. Journal of Molecular

Liquids. 281, pp.117-133.

7. Ahmad, S., Raza, S., Uddin, R. and Azam, S.S., 2017. Binding mode analysis, dynamic

simulation and binding free energy calculations of the MurF ligase from

Acinetobacter baumannii. Journal of Molecular Graphics and Modelling, 77, pp.72-85.

8. Ahmad, S., Shaker, B., Ahmad, F., Raza, S. and Azam, S.S., 2018. Moleculer dynamics

simulation revealed reciever domain of Acinetobacter baumannii BfmR enzyme as

the hot spot for future antibiotics designing. Journal of Biomolecular Structure and

Dynamics, pp.1-16.

xiv

Research Articles Not Part of the Dissertation

1. Ul Ain, Q., Ahmad, S. and Azam, S.S., 2018. Subtractive proteomics and

immunoinformatics revealed novel B-cell derived T-cell epitopes against Yersinia

enterocolitica: An etiological agent of Yersiniosis. Microbial pathogenesis, 125,

pp.336-348. (Equal Contributor)

2. Ahmad, S., Raza, S., Uddin, R., Rungrotmongkol, T. and Azam, S.S., 2018. From

phylogeny to protein dynamics: A computational hierarchical quest for potent drug

identification against an emerging enteropathogen “Yersinia enterocolitica”. Journal

of Molecular Liquids, 265, pp.372-389.

3. Asad, Y., Ahmad, S., Rungrotmongkol, T., Ranaghan, K.E. and Azam, S.S., 2018.

Immuno-informatics driven proteome-wide investigation revealed novel peptide-

based vaccine targets against emerging multiple drug resistant Providencia

stuartii. Journal of Molecular Graphics and Modelling, 80, pp.238-250. (Equal

Contributor).

4. Ahmad, S., Navid, A., Akhtar, A.S., Azam, S.S., Wadood, A. and Pérez-Sánchez, H.,

2018. Subtractive genomics, molecular docking and molecular dynamics simulation

revealed lpxc as a potential drug target against multi-drug resistant Klebsiella

pneumoniae. Interdisciplinary Sciences: Computational Life Sciences, pp.1-19.

5. Baseer, S., Ahmad, S., Ranaghan, K.E. and Azam, S.S., 2017. Towards a peptide-based

vaccine against Shigella sonnei: A subtractive reverse vaccinology based

approach. Biologicals, 50, pp.87-99. (Equal Contributor)

6. Ehsan, N., Ahmad, S., Azam, S.S., Rungrotmongkol, T. and Uddin, R., 2018. Proteome-

wide identification of epitope-based vaccine candidates against multi-drug resistant

Proteus mirabilis. Biologicals, 55, pp.27-37. (Equal Contributor).

xv

Submitted for Publication

1. Sajjad. R., Ahmad, S., Raza, S., Azam, S.S. 2019. Target based virtual screening of

asinex antibacterial library against Acinetobacter baumannii division protein- the

FtsZ. Transition on Computational Biology and Bioinformatics. (Equal Contributor).

2. Rida Sajjad, Sajjad Ahmad, Syed Sikander Azam. 2019. Subtractive proteomics based

prioritization of potential vaccine candidates and immunoinformatics design of a

multi-epitope vaccine for Acinetobacter nosocomialis. Journal of Molecualr Graphics

and Modelling (Under revisions- JMGM_2019_270).

xvi

Summary

Computer aided vaccine and drug designing are emerged as powerful approaches for over three

decades playing critical roles in the development of new vaccines and drug molecules for

bacterial pathogens, respectively. The present dissertation focused primarily on the applications

of these two fields to make available all possible vaccines and drug targets in the sequenced

genome of selected nosocomial pathogens especially the Acinetobacter baumannii. Further,

computational structure modeling, structure based high throughput screening, molecular

simulations and binding free energies calculation studies were also taken into account to elucidate

the structural and functional characteristics of shortlisted biological systems in context of

screened antigens and small drug molecules. The first chapter of this dissertation addresses a

general introduction of nosocomial infections and nosocomial pathogens with emphasis on A.

baumannii, thus providing the background and motivation for the current research objectives.

The second chapter is focused on the theoretical details of computational techniques and analysis

employed for identification of antigenic peptides and druggable protein targets/drugs. The third

chapter describes a multi-epitope peptide vaccine designing for tigecycline-resistant A.

baumannii superbug. In this chapter, a comprehensive computational framework is designed

keeping in view the limitations of conventional subunit and peptide vaccines. A multi-epitope

peptide vaccine is formulated by linking the shortlisted B-cell derived T-cell antigenic

eptiopes from prioritized vaccine proteins that fulfilled the requirements of appropiate vaccine

candidates. Further, molecular docking and molecular dynamics (MD) simulation studies have

been undertaken to probe the binding conformation and dynamics of the modeled peptide with

respect to the TLR4 receptor. In the fourth chapter, a novel virulome based reverse

vaccinology (RV) approach is demonstrated to predict broad-spectrum antigenic peptides

harboring proteins for induction of targeted immune responses against multi-drug resistant A.

baumannii. The fifth chapter deals with the identification of promising and broad-spectrum

drug targets for A. baumannii using an extensive comparative subtractive proteomics

methodology for 35 strains of A. baumannii. In total, 10 protein targets: KdsA, KdsB, LpxA,

LpxC, LpxD, GpsE, PhoB, UvrY, KdpE and OmpR were identified as ideal candidates for

designing novel antibiotics. Further in this chapter, KdsA protein from 3-deoxy-Dmanno-

octulosonate (KDO) biosynthesis pathway was used as a receptor macromolecule in computer

xvii

aided drug designing applications of structure modeling, virtual screening of Asinex antibacterial

library, dynamics understanding and binding free energy calculations. The sixth chapter of the

dissertation focuses majorly on KdsB enzyme dynamics in the presence of an inhibitor in its

cavity and binding free energy calculations. D-alanine-D-alanine ligase (Ddl) enzyme of the

peptidoglycan biosynthesis machinery was targeted for screening of potent antibacterial drugs in

the seventh chapter. Radial Distribution Function (RDF) and an in-house developed Axial

Frequency Distribution (AFD) demonstrated Lys176 and Trp177 as critical residues from the

enzyme active site for binding, anchoring and bridging strong hydrogen and hydrophobic

contacts with the virtually screened inhibitor. In the eighth chapter, MurC ligase enzyme of the

peptidoglycan biosynthesis was targeted to block its catalytic mechanism by identifying the most

promising inhibitor for the Ligand binding (LB) domain of the enzyme. The complex was further

analyzed for free energies calculation using MM(PB/GB)SA and WaterSwap. At residue level,

RDF and AFD illustrated Asp334 as the most critical amino acid that drives recognition and

binding of the shortlisted compounds. An ethyl pyridine substituted 3-cyanothiophene was

predicted in the second last chapter as the most active inhibitor for A. baumannii MurF ligase

enzyme that catalyzes the final cytoplasmic step of bacterial peptidoglycan biosynthesis. Protein

active site residues: Thr42 and Asp43 were found to show high affinity for inhibitor binding

during simulation studies. The final chapter of the dissertation revealed α4-β5-α5 face of A.

baumannii BfmR enzyme as the hot spot for future antibiotics designing. In conclusion, the

findings of this dissertation could provide new foundations for discovery of novel therapeutics

against the notorious A. baumannii.

1

Chapter # 1

Introduction to Nosocomial Infections and

Nosocomial Pathogens

1.1.Nosocomial Infections

Hospital-acquired infections (HAIs), or nosocomial infections (NIs) refer to any systemic or

localized infection that develops because of the reaction of an infectious toxin or agent, acquired

under medical care in hospital milieu, healthcare and rehabilitation facility, nursing home,

outpatient clinics and other clinical settings (Hassan, Aftab, & Riffat, 2015; Khan, Baig, &

Mehboob, 2017). Nosocomial pathogens (NPs) are the organisms responsible for such

infections, and include bacteria, mycobacteria, viruses, fungi and protozoan parasites (Elliott &

Justiz-Vaillant, 2018). The most common types of NIs are: urinary tract infections (UTIs),

bloodstream infections (BSIs), ventilator-associated pneumonia (VAP), gastroenteritis,

meningitis, skin and soft tissue infections (SSTIs) and surgical site infections (SSIs) (Hassan,

Aftab, & Riffat, 2015; Khan, Baig, & Mehboob, 2017). Transmission of these infections to the

susceptible person occur through different modes: the spread of infection through health care

staff, bed lines, air droplets and most importantly, through contaminated equipment (Ducel,

Fabry, Nicolle, Organization, & others, 2002). In fact, such infections originate from another

infected patient, the outside environment, infected staff and, sometimes, from the patient’s own

skin microbiota (Khan, Baig, & Mehboob, 2017). The CDC suggests that approximately 1.7

million NIs cause 99,000 deaths in the United States per year (Klevens et al., 2007). According

to the World Health Organization (WHO), 15% of the patients that attend hospitals got infected

with NIs (Sydnor & Perl, 2011). NIs cause 4%-56% of all deaths in neonates, with a 75%

incidence rate in the sub-Saharan Africa and South-East Asia (WHO, 2011). The incidence is

also high in developed countries i.e. 3.5%-12%, while in under-developed and developing

countries it varies between 5.7% to 19.1%. The overall infection rate in low-income countries

2

is three times higher compared to high-income countries, specifically in neonates, among which

it gets 3-20% higher (Nejad, Allegranzi, Syed, Ellis, & Pittet, 2011). In Europe, Gram-negative

NIs cause two-thirds of the 25,000 deaths annually (Singh, 2016). The majority of the NPs

exhibit resistance to broad-spectrum antibiotics, hence, largely complicating the treatment

course. Therefore, in the absence of vaccines against these microorganisms that would allow

for timely prevention, multi-drug resistant NPs pose an alarming threat to human populations

worldwide (WHO, 2017). This entails quest for the identification of novel vaccines and drug

targets to develop novel therapeutics capable of countering the dissemination of drug-resistant

NP clones (Tacconelli et al., 2018).

1.1.1. Urinary Tract Infections (UTIs)

A urinary tract infection (UTI) is an infection of the urinary system (Flores-Mireles,

Walker, Caparon, & Hultgren, 2015). Infection of the kidneys is known as pyelonephritis, while

cystitis is an infection of the urinary bladder (Lane & Takhar, 2011). Symptoms of pyelonephritis

include flank pain and fever, while frequent urination, pain during urination and urgent feeling

to urinate in spite of having an almost empty bladder are the most common symptoms of cystitis

(Jhang & Kuo, 2017). Catheter-associated UTIs are the most frequent cause of NIs and are linked

to increased health care costs and patient morbidity (Frank, Wilson, Amand, & Pace, 2009). E.

coli is the most common bacterium responsible for UTIs, though, less frequently, other bacteria

and fungi may also be the causative agents (Behzadi et al., 2010). The common risk factors for

UTIs include sexual intercourse, female anatomy, diabetes, and obesity (Flores-Mireles, Walker,

Caparon, & Hultgren, 2015). Individuals using indwelling catheters allow the spread of bacteria

to the normally sterile bladder environment. There, the uroepithelium provides an excellent

attachment site for bacteria, thus facilitating the long-term colonization to the host cell surface

mediated by fimbria adhesions (Elliott & Justiz-Vaillant, 2018). It has been reported that

indwelling medical devices are associated with about 80% of the NIs and are predicted to be

mainly biofilm mediated (Guiton, Hung, Hancock, Caparon, & Hultgren, 2010). In any given

year, about 150 million people are expected to develop UTIs (Flores-Mireles, Walker, Caparon,

& Hultgren, 2015). Women are more susceptible to bacterial borne UTIs compared to men: up

to 10% of women present with at least one UTI episode per year (Nicolle, 2008). Most

frequently, they occur between the ages of 16 and 35 years (Salvatore et al., 2011). A study

3

carried out in Spain revealed an incidence rate of 8.2% for catheter-associated UTIs, with E. coli

being the most common isolated pathogen, followed by P. aeruginosa and Enterococcus specie.

In terms of resistance, 41.9% of the E.coli cases showed resistance to quinolones, 33.3% of

which were due to the production of extended spectrum β-lactamases. The resistance rate in P.

aeruginosa was 42.1% for quinolones and 21.1% for carbapenems (Jimenez-Alcaide et al.,

2015). The first-line therapy for treating uncomplicated bacterial cystitis includes a single dose

of 3 grams of fosfomycin tromethamine or a 5-day course of nitrofurantoin, while the second-

line of treatment includes β-lactams and fluoroquinolones (Bartoletti, Cai, Wagenlehner, Naber,

& Johansen, 2016). UTIs due to AmpC- β -lactamase-producer are treated with cefepime,

carbapenems, fosfomycin, fluoroquinolones, nitrofurantoin, and piperacillin-tazobactam (Elliott

& Justiz-Vaillant, 2018). Moreover, Enterobacteriaceae species producing extended-spectrum

β-lactamases (ESBL) are treated with aminoglycosides, carbapenems, cefoxitin, ceftazidime-

avibactam, ceftolozane, tazobactam, fosfomycin, fluoroquinolones, nitrofurantoin, and

piperacillin-tazobactam (Bartoletti, Cai, Wagenlehner, Naber, & Johansen, 2016). For

carbapenem-resistant Enterobacteriaceae, treatment options include aminoglycosides,

aztreonam, colistin, ceftazidime-avibactam, polymyxin B, fosfomycin, and tigecycline.

Therapeutic options for UTIs caused by multi-drug resistant (MDR) P. aeruginosa include

cefepime, aminoglycosides, ceftazidime, fluoroquinolones, carbapenems, ceftazidime-

avibactam, ceftolozane-tazobactam, colistin, and piperacillin-tazobactam (Bartoletti, Cai,

Wagenlehner, Naber, & Johansen, 2016; Elliott & Justiz-Vaillant, 2018).

1.1.2. Nosocomial Respiratory Tract Infections (NRTIs)

In the United States, NRTIs are the cause of extreme mortality and morbidity affecting

5 to 10 patients out of every 1,000. Nosocomial pneumonia is considered as the second most

common NI and responsible for 15% to 20% of all the NI episodes (American Thoracic Society,

2005). It refers to any type of pneumonia contracted by patients at least 48-72 hours following

admission to the hospital. Nosocomial pneumonia can be categorized into two types: ventilator-

associated pneumonia (VAP), which occurs in individuals receiving mechanical ventilation,

and hospital-acquired pneumonia (HAP), that infects individuals who frequently visit

healthcare environments (Mandell, Bennett, & Dolin, 2004). The risk of VAP is highest during

the initial course of hospital stay (Malani, 2012). Frequently, these infections are caused by

4

bacteria rather than a virus and are the primary cause of death in intensive care units (Mandell,

Bennett, & Dolin, 2004). Bacterial pneumonia causes 25% of all intensive care unit (ICU)

infections (Elliott & Justiz-Vaillant, 2018). The development of such NIs is dependent on two

physiological factors: colonization of the human respiratory tract by bacteria and decreased

immunity (Blot et al., 2014). The risk factors, such as: older age, anemia, admission to the ICU,

the need for mechanical ventilation, lymphocytopenia, and sepsis strongly predispose the

patient to severe influenza A pneumonia (Zhou et al., 2018). Aspiration of throat secretions

also commonly causes NRTIs (Terpenning, 2005). Dental plaques contribute to the spread of

these infections as well (Elliott & Justiz-Vaillant, 2018). A study performed in China illustrated

that NRTIs are commonly caused by A. baumannii, S. aureus, Stenotrophomonas maltophilia,

and P. aeruginosa (Elliott & Justiz-Vaillant, 2018). MDR E.coli and Klebsiella pneumonia are

increasingly reported to cause NRTIs in neonatal ICUs (Giuffre et al., 2016). NRTIs can be

treated with fluoroquinolones and erythromycin along with ceftazidime, imipenem, amoxicillin

with clavulanic acid and piperacillin/tazobactam (Elliott & Justiz-Vaillant, 2018). Ceftolozane-

tazobactam resistant P. aeruginosa strains are associated with higher patient death rates (Haidar

et al., 2017).

1.1.3. Nosocomial Bloodstream Infections (NBSIs)

Among critically ill patients, NBSIs are the main infectious hurdle, while those acquired

in ICUs are associated with significant morbidity and mortality (Martin, Mannino, Eaton, &

Moss, 2003). Even worse, patients with central venous catheters in place and those who are

immunocompromised are more prone to NBSIs (Ulrich, Santhosh, Mogle, Young, & Rao,

2017). Klebsiella and Enterococcus commonly cause NBSIs (Elliott & Justiz-Vaillant, 2018).

Studies show that certain medical and surgical procedures increase the probability of NBSIs as

demonstrated by a study in which 35% of the patients were found to have an episode of NBSIs

during venovenous extracorporeal membrane oxygenation (Kutlesa et al., 2017).

1.1.4. Nosocomial Central Nervous System Infections (NCNSIs)

NCNSIs arise from the deep structures of the brain parenchyma, from superficial

wounds, as well as from ventricular shunts, which require foreign-body insertion in the

5

ventricular system of the brain, are responsible for increased mortality and morbidity, similarly

to other types of NIs (Morris & Low, 1999; Elliott & Justiz-Vaillant, 2018). From the most

common NCNSIs are bacterial meningitis and central nervous shunt infections (Elliott & Justiz-

Vaillant, 2018). It has been estimated that 40% of bacterial meningitis episodes are

nosocomially based and Gram-negative bacilli were the culprit in 33% of such infections

(Durand et al., 1993; Elliott & Justiz-Vaillant, 2018). NCNSIs can be grouped into two types:

device-related or surgical and non-surgical infections (Whitson, Ball, Lollis, Balkman, &

Bauer, 2014). Reportedly, Mycoplasma hominis is one of the most prominent pathogens that

cause nosocomial meningitis following surgical procedures. Other pathogens known to cause

meningitis in the hospital setting are Acinetobacter, Enterobacter, E. coli, Klebsiella,

Pseudomonas, Serratia, S. aureus, and Streptococcus pneumonia. Nosocomial shunt infections

are most commonly caused by Ascaris, Bacillus, Corynebacterium, Clostridium, Cryptococcus,

Mycobacterium, Neisseria, and Yersinia species (Mace, 2008). Infections due to MDR Gram-

negative aerobic bacilli, penicillin-resistant Pneumococci, methicillin-resistant Staphylococci,

Aspergillus, Nocardia asteroids and Scedosporium apiospermum primarily affect the CNS in

immunocompromised patients and are largely problematic due to reduced sensitivity to drugs

(Nau, Sorgel, & Eiffert, 2010). NCNSIs are treated with antibiotics such as fluoroquinolones,

fluconazole, isoniazid, linezolid, metronidazole and pyrazinamide (Elliott & Justiz-Vaillant,

2018).

1.1.5. Nosocomial Skin and Soft Tissue Infections (NSSTIs)

NSSTIs are due to microbial invasion of the skin layers and the underlying soft tissues

and occur during trauma and surgery (Ramakrishnan, Salinas, & Agudelo, 2015). NSSTIs cause

cutaneous erythema, raised local temperature, edema and pain, while some pathogens can also

result in the appearance of violaceous bullae, skin sloughing, and tissue gas (Stevens et al.,

2014). Alcohol abuse, prolonged hospitalization, old age, immunosupression, and diabetes

mellitus are significant predisposing factors for the development of NSSTIs (Ki & Rotstein,

2008). Bacterial pathogens, such as S. aureus, E. coli, P. aeruginosa, and Enterococcus are

commonly associated with NSSTIs (Elliott & Justiz-Vaillant, 2018). The infections caused by

Streptococcus pyogenes, can be treated with the first line of penicillin, besides clindamycin,

macrolides, first-generation cephalosporin and expanded spectrum fluoroquinolones. Infections

6

due to E. coli, K. pneumoniae, and Serratia marcescens are treated with aminoglycosides β-

lactamase inhibitors carbapenems, cephalosporins, and fluoroquinolones (Casellas, 2011).

1.2. Nosocomial Bacterial Pathogens

The majority of the NIs are caused by the bacterial pathogens compared to other

infectious culprits and mainly involve the following pathogens: A. baumannii, Bacillus cereus,

Enterococci, S. aureus, coagulase-negative Staphylococci, Legionella, Proteus mirabilis,

Yersinia enterocolitica, Providencia stuartii, S. marcescenes, K. pneumonia, E. coli etc. (Khan,

Baig, & Mehboob, 2017). Among the aforementioned pathogens, S .aureus is a Gram-positive

non-spore forming cocci, immotile, coagulase and catalase positive and a facultative anaerobe

(Tong, Davis, Eichenberger, Holland, & Fowler, 2015). It persistently colonizes the nasal

passages of 20% of individuals, while 30% of the population exerts intermittent colonization

(Vandenesch, Lina, & Henry, 2012). Patients with suppressed immunity are more susceptible

to S. aureus superficial infections, abscess formation and deep tissue infections (Khan, Baig, &

Mehboob, 2017). S. aureus also produces toxins: the toxic shock syndrome toxin 1, which

causes toxic shock syndrome, and exfoliative toxins, which are responsible for the

staphylococcal scalded skin syndrome, are the prominent examples of crucial virulent factors

of S. aureus (Al Laham et al., 2015). Additionally, the pathogenesis of S. aureus infections is

also mediated by virulent enzymes and immune modulators (Vandenesch, Lina, & Henry,

2012). K. pneumonia is the eighth most significant health-care-related bacterial pathogen and

accounts for 3-7% of all the hospital-acquired infections (Lin, Wang, Wang, & Fung, 2015). It

is a Gram-positive opportunistic bacillus that belongs to the Enterobacteriaceae family and

usually colonizes the skin, the pharynx and the gastrointestinal tract (Nordmann, Cuzon, &

Naas, 2009). For its pathogenesis, it uses a capsular polysaccharide, endotoxins, and cell wall

receptors (Khan, Baig, & Mehboob, 2017). This pathogen is responsible for diseases, such as

wound infections, septicemia and pneumonia (Li, Zhao, Liu, Chen, & Zhou, 2014). E. coli is a

Gram-negative facultative anaerobethat is oxidase negative (Kaper, Nataro, & Mobley, 2004).

It colonizes the gastrointestinal tract of humans and other animals and it is responsible for

urinary tract infections, gastroenteritis, peritonitis, meningitis, pneumonia, and septicemia

(Lausch, Fuursted, Larsen, & Storgaard, 2013). The pathogenic potential of this bacterium is

due to the adhesive capability of some of its strains, its capsule, its endotoxins, and its type 3

7

secretion systems (Zhao, Yang, Huang, & Cai, 2015). The Enterococci genus is second on the

NI causative microorganism list and it is the leading pathogen in the United States accounting

for 20-30% of the infections. Enterococci are mainly involved in the urinary tract and wound

infections (Karki, Leder, & Cheng, 2015; Kaiser et al., 2015). For their pathogenesis,

Enterococci use adhesions, aggregation substances, cytolysin, extracellular surface proteins,

extracellular superoxide, hemolysins, and gelatinase (Sood, Malhotra, Das, Kapil, & others,

2008). P. aeruginosa is a nosocomial pathogen associated with high mortality and morbidity

and contributes to 11% of all nosocomial infections (Hassan, Aftab, & Riffat, 2015). P.

aeruginosa colonizes the upper respiratory tract, urinary tract, and kidneys. It usually causes

diseases, such as bacteremia, cystic fibrosis, pneumonia, urinary tract, wound, and surgical-site

infections (Aloush, Navon-Venezia, Seigman-Igra, Cabili, & Carmeli, 2006). It’s most

important virulent factors include exotoxins, hemolysins, adhesions, siderophores, and

proteases (Gellatly& Hancock, 2013). Clostridium difficile is mainly involved in nosocomial

diarrhea (Hassan, Aftab, & Riffat, 2015). It is a part of normal microbiota and colonizes the

intestinal tract (Kim et al., 2013). Under overgrowth circumstances, C. difficile associated

toxins can cause inflammation of the intestinal tract, called colitis, and is responsible for 15%

to 20% of the hospital-related diarrheal cases (Hassan, Aftab, & Riffat, 2015). The virulence

factors of this pathogen include its capsule, fimbriae, hydrolytic enzymes, and toxins (Borriello,

Davies, Kamiya, Reed, & Seddon, 1990). P. mirabilis is a facultative Gram-negative anaerobe

that is urease positive and demonstrates swarming motility (Foris & Snowden, 2017). This

pathogen causes 90% of all the Proteus infections in human. About 10-20% of the P. mirabilis

strains are resistant to ampicillin and first-generation cephalosporin (Schaffer & Pearson, 2015).

It accounts for sepsis, pneumonia and wound infections in hospitalized patients (Chen et al.,

2012). P. stuartii is a member of the Providencia genus and is mainly responsible for causing

human infections (Warren, 1986). This pathogen is seen in patients with long-term indwelling

urinary catheters or severe burns (Wie, 2015). Individuals of older ages are at greater risk for

acquiring P. stuartii infections (Choi, Kim, Kim, Park, & Uh, 2015). In neonates, it causes

diarrhea and it is the most common cause of purple urine bag syndrome (C.-H. Lin, Huang,

Chien, Tzeng, & Lung, 2008). Y. enterocolitica is a rod-shaped Gram-negative bacterium and

uses the gastrointestinal tract as the portal of entry to the human body, causing severe

gastrointestinal diseases (Fàbrega & Vila, 2012). Yersiniosis, caused by Y. enterocolitica is an

8

animal-borne disease (Cover & Aber, 1989). For its treatment aminoglycosides are used in

combination with doxycycline. Chloramphenicol, ceftriaxone, fluoroquinolones,

and trimethoprim-sulfamethoxazole are among other anti-microbial agents that are active

against this pathogen (Bottone, 1997).

A. baumannii is an opportunistic Gram-negative coccobacillus that usually affects

people with compromised immune systems (McConnell, Actis, & Pachón, 2013). It is

becoming an increasingly important nosocomial pathogen and it is identified as a member of

the ESKAPE group (Pendleton, Gorman, & Gilmore, 2013). A. baumannii is colloquially

referred as Iraqi bacter because of its sudden emergence in the military treatment facilities

during the war in Iraq and since then a continuous threat for soldiers and veterans who served

in Afghanistan and Iraq (Almaghrabi, Joseph, Assiry, & Hamid, 2018). The MDR strains of the

pathogen have spread to different civilian hospitals due to the transport of the soldiers to

multiple medical facilities (McQueary et al., 2012). Few antibiotics are effective against such

MDR A. baumannii and thus the knowledge of this pathogen pathogenesis, mechanisms of

antibiotic resistance, and prospective treatment options are important to disclose. As A.

baumannii is the main focus of this thesis, we will discuss this pathogen in detail.

1.2.1. A. baumannii Infections

VAP is an important clinical manifestation of A. baumannii in patients that receive

mechanical ventilation in the intensive care setting (Peleg, Seifert, & Paterson, 2008). VAP is

the result of airway colonization by A. baumannii, followed by the development of pneumonia

that has a crude mortality rate ranging between 40% and 70% (Blot et al., 2014). The A.

baumannii community-acquired pneumonia (CAP) mainly appears in patients with chronic

obstructive pulmonary disease (COPD) and in those with alcohol abuse habits, with the death

rate ranging between 40% and 60% (Eugenin, 2013). BSIs caused by A. baumannii are very

frequent in the ICU setting and are usually caused by intravascular devices (Jung et al., 2010).

These infections have a crude mortality rate of 28%-43% (Wisplinghoff et al., 2004). A.

baumannii also represents a significant cause of burn wound infections. These infections are

highly challenging for clinicians because of the high rate of resistance of the pathogen to

antibiotics and the poor penetration of antibiotics at the burn sites (Keen et al., 2010). Among

https://en.wikipedia.org/wiki/Chloramphenicol

https://en.wikipedia.org/wiki/Ceftriaxone

https://en.wikipedia.org/wiki/Fluoroquinolone

https://en.wikipedia.org/wiki/Trimethoprim-sulfamethoxasole

9

military personnel, this pathogen accounts for 22% of burn-site infections, 53% of which

demonstrate resistance to multiple drugs (Keen et al., 2010). A. baumannii soft tissue infections

are highly problematic in the aforementioned population especially in those who sustain

battlefield-related trauma (Sebeny, Riddle, & Petersen, 2008). These infections can also lead to

cellulitis and necrotizing fasciitis that in addition to the use of antibiotics, also require urgent

surgical debridement (McConnell, Actis, & Pachon, 2013). Additionally, skin/soft tissue

infections due to A. baumannii were seen in wounded survivors of the tsunami (Maegele et al.,

2005; Cascio et al., 2010). Moreover, A. baumannii is an infrequent cause of endocarditis

associated with intravascular catheters and prosthetic valves (Kumar, Vengadassalapathy, &

Menon, 2008). A. baumannii is predominately responsible for osteomyelitis in military

personnel and is a significant issue in US military operations in Afghanistan and Iraq (Schafer

& Mangino, 2008).The different types of A. baumannii infections are summarized in Fig.1.1.

Fig.1.1. The major nosocomial infections due to A. baumannii

(Dijkshoorn, Nemec, & Seifert, 2007).

10

1.2.2. Natural Habitats

A. baumannii has been frequently recovered from animals, humans, soil, sewage, food, and

water indicating its ubiquitous distribution in nature (Peleg, Seifert, & Paterson, 2008).

Additionally, it has also been isolated from fish, birds and rainbow trout, a variety of food items,

including fruits, dairy products, milk, and raw vegetables (Almasaudi, 2018). This pathogen

normally inhabits the human skin (Towner, 2006) and is isolated from respiratory tract

secretions and the pharyngeal mucosa of hospitalized patients (Munoz-Price & Weinstein,

2008). Its existence is also reported on the ear, nose, hand, vagina, throat, trachea, and

conjunctiva (Seifert et al., 1997). Furthermore, it is also recovered from moist areas of body i.e.

groin, axillae, and toe webs (Almasaudi, 2018). Frequently, A. baumannii is isolated from

reused medical devices, such as: respirometers, tubing, plastic urinals, and humidifiers

(Kanafani & Kanj, 2014). Mattresses, the skin of healthcare personnel, and pillows could also

be the source of A. baummanni (Beggs, Kerr, Snelling, & Sleigh, 2006).

1.2.3. Adherence, Biofilm Formation and Pathogenicity

A. baumannii has tremendous potential to withstand and prosper in extreme hospital

habitats mainly because of its interaction capacity with different kinds of surfaces (McConnell,

Actis, & Pachon, 2013). These surfaces involve those from the abiotic substrata in the hospital

settings like medical equipment, linen and furniture (Borer et al., 2005) as well as from biotic

surfaces like human epithelial cell (Jawad, Seifert, Snelling, Heritage, & Hawkey, 1998; Lee et

al., 2006). A. baumannii adherence to both these surfaces leads to the formation of a complex

and multicellular three-dimensional structure called biofilm. In this structure, the cells stick to

each other in a slimy extra-cellular matrix that comprises of extra polymeric substances

(Costerton, Stewart, & Greenberg, 1999). The persistence of A. baumannii in medical

environments, resistance to antibiotics, and causing different diseases in humans is largely

because of its ability to produce biofilm on solid surfaces (Gaddy et al., 2009). A. baumannii

also produces a complex biofilm-like structure on liquid media surface called pellicles

(McQueary & Actis, 2011). Biofilm and pellicle formation on abiotic surfaces among the

clinical isolates are quite variable and there is no apparent correlation between the bacterial

surface properties and the nature of substrata (McQueary & Actis, 2011). In A. baumannii

https://www.sciencedirect.com/topics/earth-and-planetary-sciences/trachea

https://www.sciencedirect.com/science/article/pii/S1319562X16000565#b0150

11

ATCC 19606 strain, studies have shown that biofilm production and adherence to abiotic

surfaces is mediated by the formation of pilus encoded by CsuA/BABCDE usher-chaperone

assembly system (Tomaras, Dorsey, Edelmann, & Actis, 2003). According to preliminary

observations, the pili in A. baumannii ATCC 19606 strain are different from that described in

A. baumanii ATCC 17978. In the latter case, pili are thin, long and tend to bundle (McConnell,

Actis, & Pachon, 2013). In A. baumannii strain 307-0294, volume and thickness of biofilm was

seen diminished because of mutation in a conserved outer membrane protein which resembles

staphylococcal biofilm-associated protein (Bap) (Loehfelm, Luke, & Campagnari, 2008). It is

also well documented that ex-poly-b-1-6-N-acetylglucosamine (PNAG) production and

excretion is critical for the formation of a biofilm (A. H. K. Choi, Slamti, Avci, Pier, & Maira-

Litrán, 2009). The two-component system which comprises a BfmR encoded response regulator

and BfmS encoded sensor kinase is involved in cell to cell communication, adaptation and

pathogenesis (Tomaras, Flagler, Dorsey, Gaddy, & Actis, 2008). BfmR inactivation has been

observed to block CsuA/BABCDE operon expression that leads to loss of pili and biofilm

formation (McConnell, Actis, & Pachon, 2013). Little is known about the adherence factors of

A. baumannii on biotic surfaces (McConnell, Actis, & Pachón, 2013). OmpA is illustrated to

play a key role in attaching the pathogen to C. albicans filaments and human epithelial cells

and could be vital in biofilm production (Gaddy, Tomaras, & Actis, 2009). A. baumannii

adherence and biofilm formation is a well-orchestrated process and a respond to different

cellular and environmental signals. For example, the expression of blaPER-1 resistance gene

allows the adherence to human epithelial cells and biofilm formation. Adherence and biofilm

formation are also controlled by cell population density and accordingly quorum sensing

signaling molecules are produced (Gonzalez, Dijkshoorn, den Barselaar, & Nudel, 2009). Three

types of quorum sensing signaling molecules have been identified in A. baumannii: the Rf1-

type sensor which is commonly found in isolates of A. calcoaceticus-baumannii complex, N-

acyl-homoserine lactone [i.e. N-3-hydroxydodecanoyl-homoserine lactone (3-OH-C12-HSL)]

molecule which is vital for biofilm formation on abiotic surfaces (Niu, Clemmer, Bonomo, &

Rather, 2008) and lastly, BlsA photoreceptor protein.

12

1.2.4. Iron Acquisition

Iron is wide spread in biological systems and environment. However, it is unavailable

to the cells under aerobic conditions because of its poor solubility and chelation by transferrin

and lactoferrin, (McConnell, Actis, & Pachon, 2013). Aerobic bacteria expresst high affinity

iron acquisition systems in response to limited iron that produce, export and uptake Fe3+

chelators, siderophores (Wandersman & Delepelaire, 2004). Additionally, some bacteria utilize

hemoglobin and heme as a source of iron, while other remove iron from lactoferrin and

transferrin (Zimbler et al., 2009; Skaar, 2010). In ATCC strain 17978, a poly-cistronic operon

is involved in heme transportation into the cytoplasm from periplasm (Smith et al., 2007). A.

baumannii can also use ferrous ions that are available under low oxygen conditions

(McConnell, Actis, & Pachon, 2013). The best-characterized chelation system in A. baumannii

is mediated by siderophore acinetobactin described initially in ATCC 19606 (Mihara et al.,

2004). Based on genomic and functional analysis, a 26.5-kb chromosomal region comprising

genes involved in the synthesis, transportation, and secretion of acinetobactin is identified

(Eijkelkamp, Hassan, Paulsen, & Brown, 2011). Taken together, the available in silico and

experimental observations illustrate that A. baumannii can acquire iron either by capturing iron

using acinetobactin or using heme as a source of iron (McConnell, Actis, & Pachon, 2013). A

recent report illustrated the crucial role of the acinetobactin-mediated system in persistence and

in the damage caused to animal cells (Gaddy et al., 2012).

1.2.5. Virulence Factors

In contrast to other Gram-negative bacterial pathogens, not much is known about the

virulence factors of A. baumannii (McConnell, Actis, & Pachon, 2013). A. baumannii

pathogenesis is attributed to the outer membrane proteins, verotoxins, toxic slime

polysaccharides and the cell surface. The cell surface is hydrophobic that aids in adhesion and

avoiding phagocytosis by phagocytic cells (Almasaudi, 2018). Probably, the best-characterized

A. baumannii outer membrane virulent protein identified to date is OmpA (C. H. Choi et al.,

2005). This protein is known to have a role in the bacterial adaptation in host cells, while it

increases the bacterial pathogenesis and resistance to antibiotics (Doughari, Ndakidemi,

Human, & Benade, 2011). Additionally, it promotes the production of apoptosis-inducing factor

13

and proapoptotic cytochrome c thus damaging the human airway cells during infection (C. H.

Choi, Lee, Lee, Park, & Lee, 2008). The OmpA protein also plays a role in bacterial adherence,

invasion to the epithelial cells and dissemination (C. H. Choi, Lee, Lee, Park, & Lee, 2008).

Furthermore, the OmpA protein contributes to the A. baumannii ability to avoid complement

mediated killing (S. W. Kim et al., 2009). In the environment, OmpA assists in biofilm

formation and surface motility (Clemmer, Bonomo, & Rather, 2011). The lipopolysaccharide

(LPS) structure is demonstrated to play a crucial role in neutrophil cytotoxicity, as well as in

the inhibition of neutrophil migration aand phagocytosis. Pathogenesis is enhanced with the

release of cytotoxins and extracellular enzymes that cause significant harm to the host tissue

especially during respiratory tract infections (Tomaras, Dorsey, McQueary, & Actis, 2008). The

capsular polysaccharide in addition to the LPS is identified as a virulent factor that protects A.

baumannii from the host innate immune system (Doughari, Ndakidemi, Human, & Benade,

2011). Similarly, phospholipase D, which is a lipolytic enzymes and catalyzes phospholipid

cleavage, increases the bacterial serum survival and enhances its capacity to invade epithelial

cells (Jacobs et al., 2010). It also contributes in the dissemination of the pathogen from the

lungs to other tissues (McConnell, Actis, & Pachon, 2013). Among other, esterases, acid

phosphatases, and amino peptidases are the secretory products that confer virulence to the

bacterium (Doughari, Ndakidemi, Human, & Benade, 2011). Verotoxins produced by

Acinetobacter can be grouped into two antigenic types: vtx-1 and vtx-2. These toxins belong to

the RNA N-glycosidases subfamily and target ribosome machinery, thus inhibiting protein

synthesis (Almasaudi, 2018). The different virulent factors identified in A. baumannii are

summarized in Table 1.1.

1.2.6. Epidemiology

A. baumannii infections have been a substantial clinical issue in Europe (Van Looveren,

Goossens, Group, & others, 2004) and outbreaks in hospitals in France, Germany, England,

Spain and the Netherlands have been reported (Fournier, Richet, & Weinstein, 2006). The

spread of multiple drug resistant isolates has been described on a national scale. This can be

exemplified by the spread of the Oxa-23 clones 1 and 2 and the Southeast clone in the Southeast

of England (Coelho et al., 2006), the spread of VEB-1 ESBL-producing strains between

hospitals in Southeast and Northern France (Naas et al., 2006), the amikacin-resistant strain

14

dissemination in Spain (Vila et al., 1999), and the spread of MDR strains in Portugal (Da Silva,

Dijkshoorn, Van Der Reijden, Van Strijen, & Duarte, 2007). Introduction of MDR strains and

their subsequent epidemic spread among different regions are mainly associated with

international transfer of colonized patients (Schulte et al., 2005).

Table 1.1. A. baumannii virulence factors (McConnell, Actis, & Pachon, 2013).

Virulence Factors Role in Pathogenesis

OmpA Adherence and invasion of the epithelial cells, induction of

apoptosis, serum resistance, surface mobility, biofilm formation

Outer membrane

vesicles

Act as a vehicle for virulence factors delivery into the host cells

cytoplasm, genetic material transfer between the bacterial cells

Lipopolysaccharide Host immune system stimulation and evasion

Phospholipase In vivo bacterial survival, serum resistance, bacterial

dissemination

Capsular polysaccharide Growth in serum, evasion of the host immune response

Acinetobactin-mediated

iron acquisition system Cell apoptosis, iron acquisition needed to persist in the host cell

Penicillin binding

protein Cellular stability, biosynthesis of peptidoglycan, serum growth

For example, the transfer of MDR strains from Southern to Northern European countries like

Germany and Belgium (Bogaerts et al., 2006). Moreover, inter-institutional outbreaks of three

A. baumannii clones i.e. European clone I, II and III are reported from hospitals of Spain, Italy,

Greece, and Turkey (southern European countries), as well as from hospitals in Denmark,

France, and Belgium (Northern Europe) (Van Dessel et al., 2004). The prevalence of

carbapenem-resistant A. baumannii is now a major health issue in many European countries

such as Spain, Italy, Greece, Turkey, and England (Dobrewski et al., 2006). A survey conducted

15

by an industry-supported surveillance report (MYSTIC) in the period of 2002-2004 from 48

European hospitals demonstrated that 73.1% of the isolates were susceptible to meropenem

(Unal & Garcia-Rodriguez, 2005), 69.8% to imipenem (Unal& Garcia-Rodriguez, 2005),

32.4% to ceftazidime, 34.0% to ciprofloxacin, and 47.6% to gentamicin (Unal& Garcia-

Rodriguez, 2005). Polymyxins resistant isolates are detected in some of the European countries

though these remain rare (Peleg, Seifert, & Paterson, 2008).

A. baumannii infection history in the United States is long. Carbapenem-resistant

outbreaks were observed in the hospitals of New York in 1991-92 (Go et al., 1994). According

to national surveillance studies, significant trends of resistance to multiple drugs emergence

across the United States have been documented (Weinstein, Gaynes, Edwards, & System,

2005). From 1986 to 2003, Acinetobacter strains showed profound resistance to amikacin (5%

to 20%), imipenem (0% to 20%), and ceftazidime (25% to 68%) (Weinstein, Gaynes, Edwards,

& System, 2005). An industry supported surveillance study carried out in the United States

from 2004 to 2005 found that only 60.2% of the Acinetobacter species were susceptible to

imipenem (Rhomberg & Jones, 2007). Substantial non-susceptibility of Acinetobacter species

was also reported for carbapenems (10% to 15%), aminoglycosides (10% to 30%),

ciprofloxacin/levofloxacin (35% to 40%), and ceftazidime/cefepime (35% to 40%) (Rhomberg

& Jones, 2007). In North America, reported polymyxin resistance oft A. baumannii is 1.7%

(Gales, Jones, & Sader, 2006), carbapenem resistance is 2.8% and MDR is 3.2% (Gales, Jones,

& Sader, 2006). In Latin America, Acinetobacter species resistance to imipenem, meropenem,

ciprofloxacin, ceftazidime, gentamicin, and piperacillin-tazobactam are thought to be the

highest around the globe (Unal & Garcia-Rodriguez, 2005). Several different carbapenemases

have been identified in isolates of A. baummanni including IMP-6 and IMP-1 in Brazil

(Tognim, Gales, Penteado, Silbert, & Sader, 2006), OXA-58 in Argentina (J. Coelho,

Woodford, Afzal-Shah, & Livermore, 2006), and OXA-23 in Brazil and Columbia (Villegas et

al., 2007).

In Africa, 30% of bloodstream isolates were shown to be carbapenem resistant, > 30%

were non-susceptible to levofloxacin and ciprofloxacin, and > 40% were resistant to

piperacillin-tazobactam and cefepime (Brink, Moolman, Da Silva, Botha, & others, 2007).

16

In Australia, the first documented outbreak of A. baumannii was reported from Western

Australia (Riley et al., 1996) where the isolates were resistant to ciprofloxacin, cephalosporins,

ticarcillin, and gentamicin. Outbreaks have been reported from major cities including Sydney,

Brisbane, and Melbourne that involve carbapenem-resistant strains (Peleg, Franklin, Bell, &

Spelman, 2006). Specifically, in French Polynesia, frequent outbreaks of carbapenem-resistant

strains were reported (Naas, Levy, Hirschauer, Marchandin, & Nordmann, 2005). Strains

resistant to tigecycline were also reported from Australia (Iredell, Thomas, Power, & Mendes,

2007).

In the Middle East and in Asian countries, various outbreaks of pan-drug-resistant

isolates have been documented (Ying, Ling, Lee, & Ling, 2006; K. Lee et al., 2006; Peleg,

Seifert, & Paterson, 2008; Xu et al., 2015; Almasaudi, 2018). Additionally, a variety of

carbapenemases are also reported from both the aforementioned regions (Peleg, Seifert, &

Paterson, 2008). Reportedly, isolates have been shown to exceed 25% of non-susceptibility to

meropenem and imipenem, 35% to amikacin, 40% to ceftazidime and cefepime, 45% to

ciprofloxacin, and 40% to ceftazidime and cefepime (Gales, Jones, & Sader, 2006). Even more

worrisome is the existence of polymyxin B (Ko et al., 2007) and tigecycline resistant A.

baumannii strains in these regions (Navon-Venezia, Leavitt, & Carmeli, 2007).

1.2.7. Antibiotic Resistance

The main resistance mechanisms in A. baumannii are illustrated in Fig.1.2. Antibiotic

resistance mechanisms in A. baumannii have been increased substantially in the past years.

Resistance to β-lactams, including carbapenems, has been described in various studies and is

largely due to the pathogen’s capability to adopt quickly to environmental changes.

Upregulating innate resistance and acquiring foreign genetic determinants are the critical

attributes of A. baumannii that enable this pathogen to be resistant to almost all clinically used

antibiotics (Peleg, Seifert, & Paterson, 2008). Enzymatic degradation of β-lactam antibiotics

by β-lactamases is the most prevalent resistance mechanism in A. baumannii (Peleg, Seifert, &

Paterson, 2008). Chromosomally encoded AmpC cephalosporinases that are inherent to almost

all A. baumannii strains are responsible for cleaving extended-spectrum cephalosporins

(Thomson & Bonomo, 2005). Extended-spectrum Beta-lactamases (ESBLs) from the Ambler

17

class A group have also been documented for A. baumannii (Peleg, Seifert, & Paterson, 2008).

Recent focus is shifted toVEB-1, which disseminates throughout the hospitals of France and

also reported from Belgium and Argentina (Poirel, Menuteau, Agoli, Cattoen, & Nordmann,

2003; Carbonne et al., 2005). PER-1 type of ESBL has been reported from the United States,

Belgium, Turkey, Korea, Romania, and France (Vahaboglu et al., 1997; Poirel et al., 2003;

Naas, Bogaerts, et al., 2006), while PER-2 has been documented from Argentina (Pasteran et

al., 2006).

Fig.1.2. Resistance mechanisms in A. baumannii: (I) β-lactams; (II) aminoglycosides; (III)

quinolones; (IV) colistin. AME, aminoglycoside modifying enzyme; LPS, lipopolysaccharide;

OMP, outer membrane porin; PBP, penicillin-binding protein (Asif, Alvi, & Rehman, 2018).

18

Other ESBLs characterized involve TEM-116 and TEM-92 from the Netherland and Italy,

respectively (Al Naiemi et al., 2005; Endimiani et al., 2007). SHV-12 is identified from China and

the Netherland (Huang, Mao, Chen, Wu, & Wu, 2004; Peleg, Seifert, & Paterson, 2008): CTX-M-

43 from Bolivia and CTX-M-2 from Japan (Danes et al., 2002; Peleg, Seifert, & Paterson, 2008).

The β-lactamases carbapenemase potency are of great concern and include the serine

oxacillinases (Ambler class D OXA type) and Metallo-Beta-lactamases (MBLs) (Ambler class B)

(Poirel & Nordmann, 2006). The first described OXA-type with carbapenemase potency was

identified in a clinical isolate from Edinburgh, Scotland (Paton, Miles, Hood, & Amyes, 1993).

This plasmid-encoded resistant enzyme, that was found to be transferable and was later named as

blaOXA-23, contributes globally to carbapenem resistance (Peleg, Seifert, & Paterson, 2008). The

blaOXA-23 gene cluster comprises of two closely related enzymes: OXA-27 and OXA-49 (Brown

& Amyes, 2005). Other than this, two acquired OXA-type gene cluster (blaOXA-58-like and

blaOXA-24-like) with carbapenemase activity have also been documented (Bou, Oliver, &

Martinez-Beltrán, 2000: Afzal-Shah, Woodford, & Livermore, 2001; Lopez-Otsoa et al., 2002:

Peleg, Seifert, & Paterson, 2008). BlaOXA-51-like gene (encoding OXA-51, - 64, - 65, - 66, - 68,

- 69, - 70, - 71, - 78, - 79, - 80, and - 82) is chromosomally encoded and unique to A. baumannii

(Vahaboglu et al., 2006; Turton et al., 2006; Peleg, Seifert, & Paterson, 2008). Compared to OXA-

sort carbapenemases, MBLs are less prevalent in A. baumannii.However,t their hydrolytic

potential is more potent towards carbapenem (Peleg, Franklin, Bell, & Spelman, 2006). These

enzymes hydrolyze all β-lactams excluding monobactam and aztreonam (Peleg, Seifert, &

Paterson, 2008). Out of the five MBL groups, only three are identified in A. baumannii to date

including SIM, VIM and IMP (Peleg, Seifert, & Paterson, 2008). A. baumannii strains harboring

both the OXA and the MBL enzymes have been reported from Australia, Greece, Spain, and

Singapore (Canduela et al., 2006; Peleg, Franklin, Walters, Bell, & Spelman, 2006; Peleg, Seifert,

& Paterson, 2008).

The resistance to β-lactam antibiotics including carbapenem is also attributed to the nonenzymatic

resistance mechanism by altering the expression and affinity of penicillin-binding proteins,

changes in the MDR pumps and in the outer membrane proteins (OMPs) (Obara & Nakae, 1991;

Costa et al., 2000; Magnet, Courvalin, & Lambert, 2001; Fernandez-Cuenca et al., 2003; Del Mar

Tomas et al., 2005; Peleg, Seifert, & Paterson, 2008). It is demonstrated that the loss of an outer

19

membrane protein, CarO, is associated with meropenem and imipenem resistance (Limansky,

Mussi, & Viale, 2002; Siroy et al., 2005). Reduced expression of 37, 44, 47-kDa OMPs in A.

baumannii resulted into clinical carbapenem-resistant outbreaks endemic to the New York City

(Quale, Bratu, Landman, & Heddurshetti, 2003). Similarly, poor expression of 33 and 22-kDa

OMPS have been described in association with OXA-24 in Spain (German Bou, Cervero,

Dominguez, Quereda, & Martinez-Beltran, 2000). Additionally, other proteins that are relevant

from a β-lactam resistance point of view, involves the heat-modifiable protein HMP-AB (Gribun,

Nitzan, Pechatnikov, Hershkovits, & Katcoff, 2003), and OmpW (Vila, Marti, & Sanchez-

Céspedes, 2007). The genome of MDR A. baumannii encodes for a wide range of multidrug efflux

systems (P.-E. Fournier et al., 2006), among which, the resistance-nodulation-division (RND)

family-type pump AdeABC is the best described that uses β-lactams including carbapenems, as

well as fluoroquinolones, trimethoprim, aminoglycosides, chloramphenicol, fluoroquinolones,

erythromycin as substrate (Magnet et al., 2001; Peleg, Seifert, & Paterson, 2008). The AdeABC

comprises three structural components: AdeA, AdeB, and AdeC for the inner membrane fusion

protein, the transmembrane structure, and the outer membrane structure, respectively (Marchand,

Damier-Piolle, Courvalin, & Lambert, 2004). Regulation of AdeABC is achieved by a two-

component system that has a sensor kinase (AdeS) and the response regulatory protein (AdeR)

(Marchand, Damier-Piolle, Courvalin, & Lambert, 2004).

Aminoglycoside modifying enzymes are highly expressed in MDR A. baumannii and all the major

types have been identified including nucleotidyltransferases, phosphotransferases, and

acetyltransferases (Nemec, Dolzani, Brisse, van den Broek, & Dijkshoorn, 2004). An emerging

resistance mechanism, 16S rRNA methylation, from the United States, Japan and Korea has been

described that impairs the binding of aminoglycosides to its receptor and thus confers elevated

resistance to most of the clinically useful antibiotics, namely amikacin, tobramycin, and

gentamicin (Doi & Arakawa, 2007). The AdeABC efflux pumps are more hydrophilic and thus

less effective in the transport of amikacin and kanamycin (Magnet, Courvalin, & Lambert, 2001).

AbeM pump from the multidrug and toxic compound extrusion (MATE) family also uses

aminoglycosides like gentamicin and kanamycin as substrates (Su, Chen, Mizushima, Kuroda, &

Tsuchiya, 2005).

20

Changes to topoisomerase IV or DNA gyrase by mutating parC and gyrA genes interfere with the

binding sites of quinolones and have been well reported in A. baumannii (Vila, Ruiz, Goni, Marcos,

& De Anta, 1995). Quinolones also act as substrates for these pumps including MATE pump

AdeM (Su, Chen, Mizushima, Kuroda, & Tsuchiya, 2005) and RND-type pump AdeABC

(Magnet, Courvalin, & Lambert, 2001).

Tetracyclines resistance is because of either ribosomal protection or efflux (Fluit, Florijn, Verhoef,

& Milatovic, 2005). In A. baumannii, tetracycline specific efflux pumps are encoded by tet(A) and

tet(B) determinants (Guardabassi, Dijkshoorn, Collard, Olsen, & Dalsgaard, 2000). Ribosomal

protection in A. baumannii is achieved by tet(O) and tet(M) determinants (Ribera, Ruiz, & Vila,

2003). This group of antibiotics is also susceptible to efflux by AdeABC MDR efflux pump

(Magnet, Courvalin, & Lambert, 2001; Peleg, Adams, & Paterson, 2007).

Resistance to polymyxins in A. baumannii has been demonstrated, however, the mechanism is still

unknown. Possible resistant mechanisms could include the alteration of the LPS binding site, and

changes in OMPs (Peleg, Seifert, & Paterson, 2008).

Trimethoprim-sulfamethoxazole resistance is high in many geographic regions (Gu et al., 2007;

Peleg, Seifert, & Paterson, 2008). The conserved 3΄ region of the MDR phenotype integron

commonly has the sul gene fused to qac, hence conferring resistance to sulfonamides and

antiseptics, respectively (Walsh, Toleman, Poirel, & Nordmann, 2005). Chloramphenicol (cat) and

trimethoprim (dhfr) resistance genes have also been reported within the structure of integron in A.

baumannii (Peleg, Seifert, & Paterson, 2008).

1.3. References

Afzal-Shah, M., Woodford, N., & Livermore, D. M. (2001). Characterization of OXA-25, OXA-

26, and OXA-27, molecular class D β-lactamases associated with carbapenem resistance in

clinical isolates of Acinetobacter baumannii. Antimicrobial Agents and Chemotherapy, 45(2),

583–588.

Al Laham, N., Mediavilla, J. R., Chen, L., Abdelateef, N., Elamreen, F. A., Ginocchio, C. C., …

Kreiswirth, B. N. (2015). MRSA clonal complex 22 strains harboring toxic shock syndrome

toxin (TSST-1) are endemic in the primary hospital in Gaza, Palestine. PloS One, 10(3),

e0120008.

21

Al Naiemi, N., Duim, B., Savelkoul, P. H. M., Spanjaard, L., De Jonge, E., Bart, A., … de Jong,

M. D. (2005). Widespread transfer of resistance genes between bacterial species in an

intensive care unit: implications for hospital epidemiology. Journal of Clinical Microbiology,

43(9), 4862–4864.

Almaghrabi, M. K., Joseph, M. R. P., Assiry, M. M., & Hamid, M. E. (2018). Multidrug-Resistant

Acinetobacter baumannii: An Emerging Health Threat in Aseer Region, Kingdom of Saudi

Arabia. Canadian Journal of Infectious Diseases and Medical Microbiology, 2018.

Almasaudi, S. B. (2018). Acinetobacter spp. as nosocomial pathogens: Epidemiology and

resistance features. Saudi Journal of Biological Sciences, 25(3), 586–596.

Aloush, V., Navon-Venezia, S., Seigman-Igra, Y., Cabili, S., & Carmeli, Y. (2006). Multidrug-

resistant Pseudomonas aeruginosa: risk factors and clinical impact. Antimicrobial Agents and

Chemotherapy, 50(1), 43–48.

Asif, M., Alvi, I. A., & Rehman, S. U. (2018). Insight into Acinetobacter baumannii: pathogenesis,

global resistance, mechanisms of resistance, treatment options, and alternative modalities.

Infection and Drug Resistance, 11, 1249.

Bartoletti, R., Cai, T., Wagenlehner, F. M., Naber, K., & Johansen, T. E. B. (2016). Treatment of

urinary tract infections and antibiotic stewardship. European Urology Supplements, 15(4),

81–87.

Beggs, C. B., Kerr, K. G., Snelling, A. M., & Sleigh, P. A. (2006). Acinetobacter spp. and the

clinical environment. Indoor and Built Environment, 15(1), 19–24.

Behzadi, P., Behzadi, E., Yazdanbod, H., Aghapour, R., Cheshmeh, M. A., & Omran, D. S. (2010).

A survey on urinary tract infections associated with the three most common uropathogenic

bacteria. Maedica, 5(2), 111.

Blot, S., Koulenti, D., Dimopoulos, G., Martin, C., Komnos, A., Krueger, W. A., … Rello, J.

(2014). Prevalence, risk factors, and mortality for ventilator-associated pneumonia in middle-

aged, old, and very old critically ill patients. Critical Care Medicine, 42(3), 601–609.

Bogaerts, P., Naas, T., Wybo, I., Bauraing, C., Soetens, O., Piérard, D., … Glupczynski, Y. (2006).

Outbreak of infection by carbapenem-resistant Acinetobacter baumannii producing the

22

carbapenemase OXA-58 in Belgium. Journal of Clinical Microbiology, 44(11), 4189–4192.

Borer, A., Gilad, J., Smolyakov, R., Eskira, S., Peled, N., Porat, N., … Schlaeffer, F. (2005). Cell

phones and Acinetobacter transmission. Emerging Infectious Diseases, 11(7), 1160.

Borriello, S. P., Davies, H. A., Kamiya, S., Reed, P. J., & Seddon, S. (1990). Virulence factors of

Clostridium difficile. Reviews of Infectious Diseases, 12(Supplement_2), S185-S191.

Bottone, E. J. (1997). Yersinia enterocolitica: the charisma continues. Clinical Microbiology

Reviews, 10(2), 257–276.

Bou, G., Cervero, G., Dominguez, M. A., Quereda, C., & Martinez-Beltran, J. (2000).

Characterization of a nosocomial outbreak caused by a multiresistant Acinetobacter

baumannii strain with a carbapenem-hydrolyzing enzyme: high-level carbapenem resistance

in A. baumannii Is Not Due solely to the presence of β-lactamases. Journal of Clinical

Microbiology, 38(9), 3299–3305.

Bou, G., Oliver, A., & Martinez-Beltran, J. (2000). OXA-24, a novel class D β-lactamase with

carbapenemase activity in an Acinetobacter baumannii clinical strain. Antimicrobial Agents

and Chemotherapy, 44(6), 1556–1561.

Brink, A., Moolman, J., Da Silva, M. C., Botha, M., & others. (2007). Antimicrobial susceptibility

profile of selected bacteraemic pathogens from private institutions in South Africa. South

African Medical Journal, 97(4), 273–279.

Brown, S., & Amyes, S. (2005). OXA β-lactamases in Acinetobacter: the story so far. Journal of

Antimicrobial Chemotherapy, 57(1), 1–3.

Canduela, M. J., Gallego, L., Sevillano, E., Valderrey, C., Calvo, F., & Perez, J. (2006). Evolution

of multidrug-resistant Acinetobacter baumannii isolates obtained from elderly patients with

respiratory tract infections. Journal of Antimicrobial Chemotherapy, 57(6), 1220–1222.

Carbonne, A., Naas, T., Blanckaert, K., Couzigou, C., Cattoen, C., Chagnon, J.-L., … Astagneau,

P. (2005). Investigation of a nosocomial outbreak of extended-spectrum β-lactamase VEB-1-

producing isolates of Acinetobacter baumannii in a hospital setting. Journal of Hospital

Infection, 60(1), 14–18.

23

Cascio, A., Conti, A., Sinardi, L., Iaria, C., Angileri, F. F., Stassi, G., … David, A. (2010). Post-

neurosurgical multidrug-resistant Acinetobacter baumannii meningitis successfully treated

with intrathecal colistin. A new case and a systematic review of the literature. International

Journal of Infectious Diseases, 14(7), e572-e579.

Casellas, J. M. (2011). Antibacterial drug resistance in Latin America: consequences for infectious

disease control. Rev Panam Salud Pública, 30(6), 519–528.

Chen, C.Y., Chen, Y.H., Lu, P.L., Lin, W.R., Chen, T.C., & Lin, C.Y. (2012). Proteus mirabilis

urinary tract infection and bacteremia: risk factors, clinical presentation, and outcomes.

Journal of Microbiology, Immunology and Infection, 45(3), 228–236.

Choi, A. H. K., Slamti, L., Avci, F. Y., Pier, G. B., & Maira-Litran, T. (2009). The pgaABCD

locus of Acinetobacter baumannii encodes the production of poly-β-1-6-N-

acetylglucosamine, which is critical for biofilm formation. Journal of Bacteriology, 191(19),

5953–5963.

Choi, C. H., Lee, E. Y., Lee, Y. C., Park, T. I., Kim, H. J., Hyun, S. H., … Lee, J. C. (2005). Outer

membrane protein 38 of Acinetobacter baumannii localizes to the mitochondria and induces

apoptosis of epithelial cells. Cellular Microbiology, 7(8), 1127–1138.

Choi, C. H., Lee, J. S., Lee, Y. C., Park, T. I., & Lee, J. C. (2008). Acinetobacter baumannii invades

epithelial cells and outer membrane protein A mediates interactions with epithelial cells. BMC

Microbiology, 8(1), 216.

Choi, H. K., Kim, Y. K., Kim, H. Y., Park, J. E., & Uh, Y. (2015). Clinical and microbiological

features of Providencia bacteremia: experience at a tertiary care hospital. The Korean Journal

of Internal Medicine, 30(2), 219.

Clemmer, K. M., Bonomo, R. A., & Rather, P. N. (2011). Genetic analysis of surface motility in

Acinetobacter baumannii. Microbiology, 157(9), 2534–2544.

Coelho, J. M., Turton, J. F., Kaufmann, M. E., Glover, J., Woodford, N., Warner, M., … others.

(2006). Occurrence of carbapenem-resistant Acinetobacter baumannii clones at multiple

hospitals in London and Southeast England. Journal of Clinical Microbiology, 44(10), 3623–

3627.

24

Coelho, J., Woodford, N., Afzal-Shah, M., & Livermore, D. (2006). Occurrence of OXA-58-like

carbapenemases in Acinetobacter spp. collected over 10 years in three continents.

Antimicrobial Agents and Chemotherapy, 50(2), 756–758.

Costa, S. F., Woodcock, J., Gill, M., Wise, R., Barone, A. A., Caiaffa, H., & Levin, A. S. S. (2000).

Outer-membrane proteins pattern and detection of β-lactamases in clinical isolates of

imipenem-resistant Acinetobacter baumannii from Brazil. International Journal of

Antimicrobial Agents, 13(3), 175–182.

Costerton, J. W., Stewart, P. S., & Greenberg, E. P. (1999). Bacterial biofilms: a common cause

of persistent infections. Science, 284(5418), 1318–1322.

Cover, T. L., & Aber, R. C. (1989). Yersinia enterocolitica. New England Journal of Medicine,

321(1), 16–24.

Da Silva, G., Dijkshoorn, L., Van Der Reijden, T., Van Strijen, B., & Duarte, A. (2007).

Identification of widespread, closely related Acinetobacter baumannii isolates in Portugal as

a subgroup of European clone II. Clinical Microbiology and Infection, 13(2), 190–195.

Danes, C., Navia, M. M., Ruiz, J., Marco, F., Jurado, A., de Anta, M. T., & Vila, J. (2002).

Distribution of β-lactamases in Acinetobacter baumannii clinical isolates and the effect of

Syn 2190 (AmpC inhibitor) on the MICs of different β-lactam antibiotics. Journal of


Del Mar Tomas, M., Beceiro, A., Perez, A., Velasco, D., Moure, R., Villanueva, R., … Bou, G.

(2005). Cloning and functional analysis of the gene encoding the 33-to 36-kilodalton outer

membrane protein associated with carbapenem resistance in Acinetobacter baumannii.


Dijkshoorn, L., Nemec, A., & Seifert, H. (2007). An increasing threat in hospitals: multidrug-

resistant Acinetobacter baumannii. Nature Reviews Microbiology, 5(12), 939.

Dobrewski, R., Savov, E., Bernards, A. T., Van Den Barselaar, M., Nordmann, P., Van Den Broek,

P. J., & Dijkshoorn, L. (2006). Genotypic diversity and antibiotic susceptibility of

Acinetobacter baumannii isolates in a Bulgarian hospital. Clinical Microbiology and

Infection, 12(11), 1135–1137.

25

Doi, Y., & Arakawa, Y. (2007). 16S ribosomal RNA methylation: emerging resistance mechanism

against aminoglycosides. Clinical Infectious Diseases, 45(1), 88–94.

Doughari, H. J., Ndakidemi, P. A., Human, I. S., & Benade, S. (2011). The ecology, biology and

pathogenesis of Acinetobacter spp.: an overview. Microbes and Environments, 26(2), 101–

112.

Ducel, G., Fabry, J., Nicolle, L., Organization, W. H., & others. (2002). Prevention of hospital-

acquired infections: a practical guide.

Durand, M. L., Calderwood, S. B., Weber, D. J., Miller, S. I., Southwick, F. S., Caviness Jr, V. S.,

& Swartz, M. N. (1993). Acute bacterial meningitis in adults-A review of 493 episodes. New

England Journal of Medicine, 328(1), 21–28.

Eijkelkamp, B. A., Hassan, K. A., Paulsen, I. T., & Brown, M. H. (2011). Investigation of the

human pathogen Acinetobacter baumannii under iron limiting conditions. BMC Genomics,

12(1), 126.

Elliott, C., & Justiz-Vaillant, A. (2018). Nosocomial Infections: A 360-degree Review.

International Biological and Biomedical Journal, 4(2), 0.

Endimiani, A., Luzzaro, F., Migliavacca, R., Mantengoli, E., Hujer, A. M., Hujer, K. M., …

Toniolo, A. (2007). Spread in an Italian hospital of a clonal Acinetobacter baumannii strain

producing the TEM-92 extended-spectrum β-lactamase. Antimicrobial Agents and

Chemotherapy, 51(6), 2211–2214.

Eugenin, E. A. (2013). Community-acquired pneumonia infections by Acinetobacter baumannii:

how does alcohol impact the antimicrobial functions of macrophages? Virulence, 4(6), 435–

436.

Fabrega, A., & Vila, J. (2012). Yersinia enterocolitica: pathogenesis, virulence and antimicrobial

resistance. Enfermedades Infecciosas y Microbiologia Clinica, 30(1), 24–32.

Fernandez-Cuenca, F., Martinez-Martinez, L., Conejo, M., Ayala, J. A., Perea, E. J., & Pascual,

A. (2003). Relationship between β-lactamase production, outer membrane protein and

penicillin-binding protein profiles on the activity of carbapenems against clinical isolates of

Acinetobacter baumannii. Journal of Antimicrobial Chemotherapy, 51(3), 565–574.

26

Flores-Mireles, A. L., Walker, J. N., Caparon, M., & Hultgren, S. J. (2015). Urinary tract

infections: epidemiology, mechanisms of infection and treatment options. Nature Reviews


Fluit, A. C., Florijn, A., Verhoef, J., & Milatovic, D. (2005). Presence of tetracycline resistance

determinants and susceptibility to tigecycline and minocycline. Antimicrobial Agents and

Chemotherapy, 49(4), 1636–1638.

Foris, L., & Snowden, J. (2017). Proteus Mirabilis Infections.

Fournier, P.E., Vallenet, D., Barbe, V., Audic, S., Ogata, H., Poirel, L., … others. (2006).

Comparative genomics of multidrug resistance in Acinetobacter baumannii. PLoS Genetics,

2(1), e7.

Fournier, P. E., Richet, H., & Weinstein, R. A. (2006). The epidemiology and control of

Acinetobacter baumannii in health care facilities. Clinical Infectious Diseases, 42(5), 692–

699.

Frank, D. N., Wilson, S. S., Amand, A. L. S., & Pace, N. R. (2009). Culture-independent

microbiological analysis of foley urinary catheter biofilms. PloS One, 4(11), e7811.

Gaddy, J. A., Arivett, B. A., McConnell, M. J., Lopez-Rojas, R., Pachon, J., & Actis, L. A. (2012).

Role of acinetobactin-mediated iron acquisition functions in the interaction of Acinetobacter

baumannii ATCC 19606T with human lung epithelial cells, Galleria mellonella caterpillars

and mice. Infection and Immunity, IAI-06279.

Gaddy, J. A., Tomaras, A. P., & Actis, L. A. (2009). The Acinetobacter baumannii 19606 OmpA

protein plays a role in biofilm formation on abiotic surfaces and in the interaction of this

pathogen with eukaryotic cells. Infection and Immunity, 77(8), 3150–3160.

Gales, A. C., Jones, R. N., & Sader, H. S. (2006). Global assessment of the antimicrobial activity

of polymyxin B against 54 731 clinical isolates of Gram-negative bacilli: report from the

SENTRY antimicrobial surveillance programme (2001-2004). Clinical Microbiology and

Infection, 12(4), 315–321.

Gellatly, S. L., & Hancock, R. E. W. (2013). Pseudomonas aeruginosa: new insights into

pathogenesis and host defenses. Pathogens and Disease, 67(3), 159–173.

27

Giuffre, M., Geraci, D. M., Bonura, C., Saporito, L., Graziano, G., Insinga, V., … Mammina, C.

(2016). The increasing challenge of multidrug-resistant Gram-negative bacilli: results of a 5-

year active surveillance program in a neonatal intensive care unit. Medicine, 95(10).

Go, E. S., Urban, C., Burns, J., Mariano, N., Mosinka-Snipas, K., Rahal, J. J., … Eisner, W. (1994).

Clinical and molecular epidemiology of Acinetobacter infections sensitive only to polymyxin

B and sulbactam. The Lancet, 344(8933), 1329–1332.

González, R. H., Dijkshoorn, L., den Barselaar, M., & Nudel, C. (2009). Quorum sensing signal

profile of Acinetobacter strains from nosocomial and environmental sources. Rev Argent

Microbiol, 41(2), 73–78.

Gribun, A., Nitzan, Y., Pechatnikov, I., Hershkovits, G., & Katcoff, D. J. (2003). Molecular and

structural characterization of the HMP-AB gene encoding a pore-forming protein from a

clinical isolate of Acinetobacter baumannii. Current Microbiology, 47(5), 434–443.

Gu, B., Tong, M., Zhao, W., Liu, G., Ning, M., Pan, S., & Zhao, W. (2007). Prevalence and

characterization of class I integrons among Pseudomonas aeruginosa and Acinetobacter

baumannii isolates from patients in Nanjing, China. Journal of Clinical Microbiology, 45(1),

241–243.

Guardabassi, L., Dijkshoorn, L., Collard, J.M., Olsen, J. E., & Dalsgaard, A. (2000). Distribution

and in-vitro transfer of tetracycline resistance determinants in clinical and aquatic

Acinetobacter strains. Journal of Medical Microbiology, 49(10), 929–936.

Guiton, P. S., Hung, C. S., Hancock, L. E., Caparon, M. G., & Hultgren, S. J. (2010). Enterococcal

biofilm formation and virulence in an optimized murine model of foreign body-associated

urinary tract infections. Infection and Immunity, 78(10), 4166–4175.

Haidar, G., Philips, N. J., Shields, R. K., Snyder, D., Cheng, S., Potoski, B. A., … others. (2017).

Ceftolozane-tazobactam for the treatment of multidrug-resistant Pseudomonas aeruginosa

infections: clinical effectiveness and evolution of resistance. Clinical Infectious Diseases,

65(1), 110–120.

Hassan, K. A., Aftab, A., & Riffat, M. (2015). Nosocomial infections and their control strategies.

Asian Pacific Journal of Tropical Biomedicine, (7), 505–509.

28

Huang, Z. M., Mao, P. H., Chen, Y., Wu, L., & Wu, J. (2004). Study on the molecular

epidemiology of SHV type beta-lactamase-encoding genes of multiple-drug-resistant

Acinetobacter baumannii. Zhonghua Liu Xing Bing Xue Za Zhi= Zhonghua Liuxingbingxue

Zazhi, 25(5), 425–427.

Iredell, J., Thomas, L., Power, D., & Mendes, E. (2007). Tigecycline resistance in Australian

antibiotic-resistant Gram-negative bacteria. Journal of Antimicrobial Chemotherapy, 59(4),

816–818.

Jacobs, A. C., Hood, I., Boyd, K. L., Olson, P. D., Morrison, J. M., Carson, S., … Dunman, P. M.

(2010). Inactivation of phospholipase D diminishes Acinetobacter baumannii pathogenesis.

Infection and Immunity, 78(5), 1952–1962.

Jawad, A., Seifert, H., Snelling, A. M., Heritage, J., & Hawkey, P. M. (1998). Survival of

Acinetobacter baumannii on dry surfaces: comparison of outbreak and sporadic isolates.

Journal of Clinical Microbiology, 36(7), 1938–1941.

Jhang, J.F., & Kuo, H.C. (2017). Recent advances in recurrent urinary tract infection from

pathogenesis and biomarkers to prevention. Tzu-Chi Medical Journal, 29(3), 131.

Jimenez-Alcaide, E., Medina-Polo, J., Garcia-Gonzalez, L., Arrebola-Pajares, A., Guerrero-

Ramos, F., Perez-Cadavid, S., … others. (2015). Healthcare-associated urinary tract

infections in patients with a urinary catheter: Risk factors, microbiological characteristics and

patterns of antibiotic resistance. Archivos Espanoles de Urologia, 68(6), 541–550.

Jung, J. Y., Park, M. S., Kim, S. E., Park, B. H., Son, J. Y., Kim, E. Y., … others. (2010). Risk

factors for multi-drug resistant Acinetobacter baumannii bacteremia in patients with

colonization in the intensive care unit. BMC Infectious Diseases, 10(1), 228.

Kaiser, L., Emonet, S., Huttner, A., Vetter, P., Toutous, L. T., Uriel, P. V. B., & Lecompte, T.

(2015). Infections disease update: 2014 is viral. Revue Medicale Suisse, 11(456–457), 72–77.

Kanafani, A. Z., & Kanj, S. S. (2014). Ministry of Health, Kingdome of Saudi Arabia.

Kaper, J. B., Nataro, J. P., & Mobley, H. L. T. (2004). Pathogenic Escherichia coli. Nature Reviews


29

Karki, S., Leder, K., & Cheng, A. C. (2015). Should we continue to isolate patients with

vancomycin-resistant enterococci in hospitals. Med J Aust, 202(5), 234–236.

Keen, E. F., Robinson, B. J., Hospenthal, D. R., Aldous, W. K., Wolf, S. E., Chung, K. K., &

Murray, C. K. (2010). Prevalence of multidrug-resistant organisms recovered at a military

burn center. Burns, 36(6), 819–825.

Khan, H. A., Baig, F. K., & Mehboob, R. (2017). Nosocomial infections: Epidemiology,

prevention, control and surveillance. Asian Pacific Journal of Tropical Biomedicine, 7(5),

478–482.

Ki, V., & Rotstein, C. (2008). Bacterial skin and soft tissue infections in adults: a review of their

epidemiology, pathogenesis, diagnosis, treatment and site of care. Canadian Journal of

Infectious Diseases and Medical Microbiology, 19(2), 173–184.

Kim, J., Kang, J. O., Kim, H., Seo, M.R., Choi, T. Y., Pai, H., … Fawley, W. (2013). Epidemiology

of Clostridium difficile infections in a tertiary-care hospital in Korea. Clinical Microbiology

and Infection, 19(6), 521–527.

Kim, S. W., Choi, C. H., Moon, D. C., Jin, J. S., Lee, J. H., Shin, J.-H., … others. (2009). Serum

resistance of Acinetobacter baumannii through the binding of factor H to outer membrane

proteins. FEMS Microbiology Letters, 301(2), 224–231.

Klevens, R. M., Edwards, J. R., Richards Jr, C. L., Horan, T. C., Gaynes, R. P., Pollock, D. A., &

Cardo, D. M. (2007). Estimating health care-associated infections and deaths in US hospitals,

2002. Public Health Reports, 122(2), 160–166.

Ko, K. S., Suh, J. Y., Kwon, K. T., Jung, S.I., Park, K.H., Kang, C. I., … Song, J.-H. (2007). High

rates of resistance to colistin and polymyxin B in subgroups of Acinetobacter baumannii

isolates from Korea. Journal of Antimicrobial Chemotherapy, 60(5), 1163–1167.

Krewulak, K. D., & Vogel, H. J. (2008). Structural biology of bacterial iron uptake. Biochimica et

Biophysica Acta (BBA)-Biomembranes, 1778(9), 1781–1804.

Kumar, S., Vengadassalapathy, L., & Menon, T. (2008). Prosthetic valve endocarditis caused by

Acinetobacter baumannii complex. Indian Journal of Pathology and Microbiology, 51(4),

573.

30

Kutlesa, M., Santini, M., Krajinovic, V., Papic, N., Novokmet, A., Josipovic, R. M., & Barsic, B.

(2017). Nosocomial blood stream infections in patients treated with venovenous

extracorporeal membrane oxygenation for acute respiratory distress syndrome. Minerva

Anestesiologica, 83(5), 493–501.

Lane, D. R., & Takhar, S. S. (2011). Diagnosis and management of urinary tract infection and

pyelonephritis. Emergency Medicine Clinics, 29(3), 539–552.

Lausch, K. R., Fuursted, K., Larsen, C. S., & Storgaard, M. (2013). Colonisation with multi-

resistant Enterobacteriaceae in hospitalised Danish patients with a history of recent travel: a

cross-sectional study. Travel Medicine and Infectious Disease, 11(5), 320–323.

Lee, J. C., Koerten, H., Den Broek, P., Beekhuizen, H., Wolterbeek, R., den Barselaar, M., …

Dijkshoorn, L. (2006). Adherence of Acinetobacter baumannii strains to human bronchial

epithelial cells. Research in Microbiology, 157(4), 360–366.

Lee, K., Lim, C. H., Cho, J. H., Lee, W. G., Uh, Y., Kim, H. J., … others. (2006). High prevalence

of ceftazidime-resistant Klebsiella pneumoniae and increase of imipenem-resistant

Pseudomonas aeruginosa and Acinetobacter spp. in Korea: a KONSAR program in 2004.

Yonsei Medical Journal, 47(5), 634–645.

Li, B., Zhao, Y., Liu, C., Chen, Z., & Zhou, D. (2014). Molecular pathogenesis of Klebsiella

pneumoniae. Future Microbiology, 9(9), 1071–1081.

Limansky, A. S., Mussi, M. A., & Viale, A. M. (2002). Loss of a 29-kilodalton outer membrane

protein in Acinetobacter baumannii is associated with imipenem resistance. Journal of

Clinical Microbiology, 40(12), 4776–4778.

Lin, C.H., Huang, H.T., Chien, C.C., Tzeng, D.S., & Lung, F.W. (2008). Purple urine bag

syndrome in nursing homes: ten elderly case reports and a literature review. Clinical

Interventions in Aging, 3(4), 729.

Lin, Y.T., Wang, Y.P., Wang, F.D., & Fung, C.P. (2015). Community-onset Klebsiella

pneumoniae in Taiwan: clinical features of the disease and associated microbiological

characteristics of isolates from pneumonia and nasopharynx. Frontiers in Microbiology, 6,

122.

31

Loehfelm, T. W., Luke, N. R., & Campagnari, A. A. (2008). Identification and characterization of

an Acinetobacter baumannii biofilm-associated protein. Journal of Bacteriology, 190(3),

1036–1044.

Lopez-Otsoa, F., Gallego, L., Towner, K. J., Tysall, L., Woodford, N., & Livermore, D. M. (2002).

Endemic carbapenem resistance associated with OXA-40 carbapenemase among

Acinetobacter baumannii isolates from a hospital in northern Spain. Journal of Clinical

Microbiology, 40(12), 4741–4743.

Mace, S. E. (2008). Acute bacterial meningitis. Emergency Medicine Clinics of North America,

26(2), 281–317.

Maegele, M., Gregor, S., Steinhausen, E., Bouillon, B., Heiss, M. M., Perbix, W., … others.

(2005). The long-distance tertiary air transfer and care of tsunami victims: injury pattern and

microbiological and psychological aspects. Critical Care Medicine, 33(5), 1136–1140.

Magnet, S., Courvalin, P., & Lambert, T. (2001). Resistance-nodulation-cell division-type efflux

pump involved in aminoglycoside resistance in Acinetobacter baumannii strain BM4454.


Malani, P. N. (2012). Harrison’s principles of internal medicine. JAMA, 308(17), 1813–1814.

Mandell, G. L., Bennett, J. E., & Dolin, R. (2004). Mandell’s Principles and Practices of Infection

Diseases . Churchill Livingstone. ISBN 0-443-06643-4.

Marchand, I., Damier-Piolle, L., Courvalin, P., & Lambert, T. (2004). Expression of the RND-

type efflux pump AdeABC in Acinetobacter baumannii is regulated by the AdeRS two-

component system. Antimicrobial Agents and Chemotherapy, 48(9), 3298–3304.

Martin, G. S., Mannino, D. M., Eaton, S., & Moss, M. (2003). The epidemiology of sepsis in the

United States from 1979 through 2000. New England Journal of Medicine, 348(16), 1546–

1554.

McConnell, M. J., Actis, L., & Pachon, J. (2013). Acinetobacter baumannii: human infections,

factors contributing to pathogenesis and animal models. FEMS Microbiology Reviews, 37(2),

130–155.

32

McQueary, C. N., & Actis, L. A. (2011). Acinetobacter baumannii biofilms: variations among

strains and correlations with other cell properties. The Journal of Microbiology, 49(2), 243–

250.

McQueary, C. N., Kirkup, B. C., Si, Y., Barlow, M., Actis, L. A., Craft, D. W., & Zurawski, D. V.

(2012). Extracellular stress and lipopolysaccharide modulate Acinetobacter baumannii

surface-associated motility. Journal of Microbiology, 50(3), 434–443.

Mihara, K., Tanabe, T., Yamakawa, Y., Funahashi, T., Nakao, H., Narimatsu, S., & Yamamoto,

S. (2004). Identification and transcriptional organization of a gene cluster involved in

biosynthesis and transport of acinetobactin, a siderophore produced by Acinetobacter

baumannii ATCC 19606T. Microbiology, 150(8), 2587–2597.

Morris, A., & Low, D. E. (1999). Nosocomial bacterial meningitis, including central nervous

system shunt infections. Infectious Disease Clinics of North America, 13(3), 735–750.

Munoz-Price, L. S., & Weinstein, R. A. (2008). Acinetobacter infection. New England Journal of

Medicine, 358(12), 1271–1281.

Naas, T., Bogaerts, P., Bauraing, C., Degheldre, Y., Glupczynski, Y., & Nordmann, P. (2006).

Emergence of PER and VEB extended-spectrum β-lactamases in Acinetobacter baumannii in

Belgium. Journal of Antimicrobial Chemotherapy, 58(1), 178–182.

Naas, T., Coignard, B., Carbonne, A., Blanckaert, K., Bajolet, O., Bernet, C., … others. (2006).

VEB-1 extended-spectrum β-lactamase--producing Acinetobacter baumannii, France.

Emerging Infectious Diseases, 12(8), 1214.

Naas, T., Levy, M., Hirschauer, C., Marchandin, H., & Nordmann, P. (2005). Outbreak of

carbapenem-resistant Acinetobacter baumannii producing the carbapenemase OXA-23 in a

tertiary care hospital of Papeete, French Polynesia. Journal of Clinical Microbiology, 43(9),

4826–4829.

Nau, R., Sorgel, F., & Eiffert, H. (2010). Penetration of drugs through the blood-cerebrospinal

fluid/blood-brain barrier for treatment of central nervous system infections. Clinical

Microbiology Reviews, 23(4), 858–883.

Navon-Venezia, S., Leavitt, A., & Carmeli, Y. (2007). High tigecycline resistance in multidrug-

33

resistant Acinetobacter baumannii. Journal of Antimicrobial Chemotherapy, 59(4), 772–774.

Nejad, S. B., Allegranzi, B., Syed, S. B., Ellis, B., & Pittet, D. (2011). Health-care-associated

infection in Africa: a systematic review. Bulletin of the World Health Organization, 89, 757–

765.

Nemec, A., Dolzani, L., Brisse, S., van den Broek, P., & Dijkshoorn, L. (2004). Diversity of

aminoglycoside-resistance genes and their association with class 1 integrons among strains

of pan-European Acinetobacter baumannii clones. Journal of Medical Microbiology, 53(12),

1233–1240.

Nicolle, L. E. (2008). Uncomplicated urinary tract infection in adults including uncomplicated

pyelonephritis. Urologic Clinics of North America, 35(1), 1–12.

Niu, C., Clemmer, K. M., Bonomo, R. A., & Rather, P. N. (2008). Isolation and characterization

of an autoinducer synthase from Acinetobacter baumannii. Journal of Bacteriology, 190(9),

3386–3392.

Nordmann, P., Cuzon, G., & Naas, T. (2009). The real threat of Klebsiella pneumoniae

carbapenemase-producing bacteria. The Lancet Infectious Diseases, 9(4), 228–236.

Obara, M., & Nakae, T. (1991). Mechanisms of resistance to $β$-lactam antibiotics in

Acinetobacter calcoaceticus. Journal of Antimicrobial Chemotherapy, 28(6), 791–800.

WHO. (2017). Global priority list of antibiotic-resistant bacteria to guide research, discovery, and

development of new antibiotics. Geneva: World Health Organization.

Pasteran, F., Rapoport, M., Petroni, A., Faccone, D., Corso, A., Galas, M., … Cagnoni, V. (2006).

Emergence of PER-2 and VEB-1a in Acinetobacter baumannii strains in the Americas.


Paton, R., Miles, R. S., Hood, J., & Amyes, S. G. B. (1993). ARI 1:β-lactamase-mediated

imipenem resistance in Acinetobacter baumannii. International Journal of Antimicrobial

Agents, 2(2), 81–87.

Peleg, A. Y., Adams, J., & Paterson, D. L. (2007). Tigecycline efflux as a mechanism for

nonsusceptibility in Acinetobacter baumannii. Antimicrobial Agents and Chemotherapy,

34

51(6), 2065–2069.

Peleg, A. Y., Franklin, C., Bell, J. M., & Spelman, D. W. (2006). Emergence of carbapenem

resistance in Acinetobacter baumannii recovered from blood cultures in Australia. Infection

Control & Hospital Epidemiology, 27(7), 759–761.

Peleg, A. Y., Franklin, C., Walters, L. J., Bell, J. M., & Spelman, D. W. (2006). OXA-58 and IMP-

4 carbapenem-hydrolyzing β-lactamases in an Acinetobacter junii blood culture isolate from

Australia. Antimicrobial Agents and Chemotherapy, 50(1), 399–400.

Peleg, A. Y., Seifert, H., & Paterson, D. L. (2008). Acinetobacter baumannii: emergence of a

successful pathogen. Clinical Microbiology Reviews, 21(3), 538–582.

Pendleton, J. N., Gorman, S. P., & Gilmore, B. F. (2013). Clinical relevance of the ESKAPE

pathogens. Expert Review of Anti-Infective Therapy, 11(3), 297–308.

Poirel, L., Menuteau, O., Agoli, N., Cattoen, C., & Nordmann, P. (2003). Outbreak of extended-

spectrum β-lactamase VEB-1-producing isolates of Acinetobacter baumannii in a French

hospital. Journal of Clinical Microbiology, 41(8), 3542–3547.

Poirel, L., & Nordmann, P. (2006). Carbapenem resistance in Acinetobacter baumannii:

mechanisms and epidemiology. Clinical Microbiology and Infection, 12(9), 826–836.

Quale, J., Bratu, S., Landman, D., & Heddurshetti, R. (2003). Molecular epidemiology and

mechanisms of carbapenem resistance in Acinetobacter baumannii endemic in New York

City. Clinical Infectious Diseases, 37(2), 214–220.

Ramakrishnan, K., Salinas, R. C., & AGUDELO HIGUITA, N. I. (2015). Skin and Soft Tissue

Infections. American Family Physician, 92(6).

Rhomberg, P. R., & Jones, R. N. (2007). Contemporary activity of meropenem and comparator

broad-spectrum agents: MYSTIC program report from the United States component (2005).

Diagnostic Microbiology and Infectious Disease, 57(2), 207–215.

Ribera, A., Ruiz, J., & Vila, J. (2003). Presence of the Tet M determinant in a clinical isolate of

Acinetobacter baumannii. Antimicrobial Agents and Chemotherapy, 47(7), 2310–2312.

Riley, T. V, Webb, S. A. R., Cadwallader, H., Briggs, B. D., Christiansen, L., & Bowman, R. A.

35

(1996). Outbreak of gentamicin-resistant Acinetobacter baumanii in an intensive care unit:

clinical, epidemiological and microbiological features. Pathology, 28(4), 359–363.

Rodriguez Guardado, A., Blanco, A., Asensi, V., Perez, F., Rial, J. C., Pintado, V., … others.

(2008). Multidrug-resistant Acinetobacter meningitis in neurosurgical patients with

intraventricular catheters: assessment of different treatments. Journal of Antimicrobial


Salvatore, S., Salvatore, S., Cattoni, E., Siesto, G., Serati, M., Sorice, P., & Torella, M. (2011).

Urinary tract infections in women. European Journal of Obstetrics & Gynecology and

Reproductive Biology, 156(2), 131–136.

Schafer, J. J., & Mangino, J. E. (2008). Multidrug-resistant Acinetobacter baumannii osteomyelitis

from Iraq. Emerging Infectious Diseases, 14(3), 512.

Schaffer, J. N., & Pearson, M. M. (2015). Proteus mirabilis and urinary tract infections.

Microbiology Spectrum, 3(5).

Schulte, B., Goerke, C., Weyrich, P., Gröbner, S., Bahrs, C., Wolz, C., … Borgmann, S. (2005).

Clonal spread of meropenem-resistant Acinetobacter baumannii strains in hospitals in the

Mediterranean region and transmission to Southwest Germany. Journal of Hospital Infection,

61(4), 356–357.

Sebeny, P. J., Riddle, M. S., & Petersen, K. (2008). Acinetobacter baumannii skin and soft-tissue

infection associated with war trauma. Clinical Infectious Diseases, 47(4), 444–449.

Seifert, H., Dijkshoorn, L., Gerner-Smidt, P., Pelzer, N., Tjernberg, I., & Vaneechoutte, M. (1997).

Distribution of Acinetobacter species on human skin: comparison of phenotypic and

genotypic identification methods. Journal of Clinical Microbiology, 35(11), 2819–2825.

Singh, P. K. (2016). Antibiotics, handle with care. Geneva: WHO.

Siroy, A., Molle, V., Lemaitre-Guillier, C., Vallenet, D., Pestel-Caron, M., Cozzone, A. J., … De,

E. (2005). Channel formation by CarO, the carbapenem resistance-associated outer

membrane protein of Acinetobacter baumannii. Antimicrobial Agents and Chemotherapy,

49(12), 4876–4883.

36

Skaar, E. P. (2010). The battle for iron between bacterial pathogens and their vertebrate hosts.

PLoS Pathogens, 6(8), e1000949.

Smith, M. G., Gianoulis, T. A., Pukatzki, S., Mekalanos, J. J., Ornston, L. N., Gerstein, M., &

Snyder, M. (2007). New insights into Acinetobacter baumannii pathogenesis revealed by

high-density pyrosequencing and transposon mutagenesis. Genes & Development, 21(5),

601–614.

Society, A. T., of America, I. D. S., & others. (2005). Guidelines for the management of adults

with hospital-acquired, ventilator-associated, and healthcare-associated pneumonia.

American Journal of Respiratory and Critical Care Medicine, 171(4), 388.

Sood, S., Malhotra, M., Das, B. K., Kapil, A., & others. (2008). Enterococcal infections &

antimicrobial resistance. Indian Journal of Medical Research, 128(2), 111.

Stevens, D. L., Bisno, A. L., Chambers, H. F., Dellinger, E. P., Goldstein, E. J. C., Gorbach, S. L.,

… Wade, J. C. (2014). Practice guidelines for the diagnosis and management of skin and soft

tissue infections: 2014 update by the Infectious Diseases Society of America. Clinical

Infectious Diseases, 59(2), e10-e52.

Su, X.Z., Chen, J., Mizushima, T., Kuroda, T., & Tsuchiya, T. (2005). AbeM, an H+-coupled

Acinetobacter baumannii multidrug efflux pump belonging to the MATE family of

transporters. Antimicrobial Agents and Chemotherapy, 49(10), 4362–4364.

Sydnor, E. R. M., & Perl, T. M. (2011). Hospital epidemiology and infection control in acute-care

settings. Clinical Microbiology Reviews, 24(1), 141–173.

Tacconelli, E., Carrara, E., Savoldi, A., Harbarth, S., Mendelson, M., Monnet, D. L., … others.

(2018). Discovery, research, and development of new antibiotics: the WHO priority list of

antibiotic-resistant bacteria and tuberculosis. The Lancet Infectious Diseases, 18(3), 318–327.

Terpenning, M. (2005). Geriatric oral health and pneumonia risk. Clinical Infectious Diseases,

40(12), 1807–1810.

Thomson, J. M., & Bonomo, R. A. (2005). The threat of antibiotic resistance in Gram-negative

pathogenic bacteria: β-lactams in peril! Current Opinion in Microbiology, 8(5), 518–524.

37

Tognim, M. C. B., Gales, A. C., Penteado, A. P., Silbert, S., & Sader, H. S. (2006). Dissemination

of IMP-1 metallo-β-lactamase producing Acinetobacter species in a Brazilian teaching

hospital. Infection Control & Hospital Epidemiology, 27(7), 742–747.

Tomaras, A. P., Dorsey, C. W., Edelmann, R. E., & Actis, L. A. (2003). Attachment to and biofilm

formation on abiotic surfaces by Acinetobacter baumannii: involvement of a novel

chaperone-usher pili assembly system. Microbiology, 149(12), 3473–3484.

Tomaras, A. P., Dorsey, C. W., McQueary, C., & Actis, L. A. (2008). Molecular basis of

Acinetobacter virulence and pathogenicity. Acinetobacter Molecular Microbiology.

Tomaras, A. P., Flagler, M. J., Dorsey, C. W., Gaddy, J. A., & Actis, L. A. (2008). Characterization

of a two-component regulatory system from Acinetobacter baumannii that controls biofilm

formation and cellular morphology. Microbiology, 154(11), 3398–3409.

Tong, S. Y. C., Davis, J. S., Eichenberger, E., Holland, T. L., & Fowler, V. G. (2015).

Staphylococcus aureus infections: epidemiology, pathophysiology, clinical manifestations,

and management. Clinical Microbiology Reviews, 28(3), 603–661.

Towner, K. (2006). The genus acinetobacter. In The prokaryotes (pp. 746–758). Springer.

Turton, J. F., Woodford, N., Glover, J., Yarde, S., Kaufmann, M. E., & Pitt, T. L. (2006).

Identification of Acinetobacter baumannii by detection of the blaOXA-51-like

carbapenemase gene intrinsic to this species. Journal of Clinical Microbiology, 44(8), 2974–

2976.

Ulrich, R. J., Santhosh, K., Mogle, J. A., Young, V. B., & Rao, K. (2017). Is Clostridium difficile

infection a risk factor for subsequent bloodstream infection? Anaerobe, 48, 27–33.

Unal, S., & Garcia-Rodriguez, J. A. (2005). Activity of meropenem and comparators against

Pseudomonas aeruginosa and Acinetobacter spp. isolated in the MYSTIC Program, 2002--

2004. Diagnostic Microbiology and Infectious Disease, 53(4), 265–271.

Vahaboglu, H., Budak, F., Kasap, M., Gacar, G., Torol, S., Karadenizli, A., … Eroglu, C. (2006).

High prevalence of OXA-51-type class D β-lactamases among ceftazidime-resistant clinical

isolates of Acinetobacter spp.: co-existence with OXA-58 in multiple centres. Journal of


38

Vahaboglu, H., Ozturk, R., Aygun, G., Cocskunkan, F., Yaman, A., Kaygusuz, A., … Otkun, M.

(1997). Widespread detection of PER-1-type extended-spectrum beta-lactamases among

nosocomial Acinetobacter and Pseudomonas aeruginosa isolates in Turkey: a nationwide

multicenter study. Antimicrobial Agents and Chemotherapy, 41(10), 2265–2269.

Van Dessel, H., Dijkshoorn, L., van der Reijden, T., Bakker, N., Paauw, A., van den Broek, P., …

Brisse, S. (2004). Identification of a new geographically widespread multiresistant

Acinetobacter baumannii clone from European hospitals. Research in Microbiology, 155(2),

105–112.

Van Looveren, M., Goossens, H., Group, A. S., & others. (2004). Antimicrobial resistance of

Acinetobacter spp. in Europe. Clinical Microbiology and Infection, 10(8), 684–704.

Vandenesch, F., Lina, G., & Henry, T. (2012). Staphylococcus aureus hemolysins, bi-component

leukocidins, and cytolytic peptides: a redundant arsenal of membrane-damaging virulence

factors? Frontiers in Cellular and Infection Microbiology, 2, 12.

Vila, J., Marti, S., & Sanchez-Cespedes, J. (2007). Porins, efflux pumps and multidrug resistance

in Acinetobacter baumannii. Journal of Antimicrobial Chemotherapy, 59(6), 1210–1215.

Vila, J., Ruiz, J., Goni, P., Marcos, A., & De Anta, T. J. (1995). Mutation in the gyrA gene of

quinolone-resistant clinical isolates of Acinetobacter baumannii. Antimicrobial Agents and

Chemotherapy, 39(5), 1201–1203.

Vila, J., Ruiz, J., Navia, M., Becerril, B., Garcia, I., Perea, S., … others. (1999). Spread of amikacin

resistance in Acinetobacter baumannii strains isolated in Spain due to an epidemic strain.

Journal of Clinical Microbiology, 37(3), 758–761.

Villegas, M. V., Kattan, J. N., Correa, A., Lolans, K., Guzman, A. M., Woodford, N., … others.

(2007). Dissemination of Acinetobacter baumannii clones with OXA-23 carbapenemase in

Colombian hospitals. Antimicrobial Agents and Chemotherapy, 51(6), 2001–2004.

Vinh, D. C., & Embil, J. M. (2005). Rapidly progressive soft tissue infections. The Lancet

Infectious Diseases, 5(8), 501–513.

Walsh, T. R., Toleman, M. A., Poirel, L., & Nordmann, P. (2005). Metallo-β-lactamases: the quiet

before the storm? Clinical Microbiology Reviews, 18(2), 306–325.

39

Wandersman, C., & Delepelaire, P. (2004). Bacterial iron sources: from siderophores to

hemophores. Annu. Rev. Microbiol., 58, 611–647.

Warren, J. W. (1986). Providencia stuartii: a common cause of antibiotic-resistant bacteriuria in

patients with long-term indwelling catheters. Reviews of Infectious Diseases, 8(1), 61–67.

Weinstein, R. A., Gaynes, R., Edwards, J. R., & System, N. N. I. S. (2005). Overview of

nosocomial infections caused by Gram-negative bacilli. Clinical Infectious Diseases, 41(6),

848–854.

Whitson, W. J., Ball, P. A., Lollis, S. S., Balkman, J. D., & Bauer, D. F. (2014). Postoperative

Mycoplasma hominis infections after neurosurgical intervention: A review. Journal of

Neurosurgery: Pediatrics, 14(2), 212–218.

WHO. (2011). The burden of health care-associated infection worldwide.

Wie, S.H. (2015). Clinical significance of Providencia bacteremia or bacteriuria. The Korean

Journal of Internal Medicine, 30(2), 167.

Wisplinghoff, H., Bischoff, T., Tallent, S. M., Seifert, H., Wenzel, R. P., & Edmond, M. B. (2004).

Nosocomial bloodstream infections in US hospitals: analysis of 24,179 cases from a

prospective nationwide surveillance study. Clinical Infectious Diseases, 39(3), 309–317.

Xu, Y., Gu, B., Huang, M., Liu, H., Xu, T., Xia, W., & Wang, T. (2015). Epidemiology of

carbapenem resistant Enterobacteriaceae (CRE) during 2000-2012 in Asia. Journal of

Thoracic Disease, 7(3), 376.

Ying, C. M., Ling, T. K. W., Lee, C. C., & Ling, J. M. (2006). Characterization of carbapenem-

resistant Acinetobacter baumannii in Shanghai and Hong Kong. Journal of Medical

Microbiology, 55(6), 799–802.

Zhao, W., Yang, S., Huang, Q., & Cai, P. (2015). Bacterial cell surface properties: role of loosely

bound extracellular polymeric substances (LB-EPS). Colloids and Surfaces B: Biointerfaces,

128, 600–607.

Zhou, F., Li, H., Gu, L., Liu, M., Xue, C., Cao, B., … others. (2018). Risk factors for nosocomial

infection among hospitalised severe influenza A (H1N1) pdm09 patients. Respiratory

40

Medicine, 134, 86–91.

Zimbler, D. L., Penwell, W. F., Gaddy, J. A., Menke, S. M., Tomaras, A. P., Connerly, P. L., &

Actis, L. A. (2009). Iron acquisition functions expressed by the human pathogen

Acinetobacter baumannii. Biometals, 22(1), 23–32.

41

Chapter # 2

Introduction to

Computer Aided Vaccine and Drug Designing

In this chapter, we briefly discussed the main computational approaches and techniques that were

used to achieve the objectives of the current research work.

2.1. Computer Aided Vaccine Designing (Reverse Vaccinology)

Reverse vaccinology (RV) is a computer based approach that uses the genomic information of

microorganisms to rationally design vaccines without the need of cultivating the organisms (Sette

& Rappuoli, 2010). Meningococcus B (MenB), an etiological agent of meningococcal meningitis,

was the first bacterial pathogen that was addressed by RV. Because of the extreme variability of

the surface proteins and capsular polysaccharide homology to human self-antigens, the hope for

developing a vaccine against this pathogen was very low and that’s why many traditional

vaccinology based approaches failed. Sequencing of MenB genome and subsequent evaluation

unraveled 90 previously unknown proteins at the pathogen surface and 29 out of which were

capable for killing the pathogen in vitro (Tettelin et al., 2000). In the following years, the antigens

uncovered through this approach were identified to provide immune protective efficacy against

most of the MenB strains (Giuliani et al., 2006). RV has also been applied to other pathogens

including the group B Streptococcus (Maione et al., 2005), group A Streptococcus (Sette &

Rappuoli, 2010), antibiotic-resistant S. pneumoniae and S. aureus (Sette & Rappuoli, 2010), and

Chlamydia (Thorpe et al., 2007).

The process of RV, which is schematically shown in Fig.2.1, commences from the computational

analysis of genome sequence to predict the antigenic sequences that most probably can be used in

vaccine designing. The key features and advantages of this technique are the following (i) the

cultivation of pathogen, for which a vaccine needs to be designed, is not required and thus this

technique is applicable to both cultivable and non-cultivable bacteria, (ii) the pathogens can be

dealt as commensals, (iii) it offers the possibility to study the complete repertoire of the pathogen

42

antigens that expressed at any time-point of the bacterial growth (Rappuoli, 2001). In the recent

years, RV had been modified to pan genomic reverse vaccinology (PGRV) and comparative

reverse vaccinology (CRV). In PGRV, the pathogen pan-genome is investigated to prioritize

vaccine candidates by virtue of in silico tools. The pan-genome is comprised of the dispensable

genome shared by subset of pathogen isolates and the core genome is present in all the sequenced

strains of the pathogen. On the other hand, CRV applies a comparative approach, which compares

the pathogenic and non-pathogenic strains of a pathogen at a genomic level (H Tettelin, 2009). RV

is extensively utilized in Chapter 3 and 4 to identify potential antigenic epitopes in the proteome

and virulome of A. baumannii, respectively.

Fig.2.1. A schematic view of genomic based RV approach (Rappuoli, 2001).

2.2. Computer Aided Drug Designing

2.2.1. Subtractive Proteomics

Drug target identification for therapeutic intervention is the first step in drug discovery (Katsila,

Spyroulias, Patrinos, & Matsoukas, 2016). The term “target” is often used in a broader sense and

applies to different biological entities including genes, proteins and RNA (Hughes, Rees,

Kalindjian, & Philpott, 2011). A good target requires to be safe, efficacious, druggable and meeting

43

the commercial and clinical needs (Hughes, Rees, Kalindjian, & Philpott, 2011). Essentialy, non-

homology to the gut microbiota, virulence factor, involvement in biological pathways and broad

spectrum conservation are among others to be mentioned (R. Gupta, Pradhan, Jain, & Rai, 2017).

The identification of novel drug targets in the genome of bacterial pathogens with resilient features

by implying data mining of ‘omics’ data with the aid of computational tools has replaced the

traditional drug designing approach against such pathogens (Sanober, Ahmad, & Azam, 2017).

Subtractive proteomics (SP) is one these computational-based approaches, that uses the pathogen

genomic/proteomic data to extract genes/proteins that are highly suitable for rational drug design

(Sakharkar, Sakharkar, & Chow, 2004; Butt et al., 2012; Wadood et al., 2018). As a proof, many

potent targets have been described for Pseudomonas and have experimentally been validated as

essential (Perumal, Lim, Chow, Sakharkar, & Sakharkar, 2008). In the past decade or so, SP

approach has been successfully utilized in providing therapeutic targets from bacterial pathogens

proteome to design new antibacterial drugs. SP measures the genes/proteins suitability as potential

target by analyzing the criteria of “selectivity” and “essentiality” (R. Gupta, Pradhan, Jain, & Rai,

2017).

2.2.2. Modeling of Proteins Structure

In silico protein structure prediction is the inference of the protein three dimensional structure (3D)

from its amino acid sequence (Gupta, Akhtar, & Bajpai, 2014). The exponential increase in

sequence knowledge does not reflect substantial biological significance without the protein

structure and thus proves the fundamental biological concept of “Sequence implies the Structure

and Structure implies the Function” and further supports the argument that protein function is

dependent totally on its native 3D structure (Schmidt & Lamzin, 2002). Determining protein

structure viz. X-ray crystallography and NMR are accurate, however, such techniques are over-

whelming ventures that are highly expensive. Additionally, technical limitations in protein

purification and upholding the protein in its native structure subsequently after crystallization are

the vital factors that propels scientist to predict protein structures computationally (Aloy & Russell,

2006). Several different methods are applied to predict the protein structure and can be broadly

categorized into the i) homology modeling, ii) Threading and, iii) ab initio methods.

Homology or comparative modeling aims to build atomic resolution model for a given protein

from an experimental and homologous template structure (Vyas, Ukawala, Ghate, & Chintha,

44

2012). The built structure from this approach rely highly on the target-template sequence similarity

and usually have less errors in loop and side chains positioning (Gupta, Akhtar, & Bajpai, 2014).

Typically protein models generated through homology modeling are analogous to NMR resolved

structure (Gupta, Akhtar, & Bajpai, 2014). In threading or fold recognition, sequence – structure

alignment strategy and fold assignments methods are employed when the sequence similarity is

not in the desired range (Wu & Zhang, 2007). In some cases, the non-availability of a suitable

template for target modeling dictates the use of ab initio methods. These methods apply molecular

and thermodynamic energy parameters at the atomistic level for each amino acid with an ultimate

goal to propose a 3D conformation of the entire protein with maximum stability and minimum

entropy (Wu, Skolnick, & Zhang, 2007).

Currently, there are several online and offline modeling software and servers available that can be

used for protein structure modeling, however, in the current work we used the following,

MODELLER is the most widely used homology modeling program for modeling protein

structure. It implements satisfaction of spatial restraints method to produce geometrical

criteria set that in turn create density function probability for the location of atoms in the

protein (Eswar et al., 2006). Additionally, it also carries out many additional tasks, like

multiple structures/sequences alignment, de novo modeling of loops in protein structures,

optimization and comparison of protein structure etc. (Sali & Blundell, 1993; Webb & Sali,

2014).

SWISS-MODEL works on the method of local similarity and fragment assembly. It

contains three highly integrated components: 1) SWISS-MODEL pipeline, which is a

collection of databases and tools for automated protein structure modeling, 2) SWISS-

MODEL Workspace, which is a web-based graphical user workbench, and 3) SWISS-

MODEL Repository, which is an updated homology models database (Schwede, Kopp,

Guex, & Peitsch, 2003).

Phyre2 is an advanced remote homology detection method to predict 3D model of a

protein, potential ligand binding sites, and to investigate the amino acid variants effect for

user protein sequence (Kelley, Mezulis, Yates, Wass, & Sternberg, 2015).

https://en.wikipedia.org/wiki/Homology_modeling#Satisfaction_of_spatial_restraints

https://en.wikipedia.org/wiki/Probability_density_function

https://en.wikipedia.org/wiki/SWISS-MODEL

45

RaptorX is a webserver for predicting protein structure even without having a close

homolog in the PDB. Additionally, it can also be used for predicting secondary structure,

disordered regions, contact maps, binding sites and contact maps (Kallberg et al., 2012).

ModWeb is a web server designed for automated comparative protein structure modeling

using the best available template from PDB (Eswar et al., 2003).

I-TASSER is an ab initio structure prediction server and works on the method of Threading

fragment structure reassembly (Roy, Kucukural, & Zhang, 2010).

2.2.3. Protein Structure Validation

The sterochemical evaluation of the predicted models for a protein is vital before using it in

subsequent studies as any error in the modelled structure could lead to profound consequences

(Berjanskii et al., 2010). However, due to the complex nature of the protein, it causes/renders the

manual and visual detection of these errors exceedingly difficult. As a result, several different

softwares have been developed to identify these errors. The most important of them, which is

widely used in protein quality structural assessments, is the Ramachandran plot.

Ramachandran plot, initially described in 1963 by G. N. Ramachandran, V. Sasisekharan and C.

Ramakrishnan, is a method to visualize backbone torsion angles ψ (Psi, the N-Cα-C-N) against φ

(Phi, the C-N-Cα-C) of amino acid residues in protein structure (Ramachandran, Ramakrishnan,

Sasisekharan,1963). In situations, where X-ray or in silico protein structure is not refined or bad,

we may find torsion angles in the disallowed region of the Ramachandran plot. The ψ torsion angle

lies between -60° to 120° while φ is present between -60° to - 150°. The plot consists of four

quadrants each of which depict torsional angles for a specified secondary element of a protein. The

torsion angles of beta sheets mostly occupy the top left quadrant, while the right and left handed

α-helices are found bottom left and top right quadrant, respectively. For validating a given protein

structure, the inside region of Ramachandran plot can be classified into allowed, generously-

allowed, low-energy region and disallowed. A good protein model has majority of its residues in

allowed region opposed to the bad model that has most of its residues mapped in the disallowed

region of the Ramachandran graph.

Besides Ramachandran graph, ERRAT is also frequently used for verifying protein structure

determined by either experimental methods or predicted by computational software. In this

https://en.wikipedia.org/wiki/I-TASSER

https://en.wikipedia.org/wiki/Gopalasamudram_Narayana_Ramachandran

https://en.wikipedia.org/wiki/Dihedral_angle

https://en.wikipedia.org/wiki/Amino_acid

https://en.wikipedia.org/wiki/Protein_structure

46

program, statistical analysis of non-bonded interactions in different atom types are performed and

error values are plotted as a function of the residue position. The calculations are based by

comparing the statistics with database of reliable high resolution structures. ERRAT is quite

sensitive to smaller errors compared to the 3D profile analysis (Colovos & Yeates, 1993).

Verify-3D is used to determine atomic model compatibility from its own amino acid sequence.

Based on the environment and location of amino acid, a structural class is assigned to amino acid

like polar, non-polar, loop, alpha, beta etc. and then compared to the findings of a reliable structure

(Eisenberg, Lüthy, & Bowie, 1997).

ProSA determine overall z quality score for the input protein structure. When the score is out of

the range, it illustrates error in the structure. The value of input structure is displayed in the plot

along with the z-score of all experimentally determined structure in the current PDB (Wiederstein

& Sippl, 2007).

2.2.4. Molecular Docking Theory

Molecular docking is a computational method with aim to predict the preferred conformation of a

ligand to macromolecule protein (Meng, Zhang, Mezei, & Cui, 2011). The best fitting orientation

between ligand and receptor protein leads to stable complex formation and thus can be utilized in

measuring the association strength between the receptor and the ligand molecule (Lengauer &

Rarey, 1996). Molecular docking works in two stages: 1) conformational sampling and 2) scoring

function (Meng, Zhang, Mezei, & Cui, 2011).

2.2.4.1. Sampling Algorithms

Several different sampling algorithms are used in molecular docking and can be briefly discussed

below.

Matching algorithms uses a ligand molecular shape map into protein active site in terms of

chemical information and shape features (Brint & Willett, 1987; Norel, Fischer, Wolfson, &

Nussinov, 1994; Meng, Zhang, Mezei, & Cui, 2011). Molecular docking software such as FLOG

(Miller, Kearsley, Underwood, & Sheridan, 1994), DOCK (Kuntz, Blaney, Oatley, Langridge, &

Ferrin, 1982), SANDOCK (Burkhard, Taylor, & Walkinshaw, 1998) and LibDock (Diller & Merz

Jr, 2001) use a matching algorithm for ligand docking.

47

Incremental construction algorithms put the ligand into the protein active site in an incremental

and fragmental fashion (Leach & Kuntz, 1992;Rarey, Kramer, Lengauer, & Klebe, 1996). This

method has been used in FlexX (Rarey, Kramer, Lengauer, & Klebe, 1996), DOCK 4.0 (Ewing,

Makino, Skillman, & Kuntz, 2001), SLIDE (Schnecke & Kuhn, 2000), eHiTS (Zsoldos, Reid,

Simon, Sadjad, & Peter Johnson, 2006) and Hammerhead (Welch, Ruppert, & Jain, 1996).

Multiple Copy Simultaneous Search algorithms is employed for de novo design of ligands using

a fragment-based approach. Moreover, it allows modification of known ligands, to improve the

binding to the receptor (Miranker & Karplus, 1991; Eisen, Wiley, Karplus, & Hubbard, 1994).

Stochastic algorithms randomly modify a ligand/ population of ligands conformation in search of

conformational space (Meng, Zhang, Mezei, & Cui, 2011). Stochastic methods can be further

divided into two typical algorithms: Monte Carlo and genetic algorithms. Monte Carlo algorithms

generate ligand poses through bond rotation, and rigid-body rotation or translation and

subsequently subjected to energy- based selection criterion test (Hart & Read, 1992; Goodsell,

Lauble, Stout, & Olson, 1993). If it passes through the criterion, it is subjected to further

modification to generate next conformation (Meng, Zhang, Mezei, & Cui, 2011). Genetic

algorithms like natural selection in evolution process generate the accurate binding conformation

(Oshiro, Kuntz, & Dixon, 1995; Morris et al., 1998). The binding conformation is assessed by

scoring function and one that survived is used for generating new conformation. Genetic

algorithms using different docking software include DIVALI (Clark, 1995) and DARWIN (Taylor

& Burnett, 2000), AutoDock (Morris et al., 1998), and Genetic Optimization For Ligand Docking

(GOLD) (Verdonk, Cole, Hartshorn, Murray, & Taylor, 2003). The last two are the most common

and frequently used in the current research work.

Molecular dynamics (MD) is a powerful simulation approach used to refine docking findings and

efficient at local optimization (Brooks et al., 1983; Cornell et al., 1995). In this algorithm, the

flexibility of ligand and protein can be effectively achieved (Meng, Zhang, Mezei, & Cui, 2011).

2.2.4.2. Scoring Functions

Scoring functions are employed to define the correct conformation among the incorrect/inactive

ligands in a reasonable computational time. It can be grouped into empirical, knowledge-based

48

scoring functions and force-field based (Head et al., 1996; Kitchen, Decornez, Furr, & Bajorath,

2004).

Empirical scoring function involves decomposing binding energy into components such as

binding entropy, ionic interaction, hydrophobic effect, and hydrogen bond (Head et al., 1996;

Böhm, 1998). Subsequently, a final score is achieved by multiplying each component with a

coefficient (Meng, Zhang, Mezei, & Cui, 2011).

Force-field based scoring functions estimate the binding energy by summing the non-bonded

interactions such as van der Waals and electrostatics (Kollman, 1993; Carlson & Jorgensen, 1995).

Lennard-Jones potential function is used to calculate van der Waals interactions while electrostatic

terms are calculated by a Coulombic formulation (Meng, Zhang, Mezei, & Cui, 2011).

In Knowledge-based scoring functions, crystal structures of protein-ligand complexes are

evaluated statistically to extract the frequencies of interatomic contact and protein and ligand

distances (Wallqvist, Jernigan, & Covell, 1995; Muegge & Martin, 1999). The score calculation

is based on penalizing repulsive interactions and favoring the preferred contacts between protein

and each ligand atom (Meng, Zhang, Mezei, & Cui, 2011).

2.2.5. Docking Methodologies

In Rigid ligand and rigid receptor docking, both ligand and protein are treated as rigid bodies

during docking. The early versions of FLOG (Gabb, Jackson, & Sternberg, 1997), DOCK

(Shoichet, Stroud, Santi, Kuntz, & Perry, 1993), and FTDOCK (Gabb et al., 1997) use this method

for docking.

During Flexible ligand and rigid receptor docking, the ligand is treated flexible while receptor is

considered rigid. Majority of the docking programs used this method for docking such as, FlexX

(Rarey, Kramer, Lengauer, & Klebe, 1996) and AutoDock (Morris et al., 1998).

In Flexible ligand and flexible receptor docking, both ligand and receptor are treated flexible.

Protein intrinsic mobility has been proved to be strongly related to ligand binding and taking this

into account is vital (Teague, 2003). Currently, several methods are available that implements

receptor flexibility. The simplest among them is called soft docking. In this docking, the van der

Waals repulsion energy is decreased in the scoring function that allows atom-atom overlap

49

freedom between the ligand and receptor. Smooth potential in AutoDock 3.0 and the LJ 8-4

potential in GOLD belong to this class (Meng, Zhang, Mezei, & Cui, 2011).

2.2.6. MD Simulations

MD is a computer based simulation program to study the dynamic properties of atoms and

molecules (Kumar & Maiti, 2011). The system is subjected to a fixed period of time to evolve

dynamically during which the atoms and molecules are allowed to interact. Newton's equations of

motion are solved to extract the trajectories of interacting molecules (Hospital, Goni, Orozco, &

Gelpi, 2015). The forces between the particles and associated potential energy are calculated using

molecular mechanics force fields or interatomic potentials. The method of MD simulation was

initially described in the late 1950s and can be applied in material science, chemical physics and

modelling of biomolecules (Alder & Wainwright, 1959; Rahman, 1964).

The interplay between theory and experiment has been changed with the advent of computer

simulations (Allen & others, 2004). Simulation studies allow the flexibility to work with extreme

condition of pressure and temperature with possibility to extract subtle information that otherwise

not possible practically. In case of a well generated computer model, the system can be used by

experimentalists to dig out meaningful insights for interpretation of experimental findings, thus

simulation can be used to bridge theory and experiment (Allen & Tildesley, 2017). This way,

computer simulation can be effectively employed in designing and discovery of new drugs that

experimentally is time consuming and resource costly comparing to in vitro synthesis and

characterization (Liu et al., 2018). That is one of the reasons that applications of computer aided

drug designing including MD simulations are frequently utilized in pharmaceutical industries.

The substantial aspect of MD simulations is to predict system behavior at atomic level. In classical

mechanics, this prediction is made using Newton’s second law which mathematically can be

represented as,

�� = 𝑚��……………………………………………………………………. (I)

Where F, m and a are force, mass and acceleration, respectively. In quantum mechanics (QM),

Schrodinger’s equation is believed to be analogous to Newton’s equation. Schrödinger’s equation

was described in 1926 by Erwin Schrodinger (Schrodinger, 1926).

https://en.wikipedia.org/wiki/Newton%27s_laws_of_motion

https://en.wikipedia.org/wiki/Newton%27s_laws_of_motion

https://en.wikipedia.org/wiki/Molecular_mechanics

https://en.wikipedia.org/wiki/Force_field_(chemistry)

50

𝐸ψ (��, 𝑡) = H𝜓 (��)

= (− ℎ2

2𝑚 + ∇2𝑉 (��)) 𝜓 (��)…………………….(II)

Where E is system total energy, ψ (��, 𝑡) stands for the wave function that contains information

for each particle of the system and is central object in Schrödinger’s equation, H is the Hamiltonian

operator and describe system potential and kinetic energies. The computation of above equation

become complex for systems having large number of particles. Hence, approximations play a

significant role in simplifying computation of Schrödinger’s equation (Swails, 2013).

The goal of QM calculations is to extract molecular and atomic properties of a system using Born–

Oppenheimer approximation (BOA), which assume the separate motion of protein, electron and

neutron (McQuarrie, Cox, & Simon, 1997). These calculations aid experimentalists with reliable

calculations of free-energy barriers of chemical reactions, gas-phase basicity, molecular

geometries, proton affinities, and ionization energies (Range, Riccardi, Cui, Elstner, & York, 2005;

Watson & Bartlett, 2013). The QM calculations are computationally expensive and cannot be

frequently applied to large atomic systems. In that case, classical mechanics is the best suited.

Molecular simulations use force fields and initial structures, developed by experiments and QM

calculations, generate a series of atomic level system information (Gangupomu, 2010). This

microscopic atomic level information regarding its position, and velocity are converted into

macroscopic terms like internal energy and pressure using statistical mechanics (Frenkel & Smit,

2001). It is often convenient to define a multicomponent system as a single component system

where thermodynamic state is described by a small set of parameters. The instantaneous state of

particle momenta and positions are defined through coordinates in multidimensional phase space.

This can be explained by a system having N atoms then phase space will have dimensions 6N

illustrating momenta and positions in all three dimensions. Combining all the points represents

system thermodynamics and known as ensemble. There are four major types of statistical ensemble

(Schlick, 2010):

Canonical (constant-N V T) ensemble

Microcanonical (constant-N V E or) ensemble

The grand canonical (constant- μV T) ensemble

51

The isothermal-isobaric (constant-N P T) ensemble

These ensembles are characterized based on thermodynamic states fixed values like volume (V),

total energy (E) or chemical potential (μ), temperature (T) and number of particles (N).

The canonical ensemble also known as constant temperature, MD have fixed temperature, volume

and the amount of substance. The exothermic and endothermic processes in NVT ensemble is

exchanged with a thermostat.

The microcanonical ensemble is an adiabatic process with no heat exchange. This system is

isolated from changes in energy, moles, and volume.

Grand canonical ensemble is described by constant temperature and volume, temperature and

chemical potential,

The substance amount, temperature and pressure are conserved in NPT ensemble with an

additional need of a barostat. It is closely related to laboratory conditions.

The partition functions in MD simulation aid in estimation of different thermodynamics properties

by sampling representative available points to construct the distributions. In MD for a system of

interacting particles, the equations of motion are integrated into a time dependent manner to build

ensembles. The force on each atom is estimated as the gradient of empirical interatomic potential

energy. The trajectories of MD simulation provide complete information of molecules movement

within the system (Swails, 2013) and that is why MD has been successfully applied in the field of

computational biology and drug discovery.

Classical mechanics integrate Newton’s laws of motion to produce time evolution of a system

configuration and for this trajectories are discretized and an integer is employed to advance over

small steps. The accurate and stable integrator usually employed is velocity-verlet algorithm

(VVA) (Grubmüller, Heller, Windemuth, & Schulten, 1991). VVA is best in longer simulations

and is time reversible and quite simple. It is vital to notice that the choice of time step is critical as

small time scale is costly that leads to greater number of steps to complete simulation length of

time. This ultimately resulted into energy fluctuation because of too large value. The system

accuracy is checked by energy conservation in a microcanonical simulation (Gonzalez, 2011). To

avoid energy drifts due to numerical errors, a thermostat is used in MD simulation. One simplest

https://en.wikipedia.org/wiki/Canonical_ensemble

https://en.wikipedia.org/wiki/Microcanonical_ensemble

https://en.wikipedia.org/wiki/Adiabatic_process

52

way to make temperature constant is by directly controlling the kinetic energy at each time step.

This give instantaneous kinetic energy harmonizing with desired temperature. Temperature

fluctuations can be controlled through an external heat bath as suggested by Berendsen et al

(Berendsen, Postma, van Gunsteren, DiNola, & Haak, 1984). Similarly, for constant pressure

similar approach can be applied. Nose-Hoover thermostat is considered more appropriate

canonical ensemble once equilibrium is attained. Langevin thermostat (Hunenberger, 2005) is

another example of temperature constraint in which each particle is connected with a heat bath.

2.2.7. Force Fields

A force field is a mathematical expression that illustrate system energy dependence on its particles

coordinate (Gonzalez, 2011). It is quite obvious that an accurate model represents the interatomic

forces of a system. The interatomic forces are calculated first by solving all nuclei electronic

structure in a particular configuration. Methods based on empirical force field provide suitable

level of approximation for treating large systems and achieving timescale of thousands of

picoseconds. The common parameters of force fields involves angles, bonds, non-bonded

interactions and torsions and generally can be presented as,

Utotal = Uangle + Ubond + Unon-bonded interactions + Utorsions……………………..…….(III)

The two prominent empirical potentials are: CHARMM (MacKerell Jr et al., 1998) and AMBER

(Weiner & Kollman, 1981).

2.2.8. Periodic Boundary Conditions

Periodic boundary conditions are set of boundary conditions that employed for approximation of

large infinite system by using it in smaller part called unit cell. Each atom in such systems is

allowed to interact with atoms of entire unit cells including its own periodic images (M. Allen &

Tildesley, 1990). In PBC simulation, a common applied practice is to allow interaction of each

atom directly with only a single image especially the nearest of every other atom and is known as

minimum image convention. It was first used in simulation in 1953 by Metropolis et al., to consider

appropriate coordinates (Metropolis, Rosenbluth, Rosenbluth, Teller, & Teller, 1953). It is used in

imposing a limit to range of calculated interactions. The energy calculated for system with PBC

53

using minimum image convention is a single cell unit energy in a generated field by every periodic

cell.

2.2.9. Binding Free Energy Calculations

The goal of MD simulations is to extract thermodynamic and kinetic data of a model system and

several thermodynamic properties can be easily derived from an adequate sample configurations.

One vital thermodynamic aspect is the estimation of free energy that illustrate system stability

(Frenkel & Smit, 2001). Different approaches are currently available to calculate absolute and

relative binding free energy. Among them, thermodynamic integration and Free-energy

perturbation have been successfully used in predicting binding strength of the complexes though

they are computationally expensive (Simonson, Archontis, & Karplus, 2002). The MM/PBSA

(Miller et al., 2012), PROFEC (Pitera & Kollman, 2000), CMC/MD (Eriksson, Pitera, & Kollman,

1999), the ì-dynamics (Pearlman, 1999), 4D-PMF (Rodinger, Howell, & Pomes, 2005), OWFEG

method (Pearlman, 1999) and LIE method (Murcko, 1995) are among the others that are frequently

used in estimating binding free energy rapidly. In this thesis, we used the MM/PBSA method to

calculate interaction energies and solvation energies of enzyme-inhibitor complexes. This method

is now in common use for non-covalently bounded complexes where the ensembles for receptor,

ligand and complex are generated separately (Genheden & Ryde, 2015). The free energy of binding

is done from simulation trajectories and is presented as,

ΔGbinding = (ΔH solv, bound) + (ΔH binding, gas) – (ΔH solv, unbound)………………………….(IV)

The computational cost was reduced in a way to usually run a single simulation only for bounded

complexes and extraction of all the ensembles from it (Homeyer & Gohlke, 2012). In this

dissertation, we used MM/GBSA, MM/PBSA and WaterSwap methods to estimated binding free

energies of docked small drug-like molecules, the specific details and used parameters of which

can be found the afterward chapters.

2.3. References

Alder, B. J., & Wainwright, T. E. (1959). Studies in molecular dynamics. I. General method. The

Journal of Chemical Physics, 31(2), 459–466.

54

Allen, M. P., & others. (2004). Introduction to molecular dynamics simulation. Computational Soft

Matter: From Synthetic Polymers to Proteins, 23, 1–28.

Allen, M. P., & Tildesley, D. J. (2017). Computer simulation of liquids. Oxford university press.

Allen, M., & Tildesley, D. (1990). Computer simulation of liquids Oxford Science. London.

Aloy, P., & Russell, R. B. (2006). Structural systems biology: modelling protein interactions.

Nature Reviews Molecular Cell Biology, 7(3), 188.

Berendsen, H. J. C., Postma, J. P. M. van, van Gunsteren, W. F., DiNola, A., & Haak, J. R. (1984).

Molecular dynamics with coupling to an external bath. The Journal of Chemical Physics,

81(8), 3684–3690.

Berjanskii, M., Liang, Y., Zhou, J., Tang, P., Stothard, P., Zhou, Y., … others. (2010). PROSESS:

a protein structure evaluation suite and server. Nucleic Acids Research, 38(suppl_2), W633-

-W640.

Bohm, H.J. (1998). Prediction of binding constants of protein ligands: a fast method for the

prioritization of hits obtained from de novo design or 3D database search programs. Journal

of Computer-Aided Molecular Design, 12(4), 309.

Brint, A. T., & Willett, P. (1987). Algorithms for the identification of three-dimensional maximal

common substructures. Journal of Chemical Information and Computer Sciences, 27(4),

152–158.

Brooks, B. R., Bruccoleri, R. E., Olafson, B. D., States, D. J., Swaminathan, S. a, & Karplus, M.

(1983). CHARMM: a program for macromolecular energy, minimization, and dynamics

calculations. Journal of Computational Chemistry, 4(2), 187–217.

Burkhard, P., Taylor, P., & Walkinshaw, M. D. (1998). An example of a protein ligand found by

database mining: description of the docking method and its verification by a 2.3 Å X-ray

structure of a Thrombin-Ligand complex1. Journal of Molecular Biology, 277(2), 449–466.

Butt, A. M., Tahir, S., Nasrullah, I., Idrees, M., Lu, J., & Tong, Y. (2012). Mycoplasma genitalium:

a comparative genomics study of metabolic pathways for the identification of drug and

vaccine targets. Infection, Genetics and Evolution, 12(1), 53–62.

55

Carlson, H. A., & Jorgensen, W. L. (1995). An extended linear response method for determining

free energies of hydration. The Journal of Physical Chemistry, 99(26), 10667–10673.

Clark, K. P. (1995). Flexible ligand docking without parameter adjustment across four ligand-

receptor complexes. Journal of Computational Chemistry, 16(10), 1210–1226.

Colovos, C., & Yeates, T. O. (1993). ERRAT: an empirical atom-based method for validating

protein structures. Protein Sci, 2, 1511–1519.

Cornell, W. D., Cieplak, P., Bayly, C. I., Gould, I. R., Merz, K. M., Ferguson, D. M., … Kollman,

P. A. (1995). A second generation force field for the simulation of proteins, nucleic acids,

and organic molecules. Journal of the American Chemical Society, 117(19), 5179–5197.

Diller, D. J., & Merz Jr, K. M. (2001). High throughput docking for library design and library

prioritization. Proteins: Structure, Function, and Bioinformatics, 43(2), 113–124.

Eisen, M. B., Wiley, D. C., Karplus, M., & Hubbard, R. E. (1994). HOOK: a program for finding

novel molecular architectures that satisfy the chemical and steric requirements of a

macromolecule binding site. Proteins: Structure, Function, and Bioinformatics, 19(3), 199–

221.

Eisenberg, D., Luthy, R., & Bowie, J. U. (1997). VERIFY3D: Assessment of protein models with

three-dimensional profiles. Methods in Enzymology, 277, 396–404.

Eriksson, M. A. L., Pitera, J., & Kollman, P. A. (1999). Prediction of the binding free energies of

new TIBO-like HIV-1 reverse transcriptase inhibitors using a combination of PROFEC,

PB/SA, CMC/MD, and free energy calculations. Journal of Medicinal Chemistry, 42(5), 868–

881.

Eswar, N., John, B., Mirkovic, N., Fiser, A., Ilyin, V. A., Pieper, U., … others. (2003). Tools for

comparative protein structure modeling and analysis. Nucleic Acids Research, 31(13), 3375–

3380.

Eswar, N., Webb, B., Marti-Renom, M. A., Madhusudhan, M. S., Eramian, D., Shen, M., … Sali,

A. (2006). Comparative protein structure modeling using Modeller. Current Protocols in

Bioinformatics, 15(1), 5–6.

56

Ewing, T. J. A., Makino, S., Skillman, A. G., & Kuntz, I. D. (2001). DOCK 4.0: search strategies

for automated molecular docking of flexible molecule databases. Journal of Computer-Aided

Molecular Design, 15(5), 411–428.

Frenkel, D., & Smit, B. (2001). Understanding molecular simulation: from algorithms to

applications (Vol. 1). Elsevier.

Gabb, H. A., Jackson, R. M., & Sternberg, M. J. E. (1997). Modelling protein docking using shape

complementarity, electrostatics and biochemical information1. Journal of Molecular Biology,

272(1), 106–120.

Gangupomu, V. K. (2010). Molecular simulations of HIV-1 gp41 protein-membrane interactions.

Genheden, S., & Ryde, U. (2015). The MM/PBSA and MM/GBSA methods to estimate ligand-

binding affinities. Expert Opinion on Drug Discovery, 10(5), 449–461.

Giuliani, M. M., Adu-Bobie, J., Comanducci, M., Arico, B., Savino, S., Santini, L., … others.

(2006). A universal vaccine for serogroup B meningococcus. Proceedings of the National

Academy of Sciences, 103(29), 10834–10839.

Gonzalez, M. A. (2011). Force fields and molecular dynamics simulations. Ecole Thematique de

La Societe Française de La Neutronique, 12, 169–200.

Goodsell, D. S., Lauble, H., Stout, C. D., & Olson, A. J. (1993). Automated docking in

crystallography: analysis of the substrates of aconitase. Proteins: Structure, Function, and


Grubmuller, H., Heller, H., Windemuth, A., & Schulten, K. (1991). Generalized Verlet algorithm

for efficient molecular dynamics simulations with long-range interactions. Molecular

Simulation, 6(1–3), 121–142.

Gupta, C. L., Akhtar, S., & Bajpai, P. (2014). In silico protein modeling: possibilities and

limitations. EXCLI Journal, 13, 513.

Gupta, R., Pradhan, D., Jain, A. K., & Rai, C. S. (2017). TiD: Standalone software for mining

putative drug targets from bacterial proteome. Genomics, 109(1), 51–57.

Hart, T. N., & Read, R. J. (1992). A multiple-start Monte Carlo docking method. Proteins:

57

Structure, Function, and Bioinformatics, 13(3), 206–222.

Head, R. D., Smythe, M. L., Oprea, T. I., Waller, C. L., Green, S. M., & Marshall, G. R. (1996).

VALIDATE: A new method for the receptor-based prediction of binding affinities of novel

ligands. Journal of the American Chemical Society, 118(16), 3959–3969.

Homeyer, N., & Gohlke, H. (2012). Free energy calculations by the molecular mechanics Poisson-

Boltzmann surface area method. Molecular Informatics, 31(2), 114–122.

Hospital, A., Goni, J. R., Orozco, M., & Gelpi, J. L. (2015). Molecular dynamics simulations:

advances and applications. Advances and Applications in Bioinformatics and Chemistry:

AABC, 8, 37.

Hughes, J. P., Rees, S., Kalindjian, S. B., & Philpott, K. L. (2011). Principles of early drug

discovery. British Journal of Pharmacology, 162(6), 1239–1249.

Hunenberger, P. H. (2005). Thermostat algorithms for molecular dynamics simulations. In

Advanced computer simulation (pp. 105–149). Springer.

Kallberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H., & Xu, J. (2012). Template-based

protein structure modeling using the RaptorX web server. Nature Protocols, 7(8), 1511.

Katsila, T., Spyroulias, G. A., Patrinos, G. P., & Matsoukas, M.T. (2016). Computational

approaches in target identification and drug discovery. Computational and Structural

Biotechnology Journal, 14, 177–184.

Kelley, L. A., Mezulis, S., Yates, C. M., Wass, M. N., & Sternberg, M. J. E. (2015). The Phyre2

web portal for protein modeling, prediction and analysis. Nature Protocols, 10(6), 845–858.

Kitchen, D. B., Decornez, H., Furr, J. R., & Bajorath, J. (2004). Docking and scoring in virtual

screening for drug discovery: methods and applications. Nature Reviews Drug Discovery,

3(11), 935.

Kollman, P. (1993). Free energy calculations: applications to chemical and biochemical

phenomena. Chemical Reviews, 93(7), 2395–2417.

Kumar, H., & Maiti, P. K. (2011). Introduction to Molecular Dynamics Simulation. In

Computational Statistical Physics (pp. 161–197). Springer.

58

Kuntz, I. D., Blaney, J. M., Oatley, S. J., Langridge, R., & Ferrin, T. E. (1982). A geometric

approach to macromolecule-ligand interactions. Journal of Molecular Biology, 161(2), 269–

288.

Leach, A. R., & Kuntz, I. D. (1992). Conformational analysis of flexible ligands in

macromolecular receptor sites. Journal of Computational Chemistry, 13(6), 730–748.

Lengauer, T., & Rarey, M. (1996). Computational methods for biomolecular docking. Current

Opinion in Structural Biology, 6(3), 402–406.

Liu, X., Shi, D., Zhou, S., Liu, H., Liu, H., & Yao, X. (2018). Molecular dynamics simulations

and novel drug discovery. Expert Opinion on Drug Discovery, 13(1), 23–37.

MacKerell Jr, A. D., Bashford, D., Bellott, M., Dunbrack Jr, R. L., Evanseck, J. D., Field, M. J.,

… others. (1998). All-atom empirical potential for molecular modeling and dynamics studies

of proteins. The Journal of Physical Chemistry B, 102(18), 3586–3616.

Maione, D., Margarit, I., Rinaudo, C. D., Masignani, V., Mora, M., Scarselli, M., … others. (2005).

Identification of a universal Group B streptococcus vaccine by multiple genome screen.

Science, 309(5731), 148–150.

McQuarrie, D. A., Cox, H., & Simon, J. D. (1997). Physical chemistry: a molecular approach.

Sterling Publishing Company.

Meng, X. Y., Zhang, H.X., Mezei, M., & Cui, M. (2011). Molecular docking: a powerful approach

for structure-based drug discovery. Current Computer-Aided Drug Design, 7(2), 146–157.

Metropolis, N., Rosenbluth, A. W., Rosenbluth, M. N., Teller, A. H., & Teller, E. (1953). Equation

of state calculations by fast computing machines. The Journal of Chemical Physics, 21(6),

1087–1092.

Miller, B. R., McGee Jr, T. D., Swails, J. M., Homeyer, N., Gohlke, H., & Roitberg, A. E. (2012).

MMPBSA. py: an efficient program for end-state free energy calculations. Journal of

Chemical Theory and Computation, 8(9), 3314–3321.

Miller, M. D., Kearsley, S. K., Underwood, D. J., & Sheridan, R. P. (1994). FLOG: a system to

select ‘quasi-flexible’ligands complementary to a receptor of known three-dimensional

59

structure. Journal of Computer-Aided Molecular Design, 8(2), 153–174.

Miranker, A., & Karplus, M. (1991). Functionality maps of binding sites: a multiple copy

simultaneous search method. Proteins: Structure, Function, and Bioinformatics, 11(1), 29–

34.

Morris, G. M., Goodsell, D. S., Halliday, R. S., Huey, R., Hart, W. E., Belew, R. K., & Olson, A.

J. (1998). Automated docking using a Lamarckian genetic algorithm and an empirical binding

free energy function. Journal of Computational Chemistry, 19(14), 1639–1662.

Muegge, I., & Martin, Y. C. (1999). A general and fast scoring function for protein-ligand

interactions: a simplified potential approach. Journal of Medicinal Chemistry, 42(5), 791–

804.

Murcko, M. A. (1995). Computational methods to predict binding free energy in ligand-receptor

complexes. Journal of Medicinal Chemistry, 38(26), 4953–4967.

Norel, R., Fischer, D., Wolfson, H. J., & Nussinov, R. (1994). Molecular surface recognition by a

computer vision-based technique. Protein Engineering, Design and Selection, 7(1), 39–46.

Oshiro, C. M., Kuntz, I. D., & Dixon, J. S. (1995). Flexible ligand docking using a genetic

algorithm. Journal of Computer-Aided Molecular Design, 9(2), 113–130.

Pearlman, D. A. (1999). Free energy grids: a practical qualitative application of free energy

perturbation to ligand design using the OWFEG method. Journal of Medicinal Chemistry,

42(21), 4313–4324.

Perumal, D., Lim, C. S., Chow, V. T. K., Sakharkar, K. R., & Sakharkar, M. K. (2008). A combined

computational-experimental analyses of selected metabolic enzymes in Pseudomonas

species. International Journal of Biological Sciences, 4(5), 309.

Pitera, J. W., & Kollman, P. A. (2000). Exhaustive mutagenesis in silico: multicoordinate free

energy calculations on proteins and peptides. Proteins: Structure, Function, and


Rahman, A. (1964). Correlations in the motion of atoms in liquid argon. Physical Review, 136(2A),

A405.

60

Range, K., Riccardi, D., Cui, Q., Elstner, M., & York, D. M. (2005). Benchmark calculations of

proton affinities and gas-phase basicities of molecules important in the study of biological

phosphoryl transfer. Physical Chemistry Chemical Physics, 7(16), 3070–3079.

Rappuoli, R. (2001). Reverse vaccinology, a genome-based approach to vaccine development.

Vaccine, 19(17–19), 2688–2691.

Rarey, M., Kramer, B., Lengauer, T., & Klebe, G. (1996). A fast flexible docking method using

an incremental construction algorithm. Journal of Molecular Biology, 261(3), 470–489.

Rodinger, T., Howell, P. L., & Pomes, R. (2005). Absolute free energy calculations by

thermodynamic integration in four spatial dimensions. The Journal of Chemical Physics,

123(3), 34104.

Roy, A., Kucukural, A., & Zhang, Y. (2010). I-TASSER: a unified platform for automated protein

structure and function prediction. Nature Protocols, 5(4), 725.

Sakharkar, K. R., Sakharkar, M. K., & Chow, V. T. K. (2004). A novel genomics approach for the

identification of drug targets in pathogens, with special reference to Pseudomonas

aeruginosa. In Silico Biology, 4(3), 355–360.

Sali, A., & Blundell, T. L. (1993). Comparative protein modelling by satisfaction of spatial

restraints. Journal of Molecular Biology, 234(3), 779–815.

Sanober, G., Ahmad, S., & Azam, S. S. (2017). Identification of plausible drug targets by

investigating the druggable genome of MDR Staphylococcus epidermidis. Gene Reports, 7,

147–153.

Schlick, T. (2010). Molecular modeling and simulation: an interdisciplinary guide: an

interdisciplinary guide (Vol. 21). Springer Science & Business Media.

Schmidt, A., & Lamzin, V. S. (2002). Veni, vidi, vici—atomic resolution unravelling the mysteries

of protein function. Current Opinion in Structural Biology, 12(6), 698–703.

Schnecke, V., & Kuhn, L. A. (2000). Virtual screening with solvation and ligand-induced

complementarity. In Virtual Screening: An Alternative or Complement to High Throughput

Screening? (pp. 171–190). Springer.

61

Schrodinger, E. (1926). An undulatory theory of the mechanics of atoms and molecules. Physical

Review, 28(6), 1049.

Schwede, T., Kopp, J., Guex, N., & Peitsch, M. C. (2003). SWISS-MODEL: an automated protein

homology-modeling server. Nucleic Acids Research, 31(13), 3381–3385.

Sette, A., & Rappuoli, R. (2010). Reverse vaccinology: developing vaccines in the era of

genomics. Immunity, 33(4), 530–541.

Shoichet, B. K., Stroud, R. M., Santi, D. V, Kuntz, I. D., & Perry, K. M. (1993). Structure-based

discovery of inhibitors of thymidylate synthase. Science, 259(5100), 1445–1450.

Simonson, T., Archontis, G., & Karplus, M. (2002). Free energy simulations come of age: protein-

ligand recognition. Accounts of Chemical Research, 35(6), 430–437.

Swails, J. M. (2013). Free energy simulations of complex biological systems at constant pH.

University of Florida.

Taylor, J. S., & Burnett, R. M. (2000). DARWIN: a program for docking flexible molecules.

Proteins: Structure, Function, and Bioinformatics, 41(2), 173–191.

Teague, S. J. (2003). Implications of protein flexibility for drug discovery. Nature Reviews Drug

Discovery, 2(7), 527.

Tettelin, H. (2009). The bacterial pan-genome and reverse vaccinology. In Microbial

Pathogenomics (Vol. 6, pp. 35–47). Karger Publishers.

Tettelin, H., Saunders, N. J., Heidelberg, J., Jeffries, A. C., Nelson, K. E., Eisen, J. A., … others.

(2000). Complete genome sequence of Neisseria meningitidis serogroup B strain MC58.

Science, 287(5459), 1809–1815.

Thorpe, C., Edwards, L., Snelgrove, R., Finco, O., Rae, A., Grandi, G., … Hussell, T. (2007).

Discovery of a vaccine antigen that protects mice from Chlamydia pneumoniae infection.

Vaccine, 25(12), 2252–2260.

Verdonk, M. L., Cole, J. C., Hartshorn, M. J., Murray, C. W., & Taylor, R. D. (2003). Improved

protein-ligand docking using GOLD. Proteins: Structure, Function, and Bioinformatics,

52(4), 609–623.

62

Vyas, V. K., Ukawala, R. D., Ghate, M., & Chintha, C. (2012). Homology modeling a fast tool for

drug discovery: current perspectives. Indian Journal of Pharmaceutical Sciences, 74(1), 1.

Wadood, A., Jamal, A., Riaz, M., Khan, A., Uddin, R., Jelani, M., & Azam, S. S. (2018).

Subtractive genome analysis for in silico identification and characterization of novel drug

targets in Streptococcus pneumonia strain JJA. Microbial Pathogenesis, 115, 194–198.

Wallqvist, A., Jernigan, R. L., & Covell, D. G. (1995). A preference-based free-energy

parameterization of enzyme-inhibitor binding. Applications to HIV-1-protease inhibitor

design. Protein Science, 4(9), 1881–1903.

Watson, T. J., & Bartlett, R. J. (2013). Infinite order relaxation effects for core ionization energies

with a variational coupled cluster ansatz. Chemical Physics Letters, 555, 235–238.

Webb, B., & Sali, A. (2014). Comparative protein structure modeling using MODELLER. Current

Protocols in Bioinformatics, 47(1), 5–6.

Weiner, P. K., & Kollman, P. A. (1981). AMBER: Assisted model building with energy

refinement. A general program for modeling molecules and their interactions. Journal of

Computational Chemistry, 2(3), 287–303.

Welch, W., Ruppert, J., & Jain, A. N. (1996). Hammerhead: fast, fully automated docking of

flexible ligands to protein binding sites. Chemistry & Biology, 3(6), 449–462.

Wiederstein, M., & Sippl, M. J. (2007). ProSA-web: interactive web service for the recognition of

errors in three-dimensional structures of proteins. Nucleic Acids Research, 35(suppl_2),

W407-W410.

Wu, S., Skolnick, J., & Zhang, Y. (2007). Ab initio modeling of small proteins by iterative

TASSER simulations. BMC Biology, 5(1), 17.

Wu, S., & Zhang, Y. (2007). LOMETS: a local meta-threading-server for protein structure

prediction. Nucleic Acids Research, 35(10), 3375–3382.

Zsoldos, Z., Reid, D., Simon, A., Sadjad, B. S., & Peter Johnson, A. (2006). eHiTS: an innovative

approach to the docking and scoring function problems. Current Protein and Peptide Science,

7(5), 421–435.

63

Chapter # 3

Combating Tigecycline Resistant Acinetobacter baumannii: A

Leap Forward towards Multi-epitope based Vaccine

Discovery

3.1. Abstract

Global emergence of Tigecycline resistant A. baumannii (TRAB) is on the horizon and poses a

very serious threat to human health. There is a pressing demand for suitable therapeutics against

this pathogen, particularly a vaccine to protect against TRAB infections. We present a

comprehensive investigation of the complete proteome of a TRAB AB031 strain to predict

promiscuous antigenic, non-allergenic, virulent B-cell derived T-cell epitopes and formulate a

multi-epitope vaccine against the pathogen. We identified epitopes from three proteins: outer

membrane protein assembly factor (BamA), fimbrial biogenesis outer membrane usher protein

(FimD) and type IV secretion protein (Rhs) that are appropriate for vaccine design. These

proteins constitute the core proteome of the pathogen, are essential, localized at the pathogen

surface, non-homologous to humans, mice and to the beneficial probiotic bacteria that reside

the human gut. Moreover, these proteins are ideal candidates for experimental investigation as

they have favorable physicochemical properties and have strong cellular interacting networks.

The predicted epitopes: FPLNDKPGD (BamA), FVHAEEAAA (FimD) and YVVAGTAAA

(Rhs) have exo-membrane topology for efficient recognition of the host immune system and high

affinity for the most prevalent allele in human populations, the DRB*0101. These epitopes were

linked and attached to an adjuvant to enhance its antigenicity. The multi-epitope vaccine-construct

was docked with the TLR4 receptor to assess its affinity for the protein and thus its presentation

to the host immune system. Docking results were validated through MD simulations and binding

free energies were calculated using the molecular mechanics/generalized Born (MM/GBSA)

method. In conclusion, we expect the designed chimeric vaccine is highly likely to be effective

against infections caused by TRAB.

64

3.2. Introduction

A. baumannii is a Gram-negative, aerobic and non-motile bacillus that has emerged as the most

troublesome pathogen for many health care institutions globally (Peleg, Seifert, & Paterson, 2008;

Howard, O’Donoghue, Feeney, & Sleator, 2012; Almaghrabi, Joseph, Assiry, & Hamid, 2018). A.

baumannii is commonly found in blood stream, lung, urinary tract and wound infections and it

especially effects immune-compromised populations (Lee et al., 2017). The antibiotic resistance

in five bacterial pathogens: A. baumanii, E. coli, Klebsiella pneumoniae, P. aeruginosa and S.

aureus cost $2.9 billion to the United States health care industry. The cost for consuming one

standard unit (SU) of antibiotics ranged from $0.6 for broad spectrum penicillins, cephalosporins,

and quinolones to $0.1 for 0.1 SU of carbapenems (Shrestha et al., 2018). Multi-drug resistant A.

baumanniii (MDAB) is becoming a serious health concern and compromising healthcare outcomes

due to its remarkable ability of acquiring resistant genes and surviving exposure to surface

microcides (Manchanda, Sanchaita, & Singh, 2010;Almaghrabi, Joseph, Assiry, & Hamid, 2018).

It has been reported that MDAB infected patients are at an increased risk of morbidity and

mortality which contribute significantly to high health costs. According to the reports of CDC,

MDAB is a serious threat that causes approximately 7,000 infections and 500 deaths each year in

the United States (CDC, 2016). The outbreaks of MDAB are generally pervasive and cases have

been documented from countries including Brazil, China, the UK, the USA, India, Spain,

Germany, Iran, Turkey and Iraq etc. (Skariyachan, Manjunath, & Bachappanavar, 2018). The

colonization rate of A. baumannii in hospitalized patients is 75% while in healthy individuals it is

42.5% (Mohd et al., 2018). The recommended therapy for MDAB includes carbapenems such as

imipenem and meropenem; however, a significant rise of carbapenem-resistant A. baumannii

(CRAB) has been reported worldwide (Pogue, Mann, Barber, & Kaye, 2013). The crude

mortality rate observed for the hospital acquired CRAB infections in intensive care units ranged

between 26% to 76% (Skariyachan, Manjunath, & Bachappanavar, 2018). The last treatment

resort for MDAB is colistin; however, severe side effects of nephrotoxicity and neurotoxicity

have limited its use (Spapen, Jacobs, Van Gorp, Troubleyn, & Honore, 2011). The first identified

and US Food and Drug Administration (FDA) approved glycylcycline antibiotic, Tigecycline,

mimics the tetracycline structure and is effective against carbapenem-resistant bacteria

(Karageorgopoulos, Kelesidis, Kelesidis, & Falagas, 2008). Tigecycline is recommended for the

treatment of serious complications including: community acquired pneumonia, skin, and intra-

65

abdominal infections (Karageorgopoulos, Kelesidis, Kelesidis, & Falagas, 2008; Curcio &

Fernandez, 2008). Tigecycline is considered a last resort because of its bacteriostatic activity

against MDR Acinetobacter (Viehman, Nguyen, & Doi, 2014). It disrupts protein synthesis by

blocking the entry of aminoacyl-tRNA into the Aminoacyl (A) site of the 30S ribosome in

prokaryotic translation (Dixit, Madduri, & Sharma, 2014). Unfortunately, an increasing number

of cases have been reported pointing to the development of Tigecycline-resistant A. baumannii

(TRAB) since its first use in clinics (Navon-Venezia, Leavitt, & Carmeli, 2007;Kulah et al., 2009;

Taneja, Singh, Singh, & Sharma, 2011). TRAB is considered as a superbug because of its

resistance to the majority of the antibiotics that are in use today (Peleg, Seifert, & Paterson,

2008; Howard, O’Donoghue, Feeney, & Sleator, 2012; Almaghrabi, Joseph, Assiry, & Hamid,

2018).Due to several limitations of the current therapy regime, such as the lack of a vaccine and

incomplete efficacy of the drugs, there is a desperate need to develop novel approaches that speed

up the identification of novel vaccines and drug targets to curtail serious infections of TRAB

(Hashemzehi, Doosti, Kargar, & Jaafarinia, 2018). Engineering a safe vaccine is of particular

importance since preventive measures are always better than curative approaches (Andre et al.,

2008). Similarly, vaccine resistance rarely emerged compared to antibiotics. It is well-established

that vaccines are far more robust against evolution if compared to drugs. This can be explained in

two ways. First, vaccines work prophylactically in contrast to drugs that function therapeutically.

Secondly, vaccines induce immunity against multiple targets while drugs target only one in the

majority of the cases. Therefore, pathogen populations create lesser mutation for vaccine resistance

in contrast to drugs (Kennedy & Read, 2017). Conventional vaccines are developed from live

attenuated or inactivated organisms, which elicit strong humoral and cellular immunity resulting

in long lasting immunity (Nascimento & Leite, 2012; Minor, 2015). However, the use of such

vaccines has been questioned due to crucial safety concerns as the used pathogen may become

reactivated (Vitrinel & Erdaug, 2008; Chang et al., 2010). Similarly, subunit-based vaccines

containing a single whole protein are also not a good choice because of non-specific immune

responses, which may occur due to the presence of multiple antigenic determinants (Weidang Li,

Joshi, Singhania, Ramsey, & Murthy, 2014). In addition, the formulation of such vaccines often

leads to reactogenic and/or allergenic responses that are often not desired (Weidang Li, Joshi,

Singhania, Ramsey, & Murthy, 2014). The design of epitope-driven or peptide-based vaccines is

more attractive; they are comparatively easier to produce and construct, they lack any infectious

66

potential and offer chemical stability (Weidang Li, Joshi, Singhania, Ramsey, & Murthy, 2014;

Baseer, Ahmad, Ranaghan, & Azam, 2017). However, peptide-based vaccines still present many

obstacles including the low intrinsic antigenicity of using individual peptides, and the shortage of

suitable delivery routes. For proper engagement of cellular and humoral immunity, the combining

task of different peptides and poor population coverage of T-cell epitopes is also challenging

(Weidang Li, Joshi, Singhania, Ramsey, & Murthy, 2014). In addition, the unreliable prediction

of B-cell epitopes (Weidang Li, Joshi, Singhania, Ramsey, & Murthy, 2014). Circumventing these

limitations, the construction of multi-epitope vaccines by linking the safest and B-cell derived T-

cell antigenic epitope offers many significant benefits (Saadi, Karkhah, & Nouri, 2017;Nezafat,

Eslami, Negahdaripour, Rahbar, & Ghasemi, 2017). These include broader intrinsic

immunogenicity, both humoral and cellular immunity can properly be engaged and better T-cell

epitopes population coverage (Rosa et al., 2015). The novel combination of immuno-informative

approaches, together with the knowledge of host immune responses and the exponential increase

in complete genome sequencing of pathogen strains, now makes it possible to select the most

antigenic epitopes (Rosa et al., 2015). As such, it is conceived that cocktails of selected and defined

epitopes may provide foundations for the rational design of vaccines capable of eliciting

convenient cellular and humoral immunity (Sette & Rappuoli, 2010). The majority of vaccines are

based on humoral immunity; however, cellular immunity-based vaccines are preferred as the host

generates strong CD8+ T-cell immune responses against the pathogen infected cells (Soria-Guerra,

Nieto-Gomez, Govea-Alonso, & Rosales-Mendoza, 2015). T-cell immunity is vital because of the

antigenic drift phenomena may aid in foreign particle escape from B-cell immunity (Clem, 2011)

and T-cell immunity is long lasting. Thus, T-cell immunity could provide better foundations for

the design of multi-epitope effective vaccines that can trigger both arms of immunity (Amanna &

Slifka, 2011). Intracellular antigen processing pathways with linear epitopes as predominant

targets lead to cytotoxic T-cell responses. Therefore, it is essential to ensure that the selected

epitopes of T-cell have binding to more than one major histocompatible allele and cover a major

human population (Cohen & Weiner, 1988; Minor, 2015;Hassan et al., 2016; Asad, Ahmad,

Rungrotmongkol, Ranaghan, & Azam, 2018). In this study, we have used bioinformatics

techniques to predict promiscuous antigenic epitopes, which bind to multiple HLA alleles within

a heterogeneous population, therefore, applicable to a wide range of individuals (Dar et al., 2016).

Similarly, the epitopes are B-cell derived T-cell that can elicit both humoral and cellular immunity

67

against TRAB and were formulated into a multi-epitope vaccine (Cohen & Weiner, 1988; Hassan

et al., 2016; Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam, 2018). Furthermore, docking

and dynamics simulations were performed to unveil vaccine-construct binding and stability

with the TLR4. TLR4 is a member of toll-like receptor family that spot and mediate cytokine

production against molecular patterns expressed on infectious agent. The findings of this study

could be useful in the design of effective vaccines for the TRAB superbug.

3.3. Materials and Methods

The complete workflow of the current study is illustrated in Fig.3.1.

3.3.1. Proteome Retrieval and Subtractive Proteomics

The complete proteome of TRAB strain AB031 (Loewen, Alsaadi, Fernando, & Kumar, 2014)

was retrieved from the NCBI (Pruitt, Tatusova, & Maglott, 2005) and used in subtractive

proteomics hierarchy to extract proteins appropriate for peptide vaccine designing (S. Ahmad,

Raza, Uddin, & Azam, 2018). These proteins include those that are part of pathogen core

proteome, host non-homologous, and essential for pathogen survival (Naz et al., 2015).

Pathogen core proteins were identified using Cluster Database At High Identity With Tolerance

(CD-HIT) (Weizhong Li & Godzik, 2006) that removed proteins sharing sequence identity of

60% (Baseer, Ahmad, Ranaghan, & Azam, 2017; Asad, Ahmad, Rungrotmongkol, Ranaghan, &

Azam, 2018). Moving ahead, the non-paralogous proteins of TRAB were used in a BLASTp

search against H. sapiens (taxid: 9606) with the inclusion of the E-value threshold set to 0.005,

bit score ≥ 100 and sequence identity of ≤ 30% (Azam & Shamim, 2014). Proteins that were

adequately similar were filtered, while the remaining non-similar proteins were removed and

considered as host non-homologous proteins. Essential proteins of TRAB were identified by

comparing the non-similar proteins against the Database Of Essential Genes (DEG) (R. Zhang,

Ou, & Zhang, 2004; Barh et al., 2013). The BLASTp search was performed with an E-value cut-

off > 0.0001, minimum bit score of 100 and sequence identity of ≥ 35% (Azam & Shamim,

2014).

3.3.2. Exo-proteome and Secretome Prediction

The exo-proteome and secretome of TRAB consist of those proteins that are localized in the

extracellular and outer membrane, respectively (Naz et al., 2015). A comparative subcellular

68

localization strategy was applied for the prediction of the TRAB exo-proteome and secretome

(Azam & Shamim, 2014; S. Ahmad & Azam, 2018). First, PSORTb v 3.0 (N. Y. Yu et al., 2010)

was used to predict cytoplasmic, inner membrane, outer membrane, extracellular and unknown

proteins. Those proteins, which were categorized as extracellular and outer membrane, were

cross-validated by CELLO v 2.5 (C.S. Yu, Chen, Lu, & Hwang, 2006) and CELLO2GO (C.S.

Yu et al., 2014). The proteins of unknown localization from PSORTb, CELLO and CELLO2GO

analysis were discarded to achieve consistency in the results. At the end, those predicted to be

localized in extracellular and outer membrane regions by all three tools were selected and

analyzed further in subsequent steps (Naz et al., 2015; S. Ahmad & Azam, 2018).

3.3.3. Virulent Proteins Evaluation

Virulent proteins in the pool of essential exo-proteome and secretome proteins were identified

using a BLASTp search of the Virulent Factor Database (VFDB) (L. Chen et al., 2005) with a

threshold of bit score ≥ 100 and sequence identity cut-off 30% (Asad, Ahmad, Rungrotmongkol,

Ranaghan, & Azam, 2018).

3.3.4. Screening of Non-probiotic Proteins

Probiotic bacteria are friendly bacteria and inhabit the gut of humans and animals (Kechagia et al.,

2013). Selection of TRAB proteins that show significant homology with probiotic bacteria can

lead to inaccurate and non-targeted immune responses and ultimately result in accidental inhibition

of probiotic bacteria (Wadood et al., 2018). In order to avoid this, a BLASTp search of NCBI

against three reference strains of intestinal lactic acid bacterial species was performed:

Lactobacillus rhamnosus (taxid: 47715), Lactobacillus casei (taxid: 1582), and Lactobacillus

johnsonii (taxid: 33959). The identity sequence cut-off was set to ≤ 30%, E-value threshold set to

0.005 and minimum bit score of 100.

3.3.5. Screening of Non-similar Mouse Proteins

The proteins passed forward by the previous filter were further used in a BLASTp search against

the mouse proteome (taxid: 10088) and those with threshold E-value of 0.005 and sequence

similarity of ≤ 30% were categorized as mouse non-similar proteins (He, Xiang, & Mobley,

2010).

https://en.wikipedia.org/wiki/Lactobacillus_rhamnosus

https://en.wikipedia.org/wiki/Lactobacillus_casei

https://en.wikipedia.org/wiki/Lactobacillus_johnsonii

https://en.wikipedia.org/wiki/Lactobacillus_johnsonii

69

3.3.6. Physicochemical Characterization

To prioritize proteins that are most suitable for experimental validations, the proteins of the

previous filter were characterized based on several physicochemical parameters (Naz et al., 2015;

Baseer, Ahmad, Ranaghan, & Azam, 2017, Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam,

2018). These parameters include: molecular weight, number of transmembrane helices,

antigenicity and allergenicity. An online computational tool ProtParm (Gasteiger et al., 2005) was

used to estimate molecular weight, aliphatic index, extinction coefficient, instability index, atomic

composition, amino acid composition, estimated half-life and grand average of hydropathicity

(GRAVY). The number of transmembrane helices contained within the protein was computed

through TMHMM (Krogh, Larsson, Von Heijne, & Sonnhammer, 2001) and HMMTOP (Tusnady

& Simon, 2001) with the threshold set to 2 (Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam,

2018).

3.3.7. Antigenicity Prediction

The antigenicity of the proteins was determined through VaxiJen (Doytchinova & Flower, 2007).

Proteins with an antigenicity score ≥ 0.4 were considered as likely to be antigenic.

3.3.8. Prediction of B-cell derived T-cell Epitopes

The prioritized proteins then underwent B-cell epitope prediction using BCpred (EL-Manzalawy,

Dobbs, & Honavar, 2008) with epitope length set to 20. B-cell epitopes that have score > 0.8 were

selected for analysis of exo-membrane topology using TMHMM. Only surface exposed B-cell

epitopes were considered in the subsequent T-cell epitope prediction phase. B-cell derived T-cell

epitopes contain both B-cell and T-cell epitopes and can mediate both arms of immunity i.e.

humoral and cellular immunity (Naz et al., 2015). T-cell epitopes prediction involves evaluation

of B-cell epitopes for their effective binding with molecules of both classes of major

histocompatibility complex (MHC) i.e. MHC class I and MHC class II (Naz et al., 2015). Common

binding T-cell epitopes interacting with the maximum number of MHC alleles were pooled out.

Binding alleles of MHC class I were predicted by Propred1 (H. Singh & Raghava, 2003), while

those alleles of MHC class II were identified by Propred (H. Singh & Raghava, 2001). Antigenicity

and virulence of the candidate B-cell derived T-cell epitopes were determined by VaxiJen

(Doytchinova & Flower, 2007) and VirulentPred (Garg & Gupta, 2008). The allergenicity of the

70

epitopes was predicted by Allertop (Dimitrov, Bangov, Flower, & Doytchinova, 2014). Lastly, the

chemical stability and resistant sequence of the epitopes were evaluated through ProtParam

program (http://web.expasy.org/protparam/) (Gasteiger et al., 2005) and CARD

(https://card.mcmaster.ca/), respectively.

3.3.9. Targeted Proteins Structure Prediction

The antigenic, virulent and non-allergenic epitopes were then subjected to an exo-membrane

topology phase, where the surface exposure of the epitopes on targeted proteins was visualized for

immunological applications. This was achieved through use of the Pepitope server (Mayrose et

al., 2007). Prior to Pepitope analysis, the protein structures were modelled through a comparative

structure prediction approach (Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam, 2018). In this

approach, the structure of the target proteins are first predicted by different servers: Phyre2 (Kelley,

Mezulis, Yates, Wass, & Sternberg, 2015), I-Tasser (Y. Zhang, 2008), Modweb (Pieper et al.,

2006), RaptorX (Kallberg et al., 2012) and Swiss-Model (Schwede, Kopp, Guex, & Peitsch, 2003).

The best model from each server then underwent a structure evaluation phase. In this phase, the

proteins with most of their residues mapped in the favored region of the Ramachandran plot were

selected for energy minimization (Skariyachan, Manjunath, & Bachappanavar, 2018). Energy

minimization was essential to relax the protein structures and remove any steric clashes.

Minimization was carried out for total of 1500 steps: 750 steps using the steepest descent algorithm

and using 750 the conjugate gradient method with a step size of 0.02 Å under the Tripos Force

Field (TFF) (S. Ahmad, Raza, Uddin, & Azam, 2017).

3.3.10. Predicting Proteins with a Strong Interactome

The work was extended further to select therapeutic proteins that have strong cellular interactions

(Rashid, Naz, Ali, & Andleeb, 2017). Interaction analysis of physiochemically suitable proteins

was executed through Search Tool for the Retrieval of Interacting Genes/Proteins (STRING)

(Szklarczyk et al., 2014). Only interactions that have very high confidence score (>0.9) were

considered.

http://web.expasy.org/protparam/)

https://card.mcmaster.ca/

71

1

Fig.3.1. The complete hierarchy of steps applied in the current study.2

72

3.3.11. Multi-epitope Vaccine Sequence Construction

The multi-epitope sequence was constructed by linking the B-cell derived T-cell epitopes predicted

in previous steps through a GPGPG linker, while the Cholera toxin B subunit (UniProtKB -

E9RIX3 (E9RIX3_VIBCL)) as adjuvant was linked to the construct using an EAAAK linker

(Saadi, Karkhah, & Nouri, 2017). The antigenicity and allergenicity of the entire construct were

re-validated through VaxiJen and Allertope, respectively. The secondary and tertiary structure of

the construct were predicted through PSIPRED (McGuffin, Bryson, & Jones, 2000) and RaptorX

(Kallberg et al., 2012), respectively. Intrinsic disorders in the vaccine-construct were predicted

using protein disorder prediction system (PrDOS) (Ishida & Kinoshita, 2007). The tertiary

structure of the multi-epitope vaccine-construct was further refined through use of the Galaxy

Refine server (Heo, Park, & Seok, 2013). Finally, to further increase the stability of the multi-

epitope sequence construct, it was subjected to Disulfide engineering utilizing Design 2 server

(Craig & Dombkowski, 2013).

3.3.12. Molecular Docking

Molecular docking of the multi-epitope vaccine sequence was done with the receptor TRL4

(Pandey, Bhatt, & Prajapati, 2018). The 3D structure of TRL4 was retrieved from the PDB using

PDB ID of 4G8A (Ohto, Yamakawa, Akashi-Takamura, Miyake, & Shimizu, 2012). Molecular

docking was performed with PatchDock (Schneidman-Duhovny, Inbar, Nussinov, & Wolfson,

2005) and subsequently refined using Fast Interaction Refinement in Molecular Docking

(FireDock) (Mashiach, Schneidman-Duhovny, Andrusier, Nussinov, & Wolfson, 2008). The top

complex was then minimized using the same parameters discussed above.

3.3.13. MD Simulations and Binding Free Energy Calculations

The stability of multi-epitope vaccine-construct with the TLR4 receptor was assessed by

subjecting the complex to 100-ns of MD simulation carried out using the AMBER 16 package

(Pearlman et al., 1995). System preparation was initiated with the tleap tool while energies were

calculated using the MPI version of SANDER (J. Wang, Wolf, Caldwell, Kollman, & Case, 2004)

using the ff03.r1 and GAF) forcefields (Ozpinar, Peukert, & Clark, 2010). The system was solvated

in a water box made up of TIP3P water molecules, with a padding distance of 12 Å between the

protein and the edge of the box. Complex neutralization was achieved by adding 28 Na+ ions. The

73

complex was then minimized in several stages. Firstly, minimization was carried out with a

restraint of force constant k =200 kcal/mol Å2 applied to all hydrogen atoms and water molecules

for 500 and 1000 cycles, respectively, to remove steric clashes. Next, 1000 steps of minimization

were carried out with a restraint of force constant k = 5 kcal/mol Å2 applied to C-alpha atoms. In

the final stage of minimization, 300 steps of minimization were performed with a restraint of force

constant k = 100 kcal/mol Å2 applied to non-heavy atoms. With restraints of 5 kcal/mol Å2 on C-

alpha atoms, the system was then heated to 300 K for 20 picoseconds (ps). Langevin dynamics

(Paterlini & Ferguson, 1998) was used to maintain system temperature and SHAKE algorithm

(Kräutler, Van Gunsteren, & Hünenberger, 2001) was applied to bonds involving hydrogen. The

constant-temperature, constant-pressure ensemble (NPT) (Uline & Corti, 2013) was applied for

50-ps with restraints of 5 kcal/mol Å2 on C-alpha atoms to allow the pressure of the system to

equilibrate. The system was then equilibrated for 1-ns. A production run of 100-ns was lastly

performed using NVT ensemble (S. Ahmad, Raza, Abbasi, & Azam, 2018) in combination with

Berendsen’s algorithm (Berendsen, Postma, van Gunsteren, DiNola, & Haak, 1984)and a time step

of 2-fs. Deviations and fluctuations in the protein backbone were monitored throughout the

simulation e.g. Root Mean Square Deviation (RMSD), Root Mean Square Fluctuation (RMSF),

and Rg (Haq, Abro, Raza, Liedl, & Azam, 2017). The complex was visualized at different stages

of the simulation using Visual Molecular Dynamics (VMD) (Humphrey, Dalke, & Schulten, 1996)

and University Of California At San Francisco Chimera (UCSF) Chimera (Pettersen et al., 2004).

The strength of interactions between the vaccine-construct and the TLR4 receptor was assessed

using Amber16 MM/GBSA method (Miller et al., 2012). The overall objective of MM/GBSA is

to calculate the free energy difference between the vaccine-construct-TLR4 complex and TLR4

(the receptor) and vaccine-construct (the ligand). Mathematically, it can be represented as:

ΔGBinding energy = ΔGTLR4-vaccine-construct complex − [ΔGTLR4 receptor – ΔGvaccine-construct]……………...(V)

In the equation, ΔG is Gibb’s free energy and is calculated by the MM/GBSA method for the

following terms,

ΔG = Egas + ΔGsolvation – TSsolute ………………………………………………………………(VI)

In the equation, T stands for temperature and S is the entropy contribution to vaccine binding. The

gas energy is the MM energy and involves internal, electrostatic and van der Waals energies,

74

Egas = EInternal energy + EElectrostatic energy + EVan der Waals energy………………………………………(VII)

The term ΔG solvation is calculated using an implicit system and can be split into electrostatic and

non-polar contributions,

ΔGsolvation = ΔGelectrostatic + ΔGnp………………………………………………………………(VIII)

ΔGele involves electrostatic and polar solvation component and can be presented as,

ΔGelectrostatic = Eelectrostatic + ΔGGB………………………………………………………………………………………………(IX)

ΔGnp is directly proportional to the molecule solvent-accessible surface area,

ΔGnp = γSAS + β……………………………………………………………………………...(X)

Estimation of ΔGnp is achieved using the Linear Combinations of Pairwise overlaps (LCPO)

(Weiser, Shenkin, & Still, 1999) approach with the water probe radius set to 1.4 Å. In MM/GBSA

calculations, the standard values of γ and β were 0.0072 kcal/mol·Å2 and 0 kcal/mol, respectively.

3.4. Results and Discussion

3.4.1. Proteome Retrieval and Subtractive Proteomics

Vaccination is considered an alternative and effective stratagem to prevent infection caused by

drug resistant bacterial pathogens (Jansen, Knirsch, & Anderson, 2018). In the last few years, a

substantial amount of progress has been made in the development of effective vaccines against

multiple and extensive drug resistant A. baumannii isolates (Moriel et al., 2013; Chiang et al.,

2015; W. Chen, 2015; R. Singh, Garg, Shukla, Capalash, & Sharma, 2016; Ni, Chen, Ong, & He,

2017; S. Ahmad & Azam, 2018). For active and passive immunization, different vaccines based

on bacterial cell or protein antigens have been evaluated. In experimental approaches, promising

antigenic-specific antibody responses were observed with varying degree of protection in animal

models. However, technical and scientific hurdles remain in the development of an effective

vaccine for A. baumannii (Hassan et al., 2016).

Recent studies showed that pure antigens can elicit a faster immune response (Weidang Li, Joshi,

Singhania, Ramsey, & Murthy, 2014; Donev, 2015) and thus the genome of A. baumannii was

rigorously explored using in silico tools to reveal more potent ones. Subunit vaccine proteins are

thought to have potential antigenic determinants: however, their role in immune protection is still

75

questioned and further evidence is required. The appropriate combination of these shortlisted

antigens is another consideration as individual antigenic peptides have several limitations; mainly

they do not always stimulate strong and accurate immune responses (Weidang Li, Joshi, Singhania,

Ramsey, & Murthy, 2014). Additionally, it has been observed that most of the identified antigens

are immune-dominant and often overshadow the more effective one present in the pathogen

proteome (Weidang Li, Joshi, Singhania, Ramsey, & Murthy, 2014). Generally, an ideal

immunotherapy antigen against TRAB should offer several important qualities. Among these, the

most critical is being protective to the host (Weidang Li, Joshi, Singhania, Ramsey, & Murthy,

2014; Hassan et al., 2016; T. A. Ahmad, Tawfik, Sheweita, Haroun, & El-Sayed, 2016). Surface

topology, high reproducibility and cost are the other important factors (Naz et al., 2015). Proteins

that remain stable and soluble in recombinant form after preparation are desirable (Barh et al.,

2013). Proteins that create a balance between pro-inflammatory and anti-inflammatory cytokines

(essential for the development of a well-regulated immune response) and reduce the incidence of

heat shock are the most appropriate candidates (Venkatesha, Dudics, Acharya, & Moudgil, 2015).

Similarly, proteins that induce strong humoral immune response while maintaining moderate

cellular immunity are also preferable (Weidang Li, Joshi, Singhania, Ramsey, & Murthy, 2014).

The protection of immune-compromised patients also needs to be ensured and the antigen

candidate must be suitable for active and passive immunization (T. A. Ahmad, Tawfik, Sheweita,

Haroun, & El-Sayed, 2016). Since limited antigen candidates are available for a TRAB vaccine

compared to other bacterial pathogens (Hassan et al., 2016), the search for novel antigens is

imperative.

The complete proteome of TRAB strain AB031 encompasses 3,438 proteins. The CD-HIT

analysis revealed 3,354 proteins as non-redundant proteins, while 84 proteins were classified

redundant based on the sequence similarity of 60%. Redundant proteins are paralogous in function

and arise due to duplication during evolution. Because of multiplication, a defect or mutation in a

redundant protein will have a reduced adverse effect on overall fitness of the organism and as such,

they are not considered attractive targets for a vaccine design (Xie, Zhang, Zhao, Peng, & Zheng,

2018). Furthermore, redundant proteins are not well conserved across the bacterial species and

strains and hence cannot be targeted by broad-spectrum therapeutics (Sanober, Ahmad, & Azam,

2017). In contrast, orthologous proteins are non-redundant and are considered more conserved

across bacterial species and strains, therefore, could be attractive candidates for vaccine design

76

(Sanober, Ahmad, & Azam, 2017). In the next step of the framework, the non-redundant proteins

were aligned with the reference human protein. It was revealed that 556 proteins are human

homologs, while 2798 proteins are non-human homologs. Eliminating host homologous proteins

is essential as targeting such proteins could result in strong immune responses of the host. In

addition, host homologous proteins can integrate and recombine in the host genome, and therefore,

are avoided for vaccine development (Sanober, Ahmad, & Azam, 2017; Asad, Ahmad,

Rungrotmongkol, Ranaghan, & Azam, 2018). The identification of essential, non-homologous

proteins was crucial because the majority of vaccines target essential cellular process of the

pathogenic organism. The essential proteome is a set of proteins prerequisite to maintain cellular

life and as such has wider therapeutic potential (Baseer, Ahmad, Ranaghan, & Azam, 2017). In

TRAB, essential proteins are involved in major biological pathways and aid survival in extreme

environmental stress conditions, infection and persistence in the host, GTP, ATP and nucleotide

binding, ligase, phosphatase and transferase activity. The use of essential proteins in vaccine

development will ensure the continuous expression of antigenic determinants and thus will evoke

the host immune response when it encounters the pathogen (Hassan et al., 2016). DEG analysis

revealed 800 proteins as essential and removed 1998 proteins from further analysis as they were

classed as non-essential.

3.4.2. TRAB Exo-proteome and Secretome

Knowledge of protein subcellular localization is important and can significantly improve the

identification of potential vaccine candidates (Hassan et al., 2016). Secreted and membrane

proteins are easily accessible to drug molecules because of their localization in the extra-cellular

space or on the organism’s cell surface (Grandi, 2010) and are of great interest due to their potential

as vaccine or diagnostic targets (Naz et al., 2015). Similarly, targeting such proteins is also useful

as they aid pathogen adherence, invasion, survival and proliferation within the host cell (Naz et

al., 2015;Hassan et al., 2016). In a comparative subcellular localization approach, proteins located

in the extracellular matrix and outer membrane of the pathogen were identified first through

PSORTb v 3.0. The majority of proteins were cytoplasmic (577), 181 were located in the inner

membrane, 13 in the outer membrane, 9 were periplasmic, 1 extracellular and 19 could not be

classified. The 13 identified outer membrane proteins and 1 extracellular protein from the

search by PSORTb were used in CELLO and CELL2GO to cross-validate the protein’s

77

localization. According to CELLO, 2 of the proteins were categorized as extracellular while 12

were outer membrane. CELL2GO identified 6 proteins as outer membrane while the remaining

8 were predicted to have dual localization of extracellular and outer membrane. These

shortlisted 14 proteins are now non-redundant, host non-homologous, pathogen essential and

outer membrane and extracellular proteins and illustrated in Table 3.1. The Venn diagram for

number of proteins shortlisted so far is shown in Fig.3.2.

Fig.3.2. The shortlisted 14 proteins for virulent protein analysis. C.P., Complete proteome, N.

R.P., Non-redundant proteome, N.H.P., Non-homologous proteome, E.P., Essential

proteome, O.M., Outer membrane, E.C., Extracellular.

78

Table 3.1. Fourteen proteins with subcellular localization of outer membrane and extracellular

matrix.

Protein ID Protein PSORTb CELLO CELL2GO

WP_000505931.1

Methyl-

accepting

chemotaxis

protein (MACP)

Outer

Membrane

Outer

Membrane

Extracellular,

Outer

Membrane

WP_000557460.1 M23 family

peptidase

Outer

Membrane Extracellular

Extracellular,

Outer

Membrane

WP_001109856.1 Hypothetical

protein

Outer

Membrane

Outer

Membrane

Extracellular,

Outer

Membrane

WP_001174793.1

Multidrug efflux

RND transporter

AdeIJK outer

membrane

channel subunit

(AdeK)

Outer

Membrane

Outer

Membrane

Outer

Membrane


protein Extracellular

Outer

Membrane

Outer

Membrane

WP_005128115.1

Peptidoglycan-

associated

lipoprotein (Pal)

Outer

Membrane Extracellular

Extracellular

Outer

Membrane

WP_025464534.1

Outer membrane

protein assembly

factor (BamA)

Outer

Membrane

Outer

Membrane

Outer

Membrane

WP_031999435.1

Outer membrane

protein assembly

factor (BamD)

Outer

Membrane

Outer

Membrane

Extracellular

Outer

Membrane

79

WP_031999452.1 Insulinase family

protein

Outer

Membrane

Outer

Membrane

Outer

Membrane

WP_031999870.1

Outer membrane

protein assembly

factor (FimD)

Outer

Membrane

Outer

Membrane

Extracellular

Outer

Membrane


protein

Outer

Membrane

Outer

Membrane

Extracellular

Outer

Membrane

WP_032000856.1

Fimbrial

biogenesis outer

membrane usher

protein (FimD)

Outer

Membrane

Outer

Membrane

Outer

Membrane

WP_038405564.1

TonB-dependent

siderophore

receptor

Outer

Membrane

Outer

Membrane

Outer

Membrane

WP_079378209.1

Type IV

secretion protein

(Rhs)

Outer

Membrane

Outer

Membrane

Extracellular

Outer

Membrane

3.4.3. TRAB Virulent Proteins

Virulence is a key factor of bacterial pathogens that facilitates organism pathogenesis as well

as survival in adverse environmental conditions (Sparling, 1983). Compared to other bacterial

pathogens, very little is known about TRAB’s pathogenic potential and virulence determinants

(Hassan et al., 2016). In order to get insights about the virulent potential of TRAB, virulent

proteins in the essential exo-proteome and secretome were explored. In total, six proteins were

identified as virulent. These include: methyl-accepting chemotaxis protein (MACP) (Bit score:

562), outer membrane protein assembly factor (BamA) (Bit score: 241), multi-drug efflux RND

transporter (RND) (Bit score: 285), TonB-dependent siderophore receptor (TonB) (Bit score:

219), type IV secretion protein Rhs (Rhs) (Bit score: 259) and fimbrial biogenesis outer

membrane usher protein (FimD) (Bit score: 591). MACP is a sensor protein that allows bacteria

80

to sense molecule concentration in the extracellular matrix and mediate smooth swim or tumble

accordingly (Kehry & Dahlquist, 1982). In the case of rising levels of nutrients or declining

levels of toxins, the bacteria will continue to swim smooth or swim forward (Moriel et al., 2013).

On the contrary, if the concentration of the nutrients is declined or the level of toxins is on the

rise, the bacteria reorient themselves or will tumble (Kehry & Dahlquist, 1982; Derr, Boder, &

Goulian, 2006). In this way, the bacteria may swim away from toxins or towards nutrients(Kehry

& Dahlquist, 1982; Derr, Boder, & Goulian, 2006). BamA is an important part of the outer

membrane and plays a vital role in assembly and insertion of beta-barrel proteins into the outer

membrane of Gram-negative bacteria (Albrecht et al., 2014). RND is Resistance-Nodulation-

Division family transporter, widespread among Gram-negative bacteria to actively efflux

chemotherapeutic agents including antibiotics (Nikaido & Takatsuka, 2009). TonB is a bacterial

outer membrane protein that binds and transports ferric chelates called siderophores, as well as

carbohydrates, nickel complexes and vitamin B12 (Noinaj, Guillier, Barnard, & Buchanan, 2010).

Similarly, this protein provides immunity to bacteria against antibiotics, detergents, digestive

enzymes and immune surveillance (Noinaj, Guillier, Barnard, & Buchanan, 2010). Rhs is a

member of the Type IV secretion proteins and plays role in the transportation of molecules

across the membrane (Koskiniemi et al., 2013). Lastly, FimD protein is involved in pilus

biogenesis in Gram-negative bacteria (Nuccio & Baumler, 2007).

3.4.4. Screening of Non-probiotic Bacterial Proteins

Probiotic bacteria are useful organisms that line the digestive tract and perform a significant role

in promoting host health and disease prevention (Nagpal et al., 2012). In the gut, these

microorganisms play an important role in preventing the growth of harmful bacterial species by

creating hydrolytic enzymes, training the host immune system by stimulating the secretion of IgA

and regulatory T-cells, producing hormones that store body fats, producing vitamins for the host

such as biotin, butyrate and vitamin K2, and regulating the development of the gut (Kechagia et

al., 2013). Accidental inhibition of these microbes could lead to severe side effects; therefore, it is

imperative to discard proteins of such microbes from the design framework (Wadood et al., 2018).

All the proteins were reported to possess no homology with the shortlisted proteins and hence are

safe to be used in vaccine for targeted immune responses.

81

3.4.5. Screening of Non-similar Mouse Proteins

In this stage, proteins that were identical to mouse proteins were removed. This was essential

as evaluating immune protection efficacy of proteins similar between the bacteria and mouse

can lead to autoimmune diseases thus limiting experimental evaluations and producing false

positive hits (He, Xiang, & Mobley, 2010). In addition, the proteins are less likely to be

immunogenic because of epitope mimicry. All the six proteins were found non-homologous to

mouse proteome and thus analyzed further in the framework.

3.4.6. Physiochemical Prioritization of Vaccine Proteins

The most crucial factor during this phase was the estimation of the protein’s molecular weight

(Hassan et al., 2016; Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam, 2018). Proteins having

molecular weight < 110 kDa are considered more appropriate as they can easily be purified and

successfully subjected to vaccine development (Baseer, Ahmad, Ranaghan, & Azam, 2017).

Molecular weights of the proteins were estimated as: Rhs (180.35 kDa), FimD (94.7 kDa),

BamA (92.27 kDa), TonB (80.84 kDa), MACP (75.38 kDa) RND (52.82 kDa). Despite its

molecular weight falling outside the desired range, the Rhs protein was not discarded at this stage

as this protein has not previously been explored and it contains potential antigenic epitopes. The

theoretical pI for MACP, BamA, FimD, Rhs and TonB was found in the acidic category with

range 4.93 to 5.0. The RND was predicted to be acidic with a theoretical pI of 9.02 (Azam &

Shamim, 2014). The Grand average of hydropathicity (GRAVY) index score for all the six

proteins was negative indicating their hydrophilic nature (Kyte & Doolittle, 1982). The stability

index computed for the proteins was found to be < 40, thus considered as stable (Guruprasad,

Reddy, & Pandit, 1990). The aliphatic index was used to indicate the thermostability of the

proteins. The average aliphatic index value for the proteins is 83.96, indicating high

thermostability. The extinction coefficient i.e. the light amount observed by a protein at

wavelength of 280 nm in water (Gill & Von Hippel, 1989), is a useful parameter to indicate the

ease of protein purification. Two kinds of values were noted: one considered all cysteine values as

half cystines, while the other considered no cysteine as half cystines. Cystine is an amino acid

formed as a result of two cysteine joined together by a disulphide bond. The extinction coefficient

for the proteins is shown in S-Table 3.1. Further, all the six proteins were analyzed for number

of trans-membrane helices. Proteins having trans-membrane helices < 2 were selected as they

82

are easy to clone and express during experimental evaluation (Hassan et al., 2016; Baseer,

Ahmad, Ranaghan, & Azam, 2017; Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam, 2018).

All the proteins were found to have < 2 trans-membrane helices: MACP (2), BamA (0), RND

(0), TonB (0), Rhs (0) and FimD (0).

3.4.7. Antigenicity Prediction

All the six proteins were found antigenic: MACP (0.44), BamA (0.61), RND (0.63), TonB (0.60),

Rhs (0.53) and FimD (0.57).

3.4.8. Prediction of B-cell derived T-cell Epitopes

Filtration of protein candidates suitable for the production of antigenic peptides could not only

help in minimizing the time, resource and labor but also at the same time optimize the chances of

success of getting the most suitable vaccine for the pathogen (Hassan et al., 2016). The first step

in this process was to predict B-cell epitopes for each protein. The predictions were: 31 for Rhs,

23 for FimD, 15 for BamA, 14 for TonB, 6 for MACP, and 6 for RND. Only B-cell epitopes

with a B-cell epitope prediction score of > 0.8 were investigated for surface exposure as

epitopes located near the surface are easily recognized by immune cells and can be efficiently

targeted (Weidang Li, Joshi, Singhania, Ramsey, & Murthy, 2014; Hassan et al., 2016). A single

exo-membrane B-cell epitope was identified for MACP, but more were predicted for the others:

4 for RND, 6 for BamA, 9 for FimD, 9 for TonB and 13 for Rhs. Surface exposed B-cell epitopes

were then analyzed for binding alleles of MHC-I and MHC-II. Only those epitopes were

considered for which common binding alleles in both classes were predicted (S. Ahmad &

Azam, 2018). MACP was excluded from further analysis at this stage as only MHC-I epitopes

were predicted. The other epitopes contained several epitopes common to both MHC classes

[BamA 3, RND 4, TonB 5, FimD 6, and Rhs 7]. The next stage of screening involved the

prediction of the antigenicity of the epitopes. All 4 epitopes identified for the RND transporter

protein were found to be non-antigenic, excluding this target from further analysis. For BamA

all 3 epitopes were found to be antigenic. For FimD 3 out of 6 of the epitopes were antigenic

and 4 out of 5 were for TonB. Rhs had the most antigenic epitopes with 4 out of 7 found to be

antigenic. The affinity of the screened epitopes for the DRB*0101 (highly prevalent allele in

the human population) was determined. This was vital as epitopes having higher affinity for

DRB*0101 produce robust and accurate immune responses. Only epitopes with an affinity <

83

100 nM were considered. For outer membrane protein assembly factor BamA, out of 3 antigenic

epitopes only 1 was found with an IC50 value in the correct range (FPLNDKPGD IC50 = 91.62

nM). For FimD, 2 antigenic B-cell derived T-cell epitopes: FVHAEEAAA and MVLVHAPDA,

were found to have higher affinity for the DRB*0101 allele with IC50 values of 3.52 and 17.42

nM, respectively. Two of the antigenic epitopes for TonB had acceptable IC50 values:

YQLLDGQEV (IC50 = 26.36 nM) and YNSDGQRSR (IC50 = 10.16 nM). For Rhs, three

epitopes: YVVAGTAAA (IC50 = 4.90 nM), YYYFDLNGF (IC50 = 0.92 nM) and YEGITIQPL

(IC50 =11.02 nM) were selected. The virulent potential of these shortlisted epitopes was another

significant consideration as relates to ability to evoke a response from the host immune system

(Naz et al., 2015). All the epitopes of the remaining vaccine candidates: BamA, FimD, TonB

and Rhs had a virulent potential above the threshold (> 0.5). Lastly, in this phase, only non-

allergenic epitopes were considered to avoid any side effects of the epitopes that will be used

for multi-epitope vaccine construction (Chung, 2014). For BamA, “FPLNDKPGD” was found

non-allergenic therefore considered. For FimD protein, epitope “FVHAEEAAA” was found

non-allergenic, while all the screened epitopes of TonB were allergenic, removing it from

further processing in the pipeline. For Rhs, only epitope “YVVAGTAAA” was a su itable

candidate. Lastly, we found all the predicted epitopes as stable and do not have any resistant

sequence. The predicted B-cell derived T-cell epitopes for the potential three vaccine proteins

are illustrated in Table 3.2.

Table 3.2. Predicted B-cell derived T-cell epitopes for potential three vaccine proteins.

Protein B-cell epitope B-cell derived

T-cell epitope

Antigenicity MHCpred

IC50

score

Virulent

Pred

Allergenicity

BamA AFPLNDKPGDETKEIQFEIG FPLNDKPGD

1.4 91.6 1.0 Non-allergen

FimD TMPVFVHAEEAAASAPVEAE FVHAEEAAA


Rhs AASVAGYGPYVVAGTAAAGS YVVAGTAAA


84

3.4.9. Structure Prediction and Evaluation of Shortlisted Epitopes Proteins

The exo-membrane topology of the shortlisted epitopes on their respective protein surfaces was

visualized. Before this, structures of the shortlisted proteins were predicted and optimal

structures selected. Generally, structures predicted by RaptorX were selected. For BamA, the

structure modelled has majority of its residues (92.0%) are plotted in the most favorable region.

The residues in the additionally allowed region, generously allowed region, and disallowed

region are 7.2%, 0.7% and 0.1%, respectively. The overall G-factor for the protein is −0.12.

Additionally, the Errat, Verify-3D and Prosa Z-score for the protein are 53.01, 68.42 and −8.45,

respectively. Similarly, structure of FimD was analyzed to have 88.1% residues in the most

favorable region of the Ramachandran plot, while residues in the disallowed region were 0.8%.

The overall G-factor score for the protein is −0.15. The ProsA Z-score for the model was highly

acceptable i.e. −9.49, while the Errat and Verify-3D scores were 61.84 and 77.70, respectively.

In case of Rhs, the structure has 83.6% of residues in most favorable region and 1.8% in the

disallowed. The G-factor for the structure is -0.21. Errat, Verify-3D and Prosa Z-score for the

structure are 77.01, 52.06 and -1.44, respectively. The predicted structures for the three proteins

are shown in Fig.3.3, while the exo-membrane topology of the predicted epitopes are shown in

Fig.3.4.

Fig.3.3. Tertiary structure of the shortlisted potential vaccine proteins. From left to right,

BamA, FimD and Rhs.

85

Fig.3.4. Exo-membrane topology of the shortlisted antigenic epitopes on their respective

protein surface.

3.4.10. Interacting Network Analysis

Interacting network analysis was carried out for BamA, FimD and Rhs that contain desired

antigenic B-cell derived T-cell epitopes. This interacting network analysis allows the interpretation

of direct and indirect interactions between the mentioned proteins and other proteins of the bacteria

that mediate cellular process and metabolic/signaling pathways (Hassan et al., 2016). No direct

interactions among the proteins were observed. Individual clusters were constructed for BamA

(Fig.3.5A) and FimD (Fig.3.5B), no cluster could be produced for Rhs as no similar sequences

were foundFor BamA using stringent interaction criteria of 0.9, a cluster of 8 proteins directly

interacting with BamA protein was identified. Of these 8 proteins, BamB, BamD and BamE

together with BamA form the BAM complex and catalyze folding and insertion of β-barrel proteins

into the Gram-negative bacteria outer membrane (Han et al., 2016). SurA is a chaperone involved

in maturation and efficient production of several outer membrane proteins (Soltes, Schwalm, Ricci,

& Silhavy, 2016). Rsep is a zinc metalloprotease enzyme (Koide, Ito, & Akiyama, 2008). FabZ is

an enzyme of unsaturated fatty acids biosynthesis and catalyzes the dehydration of long and short

chain, saturated and unsaturated beta-hydroxyacyl-ACPs (Zhu et al., 2009). LpxD is an enzyme

86

of the Lpx family and is involved in the synthesis of lipid A molecules that make up part of the

Gram-negative outer membrane (Buetow, Smith, Dawson, Fyffe, & Hunter, 2007). Peptidoglycan

associated protein (Pal) is an outer membrane protein essential for survival and pathogenesis of

bacteria (Parsons, Lin, & Orban, 2006). Three proteins were identified in the interacting network

analysis of FimD: F911_01973 (fimbrial protein), HMPREF0010_00598 (Gram-negative pili

assembly chaperone domain protein) and F911_02226 (Gram-negative pili assembly chaperone

domain protein) (Nuccio & Baumler, 2007).

Fig.3.5. Interacting network of (A) BamA and (B) FimD. Empty balls are proteins with unknown

3D structure while filled balls are proteins with known 3D structure.

3.4.11. Secondary and Tertiary Structure of Construct

The secondary structure of the vaccine-construct contains 38% of alpha helix, 19% of beta strand

and 44% of coil structure as shown in Fig.3.6. RaptorX was used to predict 5 tertiary structures

for the vaccine-constructs which were then assessed by Ramachandran analysis to select the best

model (Table 3.3). Model 5 was selected as it had the most favorable positioning or residues in

allowed regions. The G value of 0.74 and the ProA Z-score of −5.76 also indicate that this structure

is energetically stable. The Errat quality factor and Verify-3D score for the model were 84.81 and

77.71, respectively. The tertiary structure, ProsA Z-score and Ramachandran plot of the most

optimal multi-epitope peptide structure are illustrated in Fig.3.7.

87

Fig.3.6. Secondary structure of the multi-epitope peptide vaccine.

88

Table 3.3. Structural evaluation of predicted five models of the multi-epitope peptide vaccine.

Model

Most favored

regions [A,B,L]

Additional allowed

regions [a,b,l,p]

Generously allowed

regions [~a,~b,~l,~p]

Disallowed regions

[XX]

OVERALL G-Factor

(AVERAGE

−0.71*)

Model-1 93 (65.5%)

41 (28.9%)

5 (3.5%)

3 (2.1%)

−0.71

Model-2 85(59.9%)

36 (25.4%)

15 (10.6%)

6 (4.2%)

−1.52

Model-3 85 (59.9%)

39 (27.5%)

9 (6.3%)

9 (6.3%)

−1.32

Model-4 84 (59.2%)

47 (33.1%)

8 (5.6%)

3 (2.1%)

−0.98

Model-5 94 (66.2%)

38 (26.8%)

8 (5.6%)

2 (1.4%)

−0.74

89

Fig.3.7. (A) Tertiary structure (B) ProsA Z-score and (C) Ramachandran plot of the multi-

epitope peptide vaccine.

90

3.4.12. Intrinsic Disorder Regions Prediction

Intrinsically disordered regions of proteins are those regions that lack an ordered 3D structure,

commonly found in linker or loop regions (Babu, 2016). It has been recognized that these

disordered regions allow the vaccine-construct the flexibility to adopt different conformations

required for binding to the catalytic binding pocket of the receptor protein. Fig.3.8 infers that

majority of the vaccine-construct residues are well below the threshold, indicating that the majority

of the vaccine-construct residues lie in ordered regions. Two regions: one from the adjuvant (Met1-

Leu4) and the second from the B-cell derived T-cell epitope region (Pro131-Ala166) are present

in disordered regions. These regions are important for the proper adjustment of the construct in the

receptor pocket.

Fig.3.8. Intrinsic disorder graph predicting that the majority of the multi-epitope peptide regions

are below the threshold value.

3.4.13. Tertiary Structure Refinement and Disulfide Engineering

Refinement of the construct structure was vital to reconstruct side chains, followed by repacking

MD simulations to relax the modelled structure (Heo, Park, & Seok, 2013). The Galaxy refine

server is one of the best tools for the refinement of both the global and local structural quality of

91

protein models (Pandey, Bhatt, & Prajapati, 2018). Disulfide engineering helps to increase protein

stability by reducing the conformational entropy of the unfolded state of the protein (Craig &

Dombkowski, 2013; Pandey, Bhatt, & Prajapati, 2018). The GalaxyRefine server generated five

refined models for the vaccine-construct. Model-1 was the best model according to several

parameters including: GDT-HA (0.9292), MolProbity (2.538), RMSD (0.473), Poor rotamers

(1.6), Clash score (25.1), and Ramachandran favored plot (91.5). In Disulphide engineering, 8

pairs of residues in the construct were observed that violate the recommended values of energy

(cut-off < 2.2) and Chi3 should be in between −87 to + 97. These residues include: Lys5, Gly7-

Val8, Thr11, Ser15-Ala17, His34, Tyr48, Ala59, Gln70, His78, Leu98, Cys107, Trp109 and

Ala151.

3.4.14. Codon Optimization of the Vaccine-construct

The codon usage of the final vaccine-construct in E.coli K 12 strain was optimized to assure

maximum protein expression (Khatoon, Pandey, & Prajapati, 2017). The optimal GC content

should lie in the range 30% to 70% (Khatoon, Pandey, & Prajapati, 2017). The average GC content

of the improved sequence was 50.80 %, indicating that good expression of the vaccine protein was

likely in the E.coli host. The Codon Adaptation Index (CAI) of the optimized sequence is 0.92.

The recombinant clone with inserted adopted codon sequence is shown in Fig.3.9.

3.4.15. Molecular Docking of the Vaccine-construct with the TLR4 Immune Receptor

In total 189 binding pockets were identified spanning the amino acids from all the four chains: A,

B, C and D. Details of the best binding pocket is given in Fig.3.10 and S-Table 3.2. The surface

volume of the pocket was 13287.162 Å3, surface area 8770.908 Å2, molecular area of the cavity

mouth was 11028.235 Å2 and molecular circumference was 2563 Å2. Molecular docking generated

10 models ranked according to electrostatic complementarity and geometry of the protein surface

(Table 3.4). The molecular docking solutions were further refined (Table 3.5) based on the Global

energy, the lower the Global binding energy the more stable the complex and vice versa. Two

solutions were found that have lower Global binding energies: Solution 4 and 6.

92

Fig.3.9. In silico prediction of the cloning of the final multi-epitope peptide vaccine-construct

(red) into pET28a expression vector.

Fig.3.10. The best predicted binding side of the TLR4 receptor protein. The red spheres indicate

the active site region of the protein.

93

Solution 6 was ranked top complex based on the lowest Global binding energy of −8.16 kcal/mol.

The attractive and repulsive van der Waals interactions for the complex were −18.30 kcal/mol and

22.83 kcal/mol, respectively. The atomic contact energy for the complex is 5.96 kcal/mol and

hydrogen bonding energy of −2.42 kcal/mol. Visualization of the top complex showed binding of

the vaccine-construct within the binding pocket formed by chain A, B and C. The vaccine-

construct was seen docked deeply inside the pocket with helical region (Pro1509-Tyr1519) from

the Cholera toxin B subunit sequence facing towards the wall of chain A and B of TLR4. The

epitope from FimD was used to support the construct on chain C. The binding mode and

interactions of the multi-epitope vaccine-construct at the binding site of TLR4 receptor are

illustrated in Fig.3.11.

Table 3.4. Top 10 generated models by PatchDocK.

Solution

Number Score Area

Atomic Contact

Energy (ACE)

(kcal/mol)

1 21294 3038.8 37.0

2 18344 2662.0 −137.0

3 18042 3391.8 252.7

4 17776 2436.2 403.8

5 17760 2925.7 406.9

6 17456 3335.0 −20.0

7 17290 2123.2 234.9

8 16930 2561.4 380.8

9 16874 2104.0 415.4

10 16834 2471.8 −390.7

94

Table 3.5. FireDock refinement of PatchDocK models.

Rank Solution

Number

Global Energy

(kcal/mol)

Attractive

VDW

Repulsive

VDW

ACE

(kcal/

mol)

HB

1 6 −8.1 −18.3 22.8 5.9 −2.4

2 4 −3.2 −39.4 20.9 19.0 −2.2

3 8 6.9 −26.8 21.8 8.1 −2.9

4 2 11.0 −0.5 0.0 0.0 0.0

5 7 21.3 −56.5 77.5 5.9 −5.9

6 9 69.5 −55.2 158.3 14.0 −4.5

7 10 266.8 −54.8 442.1 −1.2 −6.5

8 1 695.9 −63.8 924.1 −0.2 −7.3

9 5 4056.5 −95.4 5145.7 18.6 −10.2

10 3 9406.4 −77.5 11829.9 7.4 −12.9

VDW, van der Waals forces, ACE, Atomic contact energy, HB, Hydrogen bonding

95

Fig.3.11. Binding conformation of the multi-epitope peptide construct at the binding site of Chain A and C (A), Surface view of the binding pose (B),

Closer view of the binding pose, (D) Binding interactions between multi-epitope peptide construct and TLR4.

96

3.4.16. MD Simulations and MM/GBSA based Binding Free Energy Calculations

The dynamic behavior of the vaccine-construct and TLR4 complex was studied through MD

simulations.Three physical parameters were evaluated shown in S-Fig.3.1. The atomic

displacement of the Cα atoms for the complex was explored first by calculating the RMSD as a

function of time (Haq, Abro, Raza, Liedl, & Azam, 2017). An average RMSD value calculated for

the complex is 4.85 Å with maximum RMSD observed at time scale from 90-97 ns. The trajectory

was visually inspected to explain the pattern of RMSD variations. Snapshots at different

nanoseconds: 8th, 18th, 29th, 43th, 49th, 72th, 81th and 90th ns were extracted from the trajectories. All

these snapshots were superimposed in Chimera to investigate the structural variations and decide

whether these variations are vaccine-construct induced or part of the catalytic mechanism of TLR4.

Generally, all the variations observed were due to the orientation of the vaccine-construct in the

pocket of the TLR4 resulting in the rise of RMSD. The rise of RMSD throughout the simulation

can be explained by the relaxation of the TLR4 chains in response to the presence of the vaccine-

construct in the binding pocket. The vaccine-construct orientations in the binding cavity of TLR4

are shown in Fig.3.12. The structural mobility of the complex was investigated by plotting the

RMSF as a function of residue number. The average RMSF calculated for the system is 2.48 Å,

with the first 594 residues having RMSF values around 3 Å. Higher fluctuations are seen in the

other residues. Rg in general measures the overall packing quality and density of a macromolecular

system ( Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Haq, Abro, Raza, Liedl, & Azam, 2017).

Variation in the pattern of Rg generally indicates changes in the compactness of the protein

structure. Higher the Rg indicate loose packing of the system and low values indicate compact

structures (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Haq, Abro, Raza, Liedl, & Azam, 2017).

The mean Rg value estimated for the system is 41.51 Å with highest Rg value observed at 47th ns

with value of 42.23 Å. It was noticed that variation in the Rg value is due to the multi-epitope

vaccine-construct induced variations in the protein structure. The protein structure adjustments

mainly were noticed at the loop regions and seem necessary to properly recognize the construct.

Hence, the variations observed in the Rg analysis complements that of RMSD.

97

Fig.3.12. Vaccine-construct adjustments at the active site of TLR4 receptor during the simulation period. The

multi-epitope peptide is in gold, while the TLR4 receptor is in dark magenta.

98

Molecular docking can be employed to predict the binding mode of ligands or to discriminate

between binders and non-binders (Ferreira, dos Santos, Oliva, & Andricopulo, 2015; Genheden &

Ryde, 2015), but they cannot really discriminate between ligands of very similar binding affinity

e.g. < 6 kJ/mol (G. Wang & Zhu, 2016). More accurate methods, known as alchemical

perturbation (AP) methods, exist that are based on statistical mechanics (Ferreira, dos Santos,

Oliva, & Andricopulo, 2015), but they require extensive Monte Carlo (MC) or MD simulations to

sample conformational space, limiting their widespread application (Ferreira, dos Santos, Oliva,

& Andricopulo, 2015). Between these two extremes, there are methods with intermediate

performance that do sampling at the end states, therefore, referred as end point methods (G. Wang

& Zhu, 2016). End point methods are generally more accurate than scoring functions and are less

computationally expensive than AP methods (C. Wang, Greene, Xiao, Qi, & Luo, 2018). The

arguably most popular end point method is MM/GBSA (Genheden & Ryde, 2015; G. Wang &

Zhu, 2016). This method was developed by Kollman et al. in the late 90s and has been used

successfully in wide range of settings including conformer stability, protein-protein interactions,

protein design and re-scoring (Kollman et al., 2000). The total MM/GBSA binding energy of the

complex was estimated to be −17.48 kcal/mol. The van der Waals energy (∆EVDW) reported for

the complex is −98.33 kcal/mol and dominates the overall energies of the complex. Non-favorable

contributions for the complex were observed for the electrostatic energy (∆Eele) that was 6.01

kcal/mol. The gas phase energy (∆Egas) for the complex was revealed favorable with value −92.31

kcal/mol. The polar solvation energy contribution towards the total binding energy is non-

favorable with score of 74.83 kcal/mol. The energy for polar component of solvation energy

(∆GGB) is 87 kcal/mol, which illustrates their non-favorable contribution in vaccine-construct

binding to TLR4 receptor binding site.

To the best of our knowledge, the present study reports the first work that describes the in silico

design of a multi-epitope vaccine for TRAB. The vaccine-construct is composed of antigenic

epitopes from three surface exposed proteins: BamA, FimD and Rhs. These proteins were

prioritized based on several important parameters that are vital in vaccine design (Naz et al.,

2015; Hassan et al., 2016; Asad, Ahmad, Rungrotmongkol, Ranaghan, & Azam, 2018). The BamA

protein plays an essential role in outer membrane protein biogenesis (Albrecht et al., 2014). Outer

membrane protein biogenesis is a vital process to maintain the indispensable functions of bacterial

cell envelope, which facilitates the transmembrane movement of nutrients and signaling molecules

99

and at the same time act as a protective physical barrier (Albrecht et al., 2014). Recently, in a

combined study of both in silico and experimental immunoprotective analysis revealed that

compared to previous studies, BamA as a potential candidate for vaccine designing against

infections caused by A. baumannii (R. Singh, Capalash, & Sharma, 2017). Immunization with

BamA conferred 80% of protection against lethal doses of the bacterium and produces high titer

of macrophages depended opsonophagocytic antibodies. FimD allows the assembly and secretion

of P pili fragments (Nuccio & Baumler, 2007). Pili proteins have been recognized previously to

form the core proteome of the A. baumannii by in silico studies and as such can be potential

vaccine candidates against the bacterium. A putative pilus assembly protein, FliF was first

identified as the most promising protein using in silico studies. Following experimental follow

up, the protein was found to elicit antibody titer of > 64,000 in mouse model with 50% survival

rate from the A. baumannii challenge (R. Singh, Garg, Shukla, Capalash, & Sharma, 2016). No

previous studies were found regarding the use of Rhs protein as vaccine candidate against A.

baumannii and therefore can be considered as a novel vaccine protein against the pathogen. The

pure antigens of these shortlisted proteins were formulated in multi-epitope vaccine in the

present study that may provide a best-adopted strategy against this rogue pathogen.

3.5. Conclusions

In the present study, using bioinformatics and immunoinformatics tools, we designed a multi-

epitope vaccine for the most troublesome bacterial pathogen, TRAB. The vaccine-construct

includes the safest and antigenic epitopes from three proteins: BamA, FimD and a novel Rhs

vaccine candidate. These proteins fulfilled all the requirements for potential subunit vaccine

candidates. The predicted B-cell derived T-cell epitopes for these proteins were used to design

a multi-epitope peptide vaccine to engage properly both cellular and humoral immunity and

enhance antigenicity of the individual peptides. Molecular docking and MD simulations show

the high stability of the complex of the construct with the active site of the TLR4 receptor. The

affinity of the construct for the TLR4 receptor was further validated through MM-GBSA

calculations. This multi-epitope peptide vaccine can potentially be used against TRAB

infections in both therapeutic and prophylactic situations.

3.6. Supplementary Files

S-Fig.3.1. A. RMSD, B. RMSF, C. Rg.

100

S-Table.3.1. Extinction coefficients for the shortlisted proteins

S-Table.3.2. Active site residues of TLR4.

3.7. References

Abbasi, S., Raza, S., Azam, S. S., Liedl, K. R., & Fuchs, J. E. (2016). Interaction mechanisms of

a melatonergic inhibitor in the melatonin synthesis pathway. Journal of Molecular Liquids,

221, 507–517.

Ahmad, S., & Azam, S. S. (2018). A novel approach of virulome based reverse vaccinology for

exploring and validating peptide-based vaccine candidates against the most troublesome

nosocomial pathogen: Acinetobacter baumannii. Journal of Molecular Graphics and

Modelling, 83, 1–11.

Ahmad, S., Raza, S., Abbasi, S. W., & Azam, S. S. (2018). Identification of natural inhibitors

against Acinetobacter baumannii D-alanine-D-alanine ligase enzyme: A multi-spectrum in

silico approach. Journal of Molecular Liquids, 262, 460–475.

Ahmad, S., Raza, S., Uddin, R., & Azam, S. S. (2017). Binding mode analysis, dynamic simulation

and binding free energy calculations of the MurF ligase from Acinetobacter baumannii.

Journal of Molecular Graphics and Modelling.77, 72-85.

Ahmad, S., Raza, S., Uddin, R., & Azam, S. S. (2018). Comparative subtractive proteomics based

ranking for antibiotic targets against the dirtiest superbug: Acinetobacter baumannii. Journal

of Molecular Graphics and Modelling, 82, 74–92.

Ahmad, T. A., Tawfik, D. M., Sheweita, S. A., Haroun, M., & El-Sayed, L. H. (2016).

Development of immunization trials against Acinetobacter baumannii. Trials in Vaccinology,

5, 53–60.

Albrecht, R., Schutz, M., Oberhettinger, P., Faulstich, M., Bermejo, I., Rudel, T., … Zeth, K.

(2014). Structure of BamA, an essential factor in outer membrane protein biogenesis. Acta

Crystallographica Section D: Biological Crystallography, 70(6), 1779–1789.




101

Amanna, I. J., & Slifka, M. K. (2011). Contributions of humoral and cellular immunity to vaccine-

induced protection in humans. Virology, 411(2), 206–215.

Andre, F. E., Booy, R., Bock, H. L., Clemens, J., Datta, S. K., John, T. J., … others. (2008).

Vaccination greatly reduces disease, disability, death and inequity worldwide. Bulletin of the

World Health Organization, 86, 140–146.

Asad, Y., Ahmad, S., Rungrotmongkol, T., Ranaghan, K. E., & Azam, S. S. (2018). Immuno-

informatics driven proteome-wide investigation revealed novel peptide-based vaccine targets

against emerging multiple drug resistant Providencia stuartii. Journal of Molecular Graphics

and Modelling, 80, 238-250.

Azam, S. S., & Shamim, A. (2014). An insight into the exploration of druggable genome of

Streptococcus gordonii for the identification of novel therapeutic candidates. Genomics,

104(3), 203–214.

Babu, M. M. (2016). The contribution of intrinsically disordered regions to protein function,

cellular complexity, and human disease. Biochemical Society Transactions, 44(5), 1185–

1200.

Barh, D., Barve, N., Gupta, K., Chandra, S., Jain, N., Tiwari, S., … others. (2013). Exoproteome

and secretome derived broad spectrum novel drug and vaccine candidates in Vibrio cholerae

targeted by Piper betel derived compounds. PloS One, 8(1), e52773.

Baseer, S., Ahmad, S., Ranaghan, K. E., & Azam, S. S. (2017). Towards a peptide-based vaccine

against Shigella sonnei: A subtractive reverse vaccinology based approach. Biologicals. 50,

87-99.



81(8), 3684–3690.

Buetow, L., Smith, T. K., Dawson, A., Fyffe, S., & Hunter, W. N. (2007). Structure and reactivity

of LpxD, the N-acyltransferase of lipid A biosynthesis. Proceedings of the National Academy

of Sciences, 104(11), 4321–4326.

Chang, S., Dunn, J. R., Heidari, M., Lee, L. F., Song, J., Ernst, C. W., … Zhang, H. (2010).

102

Genetics and vaccine efficacy: Host genetic variation affecting Marek’s disease vaccine

efficacy in White Leghorn chickens. Poultry Science, 89(10), 2083–2091.

Chen, L., Yang, J., Yu, J., Yao, Z., Sun, L., Shen, Y., & Jin, Q. (2005). VFDB: a reference database

for bacterial virulence factors. Nucleic Acids Research, 33(suppl_1), D325-D328.

Chen, W. (2015). Current advances and challenges in the development of Acinetobacter vaccines.

Human Vaccines & Immunotherapeutics, 11(10), 2495–2500.

Chiang, M.H., Sung, W.C., Lien, S.P., Chen, Y.Z., Lo, A. F., Huang, J.H., … Chong, P. (2015).

Identification of novel vaccine candidates against Acinetobacter baumannii using reverse

vaccinology. Human Vaccines & Immunotherapeutics, 11(4), 1065–1073.

Chung, E. H. (2014). Vaccine allergies. Clinical and Experimental Vaccine Research, 3(1), 50–

57.

Clem, A. S. (2011). Fundamentals of vaccine immunology. Journal of Global Infectious Diseases,

3(1), 73.

Cohen, I. R., & Weiner, H. L. (1988). T-cell vaccination. Immunology Today, 9(11), 332–335.

Craig, D. B., & Dombkowski, A. A. (2013). Disulfide by Design 2.0: a web-based tool for disulfide

engineering in proteins. BMC Bioinformatics, 14(1), 346.

Curcio, D., & Fernandez, F. (2008). Tigecycline for Acinetobacter baumannii infection: other

considerations. Clinical Infectious Diseases, 46(11), 1797–1798.

Dar, H., Zaheer, T., Rehman, M. T., Ali, A., Javed, A., Khan, G. A., … Waheed, Y. (2016).

Prediction of promiscuous T-cell epitopes in the Zika virus polyprotein: An in silico

approach. Asian Pacific Journal of Tropical Medicine, 9(9), 844–850.

Derr, P., Boder, E., & Goulian, M. (2006). Changing the specificity of a bacterial chemoreceptor.

Journal of Molecular Biology, 355(5), 923–932.

Dimitrov, I., Bangov, I., Flower, D. R., & Doytchinova, I. (2014). AllerTOP v. 2—a server for in

silico prediction of allergens. Journal of Molecular Modeling, 20(6), 2278.

Dixit, D., Madduri, R. P., & Sharma, R. (2014). The role of tigecycline in the treatment of

infections in light of the new black box warning. Taylor & Francis. 397-400.

103

Donev, R. (2015). Peptide and Protein Vaccines (Vol. 99). Academic Press.

Doytchinova, I. A., & Flower, D. R. (2007). VaxiJen: a server for prediction of protective antigens,

tumour antigens and subunit vaccines. BMC Bioinformatics, 8(1), 4.

EL-Manzalawy, Y., Dobbs, D., & Honavar, V. (2008). Predicting linear B-cell epitopes using

string kernels. Journal of Molecular Recognition: An Interdisciplinary Journal, 21(4), 243–

255.

Ferreira, L. G., Dos Santos, R. N., Oliva, G., & Andricopulo, A. D. (2015). Molecular docking and

structure-based drug design strategies. Molecules, 20(7), 13384–13421.

CDC. (2016). Antibiotic/antimicrobial resistance: biggest threats.

Garg, A., & Gupta, D. (2008). VirulentPred: a SVM based prediction method for virulent proteins

in bacterial pathogens. BMC Bioinformatics, 9(1), 62.

Gasteiger, E., Hoogland, C., Gattiker, A., Duvaud, S., Wilkins, M. R., Appel, R. D., & Bairoch,

A. (2005). Protein identification and analysis tools on the ExPASy server. Springer.



Gill, S. C., & Von Hippel, P. H. (1989). Calculation of protein extinction coefficients from amino

acid sequence data. Analytical Biochemistry, 182(2), 319–326.

Grandi, G. (2010). Bacterial surface proteins and vaccines. F1000 Biology Reports, 2.

Guruprasad, K., Reddy, B. V. B., & Pandit, M. W. (1990). Correlation between stability of a

protein and its dipeptide composition: a novel approach for predicting in vivo stability of a

protein from its primary sequence. Protein Engineering, Design and Selection, 4(2), 155–

161.

Han, L., Zheng, J., Wang, Y., Yang, X., Liu, Y., Sun, C., … others. (2016). Structure of the BAM

complex and its implications for biogenesis of outer-membrane proteins. Nature Structural

& Molecular Biology, 23(3), 192.

Haq, F. U., Abro, A., Raza, S., Liedl, K. R., & Azam, S. S. (2017). Molecular dynamics simulation

studies of novel β-lactamase inhibitor. Journal of Molecular Graphics and Modelling, 74,

104

143–152.

Hashemzehi, R., Doosti, A., Kargar, M., & Jaafarinia, M. (2018). Cloning and expression of nlpA

gene as DNA vaccine candidate against Acinetobacter baumannii. Molecular Biology

Reports, 1–7.

Hassan, A., Naz, A., Obaid, A., Paracha, R. Z., Naz, K., Awan, F. M., … Ali, A. (2016).

Pangenome and immuno-proteomics analysis of Acinetobacter baumannii strains revealed

the core peptide vaccine targets. BMC Genomics, 17(1), 732.

He, Y., Xiang, Z., & Mobley, H. L. T. (2010). Vaxign: the first web-based vaccine design program

for reverse vaccinology and applications for vaccine development. BioMed Research

International, 2010.

Heo, L., Park, H., & Seok, C. (2013). GalaxyRefine: protein structure refinement driven by side-

chain repacking. Nucleic Acids Research, 41(W1), W384-W388.

Howard, A., O’Donoghue, M., Feeney, A., & Sleator, R. D. (2012). Acinetobacter baumannii: an

emerging opportunistic pathogen. Virulence, 3(3), 243–250.

Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD-Visual Molecular Dynamics. Journal of

Molecular Graphics, 14: 33-38.

Ishida, T., & Kinoshita, K. (2007). PrDOS: prediction of disordered protein regions from amino

acid sequence. Nucleic Acids Research, 35(suppl_2), W460-W464.

Jansen, K. U., Knirsch, C., & Anderson, A. S. (2018). The role of vaccines in preventing bacterial

antimicrobial resistance. Nature Medicine, 24(1), 10.

Källberg, M., Wang, H., Wang, S., Peng, J., Wang, Z., Lu, H., & Xu, J. (2012). Template-based

protein structure modeling using the RaptorX web server. Nature Protocols, 7(8), 1511.

Karageorgopoulos, D. E., Kelesidis, T., Kelesidis, I., & Falagas, M. E. (2008). Tigecycline for the

treatment of multidrug-resistant (including carbapenem-resistant) Acinetobacter infections: a

review of the scientific evidence. Journal of Antimicrobial Chemotherapy, 62(1), 45–55.

Kechagia, M., Basoulis, D., Konstantopoulou, S., Dimitriadi, D., Gyftopoulou, K., Skarmoutsou,

N., & Fakiri, E. M. (2013). Health benefits of probiotics: a review. ISRN Nutrition, 2013.

105

Kehry, M. R., & Dahlquist, F. W. (1982). The methyl-accepting chemotaxis proteins of

Escherichia coli. Identification of the multiple methylation sites on methyl-accepting

chemotaxis protein I. Journal of Biological Chemistry, 257(17), 10378–10386.



Kennedy, D. A., & Read, A. F. (2017). Why does drug resistance readily evolve but vaccine

resistance does not? Proceedings of the Royal Society B: Biological Sciences, 284(1851),

20162562.

Khatoon, N., Pandey, R. K., & Prajapati, V. K. (2017). Exploring Leishmania secretory proteins

to design B and T cell multi-epitope subunit vaccine using immunoinformatics approach.

Scientific Reports, 7(1), 8285.

Koide, K., Ito, K., & Akiyama, Y. (2008). Substrate recognition and binding by RseP, an

Escherichia coli intramembrane protease. Journal of Biological Chemistry, 283, 9562-9570.

Kollman, P. A., Massova, I., Reyes, C., Kuhn, B., Huo, S., Chong, L., … others. (2000).

Calculating structures and free energies of complex molecules: combining molecular

mechanics and continuum models. Accounts of Chemical Research, 33(12), 889–897.

Koskiniemi, S., Lamoureux, J. G., Nikolakakis, K. C., de Roodenbeke, C. t’Kint, Kaplan, M. D.,

Low, D. A., & Hayes, C. S. (2013). Rhs proteins from diverse bacteria mediate intercellular

competition. Proceedings of the National Academy of Sciences, 201300627.

Kräutler, V., Van Gunsteren, W. F., & Hunenberger, P. H. (2001). A fast SHAKE algorithm to

solve distance constraint equations for small molecules in molecular dynamics simulations.

Journal of Computational Chemistry, 22(5), 501–508.

Krogh, A., Larsson, B., Von Heijne, G., & Sonnhammer, E. L. L. (2001). Predicting

transmembrane protein topology with a hidden Markov model: application to complete

genomes. Journal of Molecular Biology, 305(3), 567–580.

Kulah, C., Celebi, G., Aktas, E., Mengeloglu, Z., Comert, F., & Ankarali, H. (2009). Unexpected

tigecycline resistance among Acinetobacter baumannii Isolates: high minor error rate by

Etest. Journal of Chemotherapy, 21(4), 390–395.

106

Kyte, J., & Doolittle, R. F. (1982). A simple method for displaying the hydropathic character of a

protein. Journal of Molecular Biology, 157(1), 105–132.

Lee, C.R., Lee, J. H., Park, M., Park, K. S., Bae, I. K., Kim, Y. B., … Lee, S. H. (2017). Biology

of Acinetobacter baumannii: pathogenesis, antibiotic resistance mechanisms, and prospective

treatment options. Frontiers in Cellular and Infection Microbiology, 7, 55.

Li, W., & Godzik, A. (2006). Cd-hit: a fast program for clustering and comparing large sets of

protein or nucleotide sequences. Bioinformatics, 22(13), 1658–1659.

Li, W., Joshi, M. D., Singhania, S., Ramsey, K. H., & Murthy, A. K. (2014). Peptide vaccine:

progress and challenges. Vaccines, 2(3), 515–536.

Loewen, P. C., Alsaadi, Y., Fernando, D., & Kumar, A. (2014). Genome sequence of a tigecycline-

resistant clinical isolate of Acinetobacter baumannii strain AB031 obtained from a

bloodstream infection. Genome Announcements, 2(5).

Manchanda, V., Sanchaita, S., & Singh, N. P. (2010). Multidrug resistant Acinetobacter. Journal

of Global Infectious Diseases, 2(3), 291.

Mashiach, E., Schneidman-Duhovny, D., Andrusier, N., Nussinov, R., & Wolfson, H. J. (2008).

FireDock: a web server for fast interaction refinement in molecular docking. Nucleic Acids

Research, 36(suppl_2), W229-W232.

Mayrose, I., Penn, O., Erez, E., Rubinstein, N. D., Shlomi, T., Freund, N. T., … others. (2007).

Pepitope: epitope mapping from affinity-selected peptides. Bioinformatics, 23(23), 3244–

3246.

McGuffin, L. J., Bryson, K., & Jones, D. T. (2000). The PSIPRED protein structure prediction

server. Bioinformatics, 16(4), 404–405.




Minor, P. D. (2015). Live attenuated vaccines: Historical successes and current challenges.

Virology, 479, 379–392.

107

Mohd, R., Nesam, T., Kamaruzaman, L., Cader, R. A., Mustafar, R., & Kong, W.-Y. (2018).

Community acquired multi drug resistant (MDR) Acinetobacter baumannii pneumonia in

Malaysia--A case report. Respiratory Medicine Case Reports, 24, 147–149.

Moriel, D. G., Beatson, S. A., Wurpel, D. J., Lipman, J., Nimmo, G. R., Paterson, D. L., &

Schembri, M. A. (2013). Identification of novel vaccine candidates against multidrug-

resistant Acinetobacter baumannii. PloS One, 8(10), e77631.

Nagpal, R., Kumar, A., Kumar, M., Behare, P. V, Jain, S., & Yadav, H. (2012). Probiotics, their

health benefits and applications for developing healthier foods: a review. FEMS Microbiology

Letters, 334(1), 1–15.

Nascimento, I. P., & Leite, L. C. C. (2012). Recombinant vaccines and the development of new

vaccine strategies. Brazilian Journal of Medical and Biological Research, 45(12), 1102–

1111.

Navon-Venezia, S., Leavitt, A., & Carmeli, Y. (2007). High tigecycline resistance in multidrug-


Naz, A., Awan, F. M., Obaid, A., Muhammad, S. A., Paracha, R. Z., Ahmad, J., & Ali, A. (2015).

Identification of putative vaccine candidates against Helicobacter pylori. Helicobacter, 3, 1.

Nezafat, N., Eslami, M., Negahdaripour, M., Rahbar, M. R., & Ghasemi, Y. (2017). Designing an

efficient multi-epitope oral vaccine against Helicobacter pylori using immunoinformatics and

structural vaccinology approaches. Molecular BioSystems, 13(4), 699–713.

Ni, Z., Chen, Y., Ong, E., & He, Y. (2017). Antibiotic resistance determinant-focused

Acinetobacter baumannii vaccine designed using reverse vaccinology. International Journal

of Molecular Sciences, 18(2), 458.

Nikaido, H., & Takatsuka, Y. (2009). Mechanisms of RND multidrug efflux pumps. Biochimica

et Biophysica Acta (BBA)-Proteins and Proteomics, 1794(5), 769–781.

Noinaj, N., Guillier, M., Barnard, T. J., & Buchanan, S. K. (2010). TonB-dependent transporters:

regulation, structure, and function. Annual Review of Microbiology, 64, 43–60.

Nuccio, S.P., & Baumler, A. J. (2007). Evolution of the chaperone/usher assembly pathway:

108

fimbrial classification goes Greek. Microbiology and Molecular Biology Reviews, 71(4),

551–575.

Ohto, U., Yamakawa, N., Akashi-Takamura, S., Miyake, K., & Shimizu, T. (2012). Structural

analyses of human Toll-like receptor 4 polymorphisms D299G and T399I. Journal of

Biological Chemistry, 287(48), 40611–40617.

Ozpinar, G. A., Peukert, W., & Clark, T. (2010). An improved generalized AMBER force field

(GAFF) for urea. Journal of Molecular Modeling, 16(9), 1427–1440.

Pandey, R. K., Bhatt, T. K., & Prajapati, V. K. (2018). Novel immunoinformatics approaches to

design multi-epitope subunit vaccine for malaria by investigating anopheles salivary protein.


Parsons, L. M., Lin, F., & Orban, J. (2006). Peptidoglycan recognition by Pal, an outer membrane

lipoprotein. Biochemistry, 45(7), 2122–2128.

Paterlini, M. G., & Ferguson, D. M. (1998). Constant temperature simulations using the Langevin

equation with velocity Verlet integration. Chemical Physics, 236(1), 243–252.

Pearlman, D. A., Case, D. A., Caldwell, J. W., Ross, W. S., Cheatham, T. E., DeBolt, S., …

Kollman, P. (1995). AMBER, a package of computer programs for applying molecular

mechanics, normal mode analysis, molecular dynamics and free energy calculations to

simulate the structural and energetic properties of molecules. Computer Physics

Communications, 91(1–3), 1–41.

Peleg, A. Y., Seifert, H., & Paterson, D. L. (2008). Acinetobacter baumannii: emergence of a

successful pathogen. Clinical Microbiology Reviews, 21(3), 538–582.

Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C., &

Ferrin, T. E. (2004). UCSF Chimera—a visualization system for exploratory research and

analysis. Journal of Computational Chemistry, 25(13), 1605–1612.

Pieper, U., Eswar, N., Davis, F. P., Braberg, H., Madhusudhan, M. S., Rossi, A., … others. (2006).

MODBASE: a database of annotated comparative protein structure models and associated

resources. Nucleic Acids Research, 34(suppl_1), D291-D295.

109

Pogue, J. M., Mann, T., Barber, K. E., & Kaye, K. S. (2013). Carbapenem-resistant Acinetobacter

baumannii: epidemiology, surveillance and management. Expert Review of Anti-Infective

Therapy, 11(4), 383–393.

Pruitt, K. D., Tatusova, T., & Maglott, D. R. (2005). NCBI Reference Sequence (RefSeq): a

curated non-redundant sequence database of genomes, transcripts and proteins. Nucleic Acids

Research, 33(suppl_1), D501--D504.

Rashid, M. I., Naz, A., Ali, A., & Andleeb, S. (2017). Prediction of vaccine candidates against

Pseudomonas aeruginosa: An integrated genomics and proteomics approach. Genomics,

109(3), 274–283.

Rosa, D. S., Ribeiro, S. P., Fonseca, S. G., Almeida, R. R., Santana, V. C., Apostolico, J. de S., …

Cunha-Neto, E. (2015). Multiple approaches for increasing the immunogenicity of an

epitope-based anti-HIV vaccine. AIDS Research and Human Retroviruses, 31(11), 1077–

1088.

Saadi, M., Karkhah, A., & Nouri, H. R. (2017). Development of a multi-epitope peptide vaccine

inducing robust T cell responses against brucellosis using immunoinformatics based

approaches. Infection, Genetics and Evolution, 51, 227–234.



147–153.

Schneidman-Duhovny, D., Inbar, Y., Nussinov, R., & Wolfson, H. J. (2005). PatchDock and

SymmDock: servers for rigid and symmetric docking. Nucleic Acids Research, 33(suppl_2),

W363-W367.



Sette, A., & Rappuoli, R. (2010). Reverse vaccinology: developing vaccines in the era of

genomics. Immunity, 33(4), 530–541.

Shrestha, P., Cooper, B. S., Coast, J., Oppong, R., Thuy, N. D. T., Phodha, T., … Lubell, Y. (2018).

Enumerating the economic cost of antimicrobial resistance per antibiotic consumed to inform

110

the evaluation of interventions affecting their use. Antimicrobial Resistance & Infection

Control, 7(1), 98.

Singh, H., & Raghava, G. P. S. (2001). ProPred: prediction of HLA-DR binding sites.


Singh, H., & Raghava, G. P. S. (2003). ProPred1: prediction of promiscuous MHC Class-I binding

sites. Bioinformatics, 19(8), 1009–1014.

Singh, R., Capalash, N., & Sharma, P. (2017). Immunoprotective potential of BamA, the outer

membrane protein assembly factor, against MDR Acinetobacter baumannii. Scientific

Reports, 7(1), 12411.

Singh, R., Garg, N., Shukla, G., Capalash, N., & Sharma, P. (2016). Immunoprotective efficacy of

Acinetobacter baumannii outer membrane protein, FilF, predicted in silico as a potential

vaccine candidate. Frontiers in Microbiology, 7, 158.

Skariyachan, S., Manjunath, M., & Bachappanavar, N. (2018). Screening of potential lead

molecules against prioritised targets of multi-drug-resistant Acinetobacter baumannii-

insights from molecular docking, molecular dynamic simulations and in vitro assays. Journal

of Biomolecular Structure and Dynamics, 1–24.

Soltes, G. R., Schwalm, J., Ricci, D. P., & Silhavy, T. J. (2016). The activity of Escherichia coli

chaperone SurA is regulated by conformational changes involving a parvulin domain. Journal

of Bacteriology, JB-00889.

Soria-Guerra, R. E., Nieto-Gomez, R., Govea-Alonso, D. O., & Rosales-Mendoza, S. (2015). An

overview of bioinformatics tools for epitope prediction: implications on vaccine

development. Journal of Biomedical Informatics, 53, 405–414.

Spapen, H., Jacobs, R., Van Gorp, V., Troubleyn, J., & Honore, P. M. (2011). Renal and

neurological side effects of colistin in critically ill patients. Annals of Intensive Care, 1(1),

14.

Sparling, P. F. (1983). Bacterial virulence and pathogenesis: an overview. Reviews of Infectious

Diseases, 5(Supplement_4), S637--S646.

111

Szklarczyk, D., Franceschini, A., Wyder, S., Forslund, K., Heller, D., Huerta-Cepas, J., … others.

(2014). STRING v10: protein-protein interaction networks, integrated over the tree of life.

Nucleic Acids Research, 43(D1), D447-D452.

Taneja, N., Singh, G., Singh, M., & Sharma, M. (2011). Emergence of tigecycline & colistin

resistant Acinetobacter baumanii in patients with complicated urinary tract infections in north

India. The Indian Journal of Medical Research, 133(6), 681.

Tusnady, G. E., & Simon, I. (2001). The HMMTOP transmembrane topology prediction server.


Uline, M. J., & Corti, D. S. (2013). Molecular dynamics at constant pressure: allowing the system

to control volume fluctuations via a “shell” particle. Entropy, 15(9), 3941–3969.

Venkatesha, S., Dudics, S., Acharya, B., & Moudgil, K. (2015). Cytokine-modulating strategies

and newer cytokine targets for arthritis therapy. International Journal of Molecular Sciences,

16(1), 887–906.

Viehman, J. A., Nguyen, M. H., & Doi, Y. (2014). Treatment options for carbapenem-resistant

and extensively drug-resistant Acinetobacter baumannii infections. Drugs, 74(12), 1315–

1333.

Vitrinel, A., & Erdaug, G. C. (2008). Principles of immunization. Turkish Journal of Family

Practice, 11(4), 149–153.

Wadood, A., Jamal, A., Riaz, M., Khan, A., Uddin, R., Jelani, M., & Azam, S. S. (2018).

Subtractive genome analysis for in silico identification and characterization of novel drug

targets in Streptococcus pneumonia strain JJA. Microbial Pathogenesis, 115, 194–198.

Wang, C., Greene, D., Xiao, L., Qi, R., & Luo, R. (2018). Recent developments and applications

of the MMPBSA method. Frontiers in Molecular Biosciences, 4, 87.

Wang, G., & Zhu, W. (2016). Molecular docking for drug discovery and development: A widely

used approach but far from perfect. Future Science.

Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A., & Case, D. A. (2004). Development and

testing of a general amber force field. Journal of Computational Chemistry, 25(9), 1157–

112

1174.

Weiser, J., Shenkin, P. S., & Still, W. C. (1999). Approximate atomic surfaces from linear

combinations of pairwise overlaps (LCPO). Journal of Computational Chemistry, 20(2),

217–230.

Xie, R., Zhang, X. D., Zhao, Q., Peng, B., & Zheng, J. (2018). Analysis of global prevalence of

antibiotic resistance in Acinetobacter baumannii infections disclosed a faster increase in

OECD countries. Emerging Microbes & Infections, 7(1), 31.

Yu, C.S., Chen, Y.C., Lu, C.H., & Hwang, J.K. (2006). Prediction of protein subcellular

localization. Proteins: Structure, Function, and Bioinformatics, 64(3), 643–651.

Yu, C.S., Cheng, C.W., Su, W.C., Chang, K.C., Huang, S.W., Hwang, J.K., & Lu, C.H. (2014).

CELLO2GO: a web server for protein subcellular localization prediction with functional gene

ontology annotation. PLoS One, 9(6), e99368.

Yu, N. Y., Wagner, J. R., Laird, M. R., Melli, G., Rey, S., Lo, R., … others. (2010). PSORTb 3.0:

improved protein subcellular localization prediction with refined localization subcategories

and predictive capabilities for all prokaryotes. Bioinformatics, 26(13), 1608–1615.

Zhang, R., Ou, H.Y., & Zhang, C.T. (2004). DEG: a database of essential genes. Nucleic Acids

Research, 32(suppl_1), D271--D272.

Zhang, Y. (2008). I-TASSER server for protein 3D structure prediction. BMC Bioinformatics,

9(1), 40.

Zhu, L., Cheng, J., Luo, B., Feng, S., Lin, J., Wang, S., … Wang, H. (2009). Functions of the

Clostridium acetobutylicium FabF and FabZ proteins in unsaturated fatty acid biosynthesis.

BMC Microbiology, 9(1), 119.

113

Chapter # 4

A Novel Approach of Virulome Based Reverse Vaccinology

For Exploring and Validating Peptide-Based Vaccine

Candidates Against the Most Troublesome Nosocomial

Pathogen: Acinetobacter baumannii

4.1. Abstract

A. baumannii is one of the major cause of nosocomial infections around the globe. The emergence

of hyper-virulent strains of the pathogen greatly narrows down therapeutic options for patients

infected with this red alert superbug. Development of a peptide-based vaccine can offers an

alternative, attractive, and cost-effective remedy for MDR A. baumannii associated complications.

Herein, we introduced a novel virulome based RV for screening peptide based vaccine candidates

against A. baumannii and its validation using a negative control. The pipeline screened

“FYLNDQPVS” of polysaccharide export outer membrane protein (EpsA) and “LQNNTRRMK”

of chaperone-usher pathway protein B (CsuB) as broad-spectrum peptides for induction of targeted

immune responses. The 9-mer epitope of both proteins was rendered virulent, antigenic, non-

allergen, and highly conserved among thirty-four completely annotated strains. Interactome

examination unravels peptides protein direct and indirect interactions with biological significant

pathways, essential for A. baumannii pathogenesis and survival. Protein-peptide docking aids in

addition by unveiling deep binding of the epitopes in the active site of the most prevalent binding

allele in the human population- the DRB1*0101. Both the proteins till to date are not characterized

for immunoprotective efficacy and desirable to be deciphered experimentally. The designed series

of in silico filters rejected few recently reported peptide and non-peptide vaccine targets and has

delivered outcomes, which we believe will enrich the existing knowledge of vaccinology against

this life-threatening human pathogen.

114

4.2. Introduction

A. baumannii is a Gram-negative bacterium and opportunistic human pathogen, associated with

myriad hospital-acquired infections (Howard, O’Donoghue, Feeney, & Sleator, 2012). A.

baumannii is resistant to the majority of first-line antibiotics by acquiring diverse resistance

mechanisms, resulting in the global emergence of MDR strains (McConnell, Actis, & Pachon,

2013). More alarming are the reports pointing toward the spread of pan-drug resistant (PDR)

strains. These strains are resistant to almost all clinically used antibiotics, which implies a high-

risk challenge for its control and treatment (Valencia et al., 2009).

The development of a safe and effective vaccine is one promising strategy that can be adopted for

reducing A. baumannii infections (Laxminarayan et al., 2013). Vaccines based on inactivated

whole microbial pathogen are historically used for provoking specific immune responses and

protecting the host from subsequent infection. However, such vaccine formulation apparently

contains an array of microbial proteins, not mandatory for inducing protective immunity. In fact,

the immune protection requires few selected proteins of the pathogen within such formulation. In

addition, these extra proteins are often associated with reactogenic and allergenic responses, thus

making live-attenuated vaccine formulation an inappropriate option. This justification has led to

the choice of subunit-based vaccines composed of a protein or protein peptides. Developing

peptide-based vaccine seems to be an attractive and alternative strategy. Engineering peptide

vaccines are safe, cost-effective, and based on short length of amino acids, representing specific

antigenic epitope that induces specific, desirable, and broad-spectrum humoral and cellular

immunity (Li, Joshi, Singhania, Ramsey, & Murthy, 2014).

As there is scarcity of antigenic peptides against A. baumannii compared to other bacterial

pathogens, it is imperative to highlight novel vaccine peptides by investigating large number of

uncharacterized proteins. The efficacy of peptide vaccines greatly depend upon protein virulent

potential, mediating host immune system efficiently compared to non-virulent proteins (Garcon,

Stern, Cunningham, & Stanberry, 2011). The availability of complete virulent protein dataset in

virulent factor databases endows the opportunity to filter the virulent proteome through a

comprehensive in silico framework, and to identify proteins, which harbor suitable antigenic

peptides. Additional factors which could aid in vaccine proteins prioritization includes:

localization in the outer membrane or extracellular matrix, non-homologous to the host proteome,

115

adhesion probability, lesser number of transmembrane helices, signature peptide presence, peptide

conservation, enhanced antigenicity, non-allergencity, and exposed surface topology. Findings of

the current study could provide a ground for further studies to test these genetically invariable

epitopes in animal models, and provide future directions for vaccine designing and development

against this pathogen.

4.3. Material and Methods

The flow chart illustrating the applied methodology of the current study is shown in Fig.4.1.

4.3.1. Virulome Retrieval and Functional Categorization

The complete virulome of A. baumannii was retrieved from VFDB (L. Chen et al., 2005). After

assembly, the virulent proteome was screened for exo-proteome and secretome of the pathogen

(Barh et al., 2013; Naz et al., 2015) using three on-line sub-cellular localization tools: PsortB (N.

Y. Yu et al., 2010), Cello (C.S. Yu, Chen, Lu, & Hwang, 2006), and Cello2GO (C.S. Yu et al.,

2014). Following this, comparative alignment of individual extracellular and outer membrane

protein with the human proteome was carried out using an on-line Protein Basic Local Alignment

Search Tool (BLAST) (Pruitt, Tatusova, & Maglott, 2005). The BLAST search was performed

against Homo sapiens (taxonomic ID: 9606) proteome as a reference with the E-value cutoff set

to 10-4. Proteins having an identity < 35% or no corresponding hit in the BLAST search were

considered as host non-homologous bacterial proteins.

4.3.2. Physicochemical Prioritization

Low molecular weight proteins are suitable candidates for vaccine development due to purification

ease (Barh et al., 2013). Keeping this in view, weight calculator tool available on ExPASY server

was employed to filter proteins with weight < 110 kDa (Gasteiger et al., 2005; Naz et al., 2015).

Prioritized proteins were then analyzed through HMMTOP (Tusnady & Simon, 2001), and

Transmembrane Helices; Hidden Markov Model (TMHMM) (Krogh, Larsson, Von Heijne, &

Sonnhammer, 2001) for topology, and computing the number of transmembrane helices. Proteins

were further characterized through SPAAN (Sachdeva, Kumar, Jain, & Ramachandran, 2004) for

predicting adhesion-like proteins.

http://db.psort.org/),%20CELLO%20%5b11%5d%20(http:/cello.life.nctu.edu.tw/)%20and%20CELLO2GO

http://db.psort.org/),%20CELLO%20%5b11%5d%20(http:/cello.life.nctu.edu.tw/)%20and%20CELLO2GO

116

4.3.3. Epitopes Mapping

Identification of antigenic epitopes that stimulates both B and T-cells immunity is imperative for

the development of peptide-based vaccines (Barh, Misra, Kumar, & Vasco, 2010; Baseer, Ahmad,

Ranaghan, & Azam, 2017). Antigenic potential of the shortlisted candidates was estimated through

VaxiJen (Doytchinova & Flower, 2007)using a cutoff ≥ 0.4. Antigenic proteins were mapped for

B-cell epitopes using BCPred with epitope length and score set to 20-mer and ≥ 0.8, respectively

(EL-Manzalawy, Dobbs, & Honavar, 2008). The resultant B-cell epitopes of each protein were

then analyzed for surface exposure through TMHMM. Antigenic and surface exposed B-cell

epitopes were further mapped for T-cell epitopes. Binding alleles of MHC-I and MHC-II for B-

cell epitopes were mapped using prophred I (H. Singh & Raghava, 2003) and prophred (H. Singh

& Raghava, 2001), respectively. Shared epitopes then underwent MHCpred (Guan, Doytchinova,

Zygouri, & Flower, 2003)and VirulentPred (Garg & Gupta, 2008)evaluation to determine the IC50

value (< 100 nm) and virulent potential, respectively. The 20-mer B-cell epitopes of proteins were

further characterized for surface accessibility, flexibility, hydrophobicity, antigenicity, and β-

turns. This was accomplished by the Immune Epitope Database and Analysis Resource (IEDB)

analysis resource server. The 20-mer B-cell epitopes were submitted to Emini surface accessibility

prediction tool (Emini, Hughes, Perlow, & Boger, 1985), Karplus and Schulz flexibility scale

(Karplus & Schulz, 1985), Parker hydrophilicity scale (Parker, Guo, & Hodges, 1986), Kolaskar

and Tongaonkar Antigenicity (Kolaskar & Tongaonkar, 1990) and Chou and Fashman β-turn

prediction tool (Chou & Fasman, 1977). The peptide sequence with the best score in aforesaid five

analysis was selected and cross-referred with its respective T-cell epitope. In order to investigate

whether the epitopes can cause allergic reactions or not, SORTALLER was employed (Zhang et

al., 2012). To examine prioritized set of epitopes for conservation among 34 completely annotated

strains of A. baumannii, CLC sequence viewer 7 (www.clcbio.com) was utilized.

4.3.4. Interactome Evaluation

Interactome evaluation was performed using STRING (Szklarczyk et al., 2014). To evaluate the

potential of proteins for their interaction with the host, Pathosystems Resource Integration System

(PATRIC) was employed (Wattam et al., 2013).

javascript:openWindow(%22ScaleInformation?ID=69%22,%20%22BcellScaleInformation%22)



http://www.clcbio.com/

117

Fig.4.1. Schematic representation of in silico framework for identification of putative vaccine candidates against A. baumannii.

118

4.3.5. Structure Prediction and Evaluation

To view the topology of prioritized epitopes on the protein surface, the 3D structure of proteins

were predicted using a comparative structure prediction approach (Iqbal, Shamim, Azam, &

Wadood, 2016). Four structure prediction tools: I-tasser (Wu, Skolnick, & Zhang, 2007, Modweb

(Pieper et al., 2006), Phyre2 (Kelley, Mezulis, Yates, Wass, & Sternberg, 2015) and Swiss-model

(Schwede, Kopp, Guex, & Peitsch, 2003) were used for this purpose. To select the most optimum

model for epitopes topology view, each structure was subjected to Procheck (Laskowski,

MacArthur, Moss, & Thornton, 1993), Errat (Colovos & Yeates, 1993), Verify-3D Eisenberg,

Lüthy, & Bowie, 1997), and ProSA (Wiederstein & Sippl, 2007). The comprehensive

stereochemical evaluation of models was first done by Procheck, which includes Ramachandran

plot (Ramakrishnan & Ramachandran, 1965) and G-Factor (Morris, MacArthur, Hutchinson, &

Thornton, 1992). The best model of each protein then go through an energy minimization phase

carried out with the UCSF Chimera (Pettersen et al., 2004), by applying Gasteiger charges with

the steepest descent and conjugate gradient set to 750 steps under TFF.

4.3.6. Epitopes Exo-membrane Topology Evaluation

Prioritized proteins were then subjected to Pepitope server (Mayrose et al., 2007) to ensure the

exo-membrane topology and non-folding of epitopes in the protein globular structure.

4.3.7. Epitopes Molecular Docking

Finally, epitopes fulfilling all parameters of the framework were docked with the 3D structure of

DRB1*0101(PDB ID.1AQD). Docking was performed with GalaxyPepDock (Lee, Heo, Lee, &

Seok, 2015). The best-docked peptide-protein complex was thoroughly visualized through UCSF

Chimera, Molecular Operating Environment (MOE) (https://www.chemcomp.com/MOE-

Structure_Based_Design.htm), and Ligplot (Laskowski & Swindells, 2011) for binding mode and

binding interactions analysis.

4.3.8. Methodology Validation through a Negative Control

To validate predictions made by the pipeline, a negative control (Mycoplasma pneumonia strain

ATCC 29342 / M129) was used. The complete proteome of the pathogen was retrieved from

Uniprot (http://www.uniprot.org/) and used in the blastp search of VFDB to screen proteins that

https://www.chemcomp.com/MOE-Molecular_Operating_Environment.htm

https://www.chemcomp.com/MOE-Structure_Based_Design.htm

https://www.chemcomp.com/MOE-Structure_Based_Design.htm

http://www.uniprot.org/

119

have identity of ≥ 35% and bit score of 100. The resulting virulome was then thoroughly

investigated for peptide vaccines using the same filters discussed in the methodology.


4.4.1. A. baumannii Virulome Assembly and Evaluation

Virulence is considered as an essential ingredient for a protein to serve as a potential vaccine

candidate (Barh et al., 2013; Naz et al., 2015). Virulent proteins can mediate severe infection

pathways and can lead to disease efficiently when compared to non-virulent proteins (Naz et al.,

2015). Likewise, virulence mechanisms also assist the pathogen in immune evasion,

immunosuppression and nutrient extraction, thus making the pathogen fittest in the competing

environment (Joly-Guillou, 2005). An organism-specific search was performed against VFDB to

extract A. baumannii virulome. The search screened out seventy virulent proteins associated with

adherence, biofilm formation, immune evasion, iron uptake, serum resistance, virulent enzymes,

and virulence factor-regulated proteins (S-Table 4.1). The majority of proteins falls into the

category of immune evasion, constituted by two subgroups, capsule, and lipopolysaccharide (LPS)

family. The capsule is regarded as one of the major virulence factors of A. baumannii, and plays a

critical role in dampening host innate immune responses (Joly-Guillou, 2005). LPS is

indispensable for survival in the desiccated environment of hospital's settings (Russo et al., 2010).

It also favors A. baumannii to evade host immune responses and trigger inflammatory reactions

(Boll et al., 2015). Iron uptake and biofilm proteins were the subsequent abundant group

comprising 18 and 14 proteins, respectively. Iron uptake proteins were mainly acinetobactin, while

biofilm proteins were sub grouped as AdeFGH efflux pump proteins, Biofilm-associated protein

(Bap) proteins, Csu fimbriae proteins, and Poly-N-acetyl glucosamine (PNAG) proteins.

Acinetobactin is a catechol-hydroxamate siderophore and competes with the host for iron

acquisition (Lin et al., 2012). The AdeFGH efflux pump proteins collectively play a potential role

in the synthesis and transportation of auto-inducers during formation of biofilm (Gaddy et al.,

2012). Bap proteins, in addition, to their role in biofilm formation, also contribute in adherence to

eukaryotic cells (X. He et al., 2015). Csu fimbriae are chaperon-usher pathway assembled fimbriae

and allow bacteria to adhere abiotic surfaces. Attachment is followed by formation of micro-

colonies, which then mature to complete the biofilm structure (Brossard & Campagnari, 2012).

PNAG is a polysaccharide secreted by several pathogens including A. baumannii at their cell

120

surface and is critical for the biofilm synthesis (Tomaras, Dorsey, Edelmann, & Actis, 2003).

Groups containing the least number of virulent proteins were serum resistance penicillin-binding

protein G and outer membrane adherence protein. The outer membrane protein A (OmpA)

contributes significantly to A. baumannii pathogenesis by adhering and invading host epithelial

cells, thus contributing to its dissemination (A. H. K. Choi, Slamti, Avci, Pier, & Maira-Litrán,

2009). In addition, the OmpA protein aids immune evasion of A. baumannii by inhibiting

alternative complement pathway, allowing the pathogen to avoid complement-mediated killing (C.

H. Choi, Lee, Lee, Park, & Lee, 2008). Beside this, OmpA induces cell apoptosis, assist biofilm

formation, and cell motility (Kim et al., 2009). Additional information of the retrieved proteins

like molecular localization, start and end residue number, strand preference (plus or minus strand),

and cluster of orthologous genes (COG) characterization was also accessed from VFDB. All the

proteins are chromosomally encoded with negative strand are comparatively enriched compared

to their distribution on the plus strand.

4.4.2. Exo-proteome and Secretome Exploration

The proceeding filter of the computational framework screen proteins that constitute exo-proteome

and secretome of A. baumannii. Exo-proteome and secretome comprise proteins present in the

extracellular matrix and outer membrane of the organism. They come in frequent contact with the

host environment and could be suitable targets for exploring potential vaccine candidates. The

virulent dataset of proteins was examined for its subcellular localization using three different on-

line localization tools. Subcellular localization of proteins was first analyzed by PSORTb and

CELLO, followed by CELLO2GO for cross-checking the consistency of results. Out of total 70

virulent proteins, PSORTb illustrated five (7.1%) proteins, localized in the outer membrane region

while four (5.7%) were present in extracellular proximity (Fig.4.2A). Likewise, 14 (20%) and four

(6%) proteins were found by CELLO to constitute exo-proteome and secretome of the organism

and are illustrated in (Fig.4.2B). Comparative analysis of results generated by all three tools

brought forth ten proteins, which can be targeted for vaccine production (illustrated in S-Table

4.2). In the majority of cases, these surface exposed and cell-secreted proteins are immunogenic

and virulent, thus targeting such proteins offers many advantages, as they play significant roles in

pathogenesis by enabling the pathogen to adhere host tissues, invade, survive, and proliferate in

the host environment.

121

Fig.4.2. Subcellular localization analysis of virulent proteins.

4.4.3. Human Non-Homologous Proteins

Exo-proteome and secretome virulome of A. buamannii were aligned with the host proteome, to

ensure non-homology of proteins. Non-homology is a prerequisite for avoiding host auto-immune

responses and formulating vaccine preparation, which can mediate accurate immune responses

(Barh et al., 2013; Naz et al., 2015). Non-homology of pathogen proteins also minimizes the

chances of integration and recombination in the host genome and provide an effective targeted

therapy for the pathogen of choice. Screened extracellular and outer membrane virulent proteins

were filtered further through subtractive proteomics, to deduct proteins revealing similarity with

the human proteome. All the shortlisted proteins were found non-homologous as no-significant

hits were observed during the analysis. This indicates that the proteins truly belong to the pathogen

and play part in pathogenic pathways, unique to A. baumannii.

122


To get an in-depth understanding of the refined list of proteins from the previous phases, the

scrutinized proteins were prioritized based on several physicochemical parameters. Prioritizing

proteins not only save time, resources, and labor expenditures, but also boost the chances of

screening proteins with high potential to act as vaccine candidates. The prime parameter for

investigating the list of 10 proteins was estimating their molecular weights. Therefore, proteins

were first characterized for their molecular weight using ExPASy weight calculation tool. Proteins

weighing < 110 kDa are preferred as effective vaccine targets. Low molecular weight proteins are

highly desirable for subsequent validation studies, as they can be easily purified during wet lab

analysis. All proteins were turned out <110 kDa except hemolysin type calcium binding domain

containing protein weighing 249 kDa, thus excluded from the dataset. The OmpA protein, which

is considered as a vital protein for A. baumannii pathogenesis and act as environment

communicator, has low molecular weight of 34.86 kDa. Similarly, CsuA/B, CsuB, EpsA, Ligated-

gated channel protein, AdeH, CsuD, CsuE , PgaB, BauA have low molecular weights of 18.71

kDa, 19.37 kDa, 40.64 kDa, 83.03 kDa, 52.41 kDa, 92.74 kDa, 36.6 kDa, 68.87 kDa, and 38 kDa,

respectively. This list of nine proteins was subjected to a second round of characterization by

predicting the number of transmembrane helices they possess. To analyze topological view of the

proteins, and to compute the number of transmembrane of proteins, two online tools; HMMTOP

and TMHMM were employed. Proteins harboring 0 to 1 transmembrane helices are favored, as

proteins with more than two transmembrane helices are less likely to clone and express. All

proteins were found to have either one or no transmembrane helix except CsuE protein, which

revealed three transmembrane helices by HMMTOP, therefore, not processed further. Remaining

eight proteins were finally brought forth to adhesion probability check, where the adhesion

potential was determined. Adhesion proteins aid bacteria to attach to host cell receptors, and allow

bacteria to colonize. Targeting such proteins is an effective approach to prevent bacterial infections

in the primary stages (Wizemann, Adamou, & Langermann, 1999). Adhesion analysis was carried

out with SPAAN, which revealed five proteins of adhesion potential. These five proteins have

values greater than the cutoff (0.5), and were termed as adhesive proteins. The remaining three

proteins scored less than the threshold, therefore, not processed further. Now the final dataset

contains five proteins (S-Table 4.3 and S-Fig.4.1), all of which fulfilled the above filters, thus

approved for antigenic peptides mining.

123

4.4.5. B-cell Epitope Derived T-cell Epitope Mapping

The ability of an epitope to provoke a strong immune response is imperative for vaccine

development. The prediction of an epitope-based vaccine truly relies on its binding potential to

both MHC molecules: MHC-I and MHC-II. Such peptide-based vaccines are safe, evoke specific

immunity, and are easy to produce (Naz et al., 2015). The physicochemical based prioritized set

of proteins was first evaluated for their antigenicity, therefore, the shortlisted protein candidates

were subjected to VaxiJen analysis. Proteins were considered potentially antigenic, if they

exhibited score higher than the cutoff (> 0.4). The analysis revealed all proteins as antigenic with

the following scores; OmpA (0.85), EpsA (0.5), CsuA/B (0.82), CusB (0.56), and ligand-gated

channel protein (0.559). After affirming the proteins as antigenic, the proteins were further

examined for B-cell epitopes. Each protein declared a number of B-cell epitopes; ligated-channel

protein (21), CsuA/B and CsuB (4) while EpsA and OmpA protein have 4 and 9 B-cell epitopes,

respectively (S-Table 4.4. Section. B-cell epitopes). To ensure exo-membrane topology of B-cell

epitopes, their amino acid sequences were analyzed through TMHMM server. The highest number

of surface exposed B-cell epitopes were disclosed for ligated-gated channel protein, i.e. 5 while

for CsuA/B, CsuB, EpsA, and OmpA the analysis unveil 3, 1, 4 and 2, respectively (S-Table 4.4.

Section. Selected B-cell epitopes).

To design B-cell derived T-cell epitopes, each B-cell epitope with exposed topology underwent T-

cell mapping to nominate peptide sequences with the best capability of stimulating both B and T-

cell immunity. The selection criteria was based on the fact that, epitopes binding to both classes

of MHC were chosen. Following this, epitopes binding to 15 or more alleles were selected. The

filter stipulated that “VRYQEVESS” epitope of ligand-gated channel protein binds to 40 (MHC-

I:2, MHC-II:38), “LQNNTRRMK” epitope of CsuB protein binds to 17 (MHC-I:1, MHC-II:16),

and “FYLNDQPVS” epitope of EpsA protein binds to 17 (MHC-I:4, MHC-II:13) (S-Table 4.4.

Section. T-cell epitopes). The binding affinity of the filtered epitopes for DRB1*0101 allele was

determined though MHCPred. DRB1*0101 is one most common allele in the human population,

and epitopes exhibiting a high propensity toward DRB1*0101 binding are considered as good

immunogens [29]. Epitopes with an IC50 value less than 100nM were selected. The peptide of

ligated channel protein (VRYQEVESS) secure an IC50 score of 95.94 nM, “LQNNTRRMK” of

CsuB has 90.36 nM and “FYLNDQPVS” of EpsA has 1.36 nM. These epitopes were further

124

subjected to VirulentPred, where they were analyzed for the second round of virulence. All three

epitopes were found antigenic by securing scores as follows; “VRYQEVESS” (1.06),

“LQNNTRRMK” (1.06) and “FYLNDQPVS” (1.05). The virulent epitopes were again analyzed

for their antigenicity through VaxiJen and declared as antigenic (S-Table 4.4. Section. Final list).

Through the second round of confirmation, the B-cell epitopes of prioritized T-cell epitopes were

again characterized via additional five tools available at IDEB server to predict the best T-cell

peptide for B-cell epitopes. In case of CsuB, “LQNNTRRMK” was selected because of its higher

accessibility, flexibility, hydrophobicity, antigenicity, and β-turn prediction score (S-Table 4.5.

Section. CsuB protein). Similarly, peptide “YLNDQPV” for EpsA (S-Table 4.5. Section. EpsA

protein) and “FYLNDQ” for ligated- channel protein (S-Table 4.5. Section. Ligated-gated

channel protein) were selected. These results were cross-referred with their respective T-cell

epitopes and found that the filtered T-cell epitopes activate both types of immunity. The 9-mer

epitope of the three proteins was confirmed as antigenic, virulent, and bind to both MHC classes,

therefore, selected for further studies (Table 4.1).

4.4.6. Epitopes Allergenicity and Conservation

One important aspect while selecting epitopes for vaccine designing is to unravel their

allergenicity. Increasing number of vaccines have been found, that can cause adverse vaccine

reactions, although it is not clear whether adverse effects after vaccination are because of the

vaccine itself or by some other factors (Chung, 2014). Whatever the reason, even mild allergies to

vaccines can lead to severe complications. Epitopes were classified non-allergen and targeted for

additional analysis. Vaccines with broad-spectrum activities require epitope to be conserved

among strains of the pathogen. The conservation of epitopes was performed by accessing

completely annotated protein positive strains of A. baumannii, available at NCBI genome database.

The selected epitope “LQNNTRRMK” in case of CsuB protein and “FYLNDQPVS” in case of

EpsA were found highly conserved throughout the strains (S-Fig.4.2A and B). Major variations

were observed for ligated-gated channel protein and seem to be less conserved, therefore, excluded

from additional analysis.

4.4.7. Cellular Interactome of CsuB and EpsA

The protein-protein network of prioritized epitopes protein was constructed using STRING. A

network of both proteins was thoroughly investigated for interpretation of direct and indirect

125

connections. No direct or indirect interactions were observed between the proteins. However, each

protein interacts either directly or indirectly with a number of other essential cellular proteins that

play significant roles in A. baumannii pathogenesis and survival (S-Table 4.6). Adherence and

attachment of A. baumannii to abiotic surfaces is an important aspect of A. baumannii

pathogenesis, mediated by several surfaces associated attributes of the pathogen (Mortensen &

Skaar, 2012). The CsuB is a secreted protein of type I pili and is a member of the Csu family. This

family of proteins is encoded by Csu operon, which contains six genes for six proteins. These

proteins altogether play a major role in CsuA/BABCDE chaperone-usher pili assembly system,

which in turn results in the formation of biofilm on abiotic surfaces (X. He et al., 2015). The CsuB

protein in the current study was found to interact with all these proteins in addition to transport

channel of sugars, flagellar assembly chaperon and outer membrane usher proteins in a direct and

indirect fashion (S-Fig.4.3A). EpsA protein together with phosphor tyrosine kinase protein is

required for biosynthesis of the capsule in bacteria including A. baumannii (Russo et al., 2010).

Capsule provides an extra physical barrier and increases bacterial virulence, persistence, and aids

bacteria in immune evasion. STRING analysis revealed its interaction with protein tyrosine kinase

(PTK) protein, proteins of tyrosine phosphatase family, sugar transferase family,

aminotransferases, and outer membrane polysaccharides associated proteins (S-Fig.4.3B). Hence,

both these targets interact with an essential and virulent set of proteins and focusing on such targets

can provide a mean to block important pathways required for A. baumannii virulence and survival.

We observed that there is no experimental evidence, which can support the association of these

proteins with host. We must believe as both these proteins are localized at the surface of the

pathogen and revealed as an effective immunogen, they must be explored for their interaction with

the host.

4.4.8. Structure Prediction and Evaluation

To view topology of epitopes on the protein surface, the final set of shortlisted proteins bearing

our epitope of interest was subjected to four online structure prediction tools. The best model

generated by each of these tools was selected, and evaluated through different online structure

evaluation servers. The model with low residue number in disallowed region, highest residues in

the most favorable region, greater Errat and Verify-3D scores and lesser Prosa Z-score was

considered as the most appropriate model.

126

Table 4.1. Prioritized epitopes that can elicit both humoral and Cell mediated immunity.

Proteins Epitopes that can

elicit both B and

T-Cell immunity

Total

Alleles

binding

Start

A.A

End

A.A

IC50 value

(MHCPred

<100 nm)

Virulency

Virulent

Pred

(>0.5)

Antigenicity

(VaxiJen

>0.4)

Allergenicity

Ligand-gated

channel protein

VRYQEVESS 40 513 521 95.9 1.1 1.27 Non-allergen

CsuB LQNNTRRMK

17 92 100 90.4 1.1 1.32 Non-allergen

EpsA FYLNDQPVS 17 182 190 1.36 1.1 0.8 Non-allergen

127

For CsuB, protein model generated by Swiss-Model was selected, as number of residues in

disallowed region was 0, residues in most favorable region was maximum compare to models

generated by other servers. The scores of Errat quality factor, Verify-3D, and Prosa Z-score were

72.034, 82.35, and -5.4, respectively (S-Table 4.7). Similarly, in case of EpsA protein, the model

generated by Modweb was selected. Residues in most favored region were maximum, and only 2

residues in disallowed region were observed. The estimated Errat quality factor, Verify-3D, and

Prosa Z-score were 74.627, 74.627, and of -7.36, respectively (S-Table 4.7). Therefore, both these

structures were selected and minimized.

4.4.9. Pepitope Analysis

To unveil exo-membrane topology of epitopes, Pepitope server was used. The analysis revealed

that the surface of both epitopes are exposed, and not folded within the 3D structure of the proteins.

The minimized 3D structure and exo-membrane topology of both epitopes can be seen in Fig.4.3.

4.4.10. Epitopes Binding Mode and Interactions Analysis

The best docked peptide-protein complex after docking with GalaxyPepDock was retrieved and

analyzed further via UCSF Chimera, MOE, and Ligplot for analyzing binding mode and binding

interactions. In case of CsuB epitope, the peptide prefers to bind in the binding pocket formed by

chain A and chain B, while for EpsA the peptide prefers to bind in the cavity formed by chain C

and D of the DRB1*0101 allele. Both the epitopes were found deeply bounded in the binding

pockets of the allele and contribute in number of hydrogen and hydrophobic interactions,

suggesting the stability of the docked complex. Furthermore, hydrogen bond interactions of the

epitopes with backbone and side chains residues of the receptor were also investigated with

varying distances (Fig.4.4). Interaction details between CsuB and EpsA epitopes and the receptor

based on MOE analysis are illustrated in S-Fig.4 and S-Fig.5, respectively. Ligplot demonstrated

that for CsuB, residues of chain A involved in hydrogen bonding with the peptide atoms such as

Ser51, Asn60, Gly56, Asp64, and Gln7, while those of chain B involves Asn258, and Arg247. For

EpsA, residues of chain C, such as Gln373, Asn426, Asn433, while Arg611, Tyr600, Gln604, and

Arg611 of chain D were found to interact via hydrogen bonds with the peptide. The Ligplot 2D

depiction of both docked epitopes can be seen in S-Fig.4.6.

128

4.4.11. Methodology Validation through a Negative Control

Sixty-two protein sequences were declared as virulent in complete proteome of M. pneumonia.

Subcellular localization and physicochemical characterization revealed five proteins illustrated in

S-Table 4.8 as potential candidates for epitope mapping. Only one protein, an adhesion protein,

was found to harbor both B-cell and T-cell epitopes. However, the epitope is non-binder of

DRB1*0101, therefore, not selected. Hence, being a negative control no potential peptide vaccine

candidates were predicted for M. pneumonia.

129

Fig.4.3. Minimized 3D structure and pepitope analysis revealing expose topology of the epitopes for CsuB (A) and EpsA (B) proteins.

130

Fig.4.4. The binding pose of CsuB (left) and EpsA (right) epitope in the binding pocket of DRB1*0101 allele. Receptor protein is in light gray while

epitopes are in cyan color. Hydrogen bonds between receptor and epitopes with distances are also shown.

131

4.4. Conclusions

In conclusion, the current immunoinformatics methodology unravels two peptides of virulent and

antigenic nature, which are highly conserved in all sequenced strains of A. baumannii.

Physicochemical characterization favors both the peptide proteins, while their interacting network

depicts stronger association with several proteins involved in vital metabolic pathways of the

organism, indispensable for pathogenicity and survival. The shortlisted epitope of the proteins can

elicit both humoral and cell-mediated immunity, are non-allergen and supported stable complex

with the most prevalent allele in human population-the HLA DRB1*0101. We believe that these

putative candidates will flourish the field of vaccinology against A. baumannii, in addition, to

targets reported in recent literature. Therefore, both the protein candidates must be explored in

future for their immune protection efficacy in animal models. Lastly, we recommend the designed

framework as simple, user-friendly pipeline, and can be extended to other infectious bacterial

pathogens for highlighting vaccine candidates and eliminating the risk of antibiotic resistance.


S-Fig.4.1.Venn diagram representation of prioritized five proteins. Each color in the diagram

represents different protein category. The 70 virulent proteins were filtered to ten outer membrane

and extracellular (O.M and E.C) proteins, followed by homology filter that ensures all 10 proteins

are human-non-homologous proteins (H.N.H.Ps). Molecular weight (MW) calculation of proteins

provided nine proteins with weight less than 110 kDa. Transmembrane helices (T.M.Hs) were

found less than one for eight proteins while only five proteins were found to have adhesive

probability (A.D.P).This Venn diagram is generated via Jvenn (http://bioinfo.genotoul.fr/jvenn/).

S-Fig.4.2. Representation of conserved, antigenic and virulent peptide sequence of CsuB (A) and

EpsA (B) protein.

S-Fig.4.3. Interactome analysis. A. The prioritized CsuB protein shows direct and indirect

interactions with other proteins of Csu operon (csuC, csuD, csuE, csuA/B, csuA), flagellar and pili

proteins (HMPREF0010_00250 and HMPREF0010_00251) and sugar transportation channel

proteins (HMPREF0010_00450, HMPREF0010_02773).B. Similarly, EpsA

(HMPREF0010_03288) protein interact with tyrosine kinase family of proteins

(HMPREF0010_03289, HMPREF0010_00298), ATPases (HMPREF0010_01283), sugar

transferases (HMPREF0010_03276), and amino acid transferases.

http://bioinfo.genotoul.fr/jvenn/

132

S-Fig.4.4. 2D depiction of docked CsuB prioritized epitope in the binding groove of DRB1*0101

allele using MOE.

S-Fig.4.5. 2D depiction of docked EpsA prioritized epitope in the binding groove of DRB1*0101

allele using MOE.

S-Fig.4.6. 2D depiction of docked CsuB (left) and EpsA (right) epitope in binding pocket of

DRB1*0101 using Ligplot.

S-Table 4.1. Seventy virulent proteins of A. baumannii reported by VFDB.

S-Table 4. 2. Ten virulent proteins localized in outer membrane or extracellular regions predicted

by subcellular localization tools.

S-Table 4.3. Five proteins prioritized for vaccine candidate identification.

S-Table 4.4. B and T-cells epitopes of prioritized five proteins.

S-Table 4.5. Characterization of selected B-cell epitopes of the ligated gated channel, CsuB and

EpsA protein.

S-Table 4.6. Interacting network of prioritized CsuB and EpsA proteins based on STRING

analysis.

S-Table 4.7. Structure evaluation of models generated by different online tools. The evaluation

was done on the basis of Procheck, Errat, Verify-3D, and Prosa- Z score.

S-Table 4.8. Five potential prioritized vaccine proteins against M. pneumonia.

4.6. References

Barh, D., Barve, N., Gupta, K., Chandra, S., Jain, N., Tiwari, S., … others. (2013). Exoproteome

and secretome derived broad spectrum novel drug and vaccine candidates in Vibrio cholerae

targeted by Piper betel derived compounds. PloS One, 8(1), e52773.

Barh, D., Misra, A. N., Kumar, A., & Vasco, A. (2010). A novel strategy of epitope design in

Neisseria gonorrhoeae. Bioinformation, 5(2), 77.



87-99.

133

Bentancor, L. V, Routray, A., Bozkurt-Guzel, C., Camacho-Peiro, A., Pier, G. B., & Maira-Litran,

T. (2012). Evaluation of the Trimeric Autotransporter Ata, as a Vaccine Candidate Against

Acinetobacter baumannii Infections. Infection and Immunity, IAI-06096.

Boll, J. M., Tucker, A. T., Klein, D. R., Beltran, A. M., Brodbelt, J. S., Davies, B. W., & Trent,

M. S. (2015). Reinforcing lipid A acylation on the cell surface of Acinetobacter baumannii

promotes cationic antimicrobial peptide resistance and desiccation survival. MBio, 6(3),

e00478-15.

Brossard, K. A., & Campagnari, A. A. (2012). The Acinetobacter baumannii biofilm-associated

protein plays a role in adherence to human epithelial cells. Infection and Immunity, 80(1),

228–233.



Chiang, M.H., Sung, W.C., Lien, S.P., Chen, Y.Z., Lo, A. F., Huang, J.H., … Chong, P. (2015).

Identification of novel vaccine candidates against Acinetobacter baumannii using reverse

vaccinology. Human Vaccines & Immunotherapeutics, 11(4), 1065–1073.

Choi, A. H. K., Slamti, L., Avci, F. Y., Pier, G. B., & Maira-Litran, T. (2009). The pgaABCD

locus of Acinetobacter baumannii encodes the production of poly-β-1-6-N-

acetylglucosamine, which is critical for biofilm formation. Journal of Bacteriology, 191(19),

5953–5963.

Choi, C. H., Lee, J. S., Lee, Y. C., Park, T. I., & Lee, J. C. (2008). Acinetobacter baumannii invades

epithelial cells and outer membrane protein A mediates interactions with epithelial cells. BMC


Chou, P. Y., & Fasman, G. D. (1977). Secondary structural prediction of proteins from their amino

acid sequence. Trends in Biochemical Sciences, 2(6), 128–131.

Chung, E. H. (2014). Vaccine allergies. Clinical and Experimental Vaccine Research, 3(1), 50–

57.



134

Doytchinova, I. A., & Flower, D. R. (2007). VaxiJen: a server for prediction of protective antigens,

tumour antigens and subunit vaccines. BMC Bioinformatics, 8(1), 4.

Eijkelkamp, B. A., Stroeher, U. H., Hassan, K. A., Papadimitrious, M. S., Paulsen, I. T., Brown,

M. H., & Lo, R. (2011). Adherence and motility characteristics of clinical Acinetobacter

baumannii isolates. FEMS Microbiology Letters, 323(1), 44–51.



EL-Manzalawy, Y., Dobbs, D., & Honavar, V. (2008). Predicting linear B-cell epitopes using

string kernels. Journal of Molecular Recognition: An Interdisciplinary Journal, 21(4), 243–

255.

Emini, E. A., Hughes, J. V, Perlow, D., & Boger, J. (1985). Induction of hepatitis A virus-

neutralizing antibody by a virus-specific synthetic peptide. Journal of Virology, 55(3), 836–

839.

Erac, B., Yilmaz, F. F., Hosgor Limoncu, M., Ozturk, I., & Aydemir, S. (2014). Investigation of

the virulence factors of multidrug-resistant Acinetobacter baumannii isolates. Mikrobiyol

Bul, 48(1), 70–81.

Fattahian, Y., Rasooli, I., Gargari, S. L. M., Rahbar, M. R., Astaneh, S. D. A., & Amani, J. (2011).

Protection against Acinetobacter baumannii infection via its functional deprivation of biofilm

associated protein (Bap). Microbial Pathogenesis, 51(6), 402–406.

Gaddy, J. A., Arivett, B. A., McConnell, M. J., Lopez-Rojas, R., Pachon, J., & Actis, L. A. (2012).

Role of acinetobactin-mediated iron acquisition functions in the interaction of Acinetobacter

baumannii ATCC 19606T with human lung epithelial cells, Galleria mellonella caterpillars

and mice. Infection and Immunity, IAI-06279.

Garcon, N., Stern, P. L., Cunningham, A. L., & Stanberry, L. R. (2011). Understanding modern

vaccines: perspectives in vaccinology. Elsevier.

Garg, A., & Gupta, D. (2008). VirulentPred: a SVM based prediction method for virulent proteins

in bacterial pathogens. BMC Bioinformatics, 9(1), 62.

135



Goel, V. K., & Kapil, A. (2001). Monoclonal antibodies against the iron regulated outer membrane

proteins of Acinetobacter baumannii are bactericidal. BMC Microbiology, 1(1), 16.

Guan, P., Doytchinova, I. A., Zygouri, C., & Flower, D. R. (2003). MHCPred: a server for

quantitative prediction of peptide-MHC binding. Nucleic Acids Research, 31(13), 3621–

3624.




He, X., Lu, F., Yuan, F., Jiang, D., Zhao, P., Zhu, J., … Lu, G. (2015). Biofilm formation caused

by clinical Acinetobacter baumannii isolates is associated with over-expression of the

AdeFGH efflux pump. Antimicrobial Agents and Chemotherapy, AAC-00877.

He, Y., Xiang, Z., & Mobley, H. L. T. (2010). Vaxign: the first web-based vaccine design program

for reverse vaccinology and applications for vaccine development. BioMed Research

International, 2010.



Huang, W., Yao, Y., Long, Q., Yang, X., Sun, W., Liu, C., … others. (2014). Immunization against

multidrug-resistant Acinetobacter baumannii effectively protects mice in both pneumonia

and sepsis models. PloS One, 9(6), e100727.

Iqbal, S., Shamim, A., Azam, S. S., & Wadood, A. (2016). Identification of potent inhibitors for

chromodomain-helicase-DNA-binding protein 1-like through molecular docking studies.

Medicinal Chemistry Research, 25(12), 2924–2939.

Jang, M. J., Kim, J.E., Chung, Y. H., Lee, W. B., Shin, Y. K., Lee, J. S., … Park, Y.M. (2011).

Dendritic cells stimulated with outer membrane protein A (OmpA) of Salmonella

typhimurium generate effective anti-tumor immunity. Vaccine, 29(13), 2400–2410.

136

Joly-Guillou, M.L. (2005). Clinical impact and pathogenicity of Acinetobacter. Clinical

Microbiology and Infection, 11(11), 868–873.

Karplus, P. A., & Schulz, G. E. (1985). Prediction of chain flexibility in proteins.

Naturwissenschaften, 72(4), 212–213.



Kim, S. W., Choi, C. H., Moon, D. C., Jin, J. S., Lee, J. H., Shin, J.H., … others. (2009). Serum

resistance of Acinetobacter baumannii through the binding of factor H to outer membrane

proteins. FEMS Microbiology Letters, 301(2), 224–231.

Kolaskar, A. S., & Tongaonkar, P. C. (1990). A semi-empirical method for prediction of antigenic

determinants on protein antigens. FEBS Letters, 276(1–2), 172–174.

Krogh, A., Larsson, B., Von Heijne, G., & Sonnhammer, E. L. L. (2001). Predicting

transmembrane protein topology with a hidden Markov model: application to complete

genomes. Journal of Molecular Biology, 305(3), 567–580.

Laskowski, R. A., MacArthur, M. W., Moss, D. S., & Thornton, J. M. (1993). PROCHECK: a

program to check the stereochemical quality of protein structures. Journal of Applied

Crystallography, 26(2), 283–291.

Laskowski, R. A., & Swindells, M. B. (2011). LigPlot+: multiple ligand-protein interaction

diagrams for drug discovery. American Chemical Society. 2778-2786.

Laxminarayan, R., Duse, A., Wattal, C., Zaidi, A. K. M., Wertheim, H. F. L., Sumpradit, N., …

others. (2013). Antibiotic resistance—the need for global solutions. The Lancet Infectious

Diseases, 13(12), 1057–1098.

Lee, H., Heo, L., Lee, M. S., & Seok, C. (2015). GalaxyPepDock: a protein--peptide docking tool

based on interaction similarity and energy optimization. Nucleic Acids Research, 43(W1),

W431-W435.

Li, W., Joshi, M. D., Singhania, S., Ramsey, K. H., & Murthy, A. K. (2014). Peptide vaccine:

progress and challenges. Vaccines, 2(3), 515–536.

137

Lin, L., Tan, B., Pantapalangkoor, P., Ho, T., Baquir, B., Tomaras, A., … others. (2012). Inhibition

of LpxC protects mice from resistant Acinetobacter baumannii by modulating inflammation

and enhancing phagocytosis. MBio, 3(5), e00312-12.

Lin, L., Tan, B., Pantapalangkoor, P., Ho, T., Hujer, A. M., Taracila, M. A., … Spellberg, B.

(2013). Acinetobacter baumannii rOmpA vaccine dose alters immune polarization and

immunodominant epitopes. Vaccine, 31(2), 313–318.

Luo, G., Lin, L., Ibrahim, A. S., Baquir, B., Pantapalangkoor, P., Bonomo, R. A., … Spellberg, B.

(2012). Active and passive immunization protects against lethal, extreme drug resistant-

Acinetobacter baumannii infection. PloS One, 7(1), e29446.

Mayrose, I., Penn, O., Erez, E., Rubinstein, N. D., Shlomi, T., Freund, N. T., … others. (2007).

Pepitope: epitope mapping from affinity-selected peptides. Bioinformatics, 23(23), 3244–

3246.



130–155.

McConnell, M. J., Rumbo, C., Bou, G., & Pachon, J. (2011). Outer membrane vesicles as an

acellular vaccine against Acinetobacter baumannii. Vaccine, 29(34), 5705–5710.

Moriel, D. G., Beatson, S. A., Wurpel, D. J., Lipman, J., Nimmo, G. R., Paterson, D. L., &

Schembri, M. A. (2013). Identification of novel vaccine candidates against multidrug-

resistant Acinetobacter baumannii. PloS One, 8(10), e77631.

Morris, A. L., MacArthur, M. W., Hutchinson, E. G., & Thornton, J. M. (1992). Stereochemical

quality of protein structure coordinates. Proteins: Structure, Function, and Bioinformatics,

12(4), 345–364.

Mortensen, B. L., & Skaar, E. P. (2012). Host--microbe interactions that shape the pathogenesis

of Acinetobacter baumannii infection. Cellular Microbiology, 14(9), 1336–1344.

Naz, A., Awan, F. M., Obaid, A., Muhammad, S. A., Paracha, R. Z., Ahmad, J., & Ali, A. (2015).

Identification of putative vaccine candidates against Helicobacter pylori. Helicobacter, 3, 1.

138

Parker, J. M. R., Guo, D., & Hodges, R. S. (1986). New hydrophilicity scale derived from high-

performance liquid chromatography peptide retention data: correlation of predicted surface

residues with antigenicity and X-ray-derived accessible sites. Biochemistry, 25(19), 5425–

5432.









Research, 33(suppl_1), D501-D504.

Ramakrishnan, C., & Ramachandran, G. N. (1965). Stereochemical criteria for polypeptide and

protein chain conformations: II. Allowed conformations for a pair of peptide units.

Biophysical Journal, 5(6), 909–933.

Russo, T. A., Luke, N. R., Beanan, J. M., Olson, R., Sauberan, S. L., MacDonald, U., …

Campagnari, A. A. (2010). The K1 capsular polysaccharide of Acinetobacter baumannii

strain 307-0294 is a major virulence factor. Infection and Immunity, 78(9), 3993–4000.

Sachdeva, G., Kumar, K., Jain, P., & Ramachandran, S. (2004). SPAAN: a software program for

prediction of adhesins and adhesin-like proteins using neural networks. Bioinformatics, 21(4),

483–491.



Singh, H., & Raghava, G. P. S. (2001). ProPred: prediction of HLA-DR binding sites.


Singh, H., & Raghava, G. P. S. (2003). ProPred1: prediction of promiscuous MHC Class-I binding

sites. Bioinformatics, 19(8), 1009–1014.

139

Singh, R., Garg, N., Shukla, G., Capalash, N., & Sharma, P. (2016). Immunoprotective efficacy of

Acinetobacter baumannii outer membrane protein, FilF, predicted in silico as a potential

vaccine candidate. Frontiers in Microbiology, 7, 158.


(2014). STRING v10: protein--protein interaction networks, integrated over the tree of life.


Tomaras, A. P., Dorsey, C. W., Edelmann, R. E., & Actis, L. A. (2003). Attachment to and biofilm

formation on abiotic surfaces by Acinetobacter baumannii: involvement of a novel

chaperone-usher pili assembly system. Microbiology, 149(12), 3473–3484.

Valencia, R., Arroyo, L. A., Conde, M., Aldana, J. M., Torres, M.-J., Fernandez-Cuenca, F., …

others. (2009). Nosocomial outbreak of infection with pan--drug-resistant Acinetobacter

baumannii in a tertiary care university hospital. Infection Control & Hospital Epidemiology,

30(3), 257–263.

Wattam, A. R., Abraham, D., Dalay, O., Disz, T. L., Driscoll, T., Gabbard, J. L., … others. (2013).

PATRIC, the bacterial bioinformatics database and analysis resource. Nucleic Acids

Research, 42(D1), D581-D591.



W407-W410.

Wizemann, T. M., Adamou, J. E., & Langermann, S. (1999). Adhesins as targets for vaccine

development. Emerging Infectious Diseases, 5(3), 395.

Wu, S., Skolnick, J., & Zhang, Y. (2007). Ab initio modeling of small proteins by iterative

TASSER simulations. BMC Biology, 5(1), 17.






140




Zhang, L., Huang, Y., Zou, Z., He, Y., Chen, X., & Tao, A. (2012). SORTALLER: predicting

allergens using substantially optimized algorithm on allergen family featured peptides.


141

Chapter # 5

Comparative Subtractive Proteomics Based Ranking for

Antibiotic Targets against the Dirtiest Superbug:

Acinetobacter baumannii

5.1. Abstract

MDR A. baumannii is indeed to be the most successful nosocomial pathogen responsible for

myriad infections in modern health care system. Computational methodologies based on genomics

and proteomics proved to be powerful tools for providing substantial information about different

aspects of A. baumannii biology that made it possible to design new approaches for treating multi,

extensive and total drug resistant isolates of A. baumannii. In this current approach, 35 completely

annotated proteomes of A. bauamnnii were filtered through a comprehensive subtractive

proteomics pipeline for broad-spectrum drug candidates. In total, 10 proteins (KdsA, KdsB, LpxA,

LpxC, LpxD, GpsE, PhoB,UvrY, KdpE and OmpR) could serve as ideal candidates for designing

novel antibiotics. The work was extended with KdsA enzyme for structure information, prediction

of intrinsic disorders, active site details, and structure based virtual screening of library containing

natural product-like scaffolds. Most of the enzyme structure has fixed three-dimensional

conformation. The selection of inhibitor for KdsA enzyme was based on druglikeness,

pharmacokinetics and docking scores. Compound-4636 (5-((3-chloro-5-methyl-2-phenyl-2,3-

dihydro-1H-pyrazol-4-yl)methoxy)-2-(((1-hydroxy-4-methylpentan-2-yl)amino)methyl)phenol)

was revealed as the most potent inhibitor against A. baumannii KdsA enzyme having GOLD

fitness score of 77.68 and Autodock binding energy of -6.2 kcal/mol. The inhibitor completely

follows Lipinski rule of five, Ghose rule, and Egan rule. MD simulation for KdsA and KdsA-4636

complex was performed for 100 ns to unveil what conformational changes the enzyme underwent

in the absence and presence of the inhibitor, respectively. The average RMSD for both systems

was found 3.5 Å, which signifies stable structure of the enzyme in both bounded and unbounded

states. Absolute binding energy using MM/GBSA reflected high affinity and vigorous interactions

of the inhibitor with enzyme active residues. Findings of the current study could open up new

142

avenues for experimentalists to design new potent antibiotics by targeting the targets screened in

this study.

5.2. Introduction

A. baumannii is a Gram-negative bacterial pathogen associated primarily with nosocomial

infections (McConnell, Actis, & Pachon, 2013). Multidrug-resistant A.baumannii (MDRAB) is a

major health concern that causes several different kinds of infection including ventilator associated

pneumonia, skin and soft infection, meningitis, and urinary tract infection in immune compromised

patients (Gonzalez-Villoria & Valverde-Garduno, 2016). A. baumannii is placed in the most

critical group of MDR bacteria that pose greater threat to patients in hospitals, nursing homes, and

among those who use blood catheters and ventilators (WHO, 2017). Alone in the United States,

this pathogen is responsible for estimated 12,000 deaths in which 500 are because of MDR strains

(CDC, 2013). A. baumannii rapidly accumulates resistance to multiple antibiotics and the

incidence of its MDR phenotypes has continued to increase worldwide (L.K. Chen et al., 2017).

Resistance to carbapenems, which is the only effective antibiotic against MDRAB infections, is

also reported rapidly (Higgins, Dammhayn, Hackel, & Seifert, 2009). Although, colistin is used as

last resort for treatment of MDRAB, however, its frequent use can cause nephrotoxicity or

neurotoxicity (Falagas, Kasiakou, & Saravolatz, 2005). Development of novel antibiotics that

target this pathogen will reduce the deaths due to its resistant infections around the globe (WHO,

2017).

The first and most important step in the discovery of novel drugs is the identification of new drug

targets (Katsila, Spyroulias, Patrinos, & Matsoukas, 2016; R. Gupta, Pradhan, Jain, & Rai, 2017).

Target identification improves efficacy, avoid side effects and help in understanding drug

candidates mode of action (Katsila, Spyroulias, Patrinos, & Matsoukas, 2016) . In post genomic

era, the applications of bioinformatics have brought out the likelihood of integrating data from

genomics, metabolomics and proteomics to detect new drug targets (Butt et al., 2012; Sanober,

Ahmad, & Azam, 2017). During the past decade or so, subtractive proteomics approach is

extensively used as a tool for identifying targets that could serve as a promising candidate for

future drug designing (Sakharkar, Sakharkar, & Chow, 2004; Sarangi, Lohani, & Aggarwal, 2015).

As a proof, subtractive proteomics based prediction of potent targets against Pseudomonas species

has been experimentally validated (Perumal, Lim, Chow, Sakharkar, & Sakharkar, 2008). Once

143

identified, the target is subjected to ligand or structure based virtual screening of chemical

molecule libraries originated from natural sources or synthesized by synthetic chemists (Vyas,

Jain, Jain, & Gupta, 2008). This is vital for screening of potent hits against the target and

optimizing lead like compounds (Shoichet, 2004). Structure based virtual screening, by using most

widely employed techinique of molecular docking, has been conducted to delineate the correct

poses from incorect one based on the scoring function (Morris & Lim-Wilby, 2008).The functional

and physical behavior of the best characterized molecules could be further deciphered using MD

simulations (Hospital, Goni, Orozco, & Gelpi, 2015). MD studies aid in deciphering the

conformations of ligand which is stable with the protein (Haq, Abro, Raza, Liedl, & Azam, 2017).

In addition, these studies shed light on the protein internal motions and how conformational

changes in the protein play role in their functions (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016).

The MD simulations could be combined further with molecular mechanics force fields for

calculating absolute binding free energies of the ligands (Abro & Azam, 2016).

In the current approach, we filtered 35 completely sequenced strains of A. baumannii using a

comprehensive subtractive proteomics to detect highly specific and selective drug targets. The

pipeline identified 10 proteins (KdsA, KdsB, LpxA, LpxC, LpxD, GspE, PhoB, UvrY, KdpE,

OmpR) and could be ideal targets for future discovery against MDRAB. Here, we only ensued

with KdsA as very little is known about this enzyme as an antibacterial drug target. KdsA is a

broad-spectrum enzyme, which catalyzes the condensation of A5P and phosphoenolpyruvate

(PEP) to make 3-deoxy-D-manno-octulosonate 8-phosphate (KDO8P) molecule in 3-deoxy-D-

manno-octulosonate (KDO) biosynthesis pathway (Cipolla et al., 2009) (Fig.5.1). KDO is a vital

component of Gram-negative bacteria LPS (Lipopolysaccharide) and its inhibition ultimately leads

cell death (Yi, 2009). Several crystal structures of KdsA are available in structure databases and

were used as a template for structure modeling of A. baumannii KdsA enzyme (Yi, 2009) since its

structure is absent. The enzyme structure was then used in structure based virtual screening of a

natural inhibitors library. The selection of best hit was based on docking scores, druglikeness, and

pharmacokinetics of inhibitors. MD simulations were used further for understanding enzyme and

complex dynamics and for analyzing their physical properties namely RMSD, RMSF, Beta Factor

(β-factor), and Rg (Ahmad, Raza, Uddin, & Azam, 2017). Absolute binding free energy for the

ligand was determined using MM/GBSA to unveil its affinity for enzyme active site (Abro &

Azam, 2016; Ahmad, Raza, Uddin, & Azam, 2017). The different energy contributions like

144

electrostatic, van der Waals and solvation energies of the complex have been explored. Further,

total energy decomposition was carried out for identifying hot spot amino acids that contribute

highly to overall binding energy of the complex (Abro & Azam, 2016). The step by step flow of

the study can be found in Fig.5.2.

Fig.5.1. Biosynthetic pathway for KDO in four sequential steps. The four Kds enzymes involved

in the pathway are also shown. KDO is a part of lipid A molecule of Gram-negative bacteria

outer membrane lipopolysaccharide.

145

Fig.5.2. Complete step by step flow of the methodology employed in the current study.

146


5.3.1. Proteome Subtraction

A subtractive proteomic approach was applied for screening highly conserved drug candidates

against A. baumannii. Total of 35 strains of A. baumannii (till the writing of the manuscript) was

processed through the designed in silico pipeline. Proteome of each strain was subjected to CD-

HIT suit with sequence identity cut-off set to 80% (Li & Godzik, 2006). The deduced non-

redundant sequences were aligned against human proteome (TaxID: 9606) to extract pathogen

specific proteins (Pruitt, Tatusova, & Maglott, 2005). The primacy of homology check was to

avoid cross reactivity of the drugs with the host proteins and prevent its binding with active site of

host homologous proteins (Pruitt, Tatusova, & Maglott, 2005). Proteins for which no hit was

observed or have identity ≤ 35% were categorized as host non-homologs (Azam & Shamim, 2014).

To pool out pathogen essential proteins, Blastp analysis was performed for host non-homologous

proteins against DEG database with E-value and homology threshold of ≥ 0.0001 and ≥ 30%,

respectively (Zhang, Ou, & Zhang, 2004). Afterward, essential protein candidates were mapped to

unique metabolic pathways using KAAS server (Moriya, Itoh, Okuda, Yoshizawa, & Kanehisa,

2007). Comparative sub-cellular localization was accomplished for essential proteins to recognize

cytoplasmic proteins (Azam & Shamim, 2014; Azam & Shamim, 2014). This analysis was based

on three online subcellular localization tools: PsortB (N. Y. Yu et al., 2010), Cello (C.S. Yu, Chen,

Lu, & Hwang, 2006), and Cello2Go (C.S. Yu et al., 2014). Conservation of unique proteins was

performed to pinpoint those which can be targeted for designing broad-spectrum drugs.

5.3.2. Drug Target Prioritization and Selection

Conserved cytoplasmic proteins were subjected to virulence factor analysis, which Blastp the

proteins against full dataset of VFDB (L. Chen et al., 2005). Proteins with identity ≥ 35 % and bit

score > 100 were considered as virulent and physicochemically characterized to investigate

experimentally favored proteins. The virulent proteins were characterized by different

physicochemical attributes using ProtParam (Gasteiger et al., 2005). The proteins were evaluated

for molecular weight, atomic composition, theoretical pI, amino acid composition, estimated half-

life, instability index, aliphatic index extinction coefficient and grand average of hydropathicity

(GRAVY). Physicochemical favored proteins were further examined for cellular interactome.

STRING was used to interpret direct (physical) and indirect (functional) interactions of the target

147

proteins (Szklarczyk et al., 2014). Interactions with confidence score of 0.9 were only considered.

Proteins, which strongly interact with essential and vital proteins of the pathogen make backbone

of cellular functions and can aid in understanding molecular basis of biological processes, thus

selected for further evaluation (Baseer, Ahmad, Ranaghan, & Azam, 2017). Above mentioned in

silico filters brought forth several vital drug candidates against A. baumannii. We focused on the

screened proteins, which play significant contributions in pathogen survival and for which very

limited literature data of antibiotics designing and dyanamics are available. Since proteins of KDO

family are widely distributed among Gram-negative bacteria, they can serve as promising

antimicrobial targets with highly selectivity (Yi, 2009). KdsA protein catalyzes the first step of

subpathway responsible for biosynthesis of KDO, resulting in formation of inner core of LPS

(Cipolla et al., 2009; Yi, 2009). Previous studies demonstrated KdsA protein as potential

therapeutic target through subtractive genomics, metabolic pathway analysis and comparative

studies (Cipolla et al., 2009; Yi, 2009; S. K. Gupta, Gross, & Dandekar, 2016), therefore, selected

for subsequent computational studies.

5.3.3. Comparative Structure Modeling And Evaluation

A comparative structure modeling approach was applied for prediction of optimal model of the

KdsA protein. The crystal structure of 2-keto-3-deoxy-d-manno-octulosonate-8-phosphate

synthase (KDO8Ps) from P. aeruginosa (Nelson, Kelleher, Robinson, Reiling, & Asojo, 2013)

was used as a template to build homology model of KdsA protein using Modeller 9v14 (Fiser &

Sali, 2003). The criteria for selection of KDO8Ps as a template was based on its query coverage

(100%), sequence identity (74%) and E-value (5e-158). Similarly, sequence identity of A.

baumannii KdsA with Neisseria meningitidis KdsA (PDB ID: 2QKF), Vibrio cholerae O1 Biovar

Eltor Str. N16961 KdsA (PDB ID: 3E9A), E. coli KdsA (PDB ID: 1G7U) and Haemophilus

influenza (PDB ID: 1060) was 60%, 62%, 67% and 60%, respectively. Initial sequence alignment

of A. baumannii KdsA and P. aeruginosa KDO8Ps was carried out using align2d function present

within the Modeller program. Twenty models for the KdsA protein was generated based on target-

template alignment and structural coordinates and top five were selected for further analysis. In

addition to Modeller, different web-based free protein structure prediction tools were also utilized

for comparative analysis. These servers include I-TASSER (Roy, Kucukural, & Zhang, 2010),

Mod-web (Pieper et al., 2006), Swiss-model (Guex & Peitsch, 1997), and Phyre2 (Kelley, Mezulis,

https://blast.ncbi.nlm.nih.gov/Blast.cgi#alnHdr_193506536



148

Yates, Wass, & Sternberg, 2015). From each server twenty models were generated and top ranked

model was selected for further analysis. Out of 120 models most reliable model for the target

protein was selected based on evaluation of structures predicted by Modeller and online structure

prediction servers. The selection criteria of the most optimal structure were primarily based upon

the Ramachandran plot results (Hooft, Sander, & Vriend, 1997). The structure for which most of

the residues mapped in most favorable region and least residue in disallowed region was

considered the best modelled structure for the KdsA protein (Abro & Azam, 2016). The structure

evaluation of the best model was additionally performed with Verify3D (Eisenberg, Lüthy, &

Bowie, 1997, ERRAT (Colovos & Yeates, 1993) and ProSA (Wiederstein & Sippl, 2007).

Furthermore, assessment of the structure was done based on energy functions (DOPE, GA341 and

molpdf score) (Azam & Shamim, 2014). Once the best model was selected, the structure was

subjected to energy minimization process for removal of possible close contacts and to relax the

modeled structure (Ahmad, Raza, Uddin, & Azam, 2017). Minimization was carried out using

UCSF Chimera (Pettersen et al., 2004) for 1500 steps (750 steps of steepest descent and 750 steps

of conjugate gradient with a dielectric constant of 1 under TFF (Ahmad, Raza, Uddin, & Azam,

2017). The target and template symmetry was verified with RMSD value. Secondary structure

elements for the best model was predicted using PDBSum (Laskowski, 2001).

5.3.4. KdsA Intrinsic Disorder Regions Prediction

KdsA enzyme intrinsic disorder regions were predicted using PrDOS (Protein Disorder Prediction

System) (Ishida & Kinoshita, 2007). This was important to analyze regions of the enzyme lie in

disorder region and what impact they have on overall enzyme function (Shaban et al., 2017).

5.3.5. Comparative Molecular Docking

As limited data of inhibitors against KdsA protein is available a ligand library of natural sources,

Asinex antibacterial library (http://www.asinex.com/antibacterial_compound_library-html/) was

used for identification of drug-like compounds against KdsA protein. In total, the library contained

4800 natural product-like scaffolds when accessed on 8/09/2016. Virtual screening of the library

was first done using Ligandscout (Wolber & Langer, 2005) to extract compounds which

completely follow Lipinski rule of five filter (MW ≤ 500, MLogP ≤ 4.15, HBD ≤ 5, HBA ≤ 10,

TPSA 40-130 Å2 ) (Lipinski, 2004). The shortlisted candidates were subjected to predocking

149

preparation phase where the compounds were energetically minimized and optimized using

MMFF94 force field (Halgren, 1996).

Active site mapping of the protein was accomplished using multiple sequence alignment approach

and different active site prediction servers including MetaPocket (http://projects.biotec.tu-

dresden.de/metapocket/) (Huang, 2009), GHECOM (http://strcomp.protein.osaka-

u.ac.jp/ghecom/) (Kawabata, 2010), and fpocket (http://fpocket.sourceforge.net/) (Le Guilloux,

Schmidtke, & Tuffery, 2009). Additional information for active site residue of the protein was

extracted multiple sequence alignment (MSA) (Raza, Sanober, Rungrotmongkol, & Azam, 2017).

A consensus active site residue of Gln111 was considered for docking with GOLD software (Jones,

Willett, Glen, Leach, & Taylor, 1997) and AutoDock Vina (Trott & Olson, 2010).

Molecular docking of drug-like compounds into KdsA protein active site was established through

GOLD 5.2 software. GOLD utilizes a genetic algorithm to investigate flexibility of receptor

hydrogen and ligand conformational flexibility. Docking was carried out with default parameters

of the wizard. These parameters include population size (1000), niche size (2), pressure (1,1),

number of islands (1), number of operations (10,000), mutate (100), crossover (100) and operator

weights for migrate (0). A set of 10 solutions for each ligand was generated using default algorithm

settings. The binding affinity of ligands toward protein active site was measured in term of GOLD

fitness score which is based on the following equation:

GOLDScore Fitness = Shb−ext + SvdW−ext + Shb−int + Sint…………………………………………(XI)

In the above equation, Shb−ext + SvdW−ext represent hydrogen and van der Waals interaction for

protein-ligand complex, respectively while Shb−int + Sint are intramolecular hydrogen bonds and

intramolecular strain in the substrate, respectively. To achieve consistency in the results, top hits

of the GOLD having score ≥ 50 were re-docked into the active site of protein using Auto-dock

Vina. Docking with Auto-dock Vina (Trott & Olson, 2010) was carried out on Intel Core (TM) i5

CPU M 540 @ 2.53 GHz with 32-bit Windows 8.1 as an operating system. Active site coordinates

used was the same for that of GOLD while the sphere (Å) along the X, Y and Z axis was set as 15

Å.

Post-docking analysis was accomplished in order to screen protein complexe with bounded ligand

that has best GOLD fitness score, binding energy, and drug-likeness. The selected compound was

http://projects.biotec.tu-dresden.de/metapocket/

http://projects.biotec.tu-dresden.de/metapocket/

http://strcomp.protein.osaka-u.ac.jp/ghecom/

http://strcomp.protein.osaka-u.ac.jp/ghecom/

http://fpocket.sourceforge.net/

150

characterized for pharmacokinetics using SwissADME (Daina, Michielin, & Zoete, 2017) and

preADMET (Lee et al., 2004) to investigate its biological fate.

Binding mode of the best characterized drug-like compound with high GOLD fitness score into

protein active site was deciphered through UCSF Chimera. The best docked ligand was further

analyzed in term of hydrogen bonds with the protein to investigate stability of the ligand with the

protein active site. Hydrogen bond analysis was carried out to label all the atoms lying with 5 Å

of the receptor protein active site involved in hydrogen bonding with the ligand through an

indigenously designed script in VMD (Humphrey, Dalke, & Schulten, 1996). The best

characterized drug-like compound with excellent pharmacokinetics and enrich hydrogen and

hydrophobic interaction was selected to decipher its dynamics.


MD simulations of A. baumannii KdsA enzyme and KdsA with the best docked drug-like

compound was performed to unveil mechanistic, dynamics and stability details of the target protein

with respect to the inhibitor. The complex was first subjected to initial system preparation phase

using AMBER 14 software (Case et al., 2014). For ligand processing, antechamber program of

AMBER was used while GAFF was employed as a forcefield (Ozpinar, Peukert, & Clark, 2010).

For KdsA protein, ff14SB force field was selected because of having improved rotamer

distribution for side chains and better empirical tweaks for backbone potentials (Maier et al.,

2015). Leap module of AMBER was used to record topology files for protein and inhibitor while

MPI and sander was employed to carry out subsequent processing. The system was

electrostatically neutralized by adding 3 Na+ ions and was placed in three-point transferable

intermolecular potential (TIP3P) water box of 12 Å size. The system was then subjected to a series

of seven steps of preprocessing, which can be divided into minimization, heating, pre-equilibrium,

pressure equilibration and post-equilibrium (Andleeb et al., 2016. In minimization phase, the

system was gradually minimized first by relaxing the hydrogen of entire system for 500 cycles;

followed by 1000 cycles to minimize water box with a restraint of 200 kcal/mol –Å2 . The system

was then subjected to 1000 cycles of minimization with a restraint of 5 kcal/mol –Å2 on carbon

alpha atoms. Then all non-heavy atoms the system were minimized using 300 cycles with restraint

of 100 kcal/mol –Å2. In the heating phase, the system underwent heating for 20 picoseconds (ps)

at a constant temperature of 300 K and 5 kcal/mol –Å2 restraint on carbon alpha atoms. Langevin

151

dynamics was used to maintain temperature with gamma value set to 1 (Paterlini & Ferguson,

1998), SHAKE algorithm was applied for constraining hydrogen bonds and NVT ensemble was

employed for heating system (Kräutler, Van Gunsteren, & Hünenberger, 2001). In pre-

equilibration, equilibration of the system was achieved for 100 ps with time step of 2 femtosecond

(fs) . In pressure equilibration phase, the system pressure was maintain using NPT ensemble along

with isotropic position scaling with restraint for alpha carbon position by 5 kcal/mol –Å2. The

pressure equilibrium of the system was attain further for 50 ps using same parameters with carbon

atoms restraint by 1 kcal/mol –Å2 for every 10 ps. In the post-equilibrium phase, using same set of

parameters the system was allowed to equilibrate for 1 ns. A total production run of 100 ns was

performed using NVT ensemble with Berendsen temperature coupling algorithm. For non-

bounded interaction a cut-off of 8.0 Å while for hydrogen bonds SHAKE algorithm was applied.

Trajectories generated by simulation were analyzed by ccptraj (Roe & Cheatham, 2013) program

of AMBER and parameters like RMSD, RMSF, radius of gyration, and β-factor were calculated.


Binding free energies of the system were calculated using MM/GBSA method of AMBER14

(Miller et al., 2012; Abro & Azam, 2016) . Computation of binding energies taken into account

the differences between binding free energy of the complex and that of inhibitor and targeted

protein (Abro & Azam, 2016). Ideally this free energy of binding can be represented as:

ΔGbind = ΔGcomplex _ [ΔGprotein + ΔGinhibitor]……………………………………………………(XII)

In equation (XII), ΔG is Gibb’s free energy computed for each term:

ΔG = Egas + ΔGsolv – TSsolute…………………………………………………………………..(XIII)

Where Egas is the MM energy from gas phase force field while taking into account internal energy,

van der Waals interaction and electrostatic energies. T and S represent temperature and ligand

entropy contribution, respectively. Egas can be further explained by an equation:

Egas = Eint + Eele + Evdw……………………………………………………………………..(XIV)

Calculation of term ΔGsolv in equation (XIII) is based on implicit solvent model and can be split

into electrostatic and non-polar contributions:

ΔGsolv = ΔGele + ΔGnp…………………………………………………………………………(XV)

152

ΔGele can be split further into electrostatic energy and polar solvation components,

ΔGele = Eele + ΔG…………………………………………………………………………..(XVI)

In equation (XV), the term ΔGnp represents non-electrostatic interactions and is proportional to the

molecule solvent accessible surface area,

ΔGnp=γsas+β…………………………………………………………………………………(XVII)

ΔGnp was calculated using LCPO of MM/GBSA method (Weiser, Shenkin, & Still, 1999). The

water probe radius was set to 1.4 Å. In MM/GBSA, the values for γ and β were 0.0072 kcal/mol.A2

and 0 kcal/mol, respectively. Binding free energy was decomposed per residue using implicit

solvent model. MMPBSA.py module of AMBER was used for approximation of binding free

energy at atomic level. This analysis aid in investigating the energy contributions of hot spot amino

acids involved in interaction with the ligand (Cipolla et al., 2009).


5.4.1. Drug Candidate’s Prioritization

A. baumannii is profoundly implicated as a major cause of nosocomial and community acquired

infections, conferring high risk of morbidity and mortality to patients infected with the pathogen

(Munoz-Price & Weinstein, 2008). Advances in sequencing technologies particularly the next

generation sequencing generate vast amount of genomic data. Till to date, 35 strains of A.

baumannii are completely sequenced and available at NCBI Genome database. In the current

investigation, the proteome of all these strains were processed through subtractive proteomics

approach to prioritized novel potential drug proteins against A. baumannii (S-Table 5.1). The total

proteins number in all the analyzed strains was calculated as 127175 while an average strain

proteome was observed to contain 3633 proteins. The low protein count was reported in D1279779

strain while highest number of proteins was revealed by R2090 strain. All the strains served as

input for subtractive proteomics framework, which step-wise subtract each proteome to dig out

protein that could be targeted for designing broad-spectrum drugs. Such a computational based

procedure is not only time friendly but could also excel the process of drug discovery against

notorious bacterial pathogen. The redundancy problem associated with the proteomes was

countered at redundancy check, where only non-redundant set of proteins was allowed to enter

153

homology filter. Redundant proteins (paralogs) arise due to duplication events during the course

of evolution and are highly prone to adaptive mutations (Sanober, Ahmad, & Azam, 2017). Non-

redundant sequences (orthologues), on the other hand are subject to speciation event with function

among strains tends to more conserved (Hassan et al., 2016). Total of 123537 non-redundant

proteins were yielded after CD-HIT analysis with an average of 3529 non-paralogs per strain. The

strain XH386 was observed to harbor the highest number of non-paralogous proteins (3732) while

strain D1279779 was found to contain the least number of non-redundant proteins (3272).

Designing drug, which could bind only to pathogen specific protein, was another consideration in

the pipeline. Binding of drugs to proteins with similar pockets results in non-specific drug targets

interaction and could lead to serious life threating complications (Azam & Shamim, 2014;

Sanober, Ahmad, & Azam, 2017). To avoid this situation, the conserved non-redundant proteins

were brought forth to the homology filter where pathogen specific proteins were pooled out by

aligning the orthologous proteins against human proteome. The host-homologous sequences were

discarded while host-non-homologous proteins were gathered (Azam & Shamim, 2014). Total of

103450 proteins was found to have identity ≤ 35 % with host proteome with an average of 2955

proteins in each strain. Likewise, the results of redundancy check, the strains XH386 and

D1279779 contain the highest (3153) and lowest (2704) number of proteins, respectively. As most

of the vaccines and antibiotics target essential pathways of infectious agents, describing and listing

essential genetic elements and proteins in an organism is considered a critical step in designing

therapeutic products (Sanober, Ahmad, & Azam, 2017). An average number of essential proteins

per strain was 803 with total essential proteome of 28105. These essential proteins are minimal set

of proteins required by A. baumannii to support cellular life in extreme stress conditions,

persistence in host environment and in causing infection (Azam & Shamim, 2014; Sanober,

Ahmad, & Azam, 2017). In context of drug and vaccine designing, this essential set have broader

potential because of their involvement in major biological pathways of the pathogen. Further

insights into functional annotation of the essential proteins with respect to their metabolic pathway

was achieved through KAAS server. On average, 224 metabolic pathways were mapped for each

strain with highest number of metabolic pathways were assigned to MDR-TJ, LAC-4 and ZW85-

1 strains. Metabolic pathways present exclusively in the pathogen were categorized as unique

metabolic pathways while those found in both host and pathogen were termed as common

pathways. The analysis mapped 39 unique pathways and 81 common pathways on average basis

http://www.ncbi.nlm.nih.gov/genome/403?genome_assembly_id=210824


154

in each strain. As unique metabolic pathways are relevant from drug discovery and vaccine point

of view, they were investigated further for distribution of essential proteins (Azam & Shamim,

2014; Hassan et al., 2016). Majority of the essential genes were found to be involved in

biosynthesis of secondary metabolites, biosynthesis of antibiotics, microbial metabolism in diverse

metabolisms and two component system. In the next step, proteins characterization was done to

screen protein targets with better potential to act as drug targets. Proteins localized in the cytoplasm

were preferred as potential drug target because of the fact that membrane proteins or secreted

proteins are not accessible to the drug for interaction (Sanober, Ahmad, & Azam, 2017). It was

found that around 33 proteins were localized intracellularly (cytoplasmic), 3 in outer membrane,

33 were present in inner membrane and 10 were unknown for localization in each proteome. The

unique cytoplasmic proteins of each strain was then subjected to conservation analysis where

completely conserved protein candidates for opted. Conservation of protein targets is

indispensable for broad-spectrum drug designing (Baseer, Ahmad, Ranaghan, & Azam, 2017). As

conserved protein are less likely to underwent variation and evolve slowly, they represent reliable

targets for drug designing (Hassan et al., 2016). The conservation analysis shortlisted 32 proteins

present in all the 35 completely sequenced strains and processed further along the pipeline.

5.4.2. Virulence Proteins Analysis

Relative to other Gram-negative bacterial pathogens, very limited information about A. baumannii

virulome is available (Howard, O’Donoghue, Feeney, & Sleator, 2012). The extraordinary success

of A. bauamnnii is recognized because of its virulent factors present in the core proteome of the

said pathogen. Virulent proteins aid in bacterial adherence to biotic and abiotic factors,

colonization, biofilm formation, escape from host immune responses and antibiotic actions and

persistence in environment (Baseer, Ahmad, Ranaghan, & Azam, 2017). Out of 32 proteins, 15

proteins were found virulent that are conserved in all the 35 strains of the pathogen and facilitate

the survival benefits in adverse environmental conditions (S-Table 5.2). In general, these virulent

proteins can be categorized as lipid A biosynthesis proteins (LpxA, LpxC and LpxD), 3-deoxy-D-

manno-octulosonate (KDO) biosynthesis proteins (KdsA, KdsB and KdsD), two component

system (OmpR, KdpE, UvrY, GlnG, RstA, PhoB,) and secretion system protein (GspE). LpxA,

LpxC and LpxD are soluble proteins and are involved in biosynthesis of lipid A molecule, an

integral component of Gram-negative bacterial lipopolysaccharides (LPS) (Cipolla et al., 2009).

155

LPS is also term as endotoxin and contribute greatly to overall stability of the membrane structure.

In addition, LPS protect bacteria from the attack of toxic chemicals promote surface adhesion,

interaction with predators and sensitivity of bacteriophages. LpxA, LpxB and LpxC catalyze fatty

acylation of UDP-GlcNAc, deacetylation of UDP-3-O-(acyl)-GlcNAc and addition of second R-

3-hydroxymyristate chain to deacetylation of UDP-3-O-(acyl)-GlcNAc to produce UDP-2,3-

diacyl-GlcN, respectively (Raetz & Whitfield, 2002). As lipid molecule is essential for growth in

majority of Gram-negative bacterial and high conserved distribution of the above mentioned lipid

A biosynthesis proteins, they could serve excellent targets for broad spectrum inhibitors designing

(Yi, 2009). The polysaccharide chains of LPS are linked to lipid A molecule through KDO. Usually

two molecules of KDO are required by Gram-negative bacteria to be incorporated into LPS for

proper growth; however, the numbers vary from species to species (Yi, 2009). Inhibiting KDO

biosynthesis can lead cell growth arrest due to accumulation of lipid A molecule precursors while

its results render the bacteria less pathogenic and more susceptible to antibiotics (Cipolla et al.,

2009;Yi, 2009). KDO biosynthesis pathway can be decomposed into four sequential steps: the

KdsD is devoted to catalyze isomerization of D-ribulose 5- phosphate (Ru5P) for production of D-

arabinose 5-phosphate (A5P), KdsA is employed for condensation of A5P and

phosphoenolpyruvate (PEP) to make 3-deoxy-D-manno-octulosonate 8-phosphate (KDO8P)

molecule, the third step is catalyzed by KdsC where phosphate esters are hydrolyzed to provide

KDO molecule and phosphate. In the last step, KDO molecule is activated by KdsB through adding

CMP from CTP (Cipolla et al., 2009;Yi, 2009). All the four proteins are essential for cell survival

and represent excellent drug candidates with high selectivity and specificity (Yi, 2009).Two

component system (TCS) is ubiquitous among bacteria and play a crucial role in sensing and

responding to intracellular and extracellular changes (Hoch & Silhavy, 1995). TCS is also

considered to play significant role in cell-cell communication and bacterial pathogenesis. The

OmpR protein functions by regulating the transcription of major outer membrane proteins (OmpF

and OmpC) (Aiba & Mizuno, 1990). KdpE protein is a member of Kdp system in bacteria and

homologous in function with OmpR protein (Walderhaug et al., 1992). UvrY protein, a key

member of UvrY/ BarA system, regulates carbon metabolism through CsrA/CsrB regulatory

system and increase virulence (Palaniyandi et al., 2012). Likewise, RstA proteins play similar

functions by regulating DNA templated transcription and is involved in phosphorelay signal

transduction system. RstA is also a potential drug target against A. baumannii (Russo et al., 2016.

http://www.ebi.ac.uk/QuickGO/GTerm?id=GO:0000160

http://www.ebi.ac.uk/QuickGO/GTerm?id=GO:0000160

156

GlnG is a nitrogen regulator protein and is a part of NtrB/NtrC system with major role in nitrogen

fixation and nitrogen utilization (Weglenski, Ninfa, Ueno-Nishio, & Magasanik, 1989). PhoB

(Phosphate regulon transcriptional regulatory protein) protein and act as a positive regulator for

phosphate regulon when there is limitation of phosphate (Lamarche, Wanner, Crepin, & Harel,

2008). Since all these proteins of TCS are highly selective and are indispensable for A. baumannii

growth, considering these targets form drug designing point of view can open up new avenues for

identification of novel drug candidates. The last screened virulent protein (GspE) belongs to the

type II secretion system (Lu, Korotkov, & Hol, 2014). Blocking the function of GspE can lead to

interrupted function of the said system, thus can influence negatively bacterial survival.


Physicochemical characterization of virulent proteins allowed filtration of proteins that could serve

as best drug proteins against the pathogen (Hassan et al., 2016; Baseer, Ahmad, Ranaghan, &

Azam, 2017). The prime parameter in this phase was to calculate the molecular weight of proteins.

Proteins exhibiting molecular weight < 110 kDa are considered ideal targets for subsequent wet

lab analysis due to their easy purification procedures (Baseer, Ahmad, Ranaghan, & Azam, 2017).

It was estimated that all the 15 virulent proteins have weight less than the threshold and can be

effectively used for novel drug development (S-Table 5.3). The next important character of

proteins was to disclose their stability index. Proteins with estimated stability index of < 40 were

regarded as stable while those having stability index > 40 were designated as unstable proteins

(Guruprasad, Reddy, & Pandit, 1990). Proteins with stable nature are usually considered in drug

discovery pipeline. Proteins stability for majority of the proteins was computed < 40 while for 5

proteins (LpxB, KdsD, KdtA, GlnG and RstA) the value was > 40 (S-Table 5.4). The GRAVY

index score for 10 stable proteins was negative, inferring their hydrophilic nature (Kyte &

Doolittle, 1982). The UvrY protein was hydrophobic because of its predicted positive value. The

in vivo half-life of shortlisted 10 proteins was predicted to estimate proteins stability and metabolic

fate during in vivo experimental investigation (Bachmair, Finley, & Varshavsky, 1986). The half-

life values were predicted in three model organisms: E. coli in vivo, yeast in vivo and human in

vitro. It was revealed that in all three model organisms, the predicted half-life duration was same.

The in vitro half-life of proteins is 30 hours in mammalian reticulocytes, in vivo half-life in E. coli

is 10 hours and 20 hours in yeast cells. Deciphering proteins extinction coefficient was important

157

to indicate how much light a protein absorbs in water at wavelength of 280 nm. The inclusion of

this analysis could be helpful in purifying a protein during purification stage of experimental

evaluations (Gill & Von Hippel, 1989). Two kind of values were produced, the first value is based

on the assumption that all cysteine values are considered as half cystine while the second value

represents no cysteine as half cystine. The first extinction coefficient value of the scrutinized 10

proteins ranged from 10,555 to 29,910. The values are indicator of high Trp and Tyr concentration

and low Cys concentration. Another factor that was considered to uncover proteins thermostability

is the estimation of aliphatic index. This value estimate the relative volume occupied by the

aliphatic amino acid site chains (leucine, isoleucine, alanine and valine) in protein globular

structure (Ikai, 1980). Aliphatic index value for proteins ranged from 111.39 to 95.54, indicating

the high thermostability of proteins. The theoretical pl value specify the pH at which a particular

protein carries no net electrical charge (Azam & Shamim, 2014). Since proteins can be purify

according to its isoelectric point, computational estimation of this value is important from wet lab

analysis of view. The theoretical pl value for proteins ranged between 6.37 to 4.86 with an average

value of 5.75, demonstrating the acidic nature of drug proteins.

5.4.5. Interacting Networks of Targeted Proteins

Unveiling cellular interacting networks for proteins can provide unprecedented opportunities to

reveal critical cross talks among therapeutic candidates (Hassan et al., 2016). This procedure was

vital in the framework as cellular interactome of targeted proteins highlights many interactions

required by the pathogen to mediate important signaling pathways (Baseer, Ahmad, Ranaghan, &

Azam, 2017). Targeting proteins having strong connectivity in the interacting network was

preferred than those having an average connectivity. Strongly connected proteins contribute

significantly to the functional integrity of a PPI (protein-protein interactions) network and have

direct relation with pathogen lethality (Hassan et al., 2016; Baseer, Ahmad, Ranaghan, & Azam,

2017). Inhibition of any node in such networks renders the entire network disturbed. The

physicochemical characterized 10 proteins were subjected to STRING analysis to map their

interacting network at cellular level (Fig.5.3). The first cluster was constructed for two proteins

of Kds and three proteins of Lpx protein family. As both the family proteins are involved in Gram-

negative bacteria LPS biosynthesis one single cluster was established (Raetz & Whitfield, 2002).

Both family proteins are strongly interrelated with each other through direct and indirect means.

158

In addition, the proteins are strongly connected with waaA protein. WaaA protein is a membrane

bounded protein and catalyzes the final step of KDO biosynthesis pathway by transferring KDO

moiety from activated sugar nucleotide onto lipid A molecule (Knirel & Valvano, 2011). The

KdsA protein further interact with PyrG protein, which is basically a CTP synthase. PyrG protein

is responsible for regulating intracellular CTP level and carried out ATP driven amination of UTP

to CTP using ammonia or L-glutamine as a source of nitrogen (Robertson & Villafranca, 1993).

The network executed for OmpR protein indicates four interacting partners all of which are

members of two component system and are sensor kinase in nature. One of the key interactor was

PhoR, a protein that act as sensor kinase in phosphate regulon (Martin, 2004). KdpE protein

interact directly and indirectly with three members of Kdp system (KdpA, KdpB and KdpC) of A.

baumannii. Kdp system in Gram-negative bacteria is ATP driven operon responsible for

transporting K+ ion (Walderhaug et al., 1992). The other three direct interactors are signal

transducing histidine kinases of two-component signaling system controlling complex process of

microorganisms. Five direct interactions were found for PhoB protein, two of the proteins belong

to the same Pho family with function of phosphate regulon sensor kinase and phosphate specific

transport system accessory protein. The remaining three are histidine kinases, playing important

role in two component system of A. baumannii. The gspE is a general secretion pathway protein

E and interact strongly with five nodes: GspD, GspF and GspG are proteins from type II secretion

system for protein transportation (Lu, Korotkov, & Hol, 2014), PilC is a type IV fimbrial assembly

protein and tapD is a type IV prepilin protein that cleaves leader sequences and methylate the N-

terminal residues (Rudel et al., 1995). UvrY protein is a response regulator with strong association

with histidine kinase and Hpt domain protein (Palaniyandi et al., 2012).

159

Fig.5.3. Interacting networks for prioritized 10 drug candidates.

5.4.6. Drug Target Selection

A comprehensive literature survey was conducted for selection of potent drug candidates among

the shortlisted targets against A. baumannii. Proteins having favorable physicochemical properties,

strong interacting network with enrich essential interacting nodes for pathogen was important from

further investigation point of view. We focused on the KDO biosynthesis pathway proteins. As

such proteins are essential and specific to pathogen, virulent and have strong cellular interactome,

their inhibition can lead to A. baumannii cell growth arrest (Yi, 2009). Moreover, it was found that

very scarce inhibitors literature is available for this pathway proteins. Keeping in view, we selected

KdsA protein of this family for molecular docking and dynamics simulation study for revealing

best docked compound. The KdsA protein is ranked as top ranked drug candidate based on network

and orthology based methods (Nelson, Kelleher, Robinson, Reiling, & Asojo, 2013). Similarly,

KdsA is a druggable target because it has high affinity of binding to drugs and have bit score of

398.32 and E. value of 1.13837e-141 in Drug bank. No trail studies are reported so far against target

160

protein till to date and represents an excellent target for future in silico and experimental drug

discovery process.

5.4.7. Comparative Structure Modelling

As no 3D structure of A. baumannii KdsA was available, a comparative structure prediction

approach was applied. This approach comprised of Modeller and web based tools and subsequent

evaluation for selection of the most optimal model. Predicting a reliable 3D structure for KdsA

protein not only provides in-depth insights of molecular recognition but could also aid in potent

drug identification against the protein (Azam & Shamim, 2014). Selection of model was truly

based on the distribution of minimal number of residues in disallowed region and high percentage

of residues in most favorable region of the Ramachandran plot produced by PROCHECK analysis

(Abro & Azam, 2016). In addition, the factor to map the maximum residues was also taken into

consideration. Based on the above mentioned facts, the first model predicted by Modeller was

recognized as best optimal model (Table 5.1). The RMSD value between minimized A. baumannii

KdsA structure and that of template was observed as 0.601 Å, indicating high accuracy of the

model (Fig.5.4a). Statistically, the model is composed of 19.4 % of beta strand, 35.5 % of alpha

helix, 5 % of 3-10 helix and 40.3 % others including beta-barrel, beta-hairpins, beta-turns, psi

loops, helices, gamma-turns, beta-alpha-beta-unit, beta-bulges, helix-helix interaction and di-

sulfides (Fig.5.4b). In terms of stereochemical quality, the structure showed 93.4% of residues in

most favorable region with overall G quality factor value of -0.03. Residues in additionally allowed

region, generously allowed region and disallowed regions were 5.8%, 0.4% and 0.4%, respectively

(S-Fig.5.1a). Non-bounded interactions for the protein was calculated through ERRAT quality

factor, which demonstrated high accuracy of the model with score of 82.56 (S-Fig.5.1b). Likewise,

high quality of the model was also supported by high VERIFY-3D value of 84.17 (S- Fig.5.1c).

Moreover, the low Z-score of -7.54 indicates overall high quality of the model (S-Fig.5.1d).

Further assessment of the structure was carried out based on DOPE score. The low DOPE score

of the minimized KdsA protein further indicates the overall stability and quality of the protein (S-

Fig.5.2).

161

Fig.5.4. (a) Superimposing the best modeled KdsA structure (blue) over 4luo template (Red). (b) Secondary structure of the best

modelled KdsA structure.

162

Table 5.1. Structure evaluation of the predicted structures for KdsA enzyme using different tools.

R.M.F.R, Residues in Most favorable region, R.A.A.R, Residues in additionally allowed region, R.G.A.R, Residues in generously allowed regions,

R.D.R, Residues in Disallowed regions.

Structure

Source R.M.F.R R.A.A.R R.G.A.R R.D.R G Factor

ERRAT

Quality

Factor

VERIFY-3D

(3D-1D score >

0.2 (%))

ProSA-Web

Z-score

I-TASSER 187 49 8 4 -0.43 94.56 84.21 -7.92

ModWeb 201 34 2 6 -0.25 74.9 86.02 -7.28

Phyre2 193 15 0 1 0.17 94.46 78.99 -3.77

Swiss-model 185 43 9 5 -0.56 82.95 88.49 -7.73

Model-1 226 14 1 1 -0.03 82.56 84.17 -7.54

Model-2 220 18 3 1 -0.09 76.42 83.09 -7.37

Model-3 220 20 2 0 -0.07 79.23 79.5 -7.57

Model-4 219 19 4 0 -0.08 77.94 89.57 -7.86

Model-5 219 19 4 0 -0.08 78.94 90.77 -7.89

163

5.4.8. KdsA Intrinsic Disorder Regions Analysis

An intrinsically disordered regions of protein are those that lacked an ordered or fixed 3D structure.

This analysis was vital as function of protein strongly depends on a fixed 3D structure. It is very

evident in the Fig.5.5 that majority of the enzyme regions have order 3D structures including the

functional domain. Only three regions “QEV”, “RSDS” and “KKLDT” were revealed as discorded

having values higher than the threshold. These regions are part of loop, which is reported to be

structurally flexible. Overall it can predict that these regions could have less effect on KdsA

enzyme structure and ultimately on function.

Fig.5.5. KdsA enzyme intrinsic disorder plot. The threshold value was set to 0.5. The residues

making disorder region of enzyme are plotted over the red line.

5.4.9. Protein Active Site Prediction

Identification of binding pocket in a protein tertiary structure is vital for efficient binding of small

molecules and to block its functionality (Iqbal, Shamim, Azam, & Wadood, 2016). Active side

identification was based on comparative strategy comprising online tools, multiple sequence

alignment and literature survey. Two potential ligand cavity was predicted by Meta pocket (S-

Table 5.5) while the GHECOM and fpocket reported 9 and 25 potential binding cavities, which

vary in length and depth (S-Fig.5.3 and S-Fig.5.4). The pockets were then analyzed further for

conserved residues that are localized in the binding cavity of protein. For this, multiple sequence

alignment was done using KdsA protein sequences from P. aeruginosa, Vibrio cholerae, Neisseria

164

meningitidis and E. coli (S-Fig.5.5). The 3D alignment of KdsA from four different organisms is

also illustrated in S-Fig.6. We also found no marked differences in the overall structure predicted

from various servers and Modeller (S-Fig.5.7A). The most active site residues of KdsA is also

conserved and are illustrated in S-Fig.5.7B. Based on such analysis, Glutamine 111 (Gln 111) was

selected as a consensus active site residue of A. baumannii KdsA binding pocket for comparative

molecular studies.

5.4.10. Comparative Molecular Docking

Compounds succeeding Lipinski’s rule of five were screened through ligand scout. Total of 1460

compounds were classified as drug-like compounds and were used to dock into the active site of

KdsA protein. Top 10 hits of the docking analysis are presented in Table 5.2. Computational drug-

like and pharmacokinetics properties evaluation was done using Swiss-ADME (Daina, Michielin,

& Zoete, 2017) and PreADMET (Lee et al., 2004). Based on the analysis, Compound 4636 was

considered as potent inhibitor with GOLD fitness score and binding energy of 77.68 and -6.2

Kcal/mol, respectively. The correlation coefficient between GOLD fitness score and Autodock

Vina binding energy is shown in Fig.5.6. Critical investigation of the complex using UCSF

Chimera, Discovery studio and VMD interpreted several important residues interacting with the

compound. Comparative visual inspection revealed that the several residues of the protein active

cavity play substantial contribution in binding and stabilizing the potent inhibitor in the catalytic

pocket (Fig.5.7).These residues include Lys53, Ala54, Ser55, Lys58, Asp93, Gln111, Pro113,

Ala114, Phe115, Lys136, Ala138, Gln139, Arg166, Leu175, Asp195, Thr197, His198, Ala199,

Leu200, Gln201, Phe233, and Glu235. Prominent hydrogen bond and hydrophobic interactions

were observed between the mentioned protein active site residues and the ligand. These

interactions are fundamental to molecular recognition and contribute towards concrete and stable

complex formation. The binding pose and interactions of the compound in GOLD and AutoDock

Vina was found different. The binding mode of the compound KdsA enzyme active cavity in

GOLD was positioned to allow deep cavity binding of 2-amino-4-methylpentan-1-ol ring

(Fig.5.8). The central anisole and 3-methy—3H-pyrazole drive all the three hydrogen bond

interactions found in the complex and covered major portion of the active site interface. The

oxygen of anisole ring was found bonded to Gln139, while both nitrogen of 3-methy—3H-

pyrazole was observed in hydrogen bond interactions with His198 and Gln201. In AutoDock Vina,

165

two hydrogen bonds were observed, one between terminal oxygen of the compound and Asn60

from enzyme active site and one between the central oxygen (that linked 2-amino-4-methylpentan-

1-ol ring and anisole ring) and Gln111. The benzene ring of the compound in both tools was not

reported in any interactions. The compound completely follows Lipinski rule of five, Ghose rule

(Daina, Michielin, & Zoete, 2017), and Egan rule (Daina, Michielin, & Zoete, 2017). The

molecular weight of compound is 457.99 g/mol with number of heavy and aromatic heavy atoms

are 32 and 17, respectively. The compound possesses 11 rotatable bonds, 5 hydrogen bond

acceptors, and 2 hydrogen bond donors. The pattern of carbon attachment to other atoms is in Csp3

pattern. The value of Csp3 fraction is 0.40. The molar refractivity of the compound is 128.78 while

TPSA value is 68.54 Å2. The consensus lipophilicity of the compound is 4.55, which is within the

rules of Lipinski. The Gastrointestinal absorption of this compound is high, which indicates its

good distribution to ensure availability of high concentration of drug at the target site. The logS

value for the compound is -5.58, which indicate its moderate solubility. From synthetic point of

view, the compound score is 3.94 that mean it’s easy synthesis. From toxicity point of view, the

compound was predicted non-mutagenic in TA1535 strain Ames test and non-carcinogenic in Rat.

Fig.5.6. Correlation coefficient between GOLD fitness score and Autodock Vina binding energy

for 10 best inhibitors shortlisted in the current study.

y = -0.1809x + 7.7421R² = 0.9585

-7

-6

-5

-4

-3

-2

-1

0

6 8 6 9 7 0 7 1 7 2 7 3 7 4 7 5 7 6 7 7 7 8 7 9

Gold fitness score

Auto

do

ck V

ina

bin

din

g e

ner

gy

(kca

l.m

ol-1

)

166

Fig.5.7. Binding mode and interactions of inhibitor-4636 in the binding pocket of KdsA enzyme.

167

Fig.5.8. (A) Binding mode of the best-characterized inhibitor in KdsA enzyme cavity. (B) Closer

view.

168

Table 5.2. Top 10 inhibitors screened in the current study along with GOLD fitness score, Autodock binding energy, and druglikeness.

Compound

ID

Structure

GOLD

fitness

score

Auto Dock

Vina

Binding

energy

(Kcal.mol-1)

Druglikeness

4636

77.68 -6.2

Lipinski: Yes

Ghose: Yes

Veber: 1 voilation

Egan: Yes

Muegge: 1 voilation

1828

76.67 -6.2

Lipinski: Yes

Ghose: 2 voilations

Veber: Yes

Egan: Yes

Muegge: Yes

169

287

76 -6.0

Lipinski: Yes

Ghose: Yes

Veber: Yes

Egan: 1 voilation

Muegge: Yes

4149

72.22 -5.4

Lipinski: Yes

Ghose: 2 voilations

Veber: Yes

Egan: Yes

Muegge: Yes

1799

70.1 -5.2

Lipinski: Yes

Ghose: Yes

Veber: Yes

Egan: Yes

Muegge: Yes

170

4642

69.44 -4.9

Lipinski: Yes

Ghose: Yes

Veber: Yes

Egan: Yes

Muegge: Yes

4105

70.51 -4.9

Lipinski: Yes

Ghose: Yes

Veber: Yes

Egan: Yes

Muegge: Yes

3766

70.0 -4.8

Lipinski: Yes

Ghose: Yes

Veber: Yes

Egan: Yes

Muegge: Yes

171

4436

68.8 -4.7

Lipinski: Yes

Ghose: 1 voilation

Veber: Yes

Egan: Yes

Muegge: Yes

4248

68.6 -4.5

Lipinski: 1 voilation

Ghose: 3 voilations

Veber: Yes

Egan: Yes

Muegge: 1 voilation

172


Very little is known about the dynamic properties of KdsA enzyme in aqueous environment of the

organism under physiological conditions (Raza, Sanober, Rungrotmongkol, & Azam, 2017). The

dynamics of enzyme alone and in complex with inhibitor was explained for 100 ns (Fig.5.9). This

comparative dynamics was vital to understand what mechanistic insights the enzyme revealed

alone and while processing inhibitor. The highest RMSD value for enzyme and complex was 4.5

Å and 4.4 Å, respectively. Interestingly, the average RMSD for both systems was revealed 3.5 Å.

This means that the structural adaptions, which the enzyme acquired during simulation, are same

in both docked and undocked enzyme. An initial increase in RMSD up to 2 ns for both system

depicts loop region displacement from initial conformation and was important for compactness of

the enzyme (Fig.5.10A). In addition, a conformational change was observed in both systems after

2 ns, where residues from Val4 to Gly7 and Ile9 to Met11 of the loop region present at the N-

terminal of the enzyme convert into helix (Fig.5.10B). After 2 ns, the enzyme and complex

remained stable and has no significant fluctuations, indicating the steady confirmation of the

enzyme and complex throughout the simulation time. The minor fluctuations in RMSD of both are

due to less stable loop regions of the enzyme. No major inhibitor induced structural variations

were observed. Further understanding of enzyme structure mobility was achieved by obtaining

residue-by-residue RMSF values for the whole course of simulation. In general, the RMSF values

for the docked enzyme was unraveled highly stable compared to undock. The average RMSF for

enzyme and complex was 1.4 Å and 1.2 Å, respectively. The highest RMSF peaks for residues in

the graph are because of their presence in loop regions. As clearly can be seen that majority of

residues from active pocket of the enzyme are highly stable and was more pronounced in the

presence of inhibitor. Here it signifies that in the presence of ligand, the enzyme residues remained

in stable confirmation leading to stable complex formation. The Rg values for both systems were

investigated to better understand enzyme conformation and to shed light on equilibrium

conformational of the systems. Rg is also used for protein structure compactness. High Rg value

implies loss packing of protein atoms while lower values favor tight packing of protein atoms. An

average Rg value for both systems are 21 Å, however, docked enzyme was found more tightly

packed. Lastly, the β-factor values for enzyme residues were in complete coherent with RMSF. As

in RMSF, β-factor depicts stability of enzyme residues in the presence of ligand and structure

flexibility of loop regions

173

Fig.5.9. Statistical properties of KdsA enzyme and KdsA-inhibitor 4636 complex.

174

Fig.5.10. (A) Superimposition of initial conformation of KdsA (dark khaki) over that obtained after 2 ns (purple). (B) Conversion of

N-terminal loop into helix of KdsA at 2 ns.

175

5.4.12. Estimation of Binding Free energy

In medicinal chemistry and drug designing, the prediction of accurate ligand binding affinity to an

intended biological target is vital for novel drug candidate’s identification (Dominy, 2008;

Kitamura et al., 2014). In CAAD, both structure and ligand based drug designing provide initial

foundations for exploiting known structural information to prioritize novel molecules for

experimentation and optimization. Molecular docking is a widely used technique for virtual

screening of structurally diverse molecules derived from commercial and public chemical libraries

(Morris & Lim-Wilby, 2008). The purpose is to identify initial hits or lead inhibitors with low µM

concentrations. Chemical modifications of these hits can further improve their binding affinity and

drug-like properties. Because of the limited success of docking calculations, the need for more

accurate and lower computational throughput methods are required that prompt these calculations

and prediction to a suitable level. MM/GBSA has been successfully applied to wide range of

targets for calculating binding energies using molecular mechanics and implicit continuum

solvation model (Massova & Kollman, 2000).

The MM/GBSA calculations were performed on entire complex MD trajectories. The summary of

calculations is given in the Table 5.3. The term entropy was not calculated due convergence

problem, where it failed to be calculated. Complex formation between KdsA enzyme and the best

screened 4636 inhibitors lead to more favorable columbic interactions (-93.66 kcal.mol-1). In

contrast, non-favorable contributions for the complex was driven by polar part of solvation free

energy (111. 94 kcal.mol-1). The total electrostatic energy contribution for the complex is still high

(-18.28 kcal.mol-1). Favorable van der Waals (-36.22) and non-polar part of solvation energy (-

5.12) for the complex was also observed. The total binding energy for the system (HtotGB = ΔGsol

– np + ΔGGB + ΔEgas) is -23.06 kcal.mol-1. Total of 500 snapshots taken after every 0.2 ns of MD

trajectories was analyzed.

176

Table 5.3. Binding free energies for Kds-inhibitor 4636 complex.

5.4.13. Free Energy Decomposition

The complex interactions at atomic level were further investigated to examine the contribution of

each residue of enzyme to the overall complex binding energy. The binding free energy of each

enzyme residue and ligand is shown in Fig.5.11. This leads to decomposing total free energy into

insightful interactions and desolvation components. Residues having a contribution < -1 kcal/mol

were regarded as hot spot amino acids due to their major role in complex stabilization (Haq, Abro,

Raza, Liedl, & Azam, 2017). Residues like His65, Arg213, and His237 have binding energy

contributions < 1 kcal/mol and contribute significantly to the stability of complex. Arg213 and

His237 are part of KdsA active pocket while His65 is present around KdsA active pocket. The

residues like Gln139, His198, and Gln201 involved in hydrogen bonding with the inhibitor in

GOLD have binding energy of 0, -0.8, and -0.5, respectively. The residues (Asn60 and Gln111) of

Autodock Vina involved in hydrogen bonding interactions have binding energy of < 0 kcal.mol-1.

This indicates, in addition, to hotspot amino acids these residues also played critical role in

complex stabilization. The thorough contribution of these hotspot residues was deciphered by

decomposing their total free energy into electrostatic, van der Waals and solvation energies (Table

Contribution Energy values

(kcal.mol-1)

∆Evdw -36.22

∆Eele -93.66

∆Egas -129.88

∆Gsolv,GB 106.81

∆GGB 111.94

∆Gsol-np -5.12

∆Gele,GB -18.28

Htot,GB -23.06

177

5.4). It is quite evident like in complex, that how electrostatic and van der Waals energies dominate

the overall energy of the complex and affirm the findings the complex energy calculation

Table 5.4. Decomposition of free energy into components for active residues.

Residue van der

Waals Electrostatic

Polar

Solvation

Non-Polar

Solvation TOTAL

Asn60 -0.17 -0.43 0.61 -0.02 -0.02

His65 -1.35 -4.12 3.42 -0.25 -2.31

Gln111 -0.04 -0.08 0.12 -0.00 -0.00

Arg213 -1.03 -9.11 8.91 -0.11 -1.35

His237 -2.16 1.46 -0.53 -0.38 -1.62

178

Fig. 5.11. MM/GBSA based free energy decomposition into KdsA residues and inhibitor.

179

5.5. Conclusions

Identification of new drug targets against MDRAB is essential for guiding future drug discovery.

The 10 proteins (KdsA, KdsB, LpxA, LpxC, LpxD, GpsE, PhoB, UvrY, KdpE, and OmpR), which

we identified in this current study could provide an excellent plateform for scientist in antiinfective

research to design novel and potent anti-MDRAB inhibitors. Using KdsA enzyme in structural

based virtual screening of a natural inhibitor library, a drug-like and pharmacokinetics favored

inhibitor-4636 is predicted. The dynamics of KdsA and KdsA-4636 inhibitor complex to the best

of our knowledge was deciphered for the first time and was aimed to better understand how to

enzyme behave in real time in the absence and presence of ligand,respectively. Based on the

statistical properties of RMSD, RMSF, Rg, and β-factor derived from MD simulations trajectories

for 100 ns, it was concluded that in both states the enzyme has demonstrated stable physical

behaviour and remained stable over the period of simulation. The minor variations in enzyme

structure were investigated due to structurally sensitive loop regions. Inhibitor induced

conformational variations were found only at the start of simulation and could be speculated for

achieving a stable conformational state. Absolute binding free energy for complex using

MM/GBSA revealed vigorous interactions between enzyme and inhibitor. The complex absolute

binding energy was found -23.06 kcal.mol-1, an agreement on stability of the complex. The

electrostatic energy was found quite dominant followed by van der Waals contributions. As a

conclusion, KdsA enzyme has a stable and fixed 3D conformation and could be used in virtual

screening of diverse drug libraries to discover and design potent inhibitor against MDRAB

infections.


S-Fig.5.1. Structure evaluation plots for KdsA enzyme (a) Ramachandran plot, (b) Errat, (c)

Verify-3D, (d) Pros-SA.

S-Fig.5.2. DOPE score KdsA template and KdA enzyme.

S-Fig.5.3. Binding pockets revealed by GHECOM for KdsA enzyme.

S-Fig.5.4. Potential binding cavities of KdsA revealed by fpocket. Residues of the most active

pocket are shown.

S-Fig.5.5. Multiple sequence alignment for KdsA enzyme and its orthologues.

180

S-Fig.5.6. (A) 3D alignment of KdsA from 4 different organisms: A. baumannii KdsA (cyan), P.

aeruginosa (blue), V. cholerae (Gold), N. meningitidis (dark cyan). (B) 3D conservation of active

site residues (Gln, Leu, Pro, Ala, Phe, Leu). For clarity, A. baumannii active site residues are only

shown.

S-Fig.5.7. (A) Superimposition of all the predicted models for KdsA: Modeller-1 (yellow),

Modeller-2 (dark khaki), Modeller-3 (cyan), Modeller-4 (forest green),Modeller-5(cornflower

blue), Phyre-2 (magenta), Swiss-Model (chartreuse), Mod-web (blue), and I-Tasser (olive drab).

(B) Superimposition of active site residues critical for inhibition. For clarity, ribbon and residues

number of Modeller-1 are only shown.

S-Table 5.1. Number of proteins screened out at each step of subtractive proteomics.

S-Table 5.2. List of 15 virulent proteins extracted through VFDB analysis.

S-Table 5.3. Molecular weight prioritization of virulent proteins.

S-Table 5.4. Physicochemical characterization of virulent proteins for potential drug targets.

S-Table 5.5. Potential ligand binding sites predicted by Meta pocket for KdsA enzyme.

5.7. References



221, 507–517.

Abro, A., & Azam, S. S. (2016). Binding free energy based analysis of arsenic (+ 3 oxidation state)

methyltransferase with S-adenosylmethionine. Journal of Molecular Liquids, 220, 375–382.



Journal of Molecular Graphics and Modelling,77, 72-85.

Aiba, H., & Mizuno, T. (1990). Phosphorylation of a bacterial activator protein, OmpR, by a

protein kinase, EnvZ, stimulates the transcription of the ompF and ompC genes in Escherichia

coli. FEBS Letters, 261(1), 19–22.

Andleeb, S., Rauf, M. K., Azam, S. S., Badshah, A., Sadaf, H., Raheel, A., … others. (2016). A

181

one-pot multicomponent facile synthesis of dihydropyrimidin-2 (1 H)-thione derivatives

using triphenylgermane as a catalyst and its binding pattern validation. RSC Advances, 6(83),

79651–79661.

Azam, S. S., & Shamim, A. (2014). An insight into the exploration of druggable genome of

Streptococcus gordonii for the identification of novel therapeutic candidates. Genomics,

104(3), 203–214.

Bachmair, A., Finley, D., & Varshavsky, A. (1986). In vivo half-life of a protein is a function of

its amino-terminal residue. Science, 234(4773), 179–186.



87-99.

Butt, A. M., Tahir, S., Nasrullah, I., Idrees, M., Lu, J., & Tong, Y. (2012). Mycoplasma genitalium:

a comparative genomics study of metabolic pathways for the identification of drug and

vaccine targets. Infection, Genetics and Evolution, 12(1), 53–62.

Case, D. A., Babin, V., Berryman, J., Betz, R. M., Cai, Q., Cerutti, D. S., … others. (2014). Amber

14.

Chen, L.K., Kuo, S.C., Chang, K.C., Cheng, C.C., Yu, P.Y., Chang, C.H., … Tseng, C.C. (2017).

Clinical Antibiotic-resistant Acinetobacter baumannii strains with higher susceptibility to

environmental phages than antibiotic-sensitive strains. Scientific Reports, 7(1), 6319.



Cipolla, L., Polissi, A., Airoldi, C., Galliani, P., Sperandeo, P., & Nicotra, F. (2009). The Kdo

biosynthetic pathway toward OM biogenesis as target in antibacterial drug design and

development. Current Drug Discovery Technologies, 6(1), 19–33.



Daina, A., Michielin, O., & Zoete, V. (2017). SwissADME: a free web tool to evaluate

182

pharmacokinetics, drug-likeness and medicinal chemistry friendliness of small molecules.

Scientific Reports, 7, 42717.

Dominy, B. N. (2008). Molecular recognition and binding free energy calculations in drug

development. Current Pharmaceutical Biotechnology, 9(2), 87–95.



Falagas, M. E., Kasiakou, S. K., & Saravolatz, L. D. (2005). Colistin: the revival of polymyxins

for the management of multidrug-resistant Gram-negative bacterial infections. Clinical

Infectious Diseases, 40(9), 1333–1341.

Fiser, A., & Sali, A. (2003). Modeller: generation and refinement of homology-based protein

structure models. Methods in Enzymology, 374, 461–491.

CDC. (2013). Antibiotic resistance threats in the United States, 2013. Centres for Disease Control

and Prevention, US Department of Health and Human Services.



Gill, S. C., & Von Hippel, P. H. (1989). Calculation of protein extinction coefficients from amino

acid sequence data. Analytical Biochemistry, 182(2), 319–326.

Gonzalez-Villoria, A. M., & Valverde-Garduno, V. (2016). Antibiotic-resistant Acinetobacter

baumannii increasing success remains a challenge as a nosocomial pathogen. Journal of

Pathogens, 2016.

Guex, N., & Peitsch, M. C. (1997). SWISS-MODEL and the Swiss-Pdb Viewer: an environment

for comparative protein modeling. Electrophoresis, 18(15), 2714–2723.



Gupta, S. K., Gross, R., & Dandekar, T. (2016). An antibiotic target ranking and prioritization

pipeline combining sequence, structure and network-based approaches exemplified for

Serratia marcescens. Gene, 591(1), 268–278.

183

Guruprasad, K., Reddy, B. V. B., & Pandit, M. W. (1990). Correlation between stability of a

protein and its dipeptide composition: a novel approach for predicting in vivo stability of a

protein from its primary sequence. Protein Engineering, Design and Selection, 4(2), 155–

161.

Halgren, T. A. (1996). Merck molecular force field. I. Basis, form, scope, parameterization, and

performance of MMFF94. Journal of Computational Chemistry, 17(5–6), 490–519.



143–152.




Higgins, P. G., Dammhayn, C., Hackel, M., & Seifert, H. (2009). Global spread of carbapenem-


Hoch, J. A., & Silhavy, T. J. (1995). Two-component signal transduction (Vol. 2). ASM press

Washington, DC:

Hooft, R. W. W., Sander, C., & Vriend, G. (1997). Objectively judging the quality of a protein

structure from a Ramachandran plot. Bioinformatics, 13(4), 425–430.



AABC, 8, 37.



Huang, B. (2009). MetaPocket: a meta approach to improve protein ligand binding site prediction.

OMICS A Journal of Integrative Biology, 13(4), 325–330.



184

Ikai, A. (1980). Thermostability and aliphatic index of globular proteins. The Journal of

Biochemistry, 88(6), 1895–1898.

Iqbal, S., Shamim, A., Azam, S. S., & Wadood, A. (2016). Identification of potent inhibitors for

chromodomain-helicase-DNA-binding protein 1-like through molecular docking studies.

Medicinal Chemistry Research, 25(12), 2924–2939.

Ishida, T., & Kinoshita, K. (2007). PrDOS: prediction of disordered protein regions from amino

acid sequence. Nucleic Acids Research, 35(suppl_2), W460-W464.

Jones, G., Willett, P., Glen, R. C., Leach, A. R., & Taylor, R. (1997). Development and validation

of a genetic algorithm for flexible docking. Journal of Molecular Biology, 267(3), 727–748.

Katsila, T., Spyroulias, G. A., Patrinos, G. P., & Matsoukas, M.-T. (2016). Computational

approaches in target identification and drug discovery. Computational and Structural

Biotechnology Journal, 14, 177–184.

Kawabata, T. (2010). Detection of multiscale pockets on protein surfaces using mathematical

morphology. Proteins: Structure, Function, and Bioinformatics, 78(5), 1195–1211.



Kitamura, K., Tamura, Y., Ueki, T., Ogata, K., Noda, S., Himeno, R., & Chuman, H. (2014).

Binding free-energy calculation is a powerful tool for drug optimization: calculation and

measurement of binding free energy for 7-azaindole derivatives to glycogen synthase kinase-

3β. Journal of Chemical Information and Modeling, 54(6), 1653–1660.

Knirel, Y. A., & Valvano, M. A. (2011). Bacterial lipopolysaccharides: structure, chemical

synthesis, biogenesis and interaction with host cells. Springer Science & Business Media.

Krautler, V., Van Gunsteren, W. F., & Hunenberger, P. H. (2001). A fast SHAKE algorithm to



Kyte, J., & Doolittle, R. F. (1982). A simple method for displaying the hydropathic character of a

protein. Journal of Molecular Biology, 157(1), 105–132.

185

Lamarche, M. G., Wanner, B. L., Crépin, S., & Harel, J. (2008). The phosphate regulon and

bacterial virulence: a regulatory network connecting phosphate homeostasis and

pathogenesis. FEMS Microbiology Reviews, 32(3), 461–473.

Laskowski, R. A. (2001). PDBsum: summaries and analyses of PDB structures. Nucleic Acids

Research, 29(1), 221–222.

Le Guilloux, V., Schmidtke, P., & Tuffery, P. (2009). Fpocket: an open source platform for ligand

pocket detection. BMC Bioinformatics, 10(1), 168.

Lee, S. K., Chang, G. S., Lee, I. H., Chung, J. E., Sung, K. Y., & No, K. T. (2004). The PreADME:

pc-based program for batch prediction of adme properties. EuroQSAR, 2004, 5–9.

Li, W., & Godzik, A. (2006). Cd-hit: a fast program for clustering and comparing large sets of

protein or nucleotide sequences. Bioinformatics, 22(13), 1658–1659.

Lipinski, C. A. (2004). Lead-and drug-like compounds: the rule-of-five revolution. Drug

Discovery Today: Technologies, 1(4), 337–341.

Lu, C., Korotkov, K. V, & Hol, W. G. J. (2014). Crystal structure of the full-length ATPase GspE

from the Vibrio vulnificus type II secretion system in complex with the cytoplasmic domain

of GspL. Journal of Structural Biology, 187(3), 223–235.

Maier, J. A., Martinez, C., Kasavajhala, K., Wickstrom, L., Hauser, K. E., & Simmerling, C.

(2015). ff14SB: improving the accuracy of protein side chain and backbone parameters from

ff99SB. Journal of Chemical Theory and Computation, 11(8), 3696–3713.

Martin, J. F. (2004). Phosphate control of the biosynthesis of antibiotics and other secondary

metabolites is mediated by the PhoR-PhoP system: an unfinished story. Journal of

Bacteriology, 186(16), 5197–5201.

Massova, I., & Kollman, P. A. (2000). Combined molecular mechanical and continuum solvent

approach (MM-PBSA/GBSA) to predict ligand binding. Perspectives in Drug Discovery and

Design, 18(1), 113–135.



186

130–155.




Moriya, Y., Itoh, M., Okuda, S., Yoshizawa, A. C., & Kanehisa, M. (2007). KAAS: an automatic

genome annotation and pathway reconstruction server. Nucleic Acids Research, 35(suppl_2),

W182-W185.

Morris, G. M., & Lim-Wilby, M. (2008). Molecular docking. Molecular Modeling of Proteins,

365–382.

Munoz-Price, L. S., & Weinstein, R. A. (2008). Acinetobacter infection. New England Journal of

Medicine, 358(12), 1271–1281.

Nelson, S. K., Kelleher, A., Robinson, G., Reiling, S., & Asojo, O. A. (2013). Structure of 2-keto-

3-deoxy-d-manno-octulosonate-8-phosphate synthase from Pseudomonas aeruginosa. Acta

Crystallographica Section F: Structural Biology and Crystallization Communications,

69(10), 1084–1088.





Palaniyandi, S., Mitra, A., Herren, C. D., Lockatell, C. V., Johnson, D. E., Zhu, X., &

Mukhopadhyay, S. (2012). BarA-UvrY two-component system regulates virulence of

uropathogenic E. coli CFT073. PloS One, 7(2), e31348.



Perumal, D., Lim, C. S., Chow, V. T. K., Sakharkar, K. R., & Sakharkar, M. K. (2008). A combined

computational-experimental analyses of selected metabolic enzymes in Pseudomonas

species. International Journal of Biological Sciences, 4(5), 309.

187










Raetz, C. R. H., & Whitfield, C. (2002). Lipopolysaccharide endotoxins. Annual Review of

Biochemistry, 71(1), 635–700.

Raza, S., Sanober, G., Rungrotmongkol, T., & Azam, S. S. (2017). The vitality of swivel domain

motion in performance of Enzyme I of phosphotransferase system; A comprehensive

molecular dynamic study. Journal of Molecular Liquids, 242,1184-1198.

Robertson, J. G., & Villafranca, J. J. (1993). Characterization of metal ion activation and inhibition

of CTP synthetase. Biochemistry, 32(14), 3769–3777.

Roe, D. R., & Cheatham, T. E. (2013). PTRAJ and CPPTRAJ: software for processing and analysis

of molecular dynamics trajectory data. Journal of Chemical Theory and Computation, 9(7),

3084–3095.

Roy, A., Kucukural, A., & Zhang, Y. (2010). I-TASSER: a unified platform for automated protein

structure and function prediction. Nature Protocols, 5(4), 725.

Rudel, T., FAcIus, D., Barten, R., Scheuerpflug, I., Nonnenmacher, E., & Meyer, T. F. (1995).

Role of pili and the phase-variable PilC protein in natural competence for transformation of

Neisseria gonorrhoeae. Proceedings of the National Academy of Sciences, 92(17), 7986–

7990.

Russo, T. A., Manohar, A., Beanan, J. M., Olson, R., MacDonald, U., Graham, J., & Umland, T.

C. (2016). The Response Regulator BfmR Is a Potential Drug Target for Acinetobacter

baumannii. MSphere, 1(3), e00082-16.

188

Sakharkar, K. R., Sakharkar, M. K., & Chow, V. T. K. (2004). A novel genomics approach for the

identification of drug targets in pathogens, with special reference to Pseudomonas

aeruginosa. In Silico Biology, 4(3), 355–360.



147–153.

Sarangi, A. N., Lohani, M., & Aggarwal, R. (2015). Proteome mining for drug target identification

in Listeria monocytogenes strain EGD-e and structure-based virtual screening of a candidate

drug target penicillin binding protein 4. Journal of Microbiological Methods, 111, 9–18.

Shaban, H., Na, I., Kislichkina, A. A., Dentovskaya, S. V, Anisimov, A. P., & Uversky, V. N.

(2017). Effect of natural polymorphism on structure and function of the Yersinia pestis outer

membrane porin F (OmpF protein): a computational study. Journal of Biomolecular Structure

and Dynamics, 35(12), 2588–2603.

Shoichet, B. K. (2004). Virtual screening of chemical libraries. Nature, 432(7019), 862.


(2014). STRING v10: protein--protein interaction networks, integrated over the tree of life.


Trott, O., & Olson, A. J. (2010). AutoDock Vina: improving the speed and accuracy of docking

with a new scoring function, efficient optimization, and multithreading. Journal of


Vyas, V., Jain, A., Jain, A., & Gupta, A. (2008). Virtual screening: a fast tool for drug design.

Scientia Pharmaceutica, 76(3), 333–360.

Walderhaug, M. O., Polarek, J. W., Voelkner, P., Daniel, J. M., Hesse, J. E., Altendorf, K., &

Epstein, W. (1992). KdpD and KdpE, proteins that control expression of the kdpABC operon,

are members of the two-component sensor-effector class of regulators. Journal of

Bacteriology, 174(7), 2152–2159.

Weglenski, P., Ninfa, A. J., Ueno-Nishio, S., & Magasanik, B. (1989). Mutations in the glnG gene

of Escherichia coli that result in increased activity of nitrogen regulator I. Journal of

189

Bacteriology, 171(8), 4479–4485.



217–230.



W407-W410.

Wolber, G., & Langer, T. (2005). LigandScout: 3-D pharmacophores derived from protein-bound

ligands and their use as virtual screening filters. Journal of Chemical Information and

Modeling, 45(1), 160–169.

Yi, L. (2009). Studies of 3-Deoxy-D-manno-octulosonate 8-phosphate phosphatase: Mechanistic

insights and a gene fusion example. University of Michigan.









Zhang, R., Ou, H.Y., & Zhang, C.T. (2004). DEG: a database of essential genes. Nucleic Acids


190

Chapter # 6

Toward Novel Inhibitors against Kdsb: A Highly Specific and

Selective Broad-Spectrum Bacterial Enzyme

6.1. Abstract

KdsB (3-deoxy-manno-octulosonate cytidylyltransferase) is a highly specific and selective

bacterial enzyme that catalyzes KDO (3-Deoxy-D-mano-oct-2-ulosonic acid) activation in KDO

biosynthesis pathway. Failure in KDO biosynthesis causes accumulation of lipid A in the bacterial

outer membrane that leads to cell growth arrest. This study reports a combinatorial approach

comprising virtual screening of natural drugs library, molecular docking, computational

pharmacokinetics, MD simulations, and binding free energy calculations for the identification of

potent lead compounds against the said enzyme. Virtual screening demonstrated 1460 druglike

compounds in a total of 4800, while molecular docking illustrated Ser13, Arg14, and Asp236 as

the anchor amino acids for recognizing and binding the inhibitors. Functional details of the enzyme

in complex with the best characterized compound-226 were explored through two hundred

nanoseconds of MD simulations. The ligand after initial adjustments jumps into the active cavity,

followed by the deep cavity, and ultimately backward rotating movement towards the initial

docked site of the pocket. During the entire simulation period, Asp236 remained in contact with

the ligand and can be considered as a major catalytic residue of the enzyme. RDF confirmed that

towards the end of the simulation, strengthening of ligand-receptor occurred with ligand and

enzyme active residues in close proximity. Binding free energy calculations via MM(PB/GB)SA

and Waterswap reaction coordinates, demonstrated the high affinity of the compound for enzyme

active site residues. These findings can provide new avenues for designing potent compounds

against notorious bacterial pathogens.

191

6.2. Introduction

Each year around the globe, antibiotic resistance kills an estimated 0.7 million people and the

number could be upstretched to 10 million by 2050 if no appropriate remedies are developed or

efforts are made to curtail resistance (Willyard, 2017). In February 2017, the WHO released a list

of serious drug-resistant bacterial pathogens that pose the greatest threat to human health (WHO,

2017). On top of this list is Carbapenem-resistant A. baumannii, a Gram-negative nosocomial

bacterium, responsible for severe life threatening complications, as it is resistant to the last resort

“Carbapenem” antibiotics. A. baumannii infections effect critically ailed individuals for which

almost no treatment exists (Hsu et al., 2017; Willyard, 2017). The pathogen is classified in the

critical group based on the urgent need for the development of new antibiotics (WHO, 2017).

Identification of novel drug targets against the pathogen can aid in the development of new

antibiotics and in reducing the death burden due to its resistant infections (Miesel, Greene, &

Black, 2003; WHO, 2017; Willyard, 2017).

Strategies based on genomic and proteomic information have garnered loads of appreciation in the

recent years for screening druggable candidates against several notorious pathogens (Gupta,

Pradhan, Jain, & Rai, 2017). One of the target that is of particularly great interest against Gram-

negative bacteria is the outer membrane layer, a structure that is essential for their virulence and

pathogenicity (Yi, 2009). This outer layer is asymmetrical as it comprises of Lipid A molecule

localized on the outer leaflet and phospholipids present on the inner leaflet (S-Fig.6.1) (Yi, 2009).

Lipid A molecule, inner and outer polysaccharide core joint by covalent bonds and the outreaching

O-antigen constitute the layer of Lipopolysaccharide (LPS), also referred as endotoxins or

lipoglycan (Osborn, 1979). In normal conditions, lipid A is a disaccharide of phosphorylated

glucosamine interlinked by β (1 → 6) linkage and is decorated with attached 6-7 fatty acid

molecules (Osborn, 1979). The lipid A region is a very conserved component of LPS and is mainly

responsible for its toxicity (Todar, 2002). The core domain of polysaccharide chain is directly

attached to lipid A and normally contains 3-Deoxy-D-manno-oct-2-ulosonic acid (KDO) and

heptose sugar (Yi, 2009). The number of KDO sugars to be incorporated into LPS for the proper

functioning of bacterial cell vary from specie to specie. For instance, two KDO sugars linked by α

(2 → 6) bond, are required to be incorporated into E. coli LPS for normal growth (Yi, 2009). At

the outreaching end of LPS oligosaccharide, is the outermost domain of O antigen (also known as

192

O side chain) (Whitfield & Trent, 2014). The O side chain consists of repetitive glycan polymer,

which varies from strain to strain (Valvano, Furlong, & Patel, 2011). The cellular machinery

required for KDO biosynthesis is mandatory and its failure to produce KDO products in bacterial

cells leads to accumulation of lipid A precursor and ultimately to cell growth arrest (Smyth &

Marchant, 2013). Any interruption in KDO biosynthesis pathway can make the bacterial cell less

pathogenic and vulnerable to antibiotic actions (Cipolla et al., 2009). Since, KDO metabolic

pathway is specific and essential for survival; it could serve as a promising therapeutics target

against Gram-negative bacteria (Yi, 2009; Smyth & Marchant, 2013; Cipolla et al., 2009). KDO

biosynthetic pathway is accomplished through four sequential enzymatic phases (S-Fig.6.2). The

first phase involves the isomerization reaction of D-ribulose-5-phosphate (Ru5P), catalyzed by

KdsD enzyme to produce arabinose 5-phosphate (A5P). In the second phase, 3-deoxy-D-manno-

octulosonate-8- phosphate (KDO8P) is generated through a condensation reaction of A5P and

phosphoenolpyruvate (PEP) in the presence of KdsA enzyme. The phosphate ester of KDO8P is

catalyzed in the third phase by mean of KdsC enzyme and in the final phase, CMP portion of CTP

is added by KdsB enzyme to KDO for its activation (Smyth & Marchant, 2013). KdsB enzyme,

also known as CMP-KDO synthetase or KDO cytidylyltransferase, is an attractive antibacterial

target for several reasons. The enzyme is highly (i) specific, (ii) selective, (iii) essential for the

growth and survival, (iv) virulent, (v) has enriched cellular protein-protein interactions, (vi) high

stability index, (vii) favorable physicochemical parameters, and (vii) broad spectrum conservation.

In addition, the crystal structure of the enzyme is reported, which makes this enzyme an excellent

target for CADD.

Here, in this present study, we employed a combinatorial approach for screening natural inhibitors

against A. baumannii KdsB (AbKdsB) enzyme. Virtual screening (John, Sivashanmugam,

Umashankar, & Natarajan, 2017; Naz et al., 2018;Vyas, Jain, Jain, & Gupta, 2008) in combination

with comparative docking approach (Jones, Willett, Glen, Leach, & Taylor, 1997; Trott & Olson,

2010) was used for the identification of the best hit inhibitors against the enzyme. Computational

druglikeness and pharmacokinetics evaluation, further aided in the selection of potent inhibitor

(Lee et al., 2004; Daina, Michielin, & Zoete, 2017). The best characterized ligand in complex with

the enzyme was subjected to MD simulations (Hospital, Goni, Orozco, & Gelpli, 2015; John,

Sivashanmugam, Umashankar, & Natarajan, 2017), where statistical parameters like, RMSD,

RMSF, β-factor, and Rg were employed to decipher the dynamics of complex and validate docking

193

results (Raj, Kumar, & Varadwaj, 2017; Panman et al., 2017). The distribution of ligand atoms

with reference to enzyme and stability of the complex, towards the end of simulation period were

determined through RDF (Hemmer, Steinhauer & Gasteiger, 1999). Furthermore, the binding free

energy calculation analysis was performed to validate complex stability and shed light on the

receptor hotspot amino acids (Genheden & Ryde, 2015;Massova & Kollman, 2000; Woods,

Malaisree, Hannongbua, & Mulholland, 2011).


The three-dimensional structure of AbKdsB protein having a resolution of 1.9 Å was retrieved

from PDB with PDB ID, 4FCU (Berman et al., 2006). The protein is present in homodimer

stoichiometry, each having an identical independent active site, therefore, only a single monomer

is used in the current study (Vukic et al., 2015). To remove steric clashes and to obtain an

energetically optimized structure the protein structure was subjected to minimization (Ahmad,

Raza, Uddin, & Azam, 2017; Vukic et al., 2015). Energy minimization was accomplished through

UCSF Chimera (Pettersen et al., 2004), by assigning Gasteiger charges (Elengoe, Naser, &

Hamdan, 2014) under TFP (Clark, Cramer, & Van Opdenbosch, 1989). Total of 1500

minimization steps were used, split into 750 steps of conjugate gradient algorithm and 750 steps

of steepest descent algorithm (Ahmad, Raza, Uddin, & Azam, 2017; Baseer, Ahmad, Ranaghan,

& Azam, 2017).


Virtual screening in comparison to de novo techniques has the advantage of evaluating best hits

from commercial sources for biological activities (Green, 2008; Shoichet, 2004; Verma, Tiwari,

& Tiwari, 2017;Wadood et al., 2017). Virtual screening of Asinex library

(http://www.asinex.com/antibacterial_compound_library-html/), which comprising 4800 natural

compounds was carried out using LigandScout 4.1 (Wolber & Langer, 2005). Only compounds

fulfilling Lipinski’s rule of five (Lipinski, 2004) were selected and minimized using MMFF94

force field (Halgren, 1996). According to this rule, compounds having molecular weight < 500

kDa, hydrogen bond acceptors (HBA), hydrogen bond donors (HBD), log p value < 5, and total

polar surface area (TPSA) in between 30-140 can be considered as druglike compounds.

http://www.asinex.com/antibacterial_compound_library-html/

194

Druglikeness of molecules is useful when characterizing novel lead compounds against a target of

interest by screening compound libraries (Kadam & Roy, 2007).

Active site information for docking of druglike compounds was gathered using multiple

approaches. These involve online ligand binding cavity prediction tools (Metapocket (Huang,

2009) and DoGSite-Scorer (Volkamer, Kuhn, Rippmann, & Rarey, 2012)), and multiple sequence

alignment (MSA) (Raza, Sanober, Rungrotmongkol, & Azam, 2017; Zvelebil, Barton, Taylor, &

Sternberg, 1987), and from comprehensive literature survey (Heyes et al., 2009). An active cavity

residue (Asp236), which is highly conserved among its orthologues was selected as a putative

active site residue for molecular docking studies. The residue was cross referred with the results

of online servers and further confirmed from literature.

For docking, two different software were utilized in order to cross-validate the docking findings

and remove the false positive hits of virtual screening from one docking routine. Therefore, it was

absolutely necessary to get rid of such hits. In the current study, top hits of GOLD docking were

cross-validated through Auto Dock Vina. Cross-validation is important prior to commence on

virtual screening experiment (Hevener et al., 2009). Molecular docking of shortlisted druglike

compounds into protein catalytic cavity was performed using GOLD software version 5.1 (Jones,

Willett, Glen, Leach, & Taylor, 1997). The GOLD fitness score calculations were done on linux

(OpenSuse 11.2) operating system installed on Intel Xeon QuadTM core processor of 3.0 GHz.

The genetic runs for each compound were set to 10 in number, while all other parameters were

kept as default. Top hit compounds of GOLD were docked into protein active pocket using

AutoDock Vina for calculating binding energies and validating the docking results of GOLD (Trott

& Olson, 2010). The grid box size was set to 15 Å along X, Y and Z-axis. The coordinates of

Asp236: OD1 atom were selected as docking center in both GOLD and AutoDock Vina.

6.3.2. Druglikeness and Computational Pharmacokinetics

The selection of enzyme-ligand complex for MD simulations was based on both docking results

and druglikeness of the compounds. The docked compound having high GOLD fitness score, low

binding energy, and following druglikeness rules of Ghose filter, Veber filter, Egan rule and

Muegge rule in addition to Lipinski’s rule of five was selected. Computational pharmacokinetics

of the best characterized compound was disclosed using SWISSADME

195

(http://www.swissadme.ch/) (Daina, Michielin, & Zoete, 2017) and PreADMET

(https://preadmet.bmdrc.kr/) (Lee et al., 2004).


The best characterized compound along with the target protein was simulated to gain mechanistic

insights to understand the dynamics (Hospital, Goni, Orozco, & Gelpi, 2015;Abro & Azam, 2016).

Simulation protocol was followed using AMBER (Assisted model building with energy

refinement) 14 package (Case et al., 2014). Initial libraries for the ligand were generated using

antechamber program, while integration of the docked complex into TIP3P water box (size 12 Å)

with ff14SB was achieved using leap program (Weiner & Kollman, 1981). The hydrated complex

was neutralized by addition of 12 Na+ ions. Minimization of the system was done in a gradual

manner (Andleeb et al., 2016). Hydrogen atoms of the entire system were relaxed first with 500

cycles, followed by water box minimization with 1000 cycles at restraint of 200 kcal/mol- Å2. The

system C-alpha atoms were then minimized for 1000 cylcles at restraint of 5 kcal/mol- Å2. The

system non-heavy atoms were minimized for 300 cycles at restraint of 100 kcal/mol- Å2. The

system was then subjected to heating where it was heated to 300 K for 20 picoseconds (ps) with

restraint on C-alpha atoms of 5 kcal/mol- Å2 and time scale of 2 femtoseconds (fs). Langevin

dynamics (Paterlini & Ferguson, 1998) were used for maintaining system temperature while

SHAKE algorithm (Kräutler, Van Gunsteren, & Hunenberger, 2001) was used for heating the

system to constrain bond with hydrogen and NVT ensemble. System equilibrium was achieved for

100 ps while time scale is set to 2 fs (Andersen, 1980). System pressure was maintained using

NPT ensemble and restraint on C-alpha atoms of 5 kcal/mol- Å2. Pressure parameters were used

again with exception to restraint on C-alpha atoms for 50 ps. These parameters were applied for

every 10 ps by 1 kcal/mol- Å2. Same set of parameters were used for 1 ns to allow the system to

equilibrate. The NVT ensemble in combination with Berendsen algorithm was used for production

run. For nonbonded interactions cut-off value of 0.8 Å was set while SHAKE algorithm was

applied for hydrogen bonds. Production run of 200 ns was carried out with a time scale of 2 fs.

Cpptraj program of AMBER was used to analyze simulation trajectories (Roe & Cheatham, 2013).

http://www.swissadme.ch/

196

6.3.4. RDF

RDF, g(r), is the probability of finding an atom of interest in a sphere volume of radius r (Donohue,

1954). In the present study, RDF between the atoms of most critical protein active site residue

(Asp236) and ligand (compound-226) atoms were plotted in the last 10 ns of simulation protocol

to unveil the stability of the complex towards the end of the simulation period. RDF was extracted

using PTRAJ program of AMBER to quantify the density of specified atoms of interest with

reference to another definite atom.


Binding free energy of the top five systems was first determined using MM(PB/GB)SA method of

AMBER 14 (Kumar, Srivastava, Negi, & Sharma, 2018; Massova & Kollman, 2000). A total of

1000 frames were extracted from simulation trajectories and analyzed using MMPBSA.py module

of AMBER. In MMPB/GBSA approach, computation of binding free energy takes into account,

the difference between the complex free energy, and that of ligand and receptor.

∆Gbinding = Gcomplex- (Gprotein + Gligand)……………………………...……………………….(XVIII)

Furthermore, the binding free energy of the top complex was decomposed into per residue to

underline hotspot amino acids that contribute as major residues in binding affinity.

Binding free energies were calculated using two different methods to cross-validate the findings,

remove false positive results and circumvent the limitations of one method. MM(PB/GB)SA

employs an implicit solvent system to fill the cavity left behind as a result of ligand decoupling

reaction. The water molecules introduced into the generated cavity interact with the protein active

site, and thus contribute significantly to overall binding free energy (Woods et al., 2014). To

circumvent this gap, Christopher et. al. proposed Waterswap, which calculate binding free energy

using explicit solvent method (Woods, Malaisree, Hannongbua, & Mulholland, 2011). MM/PBSA

utilizes the conformation generated by MD simulations for free energy calculation while

Waterswap run a Monte Carlo simulation to estimate the conformation and evaluate the binding

free energy. Form Monte Carlo simulation we attain Waterswap reaction coordination (WSRC)

that represent swapping of ligand with a water cluster. Waterswap is relatively a new methodology

for predicting absolute binding free energy of protein-ligand complex making use of a single

simulation (Woods et al., 2014).

197

This approach is based on novel reaction coordinates, which swap the bonded ligand with an equal

amount of bulk water and this give new insights to investigate the bonded behavior of the system

of interest. The method is supplemented with an identity constraint (Tyka, Sessions, & Clarke,

2007), which allows it to identify a cluster of water molecules in bulk water having same volume

and shape of ligand in protein active pocket and subsequent swapping of the two using a dual

topological algorithm (Hamelberg & McCammon, 2004). Absolute binding free energy is

calculated using replica exchange thermodynamic integration method (Woods, Malaisree, Long,

McIntosh-Smith, & Mulholland, 2013). A pair of coupled simulation boxes, coupled to the same

thermostat, are used in the method, (a) a protein box (containing protein-ligand complex solvated

in periodic boundary box of explicit water molecules) and a water box (composed of only periodic

box of water molecules) (Woods et al., 2014).

The binding energy of the system is calculated using the following statistical equation,

E (λ) = Eproteinbox + Ewaterbox + Eligand + Ecluster + (1 - λ) (Eligand:proteinbox + Ecluster:waterbox) + (λ)

(Ecluster:proteinbox + Eligand:waterbox)………………………………………………………………(XIX)

Here, Eproteinbox is energy of all the molecules in the protein box except the ligand, Ewaterbox is the

energy of all the molecules except the water cluster identified in the water box, Eligand represents

the intramolecular energy of the ligand, Ecluster is the intermolecular energy between all the water

molecules present in the water cluster, Eligand:protein box is the energy of interaction between ligand

and all atoms of the protein box, Ecluter:water box is the resultant energy of interactions between all

water molecules and water clusters of the water box, λ is reaction coordinates used to scale

Ecluter:water box by 1 – λ. The effect of the reaction is to decouple the ligand from protein box linked

with simultaneous decoupling of water cluster from water box. Concurrently, the coupling of

ligand to water box and that of water cluster to protein box occurred. Energy calculation between

the ligand and all water molecules in the protein box is inferred through Eligand:water box, while that

for water cluster and molecules in the protein box is represented by Ecluter:protein box and scaling by

λ. λ is a single coordinate reaction that is transformed from λ = 0 to λ = 1. λ = 0 condition means

the bound ligand to protein in protein box while λ = 1, represents the unbound ligand and is in bulk

water and additionally, it corresponds to a transferred cluster of water to protein box for filling the

resulting cavity.

Absolute binding free energy takes into account gradient of energy calculations with respect to λ,

198

dE/D λ = (Ecluster:proteinbox + Eligand:waterbox) – (Eligand:waterbox + Ecluster:waterbox)……………………(XX)

The ensemble average is calculated at different values of λ to obtain free energy gradient across λ,

(dG/dλ)λ= (dE/dλ)λ …………………………………………………………………………..(XXI)

In order to average free energy gradients, Monte Carlo (MC) sampling at each value of λ must be

achieved. The binding free energy is finally obtained by integrating the gradient across λ,

Gbind = - ʃo1 (dG/dλ)λ λd………………………………………………………………...……(XXII)

Here, the negative integral is employed as the Waterswap reaction coordinate models unbinding,

which indicate ligand pulling out of the protein. The quality of Waterswap results depends on the

number of Monte Carlo simulations used to calculate free energy gradients across λ. It has been

found that sixteen Monte Carlo simulations to be sufficient generating sixteen free energy

gradients spaced across λ. The gradients can be integrated either numerically or analytically. In

the latter case, the gradients can be fitted to a polynomial expansion and integrating the resulting

expression. Waterswap has the further advantage of averaging the gradient of components of the

total energy with respect to λ and would result in approximation of the gradient of the

corresponding free energy component. This can be exemplified by equation (XXIII) to (XXVI)

that show how free energy change only of the protein box, Gproteinbox, could be achieved by

integrating energy components gradient, which involve only interactions between the water cluster

and ligand and protein box molecules (Eproteinbox (λ)) (Shyu & Ytreberg, 2009).

Eproteinbox (λ) = (1- λ)Eligand:proteinbox + λEcluster:proteinbox………………………………………(XXIII)

dEproteinbox / dλ = Ecluster:proteinbox - Eligand:proteinbox……………………………………………..(XXIV)

(dGproteinbox / dλ)λ ≈ (dEproteinbox / dλ) λWSRC………………………………………………….(XXV)

Gproteinbox = - ʃo1 (dGproteinbox/dλ)λ dλ…………………………………………………………(XXVI)

Gproteinbox provides an estimate of binding free energy for only protein box molecules that

contribute to total binding free energy. The contribution of molecules from waterbox, Gwaterbox, can

be calculated using similar decomposition approach.

Ewaterbox (λ) = (1- λ)Ecluster:waterbox + λEligand:waterbox……........................................................ (XXVII)

dEwaterbox / dλ = Eligand:waterbox – Ecluster:waterbox………………………………………………. (XXVIII)

199

(dGwaterbox / dλ)λ ≈ (dEwaterbox / dλ) λWSRC…………………………………………………….(XXIX)

Gwaterbox = - ʃo1 (dGwaterbox/dλ)λ dλ……………………………………………………………(XXX)

It is important to mention that these decompositions are approximate because free binding energy

is an average over configuration sampled using total Waterswap energy function as mentioned by

λWSRC in the equations. As a result of this, the sum Gproteinbox and Gwaterbox will not exactly be equal

to Gbind (Woods et al., 2014). Waterswap module in sire package was used for Waterswap

calculations of complex of compound 226 using the same force field and solvent used in dynamics

simulation (Woods, Malaisree, Long, McIntosh-Smith, & Mulholland, 2013). Clustering analysis

was accomplished on the MD trajectories in order to find representative structures to be used for

Waterswap calculations. The DBSCAN (Density-based spatial clustering of applications with

noise) algorithm was employed for clustering with allowed sieve value of 20 and the sieved frames

were fitted to the entire cluster. Waterswap calculations were performed for default 1000 iterations

of the MD trajectories on Intel Xeon QuadTM Core processor 3.0 GHz, using linux as a workstation

and took 10 days for completion. To tackle configuration search space, Waterswap runs its own

MC calculations and 100 million moves of MC sampling for each of 16 λ (0.005, 0.071, 0.203,

0.269, 0.335, 0.401, 0.467, 0.533, 0.599, 0.665, 0.731, 0.797, 0.863, 0.929, 0.995) were performed.

Minimum of 25 members were allowed in a cluster with cluster cut off size set to 1 Å. Total of

1.6 × 109 MC move sample size was achieved for calculating Free Energy Perturbation (FEP),

Thermodynamic Integration (TI), and Bennetts algorithm. The level of agreement between these

three values gives an idea about the convergence of the results. Values within 1 kcal/mol of

deviation are regarded reasonable.


Computational screening or in silico screening of chemical databases is an attractive approach for

screening novel inhibitors against a target of interest (Shoichet, 2004). Such screening aims to

reduce libraries of chemical compounds to more manageable number having the highest chance of

inhibiting the intended target of choice (Vyas, Jain, Jain, & Gupta, 2008). These approaches are

not only time saving and cost effective but also aid in prioritizing compounds for biological assays

(Wadood et al., 2017). Majority of the drugs, which fail in clinical trials prove very detrimental

and expensive in drug development process. Failure of drugs is mainly because of the adverse

effects (Lionta, Spyrou, Vassilatis, & Cournia, 2014). Considering this, virtual screening of Asinex

200

library (a database of natural antibacterial compounds) was performed first to filter only those

compounds that completely follow Lipinski’s rule of five. This was followed by computational

pharmacokinetics to ensure filtration of compounds with no associated toxicity and improved

physicochemical properties. Compounds that follow Lipinski’s rule are believed to have enhanced

potency of membrane permeability and readily absorbed by the body (Leeson, 2012). The analysis

was done through ligandScout and brought forth 1460 druglike compounds. Minimization of KdsB

enzyme revealed improved structure stability as illustrated by superimposition of non-minimized

KdsB and minimized KdsB enzyme with RMSD value of 0.4 Å. The next phase of the study was

to predict enzyme active cavity for molecular docking studies of shortlisted druglike compounds.

Identification of enzyme functional site was vital in enzyme tertiary structure for its inhibition and

to guide drug discovery process (Copeland, 2013; Iqbal, Shamim, Azam, & Wadood, 2016). A

comparative framework utilizing online active site prediction tools, evolutionary sequence

conservation, and literature survey were conducted to estimate KdsB enzyme catalytic site. The

protein active site prediction phase revealed Asp236 as a residue of choice for molecular docking.

The residue is present in the 8th major pocket revealed by GHECOM (S-Fig.6.3A), which in total

predicted 8 major cavities. The presence of this residue in the ligand binding cavity was further

confirmed through DoGSite-Scorer (S-Fig.6.3B). Residue conservation analysis through MSA

revealed that this residue is present in a stretch of 20 conserved amino acid residues (Fig.6.1),

localized in major binding pocket across several experimentally predicted Gram-negative KdsB

enzymes.


In computational biology and computer aided drug designing, the most demanding and momentous

job is to foresee accurate binding of the inhibitor into protein catalytic pocket (Copeland, 2013).

The appropriate binding conformation of the inhibitor into enzyme binding cavity is a prerequisite

for inhibiting enzyme functionality and forming a stable enzyme-inhibitor complex. The binding

conformations of scrutinized compoundss in KdsB active site was investigated through two

prominent molecular docking software packages (GOLD and AD-Vina). Molecular docking

allows effective in silico screening of ligands to sort out appropriate ligands that best fit both

geometrically and energetically into the protein active site (Cavasotto, Orry, & Andrew, 2007).

Molecular docking of 1460 druglike compounds into protein active site cavity was first performed

201

with GOLD. It generated 10 conformations for each compound, among which only best

characterized conformation in terms of GOLD fitness score was selected. A cut off value of GOLD

fitness score > 70 was applied to filter compounds for binding energy calculations. In total, five

shortlisted hits were selected for AD Vina assay (Table 6.1). The compounds were docked into

the same cavity using GOLD. The selection of complex was based on both GOLD fitness score

and AD Vina binding energy. Further assistance was taken from druglikeness rules and ADMET

properties of compounds predicted by SWISS-ADME and preADMET servers. The binding mode

and interactions of compound-226 in GOLD complex is shown in S-Fig.6.4. Compound 226 was

regarded as an ideal druglike compound by all major rules of druglikeness mentioned in the

methodology section. GOLD fitness score of the compound was determined to be 72.2 and binding

energy of -5.7 kcal/mol. The complex was visualized through VMD, UCSF Chimera, Discovery

Studio (DS) and Ligplot. Binding mode and interactions of top docked ligand in both GOLD and

AD Vina are explained in Fig.6.2. Results from all these tools demonstrated that three main

residues (Ser13, Arg14, and Asp236) of protein active site played a major role in anchoring the

ligand within the active pocket and forming hydrogen and van der Waals interactions. Since,

hydrogen bonding ensures stable and tangible complex formation (Patil et al., 2010), they were

investigated first. The role of hydrogen bonding in binding affinity of ligand is extensively

elaborated and considered as a major driving force for inhibitors to exert their inhibitory actions

(Hubbard & Kamran Haider, 2010). The three residues (Ser13, Arg14, and Asp236) were observed

in close vicinity of the ligand. The hydrogen bond distance between these residues and ligand was

unveiled to be in strong category. The residue Ser13 interact with the oxygen of the central pyran

ring of the compound through its hydrogen atom attached to main chain nitrogen. The hydrogen

bond distance between the receptor residue atom and that of the ligand was estimated to be 2.4 Å.

The hydrogen atom of the Arg14 residue was found to form close hydrogen contact with the

oxygen of 2-methoxy-3-methyl-1-1-bi (cyclohexane) ring. The distance between Arg14 hydrogen

atom and oxygen of ligand cyclohexane ring was deduced as 2.2 Å.

202

Fig.6.1. Multiple sequence alignment of KdsB sequences from different bacterial species, 4FCU

(A. baumannii), 3K8D (E.coli), 4XWI (P. aeruginosa), 3TQD (C. Burnetii), 3QAM (V. Cholerae),

3JTJ (Y. Pestis). Sequences in green box represent the major conserved catalytic pocket of KdsB

enzyme.

Similarly, another hydrogen bond was observed between the oxygen of Asp236 and that of ligand

nitrogen from ((2-(piperidine-1-yl) ethyl) amine) methanol ring with a distance of 2.4 Å. In

addition to hydrogen bonding, hydrophobic interactions also reported to alter binding affinity and

efficacy of drugs (Patil et al., 2010), therefore, hydrophobic interactions were also analyzed. The

following residues were reported to be involved in hydrophobic interactions with the ligand: Ser11,

Ser12, Lys18, Gln98, Glu101, Gly184, Pro233, Gly234, Val235, and Asp240.

203

Table 6.1. Top five best characterized natural compounds A. baumannii KdsB enzyme.

Compound Structure

GOLD

fitness

score

AD Vina

binding

energy

(kcal/mol)

226

72.3 -5.7

2469

70.4 -5.3

3841

69.8 -5.1

1134

69.2 -5.0

204

1762

67.1 -5.1

Fig.6.2. Binding interactions of ligand into protein active pocket from GOLD analysis (left) and

AD-Vina (right).

205

The molecular weight of the compound is 438.6 g/mol, number of heavy atoms (32), number of

aromatic heavy atoms (12), number of rotatable bonds (9), hydrogen bond acceptors (5), hydrogen

bond donors (2), molar refractivity (133.2) and TPSA value of 53.9 Å². According to in silico

pharmacokinetic assays, the compound gastrointestinal absorption (GI) is very high (95.5 %) while

it’s in vivo blood-brain penetration (BBP) value is calculated as 2.5 C.bood/C.brain, suggesting its

high potential for reaching the target sites and treatment of A. baumannii CNS infections (Ahmad,

Raza, Uddin, & Azam, 2017). The high gastrointestinal absorption of the compound is due to its

lower molecular weight, TPSA value, and number of rotatable bonds (Wadood et al., 2017).

Permeability across biological membranes is one key factor in drug absorption and distribution in

biological systems (Alavijeh, Chishty, Qaiser, & Palmer, 2005). The in vitro MDCK

(Madin−Darby Canine Kidney) Cells assay demonstrated that the permeability of this compound

is 6.1 nm/sec while its Caco-2 cell permeability score is 50.9 nm/sec. The in vitro skin permeability

(log kp) of the drug is -2.9/cm/hour, while plasma protein binding efficacy of the drug is 82.4 %.

The high percent of plasma protein binding further allows the compound retention in blood for

extended period, which increases the chances of deliverying high proportion of the drug to target

sites (Wadood et al., 2017). The carcinogenicity assay in both mouse and rat revealed that the

compound is non-carcinogen and there is a medium risk of hERG (Human ether-a-go-go related

gene channel inhibition) protein inhibition (Lee et al., 2004). The Ames test experimentation in

Salmonella typhimurium strain TA100 of the drug pinpoint that the compound is non-mutagenic

and is not associated with any metabolic activation in the host cells (Lee et al., 2004). Lipophilicity

plays crucial role in design of potent compounds due to its role in solubility, penetration into tissue,

organs and through cell membranes, absorption and host cell proteins binding (Rutkowska, Pajak,

& Jóźwiak, 2013). In addition, this physicochemical property is also an important consideration in

recognition of ligand, binding to hERG protein (Yu, Zou, Wang, & Li, 2016), interaction with

PXR (pregnane X receptor) (Ma, Idle, & Gonzalez, 2008) and CYP450 of the liver (Flockhart &

Oesterheld, 2000). The consensus log P value of the shortlisted compound is 4.03, which is an

ideal parameter for druglikeness (Lipinski, 2004). From medicinal chemistry point of view, the

synthetic accessibility score of the compound is 4.05, which indicates that the synthesis of the

compound is easy (Daina, Michielin, & Zoete, 2017). Further, the compound does not possess any

structure for pan-assay interference. This means that this inhibitor can bind and affect only

206

desirable target (Baell & Holloway, 2010). The compound cannot be classified as lead like because

of higher molecular weight (>350), log P value > 3.5, and number of rotatable bonds > 7.

6.4.2. Trajectories Analysis

To further validate the docking findings and to understand the dynamics, the complex system of

compound-226 was subjected to a production run of 200 ns. Complex stability in aspect of RMSD

was monitored first, which explained the fluctuation of the complex in hydrated environment and

present important findings about ligand movement during the course of simulation as shown in the

Fig.6.3 (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016). The system was observed to have highly

appreciable stability through the entire simulation time with an average RMSD value of 2.1 Å

(Fig.6.4A). Small deviations in RMSD graphs were witnessed after an initial surge. The RMSD

scores at each ns were deduced within acceptable range of stability. The sudden minor shifts in the

pattern of RMSD were mainly because of the moving and rotating inhibitor, exerting a pressure

on structurally flexible loop regions of the enzyme. An initial brief surge in RMSD for first 5 ns

of the system determines the protein structural changes due to displacement of ligand from initial

binding site. During this time, the inhibitor takes a sudden leap into the protein cavity and this

could be hypothesized as a step towards proper binding mode in the initial docked site. Afterwards,

the protein structure was observed to be stable till 16 ns with RMSD value in the range from 2-2.4

Å. This stability can be owed to loose bonding of the inhibitor with protein active residues, as a

result of which the exerted pressure on the protein loop region was less. In the next two ns, the

compound started to move further away and deep inside the cavity and formed strong contacts

with the catalytic residues from the opposite side. Following, a sudden minor increase pattern in

RMSD value was observed and was disclosed as ligand movement induced loop regions mobility.

A downward slope was detected till 34th ns which means that the inhibitor now starts a backward

journey from the cavity floor to its original site. As intermolecular interactions at this point were

weak, the net effect on the enzyme structure loop regions was less resulting in a stable protein

structure. The highest RMSD value of 2.8 Å was reported at 64th ns and this could be possibly due

to the structural flexibility of the protein loop regions. Superimposition of complex at 64 ns over

the one obtained from the 1st ns revealed that configuration variations in shape of loop regions

extension, shortening and displacement occurred, which seems to be important for proper

adjustment of the rotating and moving inhibitor. During this, the compound tried to come closer

207

to the initial residues and in the way attached to the neighboring residue of Arg181 through two

strong hydrogen bonds at a distances of 1.9 Å and 2.2 Å. Examining further, after 64th ns and

towards the end of simulation, the ligand was observed to exhibits slow and steady moves toward

initial binding site and continuously changes its binding mode. The change of these modes reflects

to achievement of appropriate and proper orientation to get the best fitting pose in the original

binding region. Throughout the entire simulation scale, no considerable structural shift was

reported, which explicates enzyme structure stability and strong binding affinity of the compound

towards the binding pocket of AbKdsB enzyme. Further assistance was sorted from RMSF

analysis, which provided an in-depth mobility data of each residue of the protein during the whole

simulation stretch (Fig.6.4B) (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Haq, Abro, Raza, Liedl,

& Azam, 2017). The RMSF values were investigated for C-alpha atoms of each residue to unravel

their average displacement. On average basis, the system RMSF value was noted at 1 Å with

highest fluctuation was observed for residues Asp166, Gly167, Ala168 and Lys169 with RMSF

value of 4.3 Å, 4.3 Å, 4.4 Å, and 4.5 Å, respectively. The RSMF fluctuation of these values can

be summarized by the presence of these residues in the loop region. Loop regions are reported to

be highly unstable during simulation to allow the stable accommodation and tight binding of ligand

in its active catalytic cavity (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016). Majority of the residues

from enzyme active pocket were noticed as highly stable, despite of their presence in the loop

region. Asp236, which is a key functional residue of the enzyme reflects high stability of 0.8 Å.

Other vital residues i.e. Ser13, Arg14, Gln98, Glu101, Gly184, and Tyr186 have a RMSD value

of 1.4 Å, 1.2 Å, 0.5 Å, 0.5 Å, 0.6 Å, and 0.4 Å, respectively. The low RMSD of these residues

illustrates the highly stable conformation of protein active pocket and implies that the inhibitor has

a great affinity for the enzyme pocket. The thermal disorderness of the system was evaluated with

B-factor enquiry. Analogous results of B-factor were observed when compared to that of RMSF

values, suggesting that B-factor complements RMSF investigation for residue flexibility and

complex structure stability (Fig.6.4C). B-factor interprets that protein residues from 166-169

depict high fluctuations; the same trend was observed during RMSF. Overall, both the graphs

illustrated similar trend of system stability (Fig.6.4D). Further proofs of total system stability were

acquired from Rg (Haq, Abro, Raza, Liedl, & Azam, 2017). Rg determines the system equilibrium

conformation and overall compactness and variation in their values, indicate variation in the

protein structure compactness. Higher the value of Rg, loose will be the compactness of the system

208

atoms, whereas lower Rg values implies tight packing of the system atoms. The average Rg value

observed was 21.5 Å with peak value of 22.2 Å and bottom value of 20.9 Å. These findings depict

that protein-inhibitor complex underwent minor variations during simulation time scale, and

hence, the protein structure is stable.

Fig.6.3. The back and forth rotating movement of inhibitor in KdsB enzyme. The coils, helix,

strand, and inhibitor are colored green, cyan, and magenta, respectively.

209

Fig.6.4. Statistical parameters for analyzing docked enzyme-inhibitor complex through RMSD (A), RMSF (B), β-factor (C) and Rg

(D).

210

6.4.3. Binding Pattern Analysis

Binding pattern analysis of the complex, in particular, hydrogen bonding between the enzyme and

the best characterized inhibitor throughout the entire period of simulation was assessed through an

indigenously designed Perl script in VMD, which screened out all the hydrogen bonds between

enzyme active site residues and ligand atoms (S-Table 6.1). Only residues surrounding ligand

within 5 Å distance were considered. The aim of this exercise was to get insight into the role of

residues in stabilizing the complex at the start, middle, and towards the end of simulation. It was

observed that in the start, the complex stability was achieved mainly by two important residues of

enzyme active site, Asp236, and Arg14. Binding pattern analysis of the docked complex shows

that both these residues in addition to Asp100 and Ser13 are in close contact with ligand (Fig.6.2).

The ligand during the start of simulation tends to adjust itself in binding cavity of the enzyme,

losing its hydrogen bond from Asp100 and Ser13, while remaining attached to Asp236 and Arg14.

At 50 ns, the complex was observed to remain highly stable. The enzyme active site residues

Gln98, Glu101, Gly184, and Tyr186 were found in close vicinity of the ligand with a distance of

2 Å, 1.8 Å, 2.5 Å, and 2.7 Å, respectively. Proceeding further, at 100 ns the ligand was found

attached to Gln98 and additional Gly234 residue of the active cavity. This supports that during

initial 100 ns of simulation, continuous rotating movements of the ligand away, downward, and

towards its first contact point in the binding pocket happened. During this period, the ligand was

noticed to move towards the opposite wall of the active site, while maintaining its contact with the

initial binding site through both hydrogen and hydrophobic interactions. It was revealed that at

150 ns, the Aps236 residue through its oxygen atoms and Arg14 through its nitrogen propel the

ligand to move back towards it initial contact site, and anchor the ligand here. Very strong

hydrogen bonding between the Asp236 oxygen atom and ligand hydrogen and nitrogen atoms were

reported. Toward the end of 200 ns, the ligand tends to stabilize in the initial binding site by

forming strong hydrogen bonds with Asp236 oxygen atoms through its nitrogen and hydrogen

atoms. The 2D interactions between ligand and enzyme active cavity are presented in Fig.6.5.

211

Fig.6.5. Binding interactions between ligand and enzyme active site residues, during different time scale of simulation (a) at 50 ns, (b) 100 ns, (c) at

150 ns, and (d) at 200 ns.

212

6.4.4. RDF Analysis

The stability of simulated system towards the end of simulation was inspected through RDF. RDF

is used to explain the distribution of atoms, molecules, and species around a point reference

(Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Donohue, 1954). Last production run comprising the

concluding 10 ns of simulation was analyzed for complex stability. It was observed that the ligand

binds stably in the protein active pocket, mainly through close hydrogen contacts. The receptor

protein active site residues (Arg14, Ser141, Gly234, and Asp236) by virtue of hydrogen bonding

entrapped the inhibitor in its pocket. Specifically, the residue Asp236 is an important contributor

to overall stability by forming four hydrogen bonds through highly electronegative oxygen atoms

with ligand nitrogen and hydrogen. In addition, the residue was reported in the initial phases of

simulation and can be considered as a significant residue in anchoring inhibitor in protein cavity.

RDF graphs were generated for all the four hydrogen interaction of Asp236 over the course of last

10 ns (Fig.6.6). A distorted graph was generated for Asp236: OD1 and ligand N1 atom. The

distribution of ligand atom was mostly noticed at 3.9 Å. The highest distribution was explained at

3.8 Å with g(r) value of 0.8. In case of OD2 from Asp236 and Ligand N1 atom, the highest

distribution of the ligand atom was observed at 2.8 Å with an average g(r) value of 0.8. The graph

is narrow with the magnitude of 0.9 Å towards the end of the simulation, an indicator, which means

that the ligand atom comes in close contact with active cavity enhancing complex stability and

strength. For Asp236: OD1 and ligand H15, a distorted graph was reported. An average

distribution of ligand atom was revealed at 2.4 Å with g (r) value of 0.04, suggesting that the ligand

atom spend most of the time in close proximity of this catalytic residue. The highest peak was seen

at 3.5 Å with g(r) value of 0.2. The graph width was illustrated as modest with the value of 1.8 Å.

For Asp236: OD2 with respect to ligand H15 atom, a more refine graph was obtained. The highest

peak was observed at 1.8 Å with g(r) value of 1.6. The ligand atom with reference to residue atom

spends most of its time at a distance from 1.8 Å to 2 Å, which deciphered the close proximity

between the atoms, and hence, the overall stability of the complex. A distance of 2.4 Å was

revealed between the atoms when compared to the initial time of simulation with a distance of 3.7

Å.

213

Fig.6.6. RDF graphs for hotspot amino acids involved in stability of enzyme-inhibitor complex

stability towards the last 10 ns of simulation period.


Binding free energies illustrate the affinity of all the five ligands for protein active site and are

thought to be a viable evaluation parameter in the identification of lead compounds and provide

mechanistic insights about the binding ligands (Slynko et al., 2016). MM(PB/GB)SA method of

AMBER14 demonstrated robust interactions between enzyme-inhibitors complexes which can be

seen in Table 6.2. Due to convergence problem, the entropy calculations were excluded. In case

of MM/GBSA, it was analyzed that formation of complexes lead to the highly favorable

electrostatic contribution of -435.9 kcal/mol, -356.7 kcal/mol, -415.5 kcal/mol, -415.5, -285.2 and

-311.3 for KdsB-226, KdsB-2469, KdsB-3841, KdsB-1134 and KdsB-1162, respectively.

Similarly, MM/PBSA based electrostatic energy contribution was also revealed dominant and was

214

found in the following order: KdsB-226 (-435.6 kcal/mol), KdsB-2469 (-345.1 kcal/mol), KdsB-

3841 (-351.3 kcal/mol), KdsB-1134 (-204.1 kcal/mol) and KdsB-1162 (-301.5 kcal/mol). Highly

favorable contribution towards KdsB-inhibitor complexes is also aided by van der Waal energy

value of -78.0 kcal/mol (KdsB-226), -36.5 kcal/mol (KdsB-2469), -58.3 kcal/mol (KdsB-3841), -

25.1 kcal/mol (KdsB-1134) and -31.1 kcal/mol (KdsB-1162) from MM/GBSA. Likewise from

MM/PBSA, favorable van der Waal energy for complexes was observed: -78.4 kcal/mol (KdsB-

226), -32.1 kcal/mol (KdsB-2469), -47.3 kcal/mol (KdsB-3841), -8.3 kcal/mol (KdsB-1134) and

-16.5 kcal/mol (KdsB-1162). In contrast, major non-favorable contribution towards complexes

binding affinity came from solvation energy with the binding energy of 134.7 kcal/mol (KdsB-

226), 107.2 kcal/mol (KdsB-2469), 119.4 kcal/mol (KdsB-3841), 64.2 kcal/mol (KdsB-1134),

84.3 kcal/mol (KdsB-1162) and 181.3 kcal/mol (KdsB-226), 114.2 kcal/mol (KdsB-2469), 92.9

kcal/mol (KdsB-3841), 51.2 kcal/mol (KdsB-1134), 74.1 kcal/mol (KdsB-1162), respectively

form MM/GBSA and MM/PBSA methods.

Table 6.2. Binding free energy values for the top five docked complexes.

Complex MM/GBSA

Contributions

Values

(kcal/mol)

MM/PBSA

Contributions

Values

(kcal/mol)

Energy

Difference in

kcal/mol

(MM/GBSA-

MM/PBSA)

KdsB-226

ΔEele,GB -435.9 ΔEele,PB -435.6 -0.3

ΔEgas,GB -306.4 ΔEgas,PB -306.1 -0.3

ΔGsolv,GB 134.7 ΔGsolv,PB 181.3 -46.6

Htot,GB -51.2 Htot,PB -9.1 -42.1

ΔEvdw,GB -78.0 ΔEvdw,PB -78.4 0.4

KdsB-2469

ΔEele,GB -356.7 ΔEele,PB -345.1 -11.6

ΔEgas,GB -245.2 ΔEgas,PB -227.8 -17.4

ΔGsolv,GB 107.2 ΔGsolv,PB 114.2 -7

Htot,GB -36.2 Htot,PB -6.3 -29.9

ΔEvdw,GB -36.5 ΔEvdw,PB -32.1 -4.4

215

KdsB-3841

ΔEele,GB -415.5 ΔEele,PB -351.3 -64.2

ΔEgas,GB -288.8 ΔEgas,PB -198.6 -90.2

ΔGsolv,GB 119.4 ΔGsolv,PB 92.9 26.5

Htot,GB -39.4 Htot,PB -4.6 -34.8

ΔEvdw,GB -58.3 ΔEvdw,PB -47.3 -11

KdsB-1134

ΔEele,GB -285.2 ΔEele,PB -204.1 -81.1

ΔEgas,GB -165.2 ΔEgas,PB -102.3 -62.9

ΔGsolv,GB 64.2 ΔGsolv,PB 51.2 13

Htot,GB -26.6 Htot,PB 1.2 -27.8


KdsB-1162

ΔEele,GB -311.3 ΔEele,PB -301.5 -9.8

ΔEgas,GB -198.2 ΔEgas,PB -158.9 -39.3

ΔGsolv,GB 84.3 ΔGsolv,PB 74.1 10.2

Htot,GB -29.7 Htot,PB -1.0 -28.7


The overall binding free energy (Htot) obtained through MM/GBSA approach for the systems is -

51.2 kcal/mol (KdsB-226), -36.2 kcal/mol (KdsB-2469), -39.4 kcal/mol (KdsB-3841), -26.6

kcal/mol (KdsB-1134) and -29.7 kcal/mol (KdsB-1162). For MM/PBSA the total binding affinity

was reported as -9.1 kcal/mol (KdsB-226), -6.3 kcal/mol (KdsB-2469), -4.6 kcal/mol (KdsB-

3841), 1.2 kcal/mol (KdsB-1134) and -1.0 kcal/mol (KdsB-1162), an indicator of stable nature of

the complexes. The difference in total binding affinity is possibly due to polar solvation energy,

calculated in both the approaches, which is higher in MM/GBSA approach in comparison to

MM/PBSA (Abro & Azam, 2016). Molecular interactions in a polar solvent are driven by polar

and nonpolar components, where nonpolar interactions and their respective energy contributions

are favorable. The pattern of fluctuations in both the approaches was found same. In order to

achieve convergence in the values of both approaches and to obtain consistent absolute binding

energies for the system, much higher time scale simulation is required. The energy of the top

system (KdsB-226) was decomposed into per residue of the receptor enzyme to get more

216

understanding about the hot spot amino acid of the enzyme active pocket that contributes to the

overall binding energy of the complex (Fig.6.7).

Fig.6.7. Total binding free energy decomposition per residue of the receptor enzyme based on

MM/GBSA method for KdsB-226 complex.

Residues having energy contribution of <1 kcal/mol are termed as hotspot amino acids due to their

significant and indispensable role in stabilization of enzyme-ligand complex (Haq, Abro, Raza,

217

Liedl, & Azam, 2017). Active site residue of the enzyme that lies within 5 Å of the ligand include

Ser12, Ser13, Arg14, Lys18, Gln98, Asp100, Glu101, Gly234, Val235, Thr237, Asp240, and

Ser241. These residues were observed to have energy contribution <-1 kcal/mol, highlighting their

role in complex stability and anchoring the ligand in enzyme active pocket. Furthermore, per frame

total binding free energy was also considered (Fig.6.8). Total of 1000 frames each after 0.2 ns

were investigated. The highest binding affinity of -120 kcal/mol, while the lowest binding affinity

of -22 kcal/mol was observed. In MM(GB/PB)SA, binding free energy is calculated by picking

snapshots at regular intervals from the entire simulation trajectories, while Waterswap performed

its own MC simulation and calculate the binding affinity of ligand and identical water clusters

towards protein active site (Woods et al., 2014). The MM(GB/PB)SA, although is a popular

method for binding free energy calculation, yet it presents several limitations. In MM(GB/PB)SA,

the difference between the hydration energy of protein-ligand complex and that of protein are

usually large and need to be evaluated as any variation or error in their values lead to a greater

effect on the overall binding affinity. More importantly, as MM(GB/PB)SA uses an implicit water

model implicating that molecular interaction detail of protein-water and ligand-water are skipped.

This information is of particular importance in cases where water molecules are involved in

bridging interactions between the protein and ligand (Woods et al., 2014). To circumvent this

shortcoming, we used a more sophisticated assay of Waterswap. Waterswap revealed a high

affinity of the inhibitor towards the protein active site as suggested by the poor convergence values

of PEF, TI, and Bennett algorithm, as shown in Table.6.3. In general, these findings are in direct

correlation with those obtained from MM(GB/PB)SA analysis for KdsB-226 complex. In case of

MM/GBSA, the total binding free energy (Htot,GB) is much higher for the complex i.e. -51.2

kcal/mol-1 compared to that predicted by FEP (-21.6 kcal.mol-1), TI (-22.9 kcal/mol-1) and Bennett

(-22.6 kcal.mol-1) in Waterswap. However, the FEP, TI and Bennett total binding free energy in

Waterswap is higher than that estimated by MM/PBSA for KdsB-226 complex i.e. -9.1 kcal/mol-

1.

218

Table 6.3. Binding free energy from Monte Carlo calculations in Waterswap.

Algorithm Binding free energy

(Kcal/mol)

FEP - 21.6

TI - 22.9

Bennetts - 22.6

The values are different within < 1 kcal/mol, suggesting the good agreement of the algorithms on

the high affinity of the ligand in the protein active site as well as more reasonable. This further

indicates that in comparison to water cluster, the ligand is more stable, and has high affinity in the

binding pocket of the enzyme. The total, bound and free energy of the system can be seen in S-

Fig.6.5. Similar like MM/GBSA, the Waterswap binding free energy of each iteration was

decomposed into the most active protein residues involved in frequent interaction with the ligand

and allowed its stable binding in its binding in the active pocket. These residues including Asp100,

Glu101, Asp236, Asp240, and Asp242 are only discussed here, due to their higher contributions

in ligand binding towards the end of simulation (Fig.6.9). Among these, the higher contributions

to overall binding affinity came from aforementioned Asp236. Overall, in all iterations, Asp236

was found to have a binding energy of > -50 kcal/mol, indicating its major role in anchoring the

inhibitor in binding cavity of enzyme both at the start and end of the simulation period. It was

revealed that interactions with ligand were mainly driven by hydrophilic amino acids, emphasizing

further on the highly stable nature of the complex. Asp100 and Glu101 were reported to have the

same affinity in the start, however, it was depicted towards the end in terms of ligand binding

Asp100 contributed more as compared to Glu101. The Asp240 binding energy was in the range

from -27 to -47 kcal/mol, while Asp242 binding affinity was mostly consistent in -20 kcal/mol

range. The residues like Asp100, Glu101, and Asp240 were also reported by MM/GBSA to have

affinity towards ligand binding though lesser than that predicted by Waterswap.

219

Fig.6.8. Total binding free energy for 1000 frame extracted from 200 ns of simulation trajectories.

220

Fig.6.9. The binding energy of KdsB residues that contribute significantly to the overall binding affinity of the complex.

221

6.5. Conclusions

Since, KdsB is highly specific and selective; it is an attractive target for limiting the dissemination

of drug resistant clones of bacterial pathogens. Molecular docking in combination with

druglikeness and pharmacokinetics indicated compound-226 as a potent ligand against KdsB. MD

simulations for 200 ns depicted the highly stable nature of the complex with the Asp236 residue

of enzyme active pocket as a major residue in anchoring and stabilizing the ligand through the

entire course of the simulation. Binding free energy through MM(PB/GB)SA predicted that

electrostatic energy is the major contributor to the overall binding free energy of the complex. To

overcome the limitation of MM(PB/GB)SA approach, we used, in addition, a more sophisticated

Waterswap assay for the absolute binding free energy of the complex. The analysis outcomes were

in strong agreement with the complex stability and provide insights about the major contribution

of Asp100, Glu101, Asp236, Asp240, and Asp242 towards overall binding affinity. We strongly

believe that the finding of this study could provide a basis for designing novel antibiotics against

this vital Gram-negative bacterial target.


S-Fig.6.1. Gram-negative bacteria envelope.

S-Fig. 6.2. Four sequential steps of KDO biosynthesis.

S-Fig.6.3. Binding cavity prediction through GHECOM (A) and Docsite Scorer (B).

S-Fig.6.4. Binding interactions (Left) and binding mode (Right) of the top docked drug-like

compound (shown in gold color) in the catalytic pocket of KdsB enzyme (shown in dodger blue

in color).

S-Fig.6.5. Total, bound and free binding energy decomposed per iteration of Waterswap assay.

S-Table 6.1. Hydrogen bonds between KdsB enzyme and the screened inhibitor at different time

scale of simulation protocol

6.7. References



222

221, 507–517.

Abro, A., & Azam, S. S. (2016). Binding free energy based analysis of arsenic (+ 3 oxidation

state) methyltransferase with S-adenosylmethionine. Journal of Molecular Liquids, 220,

375–382.

Ahmad, S., Raza, S., Uddin, R., & Azam, S. S. (2017). Binding mode analysis, dynamic

simulation and binding free energy calculations of the MurF ligase from Acinetobacter

baumannii. Journal of Molecular Graphics and Modelling,77, 72-85.

Alavijeh, M. S., Chishty, M., Qaiser, M. Z., & Palmer, A. M. (2005). Drug metabolism and

pharmacokinetics, the blood-brain barrier, and central nervous system drug discovery.

NeuroRx, 2(4), 554–571.

Andersen, H. C. (1980). Molecular dynamics simulations at constant pressure and/or temperature.

The Journal of Chemical Physics, 72(4), 2384–2393.



using triphenylgermane as a catalyst and its binding pattern validation. RSC Advances,

6(83), 79651–79661.

Baell, J. B., & Holloway, G. A. (2010). New substructure filters for removal of pan assay

interference compounds (PAINS) from screening libraries and for their exclusion in

bioassays. Journal of Medicinal Chemistry, 53(7), 2719–2740.


against Shigella sonnei: A subtractive reverse vaccinology based approach. Biologicals,

50, 87-99.

Berman, H. M., Westbrook, J., Feng, Z., Gilliland, G., Bhat, T. N., Weissig, H., … Bourne, P. E.

(2006). The Protein Data Bank, 1999. In International Tables for Crystallography Volume

F: Crystallography of Biological Macromolecules (pp. 675–684). Springer.


14.

Cavasotto, C. N., Orry, W., & Andrew, J. (2007). Ligand docking and structure-based virtual

screening in drug discovery. Current Topics in Medicinal Chemistry, 7(10), 1006–1014.

Cipolla, L., Polissi, A., Airoldi, C., Galliani, P., Sperandeo, P., & Nicotra, F. (2009). The Kdo

biosynthetic pathway toward OM biogenesis as target in antibacterial drug design and

https://doi.org/10.1016/j.jmgm.2017.07.024

https://www.sciencedirect.com/science/journal/10451056/50/supp/C

223

development. Current Drug Discovery Technologies, 6(1), 19–33.

Clark, M., Cramer, R. D., & Van Opdenbosch, N. (1989). Validation of the general purpose

Tripos 5.2 force field. Journal of Computational Chemistry, 10(8), 982–1012.

Copeland, R. A. (2013). Evaluation of enzyme inhibitors in drug discovery: a guide for medicinal

chemists and pharmacologists. John Wiley & Sons.




Donohue, J. (1954). Radial distribution functions of some structures of the polypeptide chain.

Proceedings of the National Academy of Sciences, 40 (6), 377-381.

Elengoe, A., Naser, M. A., & Hamdan, S. (2014). Modeling and docking studies on novel mutants

(K71L and T204V) of the ATPase domain of human heat shock 70 kDa protein 1.

International Journal of Molecular Sciences, 15(4), 6797–6814.

Flockhart, D. A., & Oesterheld, J. R. (2000). Cytochrome P450-mediated drug interactions. Child

and Adolescent Psychiatric Clinics of North America, 9(1), 43–76.



Green, D. V. S. (2008). Virtual screening of chemical libraries for drug discovery. Expert Opinion

on Drug Discovery, 3(9), 1011–1026.





Hamelberg, D., & McCammon, J. A. (2004). Standard free energy of releasing a localized water

molecule from the binding pockets of proteins: double-decoupling method. Journal of the

American Chemical Society, 126(24), 7683–7689.



143–152.

Hemmer, M. C., Steinhauer, V., & Gasteiger, J. (1999). Deriving the 3D structure of organic

molecules from their infrared spectra. Vibrational Spectroscopy, 19(1), 151–164.

224

Heyes, D. J., Levy, C., Lafite, P., Roberts, I. S., Goldrick, M., Stachulski, A. V, … others. (2009).

Structure-based Mechanism of CMP-2-keto-3-deoxymanno-octulonic Acid Synthetase

Convergent evolution of a sugar-activating enzyme with dna/rna polymerases. Journal of

Biological Chemistry, 284(51), 35514–35523.

Hevener, K. E., Zhao, W., Ball, D. M., Babaoglu, K., Qi, J., White, S. W., & Lee, R. E. (2009).

Validation of molecular docking programs for virtual screening against dihydropteroate

synthase. Journal of Chemical Information and Modeling, 49(2), 444–460.

Hospital, A., Goñi, J. R., Orozco, M., & Gelpli, J. L. (2015). Molecular dynamics simulations:


AABC, 8, 37.

Hsu, L.Y., Apisarnthanarak, A., Khan, E., Suwantarat, N., Ghafur, A., & Tambyah, P. A. (2017).

Carbapenem-resistant Acinetobacter baumannii and Enterobacteriaceae in South and

Southeast Asia. Clinical Microbiology Reviews, 30(1), 1–22.



Hubbard, R. E., & Kamran Haider, M. (2010). Hydrogen bonds in proteins: role and strength.

eLS.


of a genetic algorithm for flexible docking. Journal of Molecular Biology, 267(3), 727–

748.

John, A., Sivashanmugam, M., Umashankar, V., & Natarajan, S. K. (2017). Virtual screening,

molecular dynamics, and binding free energy calculations on human carbonic anhydrase

IX catalytic domain for deciphering potential leads. Journal of Biomolecular Structure and

Dynamics, 35(10), 2155–2168.

Kadam, R. U., & Roy, N. (2007). Recent trends in drug-likeness prediction: A comprehensive

review of In silico methods. Indian Journal of Pharmaceutical Sciences, 69(5), 609.

Krautler, V., Van Gunsteren, W. F., & Hünenberger, P. H. (2001). A fast SHAKE algorithm to



https://doi.org/10.1006/jmbi.1996.0897

225

Kumar, A., Srivastava, G., Negi, A. S., & Sharma, A. (2018). Docking, molecular dynamics,

binding energy-MM-PBSA studies of naphthofuran derivatives to identify potential dual

inhibitors against BACE-1 and GSK-3β. Journal of Biomolecular Structure and

Dynamics,1–37.


pc-based program for batch prediction of ADME properties. EuroQSAR, 2004, 5–9.

Leeson, P. (2012). Drug discovery: Chemical beauty contest. Nature, 481(7382), 455–456.

Lionta, E., Spyrou, G., K Vassilatis, D., & Cournia, Z. (2014). Structure-based virtual screening

for drug discovery: principles, applications and recent advances. Current Topics in

Medicinal Chemistry, 14(16), 1923–1938.



Ma, X., Idle, J. R., & Gonzalez, F. J. (2008). The pregnane X receptor: from bench to bedside.

Expert Opinion on Drug Metabolism & Toxicology, 4(7), 895–908.

Miesel, L., Greene, J., & Black, T. A. (2003). Genetic strategies for antibacterial drug discovery.

Nature Reviews. Genetics, 4(6), 442.

Naz, S., Farooq, U., Ali, S., Sarwar, R., Khan, S., & Abagyan, R. (2018). Identification of new

Benzamide inhibitor against α-subunit of Tryptophan synthase from Mycobacterium

tuberculosis through Structure based virtual screening, Anti-tuberculosis activity and

Molecular dynamics simulations. Journal of Biomolecular Structure and Dynamics, 1–24.

Osborn, M. J. (1979). Biosynthesis and assembly of the lipopolysaccharide of the outer

membrane. Bacterial Outer Membranes, 15–34.

Panman, W., Nutho, B., Chamni, S., Dokmaisrijan, S., Kungwan, N., & Rungrotmongkol, T.

(2017). Computational screening of fatty acid synthase inhibitors against thioesterase

domain. Journal of Biomolecular Structure and Dynamics, 1–12.



Patil, R., Das, S., Stanley, A., Yadav, L., Sudhakar, A., & Varma, A. K. (2010). Optimized

hydrophobic interactions and hydrogen bonding at the target-ligand interface leads the

226

pathways of drug-designing. PloS One, 5(8), e12029




Raj, U., Kumar, H., & Varadwaj, P. K. (2017). Molecular docking and dynamics simulation study

of flavonoids as BET bromodomain inhibitors. Journal of Biomolecular Structure and

Dynamics, 35(11), 2351–2362.



molecular dynamic study. Journal of Molecular Liquids, 242, 1184-1198.

Roe, D. R., & Cheatham, T. E. (2013). PTRAJ and CPPTRAJ: software for processing and

analysis of molecular dynamics trajectory data. Journal of Chemical Theory and

Computation, 9(7), 3084–3095.

Rutkowska, E., Pajak, K., & Jozwiak, K. (2013). Lipophilicity--methods of determination and its

role in medicinal chemistry. Acta Poloniae Pharmaceutica, 70(1), 3–18.

Shoichet, B. K. (2004). Virtual screening of chemical libraries. Nature, 432(7019), 862.

Shyu, C., & Ytreberg, F. M. (2009). Reducing the bias and uncertainty of free energy estimates

by using regression to fit thermodynamic integration data. Journal of Computational

Chemistry, 30(14), 2297–2304.

Sikander Azam, S., Wajid Abbasi, S., & Tahir, S. (2014). Investigation of novel chemical

inhibitors of human lysosomal acid lipase: virtual screening and molecular docking studies.

Combinatorial Chemistry & High Throughput Screening, 17(5), 473–482.

Slynko, I., Schmidtkunz, K., Rumpf, T., Klaeger, S., Heinzlmeir, S., Najar, A., … others. (2016).

Identification of Highly Potent Protein Kinase C-Related Kinase 1 Inhibitors by Virtual

Screening, Binding Free Energy Rescoring, and in vitro Testing. ChemMedChem, 11(18),

2084–2094.

Smyth, K. M., & Marchant, A. (2013). Conservation of the 2-keto-3-deoxymanno-octulosonic

acid (Kdo) biosynthesis pathway between plants and bacteria. Carbohydrate Research,

380, 70–75.

Todar, K. (2002). Mechanisms of bacterial pathogenicity: endotoxins. Todar’s Online Textbook

of Bacteriology.

https://doi.org/10.1016/j.molliq.2017.07.086

227




Tyka, M. D., Sessions, R. B., & Clarke, A. R. (2007). Absolute free-energy calculations of liquids

using a harmonic reference state. The Journal of Physical Chemistry B, 111(32), 9571–

9580.

Valvano, M. A., Furlong, S. E., & Patel, K. B. (2011). Genetics, biosynthesis and assembly of O-

antigen. In Bacterial Lipopolysaccharides (pp. 275–310). Springer.

Verma, P., Tiwari, M., & Tiwari, V. (2017). In silico high-throughput virtual screening and

molecular dynamics simulation study to identify inhibitor for AdeABC efflux pump of

Acinetobacter baumannii. Journal of Biomolecular Structure and Dynamics, 1–13.

Vistoli, G., Pedretti, A., & Testa, B. (2008). Assessing drug-likeness-what are we missing? Drug

Discovery Today, 13(7), 285–294.

Volkamer, A., Kuhn, D., Rippmann, F., & Rarey, M. (2012). DoGSiteScorer: a web server for

automatic binding site prediction, analysis and druggability assessment. Bioinformatics,

28(15), 2074–2075.

Vukic, V., Hrnjez, D., Milanović, S., Iličić, M., Kanurić, K., & Petri, E. (2015). Comparative

molecular modeling and docking analysis of β-galactosidase enzymes from commercially

important starter cultures used in the dairy industry. Food Biotechnology, 29(3), 248–262.

Vyas, V., Jain, A., Jain, A., & Gupta, A. (2008). Virtual screening: a fast tool for drug design.

Scientia Pharmaceutica, 76(3), 333–360.

Wadood, A., Ghufran, M., Hassan, S. F., Khan, H., Azam, S. S., & Rashid, U. (2017). In silico

identification of promiscuous scaffolds as potential inhibitors of 1-deoxy-d-xylulose 5-

phosphate reductoisomerase for treatment of Falciparum malaria. Pharmaceutical

Biology, 55(1), 19–32.




Whitfield, C., & Trent, M. S. (2014). Biosynthesis and export of bacterial lipopolysaccharides.

https://doi.org/10.1080/08905436.2015.1059766

https://doi.org/10.1080/13880209.2016.1225778

228

Annual Review of Biochemistry, 83, 99–128.



Willyard, C. (2017). The drug-resistant bacteria that pose the greatest health threats. Nature News,

543(7643), 15.



Modeling, 45(1), 160–169.

Woods, I., & Kollman, P. A. (2000). Combined molecular mechanical and continuum solvent

approach (MM-PBSA/GBSA) to predict ligand binding. Perspectives in Drug Discovery

and Design, 18(1), 113–135.

Woods, C. J., Malaisree, M., Hannongbua, S., & Mulholland, A. J. (2011). A water-swap reaction

coordinate for the calculation of absolute protein-ligand binding free energies. The Journal

of Chemical Physics, 134(5), 02B611

Woods, C. J., Malaisree, M., Long, B., McIntosh-Smith, S., & Mulholland, A. J. (2013).

Computational assay of H7N9 influenza neuraminidase reveals R292K mutation reduces

drug binding affinity. Scientific Reports, 3, 3561.

Woods, C. J., Malaisree, M., Michel, J., Long, B., McIntosh-Smith, S., & Mulholland, A. J.

(2014). Rapid decomposition and visualisation of protein -ligand binding free energies by

residue and by water. Faraday Discussions, 169, 477–499.

Yi, L. (2009). Studies of 3-Deoxy-D-manno-octulosonate 8-phosphate phosphatase: Mechanistic

insights and a gene fusion example. University of Michigan.

Yu, H., Zou, B., Wang, X., & Li, M. (2016). Investigation of miscellaneous hERG inhibition in

large diverse compound collection using automated patch-clamp assay. Acta

Pharmacologica Sinica, 37(1), 111.

Zvelebil, M. J., Barton, G. J., Taylor, W. R., & Sternberg, M. J. E. (1987). Prediction of protein

secondary structure and active sites using the alignment of homologous sequences. Journal

of Molecular Biology, 195(4), 957–961.

https://doi.org/10.1016/0022-2836(87)90501-8

229

Chapter # 7

Identification of Natural Inhibitors against Acinetobacter

Baumannii D-Alanine-D-Alanine Ligase Enzyme: A Multi-

Spectrum In Silico Approach

7.1. Abstract

D-alanine-D-alanine ligase (Ddl), an enzyme that catalyzes the D-ala-D-ala dipeptide formation

in UDPMurNAc pentapeptide, is a part of peptidoglycan biosynthesis machinery. Inhibition of

enzyme leads to bacterial growth arrest making it a viable and attractive target for screening of

potent antibacterial drugs. Combination of virtual screening, molecular docking, in silico

pharmacokinetics, MD simulations, and binding free energy calculations based on MM/GBSA and

WaterSwap were applied in the current framework for the detailed analysis of potent natural

inhibitors against Ddl enzyme. Comparative molecular docking supported with computational

druglikeness revealed compound-331 (6-(4-((3-methoxyphenylsulfonamido) methyl) phenyl)-2-

methylnicotinamide) as the best docked inhibitor. The inhibitor has GOLD fitness score of 84.2

and AutoDock Vina binding energy of -7.2 kcal/mol. The inhibitor exhibited to have a better

druglikeness by adhering to Lipinski rule of five, Ghose rule, Veber filter, Egan filter, and Muegge

filter in contrast to D-cycloserine, which violates Ghose and Muegge rule. MD simulations unravel

that in complex over the course of 100-ns, the enzyme remained highly stable with mean RMSD

of 1.4 Å when compared to an undocked structure having RMSD of 1.6 Å. RMSF predicted stable

behaviour of the active site residues in both undocked and docked system with average RMSD

values of 1.2 Å and 0.7 Å, respectively. RDF and AFD demonstrated Lys176 and Trp177 as critical

residues of enzyme for binding, anchoring and bridging strong hydrogen and hydrophobic contacts

between enzyme and the inhibitor. The estimated MM/GBSA based binding free energy for the

complex is -47.36 kcal/mol, signifying its stable nature. WaterSwap further indicated a worthy

agreement on the affinity of inhibitor towards enzyme active pocket. This newly discovered natural

230

inhibitor might serve a parent structure for the development of more potent derivatives with

encouraging biological activity.

7.3. Introduction

A. baumannii is a Gram-negative bacilli that has emerged as a MDR pathogen, tough to control

and treat (Gonzalez-Villoria & Valverde-Garduno, 2016). The organism produced several types

of hospital-acquired infections including hospital-acquired pneumonia, bacteremia, endocarditis,

meningitis, soft and skin tissue infections, and UTIs (McConnell, Actis, & Pachon, 2013).

According to CDC, an estimated 12,000 A. baumannii associated infections occurred in the United

States each year. Approximately, 7,000 (63%) of these are MDR, which leads to 500 deaths (CDC,

2013). The clinical isolates of A. baumannii have been shown to confer resistance to majority of

the currently used antibiotics, including aminoglycosides, β-lactams, diaminopyrimidines,

macrolides, phenicols, quaternary amines, sulfonamides, and streptothricins (Taitt et al., 2013).

Moreover, A. baumannii is recently placed in “Critical” category based on the urgency of need for

the development of novel antibiotics (WHO, 2017).

The pathways that govern cell wall biosynthesis are generally a target of great interest against

bacterial pathogens for designing of novel antibiotics since they are essential and specific for

bacterial survival (Barton, 1999; Bugg, Braddick, Dowson, & Roper, 2011). One of the integral

structures of the cell wall is peptidoglycan that preserves cell integrity and ensures defined cell

shape (Barreteau et al., 2008). Peptidoglycan is a common feature of both Gram-positive and

Gram-negative bacteria and is an attractive target for antibacterial (Bugg, Braddick, Dowson, &

Roper, 2011). Peptidoglycan biosynthesis is broadly divided into three sequential steps; (i)

synthesis of nucleotide precursors in the cytoplasm, (ii) synthesis of lipid-linked intermediated on

the inner side of cytoplasm membrane, and (iii) polymerization reactions that occurred at outside

of cytoplasmic membrane (Barreteau et al., 2008). Each of these steps encompasses a series of

enzymatically catalyzed reactions, which occurred at different locations (Bugg, Braddick,

Dowson, & Roper, 2011; Skedelj et al., 2012). D-alanine-D-alanine ligase (Ddl) enzyme acts in

the first stage and utilized D-alanine as a substrate (Zawadzke, Bugg, & Walsh, 1991; Gholizadeh

et al., 2001). Ddl functions by ligating two D-alanine units to produce dipeptide D-alanyl-D-

alanine (D-ala-D-ala) as shown in S-Fig.1. D-ala-D-ala is a fundamental component of bacterial

231

cell that is essential for maintaining cell wall stability by cross-linking peptide chains of

peptidoglycan (Tytgat et al., 2009).

So far, four main kinds of Ddl inhibitors are reported: (i) D-Ala analogues, (ii) D-ala-D-ala

analogues, (iii) transition state analogues, and (iv) more recently those discovered by screening

and modeling methods (Tytgat et al., 2009; Skedelj et al., 2012). Among these, D-cycloserine (D-

4-amino-3-isoxazolidine) is the first discovered and most important (Neuhaus & Lynch, 1962). D-

cycloserine is an analogue of D-alanine having a Ki value of 27 µM and is the only clinically used

inhibitor mostly employed in a combination therapy for treatment of M. tuberculosis associated

infections (Neuhaus & Lynch, 1962). The use of D-cycloserine is now almost completely

abounded due to its high Minimum Inhibitory Concentration (MIC) value and neurological side

effects (Lu, Peng, Hwang, & Chen, 2008). Over the last decade, virtual screening (Kovač et al.,

2007; Kovač et al., 2008; Wu et al., 2008; Triola et al., 2009; Vehar et al., 2011) and de novo

structural-based drug design (Besong et al., 2005; Sova et al., 2009) have reported several new

scaffolds of Ddl inhibitors. Most of these inhibitors show no similarity with the substrate, reaction

intermediates, and product (Skedelj et al., 2012). Due to lack of potent Ddl inhibitors, the search

for novel inhibitor scaffolds against this target enzyme is urgently required for the routine use in

antibacterial therapy. In this current investigation, we documented the potential of natural

inhibitors against Ddl enzyme based on the applications of in silico drug discovery. The library of

natural inhibitors was subjected first to Lipinski’s rule of five to screen only drug-like compounds.

Structure based virtual screening was then followed to identify compounds with the best

orientation in enzyme active site and docking score. The functionality and physical behavior of the

enzyme were further deciphered to understand its dynamics in docked and undocked form. This

was especially important for the behavior of active pocket of the enzyme in the presence and

absence of inhibitor. Further, in complex the enzyme was analyzed for the first time through RDF

(Donohue, 1954) and AFD (Abro & Azam, 2016; Ahmad, Raza, Uddin, & Azam, 2017) to

examine the hydrogen-bonding patterns of vital residues from enzyme active site. Lastly, binding

free energy calculations were performed for enzyme complex to understand the inhibitor affinity

for active site. MM/GBSA (Miller et al., 2012) and a more recent and sophisticated approach of

WaterSwap (Woods, Malaisree, Hannongbua, & Mulholland, 2011; Woods et al., 2014) were

applied.

232

7.4. Materials and methods

7.4.1. Protein and Inhibitors Preparation

Since the crystal structure of A. baumannii Ddl is resolved at a resolution of 2.2 Å and is available

in PDB (ID: 5D8D) (Huynh et al., 2015), it was retrieved and subjected to initial preparation phase.

The enzyme is present in homodimer stereochemistry; each monomer with the independent

identical active site, so only one monomer was retained (Vukic et al., 2015). The protein structure

was relaxed by assigning Gasteiger charges and steric clashes were removed by minimizing the

protein through UCSF Chimera for 1500 rounds of minimization (750 steepest descent steps and

750 conjugate gradient steps). The steepest descent and conjugate gradient step size were set to

0.02 Å under TFF (Ahmad, Raza, Uddin, & Azam, 2017). For inhibitors, Asinex antibacterial

library (http://www.asinex.com/antibacterial_compound_library-html/) of natural compounds was

used. Virtual screening of the library based on Lipinski rule of five filters (Lipinski, 2004) was

first accomplished through Ligand scout 4.1(Wolber & Langer, 2005). The screened compounds

were minimized by the MMFF94 force field (Halgren, 1996).

7.4.2. Binding Cavity Prediction

Binding cavity prediction for docking of inhibitors was done through multiple approaches. First,

the catalytic cavity was predicted through online tools (Metapocket (Huang, 2009), DoGSiteScorer

(Volkamer, Kuhn, Rippmann, & Rarey, 2012). The predicted cavities were reconfirmed thorough

literature search (Huynh et al., 2015). After confirmation, coordinates of a highly conserved

catalytic residue, as point of inhibitors docking, was identified via multiple sequence alignment

(MSA) approach (Raza, Sanober, Rungrotmongkol, & Azam, 2017).


Molecular docking was performed to unveil the preferred binding mode of drug-like inhibitors into

protein active cavity and to understand their affinity toward the active pocket. Docking was carried

out using GOLD (Jones, Willett, & Glen, 1995) and AutoDock Vina (Trott & Olson, 2010). In

both cases, the receptor protein was kept rigid while the binding site for compounds was defined

to include all the residues for binding present within 10 Å of Glu247. GOLD docking was

accomplished through the automated GOLD wizard on Intel Xeon QuadTM Core processor of 3.0

GHz with Linux workstation. Default parameters were employed for genetic algorithm while

http://www.asinex.com/antibacterial_compound_library-html/

233

number of iterations was set to 10. GOLD fitness score was assigned to each compound, which is

based on the following equation.

GOLD fitness score = Shb-ext + SvdW-ext +Shb-int + Sint……………………………………...…(XXXI)

In the above equation, Shb-ext and SvdW-ext are hydrogen and van der Waals interaction score,

respectively for the protein-inhibitor complex. Shb-int is the intramolecular hydrogen bonding

contributions to overall GOLD fitness score while Sint is an intramolecular strain of the inhibitor.

AutoDock Vina calculations were done on Intel Core (TM) i5 CPU M 540 @ 2.53 GHz processor.

The binding site coordinates were the same used in GOLD. The inhibitors binding affinity was

calculated in terms of binding energy in kcal/mol. The selection of a compound for MD simulations

was based on GOLD fitness score, AutoDock Vina binding energy, and druglikenss predicted by

SwisADME (Daina, Michielin, & Zoete, 2017) and preADMET (Lee et al., 2004).


To further gain dynamic and mechanistic insights of enzyme and its docked complex, a 100-ns

MD simulation was carried out (Abro & Azam, 2016). Simulation protocol was accomplished

using Assisted Model Building with Energy Refinement version 14 (AMBER14) (Case et al.,

2014). Initial libraries and parameters for the inhibitor were generated through antechamber

program of the AMBER. The integration of complex in TIP3P water box (size of 12 Å) was

assisted with Leap program using ff14SB force field to explain molecular interactions of the

system (Weiner & Kollman, 1981). The energy of systems was minimized in a gradual manner

(Andleeb et al., 2016) started with relaxing the entire system hydrogens for 500 rounds of

minimization followed by water box minimization for 1000 cycles with a restraint of 200 kcal/mol.

The systems were then subjected to restraint of 5 kcal/mol-Å2 on alpha carbon atoms for 1000

steps. In the last phase of minimization, all non-heavy atoms were minimized for 300 cycles with

restraint of 100 kcal/mol- Å2. Followed by minimization, each system was heated gradually to 300

K for a time scale of 20-ps with restraint on carbon alpha atoms of 5 kcal/mol- Å2 and a time step

set to 2-fs. To maintain temperature of the systems, Langevin dynamics (Paterlini & Ferguson,

1998) was used. Running Langevin dynamics at constant pressure (NPT) is quite useful to maintain

a constant density and let the box reduce to the "proper" size for the number of solvent molecules

in system. The gamma value was set to 1.0, while NVT ensemble and SHAKE algorithm (Kräutler,

234

Van Gunsteren, & Hunenberger, 2001) were used for heating and applying constraints on hydrogen

bonds of the system. Systems equilibration was achieved for 100-ps and time step of 2-ns

(Andersen, 1980). Systems pressure was maintained using NPT ensemble along with isotropic

position scaling and restrained on carbon atoms of 5 kcal/mol- Å2. The pressure phase was

extended for another 50-ps with restraint on carbon atom, which is reduced by 1 kcal/mol- Å2

applied after every 10-ps. The systems were then allowed to equilibrate for 1 ns using the same

conditions discussed above. Finally, the systems were subjected to a production run of 100-ns

using Berendsen temperature coupling with NVT ensemble. As equilibration in density is attained,

we used Berendsen to compute potential energies from a trajectory file. We also used NVT to

tighten the shake tolerance, the particle-mesh-Ewald method (PME) tolerances and reduce the time

step to get ideal energy conservation. This was important to compute potential energies from a

trajectory file and tighten the shake tolerance and reduce the time step to get ideal energy

conservation. The production run was performed with time step value of 2-fs and non-bounded

interaction cut off 8.0 Å. Simulated trajectories were analyzed using CCPTRAJ (Roe & Cheatham,

2013) of AMBER. The snapshots taken from trajectories were visualized using UCSF Chimera

(Pettersen et al., 2004) and VMD (Humphrey, Dalke, & Schulten, 1996).

7.5.5. RDF and AFD

Hydrogen bond interactions in addition to intermolecular close contacts between enzyme active

site residues and that of inhibitor crucial for complex stabilization were scrutinized through a Perl

written script used in VMD. This was vital in highlighting enzyme active residues that played

significant role in recognizing and binding inhibitor over the period of simulation. The screened

interactions were then used in RDF (Donohue, 1954) and AFD (Ahmad, Raza, Uddin, & Azam,

2017). RDF was applied for finding the probability and to quantify ligand atoms present in the

vicinity of protein pocket. RDF assists in getting valuable insights about the conformation

variations induced as a result of intermolecular interactions between the inhibitor and protein

active residues. This was achieved by employing the PTRAJ module of AMBER14. To further

express the coordination geometry of the ligand atoms with respect to protein in a 3D histogram,

an indigenously engineered analytical tool termed as AFD was used. AFD is a novel analytical

tool designed by the Computational Biology Group at National Center for Bioinformatics, Quaid-

i-Azam University, Islamabad, Pakistan. AFD is very sensitive to local structural rearrangements

235

and can recognize local displacements, providing in-depth informations about system stability.

The output of the analysis is in the form of a 3D histogram, in which ligand atom coordinates are

plotted on XY plane while protein atom was set as a reference (Ahmad, Raza, Uddin, & Azam,

2017).


In rational drug designing, structure-based virtual screening of drug libraries, combined with MD

simulations and binding free energy computation is a widely recommended approach for

identification of potent inhibitors against the intended biological target (Dominy, 2008).

MM/GBSA (Miller et al., 2012) is a popular technique for predicting estimated binding affinity of

inhibitors at the expense of modest computational efforts. The net values of binding free energies

are extracted by considering the difference between complex binding energy and that combined

for receptor and ligand. This can be represented as below,

∆Gbinding energy = ∆Gcomplex– [∆Greceptor + ∆Gligand]…………………………………………(XXXII)

Although quite successful in improving the findings of molecular docking, yet MM/GBSA

contains several questionable and crude approximations. Among these, the most disregard

parameter is the use of implicit water model, which ignores the free energy of water molecules in

the protein active site. This could result remarkably on the approximation of binding free energy.

To circumvent the limitations of MM/GBSA, Christopher et al (Woods, Malaisree, Hannongbua,

& Mulholland, 2011; Woods, Malaisree, Hannongbua, & Mulholland, 2011;Woods et al., 2014)

introduced a more sophisticated method of “WaterSwap” based on the explicit solvent model

system. This method is based on the principle of swapping the ligand with an equal shape and

volume of explicit water molecules present in the protein active cavity. A couple of simulation

boxes are used in the method. The first box contains protein, ligand, and explicit water molecules

while the second box is composed of only explicit water molecules. Both the boxes are placed in

a heat bath to ensure the transfer of heat and entropy used for decoupling between the boxes.

WaterSwap was performed for 1000 iterations with sampling size for Monte Carlo simulation set

to 1.6 × 109. Three algorithms for energy calculation like Bennets, Free energy perturbation (FEP),

and Thermodynamic Integration (TI) were used. Agreement value of < 1 Kcal/mol among these,

is regarded reasonable and indicates the highly stable nature of the complex.

236


Current study presents the applications of CADD for exploring Ddl as a potential drug target

against many intractable complications of A. baumannii. With the recently reported Ddl crystal

structure, there is a dire need for understanding the conformational changes, which this enzyme

undergoes in the catalytic mechanism of the drug-like natural inhibitors. This could provide

meaningful insights, guiding the discovery of novel antibiotics against A. baumannii. The protein

after retrieval from PDB was subjected to a preparation phase, where only one monomer of the

structure was retained while the second monomer was removed (Fig.7.1). This was logical as both

the monomers contain the same active cavity and proceeding further with one subunit could reduce

the computational efforts. Afterward, the protein was minimized for total steps of 1500.

Superimposing the minimized protein over the original structure revealed RMSD value of 0.4 Å,

which indicates the improved quality of the protein after minimization.

7.4.1. Active Site Prediction

The mapping of protein active site for ligand binding is significant for a wide range of applications

including de novo drug designing, functional site comparison, and molecular docking studies

(Capra, Laskowski, Thornton, Singh, & Funkhouser, 2009). The protein active site was predicted

through multiple ways. First, an online server, Metapocket (Huang, 2009) was used. Two potential

sites were predicted (S-Table 7.1). The first cavity was large comprises 31 amino acids while the

second cavity was composed of 15 amino acids. It was observed that the second cavity lies within

the first predicted cavity as a sub-cavity. The cavity was further confirmed from DoGSiteScorer

(S-Fig.7.2) (Volkamer, Kuhn, Rippmann, & Rarey, 2012) and literature search. In order to select

an amino acid atom coordinates for molecular docking through GOLD and AutoDock Vina, MSA

approach was utilized (Fig.7.2) (Raza, Sanober, Rungrotmongkol, & Azam, 2017). The basic idea

behind this was to look for the residue in protein active site that is highly conserved among Ddl

enzyme orthologues in bacterial pathogens. In this regard, OE2 atom of residue Glu247 was set as

a site for docking of compounds. The residue is present in the central domain and is completely

conserved among the various orthologues of Ddl enzymes. As the residue is highly conserved and

vital for enzyme reaction mechanism, any point mutation of this residue will lead to loss of enzyme

catalytic function (Krynetski et al., 1995; Kairys, Gilson, Lather, Schiffer, & Fernandes, 2009),

237

indicating less chances of point mutation in this residue of the enzyme and hence lesser opportunity

for generating resistance (Singh, Frushicheva, & Warshel, 2012).

Fig.7.1. The 3D structure of Ddl monomer. The C-terminal, Central and N-terminal domains are

shown in orange, purple and cyan, respectively.


Molecular docking phase was initiated with a virtual screening of Asinex antibacterial library. The

library contains 4800 compounds at the time of retrieval. To proceed only with drug-like

compounds, the library was screened for inhibitors that fulfill all the parameters of Lipinski rule

of five. According to this rule, compounds will be considered drug-like if their molecular weight

is less than 500 dalton (Da), log P score < 5, and hydrogen bond donors and acceptors < 5 and 10,

respectively. This screening was accomplished using LigandScout 4.1, which revealed 1440

compounds that completely satisfy Lipinski rule of five. The compounds were minimized and

subjected to molecular docking. In molecular docking, structure-based virtual screening of drug-

like compounds was performed to identify compounds having better affinity for the receptor

enzyme active site. Inhibitors were first docked into enzyme active site using GOLD, followed by

docking of the best sorted 10 compounds using AutoDock Vina. The top ten best inhibitors based

238

on GOLD fitness score and AutoDock Vina binding energy along with druglike rules which they

voilates are shown in Table 7.1. The correlation coefficient between GOLD fitness score and

AutoDock Vina binding energy can be found in Fig.3. After comparative analysis, compound-

331(6-(4-((3- methoxyphenylsulfonamido) methyl) phenyl)-2-methylnicotinamide)) was found

the best docked inhibitor in both GOLD and AutoDock Vina. The GOLD fitness score and

AutoDock Vina binding energy attained by the compound was 84.2 and -7.2 kcal/mol,

respectively.

Fig.7.2.Multiple sequence alignment of Ddl enzymes from different bacterial enzymes. (2187, S.

aureus), (3E5N, X. oryzae), (3LWB, M. tuberculosis), (3R23, B. anthracis),(4FUO, E. faecalis),

(5D8D, A. baumannii), (3V4Z, Y. pestis), (4DGJ, B. xenovorans), (4EGO, B. ambifaria), (4EGQ,

B. pseudomallei).

In both tools, the compound was investigated to docked in the same position and interacts with

almost same residues of the active site (Fig.7.4). Importantly, the compound follows all prominent

239

rules of druglikenss i.e. Lipinski (Lipinski, 2004), Ghose (Daina, Michielin, & Zoete, 2017),

Muegge (Daina, Michielin, & Zoete, 2017), Veber (Daina, Michielin, & Zoete, 2017), and Egan

rules (Daina, Michielin, & Zoete, 2017) and increases the likelihood of surviving the well-

documented high rates of attrition in drug discovery. In GOLD, the compound was found docked

deep inside the cavity of the enzyme (Fig.7.5), with minor effect on the overall structure. It was

observed the compound shields majority of the active cleft length present between the Central and

C-domains of the protein. The methoxybenzene ring (anisole) of the compound was found

extended towards N-terminal of the enzyme. These findings support the compound credibility in

hindering the accessibility of the active site to the substrate and subsequent enzyme-substrate

complex formation and final product formation. The preferred orientation of the compound was in

such a way that formamide moiety of the 2-methylnicotinamide interacts with Trp177 residue of

the loop region from the Central domain. This results in tilting the 2,6-dimethylpyridine ring to

make possible a strong hydrogen bond of 1.6 Å between nitrogen atom of 2,6-dimethylpyridine

and hydrogen of Lys140 from the Central domain. The major contribution from the compound

towards interaction with the enzyme active site came from sulfonamide group. This moiety

interacts with residues from both the Central domain and the C-terminal domain. All the atoms of

sulphonamide group tend to strongly interact with both these domains notably the Arg232 from

C-terminal domain and Ser146 from the Central domain. The anisole ring was positioned in such

a way to make possible interactions between its oxygen atom and Glu20 and Glu71 residue of N-

terminal domain of the enzyme. In comparison to GOLD, AutoDock Vina generated pose binds

totally in opposite direction. The formamide moiety of 2-methylnicotinamide ring of the inhibitor

tends to move towards the N-terminal of the enzyme and favors to interact with this domain

residues through its amino ethanol site chain. The hydrogen atom of pyridine ring of the inhibitor

interacts with Ser146 oxygen through a strong hydrogen bond of strength 2.5 Å. The middle

benzene ring of the inhibitor in contrast to GOLD move a bit up and rotate in clockwise direction.

This movement allows the inhibitor to positioned the sulfonamide group upward. In AutoDock

Vina, this pose resulted in no interaction between sulphonamide atoms and receptor residues. The

behaviour could be assumed to continue for proper positioning of the inhibitor in the active cavity

to achieve maximum interactions between the inhibitor atoms and receptor active site residues.

240

Fig.7.3. Correlation coefficient between GOLD fitness score and AutoDock Vina binding energy

for the top ten best inhibitors.

y = -0.3157x + 18.654

R² = 0.7575

-10

-9

-8

-7

-6

-5

-4

-3

-2

-1

0

81 82 83 84 85 86 87 88 89

Au

toD

ock

Vin

a b

ind

ing e

ner

gy

(kca

l/m

ol)

GOLD fitness score

241

Fig.7.4. 2D representation of binding mode and interactions of the best characterized drug-like inhibitor in the active pocket of Ddl enzyme.

242

Fig.7.5. (A) Binding mode of compound-331 in active pocket of KdsA enzyme. (B) Closer view.

243

Table 7.1. Top ten best natural inhibitors shortlisted in the current study.

Compound GOLD fitness score AutoDock Vina binding energy

(kcal.mol-1) Drugllikenss rule(s) violation

Compound-965

88.3 -9.2

Ghose rule

3 violations:

MW>480,

MR>130,

#atoms>70

Compound-1150

85.6 -8.5

Ghose rule

3 violations:

MW>480,

MR>130,

#atoms>70

Compound-1137

85.1 -8.8

Ghose rule

1 violation:

WLOGP>5.6

Muegge filter

1 violation:

XLOGP3>5

244

Compound-1131

85.0 -8.1

Ghose rule

2 violations:

MR>130,

#atoms>70

Compound-331

84.2 -7.2 No violations

Compound-1130


Compound-355


245

Compound-1139

82.2 -7,4

Ghose rule

1 violation:

MW>480

Compound-1133


Compound-370

-81.4 -7.4 No violations

MW, molecular weight, MR, molecular refractivity, #atoms, number of atoms

246

7.4.3. SwissADME and PreADMET Analysis

The high absorption and well distribution of a drug in required time are important for its effective

metabolism and action. In addition, toxicity that often overshadows ADME behaviour is another

important consideration (Nisha et al., 2016). This is because majority of the drugs, which fails in

clinical stages due to high toxicity proves very expensive and detrimental in the drug designing

process (Wadood et al., 2017). Computational prediction of druglikness together with ADMET

properties predication at the drug design stage could possibly provide an array of opportunities for

accelerating lead compounds identification with predicted biological activities (Wadood et al.,

2017). Druglikeness is the qualitative inspection of compound physicochemical or structural

properties, to investigate the compound likelihood as an oral drug-like candidate (Vistoli, Pedretti,

& Testa, 2008). Five filters were used in order to ensure the improved quality of the inhibitors for

future lead optimization. The compound was unveiled to follow all the prominent druglike rules

including Lipinski rule of five (Lipinski, 2004), Ghose rule (Daina, Michielin, & Zoete, 2017),

Veber filter (Daina, Michielin, & Zoete, 2017), Egan filter (Daina, Michielin, & Zoete, 2017), and

Muegge filter (Daina, Michielin, & Zoete, 2017).The molecular weight of compound is 419.5

g/mol, number of heavy atoms (29), number of aromatic heavy atoms (12), number of rotatable

bonds (7), number of hydrogen bond acceptor and donor 7 and 3, respectively. The presence of

rotatable bonds pointed the inhibitor to be a good adaptive inhibitor candidate. Adaptive inhibitors

achieve serval alternative functionalities at variable regions of the targeted proteins and as such

are regarded as to overcome point mutations that are responsible for decreasing inhibitor binding

affinity (Ohtaka & Freire, 2005). The TPSA of the compound is 122.92 Å². TPSA calculates the

presence of polar amino acids at surface of compound (Ertl, Rohde, & Selzer, 2000). Lower TPSA

values for a given compound is preferred as higher values diminishes compound membrane

permeability and act as a substrate for p-glycoprotein (Nisha et al., 2016). This protein is a drug

efflux protein and aids in lowery intracellular drug concentration (Daina, Michielin, & Zoete,

2017; Nisha et al., 2016). The ADME behaviour of the compound was evaluated from its

pharmacokinetics. Better absorption of the compound illustrated higher absorption of the

compound from the intestinal tract upon oral administration (Nisha et al., 2016). The

carcinogenicity test in rats revealed the compound non-carcinogenic, while the in vitro Ames test

in TA100 and TA1535 strains of Salmonella typhimurium declared the compound non-mutagen.

The in vitro Human ether-a-go-go-related gene channel inhibition (Wadood et al., 2017) was found

247

ambiguous for the compound. The water solubility of the compound was investigated through

three models of SwissADME server. According to Estimated Solubility (ESOL) model (Delaney,

2004), the log S value for the compound is -3.38 and is placed in soluble category. Log S (Ali,

Camilleri, Brown, Hutt, & Kirton, 2012); indicates compound solubility; less the log S score,

higher will be the absorption and viseversa. The second model, which adapted from Ali et al,

predicted the log S value of -3.94 with category of soluble. In third model, which proposed by

SILICOS-IT (http://silicos-it.be.s3-website-eu-west-1.amazonaws.com/software/filter-

it/1.0.2/filter-it.html), the compound score log S value of -5.66 that placed the compound in

moderate soluble category. The inhibitor molecule was found less skin permeate as indicated by

the negative log Kp value (-7.6 cm/s). One of the concerns associated with the inhibitor is being

a substrate for the P-glycoprotein. This is particularly important as the action of P-glycoprotein

results in lowering the intracellular concentration of drug. Further, the compound was unraveled

as non-inhibitor of cytochrome P450 superfamily isozymes. Solubility of compounds greatly

facilitates drug development process particularly the handling and formulationn phase. The log

Po/w value, which is a partition coefficient between n-octanol and water is an important descriptor

for lipophilicity (Arnott & Planey, 2012). In SwissADME, the log Po/w is the arithmetic mean of

values generated by five methods (iLOGP, XLOGP3, WLOGP, MLOGP, and SILICOS-IT). The

value of log Po/w for the compound is 2.21. The bioavailability score of the compound is 0.55. The

druglikness and pharmacokinetics of this compound were compared with D-cycloserine (a broad

spectrum antibiotic used for the treatment of tuberculosis and inhibit Ddl and Alanine recemase

enzyme). It was revealed that in contrast to the compound of our study that follows all drug rules,

D-cycloserine violates two drug rules: Ghose and Muegge. D-cycloserine has low Gastro-

Intestinal absorption when compared to our scrutinized compound and placed in low category. The

bioavailability score of D-cycloserine was found the same predicted for our compound of interest.

Lipophilicity of D-cycloserine was described as improved, and have enhanced solubility in water

similar to the compound reported in the study. One of the major limitation in the use of D-

cycloserine is its neurological side effects (Lu, Peng, Hwang, & Chen, 2008), as D-cycloserine

penetrates into CNS and results in several adverse effects. The screened compound in this study,

on the other hand, could be assumed to have low CNS penetration due to its molecular weight >

400 Da and TPSA value of higher than > 90 Å², thus can be less likely be related with CNS

penetration associated effects. As the protein and compound contains ionisable groups, the

http://silicos-it.be.s3-website-eu-west-1.amazonaws.com/software/filter-it/1.0.2/filter-it.html

http://silicos-it.be.s3-website-eu-west-1.amazonaws.com/software/filter-it/1.0.2/filter-it.html

248

knowledge of their pKa was essential (Sham, Chu, & Warshel, 1997; Borsstnar, Repicc, Kamerlin,

Vianello, & Mavri, 2012; Repic, Purg, Vianello, & Mavri, 2014). Therefore, pKa value, which can

be define as the negative log of the dissociation constant, calculation of the drug and the entire

protein molecule was determined using online version of Chemicalize ChemAxon (Swain, 2012)

and H++ software (Anandakrishnan, Aguilar, & Onufriev, 2012). The pKa calculations results for

the compound can be found in S-Fig.7.3 while titration curve for the protein can be seen in S-

Fig.7.4. List of all computed pKs can be found S-Table 7.2 while residues that contribute the most

to each pK shift is tabulated S-Table 7.3. The strongest acidic and basic pKa value of the

compound was revealed 9.99 and 8.15, respectively. The strongest acidic pKa value was for the

NH of sulphonamide while the strongest pKa was for NH2 group of 2-methylnicotinamide ring.

The strongest acidic pKa value of NH of sulphonamide is due to resonance of electron pair towards

the Sulphur after dissociation of the proton while the strongest basic pKa value of NH2 is due to

high ability of accepting protons. Six microspecies of the compound was observed and there

distribution at different pH can be seen in S-Fig.7.3. Overall, the predicted compound of this study

have improved druglikenss and pharmacokinetics and can be targeted further for designing potent

derivatives having improved pharmacokinetics and ADMET properties.


Although several static structures of the Ddl enzyme have been proposed, alone or with ligand, yet

the dynamics of the enzyme in aqueous milieu is still needed to decipher under physiological

conditions. The dynamics of protein alone and in complex with the best inhibitor were investigated

for a timescale of 100-ns.This leads to in-depth understanding of the structural adjustment of the

protein adopted in the presence or absence of ligand and can potentially guide the discovery of

novel antibiotics against the enzyme. Four statistical parameters; RMSD, RMSF, β-factor and Rg

were performed to unveil the protein stability over the course of simulation period (Fig.7.6).

RMSD is the average measure of distances between backbone atoms of superimposed proteins.

The relative fluctuations in the RMSD of Cα atoms of undocked protein was seen higher when

compared to the docked, indicating the stability of enzyme-inhibitor complex. The mean RMSD

value estimated for docked complex was 1.6 Å and the maximum RMSD value was 2.4 Å at 67th

ns while the mean RMSD for undocked system was 1.9 Å. Intriguingly, two main deviations were

noted in undocked system when compared to docked. The first deviation occurred from 0 to 27-ns

249

with maximum RMSD of 2.3 Å (25th ns). The second deviation was observed from 62th ns till 100-

ns with maximum RMSD of 2.4 Å. On the contrary, it was observed that studied inhibitor sustained

its interactions and binding with Ddl enzyme with no aforesaid deviations. The convergence values

for both the systems and to reveal the structural mobility for each residue of the system, RMSF

values were monitored. It is quite evident that in bounded form the enzyme residues remained

highly stable in contrast to unbounded form. The mean RMSF values for undocked and docked

system were reported 1.08 Å and 0.90 Å, respectively. As depicted clearly in Fig.7.6B, majority

of the residues of docked protein have RMSF values less than 2 Å. This further demonstrates the

highly stable nature of Ddl enzyme. In undocked protein, the residues from Ser2, Glu155 to Gln169

and Ala172 to Gly182 of the Central domain, Glu21 to Glu44, and Gly74 of the N-terminal

domain, were found to have comparatively high flexibility than others but still have in stable range.

In docked protein, residue Lys18, 47-54, 189-190, 197-200, 239-240, were reported to have

comparatively high RMSF values than in undocked form. Binding of ligand modified the enzyme

structure in such a way that it reduces the distances between the interacting active residues and

move the other residues mentioned above bit far. The most important active pocket residues;

Glu20, Glu71, Ser146, Lys140, Trp177, Arg232 in undocked system were revealed with an

average value of 1.2 Å while in docked system their value was found 0.7 Å. The systems were

further subjected to β-factor calculation. β-factor estimates flexibility and thermal stability of

protein and its side chains with respect to time. A complete coherence was observed between β-

factor values and RMSF values for the systems. The mean β-factor value for undocked and docked

systems were determined as 38.2 Å and 26.5 Å, respectively. In last, equilibrium conformation of

the systems was investigated for better understanding of the systems. This was achieved by

calculating Rg, which estimates the compactness of protein structure. Variations in the Rg

measures determines compactness of protein during simulation run. Higher Rg value implies less

tight packing of protein amino acids whereas lower Rg value entails tight packing of protein amino

acids. The mean Rg value computed for undocked system was 22.4 Å, with maximum value of

23.06 Å at 49thns. For docked system, the maximum value of 22 Å was observed at 48 ns with

mean value of 22.3 Å. The Rg values reported herein for both the systems were analogous to the

RMSD values and further validate the stability of the protein in docked form.

250

7.4.5. RDF and AFD Analysis

A Tcl/Tk script was designed to screen protein residues involved in hydrogen bonding and

intermolecular contacts at the start (first ten nanoseconds) and towards the end of simulation (last

ten nanosecond) was utilized in the VMD software. The purpose of this was to shed light on vital

residues of the enzyme active site, which played a significant role in recognizing, binding, and

stabilizing the inhibitor over the entire course of simulation. The shortlisted residues were then

subjected to RDF (Donohue, 1954) and AFD (Abro & Azam, 2016) to complement their role in

inhibitor binding. Analysis comprised of a correlation between enzyme active residues and ligand

atoms on a XY plane in the form of 3D histogram. Only four interactions involving Lys176 and

Trp177 residues of the pocket and inhibitor hydrogen and nitrogen atoms were selected for RDF

and AFD, due their profound role in complex stability and structural integrity. Both the protein

residues were revealed to efficiently trap the inhibitor by virtue of hydrogen bonds. RDF graphs

were generated for the four aforesaid interactions and are illustrated in Fig.7.7.The first plot in the

figure represents the interaction distance between Lys176-O and inhibitor-H13. At the start 10-ns

of simulation, the largest peak appears at 1.71 Å having a g (r) value of 1.89 while towards the end

10-ns, the largest peak appears at 2.00 Å having a g (r) value of 1.30. The graph demonstrates that

both at the start 10-ns and end 10-ns of simulation period, the oxygen atom of Lys176 and H13 of

inhibitor spend majority of the time in close proximity of each other to exhibit the activity. For

Lys176-O and inhibitor H14 atom, the largest peak at the start 10-ns of simulation was observed

at 3.39 Å with a g (r) value of 0.4. Towards the end 10-ns of simulation, the peak shortened and

broadened abit with the highest peak noted at 3.60 Å having a g (r) value of 0.36. The graph

explains that towards the end 10-ns of simulation the distribution of inhibitor H14 atom around

Lys176 oxygen is more dispersed and moved away about 0.2 Å from each other. The RDF for

Lys176-O and inhibitor-N revealed that at the starting 10-ns time of simulation the largest peak is

present at 2.84 Å having a g (r) value of 0.92, while towards the end 10-ns the largest peak was

observed at 2.85 Å having a g (r) value of 0.56. Although towards the end 10-ns of simulation, the

distribution of inhibitor atom is dispersed more than at start 10-ns of simulation, however, it was

noted that atoms tend to come close to each other. The graphs generated for Trp177 nitrogen atom

and inhibitor H13 atoms were found bit distorted at the start 10-ns and towards the end 10-ns of

simulation. The maximum possibility of finding inhibitor H13 atom around Trp177 nitrogen was

found at 3.34 Å with g (r) value of 0.29 while towards the end, the H13 atom of inhibitor was

251

found at maximum of 3.64 Å with g (r) value of 0.28. Towards the end, the increase distance of

0.1 Å was observed. All the graphs infer that these four interactions held the ligand in the close

proximity of the protein active cavity, and are contributing factors in formation of stable complex.

The minor variations depict little adjustment of inhibitor in the protein active pocket for achieving

the best orientation. Further, the RDF demonstrate the system composition in a correct way as

indicated by the lower flexibility of the interaction and showed good agreement between the

findings of docking and MD simulations. In simulation trajectories particles are positioned with

respect to time, which are subjected to several different statistical parameters for explaining the

molecular nature of the system under study. Among these, RDF gives insights about the local

displacement of an atom with respect to a reference. RDF is less responsive to local structural

reorganizations that are mainly because of the moving inhibitor in protein active cavity to achieve

an energetically stable confirmation as the simulation advances. To get more understanding of

local movements and structural changes, AFD was utilized. Generally, the pattern of AFD graph

signifies that the inhibitor is retained in enzyme active site for the whole time of simulation. The

contour plot for AFD analysis is illustrated in Fig.7.8. The AFD plots were observed with

maximum values having little variation and small distribution area from top to down graph

perspective that indicate highly stable nature of inhibitor in enzyme active pocket. In the AFD plot

generated for oxygen atom of Lys176 and ligand H13 atom, it was observed that the inhibitor

moved closer to protein active site both on X-axis and Y-axis towards the end 10-ns of simulation.

Although, the maximum value is decreased towards the end 10-ns of simulation yet the density

distribution of inhibitor atom is more refine. Area of distribution was revealed to decreased by 0.5

Å. A slight shift in inhibitor displacement was also observed that may be due to increased affinity

for the protein. In case of oxygen of Lys176 and inhibitor H14 atom, the distribution area towards

the end of simulation is slightly increased. The inhibitor moved both on X-axis and Y-axis towards

the center of Lys176 oxygen atoms, which signifies the stability of this interaction towards the end

10-ns of simulation. The tilting behavior of the inhibitor was seen in all the distribution. For oxygen

atom Lys176 and inhibitor N3, the inhibitor atom was revealed to come close in vicinity of Lys176

oxygen atom mediated by inhibitor rotation. Towards the end 10-ns of simulation the distribution

area was more refined when compared to the start 10-ns of simulation, however, its magnitude

decreases. In case of Trp177-N and inhibitor-H13, the distribution area had increased a little

towards simulation end that indicate the inhibitor atom movements towards on X-axis plane.

252

Fig.7.6. RMSD (A), RMSF (B), β-factor (C), and Rg (D) for enzyme and enzyme-inhibitor complex.

253

The magnitude of graph was also found decreased at the end of simulation. By comparison of RDF

graphs there is no significant shift in the distance between the interacting partner but the AFD

graphs represent a tilting behaviour of inhibitor towards the protein active site. This behavior of

inhibitor can be seen with visual representation of snapshots of simulation at 1-ns and at 100-ns

(Fig.7.9). From the figure side groups of the inhibitor are shifting more towards the protein. AFD

highlighted these movement which RDF cannot represent using only distance for analysis. Overall

these graphs interpret that through the entire simulation period, the inhibitor is present in the close

vicinity of active pocket and remained bonded through multiple strong hydrogen bond interactions

with protein active site residues.

Fig.7.7. RDF graph for: A. Lys-176-O-Lig-H13, B. Lys176-O-Lig-H14, C. Lys176-O-Lig-N3,

D. Trp177-N-Lig-H13.

254

Fig.7.8. AFD:A. Lys-176-O-Lig-H13, B. Lys176-O-Lig-H14, C. Lys176-O-Lig-N3, D. Trp177-N-Lig-H13

255

Fig.7.9. Inhibitor movement from 0-ns to 100-ns.

7.4.6. MM/GBSA Based Energy Calculations

The accurate prediction of a drug affinity for its intended biological protein is considered as the

“holy grails” in CADD (Aldeghi, Heifetz, Bodkin, Knapp, & Biggin, 2016). In addition to the

development of pharmaceutically active molecules with improved bioavailability and toxicity, the

high affinity of a drug for its intended target holds promising outcomes in designing highly potent,

selective, and eventually efficacious compounds. Though the presence of structural information

aided the rationalization of target-ligand but still conformational changes of protein and ligand,

the solvent effect, and the entropy-enthalpy compensation make this process a complex task

(Bissantz, Kuhn, & Stahl, 2010). Advances in computational power and theory especially in the

last decade or so, the physics-driven computer simulation holds a great potential in predicting the

accurate binding affinities of compound for their targeted biological macromolecules

(Wereszczynski & McCammon, 2012). These methods naturally encounter the complicating

effects generated because of discrete nature of entropy and solvent (Aldeghi, Heifetz, Bodkin,

Knapp, & Biggin, 2016). The estimated binding free energy calculations and intermolecular

interactions between the compound-331 and the enzyme were elucidated using MM/GBSA (Miller

et al., 2012). Other elegant method for prediction binding free energies is Linear Interaction

256

Energy (LIE) (Wang, Wang, & Kollman, 1999; Aqvist & Marelius, 2001; Gutierrez-de-Teran &

Aqvist, 2012). The comparison of MM/GBSA with LIE and other binding free energy calculation

methods have been done in several studies and was revealed to be highly depended on the tested

system (Kuhn & Kollman, 2000; Barril, Gelpi, Lopez, Orozco, & Luque, 2001; Wong, Amaro, &

McCammon, 2009; Genheden, 2011; Mikulskis et al., 2012; Muddana et al., 2012; Genheden &

Ryde, 2015). Therefore, it is difficult to predict which of the method is more accurate and elegant

(Genheden, 2015). Generally, the accuracy of MM/GBSA is comparable with LIE (Genheden,

2011). From expert point of view, MM/GBSA is more preferred in many of the present studies

thanks to its fast nature and often give better accuracy (Genheden, 2015). Compared to

MM/GBSA, LIE is highly system dependent and cannot readily be employed in studying

molecules without introducing explicit salt (Genheden, 2015). Altogether, MM/GBSA was

preferred in this present study. The calculations were performed on entire trajectories of MD

simulations of enzyme-inhibitor complex. The estimated binding free energies obtained for the

complex can be visualized in Table 7.2.

Table 7.2. Estimated MM/GBSA based binding free energy values for enzyme-inhibitor complex.

Contribution

Estimated binding free

energy values

(kcal.mol-1)

∆Evdw -50.30

∆Eele -38.89

∆Egas -89.2

∆Gsolv,GB 41.83

∆GGB 50.8

∆Gsol-np -8.96

Htot,GB -47.36

The term entropy was excluded due to convergence problems in some cases where it failed to be

calculated (Genheden, 2015; Abro & Azam, 2016). The average of estimated binding free energies

calculated for the complex was determined as -47.36 kcal/mol. Usually the calculated values of

binding free energy are greater than the actual values and as such are referred as estimated or

257

calculated values rather than absolute binding free energy in the manuscript (Hou, Wang, Li, &

Wang, 2010; Genheden, 2015). Major contributions towards complex stability came from van der

Waals energy that dominates the overall calculated binding free energy. Van der Waals energy

determined for the system was -50.30 kcal/mol. In contrast, non-favorable contributions arise from

polar component of solvation free energy with value of 50.8 kcal/mol. The non-polar solvation

free energy deduced from the system was -8.96 kcal/mol. In polar solvent, all molecular

interactions are driven by polar and non-polar components, where non-favorable interactions and

its associated energies are more promising. Electrostatic energies were also found dominant and

aid significantly to the overall calculated binding energy of the system. The electrostatic energy

calculated for the system was -38.89 kcal/mol. ∆Egas is gas-phase interaction energy and

important because of its contribution from internal energy as well as from electrostatic and van der

Waals interaction energies and can be represented as Egas = Eint +EELE +EVDW. The ∆Egas energy

for the complex was -89.2 kcal.mol-1. Further understanding of the complex interactions at the

atomic level was achieved by decomposing total energy of the system into each residue of the

enzyme and inhibitor (Fig.7.10). Residues having estimated binding affinity of < -1 kcal/mol were

referred as hotspot amino acids due to their major contributions in stabilizing the complex [81].

The following residues were found to have calculated binding energy of < -1 kcal/mol; Lys100 (-

2 kcal/mol), Ile138 (-1 kcal/mol), Gly145 (-1 kcal/mol), Glu175 (-1.6 kcal/mol), Trp177 (-2

kcal/mol), Ile178 (-1 kcal/mol),Leu246 (-2 kcal/mol), and Asn249 (-1.3 kcal/mol). All these

residues are important constituents of active pocket and have considerably low estimated binding

energy. These residues are revealed to be present within or around the enzyme active pocket and

contribute substantially to hydrogen and hydrophobic interactions. It was also noticed that loss of

these residues could result in loss of catalytic mechanism for substrate. Moreover, the calculated

binding free energy of the active site residues that contribute significantly was decomposed into

constituent like electrostatic, van der Waals, and polar solvation, and non-polar solvation

contribution (S-Table 7.4).The total calculated binding energy of complex, protein, inhibitor in

500 frames of simulation is shown in S-Fig.7.5.

7.4.7. WaterSwap Based Energy Calculations

WaterSwap is a method of finding absolute binding free energy by overcoming the problems

associated with cavitation and large value differences of double decoupling (Woods, Malaisree,

258

Hannongbua, & Mulholland, 2011). The use of explicit water model in WaterSwap means to

include molecular details of protein-water-ligand, protein-water, and ligand-water interactions that

are omitted in continum solvent methods like MM/GBSA (Woods, Malaisree, Hannongbua, &

Mulholland, 2011; Woods et al., 2014). This is of particular importance in cases where water

molecules act as bridging interactions between the protein and the ligand. WaterSwap has been

successfully applied for mutants prediction of influenza neuraminidase enzyme that showed

reduced drug affinity (Woods, Malaisree, Long, McIntosh-Smith, & Mulholland, 2013). The

binding free energy calculated through WaterSwap is presented in Table.7.3. It is evident from

the findings of three algorithms that the agreement values is < 1 kcal.mol-1, indicating the high

affinity of the inhibitor towards enzyme active pocket. For further understanding of the binding

energy at protein residues level, the total energy was decomposed into residue level to shed light

on the hotspot amino acids involved in inhibitor binding.It was revealed that active site

residues:Lys176 and Trp177 have considerably low binding energy and played vital role in

inhibitor recognition and binding throughout the period of simulation. The mean binding energy

of Lys176 and Trp177 is -1.94 kcal.mol-1and -2.73 kcal.mol-1, respectively.

Table 7.3. WaterSwap based absolute binding free energy calculation for enzyme-inhibitor

complex.


Kcal.mol-1

Bennetts -19.01

FEP -18.02

TI -18.99

259

Fig.7.10. MM/GBSA based binding free energy decomposition into each residue of the enzyme. Amino acids are reppresented by a single lettere code.

-28.0

-23.0

-18.0

-13.0

-8.0

-3.0

S2T5 G8 A11

L14K17

E20V23

D26

Q29

L32

L35

S38

Q41

A44

P47

R50

T53

V56

D59

F62

L65

R68

E71

Q74

G77

E80

N83

Y86

T89

Q92

A95

M98

V101

K104

W107

S110

P113

P116

I119

K122D125

S128A131

G134V137K140H143S146M149V152A155

F158A161

K164Q167

A170

M173

K176

T179

E182

I185

L188

Q191

P194

R197

Y200

P203

L206

T209

K212

Q215

C218

A221

A224

A227

W230

I233

M236

E239

N242

L245

V248

V251

M254

H257

V260

A263A266

Y269D272

C275 I278 Q281Lig284

`260

7.7. Conclusions

Among several targetable bacterial species, Ddl enzyme is one of the systematically searched and

attracted candidate holding great promise for controlling the dissemination of A. baumannii

resistant clones. The potential of natural antibacterial against DDI was revealed to be promising

therapy and could serve as an alternative to D-cycloserine because of improved druglikeness and

pharmacokinetics. Molecular docking unravels the deep binding of the inhibitors in the enzyme

active cavity, validated by MD simulations as studied by the formation of highly stable complex

between the enzyme and the best screened natural inhibitor over the course 100-ns of simulation.

The active site core of the enzyme was found highly stable and showed great affinity towards the

inhibitor by making strong hydrogen and hydrophobic interactions. RDF analysis demonstrated

Lys176 and Trp177 as the main residues involved in stable interactions with the inhibitor while

novel analytical AFD unveiled the retention of hydrogen between Lys176 and Trp177 and inhibitor

throughout the simulation time. Estimated binding free energy based on MM/GBSA depicted

further, the exceptionally high stable value for the complex of -47.36 kcal.mol-1 dominated mainly

by van der Waals interactions. The use of more sophisticated and recently introduced approach of

WaterSwap discovered a more reliable agreement on the affinity of inhibitor towards enzyme

active residues. The agreement value among Bennetts, FEP, TI algorithm is < 1 kcal.mol-1,

signifying the highly stable nature of complex. Understanding the functionality of antibacterial

enzymes in aspect of inhibitor binding and catalysis in addition to the knowledge of binding

thermodynamics is vital in structure based medicinal chemistry research. Therefore, the outcomes

of this study will provide significant directions for designing novel potent inhibitors with improved

druglikeness, pharmacokinetics, and antibacterial activity that may serve as a treatment for A.

baumannii related infections.


S-Fig.7.1. The reaction mechanism of Ddl enzyme.

S-Fig.7.2. DoGSiteScorer predicted binding cavity for Ddl enzyme.

S-Fig.7.3. PKa calculations for the compound-331.

S-Fig.7.4. Titration curve for the entire DDl molecule.

S-Fig.7.5. MM/GBSA free energy versus number of frames.

`261

S-Table.7.1. The potential ligand binding sites in Ddl enzyme predicted by Metapocket.

S-Table.7.2. List of all computed pKs for the DDl enzyme.

S-Table.7.3. Significant contributions to pK of each site from neighbouring groups.

S-Table.4. MM/GBSA free energy decomposition into most active residue of the enzyme.

7.9. References





Journal of Molecular Graphics and Modelling, 77, 72-85.

Aldeghi, M., Heifetz, A., Bodkin, M. J., Knapp, S., & Biggin, P. C. (2016). Accurate calculation

of the absolute free energy of binding for drug molecules. Chemical Science, 7(1), 207–218.

Ali, J., Camilleri, P., Brown, M. B., Hutt, A. J., & Kirton, S. B. (2012). Revisiting the general

solubility equation: in silico prediction of aqueous solubility incorporating the effect of

topographical polar surface area. Journal of Chemical Information and Modeling, 52(2), 420–

428.

Anandakrishnan, R., Aguilar, B., & Onufriev, A. V. (2012). H++ 3.0: automating p K prediction

and the preparation of biomolecular structures for atomistic molecular modeling and

simulations. Nucleic Acids Research, 40(W1), W537-W541.






79651–79661.

Aqvist, J., & Marelius, J. (2001). The linear interaction energy method for predicting ligand

binding free energies. Combinatorial Chemistry & High Throughput Screening, 4(8), 613–

626.

`262

Arnott, J. A., & Planey, S. L. (2012). The influence of lipophilicity in drug discovery and design.

Expert Opinion on Drug Discovery, 7(10), 863–875.

Barreteau, H., Kovac, A., Boniface, A., Sova, M., Gobec, S., & Blanot, D. (2008). Cytoplasmic

steps of peptidoglycan biosynthesis. FEMS Microbiology Reviews, 32(2), 168–207.

Barril, X., Gelpi, J. L., Lopez, J. M., Orozco, M., & Luque, F. J. (2001). How accurate can

molecular dynamics/linear response and Poisson--Boltzmann/solvent accessible surface

calculations be for predicting relative binding affinities? Acetylcholinesterase huprine

inhibitors as a test case. Theoretical Chemistry Accounts, 106(1–2), 2–9.

Barton, D. (1999). Comprehensive natural products chemistry. Newnes.

Besong, G. E., Bostock, J. M., Stubbings, W., Chopra, I., Roper, D. I., Lloyd, A. J., … Johnson,

A. P. (2005). A De Novo Designed Inhibitor of D-Ala-D-Ala Ligase from E. coli.

Angewandte Chemie International Edition, 44(39), 6403–6406.

Bissantz, C., Kuhn, B., & Stahl, M. (2010). A medicinal chemist’s guide to molecular interactions.

Journal of Medicinal Chemistry, 53(14), 5061–5084.

Borstnar, R., Repic, M., Kamerlin, S. C. L., Vianello, R., & Mavri, J. (2012). Computational Study

of the p K a Values of Potential Catalytic Residues in the Active Site of Monoamine Oxidase

B. Journal of Chemical Theory and Computation, 8(10), 3864–3870.

Bugg, T. D. H., Braddick, D., Dowson, C. G., & Roper, D. I. (2011). Bacterial cell wall assembly:

still an attractive antibacterial target. Trends in Biotechnology, 29(4), 167–173.

Capra, J. A., Laskowski, R. A., Thornton, J. M., Singh, M., & Funkhouser, T. A. (2009). Predicting

protein ligand binding sites by combining evolutionary sequence conservation and 3D

structure. PLoS Computational Biology, 5(12), e1000585.


14.




`263

Delaney, J. S. (2004). ESOL: estimating aqueous solubility directly from molecular structure.

Journal of Chemical Information and Computer Sciences, 44(3), 1000–1005.

Dominy, B. N. (2008). Molecular recognition and binding free energy calculations in drug

development. Current Pharmaceutical Biotechnology, 9(2), 87–95.


Proceedings of the National Academy of Sciences, 40(6), 377–381.

Ertl, P., Rohde, B., & Selzer, P. (2000). Fast calculation of molecular polar surface area as a sum

of fragment-based contributions and its application to the prediction of drug transport

properties. Journal of Medicinal Chemistry, 43(20), 3714–3717.

CDC. (2013). Antibiotic resistance threats in the United States, 2013. Centres for Disease Control

and Prevention, US Department of Health and Human Services.

Genheden, S. (2011). MM/GBSA and LIE estimates of host-guest affinities: dependence on

charges and solvation model. Journal of Computer-Aided Molecular Design, 25(11), 1085–

1093.



Gholizadeh, Y., Prevost, M., Van Bambeke, F., Casadewall, B., Tulkens, P. M., & Courvalin, P.

(2001). Sequencing of the ddl gene and modeling of the mutated D-alanine: D-alanine ligase

in glycopeptide-dependent strains of Enterococcus faecium. Protein Science, 10(4), 836–844.



Pathogens, 2016.

Gutierrez-de-Teran, H., & Aqvist, J. (2012). Linear interaction energy: method and applications

in drug design. In Computational drug discovery and design (pp. 305–323). Springer.




`264


143–152.

Hou, T., Wang, J., Li, Y., & Wang, W. (2010). Assessing the performance of the MM/PBSA and

MM/GBSA methods. 1. The accuracy of binding free energy calculations based on molecular

dynamics simulations. Journal of Chemical Information and Modeling, 51(1), 69–82.





Huynh, K.H., Hong, M., Lee, C., Tran, H.T., Lee, S. H., Ahn, Y.J., … Kang, L.W. (2015). The

crystal structure of the D-alanine-D-alanine ligase from Acinetobacter baumannii suggests a

flexible conformational change in the central domain before nucleotide binding. Journal of

Microbiology, 53(11), 776–782.

Jones, G., Willett, P., & Glen, R. C. (1995). Molecular recognition of receptor sites using a genetic

algorithm with a description of desolvation. Journal of Molecular Biology, 245(1), 43–53.

Kairys, V., Gilson, M. K., Lather, V., Schiffer, C. A., & Fernandes, M. X. (2009). Toward the

design of mutation-resistant enzyme inhibitors: further evaluation of the substrate envelope

hypothesis. Chemical Biology & Drug Design, 74(3), 234–245.

Kovacc, A., Konc, J., Vehar, B., Bostock, J. M., Chopra, I., Janezzic, D., & Gobec, S. (2008).

Discovery of new inhibitors of D-alanine: D-alanine ligase by structure-based virtual

screening. Journal of Medicinal Chemistry, 51(23), 7442–7448.

Kovac, A., Majce, V., Lenarsic, R., Bombek, S., Bostock, J. M., Chopra, I., … Gobec, S. (2007).

Diazenedicarboxamides as inhibitors of D-alanine-D-alanine ligase (Ddl). Bioorganic &

Medicinal Chemistry Letters, 17(7), 2047–2054.




`265

Krynetski, E. Y., Schuetz, J. D., Galpin, A. J., Pui, C.H., Relling, M. V, & Evans, W. E. (1995). A

single point mutation leading to loss of catalytic activity in human thiopurine S-

methyltransferase. Proceedings of the National Academy of Sciences, 92(4), 949–953.

Kuhn, B., & Kollman, P. A. (2000). Binding of a diverse set of ligands to avidin and streptavidin:

an accurate quantitative prediction of their relative affinities by a combination of molecular

mechanics and continuum solvent models. Journal of Medicinal Chemistry, 43(20), 3786–

3791.





Lu, P.L., Peng, C.F., Hwang, J.J., & Chen, Y.H. (2008). Activity of twelve second-line

antimicrobial agents against Mycobacterium tuberculosis in Taiwan. Journal of




130–155.

Mikulskis, P., Genheden, S., Rydberg, P., Sandberg, L., Olsen, L., & Ryde, U. (2012). Binding

affinities in the SAMPL3 trypsin and host-guest blind tests estimated with the MM/PBSA

and LIE methods. Journal of Computer-Aided Molecular Design, 26(5), 527–541.




Muddana, H. S., Varnado, C. D., Bielawski, C. W., Urbach, A. R., Isaacs, L., Geballe, M. T., &

Gilson, M. K. (2012). Blind prediction of host-guest binding affinities: a new SAMPL3

challenge. Journal of Computer-Aided Molecular Design, 26(5), 475–487.

Neuhaus, F. C., & Lynch, J. L. (1962). Studies on the inhibition of D-alanyl-D-alanine synthetase

by the antibiotic D-cycloserine. Biochemical and Biophysical Research Communications,

`266

8(5), 377–382.

Nisha, C. M., Kumar, A., Nair, P., Gupta, N., Silakari, C., Tripathi, T., & Kumar, A. (2016).

Molecular Docking and In Silico ADMET Study Reveals Acylguanidine 7a as a Potential

Inhibitor of β-Secretase. Advances in Bioinformatics, 2016.

Ohtaka, H., & Freire, E. (2005). Adaptive inhibitors of the HIV-1 protease. Progress in Biophysics

and Molecular Biology, 88(2), 193–208.











Repic, M., Purg, M., Vianello, R., & Mavri, J. (2014). Examining Electrostatic Preorganization in

Monoamine Oxidases A and B by Structural Comparison and pK a Calculations. The Journal

of Physical Chemistry B, 118(16), 4326–4332.



3084–3095.

Sham, Y. Y., Chu, Z. T., & Warshel, A. (1997). Consistent calculations of pKa’s of ionizable

residues in proteins: semi-microscopic and microscopic approaches. The Journal of Physical

Chemistry B, 101(22), 4458–4472.

Singh, N., Frushicheva, M. P., & Warshel, A. (2012). Validating the vitality strategy for fighting

drug resistance. Proteins: Structure, Function, and Bioinformatics, 80(4), 1110–1122.

`267

Skedelj, V., Arsovska, E., Tomasic, T., Kroflic, A., Hodnik, V., Hrast, M., … others. (2012). 6-

Arylpyrido [2, 3-d] pyrimidines as novel ATP-competitive inhibitors of bacterial D-alanine:

D-alanine ligase. PLoS One, 7(8), e39922.

Sova, M., Cadez, G., Turk, S., Majce, V., Polanc, S., Batson, S., … Gobec, S. (2009). Design and

synthesis of new hydroxyethylamines as inhibitors of D-alanyl-D-lactate ligase (VanA) and

D-alanyl-D-alanine ligase (DdlB). Bioorganic & Medicinal Chemistry Letters, 19(5), 1376–

1379.

Swain, M. (2012). Chemicalize. org. ACS Publications.

Taitt, C. R., Leski, T., Stockelman, M. G., Craft, D. W., Zurawski, D. V, Kirkup, B. C., & Vora,

G. J. (2013). Antimicrobial resistance determinants in Acinetobacter baumannii isolates taken

from military treatment facilities. Antimicrobial Agents and Chemotherapy, AAC-01897.

Triola, G., Wetzel, S., Ellinger, B., Koch, M. A., Hübel, K., Rauh, D., & Waldmann, H. (2009).

ATP competitive inhibitors of D-alanine-D-alanine ligase based on protein kinase inhibitor

scaffolds. Bioorganic & Medicinal Chemistry, 17(3), 1079–1087.




Tytgat, I., Colacino, E., Tulkens, P. M., Poupaert, J. H., Prevost, M., & Van Bambeke, F. (2009).

D-ligases as a potential target for antibiotics: past, present and future. Current Medicinal

Chemistry, 16(20), 2566–2580.

Vehar, B., Hrast, M., Kovac, A., Konc, J., Mariner, K., Chopra, I., … Gobec, S. (2011). Ellipticines

and 9-acridinylamines as inhibitors of D-alanine: D-alanine ligase. Bioorganic & Medicinal

Chemistry, 19(17), 5137–5146.

Vistoli, G., Pedretti, A., & Testa, B. (2008). Assessing drug-likeness--what are we missing? Drug


Volkamer, A., Kuhn, D., Rippmann, F., & Rarey, M. (2012). DoGSiteScorer: a web server for

automatic binding site prediction, analysis and druggability assessment. Bioinformatics,

28(15), 2074–2075.

`268

Vukic, V., Hrnjez, D., Milanović, S., Ilicic, M., Kanuric, K., & Petri, E. (2015). Comparative

Molecular Modeling and Docking Analysis of β-galactosidase Enzymes from Commercially

Important Starter Cultures Used in the Dairy Industry. Food Biotechnology, 29(3), 248–262.



phosphate reductoisomerase for treatment of Falciparum malaria. Pharmaceutical Biology,

55(1), 19–32.

Wang, W., Wang, J., & Kollman, P. A. (1999). What determines the van der waals coefficient β

in the LIE (linear interaction energy) method to estimate binding free energies using

molecular dynamics simulations? Proteins: Structure, Function, and Bioinformatics, 34(3),

395–402.




Wereszczynski, J., & McCammon, J. A. (2012). Statistical mechanics and molecular dynamics in

evaluating thermodynamic properties of biomolecular recognition. Quarterly Reviews of

Biophysics, 45(1), 1–25.



Modeling, 45(1), 160–169.

Wong, S., Amaro, R. E., & McCammon, J. A. (2009). MM-PBSA captures key role of intercalating

water molecules at a protein- protein interface. Journal of Chemical Theory and Computation,

5(2), 422–429.



of Chemical Physics, 134(5), 02B611.


Computational assay of H7N9 influenza neuraminidase reveals R292K mutation reduces drug

`269

binding affinity. Scientific Reports, 3, 3561.

Woods, C. J., Malaisree, M., Michel, J., Long, B., McIntosh-Smith, S., & Mulholland, A. J. (2014).

Rapid decomposition and visualisation of protein-ligand binding free energies by residue and

by water. Faraday Discussions, 169, 477–499.

Wu, D., Kong, Y., Han, C., Chen, J., Hu, L., Jiang, H., & Shen, X. (2008). D-Alanine: D-alanine

ligase as a new target for the flavonoids quercetin and apigenin. International Journal of

Antimicrobial Agents, 32(5), 421–426.

Zawadzke, L. E., Bugg, T. D. H., & Walsh, C. T. (1991). Existence of two D-alanine: D-alanine

ligases in Escherichia coli: cloning and sequencing of the ddlA gene and purification and

characterization of the DdlA and DdlB enzymes. Biochemistry, 30(6), 1673–1682.

`270

Chapter # 8 Blocking the Catalytic Mechanism of MurC Ligase Enzyme

from Acinetobacter baumannii: An in Silico Guided Study

towards the Discovery of Natural Antibiotics

8.1. Abstract

A. baumannii belongs to the most critical group of bacterial pathogens that are resistant to large

number of antibiotics, including the highly effective carbapenems and third generation

cephalosporins. Novel antibiotics targeting unique pathways of the said pathogen could reduce the

mortality rate due to its infections that are impervious to the current antibiotics. Herein, we

explored a library of natural compounds that was filtered first for drug-like, followed by lead-like

molecules, which were finally utilized in structure based virtual screening to identify the most

promising inhibitor for the Ligand binding (LB) domain of MurC ligase enzyme. The inhibitor (1'-

((2H-imidazol-2-yl)methyl)-N-(pyridin-2-yl)-1',2'-dihydro-[4,4'-bipyridin]-2-amine) was

observed with a conformation to target most conserved and catalytically critical residues of both

the intended LB as well as the ATP binding domain of MurC. This multi-domain inhibitor revealed

to have an excellent pharmacokinetics profile thus likely to have safe and effective therapeutic

applications for the future. MD simulations in aqueous solution further supported the high affinity

of the compound for the target site involving strong hydrogen bonding. At residue level, RDF and

AFD illustrated Asp334 as the most critical amino acid that drives recognition, binding, and

activity of the compound. The complex stability was validated by subjecting it to MM/PBSA,

MM/GBSA and WaterSwap based binding free energy calculations. The system was observed

with high stability: total binding energy in MM/GBSA (-48.45 kcal/mol) and MM/PBSA (-3.62

kcal/mol). The columbic interactions were noticed to dominate (-336.90 kcal/mol), followed by

van der Waals energies (-45.52 kcal/mol) in MurC-inhibitor binding. The absolute binding free

energy estimated by WaterSwap was -43.2 kcal/mol, depicting higher complex stability. The

screened scaffold might be used in functional groups substitution to achieve further lead

optimization.

`271

8.2. Introduction

The emergence and spread of multi-drug and extensive-drug resistant bacterial strains have

become the most critical global health problem that resulted into complicating the treatment,

increasing health care cost, human mortality and morbidity (Livermore, 2009; Fauci & Morens,

2012). The development of new antibiotics and their approval by pharmaceutical industries are

largely abandoned in the last forty years and since 1987; no potent antibiotics have been discovered

(Silver, 2011). On the shelf, we now have an inadequate arsenal of potent antibiotics and

development of novel antibiotics with unique mechanism of action must be given top priority

(Kouidmi, Levesque, & Paradis-Bleau, 2014; WHO, 2017).

The peptidoglycan (PG) biosynthesis enzymes are indispensable for growth and function among

the most pathogenic bacteria (Schneider & Sahl, 2010). Being unique to bacteria, these enzymes

are excellent targets as antibacterials (Schneider & Sahl, 2010; WHO, 2017). The two well-

familiar classes of antibiotics can exemplify this: β-lactams (KONG, Schneper, & Mathee, 2010)

and glycopeptides (Binda, Marinelli, & Marcone, 2014) that impede PG cross-linking in the last

steps of PG biosynthesis. Ramoplanin, a glycolipodepsi peptide, inhibits transglycosylase reaction

and MurG unveils broad-spectrum anti Gram-positive activity (Cheng, Huang, Ramu, Butler, &

Cooper, 2014). In principle, the initial cytoplasmic steps of PG biosynthesis pathway have not

been fully explored, despite being essential and unique (Kouidmi, Levesque, & Paradis-Bleau,

2014). The only antibacterial agent that inhibits any of the cytoplasmic steps is Fosfomycin.

Fosfomycin is a fully developed antibiotic derived from a natural product and functions by

blocking the functionality of the first enzyme, UDP-N-acetylglucosamine enolpyruvyltransferase

(MurA), of the pathway (Gadebusch, Stapley, & Zimmerman, 1992). The amide ligases: UDP-

Nacetylmuramate: L-alanine ligase (MurC), UDP-N-acetylmuramoylalanine-D-glutamate ligase

(MurD), UDP-N-acetylmuramoyl-L-alanyl-D-glutamate-2,6-diaminopimelate ligase (MurE) and

UDP-N-acetylmuramoyl-tripeptide-D-alanyl-D-alanine ligase (MurF) are attractive antibacterial

targets as they have broad-spectrum conservation among medically important bacteria and have

no homologous counterpart in the host (N Sangshetti, S Joshi, H Patil, G Moloney, & B Shinde,

2017). These enzymes ligate the incoming peptide moieties through a non-ribosomal peptide bond

to the growing PG chain (Kouidmi, Levesque, & Paradis-Bleau, 2014). MurC, in particular, is

involved in the first ATP dependent ligation of L-alanine (Ala) to UDPN-acetylmuramic acid

(UNAM) to produce UDP-N-acetylmuramoyl-L-alanine (UMA) (Humnabadkar et al., 2014).

`272

Subsequent addition of peptides to the C-terminus of UMA is accomplished in a sequential

manner: MurD adds D-glutamate (Perdih et al., 2014), MurE adds diamino acids of L-lysine or

meso-diaminopimelate (Tomasic et al., 2012), and MurF that adds a dipeptide of D-Ala-D-Ala

(Cha, An, Jeong, Yu, & Chung, 2014).

MurC is a three-domain protein: Uridine diphosphate (UDP) binding domain (the N-terminal),

Adenosine triphosphate (ATP) binding domain (the Central) and Ligand binding (LB) domain (the

C-terminal) (Mol et al., 2003). The UDP domain binds the UNAM substrate, while ATP binding

domain and LB domain bind with the ATP and incoming amino acid ligand, respectively (Fiuza

et al., 2008). The residues of LB domain play an important role in orientation and positioning of

the incoming peptide with the growing chain of PG (Mol et al., 2003; Fiuza et al., 2008). This

domain is highly conserved among the enzyme of Mur family as: MurF has equivalent 107 Cα

atoms with RMSD of 2.7 Å (Yan et al., 2000; Mol et al., 2003), while MurE and MurD share 106

(RMSD, 2.2 Å) (Gordon et al., 2000; Mol et al., 2003) and 97 (RMSD, 2.6 Å) (Bertrand et al.,

1997; Mol et al., 2003) equivalent Cα atoms, respectively. As this LB domain is vital for the

catalytic mechanism of MurC, its inhibition by molecular adducts could provide a novel baseline

for providing novel antibiotics (Mol et al., 2003; Kouidmi, Levesque, & Paradis-Bleau, 2014).

Previously, to block catalytic mechanism of Mur ligases, derivatives of phosphinate were designed

with promising activity against MurD and MurE having IC50 value in micromolar (Strancar,

Boniface, Blanot, & Gobec, 2007). Naphthalene-N-sulphonyl-D-glutamic based derivatives of

sulphonamide were observed to have anti-MurD activity (Simcic et al., 2012). Perdih et al. using

a virtual screening approach identified derivatives of benzene 1,3-dicarboxylic acid that inhibits

both MurD and MurE with IC50 values in the micromolar (Perdih et al., 2009). Quite promptly,

the same group reported structural derivatives of benzene-1,3-dicarboxylic acid 2,5-

dimethylpyrrole as micromolar inhibitor of Mur ligases (Perdih et al., 2014).

Considering the need of developing new antibiotics, in the present study, we modeled the MurC

enzyme from A. baumannii and used it in structure based screening of natural small molecules,

MD simulations and binding free energy calculations. The screened lead structure can conceivably

be explored for the design of novel antibiotics. Additionally, the potential of this lead structure to

inhibit the ATP binding domain was investigated to unveil its ability as a multi-domain inhibitor.

This could provide an opportunity to simultaneously block more than one Mur ligase enzyme

rendering the reduced likelihood of target mediated resistant development.

`273


8.3.1. Receptor Protein Structure Modelling and Evaluation

As the experimentally determined 3D structure of the receptor protein was absent, a comparative

structure prediction approach was used to predict the most optimal model for the MurC protein

(Dorn, e Silva, Buriol, & Lamb, 2014;Raza, Sanober, Rungrotmongkol, & Azam, 2017). For this

purpose, the protein sequence from A. baumannii strain ATCC 17978 was retrieved from UniProt

(UniProt ID: A3M9Y0) (Consortium & others, 2017). The template used in Modeller 9.14 to

model the MurC was identified using the online Blastp search of the National Center for

Biotechnological Information (NCBI) (Lobo, 2008). The template having maximum sequence

identity and query coverage with the target was considered as the most suitable template (Vyas,

Ukawala, Ghate, & Chintha, 2012). For comparative analysis, online structure prediction tools:

Modweb, Phyre2, and Swiss-model were employed (Ahmad, Raza, Uddin, & Azam, 2018). The

structure with maximum residues in the favorable region and minimum residues in the disallowed

region was considered as the suitable structure (Skariyachan, Manjunath, & Bachappanavar,

2018). Moreover, high Errat (Colovos & Yeates, 1993), Verify-3D (Eisenberg, Lüthy, & Bowie,

1997) and least Z-score (Wiederstein & Sippl, 2007) were the additional factors that were

considered for selection of the model (Ahmad, Raza, Uddin, Rungrotmongkol, et al., 2018). The

modeled structure was then minimized in UCSF Chimera for total of 1500 steps (Ahmad, Raza,

Abro, Liedl, & Azam, 2018). The protein was first minimized for 750 steps of steepest descent

algorithm to relieve unfavorable clashes; followed 750 rounds of slow conjugate gradient method

to remove the severe clashes remained during the first phase under TFF. In both algorithms, the

step size was set to 0.02 Å (Ahmad, Raza, Abbasi, & Azam, 2018).

8.3.2. Inhibitors Dataset Preparation

Asinex library (http://www.asinex.com/) containing natural scaffolds was used to identify the best

fitting molecule in the LB domain of the MurC protein. Prior to being used in docking, the library

was first filtered for drug-like and lead-like molecules (Lipinski, 2004;Vistoli, Pedretti, & Testa,

2008). The filters for druglikeness and leadlikeness were retrieved from Swiss-ADME (Daina,

Michielin, & Zoete, 2017) and used in Ligand Scout 4.1 (Wolber & Langer, 2005). The shortlisted

molecules were then minimized using MMFF94 force field (Halgren, 1996) in the Ligand Scout.

The inhibitors were used in either charged or neutral depending on their chemistry.

http://www.asinex.com/

`274

8.3.3. Structure based Virtual Screening using GOLD

Structure based virtual screening of the shortlisted lead-like compounds with MurC LB domain

was executed using GOLD 5.2 (Jones, Willett, Glen, Leach, & Taylor, 1997). The coordinates of

Asp334: OD1 were used for inhibitors binding with grid size set to 10 Å. This residue is highly

conserved among the LB domain of Mur ligases (Mol et al., 2003). GOLD fitness score (Jones,

Willett, Glen, Leach, & Taylor, 1997), ChemScore (Verdonk, Cole, Hartshorn, Murray, & Taylor,

2003), Astex Statistical Potential (ASP) (Mooij & Verdonk, 2005) and ChemPLP score (Korb,

Stutzle, & Exner, 2009) for the compounds were estimated in GOLD. Additionally,

Autodock/Vina binding energy (Trott & Olson, 2010) was also determine for the compounds. All

the docking scores were used in statistical correlation coefficient analysis to determined correlation

among them (Ahmad, Raza, Uddin, & Azam, 2017). In each scoring function, twenty iterations

were generated for each molecule. The selection of docked complex was based on the highest

GOLD score and the lowest binding energy, in addition, to maximum number of interacting

residues and hydrogen bonds (Verma, Tiwari, & Tiwari, 2018). For comparative analysis, two

controls: Aztreonam (monobactam) and Cefclidin (third-generation cephalosporin) were utilized.

Both controls are known inhibitors of bacterial peptidoglycan.

8.3.4. Computational Pharmacokinetics

Pharmacokinetics of the shortlisted best-docked molecule was determined using

http://biosig.unimelb.edu.au/pkcsm/prediction in order to unveil the best chemical moieties

bearing biological activity for future lead optimization (Hosea & Jones, 2013).

8.3.5. MD Simulations Setup

Molecular docking studies offer a single conformation of receptor-ligand complex and have a

limitation of not depicting the flexible nature of both entities (Hospital, Goni, Orozco, & Gelpi,

2015). As a result, docking findings might not be reciprocated in in vivo state (Ambure, Bhat,

Puzyn, & Roy, 2018). Thus, to make the docking studies more statistically significant, the best

docked conformation was subjected to MD simulations in an explicit water model (Karplus &

McCammon, 2002). MD simulations were performed using AMBER16 (Salomon-Ferrer, Case, &

Walker, 2013). The antechamber (J. Wang, Wolf, Caldwell, Kollman, & Case, 2004) and leap

program (Pearlman et al., 1995) of AMBER were used to prepare inhibitor parameters and specify

complex molecular interactions, respectively. To bring electrostatic neutrality, 9 Na+ ions were

https://en.wikipedia.org/wiki/Aztreonam

http://biosig.unimelb.edu.au/pkcsm/prediction

`275

added to the MurC-inhibitor complex. The complex was solvated within 12 Å TIP3P water box

(Fig.8.1). The complex was minimized in three consecutive phases where minimization of the

whole system, water box and non-heavy atoms of the complex was achieved (Andleeb et al., 2016).

The system was then gradually heated to 300 K with a time scale of 20 picoseconds (ps). With a

time step of 2-ns, the system was then equilibrated for 100-ps. System pressure was maintained

for 50-ps using constant pressure and temperature (NPT) ensemble (Uline & Corti, 2013). To

regulate temperature and pressure, Langevin dynamics (Pastor, Brooks, & Szabo, 1988) was used,

while SHAKE algorithm (Krautler, Van Gunsteren, & Hunenberger, 2001) was used to correct

bond length. For the production phase of 100 nanoseconds (ns), Berendsen algorithm (Berendsen,

Postma, van Gunsteren, DiNola, & Haak, 1984) coupled with the constant volume and temperature

(NVT) ensemble (Ahmad, Raza, Abbasi, & Azam, 2018) was used. The CPPTRAJ module (Roe

& Cheatham, 2013) of the AMBER was used to analyze the generated trajectories. Different

statistical parameters, like: RMSD (Haq, Abro, Raza, Liedl, & Azam, 2017), RMSF (Haq, Abro,

Raza, Liedl, & Azam, 2017), β-factor (Haq, Abro, Raza, Liedl, & Azam, 2017) and Rg (Haq, Abro,

Raza, Liedl, & Azam, 2017) were estimated as a function of time to look for structural variations

of the

`276

Fig.8.1. MurC-inhibitor complex in the TIP3P water box. The MurC enzyme is shown in yellow cartoon while the inhibitor is in red CPK.

`277

protein. RDF plots (Donohue, 1954) were generated to estimate the distribution of inhibitor atoms

with respect to protein atoms crucial for complex stability at the beginning and end of the

simulation. Additionally, the local structural rearrangements and displacements of the inhibitor

with respect to the reference protein point were investigated using a novel analytical tool of AFD

that provides in-depth information about system stability (Raza & Azam, 2018).

Mathematically, AFD can be presented as,

………………………..…………………………………………( XXXIII)

In the equation, k and l on X and Y-axis, respectively are the cut-off values with respect to the

protein reference atom, while j and I represent ligand distribution on the X and Y-plane,

respectively.

8.3.6. Binding Energies Calculation

The binding free energies of the complex were estimated using MM/GBSA and MM/PBSA

approaches of AMBER16 (Miller et al., 2012; Genheden & Ryde, 2015). Estimation of binding

free energies is based on the following mathematical equation,

∆Gbinding = Gcomplex – Greceptor – Gligand……………………………………………………( XXXIV)

The above equation computes the energy difference between the complex free energy and that of

receptor and ligand. ∆G represents the binding free energy in kcal/mol of the complex, while

Gcomplex, Greceptor, and Gligand is the free energy of the respective state. Each energy term is calculated

using the following equation,

G = GGB/PB + EMM + GSA……….....………........................................................................(XXXV)

where GGB/PB is the polar contribution towards solvation energy, EMM is molecular mechanical

energy and GSA is the contributions from non-polar terms towards solvation energy. The EMM is

the product of three terms.

EMM = Eint + Eele + E VWD………………………………………………………………….(XXXVI)

Eint is the internal energy (bond, angle, and torsional angle energy), Eele is the electrostatic energy

and EVDW is the van der Waals energy. In equation (3), the GSA can be split into,

G SA = γ × SASA × b ……………………………………………………………….……...(XXXVII)

Where γ is the surface tension proportionality constant (0.0072 kcal/mol·Å2 in MM/GBSA and

`278

0.0052 kcal/mol·Å2 in MM/PBSA), SASA is the solvent-accessible surface area computed using

LCPO (Weiser, Shenkin, & Still, 1999), and b: non-polar solvation energy for a point solute (0

kcal/mol in MM/GBSA and 0.92 kcal/mol in MM/PBSA).

8.5.7. WaterSwap based Energy Calculations

To further validate the inhibitor affinity for the docked site, the binding free energy of the complex

was re-calculated with a relatively new and practically a more accurate method of WaterSwap

(Woods, Malaisree, Hannongbua, & Mulholland, 2011; Woods et al., 2014). WaterSwap uses a

novel reaction coordinate, which allows the swapping of a bounded ligand to a protein with a

cluster of water molecule selected from bulk water (Woods, Malaisree, Hannongbua, &

Mulholland, 2011). The WaterSwap reaction coordinate (WSRC) works on the principle of

identity constraint: that identify an identical cluster of water molecules from the bulk water in an

equivalent volume of the ligand in protein active site (Woods, Malaisree, Hannongbua, &

Mulholland, 2011; Woods et al., 2014) and swap it with each other using a dual topology algorithm

(Woods, Malaisree, Long, McIntosh-Smith, & Mulholland, 2013). Ultimately, the free energy is

estimated using a replica exchange thermodynamic integration (Woods, Malaisree, Hannongbua,

& Mulholland, 2011; Woods, Malaisree, Long, McIntosh-Smith, & Mulholland, 2013; Woods et

al., 2014). WaterSwap has been successfully applied to calculate the binding free energy of a

neuraminidase inhibitor (oseltamivir) and the results were found comparable with that of

experimental ones (Woods, Malaisree, Long, McIntosh-Smith, & Mulholland, 2013). WaterSwap

calculations were performed for 1000 iterations using WSRC module in Sire with the same solvent

and force fields parameters of dynamics simulation. The input files for WaterSwap were the Amber

format coordinate and topology files, which represent MurC-inhibitor complex solvated in the

TIP3P water box. WaterSwap calculates the absolute binding free energy for the complex using

four statistical methods: Bennetts acceptance ratio (BAR), thermodynamic integration (TI),

quadrature-based integration of TI, and free energy perturbation (FEP) method. The absolute

binding free energy is the negative of the mean of these four estimates (Woods, Malaisree,

Hannongbua, & Mulholland, 2011; Woods, Malaisree, Long, McIntosh-Smith, & Mulholland,

2013; Woods et al., 2014). In case if the simulation is well-converged, then the four estimates

should be roughly equal (Ahmad, Raza, Uddin, Rungrotmongkol, et al., 2018; Ahmad, Raza,

Abbasi, & Azam, 2018).

`279


With the alarming increase of antibiotic resistance among bacterial pathogens, the need for the

development of antibiotics with novel mechanism of action against the highly specific and

selective bacterial target is a high priority task (WHO, 2017; Almaghrabi, Joseph, Assiry, &

Hamid, 2018). Among several targets of pathogenic bacteria, the biochemical pathway that

governs PG biosynthesis is a target of high suitability (Vollmer, Blanot, & De Pedro, 2008;

Barreteau et al., 2008). Within this pathway, the amide ligases: MurC, MurD, MurE, and MurF

are homologous in catalytic mechanism and shared conserved structural features and amino acid

regions that can be explored to design inhibitors that target simultaneously more than one enzyme

(Mesleh et al., 2016). This not only could provide an opportunity for designing multi-target

antibacterial weapons but also at the same time could minimize the probability of target mediated

resistant development (Kouidmi, Levesque, & Paradis-Bleau, 2014).

8.4.1. MurC Structure Modelling

Predicting a correct and reliable 3D structure for a protein of interest is an arduous and intricate

task. However, once predicted it provides valuable insights into the principle of molecular

recognition and the possibility of finding inhibitors (Breda, Valadares, de Souza, & Garratt, 2007;

Dorn, Buriol, & Lamb, 2013). As stated in the methodology, a comparative structure modeling

approach was used for predicting the most suitable model of MurC. The experimentally

determined template used in Modeller was MurC from Y. pestis available in PDB (ID: 4HV4) with

good resolution of 2.25 Å. The sequence identity and query coverage of the template with the

target was 61% and 97%, respectively. The same template was used in the remaining online

structure prediction servers. In total, 80 MurC models were generated, 20 each from Phyre2, Mod-

Web, and Swiss-Model. In case of Modeller, top five were selected, while from Phyre2, Mod-

Web, and Swiss-Model only the best predicted structure was selected. Evaluating the quality of

these shortlisted structures revealed ModWeb model as the most reliable model for the targeted

MurC. As can be seen in S-Table 8.1 that majority of the model residues are mapped in the most

favorable region compared to the rest of the structures. More importantly, only one residue of the

protein was found in the disallowed region. Additionally, the model has appreciable Errat score of

83.76, Verify-3D score of 90.97 and ProSA Z-score of -11.13. Superimposition of the modeled

protein over the template revealed an RMSD of 0.622 Å depicting the high reliability of the

`280

modeled protein (Fig.8.2A). The 3D structure of the modeled MurC is shown in Fig.8.2B. The

modeled protein has three domain: LB, ATP, and the UDP binding domain. The UDP domain has

the residues from Val1 to Tyr106, ATP binding domain has the residues from Arg107 to Arg314,

and LB domain compromises of residues form 315 to Gln461. The secondary structure of the

protein was predicted using PDBSum that unveiled 23.3% of the strand, 31.3% of alpha helix,

2.7% of 3-10 helix and 42.6% of other secondary structure elements (S-Fig.8.1A). The

Ramachandran plot of the modeled MurC protein is shown in S-Fig.8.1B.

8.4.2. Inhibitor Library Preparation

Asinex library is a valuable resource of antibacterial compounds based on natural product like

scaffolds that provide great skeletal diversity combined with the presence of multiple stereogenic

centers and polar functional groups. The final compounds in the library have structural elements

inherent to known antibiotics. The Asinex library contains 8044 compounds. Prior to use in

structure based virtual screening, the library was screened first for drug-like molecules. Drug-like

compounds have functional groups/physical properties similar to the majority of the drugs and

thus can be inferred to show therapeutic potential (Kadam & Roy, 2007). Five prominent drug-

like rules: Lipinski rule of five (Lipinski, 2004), Ghose filter (Ghose, Viswanadhan, &

Wendoloski, 1999), Veber filter (Veber et al., 2002), Egan rule (Egan, Merz, & Baldwin, 2000)

and Muegge rule (Muegge, Heald, & Brittelli, 2001). According to Lipinski rule, a molecule will

be drug-like if its molecular weight (MW) ≤ 500, octanol-water partition coefficient (LogP) ≤ 4.15,

hydrogen bond acceptors (HBAs) ≤ 10, hydrogen bond donors (HBDs) ≤ 5 and TPSA 40-130 Å2

(Lipinski, 2004). For Ghose filter, the drug-like molecules have the following descriptors: MW ≥

160 - ≤ 480, LogP ≥ 0.4 - ≤ 5.6, number of atoms ≥ 20 - ≤ 70, molecular refractivity (MR) ≥ 40

- ≤ 130 (Ghose, Viswanadhan, & Wendoloski, 1999). According to Veber filter, the compounds

having rotatable bonds ≤ 10 and TPSA ≤ 140 (Veber et al., 2002) are considered as drug-like. The

parameters: TPSA ≤ 131.6 and Wildman and Crippen LogP (WLogP) ≤ 5.88 of compounds

represent the Egan rule (Egan, Merz, & Baldwin, 2000). For Muegge rule, those having MW ≥

200 - ≤ 600, TPSA ≤ 150, HBDs ≤ 5, HBAs ≤ 10, number of rings ≤ 7, number of carbons > 4,

number of heteroatoms > 1, number of rotatable bonds ≤ 15, and octanol-water partition

coefficient of organic compounds (XLogP) ≥ -2 - ≤ 5 are drug-like (Muegge, Heald, & Brittelli,

2001). The molecules were subjected to further filtration to shortlist only lead-like compounds.

Lead-like compounds in drug discovery are pharmacological and biologically active chemical

`281

entity that is likely to be therapeutically useful. The chemical structure of lead-like compounds

could serve as a starting point for chemical modification to improve selectivity, potency and

pharmacokinetics properties. The lead-like descriptors: XLogP ≤ 3.5, 250 ≤ MW ≤ 350 and

rotatable bonds ≤ 7 (Teague, Davis, Leeson, & Oprea, 1999) were applied in Ligand Scout, which

revealed total of 2995 lead-like inhibitors. These inhibitors were energetically minimized and used

in molecular docking.


In rational drug designing, molecular docking is an attractive tool to understand bimolecular-drug

interactions (Meng, Zhang, Mezei, & Cui, 2011). The holy grail of docking is to predict the

predominant binding pose of a ligand into the preferred binding pocket of the target protein with

known 3D structure (Morris & Lim-Wilby, 2008). Such understanding of small molecules

recognition and interactions with the macromolecule is of great importance in pharmaceutical

research (Kitchen, Decornez, Furr, & Bajorath, 2004). For instance, investigation of critical

molecular events such as those intermolecular interactions involved in stabilizing receptor-ligand

complex can be unveiled through molecular docking (Kitchen, Decornez, Furr, & Bajorath, 2004;

Morris & Lim-Wilby, 2008;Meng, Zhang, Mezei, & Cui, 2011). Additionally, molecular docking

algorithms can perform quantitative prediction of docking scores and binding energies that ranked

docked compounds based on its affinity for the receptor (Kitchen, Decornez, Furr, & Bajorath,

2004; Morris & Lim-Wilby, 2008;Meng, Zhang, Mezei, & Cui, 2011). Identification of the most

likely binding conformation of the ligand among the several generated is based on two rules: (i)

representing various binding mode by exploring the large conformational space and (ii) predicting

the accurate interaction energy of each binding conformation (Ferreira, dos Santos, Oliva, &

Andricopulo, 2015). These tasks are accomplished through a cyclic process during which the

conformations of a ligand is evaluated through specific scoring functions (Ferreira, dos Santos,

Oliva, & Andricopulo, 2015). This process is repeated until a solution of minimum energy is

converged (Ferreira, dos Santos, Oliva, & Andricopulo, 2015).

`282

Fig.8.2. A. Superimposition of Modeled MurC structure (Red) over 4HVA template (Blue), B. 3D structure of the modeled MurC protein three

different domains are shown.

`283

In GOLD, the docking was performed with four scoring functions: GoldScore, ChemScore, ASP

and ChemPLP score. These scores are dimensionless; however, each score represents the scale to

guide how worthy the pose is. High score illustrates better docking results. The GoldScore fitness

function has been optimized to predict ligand binding conformations by taking into account

factors, like: ligand torsion strain, H-bonding energy, metal interaction, and van der Waals energy

(Verdonk, Cole, Hartshorn, Murray, & Taylor, 2003). The ChemScore fitness function includes a

term, dG, which is the change in total free energy, occurred upon ligand binding. It also

incorporates internal energy and protein-ligand atom clash term. ChemScore consists of hydrogen

bonding, metal interaction, hydrophobic-hydrophobic contact area, and ligand flexibility

(Verdonk, Cole, Hartshorn, Murray, & Taylor, 2003). ASP fitness function is the atom-atom

distance potential calculated from protein-inhibitor complex (Mooij & Verdonk, 2005).

Comparatively, ChemPLP is four times faster than GoldScore and employs ChemScore multiple

linear potentials and hydrogen bonding term to model repulsive and van der Waals terms (Korb,

Stutzle, & Exner, 2009). Additionally, the binding affinity of the compounds was determined using

Autodock/Vina. Complex with the highest GOLDScore and lowest Autodock/Vina binding energy

was categorized as the most promising complex with complementary partners. Different scoring

functions and Autodock/Vina binding energies of shortlisted 10 best complexes are illustrated in

S-Table 8.2. Correlation coefficient between the GOLDScore and different scoring functions and

Autodock/Vina binding energies was evaluated to measure the strength and direction of

relationship among them on a scatter plot as depicted in Fig.8.3.

`284

Fig.8.3. Correlation coefficient among scoring functions.

`285

As shown, no linear relationship was found to exist among GOLDScore, ChemScore, and ASP,

while between GOLDScore and AutoDock/Vina binding energies a strong uphill positive

correlation was revealed. The inhibitor (1'-((2H-imidazol-2-yl)methyl)-N-(pyridin-2-yl)-1',2'-

dihydro-[4,4'-bipyridin]-2-amine) is characterized as top molecule with GOLDScore and

AutoDock/Vina binding energy of 72.05 and -9.5 kcal/mol, respectively. The docking scores of

this molecule as compared to the standards are better: Aztreonam (GOLDScore, 42.12 and

AutoDock/Vina binding energy, -6.1 kcal/mol) and Cefclidin (GOLDScore, 45.23 and

AutoDock/Vina binding energy, -6.5 kcal/mol). The binding mode and interactions of the inhibitor

are depicted in Fig.8.4 and Fig.8.5, respectively. Visual inspection of the top complex revealed

deep binding of the compound in the binding cavity of the protein. In GOLD docking, the inhibitor

was bound in such a way to cover both ATP and LB domain. Visual inspection of the complex

revealed that the 2-methyl-2H-imidazole ring of the inhibitor shows affinity for the interface of

the ATP and LB domain, the central structure of 1-methyl-1,2-dihydro-4,4-bipyridine shows an

appreciable affinity for the LB domain, whereas the terminal pyridine favored interacting with the

residues of ATP binding domain. In AutoDock/Vina, the inhibitor was posed in complete opposite

orientation to that of GOLD. The compound was seen positioned more at the interface of ATP

binding domain and LB domain. The 2-methyl-2H-imidazole ring of the inhibitor was positioned

in the enzyme cavity space and favored to interact with the residues of the ATP binding domain.

However, the central structure of 1-methyl-1,2-dihydro-4,4-bipyridine of the inhibitor is placed

like in GOLD towards the LB domain. Within 5 Å of the inhibitor in GOLD, from LB domain the

following residues: Arg314, Phe316, Asp334, Gly336, Hie337, Ile338, Glu341, Val342, Thr345,

Gln447 and Ala348 while from ATP binding domain residues, like: Hie116, Gly117, Lys118,

Thr119, Gly141, Glu162, Gln276, Pro277, Gly278 and Asn281 interact with the inhibitor.

8.4.4. Computational Pharmacokinetics of the Top Most Inhibitor

Pharmacokinetics is the fate of an administered drug in the living organism (Tozer & Rowland,

2006). Understanding pharmacokinetics of a given compound is important at the drug design stage

as it can aid in future optimization of the lead compound with improved pharmacokinetics

(Wadood et al., 2017). The pharmacokinetics of the compound can be divided into the following

aspects: absorption, distribution, metabolism, excretion, and toxicity.

8.4.4.1. Absorption

Absorption of the compound was predicted first through six parameters: logS, logKp, Coca-2


`286

permeability, intestinal absorption, P-glycoprotein substrate, and P-glycoprotein I and II inhibitors

(Pires, Blundell, & Ascher, 2015). The logS value is the solubility of a drug in water at 25 °C and

is expressed as the logarithm of the molar concentration (log mol/L) (Savjani, Gajjar, & Savjani,

2012). Drugs with logS value > -4 are considered to have better absorption. The predicted logS

value for the compound is -2.81. LogKp represents the skin permeability for an intended drug

molecule (Pires, Blundell, & Ascher, 2015). Compounds with LogKp value of > -2.5 cm/h usually

have low skin permeability (Pires, Blundell, & Ascher, 2015). For our shortlisted compound, the

predicted LogKp value is -2.73, which means that the compound has good skin permeability. The

monolayer of Coca-2 is frequently employed as in vitro model for predicting the absorption of an

orally administered drug (Pires, Blundell, & Ascher, 2015). A drug molecule with value > 0.90 is

believed to have higher permeability (Pires, Blundell, & Ascher, 2015). Our compound of interest

was revealed with Coca-2 permeability score of 1.14 (log Papp in 10-6 cm/s), indicating the high

potency of the drug to permeate into the Coca-2 cells. The intestinal absorption of the compound

is very high i.e. 96.68 % that illustrate the higher quantity of drug can reach to the target site. The

compound was unveiled to act as a substrate for P-glycoprotein. P-glycoprotein is an ATP

transporter protein and is involved in effluxing toxins and xenobiotics out of the cell (Lin &

Yamazaki, 2003). Further, the compound is non-inhibitor of both P-glycoprotein variants: P-

glycoprotein I and P-glycoprotein II.

8.4.4.2. Distribution

The distribution of compound was explained by volume of distribution (VD), Fraction unbound

(Fu), blood-brain barrier (BBB) and CNS permeability (logPS) (Pires, Blundell, & Ascher, 2015).

The VD is a theoretical volume that a drug dose required for uniform distribution in the body by

achieving a constant concentration in the blood plasma (Pires, Blundell, & Ascher, 2015). The unit

of VD is log L/kg. A compound with value > 0.45 has higher distribution of the drug in tissues

compared to plasma, while lower VD (< -0.15) value indicate the reverse (Pires, Blundell, &

Ascher, 2015). The predicted VD value for the compound is -0.27 that indicates maximum

compound concentration in the blood plasma. Fu is a fraction of unbounded compound in the blood

plasma that can transverse biological membranes for its target site action (Pires, Blundell, &

Ascher, 2015). The Fu value for the compound is 0.377. The ability of compounds to penetrate the

blood-brain barrier (BBB) is important to evaluate as exogenous compounds with pharmacological

activities in the brain could lead to toxic effects (Pires, Blundell, & Ascher, 2015).

`287

Fig.8.4. Binding mode of the inhibitor in GOLD (A) and AutoDock/Vina (B).

`288

Fig.8.5. Binding interactions of the inhibitor in GOLD (A) and AutoDock/Vina (B).

`289

In animal models, this blood-brain barrier permeability is expressed as logBB. The predicted

logBB for the compound is -1.12 and thus has very low brain permeability. Compounds with

logBB value of > 0.3 are usually readily cross the BBB, while those with logBB < -1 has poor

BBB permeability. The CNS permeability of the compound is -3.39, which illustrates the

compound inability to permeate the CNS.

8.4.4.3. Metabolism

From a metabolic point of view, the compound potential to inhibit the Cytochrome P450 enzyme

was predicted. This is important as this enzyme is involved in detoxifying xenobiotics and assists

in their excretion (Nebert, Wikvall, & Miller, 2013; Pires, Blundell, & Ascher, 2015). It was found

that the compound is non-inhibitor of the majority of Cytochrome P450 enzyme isoforms:

CYP3A4, CYP2D6, CYP2C19, CYP2D6, CYP2C9, and CYP3A4 except CYP1A2.

8.4.4.4. Excretion

The excretion of compound was studied from two detrimental aspects: its ability to block Renal

cation transporter 2 (Renal OCT2) and total clearance of the drug measured as log CLtot (Pires,

Blundell, & Ascher, 2015). Renal OCT2 is a transporter protein, which plays role in the renal

clearance of endogenous drugs (Pires, Blundell, & Ascher, 2015). The shortlisted compound in

our study was found as non-substrate for this protein. Total clearance is combination of renal and

hepatic clearance of a drug (Pires, Blundell, & Ascher, 2015). The total renal clearance for the

compound is 1.199 (log ml/min/kg).

8.4.4.5. Toxicity

Toxicity is estimated to be responsible for attrition of approximately 1/3 drugs and is associated

with increased cost of drug development (Spławinski, Kuzniar, Filipiak, & Zielinski, 2006;Pires,

Blundell, & Ascher, 2015). The drug toxicity can be grouped into several ways, including: immune

hypersensitivity, on-target toxicity, off-target toxicity, and covalent modification (Pires, Blundell,

& Ascher, 2015). The acute toxicity was measured by lethal dosage 50 (LD50) and can be

expressed as mol/kg (Pires, Blundell, & Ascher, 2015). The oral rat acute toxicity is 2.45 mol/kg.

The oral rat chronic toxicity in log mg/kg_bw/day for the compound is 1.59. The AMES toxicity

in bacteria is the most widely used approach to investigate drug mutagenicity. Computational

AMES test predicted the compound as mutagenic. The compound was predicted as non-sensitizer

of skin. The maximum tolerated dose (log mg/kg/day) of the compound in human is 0.75.

`290

Similarly, the compound was categorized as hepatotoxic. The end point toxicity of Tetrahymena

pyriformis was evaluated for the compound and predicted that the compound as toxic as it scored

as 0.28. Compounds with > -0.5 log ug/L are usually considered toxic. The minnow toxicity of

drug represents drug concentration required to kill 50% of flathead minnows. The LD50 value < -

0.3 for a compound is considered to have high toxicity. The predicted LD50 score for the

compound is 2.81 that indicated the compound as least toxic. In the end, toxicity of the compound

for human ether-a-go-go gene (hERG) isoforms: hERG I and hERG II (Pires, Blundell, & Ascher,

2015;Yu, Zou, Wang, & Li, 2016; Wadood et al., 2017) was evaluated. The compound was

predicted toxic for hERG II and non-toxic for hERG I.

8.4.5. MD Simulations of the Complex

The technique of MD simulation is used to study the physical movements of atoms and molecules

and can be effectively employed to understand the structure-to-function relationship of a given

macromolecule (Hospital, Goñi, Orozco, & Gelpi, 2015; Ambure, Bhat, Puzyn, & Roy, 2018).

Additionally, MD simulation can be used to describe the strength, pattern, and properties of

receptor-drug interactions, the different conformational changes of a protein or molecule undergo

under various conditions and the solvation of molecules (Abbasi, Raza, Azam, Liedl, & Fuchs,

2016). The dynamics of proteins are difficult to study in the lab as the process is complicated, time

and resource costly. Thus, utilizing the power of computers in simulation or in silico MD

simulation, on the contrary, holds great opportunities in deciphering the dynamics of protein

molecules complex with inhibitors. Four statistical parameters were evaluated for the protein

stability as depicted in Fig.6.

8.4.5.1. RMSD Analysis

RMSD measures the average distance between the Carbon alpha atoms from the backbone chain

of the superimposed proteins (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016;Abro & Azam, 2016).

The mean RMSD of 2.53 Å was noticed for the protein; with maximum RMSD reaches to 3.59 Å

at 73th-ns. As can be observed that the RMSD graph fluctuates consistently until 74th-ns of the

simulation time, followed by a brief phase (75th-ns to 96th-ns) of equilibrium where the protein

remained stable. At 97th-ns, the RMSD fluctuates with a minor jump and in order to investigate

this, the simulation was extended. The extension till 150th-ns validated again the stability in general

(S-Fig.8.2).

`291

Fig.8.6. MD simulations trajectories analysis. A. RMSD, B. RMSF, C. Rg, D. β-factor.

`292

To thoroughly investigate this inconsistent trend of 100-ns RMSD, snapshots at different periods

i.e. 0-ns, 10th-ns, 20th-ns, 30th-ns, 40th-ns, 50th-ns, 60th-ns, 70th-ns, and 100th-ns were extracted from

simulation trajectories and visualized to look for the structural adjustments of the protein.

Superimposition of 10th-ns snapshot over 0-ns revealed inhibitor induced minor structural

variations of the protein from its original position (Fig.8.7). This could be explained as an approach

utilized by the protein to properly hold the rotating and moving inhibitor in its active pocket. The

displacement of protein structure from its original position after 10th-ns was 1.03 Å leading to

increased stability of the complex due to the formation of two additional hydrogen bonds along

with Asp334 of active site residues: Arg315 with distance of 2.2 Å and Thr345 with distance of

2.0 Å. The conformational shift observed after 10th-ns involves residues: Ala61-Ile64 from UDP

binding domain and Met188-Glu192 from ATP binding domain where loop is converted into a

helical structure. These regions are not part of the inhibitor binding region and thus not affecting

the inhibitor binding.

Fig.8.7. A. Superimposed protein at 10th-ns (Dark Khaki) over 0-ns (Coral). B. Superimposed

inhibitor at 10th-ns (Dark Khaki) over 0-ns (Coral).

`293

The superimposed structure obtained at 20th-ns over 10th-ns can be seen in Fig.8.8. After 20th-ns,

the inhibitor was observed to come in close contact with the LB domain leading to the conversion

of a sheet comprising residues from Pro136-Val139 to loop. Another conformational shift

observed was the conversion of helical region residues Ala61-Ile64 to loop. This movement of the

inhibitor towards the LB domain resulted into structural displacement of 1.03 Å. As a result of this

displacement, hydrogen bond with the previously bounded residues: Thr335 and Arg315 were

broken, while the strength of hydrogen bond interaction with that of Asp334 increased by reducing

the distance to 2.0 Å. Additionally, a new hydrogen bond with Arg314 from the LB domain was

formed at a distance of 1.9 Å.

Fig.8.8. The superimposed complex of 20th-ns (Dark Khaki) over 10th-ns (Coral).

`294

Superimposition of the 30th-ns complex over 20th-ns (Fig.8.9) revealed minor protein structural

variability as depicted by the low RMSD of 0.92 Å. No conformational shifts were found. The

structural adjustments of the protein are due to inward movements of the inhibitor towards the

interface of ATP and LB domain. After 30th-ns, the inhibitor was seen to interact through very

strong hydrogen bonding with Asp334 (distance 1.6 Å) and Glu341 (distance 2.0 Å).


`295

No conformational shifts were observed for the protein after 40th-ns (Fig.8.10). The displacement

of structure from its original position at 30th-ns was 1.0 Å and is because of the inhibitor movement

towards the ATP binding domain.


`296

The 50th-ns complex when overlaid over that of 40th-ns (Fig.8.11), an RMSD of 1.03 Å was

revealed. The region from Glu165- Asp167 and Met188-Glu192 were converted from loop to coil

regions. The inhibitor position remained the same. Three hydrogen bonds were observed between

the inhibitor and protein. Two hydrogen bonds between Asp334 atoms: OD1 and OD2 and

inhibitor at distance 2.1 Å and 1.9 Å, respectively and one between Arg315 and the inhibitor were

seen.


`297

After 60th-ns (Fig.8.12), different conformational shifts were observed. A sheet comprising

residues from Glu252 to Gly255 in the ATP binding domain and a helical region comprising

residues from Pro425 to Gly428 in the LB domain converted to loop. Similarly, the reversion of

helical regions: Met188-Tyr191 and Glu165 to Ala168 to loop and conversion of a sheet (Ile93-

Val94) to loop in the ATP binding domain were noticed. Minor inward movement of the inhibitor

towards the interface of LB domain was noticed resulting in an additional hydrogen bond between

the terminal 2-methyl-2H-imidazole ring and Asp334 residue.


`298

The complex after 70th-ns (Fig.8.13) was noticed with structural adjustment of 1.04 Å:

conformation shift of loops: Val251 to Gly255 and Met188-Glu192 to sheet was revealed. Four

hydrogen bonds were observed: three between Asp334 and inhibitor and one between Arg314 and

the inhibitor. The inhibitor was seen a bit towards the enzyme main cavity.


`299

At the last 100th-ns (Fig.8.14), the complex when superimposed over that of 70th-ns was noticed

with an RMSD of 1.07 Å. Two conformational shifts were observed. The loop region residues

from Asp164 to Ala168 converted into a sheet, while the sheet region containing residues from

Asp167 to Tyr172 converted into the loop. The hydrogen bonding pattern was seen the same that

observed for the complex at 70th-ns.


`300

The frequency of snapshots was reset to every 5-ns and superimposed to look for stable protein

RMSD (S-Fig.8.3-S-Fig.8.7). Further, each snapshot was subjected to hydrogen bond analysis to

validate the complex stability and strength of interaction between the inhibitor and protein active

residues (S-Table.8.3). Each superimposition revealed high stable RMSD and an appreciable

number of strong hydrogen bonding thus revealing high complex stability along the simulation

time.

8.4.5.2. RMSF Analysis

RMSF is the measure of deviations of a particle from its original position (Abbasi, Raza, Azam,

Liedl, & Fuchs, 2016). RMSF analysis was carried out for C-alpha atom of each residue

representing the average displacement of each atom (Abro & Azam, 2016). The average RMSF

value calculated for the protein is 1.18 Å: with maximum RMSF value of 4.5 Å observed for

Pro425 present in the loop region of the LB domain. The loop regions are highly flexible

components of protein and their flexibility is necessary to properly accommodate the ligand at the

binding site. Majority of the protein residues were noticed stable with RMSF values < 3 Å.

Especially, residues of the active site that were seen in interactions with the inhibitor remained

highly stable during the whole course of simulation. The residues with RMSF value > 3 Å mainly

belong to those regions that showed conformational shifts at different ns of simulation.

8.4.5.3. Rg Analysis

The Rg analysis was done to better understand the equilibrium conformation and compactness of

a given protein (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Haq, Abro, Raza, Liedl, & Azam,

2017). Variations in the Rg values represent variation in the protein structure compactness during

the simulation period (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016). The high Rg value implies

protein structure with less tight packing, while the lower Rg value represents protein structure with

tight packing (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016). The average Rg value estimated for

the protein is 23.40 Å: with a maximum value of 24.27 Å at 30th-ns. Variations in the Rg trend

were seen between 10th-ns to 37th-ns, while the after trend of Rg towards the end period remained

highly stable. The variations in Rg could be explained by the conformational shifts that resulted in

the conversions of loops to sheet and sheet to loops during the simulation period.

8.4.5.4. β-factor Analysis

β-factor is the thermal disorders of a protein and specify structural stability of residue in term of

RMSF (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Haq, Abro, Raza, Liedl, & Azam, 2017).The

`301

findings of β-factor show consistency with RMSF. The mean β-factor calculated for the protein is

46.12 Å²: with maximum value observed is 541.6 Å² for residues Pro425 and Val426 present in

the loop region of the LB domain of the enzyme.

8.4.6. RDF and AFD Analysis

The binding mechanism of the inhibitor enzyme complex was elucidated using RDF. RDF has

been utilized in several studies to unveil the bonding and distribution of inhibitor atoms with

respect to the reference protein residue or (Abbasi, Raza, Azam, Liedl, & Fuchs, 2016; Haq, Abro,

Raza, Liedl, & Azam, 2017). RDF is a function of distance ‘r’ between two entities and is

represented by g (r). An in-house script was used in VMD to screen the inhibitor atom(s) and

enzyme active site residue involved in strong hydrogen bonding at the beginning and end of the

simulation period. It was found that Asp334 from MurC was the major residue in inhibitor binding

throughout the simulation time. Especially, the atoms: OD1 and OD2 from Asp334, while NH

atom of the inhibitor was seen in the binding. RDF for both interactions was generated and

analyzed to present hypothesis about the strength of interactions as simulation proceeds. For

Asp334: OD1 and inhibitor HN atom, the largest peak of RDF in starting 10-ns of simulation

appears at 3 Å with g(r) value of 0.45 whereas, towards the end 10-ns of simulation, the largest

peak appears at 2 Å with g(r) value of 0.88 (Fig.8.15A). It is very apparent from the figure that in

the last 10-ns of simulation, the density distribution of Asp334: OD1 and inhibitor HN interaction

is high. The narrow magnitude of the plot indicated the shortening of hydrogen bond distance

between the atoms rendering to the greater affinity of the interaction and more stability of the

complex. In case of Asp334:OD2 and inhibitor HN atom, the highest peak of RDF noticed for the

first 10-ns of simulation at a distance 1.75 Å with g(r) value of 2.80 whereas for the last 10-ns of

simulation, the highest RDF for the said interaction was seen at 1.79 Å with g(r) value of 1.68

(Fig.8.15B). The figure depicts that at the start of simulation, the magnitude of the graph was

narrow and sharp illustrating that density distribution of the interacting atoms was refined and in

very close proximity. In the final 10-ns of simulation, the density distribution expanded; however,

the distance of interaction decreases by 0.1 Å, an affirmation of hydrogen bond strengthening and

increase complex stability.

`302

Fig.8.15.A. Asp334-OD1-Lig-HN, B. Asp334-OD2-Lig-HN.

`303

The RDF distance being a scalar quantity cannot properly address the orientation of a given

molecule with respect to the reference (Raza & Azam, 2018). Thus, RDF has this limitation of

accessing coordinate geometry. To overcome this limitation of RDF, AFD that is a novel analytical

tool was used to investigate the distribution of ligand atoms on different coordinate planes

providing deeper information of enzyme mechanism of action (Raza & Azam, 2018). For

interactions, three point and top down perspective graphs were generated. In the case of Asp334:

OD1 and inhibitor HN atom (Fig.8.16), consistent to the RDF, the intensity of distribution

increased towards the simulation end with inhibitor move close to the protein active cavity.

Generally, the inhibitor conformation in the active pocket of the enzyme was not rigid and stable

at the start and end of simulation time. The figure implies the tilting behavior of the core structure

of the inhibitor on its axis allowing the strong hydrogen interaction between the atoms. It was

observed during simulation, that this behavior could possibly because of the inhibitor affinity for

the Asp334 residue. The deep moves of the inhibitor inside the cavity are due to structural

adjustment of the enzyme active site region and according to that, the inhibitor rotates to get access

for the mentioned residue. The distribution of the interaction was reduced as well towards end of

the simulation. For Asp 334: OD2 and inhibitor HN atom (Fig.8.17), the distribution area of the

graph was seen expanded illustrating the close adjustment of the inhibitor atom with respect to

OD2. The density distribution was reduced similarly to that noted in RDF. Reduction in bonding

distance was another vital aspect of the interaction at the end of simulation time and can be

explained by the same rotating movement visualized for Asp334: OD1 atom. Thus, these local

structural movements of the ligand could be only inferred using AFD and provide an in-depth

understanding of inhibitor structural changes in the complex.

8.4.7. Binding Free Energies Calculation

The development of new medicine starts from discovery of lead compounds, which bind to the

biological macromolecule active site efficiently during the screening process (Kitamura et al.,

2014). An important aspect in the process is to obtain a significant set of lead compounds necessary

to synthesize chemical structures, as a small number of lead compounds does not always lead to

molecules with considerable biological activity and solubility (Cournia, Allen, & Sherman, 2017).

In this context, the high-throughput screening (HTS) has been widely applied in pharmaceutical

industries to identify lead compounds from a large dataset (Janzen, 2014).

`304

Fig.8.16. AFD for Asp334-OD1 and inhibitor HN atom.

`305

Fig.8.17. AFD for Asp334-OD2 and inhibitor HN atom.

306

The molecular docking studies in this regard could provide a variety of structural information,

such as: electrostatic interactions, hydrogen bonding formation, molecular surface

complementarity and so on (Ferreira, dos Santos, Oliva, & Andricopulo, 2015). However, it has

been observed that drugs designed by such methods have hardly any significant biological

activities (G. Wang & Zhu, 2016). To circumvent the limitations of docking studies, binding free

energies calculation for the complex was achieved using renowned methods of MM/GBSA and

MM/PBSA (Genheden & Ryde, 2015). The MM/GBSA and MM/PBSA calculations were

performed for total of 400 frames. The MM/GBSA calculations were performed using LCPO,

while the MM/PBSA calculations were performed through internal PBSA solver in Sander. The

summary of different binding energies calculated in both the methods is illustrated in Table 8.1.

Both approaches revealed robust interactions between the inhibitor and the receptor MurC protein.

The columbic interactions (∆Eele = -336.90 kcal/mol) are quite high and dominates the total system

energy. The van der Waals energies (∆EVDW = -45.52 kcal/mol) and non-polar solvation energy

(∆Gsol-np = -7.62 kcal/mol) of the complex were also found dominant. The non-favorable

contributions were observed for polar solvation free energy that in case of MM/GBSA (∆GGB) was

341.62 kcal/mol and in case of MM/PBSA (∆GPB) it was 252.78 kcal/mol. The total binding energy

determined in MM/GBSA (Htot, GB =-48.45 kcal/mol) and MM/PBSA (Htot, PB =-3.62 kcal/mol).

The difference in total energy value of both systems could be possibly due to polar solvation energy

that for MM/PBSA (∆Gsolv,PB) was 378.81 kcal/mol, higher than that for MM/GBSA (∆Gsolv,PB)

i.e. 333.97 kcal/mol. The total binding energy of the complex in both the methods was

decomposed into the protein residues and inhibitor to unveil the residues that play part in inhibitor

binding through the simulation period. Such residues with energy value (< -1 kcal/mol) are

considered as the hot spot amino acids and are vital for catalytic processing of the ligand. This

information can give insights into the relative contributions of atoms in interactions between the

inhibitor and protein.

307

Table 8.1. The contribution of different energies involved in complex formation between MurC

and the top inhibitor.

Contribution Energy values (kcal.mol-1)

Standard Deviation

(kcal.mol-1)

∆EVDW -45.52 3.11

∆Eele -336.90 18.98

∆Egas

∆GPB

∆Gsolv,PB

Htot,PB

-382.43

352.78

378.81

-3.62

19.56

15.71

15.61

9.28

∆Gsolv,GB 333.97 18.07

∆GGB 341.62 18.05

∆Gsol-np -7.62 -

Htot, GB -48.45 4.48

∆Gele,GB 4.71 -

∆Gele,PB 15.88 -

It was found in the MM/GBSA, that residues: Arg314 (-2.2 kcal/mol), Arg315 (-2.5 kcal/mol),

Phe316 (-1.9 kcal/mol), Asp334 (-7.9 kcal/mol), Tyr335 (-2.8 kcal/mol), Gly336 (- 1.0 kcal/mol),

Hie337 (-1.1 kcal/mol), and Thr345 (-4.3 kcal/mol) are hot spot amino acid that contribute

significantly in inhibitor recognition and binding throughout the course of simulation period. The

binding affinity of the inhibitor was also found very high with a binding free energy value of -22.1

kcal/mol. In case of MM/PBSA, the following protein residues: Arg314 (-1.8 kcal/mol), Phe316

(-1.7 kcal/mol), Asp333 (-1.1 kcal/mol), Tyr335 (-2.4 kcal/mol), Gly336 (-1.3 kcal/mol), Hie337

(-1.0 kcal/mol) and Thr345 (-1.8 kcal/mol) were entitled as the hot spot amino acids. Similar to

MM/GBSA, in MM/PBSA the inhibitor affinity for the protein cavity was seen high (-20.5

308

kcal/mol). In both the methods, Arg314, Phe316, Tyr335, Gly336, Hie337 and Thr345 were found

common and contribute majorly to the total binding energy of the complex (S-Fig.8.8). All these

residues are part of MurC active site and directly play role in inhibitor recognition and binding.

Especially, the residue Asp334 in MM/GBSA has the lowest binding free energy and involved in

strong hydrogen bonds throughout the simulation time. The detailed contribution of these residues

was identified by decomposing the total energy into its constituents, like: electrostatic, van der

Waals, polar solvation and non-polar contribution (S-Fig.8.9). As an affirmation of the previous

results, electrostatic and van der Waals energies were found to dominate the overall binding free

energy of these residues. In case of MM/GBSA, the highest electrostatic energy was revealed for

Asp334 (-48.66 kcal/mol), while in case of MM/PBSA Asp333 was the residue with the highest

electrostatic energy (-17.46 kcal/mol).

8.4.8. WaterSwap based Binding Free Energy Calculations

In MM/GBSA and MM/PBSA, snapshots at regular interval are extracted from simulation

trajectories and used to estimate binding free energy of a given complex. In contrast, WaterSwap

runs its own Monte Carlo (MC) simulation to calculate the binding affinity of water cluster and

ligand for the protein active pocket (Woods, Malaisree, Hannongbua, & Mulholland, 2011; Woods

et al., 2014). Though very popular, MM/GBSA and MM/PBSA have serval limitations. First, the

hydration energy difference between the protein and protein-ligand complex in MM/GBSA and

MM/PBSA is large and critical to probe, because errors in their values could influence total binding

energy (Woods, Malaisree, Hannongbua, & Mulholland, 2011; Woods et al., 2014; Ahmad, Raza,

Uddin, Rungrotmongkol, et al., 2018) . Second, the implicit water model in MM/GBSA and

MM/PBSA skips details of molecular interaction between protein and water, and ligand and water

(Woods et al., 2014; Ahmad, Raza, Uddin, Rungrotmongkol, et al., 2018; (Ahmad, Raza, Abbasi,

& Azam, 2018). This holds vital importance in particular where water act as a bridge between the

protein and ligand (Woods et al., 2014). These shortcomings are overcome in WaterSwap that uses

explicit water system. The total binding free energy calculated for the complex by WaterSwap is

-43.2 kcal/mol, which is an average of the four estimates: BAR (-44.2 kcal/mol), FEP (-42.6

kcal/mol), TI (-43.0 kcal/mol) and Quadrature-based integration of TI (-43.3 kcal/mol). It can be

seen in Table 8.2 that the simulation is well converged as illustrated by almost the same estimates

of the four different methods. It is important to note that WaterSwap sometimes overestimate the

true value typically by 15 to 20 kcal/mol and care should be taken while interpreting the results.

309

Table 8.2. WaterSwap estimation of absolute binding free energy for the MurC-inhibitor complex.

Method Binding free energy

(kcal/mol)

Standard deviation

(kcal/mol)

BAR -44.2 0.05

FEP -42.6 13.9

TI -43.0 2.4

Quadrature-based

integration of TI

-43.3 -

Absolute binding free

energy

-43.2 0.58

8.7. Conclusions

Each year, A. baumannii causes 12,000 infections in the United States, 500 out of which lead to

death. In intensive care units, carbapenem-resistant A. baumannii is reported with crude mortality

of 26–76%. Due to resistance to majority of the clinically used antibiotics, the WHO ranked this

bacterium as top priority bacterial pathogen that urgently required the development of new

antibiotics. MurC ligase enzyme, which catalyzes the first step of PG biosynthesis in the

cytoplasm, is an attractive drug target. The LB domain of this enzyme is conserved among the Mur

ligase family and is not explored: thus could be an appealing target to inhibit the catalytic

mechanism of the mentioned enzyme. In the present study, structure based virtual screening was

performed that shortlisted a natural drug-like and lead-like compound that exhibits high affinity

for the MurC LB domain. Additionally, the inhibitor interacts with the conserved residues of the

ATP-binding domain making it a dual domain inhibitor. From pharmaceutics perspectives, the

inhibitor has a save therapeutic profile, while MD simulations unraveled the high protein stability

with the inhibitor well placed at the docked site. RDF and AFD analysis reflected the critical role

of Asp334 in inhibitor binding and stability at the LB domain. For validated docking and

simulation results, free energies for the complex were estimated using three popular methods:

MM/GBSA, MM/PBSA, and WaterSwap all of which supported the high complex stability. This

screened natural inhibitor could be further subjected to chemical substitutions to enhance its

310

affinity for the target site.


S-Table 8.1. Structure evaluation of the predicted MurC models.

S-Table 8.2. Structure and docking scores for the shortlisted Top-10 inhibitors.

S-Table 8.3. Hydrogen bond analysis for complex at frequency of 5-ns.

S-Fig.8.1. A. Secondary structure of the modeled MurC and B. Ramachandran plot for the

predicted MurC enzyme.

S-Fig.8.2. Extended RMSD of 150-ns for MurC enzyme.

S-Fig.8.3-S-Fig.8.7. Superimposition of complex at frequency of 5-ns.

S-Fig.8.8. Hot Spot residues in MM/GBSA (A) and MM/PBSA (B). Energy values are in kcal/mol.

S-Fig.8.9. Binding free energy decomposition of hot spot amino acids into its components.

8.9. References



221, 507–517.



Ahmad, S., Raza, S., Abbasi, S. W., & Azam, S. S. (2018). Identification of natural inhibitors

against Acinetobacter baumannii D-alanine-D-alanine ligase enzyme: A multi-spectrum in

silico approach. Journal of Molecular Liquids, 262, 460–475.

Ahmad, S., Raza, S., Abro, A., Liedl, K. R., & Azam, S. S. (2018). Towards novel inhibitors

against KdsB: A highly specific and selective broad-spectrum bacterial enzyme. Journal of

Biomolecular Structure and Dynamics, (just-accepted), 1–47.



Journal of Molecular Graphics and Modelling.

311

Ahmad, S., Raza, S., Uddin, R., & Azam, S. S. (2018). Comparative subtractive proteomics based

ranking for antibiotic targets against the dirtiest superbug: Acinetobacter baumannii. Journal

of Molecular Graphics and Modelling, 82, 74–92.

Ahmad, S., Raza, S., Uddin, R., Rungrotmongkol, T., Azam, S. S., & others. (2018). From

phylogeny to protein dynamics: A computational hierarchical quest for potent drug

identification against an emerging enteropathogen “Yersinia enterocolitica.” Journal of

Molecular Liquids.




Ambure, P., Bhat, J., Puzyn, T., & Roy, K. (2018). Identifying natural compounds as multi-target-

directed ligands against Alzheimer’s disease: an in silico approach. Journal of Biomolecular

Structure and Dynamics, 1–25.




79651–79661.





81(8), 3684–3690.

Bertrand, J. A., Auger, G., Fanchon, E., Martin, L., Blanot, D., van Heijenoort, J., & Dideberg, O.

(1997). Crystal structure of UDP-N-acetylmuramoyl-L-alanine: D-glutamate ligase from

Escherichia coli. The EMBO Journal, 16(12), 3416–3425.

Binda, E., Marinelli, F., & Marcone, G. L. (2014). Old and new glycopeptide antibiotics: action

and resistance. Antibiotics, 3(4), 572–594.

Breda, A., Valadares, N. F., de Souza, O. N., & Garratt, R. C. (2007). Protein structure, modelling

312

and applications.



Cha, S. S., An, Y. J., Jeong, C.S., Yu, J. H., & Chung, K. M. (2014). ATP-binding mode including

a carbamoylated lysine and two Mg 2+ ions, and substrate-binding mode in Acinetobacter

baumannii MurF. Biochemical and Biophysical Research Communications, 450(2), 1045–

1050.

Cheng, M., Huang, J. X., Ramu, S., Butler, M. S., & Cooper, M. A. (2014). Ramoplanin at

bactericidal concentrations induces bacterial membrane depolarization in Staphylococcus

aureus. Antimicrobial Agents and Chemotherapy, 58(11), 6819–6827.



Consortium, U., & others. (2017). UniProt: the universal protein knowledgebase. Nucleic Acids

Research, 45(D1), D158-D169.

Cournia, Z., Allen, B., & Sherman, W. (2017). Relative binding free energy calculations in drug

discovery: recent advances and practical considerations. Journal of Chemical Information

and Modeling, 57(12), 2911–2937.





Proceedings of the National Academy of Sciences, 40(6), 377–381.

Dorn, M., Buriol, L. S., & Lamb, L. C. (2013). A molecular dynamics and knowledge-based

computational strategy to predict native-like structures of polypeptides. Expert Systems with

Applications, 40(2), 698–706.

Dorn, M., e Silva, M. B., Buriol, L. S., & Lamb, L. C. (2014). Three-dimensional protein structure

prediction: Methods and computational strategies. Computational Biology and Chemistry, 53,

313

251–276.

Egan, W. J., Merz, K. M., & Baldwin, J. J. (2000). Prediction of drug absorption using multivariate

statistics. Journal of Medicinal Chemistry, 43(21), 3867–3877.



Fauci, A. S., & Morens, D. M. (2012). The perpetual challenge of infectious diseases. New England

Journal of Medicine, 366(5), 454–461.

Ferreira, L. G., dos Santos, R. N., Oliva, G., & Andricopulo, A. D. (2015). Molecular docking and

structure-based drug design strategies. Molecules, 20(7), 13384–13421.

Fiuza, M., Canova, M. J., Patin, D., Letek, M., Zanella-Cleon, I., Becchi, M., … Gil, J. A. (2008).

The MurC ligase essential for peptidoglycan biosynthesis is regulated by the serine/threonine

protein kinase PknA in Corynebacterium glutamicum. Journal of Biological Chemistry,

283(52), 36553–36563.

Gadebusch, H. H., Stapley, E. O., & Zimmerman, S. B. (1992). The discovery of cell wall active

antibacterial antibiotics. Critical Reviews in Biotechnology, 12(3), 225–243.



Ghose, A. K., Viswanadhan, V. N., & Wendoloski, J. J. (1999). A knowledge-based approach in

designing combinatorial or medicinal chemistry libraries for drug discovery. A qualitative

and quantitative characterization of known drug databases. Journal of Combinatorial

Chemistry, 1(1), 55–68.

Gordon, E., Flouret, B., Chantalat, L., van Heijenoort, J., Mengin-Lecreulx, D., & Dideberg, O.

(2000). Crystal structure of UDP-N-acetylmuramoyl-L-alanyl-D-glutamate: meso-

diaminopimelate ligase from Escherichia coli. Journal of Biological Chemistry.

Gundert-Remy, U., & Stahlmann, R. (n.d.). The blood-brain barrier in toxicology.



314



143–152.

Hosea, N. A., & Jones, H. M. (2013). Predicting pharmacokinetic profiles using in silico derived

parameters. Molecular Pharmaceutics, 10(4), 1207–1215.



AABC, 8, 37.

Humnabadkar, V., Prabhakar, K. R., Narayan, A., Sharma, S., Guptha, S., Manjrekar, P., … others.

(2014). UDP-N-Acetylmuramic Acid L-Alanine ligase (MurC) inhibition in E. coli tolC-leads

to cell death. Antimicrobial Agents and Chemotherapy, AAC-02890.

Janzen, W. P. (2014). Screening technologies for small molecule discovery: the state of the art.

Chemistry & Biology, 21(9), 1162–1170.



Kadam, R. U., & Roy, N. (2007). Recent trends in drug-likeness prediction: A comprehensive

review of In silico methods. Indian Journal of Pharmaceutical Sciences, 69(5), 609.

Karplus, M., & McCammon, J. A. (2002). Molecular dynamics simulations of biomolecules.

Nature Structural and Molecular Biology, 9(9), 646.



measurement of binding free energy for 7-azaindole derivatives to glycogen synthase kinase-

3β. Journal of Chemical Information and Modeling, 54(6), 1653–1660.

Kitchen, D. B., Decornez, H., Furr, J. R., & Bajorath, J. (2004). Docking and scoring in virtual

screening for drug discovery: methods and applications. Nature Reviews Drug Discovery,

3(11), 935.

KONG, K.F., Schneper, L., & Mathee, K. (2010). Beta-lactam antibiotics: from antibiosis to

315

resistance and bacteriology. Apmis, 118(1), 1–36.

Korb, O., Stutzle, T., & Exner, T. E. (2009). Empirical scoring functions for advanced protein-

ligand docking with PLANTS. Journal of Chemical Information and Modeling, 49(1), 84–

96.

Kouidmi, I., Levesque, R. C., & Paradis-Bleau, C. (2014). The biology of Mur ligases as an

antibacterial target. Molecular Microbiology, 94(2), 242–253.

Krautler, V., Van Gunsteren, W. F., & Hünenberger, P. H. (2001). A fast SHAKE algorithm to



Lin, J. H., & Yamazaki, M. (2003). Role of P-glycoprotein in pharmacokinetics. Clinical

Pharmacokinetics, 42(1), 59–98.



Livermore, D. M. (2009). Has the era of untreatable infections arrived? Journal of Antimicrobial

Chemotherapy, 64(suppl_1), i29-i36.

Lobo, I. (2008). Basic local alignment search tool (BLAST). Nature Education, 1(1).

Meng, X. Y., Zhang, H.-X., Mezei, M., & Cui, M. (2011). Molecular docking: a powerful approach

for structure-based drug discovery. Current Computer-Aided Drug Design, 7(2), 146–157.

Mesleh, M. F., Rajaratnam, P., Conrad, M., Chandrasekaran, V., Liu, C. M., Pandya, B. A., …

others. (2016). Targeting bacterial cell wall peptidoglycan synthesis by inhibition of

glycosyltransferase activity. Chemical Biology & Drug Design, 87(2), 190–199.




Mol, C. D., Brooun, A., Dougan, D. R., Hilgers, M. T., Tari, L. W., Wijnands, R. A., … Swanson,

R. V. (2003). Crystal structures of active fully assembled substrate-and product-bound

complexes of UDP-N-acetylmuramic acid: L-alanine ligase (MurC) from Haemophilus

316

influenzae. Journal of Bacteriology, 185(14), 4152–4162.

Mooij, W. T. M., & Verdonk, M. L. (2005). General and targeted statistical potentials for protein-

-ligand interactions. Proteins: Structure, Function, and Bioinformatics, 61(2), 272–287.


365–382.

Muegge, I., Heald, S. L., & Brittelli, D. (2001). Simple selection criteria for drug-like chemical

matter. Journal of Medicinal Chemistry, 44(12), 1841–1846.

N Sangshetti, J., S Joshi, S., H Patil, R., G Moloney, M., & B Shinde, D. (2017). Mur Ligase

Inhibitors as Anti-bacterials: A Comprehensive Review. Current Pharmaceutical Design,

23(21), 3164–3196.

Nebert, D. W., Wikvall, K., & Miller, W. L. (2013). Human cytochromes P450 in health and

disease. Phil. Trans. R. Soc. B, 368(1612), 20120431.



Pastor, R. W., Brooks, B. R., & Szabo, A. (1988). An analysis of the accuracy of Langevin and

molecular dynamics algorithms. Molecular Physics, 65(6), 1409–1419.






Perdih, A., Hrast, M., Barreteau, H., Gobec, S., Wolber, G., & Solmajer, T. (2014). Benzene-1, 3-

dicarboxylic acid 2, 5-dimethylpyrrole derivatives as multiple inhibitors of bacterial Mur

ligases (MurC-MurF). Bioorganic & Medicinal Chemistry, 22(15), 4124–4134.

Perdih, A., Hrast, M., Barreteau, H., Gobec, S., Wolber, G., & Solmajer, T. (2014). Inhibitor design

strategy based on an enzyme structural flexibility: a case of bacterial MurD ligase. Journal of

Chemical Information and Modeling, 54(5), 1451–1466.

317

Perdih, A., Kovac, A., Wolber, G., Blanot, D., Gobec, S., & Solmajer, T. (2009). Discovery of

novel benzene 1, 3-dicarboxylic acid inhibitors of bacterial MurD and MurE ligases by

structure-based virtual screening approach. Bioorganic & Medicinal Chemistry Letters,

19(10), 2668–2673.

Pires, D. E. V, Blundell, T. L., & Ascher, D. B. (2015). pkCSM: predicting small-molecule

pharmacokinetic and toxicity properties using graph-based signatures. Journal of Medicinal

Chemistry, 58(9), 4066–4072.

Raza, S., & Azam, S. S. (2018). AFD: an application for bi-molecular interaction using axial

frequency distribution. Journal of Molecular Modeling, 24(4), 84.






3084–3095.

Salomon-Ferrer, R., Case, D. A., & Walker, R. C. (2013). An overview of the Amber biomolecular

simulation package. Wiley Interdisciplinary Reviews: Computational Molecular Science,

3(2), 198–210.

Savjani, K. T., Gajjar, A. K., & Savjani, J. K. (2012). Drug solubility: importance and enhancement

techniques. ISRN Pharmaceutics, 2012.

Schneider, T., & Sahl, H.-G. (2010). An oldie but a goodie-cell wall biosynthesis as antibiotic

target pathway. International Journal of Medical Microbiology, 300(2), 161–169.

Silver, L. L. (2011). Challenges of antibacterial discovery. Clinical Microbiology Reviews, 24(1),

71–109.

Simcic, M., Sosic, I., Hodosček, M., Barreteau, H., Blanot, D., Gobec, S., & Grdadolnik, S. G.

(2012). The binding mode of second-generation sulfonamide inhibitors of MurD: clues for

rational design of potent MurD inhibitors. PloS One, 7(12), e52817.

318


molecules against prioritised targets of multi-drug-resistant-Acinetobacter baumannii

insights from molecular docking, molecular dynamic simulations and in vitro assays. Journal

of Biomolecular Structure and Dynamics, 1–24.

Spławinski, J., Kuzniar, J., Filipiak, K., & Zielinski, W. (2006). Evaluation of drug toxicity in

clinical trials. Science and Engineering Ethics, 12(1), 139–145.

Štrancar, K., Boniface, A., Blanot, D., & Gobec, S. (2007). Phosphinate inhibitors of UDP-N-

acetylmuramoyl-L-alanyl-D-glutamate: L-lysine ligase (MurE). Archiv Der Pharmazie: An

International Journal Pharmaceutical and Medicinal Chemistry, 340(3), 127–134.

Teague, S. J., Davis, A. M., Leeson, P. D., & Oprea, T. (1999). The design of leadlike

combinatorial libraries. Angewandte Chemie International Edition, 38(24), 3743–3748.

Tomasic, T., Sink, R., Zidar, N., Fic, A., Contreras-Martel, C., Dessen, A., … others. (2012). Dual

inhibitor of MurD and MurE ligases from Escherichia coli and Staphylococcus aureus. ACS


Tozer, T. N., & Rowland, M. (2006). Introduction to pharmacokinetics and pharmacodynamics:

the quantitative basis of drug therapy. Lippincott Williams & Wilkins.




Uline, M. J., & Corti, D. S. (2013). Molecular dynamics at constant pressure: allowing the system

to control volume fluctuations via a “shell” particle. Entropy, 15(9), 3941–3969.

Veber, D. F., Johnson, S. R., Cheng, H.-Y., Smith, B. R., Ward, K. W., & Kopple, K. D. (2002).

Molecular properties that influence the oral bioavailability of drug candidates. Journal of


Verdonk, M. L., Cole, J. C., Hartshorn, M. J., Murray, C. W., & Taylor, R. D. (2003). Improved

protein--ligand docking using GOLD. Proteins: Structure, Function, and Bioinformatics,

52(4), 609–623.

319



Acinetobacter baumannii. Journal of Biomolecular Structure and Dynamics, 36(5), 1182–

1194.

Vistoli, G., Pedretti, A., & Testa, B. (2008). Assessing drug-likeness-what are we missing? Drug


Vollmer, W., Blanot, D., & De Pedro, M. A. (2008). Peptidoglycan structure and architecture.

FEMS Microbiology Reviews, 32(2), 149–167.

Vyas, V. K., Ukawala, R. D., Ghate, M., & Chintha, C. (2012). Homology modeling a fast tool for

drug discovery: current perspectives. Indian Journal of Pharmaceutical Sciences, 74(1), 1.




55(1), 19–32.

Wang, G., & Zhu, W. (2016). Molecular docking for drug discovery and development: A widely

used approach but far from perfect. Future Science. 1707-1710.

Wang, J., Wolf, R. M., Caldwell, J. W., Kollman, P. A., & Case, D. A. (2004). Development and

testing of a general amber force field. Journal of Computational Chemistry, 25(9), 1157–

1174.



217–230.



W407-W410.



Modeling, 45(1), 160–169.

320



of Chemical Physics, 134(5), 02B611.


Computational assay of H7N9 influenza neuraminidase reveals R292K mutation reduces drug

binding affinity. Scientific Reports, 3, 3561.

Woods, C. J., Malaisree, M., Michel, J., Long, B., McIntosh-Smith, S., & Mulholland, A. J. (2014).

Rapid decomposition and visualisation of protein--ligand binding free energies by residue

and by water. Faraday Discussions, 169, 477–499.

Yan, Y., Munshi, S., Leiting, B., Anderson, M. S., Chrzas, J., & Chen, Z. (2000). Crystal structure

of Escherichia coli UDPMurNAc-tripeptide d-alanyl-d-alanine-adding enzyme (MurF) at 2.3

Å resolution. Journal of Molecular Biology, 304(3), 435–445.

Yu, H., Zou, B., Wang, X., & Li, M. (2016). Investigation of miscellaneous hERG inhibition in

large diverse compound collection using automated patch-clamp assay. Acta Pharmacologica

Sinica, 37(1), 111.

321

Chapter # 9

Binding Mode Analysis, Dynamic Simulation and Binding

Free Energy Calculations of the Murf Ligase from

Acinetobacter baumannii

9.1. Abstract

MurF ligase catalyzes the final cytoplasmic step of bacterial peptidoglycan biosynthesis and, as

such, is a validated target for therapeutic intervention. Herein, we performed molecular docking

to identify putative inhibitors of A. baumannii MurF (AbMurF). Based on comparative docking

analysis, compound 114 (ethyl pyridine substituted 3-cyanothiophene) was predicted to potentially

be the most active ligand. Computational pharmacokinetic characterization of drug-likeness of the

compound showed it to fulfil all the parameters of Muegge and the MDDR rule. A MD simulation

of 114 indicated the complex to be stable on the basis of an average RMSD value of 2.09 Å for

the ligand. The stability of the complex was further supported by RMSF, β-factor and Rg values.

Analyzing the complex using RDF and a novel analytical tool termed the AFD illustrated that after

simulation the ligand is positioned in close vicinity of the protein active site where Thr42 and

Asp43 participate in hydrogen bonding and stabilization of the complex. Binding free energy

calculations based on the Poisson-Boltzmann or Generalized–Born Surface Area Continuum

Solvation (MM(PB/GB)SA) method indicated the van der Waals contribution to the overall

binding energy of the complex to be dominant along with electrostatic contributions involving the

hot spot amino acids from the protein active site. The present results indicate that the screened

compound 114 may act as a parent structure for designing potent derivatives against AbMurF in

specific and MurF of other bacterial pathogens in general.

322

9.2. Introduction

A. baumannii is a Gram-negative bacterium and opportunistic human pathogen responsible for an

increasing number of infections, particularly in a hospital milieu (McConnell, Actis, & Pachon,

2013). Extensive overuse and misuse of broad-spectrum antibiotics stimulated the intrinsic

resistance mechanisms of this organism that, in combination with the potential of acquiring foreign

virulent determinants, lead to this superbug being classified as a “red alert” pathogen (Howard,

O’Donoghue, Feeney, & Sleator, 2012). Considering the rise of serious MDR strains of A.

baumannii and a decline in the efficacy of most first-line antibiotics, novel approaches to identify

new targets are required to treat A. baumannii associated infections.

Peptidoglycan (PG), an essential component of the bacterial cell wall, is a target of choice for

several antibacterial drugs (Gu et al., 2004). The main function of PG is to ensure cell integrity in

the hypotonic environment and provide the flexibility and strength required by the cell to divide

and grow (Typas, Banzhaf, Gross, & Vollmer, 2012). Structurally, PG is a mesh-like structure of

long crosslinked glycan polymers made of disaccharide units ((N-acetylmuramic acid (MurNAc)

and N-acetylglucosamine (GlcNAc) in alternating fashion. MurNAc and GlcNAc are linked via β-

1, 4-glycosidic bonds, while the glycan polymers are attached via a short MurNAc linked peptide.

The peptide is composed of D-glutamic acid, D-alanine, L-alanine and either meso-diamino-pimelic

acid (m-DAP) (in case of Gram-negative bacteria) or L-lysine (in case of Gram-positive bacteria)

(Barreteau et al., 2008). PG formation is a multifaceted phenomenon and can be divided into three

important phases: the initial intracellular phase involves the assembly of uridine diphosphate

GlcNAc (UDP- GlcNAc) from fructose-6-phosphate and uridine diphosphate MurNAc (UDP-

MurNAc) from UDP- GlcNAc to serve as precursor nucleotides for GlcNAc and MurNAc,

respectively. The attachment of the short peptide to UDP-MurNAc is done in a sequential manner

by enzymes of Mur ligase family. Briefly, the addition of L-alanine to D-lactoyl group of UDP-

MurNAc is done by MurC followed by D-glutamic acid, L-alanine or meso-diamino-pimelic acid

addition under the supervision of MurD. Finally, the catalysis of D-alanine-D-alanine to previously

added amino acids is done by MurF to produce the final product of UDP-MurNAc-pentapeptide

(Park’s nucleotide). The first phase is followed by a translocation phase during which a Park’s

nucleotide is translocated across the cytoplasmic membrane and in the final phase is incorporated

in the growing chain of the glycan polymer (Smith, 2006; Barreteau et al., 2008).

323

In the past several years, Mur ligases appear to be the center of antibacterial therapy research

(Schneider & Sahl, 2010; Bugg, Braddick, Dowson, & Roper, 2011). The Mur family of enzymes

have similar reaction mechanisms, 3D structures and catalyze intracellular ATP-driven reactions

(Barreteau et al., 2008). MurF, responsible for the catalysis of the final cytoplasmic step of PG

biosynthesis, is reported to be an attractive target for several reasons (Hrast et al., 2014). First,

MurF is essential for the survival of bacteria (Lugtenberg & Van Schijndel-van Dam, 1972).

Secondly, MurF is extensively conserved in both Gram-positive and Gram-negative bacteria

suggesting the potential for designing broad-spectrum drugs (El Zoeiby, Sanschagrin, & Levesque,

2003). Third, MurF function seems to be necessary for resistance against β-lactam antibiotics

(Sobral et al., 2003). Lastly, interfering with MurF function disturbs the replication ability of

bacteria (El Zoeiby, Sanschagrin, & Levesque, 2003). Several kinds of MurF inhibitors have been

proposed. The first among these were pseudo-tripeptides and pseudo-tetrapeptides of aminoalkyl

phosphinate followed by sulfonamide inhibitors introduced by Abbott Laboratories through

affinity selection screening (Sobral et al., 2003; Comess et al., 2006). Very shortly, structure-based

MurF inhibitor optimization lead to a compound that exhibits an IC50 of 22 nM and contains

morpholino sulphonomide and dichlorophenyl rings (Stamper et al., 2006). A series of thiazolyl

aminopyrimidine compounds using muropeptide ligand based assay were then identified with an

IC50 as low as 2.5 µM (Baum et al., 2006). Through a MurF binding assay, 8-hydroxyquinolines

inhibitors with the ability to bind and inhibit MurF activity in E. coli were reported (Baum et al.,

2007). Based on structure-based virtual screening, a novel inhibitor against S. pneumoniae MurF

was revealed (Turk et al., 2009). Inhibitors with modest potential against E.coli MurF belongs to

structurally diverse classes and includes 4-phenylpiperidine derivatives and diaryquinolines

having minimum inhibitory concentration (MIC) values in the range of 8-16 µg/ml (Kerwin,

2010). Similarly, employing ligand-based virtual screening, a new inhibitor with inhibitory

capacity in micromolar against E.coli and S. pneumoniae was identified (Baum et al., 2009). A

series of cyanothiophene based inhibitors with inhibitory potential were designed and found to

have an increased polarity when compared with the first generation of inhibitors (Hrast et al.,

2014).

Considering the importance of AbMurF, we performed molecular docking analysis with reported

MurF inhibitors to characterize their potential mode of interaction with the protein. Results include

the predicted binding modes, fitness scores, and computational pharmacokinetic profile to predict

324

the in vivo fate of the best-docked ligand. Furthermore, mechanistic insights of the protein were

revealed using MD simulations. Lastly, binding free energies of the complex, protein, and ligands

were calculated using the MM(GB/PB)SA method implemented in AMBER. Overall, the study

could assist the design and development of potent drugs to inhibit MurF and potentially tackle the

superbug A. baumannii.


The complete schematic workflow of the current study is illustrated in Fig.9.1.

9.3.1. Receptor Protein Preparation

The 3D crystal complex of AbMurF protein was retrieved from the PDB (ID 4QDI) and subjected

to an initial preparation phase where all water and solutes in the structure were removed with the

exception of the Mg2+ ion, which was retained. The protein was then minimized using UCSF

Chimera (Pettersen et al., 2004) by assigning Gasteiger charges and relaxing the structure by

applying a 1500 step minimization using the TFF.

9.3.2. Ligands Search and Preparation

A total of one hundred and thirty-two putative inhibitors (S-Table 9.1) against MurF protein of

different bacterial species were identified from a thorough literature search (Miller, Hammond,

Bugg, & others, 1998;Comess et al., 2006; Stamper et al., 2006; Baum et al., 2006; Baum et al.,

2007; Turk et al., 2009; Baum et al., 2009; Sharma & Pan, 2012; Hrast et al., 2014). To utilize the

inhibitors in docking procedure, their 2D structures with standard bond angles and lengths were

drawn using ChemDraw Ultra (Kerwin, 2010). Finally, all the structures were energetically

minimized using ChemDraw Ultra.


Molecular docking is a computational-based procedure to predict non-covalent interactions of a

receptor and a ligand. The goal of the protocol is to predict bound conformations and determine

binding affinities of inhibitors (Morris & Lim-Wilby, 2008). Docking was done with AutoDock-

Vina (AD-Vina) and GOLD. Initially, docking of the one hundred and thirty-two inhibitors into

the active site of AbmurF receptor was carried out with AD-Vina on an Intel Core (TM) i5 CPU

M 540 @ 2.53 GHz with 32-bit Windows 8.1 as an operating system. The grid was set manually

to cover the active site with search space coordinates at the center along the X, Y and Z axis as -

29.60, 2.43 and -2.59 and dimensions (Å) along the X, Y and Z axis were set to 10 Å. Other

parameters were kept as the Vina default values. In the case of GOLD, all docking calculations

325

were accomplished with default genetic algorithm settings of GOLD 5.1 on Intel Xeon QuadTM

Core processor 3.0 GHz using Linux as a workstation. Default settings employed during docking

include standard parameters such as selection pressure (1.1), population size (100), island numbers

(1), size of niche (2), operation numbers (10,000), crossover (100), mutate (100), operator weights

for migrate (0) and number of docking poses for each ligand (10). GOLD uses four types of scoring

functions; GOLDScore, CHEMLP, ChemScore, and a user-defined scoring function to predict the

preferred orientation of docked compounds in the active site of a protein. In the present study,

poses of all the ligands were sorted on the basis of the GOLD fitness score. The binding site was

defined in such a way to cover all possible binding sites of the ligand and includes all the residues

or atoms that might be involved in ligand binding. For this reason, an appropriate binding sphere

was specified by selecting Thr42: C and all atoms lying within 10 Å of that residue were taken

into consideration. Docking results were analyzed through UCSF Chimera (Pettersen et al., 2004)

and VMD (Humphrey, Dalke, & Schulten, 1996). The interaction pattern of the best-docked

ligands into the target protein active site were visualized with the Discovery Studio (DS) Visualizer

(BIOvIA, 2015) and LigPlot+ (Laskowski & Swindells, 2011).

9.3.4. Computational Pharmacokinetics Evaluation

Computational pharmacokinetics of the best ligand was performed using two online tools:

preADMET (https://preadmet.bmdrc.kr/) (Lee et al., 2004) and SwissADMET

(http://www.swissadme.ch/) (Daina, Michielin, & Zoete, 2017).


The dynamic behavior of the docked complex of the top ranking compounds were studied using

MD simulations. The simulation protocol was performed using AMBER (Assisted model building

with energy refinement) while subsequent analysis was accomplished through its different

modules (Weiner & Kollman, 1981). Primary coordinates were from the docked complexes and

topology files were prepared using the tLEAP interface of AMBER12 (Weiner & Kollman, 1981).

Similarly, tLeap was also utilized for the addition of missing atoms. A total of 708 atoms were

added to the system including 66 heavy and 642 hydrogens. Specifically heavy atom side chains

of leucine, glutamine, lysine and aspartate and hydrogen to all residues side chains were added.

System solvation was executed with three-point transferable intermolecular potential (TIP3P)

water while the force fields employed for calculations include GAFF (Ozpinar, Peukert, & Clark,

2010), ff99SB and ff03.rl (Salomon-Ferrer, Case, & Walker, 2013). The docked protein system

http://www.swissadme.ch/)%20%5b29%5d.

326

was neutralized by adding 11 Na+ ions while the ligand atom types and parameters were corrected

through antechamber program. To remove steric clashes, minimization of the docked protein

complex was performed by applying 1500 steepest descent and 1000 conjugate gradient steps. For

heating the system, Langevin dynamics was employed for 10 ps while system equilibration was

done for 100 ps (Pastor, Brooks, & Szabo, 1988) in the canonical (NVT) ensemble. In the

production run, SHAKE algorithm was deployed to restrain lengths of covalent bonds involving

hydrogens (Berendsen, Postma, van Gunsteren, DiNola, & Haak, 1984). The production run was

performed for a total of 90 ns. Periodic boundary conditions were used for all minimizations and

simulations with the system size set to be a minimum of 8 Å from the protein while volume,

temperature (300K) and pressure (1 atm) were maintained constant (Weiner & Kollman, 1981).

The generated trajectories were analyzed through the AMBER PTRAJ module. Analysis included

RMSD, RMSF, β-factor, Rg and RDF. For graphical representation and analytical evaluation of

these quantities, Xmgrace was used (Vaught, 1996).

9.3.6. AFD

AFD was performed to highlight significant interactions using coordination geometry of the ligand

with respect to the protein. AFD is a sensitive tool for determining local reorganization of ligand

atoms plotted on XY plane while considering protein atom as a point of reference (Azam, Abro,

& Raza, 2015; Abbasi, Raza, Azam, Liedl, & Fuchs, 2016). The generated graph is in the form of

3D histogram and the axial distribution of atoms is represented using the following equation

………………………………………………………………....(XXXVIII)

where, i and j represent the distribution of the ligand atom coordinates on the X and Y plane,

respectively, with respect to the reference atom of the protein. The cutoff values are designated by

k and l values on X and Y axes relative to the reference atom of the protein, respectively. Lastly,

mi, j is number of observations the fall in the (i j) coordinates.

9.3.7. MM(PB/GB)SA Analysis

In order to calculate binding free energies of the system, trajectories of MD simulation were

subjected to MM(PB/GB)SA package of AMBER12 (Massova & Kollman, 2000).

MM(PB/GB)SA calculates free energies by taking into account the free energy difference between

complex and that of the protein and ligand alone. Analysis was done individual using both the

327

Poisson-Boltzmann (PB) or Generalized-Born (GB) approaches. A total of 450 frames each after

0.2 ns were extracted from the complete MD trajectory and subjected to MM(PB)SA calculation

using MMPBSA.py module of AMBER12 (Miller et al., 2012). The hot spot amino acids

contributing significantly in binding free energy can provide a better insight into components

involved in the interaction between a protein and its ligand. The binding free energy was therefore

decomposed into per residue and pair residues using MMPBSA.py module of AMBER12.

Fig.9.1. The schematic workflow illustrating complete hierarchy of docked protein analysis.

328


9.4.1. Comparative Molecular Docking Analysis

Through a comprehensive subtractive proteomic approach, MurF protein has been characterized

as a human-non-homolog, non-paralog, essential for survival, involved in peptidoglycan

biosynthesis, drugabble and localized intracellularly by our group members in Computational

Biology Lab, National Center of Bioinformatics, Quaid-i-Azam University, Pakistan. Hence,

targeting such a protein will help in designing potent leads that may block the peptidoglycan

synthesis pathway and ultimately inhibit or retard the bacterial growth. The availability of the

crystallographic 3D structure of AbMurF provided a platform to target this protein using molecular

docking and subsequent detailed analysis of the top ligand with respect to its mode of binding in

the active site of the protein.

Crystal structures of MurF from S. pneumoniae, E. coli and Thermotoga maritima are available

(Yan et al., 2000; Longenecker et al., 2005; Favini-Stabile, Contreras-Martel, Thielens, & Dessen,

2013; Turk et al., 2013) and shared identity of 29%, 41% and 32% identity respectively with

AbMurF. The crystal structures of all these MurF variants were compared with that of AbMurF (S-

Fig.9.1). The MurF crystal structure from S. pneumoniae revealed a closed conformation where

the three domains are rolled up. In such a conformation, the C-terminal domain is connected with

both central and N-terminal domain (Longenecker et al., 2005). The ATP binding site is collapsed,

pointing to the fact that the closed conformation is a non-active conformation. In contrast, the

crystal structure of MurF from E. coli revealed a more open conformation where all three domains

are observed to a have linear arrangement forming a crescent-like conformation (Yan et al., 2000).

The MurF from T, maritima (Favini-Stabile, Contreras-Martel, Thielens, & Dessen, 2013) is in a

somewhat intermediate conformation that lies between the open and closed states. Structural

comparison of S. pneumoniae, E. coli and T. maritima with AbMurF demonstrated that the

observed differences among MurF structures are because of variations in the relative positions at

the central domain. Few direct interactions between domains through a connecting loop were

observed in the E. coli apo MurF structure; however, the other structures revealed extensive

contacts between the central and C-terminal domains. The flexibility of the hinge between the

central and C-terminal domains is supported by the interdomain interactions and the flexible

connection loop (Cha, An, Jeong, Yu, & Chung, 2014).

329

Binding of compounds to the protein active site was calculated in term of the binding energy when

using AD-Vina (Trott & Olson, 2010) while for GOLD it was represented by the fitness score

(Jones, Willett, Glen, Leach, & Taylor, 1997). The correlation coefficient between the GOLD

fitness score and AD- Vina binding energies was first calculated to investigate the strength and

direction of the relationship. A strong negative correlation of R2 = x was observed between the

two scoring methods (Fig.9.2) indicating that the methods were ranking the compounds in a similar

manner. To filter the best-docked compounds, a cut-off value for binding energy higher than -7

kcal/mol and GOLD fitness score of more than 60 was defined. The 2D structures of the top 5

compounds is presented in Table 9.1. Based on comparative analysis, Compound 114 was

identified as the top-ranked ligand with a binding energy of -9.6 kcal/mol and fitness score of 69.5

(S-Table 9.2).

Fig.9.2. Correlation coefficient between GOLD fitness scores and AD-Vina binding energies.

330

Table 9.1. Top 5 best ranked and active compounds against AbMurF.

S.

No

Ligands Structure GOLD

Score

Autodock Vina

Binding Energy

(Kcal/mol)

1. Compound

114

69.5 -9.6

2. Compound

132

68.4 -7.4

3. Compound

116

65.7 -6.5

4. Compound

8

65.3 -6.3

5. Compound

68

65.2 -6.9

331

The chemical nature of the top ligand is an ethyl pyridine substituted 3-cyanothiophene derived

from 3-cyanothiophene. According to literature, the compound demonstrated good potency against

MurF with IC50 value 6 µM (Stamper et al., 2006). The potency of this compound is based on 3-

cyanothiophene and ethyl pyridine ring attached with the main ring on the 5th position. The core

interactions with active site residues are mainly because of 3-cyanothiophene while the enhanced

activity of this compound could be assumed due to substitution on the 5 position of the 3-

cyanothiophene ring. The ethyl pyrimidine ring allows the compound to bind deeply in the binding

pocket along the hydrophobic surface of the active cavity. Compounds of this series exhibit

consistent binding modes and underwent multiple phenyl substitution of the saturated ring to

develop more potent inhibitors (Stamper et al., 2006).

The complex of the top compound was thoroughly inspected using UCSF Chimera

(Pettersen et al., 2004), VMD (Humphrey, Dalke, & Schulten, 1996), Discovery Studio (DS)

(BIOvIA, 2015) and Ligplot+ (Laskowski & Swindells, 2011) for interpretation of the binding pose

and interactions. Among the binding interactions, hydrogen bonding plays a crucial role in the

proposed inhibition mechanism by stabilizing the protein-ligand complex through specific

important interactions between the protein and interacting ligand (Hubbard & Kamran Haider,

2010).

Fig.9.3. 2D depiction of docked compound 114 into the active site of AbMurF.

332

Docked complex of Compound 114 with the AbMurF provided by Vina and GOLD was visualized

for interactions using DS (Fig.9.3). The ethyl pyridine ring of 114 is located deep inside the

binding cavity of the receptor protein, making an important contribution to binding. The pyridine

ring was pulled towards the floor of the active site to maximize hydrogen bonding between

nitrogen and Ile40 of the protein active site. Similarly, the morpholine ring of the inhibitor was

posed in an inverted form towards the floor interacting with the cavity residues through its oxygen

(S-Fig.9.2A). In order to have an insight into the active site residues involved in interactions with

the ligand, residues lying within 3 Å of the ligand were identified. The ligand was found to form

interactions Leu11, Arg39, Ile40, Leu41, Thr42, Asp43, Ser44, His46, Leu56, Phe61, Ala63,

Phe66, Met157, Leu160, and Glu161. Ligplot analysis illuminates hydrogen bonding between the

backbone of Ile40 and the ligand at a distance of 2.97 Å, while multiple hydrophobic interactions

were reported involving Leu11, Arg39, Leu41, Thr42, Asp43, Leu56, Phe61, Ala63, Phe66,

Leu160, Met157, Glu161 and ligand (S-Fig.9.2B). The most important residues involved in

hydrogen bonding with the ligand was Thr42 and Asp43 (Table 9.2). The analysis indicated that

all these interactions played very critical role in stabilizing the docked complex and can guide

further drug development.

Table 9.2. Residues of target protein involved in hydrogen, hydrophobic and ionic interactions

with the ligand

Hydrogen Bonding Hydrophobic Interactions

Ligand Protein residues Distance Ligand Protein Distance

LIG:S1 THR42:H 2.8 LIG: C1 LEU41: CA 3.6

LIG:N1 THR42:H 3.7 LIG: C1 LEU41: CB 3.1

LIG:N1 ASP43:H 2.5 LIG: C2 LEU41: CA 3.5

LIG:N3 ASP43:H 3.9 LIG: C2 LEU41: CB 3.6

LIG:H2 THR42:N 3.3 LIG: C3 LEU41: CA 3.8

LIG:H2 ASP43:N 2.7 LIG: C3 LEU41: CB 3.2

LIG:S1 THR42: HG1 3.0 LIG: C4 LEU41: CA 3.8

LIG:N1 THR42: HG1 3.1 LIG: C4 LEU41: CB 3.7

LIG:O1 THR42: HG1 2.9 LIG: S1 LEU41: CA 3.7

LIG:N3 HIS46: HD1 3.0 LIG: C5 LEU41: CB 3.5

LIG:H2 THR42: OG1 3.1 LIG: C5 LEU41: CD1 3.9

333

LIG: C5 ASP43: CB 3.8

LIG:C6 MET157:CE 3.8

LIG:C8 ASP43:CB 3.6

LIG:C8 ASP43:CG 3.8

LIG:C9 ASP43:CB 3.9

LIG:C9 ASP43:CG 3.7

LIG:C12 THR42:CB 3.7

There is high variation in interacting residues of MurF in different species. Previous work done on

various MurF belonging to different species yielded different interacting residues. Still, the active

site cavity of MurF ligase enzymes is generally the same according to their tertiary structure and

the compounds are seen to be interacting in the same cavity. Compounds 37 (4- phenyl piperidine)

and 16 (Diaryquinolines) are reported inhibitors of MurF; however, their fitness score and binding

energy was found low in the current study. The Thr42 residue of AbMurF protein active has been

reported to involve in interaction with substrate (Cha, An, Jeong, Yu, & Chung, 2014. The other

reported important residue (Phe61) of AbMurF active sites was seen frequently in van der Waals

interactions with the ligand; however, the Arg97 interaction was not observed at all in the current

study.

As Mur ligase enzymes shared identical ATP binding site, it would be ideal to identify compounds

with multispectral activity targeting this site. MD simulation studies also suggested that such

broad-spectrum anti-MurF inhibitors are energetically favorable when compared compounds

bound in substrate binding site (Perdih et al., 2014;Perdih et al., 2015). In this context, future

characterization of the compounds in this study needs to involve docking in the ATP binding site

of Mur ligase enzymes and subjected to MD simulation assays to unveil if their mode of inhibition

may be occurring at this site.


In the drug development process the number of unsuccessful drug candidates from clinical studies

has increased profoundly in the past decade (Wadood et al., 2017). In silico estimation of

pharmacokinetics properties during the design stage have facilitated the selection of suitable drug

candidates for lead optimization. The compound was classified as a drug like by Muegge and

MDDR rule and completely fulfills both of these rules. According to computational

334

pharmacokinetics analysis, gastrointestinal absorption (GI) of the best-screened ligand was low

while its ability to penetrate the blood-brain barrier (BBB) is negligible. The compound is

classified as non-mutagenic, non-carcinogenic and has a low risk of hERG (Ether-a-go-go related

gene) inhibition. The high tendency of a drug to be retained for an extended duration in the blood

stream is an important property for its delivery to the target sites (Yang, Engkvist, Llinas, & Chen,

2012). This feature of the compound was evaluated for its binding efficacy to plasma protein

binding (PPB) or log k value. The high percentage of PPB and low log k value indicates that the

compounds will be retained in the blood stream for a high period of time thus maximizing its

availability to the target sites. The high value of PPB is because of the high log D value and high

molecular weight (Na-Bangchang, Ruengweerayut, Karbwang, Chauemung, & Hutchinson,

2007). Similarly, the absorption ability of the compound in the skin and the intestinal compartment

was measured through Sklog P and log D values (Van De Waterbeemd & Gifford, 2003). The high

Sklog P and log D values further demonstrate that the compound can be efficiently absorbed

through the skin and from intestinal compartments as compared to the gut where its log P value is

less. For the treatment of CNS infections of A. baumanii, the efficiency of the compound to cross

CNS barrier was computed on the basis of the molecular weight, TPSA, number of rotatable bonds

(NRBs) and log P values. It is believed that decrease in molecular weight, TPSA and NRBs values

are associated with increased CNS penetration. Based on this analysis the CNS penetration of the

compound is predicted to be low due to higher molecular and TPSA values. Similarly, it is also

noted that increase log P value decreases drug CNS penetration. Based on this analysis, we believe

that the compound can be a suitable candidate for designing derivatives with improved

physiochemical, pharmacokinetics and ADMET properties. Furthermore, the compound must be

targeted for in vitro analysis to determine its anti-bacterial activity.


The dynamic behavior of ligand-AbMurF was studied using MD to obtain a detailed analysis of

the structural adjustment adopted by the protein in the presence of the ligand. Structural changes

in the protein were analyzed through RMSD, RMSF, β-factor, and radius of gyration. The dynamic

behavior and stability of proteins without bound ligand can also be studied, however, studies

suggest that such system tends to have high RMSD values (Saleem, Azam, & Zarina, 2012; Azam,

Abro, & Raza, 2015). Accordingly, simulation of the protein without ligand was not performed.

The stability of the protein system was initially evaluated using the RMSD for atomic positions

335

and was plotted as a function of time (Fig.9.4A). The average RMSD value of the system was 2.09

Å with maximum RMSD of 3.60 Å was observed at 13 ns. That the RMSD value of the system

reached an approximately constant value suggests the overall stability of the system. Furthermore,

the ligand is tightly bounded in the active site of the protein and does not affect the overall topology

of the protein. Frequent extension and reduction of α-helices and β-sheets and conversion of α-

helices and β-sheets into loops were observed throughout 90 ns of simulation. To calculate an

average fluctuation of the residues of the protein over time, an RMSF graph was generated for the

simulated system (Fig.9.4B). The average RMSF value for the system was 1.33 Å with maximum

fluctuation value described for Arg206 as 4 Å. The RMSF value for the protein active site pocket

residues was found in the following order: Leu11(1.1 Å), Glu12 (1.0 Å), Trp14 (1.0 Å), Arg39

(1.18 Å), Leu41 (1.35 Å), Thr42 (1.4 Å), Asp43 (1.71 Å), Arg60 (3.1 Å), Phe61 (2.4 Å), Arg97

(1.3 Å), Arg107 (0.8 Å), Arg146 (0.9 Å), Met158 (0.9 Å), Leu160 (0.8 Å) and Glu161 (0.8 Å).

Residues from 58-61, 201-207 and 373-376 showed the highest fluctuation with an average value

of 2.96 Å, 3.32 Å and 2.73 Å respectively, which is expected as these regions are loops. The

thermal stability and flexibility of the protein and its side chain with respect to time were

deciphered through calculation of β-factor. A complete coherence was depicted by β-factor with

RMSD and RMSD values for the system (Fig.9.4C). The average β-factor value for the system

was noticed as 55.94 Å with a maximum value of 419.87 Å was observed for residue Arg86. In

RMSD and RMSF, the active site residues and the entire protein structure remain stable throughout

the 90 ns time scale of simulation and same is deciphered by the β-factor values. Three main

fluctuations were observed from residue 57-62, 200-208 and 372-380 with average β-factor values

of 199.07 Å, 259.14 Å and 162.33 Å respectively, which further confirmed the overall stability of

the protein structure. The stability of the system was further indicated by the radius of gyration,

which describes overall compactness of the protein. The average value of the Rg for the system

was 24.5 Å with the highest fluctuation seen at 13 ns, with the leveling off of the value over the

remainder of the simulation suggesting the overall stability of the system (Fig.9.4D).

336

Fig.9.4. RMSD (A), RMSF (B), β-factor (C) and Rg (D) plot for the GOLD docked complex.

9.4.4. RDF and AFD analysis

RDF analysis depicts the relative location of residues and molecules in the system. It helps in

identifying protein residues which play crucial roles in ligand binding and stability. For this,

hydrogen bonding analysis was done to find interactions that occur with the inhibitor over the

course of the simulation. The RDF between the amide hydrogen, oxygen, nitrogen of Thr42 and

the ligand was calculated during the course of 90 ns simulation to understand the interactions of

this catalytic residue. Similarly, RDF analysis was generated for the Asp35 amide hydrogen. In

the case of the amide hydrogen atom of Thr42 and ligand oxygen, the highest peak in first 10 ns

of simulation was observed at 2 Å with g (r) value of 0.71 while in last 10 ns of simulation the

highest peak was observed at 1.9 Å with g (r) value of 1.02. At the start of simulation, fluctuation

in the graph was observed from 1.6 Å up to 3.9 Å while at the end of simulation fluctuations were

seen from 1.6 Å up to 3.7 Å. The figure suggests that the graph is more refine and tend to narrow

337

down at the end of simulation, indicating that the distance between Thr42 hydrogen atom and

ligand oxygen decreases. It further means that the strength hydrogen interaction between these two

residues increases in the last 10 ns of simulation period (Fig.9.5A). In the case of a Thr42 nitrogen

atom and ligand oxygen atom, at start of simulation, the highest peak was observed at 3.4 Å with

g (r) value of 0.39 while at the end, highest peak was noticed at 2.8 Å with g (r) value of 0.54. The

figure indicates that in initial phases of simulation, the graph was refined over most of the period

while it tends to be distorted towards later phases of simulation, which may be because of steric

hindrance. It is evident that in last 10 ns of simulation, the graph was refined initially with a peak

higher than that of at start of simulation, indicating the increase strength of the interaction and

close proximity of the ligand in the protein active site. However, afterward, distortion was

observed in the middle of simulation period which may be due to steric hindrance phenomena.

Towards the end of the simulation period, graph line was observed refine suggesting a closest

association of ligand atom and receptor protein residue (Fig.9.5B). For oxygen atom of Thr42 and

the ligand, at initial phases of simulation, a more distorted graphs was observed when compared

to the end phases of simulation and probably it is because of steric hindrances effects. At the start

of simulation, the highest peak was observed at 2.8 Å with g (r) value of 0.31 while towards the

end of simulation the highest peak was found at 3.2 Å with g (r) value of 0.26. The peak at the

start of simulation was greater in magnitude and wider in width, suggesting that in initial 10 ns of

simulation the interaction between this atom of Thr42 and ligand oxygen atom was not defined

with increased distance between both. The similar trend between Thr42 oxygen and ligand oxygen

was observed in last 10 ns simulation, with increased distance and weak interaction (Fig.9.5C).The

other residue of the protein involved in interaction with the ligand during the course of simulation

was Asp43. A highly distorted graph was observed in the beginning and end of simulation period

for this residue of receptor protein which is possibly because of steric hindrances. The highest peak

was the system was observed at 3.5 Å with g (r) value of 0.17 in the start of simulation while at

the end of simulation the highest peak was observed at 3.25 Å with g (r) value of 0.23. Furthermore,

at the later stages simulation the graph narrow down a bit suggesting that distance between the

ligand atom and protein residue decreases and spend more time in close proximity to exhibit

inhibition activity (Fig.9.5D).

To further understand the pattern of hydrogen bonding between ligand and protein, AFD analysis

was employed. In general, the outcomes deciphered by this analysis was consistent with that of

338

RDF results. In the case of a hydrogen bond between hydrogen atom of Thr42 and ligand oxygen

atom, a similar trend of hydrogen bonding was observed as depicted by RDF graph. With the

period of simulation, the distance between the ligand and the active site residue decreases, making

the complex more stable and the interaction more strengthen (Fig.9.6). In the figure, it can be

easily noticed that in the initial phases of simulation, the distance between the ligand and protein

residue was large but at the end of simulation period this distance reduces. Similarly, it can easily

be observed that the distribution area of the ligand atoms is reduced with simulation time. The high

intensity of the peaks at the end of simulation was 0.1 and -0.1 Å which further supported the

overall stability and refinement of the complex. The AFD graph generated for nitrogen atom of

Thr42 of protein and oxygen of ligand can be seen in Fig.9.6. The graph indicated that the high

intensity of peaks in later period of simulation, suggesting its increased stability. The ligand tend

to move towards the protein active site center of the receptor protein, thus making the interaction

more stable and stronger. In case of AFD graph for oxygen atom of protein and ligand, it was

noticed that the surface area of simulation of ligand atoms decreases while intensity remains the

same. Similarly, it can be noted in Fig.9.6 that the distance between the ligand oxygen atom and

the receptor residue Thr42 oxygen atom increases on X axis while it remains the same on the Y

axis. AFD graph for a hydrogen atom of Asp43 and the oxygen atom of the ligand is presented in

Fig.9.6. The graph clearly indicates that toward simulation end the distance between ligand and

receptor atom decreases making the interaction more stable and refined.

339

Fig.9.5. A. RDF graph of Thr42 (hydrogen) and ligand (oxygen).B. RDF graph of Thr42 (nitrogen) and ligand (oxygen).C. RDF graph of Thr42

(oxygen) and ligand (oxygen). D. RDF graph of Asp43 (hydrogen) and ligand (oxygen).

340

Fig.9.6. AFD graphs for Thr42 and Asp43 before simulation (A) and after simulation (B).

341


The binding free energies and interactions for the docked complex were elucidated using

MM(PB/GB)SA of MD trajectories. The summary of binding energies obtained after the analysis

can be found in Table 9.3.

Table 9.3. Binding energy values for the complex


(kcal·mol− 1)

ΔEvdw -56.6126

ΔEele -31.6575

ΔEgas -88.2701

ΔGPB 56.9168

ΔGsolv,PB 84.3766

ΔGele,PB − 19.1567

Htot,PB -3.8935

ΔGGB 50.8142

ΔGsolv,GB 43.8624

ΔGele,GB − 25.2593

Htot,GB -44.4077

The entropy term was not considered during calculation due to convergence problems in some

cases where calculations were not accomplished. It can be easily concluded from the table that van

der Waals energy is quite favorable and dominates the overall binding energies (-56.6 kcal/mol).

The major non-favorable contribution towards free binding energy comes from a solvated part of

PB and GB (56.9 kcal/mol and 50.8 kcal/mol respectively). The contribution of electrostatic

energy in total binding free energy is also favorable and high in both PB and GB calculations (-

31.7 kcal/mol and -31.6 kcal/mol respectively). The average binding energy (Htot,PB or

GB = ΔGsol – np + ΔGPB/ or GB + ΔEgas) in the case of PB (-3.9 kcal/mol) is less compared to

GB (-44.4 kcal/mol), both of which indicates the stable nature of the complex. The difference in

the total energy in both calculation may due to polar solvation energy, which is in the case of PB

342

is 84.3 kcal/mol while in the case of GB is 43.8 kcal/mol. Molecular interaction between polar and

non-polar molecules can be directed in polar solvents however, non-polar interactions and their

contribution to the overall binding energy is more favorable. A total of 450 frames were selected

each after 0.2 ns and subjected to calculation of binding free energies using MM(PB/GB)SA

method. Fig.9.7. and Fig.9.8. show the binding free energy for GB and PB respectively calculated

over 450 frames of complex, receptor and ligand.

We compare the binding free energies from the present study with those obtained for MurD ligase

enzyme. Binding free energies for MurD ligase were calculated for both free and bounded state

using linear interaction energy (LIE) and experimental methodology. Initial analysis indicated that

in both states the binding is quite favorable and the system is stable. The van der Waals

contribution to overall energy was observed dominant in the bound state and drive the inhibitory

effects of the drugs. The total binding energies for 4 compounds were in the range from -4.9 to -

6.5 kcal/mol (Perdih, Wolber, & Solmajer, 2013). Analogous observations were revealed in the

current study. The total binding free energy for both PB and GB were (-3.8 kcal/mol) and (-44.4

kcal/mol) respectively which is an indication of high complex stability. Similarly, as stated earlier,

van der Waals interaction was found preferred and dominates the overall binding energy in case

of MurF complex. This indicated their significant contribution toward ligand binding and complex

stability.

343

Fig.9.7. Energy values vs number of frames from MM/GBSA calculations.

344

Fig.9.8. Energy values vs number of frames from MM/PBSA calculations.

345

In both GB and PB method, the pattern of fluctuation was found same. The highest total free energy

in case of the complex was found for frame 312 (-7784.8 kcal/mol) in GB while for the same frame

the binding free energy was as -7684.6 kcal/mol. For receptor protein, frame 312 showed highest

binding energy in GB (-7784.6 kcal/mol) while in PB frame 233 revealed to possess the highest

binding energy of -6718.3 kcal/mol. For ligand, frame 404 in both GB and PB possess the highest

binding free energy of -55.8 kcal/mol and -59.9 kcal/mol respectively. The total binding energy in

case of GB for frame 435 has the highest free energy of -64.0 while in the case of PB the total

binding free energy of frame 418 was found -18.6 kcal/mol.

To further decipher binding free energy at the atomic level, the binding free energy was

decomposed into per residue using MM/GBSA method and is plotted in Fig.9.9. A residue with

binding free energy < – 2 kcal/mol are considered as hot spot amino acids due to a major role in

stabilizing the complex. The active site residues involved in interaction with the ligand were found

to contribute extremely high and played significant contribution towards stabilization of the

complex. The binding free energy of these residues was found in the following order; Glu12 (-69.4

kcal/mol), Arg39 (-120.5 kcal/mol), Thr42 (-38.0 kcal/mol), Asp43 (-54.5 kcal/mol), Arg60 (-

112.7 kcal/mol), Arg97 (-117.2 kcal/mol), Arg107 (-115.0 kcal/mol), Glu161 (-79.0 kcal/mol) and

Arg146 (-118.4 kcal/mol). Furthermore, for a thorough investigation of the binding energy from

these residues, we decompose the binding free energy of these residues into its constituents like

electrostatic, van der Waals and solvation contribution (S-Table 9.3).

It is evident from the data the electrostatic energy of amino acid residues present in the active site

of proteins dominates the overall contribution. Residue like Leu41 and Phe61 was found to

contribute equally in terms of both van der Waals and electrostatic energies while Asp43 and Ser44

dominate the van der Waals energy. The total binding energy of Ile40, Leu41, Ser44, Phe61, and

Leu160 was found very low suggesting their lesser role in stabilizing the overall complex. The

most active amino acids of the proteins involved in interaction with the ligand tend to have higher

polar solvation contribution (≤ 2 kcal/mol). It was revealed from the analysis that Glu12 (-72.4

kcal/mol), Thr42 (-2.8 kcal/mol), Asp43 (-86.4 kcal/mol), Ser44 (-7.0 kcal/mol) and Glu161 (-52.7

kcal/mol) was found to have higher polar solvation energy, contributing maximum to overall free

binding energy of the complex. To get more understanding about the residues involved in

interaction with the ligand, decomposition of binding free energy into pair residues was performed.

346

Only two residues, Thr42 and Asp43, was taken into consideration for analysis because of their

significant involvement in interaction with the ligand while residues of protein active site were

considered for pairs. The pair-wise binding free energy for both the residues can be found in S-

Fig.9.3. In almost all the cases, the calculated total binding free energy was found < 0 kcal/mol,

suggesting the crucial role of all the adjacent residues in binding with the ligand and stabilizing

the complex. Further, both the residues were investigated for conservation among E.coli, S.

pneumoniae and T. maritima MurF ligase enzyme (S-Fig.9.4). It was found that Thr42 and Asp43

were conserved in 50 % and 75% of the strains which indicates that there is a significant difference

in the conservation of ligand binding sites among MurF of different bacterial species (Mol et al.,

2003). This further signifies the potential role of both these active site residue for designing

AbMurF specific inhibitors.

Fig.9.9. Decomposition of MM/GBSA free energy per residue of the protein.

347

9.5. Conclusions

Virtual screening followed by MD simulations illustrated the nature of the interactions of the top-

ranked ethyl pyridine substituted 3-cyanothiophene with A. baumannii MurF ligase enzyme. The

compound is suggested to be an ideal candidate for future investigation due to its pharmacokinetics

properties. The predicted potency of the compound is due to the ethyl pyrimidine ring substitution

on the 5 end of the compound, which allows the compound to bind deeply into a binding cavity on

the enzyme. During the course of the MD simulation the compound come in close contact with the

protein active site by forming strong hydrogen bonds with Thr43 and Asp43. Given the predicted

binding affinity of the compound, future experimental validation of binding energies and

biological activity is strongly suggested. Furthermore, the best hits from the current study could

be used as parent structure for designing derivatives with an improved potency which may act as

therapeutics for this notorious pathogen.


S-Fig.9.1. Structure comparison of MurF ligase enzyme from E.coli (red), S. pneumoniae

(magenta) and T. maritima (light green) with AbMurF (red).

S-Fig.9.2.(A) Docked pose of compound 114 in binding cavity of AbMurF. B. Ligplot highlighting

residues of a protein involved in interaction with the ligand.

S-Fig.9.3. Decomposition of MM/GBSA free energy for pair-wise residues.

S-Fig.9.4. Multiple sequence alignment of MurF ligase enzyme from different bacterial species.

Thr42 and Asp43 are indicated by a red box.

S-Table 9.1. Structures of anti-MurF compounds used in the current study.

S-Table 9.2. GOLD fitness score and binding energies of compounds used in the current study.

S-Table 9.3. Free energy decomposition into active site residues of the protein.

9.7. References



221, 507–517.

Azam, S. S., Abro, A., & Raza, S. (2015). Binding pattern analysis and structural insight into the

https://en.wikipedia.org/wiki/Thermotoga_maritima

348

inhibition mechanism of Sterol 24-C methyltransferase by docking and molecular dynamics

approach. Journal of Biomolecular Structure and Dynamics, 33(12), 2563–2577.



Baum, E. Z., Crespo-Carbone, S. M., Abbanat, D., Foleno, B., Maden, A., Goldschmidt, R., &

Bush, K. (2006). Utility of muropeptide ligase for identification of inhibitors of the cell wall

biosynthesis enzyme MurF. Antimicrobial Agents and Chemotherapy, 50(1), 230–236.

Baum, E. Z., Crespo-Carbone, S. M., Foleno, B. D., Simon, L. D., Guillemont, J., Macielag, M.,

& Bush, K. (2009). MurF inhibitors with antibacterial activity: effect on muropeptide levels.


Baum, E. Z., Crespo-Carbone, S. M., Klinger, A., Foleno, B. D., Turchi, I., Macielag, M., & Bush,

K. (2007). A MurF inhibitor that disrupts cell wall biosynthesis in Escherichia coli.




81(8), 3684–3690.

BIOvIA, D. S. (2015). Discovery studio modeling environment. San Diego, Dassault Systemes,

Release, 4.



Cha, S. S., An, Y. J., Jeong, C. S., Yu, J. H., & Chung, K. M. (2014). ATP-binding mode including

a carbamoylated lysine and two Mg 2+ ions, and substrate-binding mode in Acinetobacter

baumannii MurF. Biochemical and Biophysical Research Communications, 450(2), 1045–

1050.

Comess, K. M., Schurdak, M. E., Voorbach, M. J., Coen, M., Trumbull, J. D., Yang, H., … others.

(2006). An ultraefficient affinity-based high-throughout screening process: application to

bacterial cell wall biosynthesis enzyme MurF. Journal of Biomolecular Screening, 11(7),

743–754.

349




El Zoeiby, A., Sanschagrin, F., & Levesque, R. C. (2003). Structure and function of the Mur

enzymes: development of novel inhibitors. Molecular Microbiology, 47(1), 1–12.

Favini-Stabile, S., Contreras-Martel, C., Thielens, N., & Dessen, A. (2013). MreB and MurG as

scaffolds for the cytoplasmic steps of peptidoglycan biosynthesis. Environmental

Microbiology, 15(12), 3218–3228.

Gu, Y. G., Florjancic, A. S., Clark, R. F., Zhang, T., Cooper, C. S., Anderson, D. D., … others.

(2004). Structure--activity relationships of novel potent MurF inhibitors. Bioorganic &




Hrast, M., Anderluh, M., Knez, D., Randall, C. P., Barreteau, H., O’Neill, A. J., … Gobec, S.

(2014). Design, synthesis and evaluation of second generation MurF inhibitors based on a

cyanothiophene scaffold. European Journal of Medicinal Chemistry, 73, 83–96.

Hrast, M., Turk, S., Sosic, I., Knez, D., Randall, C. P., Barreteau, H., … others. (2013). Structure-

-activity relationships of new cyanothiophene inhibitors of the essential peptidoglycan

biosynthesis enzyme MurF. European Journal of Medicinal Chemistry, 66, 32–45.

Hubbard, R. E., & Kamran Haider, M. (2010). Hydrogen bonds in proteins: role and strength. ELS.

Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD-Visual Molecular Dynamics J Mol Graph

14: 33-38.



Kerwin, S. M. (2010). ChemBioOffice ultra 2010 suite. ACS Publications.

Laskowski, R. A., & Swindells, M. B. (2011). LigPlot+: multiple ligand-protein interaction

diagrams for drug discovery. American Chemical Society.

350



Longenecker, K. L., Stamper, G. F., Hajduk, P. J., Fry, E. H., Jakob, C. G., Harlan, J. E., … others.

(2005). Structure of MurF from Streptococcus pneumoniae co-crystallized with a small

molecule inhibitor exhibits interdomain closure. Protein Science, 14(12), 3039–3047.

Lugtenberg, E. J. J., & Van Schijndel-van Dam, A. (1972). Temperature-sensitive mutants of

Escherichia coli K-12 with low activities of the L-alanine adding enzyme and the D-alanyl-

D-alanine adding enzyme. Journal of Bacteriology, 110(1), 35–40.

Massova, I., & Kollman, P. A. (2000). Combined molecular mechanical and continuum solvent

approach (MM-PBSA/GBSA) to predict ligand binding. Perspectives in Drug Discovery and

Design, 18(1), 113–135.



130–155.

Miller, D., Hammond, S., Bugg, T. H., & others. (1998). Aminoalkylphosphinate inhibitors of D-

Ala-D-Ala adding enzyme. Journal of the Chemical Society, Perkin Transactions 1, (1), 131–

142.




Mol, C. D., Brooun, A., Dougan, D. R., Hilgers, M. T., Tari, L. W., Wijnands, R. A., … Swanson,

R. V. (2003). Crystal structures of active fully assembled substrate-and product-bound

complexes of UDP-N-acetylmuramic acid: L-alanine ligase (MurC) from Haemophilus

influenzae. Journal of Bacteriology, 185(14), 4152–4162.


365–382.

Na-Bangchang, K., Ruengweerayut, R., Karbwang, J., Chauemung, A., & Hutchinson, D. (2007).

Pharmacokinetics and pharmacodynamics of fosmidomycin monotherapy and combination

351

therapy with clindamycin in the treatment of multidrug resistant falciparum malaria. Malaria

Journal, 6(1), 70.





Perdih, A., Hrast, M., Barreteau, H., Gobec, S., Wolber, G., & Solmajer, T. (2014). Benzene-1, 3-

dicarboxylic acid 2, 5-dimethylpyrrole derivatives as multiple inhibitors of bacterial Mur

ligases (MurC-MurF). Bioorganic & Medicinal Chemistry, 22(15), 4124–4134.

Perdih, A., Hrast, M., Pureber, K., Barreteau, H., Grdadolnik, S. G., Kocjan, D., … Wolber, G.

(2015). Furan-based benzene mono-and dicarboxylic acid derivatives as multiple inhibitors

of the bacterial Mur ligases (MurC--MurF): experimental and computational characterization.

Journal of Computer-Aided Molecular Design, 29(6), 541–560.

Perdih, A., Wolber, G., & Solmajer, T. (2013). Molecular dynamics simulation and linear

interaction energy study of D-Glu-based inhibitors of the MurD ligase. Journal of Computer-

Aided Molecular Design, 27(8), 723–738.




Saleem, A., Azam, S. S., & Zarina, S. (2012). Docking and molecular dynamics simulation studies

on glycation-induced conformational changes of human paraoxonase 1. European Biophysics

Journal, 41(2), 241–248.

Salomon-Ferrer, R., Case, D. A., & Walker, R. C. (2013). An overview of the Amber biomolecular

simulation package. Wiley Interdisciplinary Reviews: Computational Molecular Science,

3(2), 198–210.

Schneider, T., & Sahl, H.G. (2010). An oldie but a goodie-cell wall biosynthesis as antibiotic target

pathway. International Journal of Medical Microbiology, 300(2), 161–169.

352

Sharma, A., & Pan, A. (2012). Identification of potential drug targets in Yersinia pestis using

metabolic pathway analysis: MurE ligase as a case study. European Journal of Medicinal

Chemistry, 57, 185–195.

Smith, C. A. (2006). Structure, function and dynamics in the mur family of bacterial cell wall

ligases. Journal of Molecular Biology, 362(4), 640–655.

Sobral, R. G., Ludovice, A. M., Gardete, S., Tabei, K., De Lencastre, H., & Tomasz, A. (2003).

Normally functioning murF is essential for the optimal expression of methicillin resistance in

Staphylococcus aureus. Microbial Drug Resistance, 9(3), 231–241.

Stamper, G. F., Longenecker, K. L., Fry, E. H., Jakob, C. G., Florjancic, A. S., Gu, Y.G., … others.

(2006). Structure-based Optimization of MurF Inhibitors. Chemical Biology & Drug Design,

67(1), 58–65.




Turk, S., Hrast, M., Sosic, I., Barreteau, H., Mengin-Lecreulx, D., Blanot, D., & Gobec, S. (2013).

Biochemical characterization of MurF from Streptococcus pneumoniae and the identification

of a new MurF inhibitor through ligand-based virtual screening. Acta Chimica Slovenica,

60(2), 294–299.

Turk, S., Kovac, A., Boniface, A., Bostock, J. M., Chopra, I., Blanot, D., & Gobec, S. (2009).

Discovery of new inhibitors of the bacterial peptidoglycan biosynthesis enzymes MurD and

MurF by structure-based virtual screening. Bioorganic & Medicinal Chemistry, 17(5), 1884–

1889.

Typas, A., Banzhaf, M., Gross, C. A., & Vollmer, W. (2012). From the regulation of peptidoglycan

synthesis to bacterial growth and morphology. Nature Reviews Microbiology, 10(2), 123–

136.

Van De Waterbeemd, H., & Gifford, E. (2003). ADMET in silico modelling: towards prediction

paradise? Nature Reviews Drug Discovery, 2(3), 192–204.

Vaught, A. (1996). Graphing with Gnuplot and Xmgr: two graphing packages available under

353

linux. Linux Journal, 1996(28es), 7.




55(1), 19–32.




Yan, Y., Munshi, S., Leiting, B., Anderson, M. S., Chrzas, J., & Chen, Z. (2000). Crystal structure

of Escherichia coli UDPMurNAc-tripeptide d-alanyl-d-alanine-adding enzyme (MurF) at 2.3

Åresolution1. Journal of Molecular Biology, 304(3), 435–445.

Yang, Y., Engkvist, O., Llinas, A., & Chen, H. (2012). Beyond size, ionization state, and

lipophilicity: influence of molecular topology on absorption, distribution, metabolism,

excretion, and toxicity for druglike compounds. Journal of Medicinal Chemistry, 55(8),

3667–3677.

354

Chapter # 10

Moleculer Dynamics Simulation Revealed Reciever Domain

of Acinetobacter baumannii BfmR Enzyme as the Hot Spot for

future Antibiotics Designing

10.1. Abstract

A. baumannii is an alarming nosocomial pathogen that is resistant to multiple drugs. The pathogen

is forefront of scientific attention because of high mortality and morbidity found for its

complications in the past decade. As a consequence, identification of novel drug candidates and

subsequent designing of novel chemical scaffolds is an imperative need of time. In the present

study, we used a recently reported structure of BfmR enzyme and performed structure based virtual

screening, MD simulation and binding free energies calculations. MD simulation revealed a

profound movement of the best-characterized inhibitor towards the α4-β5-α5 face of the enzyme

receiver domain, thus indicating its high affinity for this site compared to phosphorylation.

Furthermore, it was observed that the enzyme and enzyme-inhibitor complex have high structure

stability with mean RMSD of 1.2 Å and 1.1 Å, respectively. Binding free energy calculations for

the complex unraveled high stability with MM/GBSA score of -26.21 kcal/mol and MM/PBSA

score of -1.47 kcal/mol. Van der Waal energy was found highly favorable with value of -30.25

kcal/mol and dominated significantly the overall binding energy. Furthermore, a novel WaterSwap

assay was used to circumvent the limitations of MMGB/PBSA that complements the inhibitor

affinity for enzyme active pocket as depicted by the low convergence of Bennett, TI and FEP

algorithms. Results yielded from this study will not only give insight into the phenomena of

inhibitor movement towards the enzyme receiver domain, but will also provide a useful baseline

for designing derivatives with improved biological and pharmacokinetics profiles.

355

10.2. Introduction

A. baumannii is a member of the group ESKAPE and is contemplated as the most difficult to treat

Gram-negative bacilli ( Ahmad, Raza, Abro, Liedl, & Azam, 2018; Gonzalez-Villoria & Valverde-

Garduno, 2016; Lob et al., 2017; Wong et al., 2017). The bacterium caused different types of

infections including: urinary tract infections, pneumonia, skin and soft tissue infections,

bacteremia, endocarditis, and meningitis (Howard, O’Donoghue, Feeney, & Sleator, 2012).The

pathogen is resistant to almost all major classes of antibiotics including: aminoglycosides, colistin,

cephalosporins, carbapenems, chloramphenicol, fluoroquinolones, tetracyclines and penicillins

(Weber et al., 2017). In recent decades, the resistance to broad range of antibiotics is exhibited to

the presence of AbaR resistance islands in the genome, the presence of protective

lipopolysaccharide, capsule, efflux pumps, OmpA and biofilm (Kenyon et al., 2017; Lee et al.,

2017). Patients infecting from A. baumannii are generally at high risk of morbidity, mortality and

associated with high health care costs (Lashinsky, Henig, Pogue, & Kaye, 2017). In particular,

carbapenem-resistant A. baumannii in the intensive care units are observed with crude mortality

rate in the range of 26–76% (Schultz et al., 2016). Globally, the outbreaks of A. baumannii have

been reported in several countries including: Brazil, China, Germany, India, Iran, Italy, Iraq, Japan,

Spain, Turkey, the USA, and the UK (Ardoino et al., 2016; Hsu et al., 2017; Lei et al., 2016). In

the USA, each year A. baumannii causes approximately 12,000 infections and nearly 7000 from

these leads to 500 deaths (https://www.cdc.gov/drugresistance/threat-report-2013/pdf/ar-threats-

2013-508.pdf). Very recently, A. bumannii is ranked top on the list of bacterial pathogens that

urgently required development of new antibiotics (WHO, 2017). Among several protein targets

that can be targeted for novel inhibitors designing, the response regulator (BfmR) (also known as

RstA), is currently reported as the most promising anti-A. baumannii drug target (Russo et al.,

2016). Unique from others, BfmR inhibition leads to a dual effect on A. baumannii; increase

sensitivity to antibacterial and decrease in vivo survival (Russo et al., 2016). Additionally, BfmR

conferred phenotypes are resistant to bactericidal activity of complement system, important for

biofilm formation and allows outer membrane protein A (OmpA) scaffolding to antibiotics like

carbapenems ( Liou et al., 2014; Geisinger & Isberg, 2015). BfmR is a two-component signal

(TCS) response regulator and amplify signals from a sensor histidine kinase, BfmS (Laub &

Goulian, 2007; Tomaras, Flagler, Dorsey, Gaddy, & Actis, 2008). Bacterial two-component

systems are attractive drug targets (Stephenson & Hoch, 2004) and as such, a histidine kinase,

https://www.cdc.gov/drugresistance/threat-report-2013/pdf/ar-threats-2013-508.pdf

https://www.cdc.gov/drugresistance/threat-report-2013/pdf/ar-threats-2013-508.pdf

356

QseC, has been targeted successfully by LED209 molecule (Rasko et al., 2008). One of the

concerns regarding LED209 is that it only inhibits bacterial in vivo virulence and does not inhibit

its growth. Since, BfmR proper functioning is vital for in vivo viability (Umland et al., 2012) and

enhance in vivo pathogenicity of the pathogen in a murine pneumonia model (Wang, Ozer, Mandel,

& Hauser, 2014), its successful inactivation could results into even better anti- A. baumannii

potency compared to that observed for QseC (Russo et al., 2016).

In the present study, the crystal structure of BfmR was utilized in a computationally designed

hierarchy to identify natural lead-like compounds for the enzyme and to understand its dynamics.

The affinity of the best-characterized inhibitor for enzyme active pocket was further evaluated

through MM/GBSA, MM/PBSA (John, Sivashanmugam, Umashankar, & Natarajan, 2017), and a

novel WaterSwap approach (Woods et al., 2014). The approach applied in the current study was

compared with those CAAD approaches recently published (Verma, Tiwari, & Tiwari,

2017;Skariyachan, Manjunath, & Bachappanavar, 2018). As it has been reported that molecules

that follow drug-like rules are more promising and have higher chances to clear clinical trials and

reaching the market, thus the present study was designed to screen only those molecules that

completely follow the prominent drug-like rules: Lipinski rule of five (Lipinski, 2004), Ghose

filter (Daina, Michielin, & Zoete, 2017), Veber filter (Daina, Michielin, & Zoete, 2017), Egan

rule (Daina, Michielin, & Zoete, 2017) and Muegge rule (Daina, Michielin, & Zoete, 2017). Once

drug-like molecules were filtered they were further subjected to the parameters of leadlikeness

(Daina, Michielin, & Zoete, 2017). The lead-like molecules are found to be pharmacological and

biologically active and have suboptimal chemical entities that required minor structural

modifications to be able for best fitting into the biological active target pocket for its inhibition

(Shaikh, Jain, Sandhu, Latha, & Jayaram, 2007). Secondly, we used a library of natural

compounds. Natural compounds are preferred over the synthetic due to less adverse reactions and

toxic effects (Stratton, Newman, & Tan, 2015). Thirdly, we used the crystal structure of BfmR

enzyme (Russo et al., 2016). The advantage of using experimentally determined structure is that

such structures undoubtedly could provide the best starting point for drug discovery (Schmidt,

Bergner, & Schwede, 2014). Lastly, in the current study, to decipher the best-characterized

inhibitor affinity for the enzyme active pocket, we used a novel methodology of “WaterSwap”.

The applications of CADD including molecular docking, MD simulations and binding free

calculation studies have been providing significant insights on novel drug discovery against

357

bacterial pathogens (Skariyachan, Manjunath, & Bachappanavar, 2018). Experimental follow up

by testing the inhibition efficacy of the screened inhibitors in in vitro and in vivo studies is open

for experimentalists and this study will definitely speed up the development of novel antibiotics

process against the pathogen.


The complete step-wise workflow applied in this study is illustrated in Fig.10.1.

10.3.1. BfmR Enzyme Retrieval and Minimization

The 3D structure of BfmR enzyme predicted by X-ray diffraction is available in the PDB under

PDB ID, 5E3J (Russo et al., 2016). The crystal structure of BfmR exhibits two distinct interfaces:

α4-β5-α5 face in the active homodimer and α1-β2-α2 face in an inactive homodimer and can be

seen in Fig.10.2. The structure is resolved with a good resolution of 2.1 Å, therefore, retrieved and

subjected to energy minimization. Minimization was performed for relaxing the structure

energetically and to remove steric clashes (Ahmad, Raza, Uddin, & Azam, 2017; Baseer, Ahmad,

Ranaghan, & Azam, 2017; Ahmad et al., 2018). In total, 1500 rounds of minimization were

accomplished: 750 steps of steepest descent algorithm was first performed to relieve unfavorable

clashes. However, due to local minima and heuristic approach some clashes can remain even

during the steepest minimization and that must be removed before further processing. To do so,

alternate method with a different approach is needed to overcome local minima for which

conjugate gradient minimization was utilized that in general is a slower process but more effective

compared to steepest as it can reach to an energy minimum after steric clashes have been relieved.

Then 750 rounds of slow conjugate gradient method was applied to remove the severe clashes

remained during the steepest minimization under TFF in UCSF Chimera. During the process,

Gasteiger charges were assigned to the enzyme while the step size was set to 0.02 Å in both the

algorithms (Ahmad, Raza, Uddin, & Azam, 2017; Asad, Ahmad, Rungrotmongkol, Ranaghan, &

Azam, 2018; Ahmad et al., 2018).

10.3.2. Inhibitors Preparation

Asinex antibacterial library (http://www.asinex.com/) comprising natural scaffolds was used in a

structure based virtual screening. Prior to that, the library was filtered first for drug-like followed

by lead-like inhibitors. This was achieved by applying drug-like and lead-like rules in Ligand

Scout 4.1 ( Panman et al., 2017; Wolber & Langer, 2005). The parameters for drug-like and lead-

http://www.asinex.com/

358

like rules were extracted from an online SWISS-ADME, database (http://www.swissadme.ch/)

(Daina, Michielin, & Zoete, 2017). Drug-like rules used include: Lipinski rule of five (octanol-

water partition coefficient (LogP) ≤ 4.15, molecular weight (MW) ≤ 500, TPSA 40-130 Å2,

hydrogen bond donors (HBDs) ≤ 5, hydrogen bond acceptors (HBAs) ≤ 10) (Lipinski, 2004),

Ghose filter (LogP ≥ 0.4 - ≤ 5.6, MW ≥ 160 - ≤ 480, atoms ≥ 20 - ≤ 70, molecular refractivity

(MR) ≥ 40 - ≤ 130) (Daina, Michielin, & Zoete, 2017), Veber filter (rotatable bonds ≤ 10, TPSA

≤ 140) (Daina, Michielin, & Zoete, 2017), Egan rule (Wildman and Crippen LogP (WLogP) ≤

5.88, TPSA ≤ 131.6) (Daina, Michielin, & Zoete, 2017) and Muegge rule (TPSA ≤ 150, number

of rings ≤ 7, number of carbons > 4, number of heteroatoms > 1, number of rotatable bonds ≤ 15,

HBDs ≤ 5, HBAs ≤ 10, MW ≥ 200 - ≤ 600, octanol-water partition coefficient of organic

compounds (XLogP) ≥ -2 - ≤ 5) (Daina, Michielin, & Zoete, 2017s). Drug-like inhibitors were

scrutinized fruther based on filters of lead-likeness (XLogP ≤ 3.5, 250 ≤ MW ≤ 350 and rotatable

bonds ≤ 7) (Daina, Michielin, & Zoete, 2017). The resultant set of inhibitors were then minimized

using MMFF94 force field (Halgren, 1996) in Ligand Scout.

10.3.3. Molecular Docking of Lead-like Inhibitors

Molecular docking of shortlisted lead-like compounds with BfmR enzyme was performed using a

comparative docking approach (Ahmad, Raza, Uddin, & Azam, 2017; Verma, Tiwari, & Tiwari,

2017). First, an automated docking wizard of GOLD 5.2 (Jones, Willett, & Glen, 1995) was

utilized for structural based virtual screening, lead optimization and identifying correct binding

pose of the inhibitors relative to the BfmR enzyme active site. The binding site for inhibitors was

defined as such to include all atoms present within 10 Å of OD1 on Asp58 (Russo et al., 2016).

Russo et al. reported Asp58, previously as a critical residue of the enzyme for phosphorylation

activity (Russo et al., 2016). For each compound, ten docked iterations were obtained using default

generic algorithm parameters. These include number of operations 100,000, population size 100,

niche size 2, and islands number 5. For hydrogen bonds and van der Waal bond distances, a

threshold of 2.5 Å and 3.0 Å, respectively, were employed (Abbasi, Raza, Azam, Liedl, & Fuchs,

2016). The compounds affinity for the BfmR enzyme active site was measured by GOLD score,

Chem score, Astex Statistical Potential (ASP) and ChemPLP score (Shamim, Abbasi, & Azam,

2015). The binding affinity of top ten shortlisted inhibitors of GOLD was calculated using

AutoDock/Vina (Trott & Olson, 2010). The binding coordinates of the enzyme in AutoDock/Vina

were the same used in GOLD.

359

Fig.10.1. Step-wise flow of the methodology used in the current study.

360

Fig.11.2. Active and inactive homodimer of A. baumannii BfmR enzyme (Russo et al., 2016).

361

The selection of best-docked conformation was based on the four docking scores used in GOLD

along the AutoDock/Vina binding free energy, number of hydrogen bonds and interacting residues.

The docked complex with highest GOLD score, Chem score, ChemPLP, and lowest

AutoDock/Vina binding free energy was selected as the best-docked conformation. Moreover,

complex with maximum number of hydrogen bonds and interacting residues were also favored.


The absorption, distribution, metabolism, and excretion-toxicity (ADMET), physicochemical

descriptors, and medicinal chemistry friendliness of the shortlisted hits were deciphered through

SwisADME (Daina, Michielin, & Zoete, 2017) and preADMET (Lee et al., 2004).


For understanding the dynamics and unraveling mechanistic insights of BfmR enzyme and its

docked complex with the best-characterized inhibitor, a 100-ns MD simulation was carried out

using AMBER14 (Case et al., 2014). The antechamber program of AMBER14 package was used

to generate initial inhibitor library and parameters (Pearlman et al., 1995). Leap program was used

to explain molecular interactions of the complex using ff14SB force field (Pearlman et al., 1995).

The enzyme was integrated in TIP3P water box, with padding distance size set to 12 Å between

the box boundary and protein (Pearlman et al., 1995). System energies were minimized gradually

(Andleeb et al., 2016) : first, the systems were minimized for 500 steps, followed by 1000 cycles

of minimization for water box with a restraint on hydrogen atoms of 200 kcal/mol. Non-heavy

atoms of the systems were minimized for 300 cycles. Heating of systems was achieved by

upscaling the temperature to 300 k with time scale set to 20-ps. For maintaining temperature,

Langevin dynamics was used (Pastor, Brooks, & Szabo, 1988). The gamma value was set to 1.0,

while the SHAKE algorithm (Kräutler, Van Gunsteren, & Hunenberger, 2001) and NVT ensemble

were used for applying constrains on hydrogen bonds of the systems and heating, respectively.

Systems equilibration were achieved for time step of 2-ns and 100-ps (Andersen, 1980). NPT

ensemble was employed for sustaining systems pressure for 50-ps. Systems equilibration for 1-ns

were achieved using the same conditions discussed above. At the end, Berendsen temperature

coupled with the NVT ensemble was utilized for production run of 100-ns for each system. During

the production run, time step of 2-fs and non-bonded interactions cut off 8.0 Å were used. Analysis

of the simulation trajectories was done through CCPTRAJ (Roe & Cheatham, 2013) while

362

snapshots at different nanoseconds were visualized through VMD (Humphrey, Dalke, & Schulten,

1996) and UCSF Chimera (Pettersen et al., 2004). Different parameters like RMSD, RMSF, Rg

and β-factor were computed for both the systems. Hydrogen bond analysis was performed to depict

complex stability.

11.3.6. MMGB\PBSA Analysis

The binding free energies for the BfmR and best-characterized inhibitor complex was estimated

using MMGB\PBSA method of AMBER14 (Miller et al., 2012; Raj, Kumar, & Varadwaj, 2017).

Initial prmtop files for the complex, receptor and ligand were generated through ante-

MMPBSA.py module, while the subsequent calculation of total binding free energies and

decomposing binding free energies into specific residue contribution were accomplished using

MMPBSA.py module of AMBER14. The total binding energy takes the difference between the

complex free energy and that of receptor and ligand.

∆Gbind = ∆Gcomplex – [∆Greceptor - ∆Gligand]………………………………………………....(XXXIX)

Here, ∆G represents Gibb’s free energy and calculated from the following terms,

∆G = Egas + ∆Gsolv - TSsolute……………………………………………………………………(XL)

Where T in the equation stands for temperature, S is the entropy contribution to ligand binding,

and Egas is the gas phase energy that can represent by the following equation,

Egas = Eele + Evdw + Eint…………………………………………………………………………(XLI)

Eele, Evdw and Eint, are the contributions from electrostatic, van der Waals interaction energies and

internal energy, respectively.

The term ∆Gsolv in the equation is estimated implicitly and consists of electrostatic and non-polar

contributions,

∆Gsolv = ∆Gnp + ∆Gele…………………………………………………………………………(XLII)

Gele involves the inclusion of electrostatic energy and components of polar solvation in both GB

and PB,

∆Gele (PB/GB) = Eele + ∆GGB/PB..................................................................................................(XLIII)

363

∆Gnp represents contribution from non-electrostatic interactions and is proportional to molecule

solvent accessible surface,

Gnp = γSAS + β………………………………………………………………………………(XLIV)

LCPO was used for estimating Gnp value (Weiser, Shenkin, & Still, 1999). The standard values for

γ and β in MM/GBSA is set to 0.007 kcal.mol. Å2 and 0 kcal.mol-1, respectively. In case of

MM/PBSA, γ value of 0.005 kcal.mol. Å2 and β value of 0.92 kcal.mol-1 are used. In total, 500

frames were analyzed in both GB and PB.

11.3.7. WaterSwap Analysis

To avoid the limitations of MMGB/PBSA and to remove false positive results, binding free

energies were cross-validated using a novel approach of WaterSwap. An implicit solvent system

is employed in MMGB/PBSA to fill the cavity left behind due to ligand decoupling reaction. The

generated cavity is filled with water molecules and interacts with the active residues of the protein,

thus contributing to the total binding free energy (Woods et al., 2014). To overcome this,

WaterSwap reaction coordinate assay intrduced by Christopher et. al. was employed that compute

binding free energy using an explicit water model (Woods, Malaisree, Hannongbua, &

Mulholland, 2011). MMGB\PBSA uses the MD simulation generated conformations to estimate

binding free energy in contrast to WaterSwap that run its own Monte Carlo (MC) simulation for

generating trajectories and estimating the binding free energy. WaterSwap is comparatively a new

methodology that uses a single simulation for calculating protein-ligand complex binding energy

(Woods et al., 2014).

This new methodology is based on identity constraint that identify water cluster of an equal shape

and volume of ligand in the protein cavity, followed by swaping of the two using an algorithm of

dual topology. A replica exchange thermodynamic integration approach is integrated in the

WaterSwap to estimate absolute binding free energy of the system of interest (Woods, Malaisree,

Long, McIntosh-Smith, & Mulholland, 2013). Two simulation boxes: a protein box and a water

box connected to the same thermostat are used (Woods et al., 2014).

WaterSwap estimates binding energy using the following statistical equation:

364

E (λ) = Ewaterbox + Eproteinbox + Eligand + Ecluster + (1 - λ) (Eligand:proteinbox + Ecluster:waterbox) + (λ)

(Ecluster:proteinbox + Eligand:waterbox)………………………………………………………………(XLV)

In the equation, Ewaterbox is the energy of all the molecules except those present in the water cluster

of water box, Eproteinbox is protein box molecules energy apart from the ligand, Eligand is ligand

intramolecular energy, Ecluster is the energy of water molecules of the water cluster, Eligand:protein box

is the energy between protein box atoms and the ligand, Ecluter:water box is the interactions energy

between water clusters of the water box and water molecules, λ stands for reaction coordinates.

The basic principle is to simultaneously decouple water cluster from water box and ligand from

protein box. Eligand:water box represents ligand and water molecules energy in the protein box, while

energy of water molecules in water cluster and protein box is illustrated by Ecluter:protein box. λ is a

single coordinate reaction transferred from λ = 0 to λ = 1. λ = 0 indicates protein bounded ligand

in protein box while λ = 1, is for unbound ligand in bulk water.

At each value of λ, MC sampling must be attained to get average free energy gradients. Finally,

by integrating the gradient across λ the binding free energy is attained,

Gbind = - ʃo1 (dG/dλ)λ λd……………………………………………………………………(XLVI)

In the above equation, the negative integral is the ligand pulling out of the protein. Usually, sixteen

MC simulations have been found to be enough for generating sixteen free energy gradient spaced

across λ.

Waterswap calculations were performed using the same solvent and force field used in the MD

simulation (Woods, Malaisree, Long, McIntosh-Smith, & Mulholland, 2013). For clustering,

Density-based spatial clustering of applications with noise (DBSCAN) algorithm was used with

an acceptable sieve value set to 20. The calculations were run for 1000 iterations with cluster cut-

off size set to 1 Å and 25 members were permitted in each cluster. Total of 1.6 × 109 MC moves

were performed for calculating Thermodynamic Integration (TI), Free Energy Perturbation (FEP)

and Bennetts algorithm. The agreement value within 1 kcal/mol between these algorithms is

regarded reasonable.

365


The present study based on a combinatorial approach of molecular docking, dynamics simulation

and binding free energy calculation revealed BfmR enzyme as the most promising candidate for

novel antibiotics designing against A. baumannii. The findings from this study could provide

foundations for the treatment of many intractable diseases that are so far incurable. BfmR enzyme

is an intriguing antimicrobial target because of its inhibition that leads to dual effects: increase

sensitivity to antibiotics and decrease in vivo survival (Russo et al., 2016). Superimposition of

minimized structure over the non-minimized revealed an RMSD of 0.34 Å, suggesting improved

stability of the minimized enzyme. The Asinex library originally has 8044 natural scaffolds of

antibacterial inhibitors. Before utilizing the library, compounds that do not follow drug-like and

lead-like rules are excluded because they have high tendency to fail in clinical trial phases and

have less chances to reach the market. On the other hand, drug-like and lead-like compounds have

the maximum likelihood to reach the market thus reducing the cost of unsuccessful testing

(Proudfoot, 2002; Vora et al., 2018). Screening of the library based drug-like and lead-like filters

identified 2995 inhibitors. The lead-like compound is pharmacological and biologically active and

has suboptimal molecule, which requires structural modification to fit best to the target (Shaikh,

Jain, Sandhu, Latha, & Jayaram, 2007). The lead-like inhibitors were then utilized in molecular

docking for the identification of best hits against BfmR enzyme.

10.4.1. Molecular Docking of Lead-like Inhibitors

Molecular docking is a key approach in CAAD and structural molecular biology for predicting the

preferred binding mode of inhibitors with target protein of known 3D structure; where the correct

pose of inhibitors are ranked based on scoring functions ( Chen et al., 2017; Morris & Lim-Wilby,

2008). In addition, docking tools can be used in virtual screening of inhibitors library and proposed

structural hypothesis about ligand-driven inhibition mechanism of the targets (Ma, Chan, & Leung,

2011; Kumar, Srivastava, Negi, & Sharma, 2018). All the lead-like inhibitors were docked with

BfmR enzyme using Asp58 coordinates as described in the methodology section. In the

OmpR/PhoB family of enzymes, the aspartate residue is highly conserved and is responsible for

the catalytic mechanism of the enzyme through auto-phosphorylation (Russo et al., 2016). The

process of phosphorylation leads to an equilibrium shift and favor the formation of a stable

366

symmetrical homodimer conformation, which allow the effector domain to bind DNA and

modulate transcription (Russo et al., 2016). The docking scores of top 10 best inhibitors screened

in the study are tabulated in Table 10.1. The preferred orientation of binding molecules in BfmR

enzyme was evaluated based on a comparative assessment of five docking scores as mentioned in

the methods. Statistical Pearson correlation coefficients were estimated for the docking scores and

are illustrated in Fig.10.3. It can be seen that GOLD score is positively correlated to Chem score,

CHEMPLP, ASP and AutoDock/Vina score. A strong positive correlation between

AutoDock/Vina binding energy and GOLD fitness score was found opposed by weak relationship

between GOLD fitness score and ASP. Compound-530 (5-(2-hydroxyethyl)-3-(1-(pyrimidine-5-

carbonyl)-1,4-dihydropyridin-3-yl)-4H-pyrazol-1-ium-4-ide) was ranked as the most active

molecule against BfmR enzyme having GOLD score of 57.37 and Autodock/Vina binding energy

of -5.32 kcal.mol-1. A comparative binding interaction of Compound-530 in BfmR enzyme active

site in GOLD and Autodock/Vina can be visualized in Fig.10.4. In GOLD, the compound preferred

to bind through 4H-pyrazol-1-ium-4-ide ring to the bottom surface of the active site allowing the

propan-1-ol to interact hydrophilically with Asp58, Met60 and Ala86 of the pocket wall. The

presence of strong hydrogen bond of inhibitor that interacts with the enzyme active site residues

leads to enhanced stability of enzyme-ligand complex that increases the interaction specificity, a

fundamental aspect of molecular recognition. The pyridin-1(4H)-yl (pyrimidin-5-yl) methanone

ring of the inhibitor obstruct the interface of the active site channel for substrate binding. Critical

evaluation of the inhibitor binding within the BfmR cavity revealed strong binding with the

following enzyme residues: Glu14, Asp15, Asp58, Met60, Leu61, Pro62, Thr85, Ala86 Lys107.

In contrast to GOLD, Autodock/Vina revealed binding of the inhibitor near the phosphorylation

site, but not exactly at the site. The propan-1-ol moiety of the compound interacts strongly with

Asp65. In both GOLD and Autodock/Vina, the compound has hydrophobic interactions with

following common protein residues: Leu61, Pro62, and Arg87.

367

Fig.10.3. Correlation coefficient among different docking scores used in the study.

368

Fig.10.4 Binding interactions of the best screened inhibitor at BfmR docked site.

369

Table 10.1. Docking scores of shortlisted ten inhibitors.

Compound GOLDScore Chem

Score CHEMPLP ASP

AutoDock/Vina

binding energy

(kcal.mol-1)

Compound-530

57.37 18.73 53.61 19.94 -5.32

Compound-1610

56.02

13.44

36.03

15.41

-6.02

Compound-2242

55.84 21.74 45.92 19.98 -7.01

Compound-1724

55.09 16.53 44.79 18.42 -4.89

370

Compound-2805

55.09 17.00 59.67 2.12 -4.89

Compound-2755

55.01 19.47 52.30 13.37 -4.55

Compound-2884

54.77 12.43 47.85 18.62 -4.34

Compound-817

54.01 15.89 34.52 15.34 -4.19

371

Compound-2134

53.87 14.32 39.29 16.18 -4.00

Compound-1008

53.20 13.86 39.02 17.25 -3.98

10.4.2. SwissADME and preADMET Analysis

Toxicity of drugs at design and development stage has greatly outnumbered success rate (Wadood

et al., 2017). In the last decade, unsuccessful drugs in clinical stages have increased exponentially.

With the advent of computational tools, medicinal chemists can now predict the physiochemical

properties and hence selection of an appropriate drug candidate for lead optimization. The ultimate

objective of this exercise was to predict the ADMET properties and in vivo pharmacokinetics of

the shortlisted 10 drug candidates of the present study. The structural and physicochemical

information will enable medicinal chemists for designing chemical potent entities against the

pathogen in the future. All the compounds were found to follow prominent drug-like rules,

including Lipinski rule of five, Egan filter, Ghose rule, Muegge filter and Veber filter. TPSA is

used to calculate the presence of polar amino acid at compound surface (Fernandes & Gattass,

2009; Ertl, Rohde & Selzer, 2000). Compounds with lower TPSA are preferred as those with

higher value diminishes membrane permeability (Nisha et al., 2016). It can be seen in Table 10.2

all the compounds have low TPSA values and thus can be considered to have efficient membrane

permeability. High absorption value of the inhibitors illustrated greater absorption from the

intestinal tract when administered orally and are capable of reaching to the target sites in high

372

concentration (Nisha et al., 2016). Water solubility greatly influences compound absorption and

provide many advantages at drug designing stage (Daina, Michielin, & Zoete, 2017). All the

inhibitors were predicted to have high solubility. Additionally, carcinogenicity prediction revealed

only compound (inhibibitor number 1610) as carcinogenic, while the remaining nine compounds

were non-carcinogenic. Computational Ames test in TA1535 and TA100 strains of Salmonella

typhimurium predicted only two compounds: 1724 and 2134 as non-mutagenic. All the compounds

were predicted to have less skin permeability as indicated by the negative logKp values. The

screened compounds in this study could be assumed to have low CNS penetration due to their low

molecular weight and improved TPSA, thus can be less likely to cause CNS penetration associated

effects (Ahmad, Raza, Uddin, & Azam, 2017; Wadood et al., 2017). The synthetic accessibility of

the inhibitors is promising and can be easily synthesized for biological evaluations. Pan Assay

Interference (PAINS) compounds are molecules that have substructures, which yield false positive

biological results. All the inhibitors were found to have no PAINS substructures and ideal

candidate for experimental evaluations (Baell & Holloway, 2010). Overall, in this study, the

shortlisted predicted compounds have improved pharmacokinetics and druglikeness, thus can be

used for designing derivatives having improved biological potency.

10.4.3. MD Simulation

The dynamics insights of the enzyme and its complex with the best-characterized inhibitor were

unveiled for a period of 100-ns. Four statistical parameters: RMSD, RMSF, Rg, β-factor were

calculated to unveil complex stability (Fig.10.5). The relative fluctuations in RMSD were observed

almost the same revealing high stability of both systems. For unbounded enzyme, the mean RMSD

was 1.2 Å, while for bounded enzyme RMSD value of 1.1 Å was reported. The simulation time

for both the systems was extended to 200-ns to check for any variations in the mean RMSD. No

major RMSD variations were observed. The mean RMSD estimated for the enzyme is 1.2 Å, while

for the bounded enzyme the mean RMSD is 1.3 Å (S-Fig.10.1). RMSF is a vital parameter in

yielding informations about Cα atom structural flexibility of each residue. It is quite apparent from

the figure that along the whole trajectories, fluctuations of individual amino acids of the enzyme

are lesser in bounded form as compared to enzyme alone, depicting complex stability during MD

simulation. The equilibrium conformation of both systems was studied by calculating the Rg

values. Rg also aided in understanding compactness of the enzyme structure. Lower Rg value is

373

the depiction of tight packing while higher Rg value is in favor of less tight packing of the protein.

The mean Rg value for the enzyme is 22.057 Å, while for enzyme-inhibitor complex it was 21.96

Å. The highest Rg value for the enzyme is 23.90 Å in contrast to the lowest Rg value of 21.39 Å.

For complex, the maximum Rg value observed is 22.66 Å, while the minimum Rg is noticed with

value of 21.35 Å. β-factor represents thermal disorderness of the system and is used to specify

structural stability at atomic position in term of RMSF. RMSF and β-factor actually complements

each other in term of overall system stability and residual flexibility. The mean β-factor value for

the enzyme is 24.27 Å² with maximum value seen for Thr120 that is 289.55 Å². The average β-

factor value for the complex is 17.38 Å² with maximum value observed 102.34 Å² for residue

Thr120. To further evaluate stability of the complex, hydrogen bond analysis was performed.

Hydrogen bonding plays a vital role in complex stability and increase in hydrogen bonds increases

affinity of the drug molecules for the protein active site and thus strengthening complex stability.

As can be seen in the Fig.10.6 initially the hydrogen bonds formation between the enzyme and

inhibitor were rare and this can be explain by the movement of the docked inhibitor after 20-ns of

simulation towards α4-β5-α5 face of the receiver domain. After 20-ns,when the inhibitor adjust

itself in the receiver domain hydrogen bonds formation between Ala106, Ala226, Gln230 and

inhibitor atoms were seen consistently over the time of simulation period with increasing strength.

During the period of simulation, a profound ligand movement was observed from the initial docked

site and could be possibly explained by the high affinity of the inhibitor for α4-β5-α5 face of the

receiver domain as can be seen in Fig.10.7. Previously, based on computational method, it was

hypothesized that BfmR inhibitors would possibly bind to either α4-β5-α5 face of the receiver

domain to prevent formation of productive active homodimer or they may bind near Asp58 to

inhibit phosphorylation (Russo et al., 2016). Here in this study, we observed that the inhibitors

tend to docked with the receiver domain efficiently compared to the phosphorylation site, which

mean that this site could be a hot spot for designing future antibiotics against this pathogen.

374

Fig.10.5. MD simulation analysis for BfmR-inhibitor complex. A. RMSD, B. RMSF, C. Rg, D. β-factor.

375

Fig.10.6. Hydrogen bonds analysis of BfmR-inhibitor complex over the course of simulation.

376

Table.10.2. SwissADME and preADMET analysis of top ten hits screened in this study.

MW, Molecular weight, TPSA, Topological polar surface area, #R.B, number of rotatable bonds, #H.B.A, number of hydrogen bond acceptor,

#H.B.D, number of hydrogen bond donors, G.I.A, Gastrointestinal absorption, L.V, Lipinski rule of Five violations, G.V, Ghose filter violations,

V.V, Veber filter violations,E.V, Egan rule violations, M.V, Muegge rule violations, B.S, Biosynthetic score, S.A, Synthetic accessibility

Molecule

Physiochemical Properties Water Solubility Pharmacokinetics Drug likeness Medsicinal Chemistery Toxicity

MW TPSA #R.B #H.B.A #H.B.D ESOL

Log S

Ali

Log S

Silicos-IT

LogSw G.I.A

log Kp

(cm/s) SKLogD L.V G.V V.V E.V M.V B.S

PAINS

#alerts

Brenk

#alerts

Leadlikeness

#violations S.A

Ames

Test Carcinogenic

530 301 95 5 5 2 V.S V.S S High -8.1 -0.3 0 0 0 0 0 0.5 0 0 0 2.9 Mutagen Negative



1610 313 82 6 4 2 V.S V.S S High -7.7 -0.9 0 0 0 0 0 0.5 0 0 0 4.5 Mutagen Positive

1724 281 103 3 5 2 V.S V.S S High -8. -0.2 0 0 0 0 0 0.5 0 0 0 3.1 Non-

mutagen Negative

2134 349 85 4 6 1 V.S V.S S High -6.7 -2.0 0 1 0 0 0 0.5 0 2 0 5.3 Non-

mutagen Negative

2242 349 82 6 5 2 S S P.S High -7.8 0.7 0 0 0 0 0 0.5 0 0 0 3.4 Mutagen Negative

2755 324 94 4 4 4 S S S High -8.7 0.9 0 0 0 0 0 0.5 0 0 0 3.3 Mutagen Negative


2884 290 101 8 5 3 V.S S S High -8.4 -0.7 0 0 0 0 0 0.5 0 1 1 3.4 Mutagen Negative

377

Fig.10.7. Inhibitor movement from phosphorylation site to α4-β5-α5 face of the enzyme receiver domain during simulation.

378

10.4.4. Binding Free Energies Calculation

Binding free energies calculation is among one of the great challenge in computational biology

and measure the strength of binding between the ligand and protein (Kitamura et al., 2014).

Binding free energy is considered as a power tool in rational drug designing (Kitamura et al.,

2014). Molecular interactions and binding free energies between the best-characterized inhibitor

and BfmR enzyme were elucidated using MMGB/PBSA method of AMBER14. The summary of

different binding free energies is tabulated in Table 10.3. As can be seen, robust interactions were

observed between the inhibitor and the enzyme.

Table 10.3. Binding free energies for the complex.


(kcal.mol-1)

∆Evdw -30.25

∆Eele -18.11

∆Egas

∆Esol-np

∆GPB

∆Gsolv,PB

Htot,PB

-48.36

4.42

32.11

49.84

-1.47

∆Gsolv,GB 22.14

∆GGB 26.56

∆Gsol-np -3.68

Htot,GB -26.21

During the calculation, the term entropy was excluded due to convergence problem. Van der Waals

energy was found the most favorable with a value of -30.25 kcal/mol and dominates overall binding

energy. In addition, the total electrostatic energy was still high and contribute favorably. The total

binding free energy in the case of GB is -26.21 kcal/mol, while in PB it is -1.47 kcal/mol. The

difference in energy of GB and PB could be possibly due to difference in polar solvation energy

in both methods. In PB, polar solvation energy is 49.84 kcal/mol and is higher than in GB, 22.14

379

kcal/mol. Understanding the complex interactions at atomic level, the total energy was

decomposed into each residue of a protein (Abro & Azam, 2016). Protein residues having the

binding energy of < -1 kcal/mol were considered as hot spot amino acids because of their major

contribution to overall binding energy (Gonzalez-Villoria & Valverde-Garduno, 2016). In GB,

residues: Glu93 (-1), Trp104 (-1), Val105 (-1.43), Ala106 (-1.68), Val225 (-1.81), Ala226 (-2.55),

Val229 (-1), Gln230 (-1.46), Val233 (-1) were found to have a binding energy of < -1 kcal/mol

(Fig.10.8). In PB, six residues were found that have < -1 kcal/mol. These residues include: Ala106

(-1), Val205 (-1.4), Val225 (-2), Ala226 (-3),Gln230 (-1) and Val233 (-1) (Fig.10.9).

MMGB/PBSA based binding free energy is accomplished by taking snapshots of trajectories at

regular intervals from the entire simulation, while WaterSwap perform its own MC simulation to

estimate the binding affinity of water cluster and ligand towards protein active pocket. The

difference of hydration energy of protein and protein-ligand complex in MMGB/PBSA is usually

large and vital to be investigated as any error in their values could have a greater effect on overall

binding energy of the system. MMGB/PBSA uses an implicit water model that skips molecular

interaction details of protein-water and ligand-water. Therefore, information regarding bridging

interactions of water molecule between protein and ligand is of great importance (Woods et al.,

2014). To avoid such limitation, we used a more convenient and advanced assay of WaterSwap.

WaterSwap exhibits a high affinity of the inhibitor towards the protein active site as proposed by

the low grade convergence values of TI, PEF, and Bennett algorithm, as shown in Table 10.4.

Table 10.4. WaterSwap calculations for the complex.


(kcal/mol)

FEP - 18.6

TI - 19.9

Bennetts - 20.6

380

Fig.10.8.MM/GBSA based decomposition of binding free energies into enzyme residues. The number around the circle represents the

residue number of enzyme.

381

Fig.10.9. MM/PBSA based decomposition of binding free energies into enzyme residues. The number around the circle represents the

residue number of enzyme.

382

10.5. Conclusions

The current study was designed with the aim to predict lead-like molecules against BfmR enzyme

of A. baumannii. BfmR is an appealing drug target against A. baumannii for several reasons:(i) it

has no sequence similarity with the human proteins, (ii) highly conserved among the bacterium

strains, (iii) high conservation of the receiver domain among BfmR enzymes across the bacterial

species and (iv) its inhibition could have dual effect: increase antibacterial sensitivity and decrease

in vivo survival. Because of these properties, the recent reported crystal structure of BfmR was

used in a structure based virtual screening of shortlisted natural lead-like inhibitors that revealed

several important lead-like molecules for designing more potent derivatives with improved

physicochemical parameters and biological activity. To validate the docking findings and

understand the enzyme and enzyme-inhibitor complex dynamics, MD simulation was performed

that unraveled both the systems as highly stable with mean RMSD of 1.2 Å and 1.1 Å, respectively.

The best-characterized inhibitor was seen to show high affinity for the receiver domain of the

enzyme and thus is a hotspot site for its inhibition. The affinity of the inhibitor for the receiver

domain was validated using MMGB/PBSA that predicted highly appreciable affinity of the

inhibitor with MM/GBSA value of -26.21 kcal/mol and MM/PBSA score of -1.47 kcal/mol.

Moreover, a novel and more sophisticated method of Waterswap was used to cross validate the

findings of MMGB/PBSA. WaterSwap is in strong agreement of the inhibitor affinity for the

receiver domain as depicted by the low convergence of Bennett, TI and FEP algorithms. Findings

of the current study may provide a foundation for designing future antibiotics against the pathogen.


S-Fig.10.1. Mean RMSD estimated for the enzyme (red line) and for the bounded enzyme (blue

line).

10.7. References

Ahmad, S., Raza, S., Abro, A., Liedl, K. R., & Azam, S. S. (2018). Towards novel inhibitors

against KdsB: A highly specific and selective broad-spectrum bacterial enzyme. Journal

of Biomolecular Structure and Dynamics, 17, 1–20.

383



221, 507–517.


methyltransferase with S-adenosylmethionine. Journal of Molecular Liquids, 220, 375–

382.



Journal of Molecular Graphics and Modelling,77, 72-85.

Ardoino, I., Zangirolami, F., Iemmi, D., Lanzoni, M., Cargnelutti, M., Biganzoli, E., & Castaldi,

S. (2016). Risk factors and epidemiology of Acinetobacter baumannii infections in a

university hospital in northern Italy: A case-control study. American Journal of Infection

Control, 44, 1600–1605.





using triphenylgermane as a catalyst and its binding pattern validation. RSC Advances,

6(83), 79651–79661.

Asad, Y., Ahmad, S., Rungrotmongkol, T., Ranaghan, K. E., & Azam, S. S. (2018). Immuno-

informatics driven proteome-wide investigation revealed novel peptide-based vaccine

targets against emerging multiple drug resistant Providencia stuartii. Journal of Molecular

Graphics and Modelling, 80, 238–250.

Baell, J. B., & Holloway, G. A. (2010). New substructure filters for removal of pan assay

interference compounds (PAINS) from screening libraries and for their exclusion in

bioassays. Journal of Medicinal Chemistry, 53(7), 2719–2740.


against Shigella sonnei: A subtractive reverse vaccinology based approach. Biologicals,

50, 87-99.

https://doi.org/10.1016/j.molliq.2016.04.109

https://www.sciencedirect.com/science/journal/10451056/50/supp/C

https://doi.org/10.1016/j.biologicals.2017.08.004

384

Chen, R., Wan, J., Song, J., Qian, Y., Liu, Y., & Gu, S. (2017). Rational screening of peroxisome

proliferator-activated receptor-γ agonists from natural products: potential therapeutics for

heart failure. Pharmaceutical Biology, 55(1), 503–509.


14.




Ertl, P., Rohde, B., & Selzer, P. (2000). Fast calculation of molecular polar surface area as a sum

of fragment-based contributions and its application to the prediction of drug transport

properties. Journal of Medicinal Chemistry, 43(20), 3714–3717.

Fernandes, J., & Gattass, C. R. (2009). Topological polar surface area defines substrate transport

by multidrug resistance associated protein 1 (MRP1/ABCC1). Journal of Medicinal

Chemistry, 52(4), 1214–1218.

Geisinger, E., & Isberg, R. R. (2015). Antibiotic modulation of capsular exopolysaccharide and

virulence in Acinetobacter baumannii. PLoS Pathogens, 11(2), e1004691.



Pathogens, 2016.





Hsu, L. Y., Apisarnthanarak, A., Khan, E., Suwantarat, N.,Ghafur, A., & Tambyah, P. A. (2017).

Carbapenem-resistant Acinetobacter baumannii and Enterobacteriaceae in south and

Southeast Asia. Clinical Microbiology Reviews, 30, 1–22.

Humphrey, W., Dalke, A., & Schulten, K. (1996). VMD: visual molecular dynamics. Journal of

Molecular Graphics, 14(1), 33–38.

385

John, A., Sivashanmugam, M., Umashankar, V., & Natarajan, S. K. (2017). Virtual screening,

molecular dynamics, and binding free energy calculations on human carbonic anhydrase

IX catalytic domain for deciphering potential leads. Journal of Biomolecular Structure and

Dynamics, 35(10), 2155–2168.

Jones, G., Willett, P., & Glen, R. C. (1995). Molecular recognition of receptor sites using a genetic

algorithm with a description of desolvation. Journal of Molecular Biology, 245(1), 43–53.

Kenyon, J. J., Kasimova, A. A., Shneider, M. M., Shashkov, A. S., Arbatsky, N. P., Popova, A.

V., … Knirel, Y. A. (2017).The KL24 gene cluster and a genomic island encoding a

wzy polymerase contribute genes needed for synthesis of the K24 capsular polysaccharide

by the multiply antibiotic resistant Acinetobacter baumannii isolate RCH51. Microbiology,

163, 355–363.



measurement of binding free energy for 7-azaindole derivatives to glycogen synthase

kinase-3β. Journal of Chemical Information and Modeling, 54(6), 1653–1660.




Kumar, A., Srivastava, G., Negi, A. S., & Sharma, A. (2018). Docking, molecular dynamics,

binding energy-MM-PBSA studies of naphthofuran derivatives to identify potential dual

inhibitors against BACE-1 and GSK-3β. Journal of Biomolecular Structure and

Dynamics,1–37.

Lashinsky, J. N., Henig, O., Pogue, J. M., & Kaye, K. S. (2017).Minocycline for the treatment of

multidrug and extensively drug-resistant A. baumannii: A review. Infectious Diseases and

Therapy, 6, 199–211.

Laub, M. T., & Goulian, M. (2007). Specificity in two-component signal transduction pathways.

Annu. Rev. Genet., 41, 121–145.

Lee, C. R., Lee, J. H., Park, M., Park, K. S., Bae, I. K., Kim, Y.B., … Lee, S. H. (2017). Biology

of Acinetobacter baumannii: Pathogenesis, antibiotic resistance mechanisms, and

386

prospective treatment options. Frontiers in Cellular and Infection Microbiology, 7, 55.



Liou, M.L., Soo, P.C., Ling, S.R., Kuo, H.Y., Tang, C. Y., & Chang, K.C. (2014). The sensor

kinase BfmS mediates virulence in Acinetobacter baumannii. Journal of Microbiology,

Immunology and Infection, 47(4), 275–281.

Lei, J., Han, S., Wu, W., Wang, X., Xu, J., & Han, L. (2016). Extensively drug-resistant

Acinetobacter baumannii outbreak cross-transmitted in an intensive care unit and

respiratory intensive care unit. American Journal of Infection Control, 44, 1280–1284.



Lob, S. H., Hackel, M. A., Kazmierczak, K. M., Young, K., Motyl, M. R., Karlowsky, J. A., &

Sahm, D. F. (2017). In vitro activity of imipenem-relebactam against Gram-negative

ESKAPE pathogens isolated by clinical laboratories in the United States in 2015 – results

from the SMART global surveillance program. Antimicrobial Agents and Chemotherapy,

62, e02209-16.

Ma, D.L., Chan, D. S.H., & Leung, C.H. (2011). Molecular docking for virtual screening of natural

product databases. Chemical Science, 2(9), 1656–1665.




Morris, G. M., & Lim-Wilby, M. (2008). Molecular docking. In Molecular modeling of proteins

(pp. 365–382). Springer.

Nisha, C. M., Kumar, A., Nair, P., Gupta, N., Silakari, C., Tripathi, T., & Kumar, A. (2016).

Molecular Docking and In Silico ADMET Study Reveals Acylguanidine 7a as a Potential

Inhibitor of β-Secretase. Advances in Bioinformatics, 2016.



387

Panman, W., Nutho, B., Chamni, S., Dokmaisrijan, S., Kungwan, N., & Rungrotmongkol, T.

(2017). Computational screening of fatty acid synthase inhibitors against thioesterase

domain. Journal of Biomolecular Structure and Dynamics, 1–12.











Proudfoot, J. R. (2002). Drugs, leads, and drug-likeness: an analysis of some recently launched

drugs. Bioorganic & Medicinal Chemistry Letters, 12(12), 1647–1650.

Raj, U., Kumar, H., & Varadwaj, P. K. (2017). Molecular docking and dynamics simulation study

of flavonoids as BET bromodomain inhibitors. Journal of Biomolecular Structure and

Dynamics, 35(11), 2351–2362.

Rasko, D. A., Moreira, C. G., Li, D. R., Reading, N. C., Ritchie, J. M., Waldor, M. K., … others.

(2008). Targeting QseC signaling and virulence for antibiotic development. Science,

321(5892), 1078–1080.



3084–3095.

Russo, T. A., Manohar, A., Beanan, J. M., Olson, R., MacDonald, U., Graham, J., & Umland, T.

C. (2016). The response regulator BfmR is a potential drug target for Acinetobacter

baumannii. MSphere, 1(3), e00082-16.

Schultz, M. B., Pham-Thanh, D., Tran-Do-Hoan, N., Wick, R. R., Ingle, D. J., Hawkey, J., … Holt,

388

K. E. (2016). Repeated local emergence of carbapenem-resistant Acinetobacter baumannii

in a single hospital ward. Microb. Genom., 2. Article id: e000050.

Schmidt, T., Bergner, A., & Schwede, T. (2014). Modelling three-dimensional protein structures

for applications in drug design. Drug Discovery Today, 19(7), 890–897.

Shaikh, S. A., Jain, T., Sandhu, G., Latha, N., & Jayaram, B. (2007). From drug target to leads-

sketching a physicochemical pathway for lead molecule design in silico. Current

Pharmaceutical Design, 13(34), 3454–3470.

Shamim, A., Abbasi, S. W., & Azam, S. S. (2015). Structural and dynamical aspects of

Streptococcus gordonii FabH through molecular docking and MD simulations. Journal of

Molecular Graphics and Modelling, 60, 180–196.


molecules against prioritised targets of multi-drug-resistant Acinetobacter baumannii-

insights from molecular docking, molecular dynamic simulations and in vitro assays.

Journal of Biomolecular Structure and Dynamics, 1–24.

Stephenson, K., & Hoch, J. A. (2004). Developing inhibitors to selectively target two-component

and phosphorelay signal transduction systems of pathogenic microorganisms. Current


Stratton, C. F., Newman, D. J., & Tan, D. S. (2015). Cheminformatic comparison of approved

drugs from natural product versus synthetic origins. Bioorganic & Medicinal Chemistry

Letters, 25(21), 4802–4807.

Tomaras, A. P., Flagler, M. J., Dorsey, C. W., Gaddy, J. A., & Actis, L. A. (2008). Characterization

of a two-component regulatory system from Acinetobacter baumannii that controls biofilm

formation and cellular morphology. Microbiology, 154(11), 3398–3409.




Umland, T. C., Schultz, L. W., MacDonald, U., Beanan, J. M., Olson, R., & Russo, T. A. (2012).

In vivo-validated essential genes identified in Acinetobacter baumannii by using human

389

ascites overlap poorly with essential genes detected on laboratory media. MBio, 3(4),

e00113-12.



Acinetobacter baumannii. Journal of Biomolecular Structure and Dynamics, 1–13.

Vora, J., Patel, S., Sinha, S., Sharma, S., Srivastava, A., Chhabria, M., & Shrivastava, N. (2018).

Molecular docking, QSAR and ADMET based mining of natural compounds against prime

targets of HIV. Journal of Biomolecular Structure and Dynamics, 1-16.




55(1), 19–32.

Wang, N., Ozer, E. A., Mandel, M. J., & Hauser, A. R. (2014). Genome-wide identification of

Acinetobacter baumannii genes necessary for persistence in the lung. MBio, 5(3), e01163-

-14.

Weber, B. S., Kinsella, R. L., Harding, C. M., & Feldman, M. F. (2017). The secrets of

Acinetobacter secretion. Trends in Microbiology, 25, 532–545.



217–230.



Modeling, 45(1), 160–169.

Wong, D., Nielsen, T. B., Bonomo, R. A., Pantapalangkoor, P., Luna, B., & Spellberg, B. (2017).

Clinical and pathophysiological overview of Acinetobacter infections: A century of

challenges. Clinical Microbiology Reviews, 30, 409–447.

390

Chapter # 12

Supplementary Data

391

Chapter # 3

S-Fig.3.1. A. RMSD, B. RMSF, C. Rg.

392

S-Table.3.1. Extinction coefficients for the shortlisted proteins.

Protein

Extinction coefficients

(assuming all pairs of Cys residues

form cystines)

(M-1 cm-1)

Extinction coefficients

(assuming all Cys residues are

reduced)

(M-1 cm-1)

Methyl-accepting chemotaxis protein

(MACP)

27515

27390

Outer membrane protein assembly factor

(BamA)

90790

90540

Multidrug efflux RND transporter

(RND)

54320

54320

TonB-dependent siderophore receptor

(TonB)

92265

92140

Type IV secretion protein (Rhs) 239100

238600

Fimbrial biogenesis outer membrane

usher protein (FimD)

139720

139470

393

S-Table.3.2. Active site residues of TLR4.

Chain Residues of the active site

Chain A

Arg264-Asn265, Tyr292-Asp294, Tyr296, Ser317-Lys324, Val338-Phe345,

Thr347-Lys349,Ser360-Asp371, Arg382-Gly384, Ser386-Gly389, Ser392,

Ser394-Asp395, Phe408-Val411, Thr413-Asn417, His431-Leu444,

His456,.His458-Phe463, Gly480, Asn481-Phe487, Gln505-Leu511, His529-

Ser534, Asp536, Pro539,Tyr551,Leu553, His555,Met557-Thr558, Lys560,

Gln578,Asp580, Ala582, Thr584,Glu586-His587,Arg606,Ala610, Thr611-

Gln617

Chain B

Arg264-Asn266, Asn268, Glu270, Tyr292-Tyr296, Ser317-Lys324, Asn339-

Phe345, Thr347-Lys349, Ser360-Asp371, Arg382-Gly384, Ser386-Gly389,

Ser392, Ser394,-Asp395, Phe408-Val411, Thr413-Asn417, His431-Ser441,

His456, His458-Arg460, Ala462-Phe463, Asn468, Gly480-Asn486, Gln505,

Gln507, Glu509-Leu511, His529-His534, Asp536, Pro539, Tyr551,Leu553-

Glu563, Gln578-Asp580, Thr584, Glu586-His587,Arg606, Glu608-Pro612,

Gln616-Gly617

Chain C

Val24, Ile32, Lys39, Ile44,Ile46, Val48, Ile52, Leu54,Lys58, Leu61,

Ile63,Tyr65,Leu71, Leu74, Phe76-Ile80, Val82, Met85,Leu87-Arg96,

Asp100-Phe104,Val113,Thr115,Ile117-Tyr127,Tyr131, Cys133,Val135-

Glu136, Ile138 and Glu143-Leu153

Chain D

Val24, Ile32, Ile44, Ile46, Val48, Ile52, Leu54-Ile65,

Tyr65,Leu71,Leu74,Phe76,Leu78,Ile80,Val82,Met85,Leu87-Arg96,Asp101-

Phe104,Val113,Thr115-Val135, Leu146, Phe147, Leu149,Phe151 and Ile153

394

Chapter # 4

S-Fig.4.1.Venn diagram representation of prioritized five proteins. Each color in the diagram represents different protein category. The

70 virulent proteins were filtered to ten outer membrane and extracellular (O.M and E.C) proteins, followed by homology filter that

ensures all 10 proteins are human-non-homologous proteins (H.N.H.Ps). Molecular weight (MW) calculation of proteins provided nine

proteins with weight less than 110 kDa. Transmembrane helices (T.M.Hs) were found less than one for eight proteins while only five

proteins were found to have adhesive probability (A.D.P).This Venn diagram is generated via Jvenn (http://bioinfo.genotoul.fr/jvenn/).

http://bioinfo.genotoul.fr/jvenn/

395

S-Fig.4.2. Representation of conserved, antigenic and virulent peptide sequence of CsuB (A) and EpsA (B) protein.

396

S-Fig.4.3. Interactome analysis. A. The prioritized CsuB protein shows direct and indirect interactions with other proteins of Csu operon

(csuC, csuD, csuE, csuA/B, csuA), flagellar and pili proteins (HMPREF0010_00250 and HMPREF0010_00251) and sugar

transportation channel proteins (HMPREF0010_00450, HMPREF0010_02773).B. Similarly, EpsA (HMPREF0010_03288) protein

interact with tyrosine kinase family of proteins (HMPREF0010_03289, HMPREF0010_00298), ATPases (HMPREF0010_01283),

sugar transferases (HMPREF0010_03276), and amino acid transferases.

397

S-Fig.4.4 2D depiction of docked CsuB prioritized epitope in the binding groove of DRB1*0101 allele using MOE.

398

S-Fig.4.5. 2D depiction of docked EpsA prioritized epitope in the binding groove of DRB1*0101 allele using MOE.

399

S-Fig.6. 2D depiction of docked CsuB (left) and EpsA (right) epitope in binding pocket of DRB1*0101 using Ligplot.

400

S-Table.4. 1. Seventy virulent proteins of A. baumannii reported by VFDB

S.NO Category Sub-Category Gene Protein Location Strand COG

1 Adherence Adherence ompA outer membrane protein chromosome Minus COG2885

2 Biofilm

formation

AdeFGH

efflux pump adeF membrane-fusion protein chromosome Plus COG0845

3 adeG cation/multidrug efflux pump chromosome Plus COG0841

4 adeH outer membrane protein chromosome Plus COG1538

5 Bap (biofilm-associated

protein)

bap hemolysin-type calcium-binding domain-containing protein chromosome Minus N.A

6 Csu fimbriae csuA CsuA chromosome Minus

7 csuA/B hypothetical protein chromosome Minus COG5430

8 csuB putative type 1 pili subunit CsuA/B protein chromosome Minus COG5430

9 csuC P pilus assembly protein, chaperone PapD chromosome Minus COG3121

10 csuD P pilus assembly protein, porin PapC chromosome Minus COG3188

11 csuE hypothetical protein chromosome Minus COG5430

12 PNAG pgaA hypothetical protein chromosome Minus

13 pgaB xylanase/chitin deacetylase chromosome Minus COG0726

14 pgaC glycosyltransferase chromosome Minus COG1215

15 pgaD hypothetical protein chromosome Minus

16 Enzymes

phospholipases plc phospholipase C chromosome Plus COG3511

17 plc phospholipase C chromosome Minus COG3511

18 plcD phosphatidylserine/phosphatidylglycerophosphate/cardiolipin

synthase chromosome Plus COG1502

19 Immune

evasion Capsule ACICU_00071 ATPase chromosome Minus COG0489

20 ACICU_00072 protein-tyrosine-phosphatase chromosome Minus COG0394

21 ACICU_00073 periplasmic protein chromosome Minus COG1596

http://www.ncbi.nlm.nih.gov/COG/

http://www.ncbi.nlm.nih.gov/nuccore/NC_010611

















http://www.mgc.ac.cn/cgi-bin/VFs/vfs.cgi?Genus=Acinetobacter&Keyword=Immune%20evasion




401

22 ACICU_00074 UDP-N-acetyl-D-mannosaminuronate dehydrogenase chromosome Plus COG0677

23 ACICU_00075 nucleoside-diphosphate sugar epimerase chromosome Plus COG1086

24 ACICU_00076 pyridoxal phosphate-dependent enzyme chromosome Plus COG0399

25 ACICU_00077 CMP-N-acetylneuraminic acid synthetase chromosome Plus COG1083

26 ACICU_00078 spore coat polysaccharide biosynthesis protein,

glycosyltransferase chromosome Plus COG3980

27 ACICU_00079 acetyltransferase chromosome Plus COG1670

28 ACICU_00080 sialic acid synthase chromosome Plus COG2089

29 ACICU_00081 membrane protein chromosome Plus COG2244

30 ACICU_00082 hypothetical protein chromosome Plus COG3562

31 ACICU_00083 hypothetical protein chromosome Plus



34 ACICU_00086 glycosyltransferase chromosome Plus COG1215

35 ACICU_00087 sugar transferase chromosome Plus COG2148

36 ACICU_00088 UDP-glucose pyrophosphorylase chromosome Plus COG1210

37 ACICU_00089 UDP-glucose 6-dehydrogenase chromosome Plus COG1004

38 ACICU_00091 UDP-glucose 4-epimerase chromosome Plus COG1087

39 ACICU_00092 phosphomannomutase chromosome Minus COG1109

40 pgi glucose-6-phosphate isomerase chromosome Plus COG0166

41 LPS lpsB glycosyltransferase chromosome Minus COG0438

42 lpxA UDP-N-acetylglucosamine acyltransferase chromosome Minus COG1043

43 lpxB lipid-A-disaccharide synthase chromosome Plus COG0763

44 lpxC UDP-3-O-[3-hydroxymyristoyl] N-acetylglucosamine

deacetylase chromosome Minus COG0774

45 lpxD UDP-3-O-[3-hydroxymyristoyl] glucosamine N-acyltransferase

chromosome Minus COG1044

46 lpxL lauroyl/myristoyl acyltransferase chromosome Minus COG1560























402

47 lpxM lauroyl/myristoyl acyltransferase chromosome Plus COG1560

48 Iron

Uptake Acinetobactin barA multidrug ABC transporter ATPase and permease chromosome Minus COG4988

49 barB multidrug ABC transporter ATPase and permease chromosome Minus COG2274

50 basA acinetobactin biosynthesis protein chromosome Minus COG1020

51 basB non-ribosomal peptide synthetase module chromosome Plus COG1020

52 basC lysine/ornithine N-monooxygenase chromosome Plus COG3486

53 basD acinetobactin biosynthesis protein chromosome Plus COG1020

54 basF isochorismate hydrolase chromosome Minus COG1535

55 basG histidine decarboxylase chromosome Minus COG0076

56 basH acinetobactin biosynthesis protein chromosome Minus COG3208

57 basI phosphopantetheinyl transferase component of siderophore

synthetase chromosome Minus COG2977

58 basJ acinetobactin biosynthesis protein chromosome Minus COG1169

59 bauA outer membrane receptor chromosome Minus COG1629

60 bauB enterochelin ABC transporter periplasmic protein chromosome Minus COG4607

61 bauC enterochelin ABC transporter permease chromosome Minus COG4605

62 bauD enterochelin ABC transporter permease chromosome Minus COG4606

63 bauE enterochelin ABC transporter ATPase chromosome Minus COG4604

64 bauF siderophore-interacting protein chromosome Plus COG2375

65 entE enterobactin synthase subunit E chromosome Minus COG1021

66 Regulation BfmRS bfmR response regulator chromosome Plus COG0745

67 bfmS Signal transduction histidine kinase chromosome Plus COG0642

68

Quorom

sensing

(autoinducer-

receptor

mechanism)

abaI N-acyl-L-homoserine lactone synthetase chromosome Minus COG3916

69 abaR DNA-binding HTH domain-containing protein chromosome Plus COG2197





















http://www.mgc.ac.cn/cgi-bin/VFs/vfs.cgi?Genus=Acinetobacter&Keyword=Regulation





403

70 Serum

resistance

PbpG

(Penicillin-

binding

protein)

pbpG D-alanyl-D-alanine carboxypeptidase chromosome Plus COG1686

S-Table.4. 2. Ten virulent proteins localized in outer membrane or extracellular regions predicted by subcellular localization tools.

S.NO Gene Protein CELLO PSORTb CELLO2GO Final Decision

1 OmpA Outer membrane protein-related peptidoglycan-associated (lipo)protein O.M O.M O.M Selected

2 AdeH Multidrug efflux system outer membrane protein O.M O.M O.M Selected

3 Bap Hemolysin-type calcium-binding domain-containing protein E.C E.C E.C Selected

4 CsuA/B CsuA/B E.C U.K E.C Selected

5 CsuB CsuB E.C E.C E.C Selected

6 CsuD P pilus assembly protein, porin PapC O.M O.M O.M Selected

7 CsuE CsuE O.M U.K E.C Selected

8 PgaB Xylanase/chitin deacetylase O.M U.K O.M Selected

9 EpsA EpsA O.M O.M O.M Selected

10 BauA Outer membrane receptor O.M O.M O.M Selected

O.M=Outer membrane, EC=Extracellular, UK=Unknown

http://www.mgc.ac.cn/cgi-bin/VFs/vfs.cgi?Genus=Acinetobacter&Keyword=Serum%20resistance

http://www.mgc.ac.cn/cgi-bin/VFs/vfs.cgi?Genus=Acinetobacter&Keyword=Serum%20resistance


404

S-Table. 4.3. Five proteins prioritized for vaccine candidate identification.

Proteins Cello PSORTb Cello2GO Molecular weight

TMHMM HMMTOP Adhesion probability

(kDa) (SPAAN)

OmpA O.M O.M O.M 38.46 1 1 0.5

CsuA/B EC UK EC 18.71 0 1 0.56

CsuB EC EC EC 19.37 0 0 0.58

EpsA O.M O.M O.M 40.64 0 1 0.5

Ligated-gated

channel

protein

O.M O.M O.M 83.03 0 0 0.62

O.M=Outer membrane, EC=Extracellular, UK=Unknown

S-Table. 4.4. B and T-cells epitopes of prioritized five proteins.

Section: B-cell Epitopes

Ligand-gated channel protein

Position Epitope Score Surface exposed Antigenicity Final prediction

321 GKDIAIPKPPKPETLLNPDW 1 Yes 0.18 Probabble non antigen

510 VPTSTLTKYKKNATTPGAAI 0.999 No 0.38 Probabble non antigen

686 AVSKQYIDQENTLHIPGRTL 0.993 No 0.18 Probabble non antigen

153 TNAASGGWSENYYIRGFESS 0.977 Yes 0.7 Probabble antigen

265 RANAVYRDGSGPVEKQDLKT 0.971 NO 1.17 Probabble antigen

602 VTNIYTSGGEQRNRGIEWSF 0.965 NO 1.46 Probabble antigen

425 WVVNATYYKHNQDDYGVRNV 0.954 No 0.63 Probabble antigen

57 CASNTYAAVIDNSAKTLEQQ 0.953 No 0.58 Probabble antigen

224 TTKRATDEPLTRLTTTYMSD 0.949 No 0.19 Probabble non antigen

11 NNVSYNRNFKTGNDQRINHR 0.945 No 1.72 Probabble antigen

178 MNGLFGITPYYRTSPEMFEK 0.942 Yes 0.28 Probabble non antigen

489 EDRLQLTLGVRYQEVESSNL 0.941 Yes 1.11 Probabble antigen

633 GGASYIEPEVTKTAIQSNEG 0.939 Yes 0.62 Probabble antigen

556 TAANYPTIFAPYKTKQTEFG 0.935 No 0.15 Probabble non antigen

405

449 WTTNLYDPIWGKAVPFNAPL 0.929 Yes 0.46 Probabble antigen

288 SLGMDWHGDRARVSTDLYTS 0.925 Yes 1 Probabble antigen

91 KAEQDDTYAGGQVSSKSSVG 0.921 No 1.39 Probabble antigen

352 IRGEYDLTNNVMGYATYGQS 0.905 No 0.65 Probabble antigen

708 VGARYKTSISNHPITFKADI 0.864 No 0.9 Probabble antigen

115 KTVMETPFNTIAYTDTYIAD 0.843 No 0.27 Probabble non antigen

201 LKGPSALLNGMPPTGSIGGT 0.829 Yes 0.78 Probabble antigen

Protein CsuA/B; putative secreted protein related to type I pili

Position Epitope Score Surface exposed Antigenicity

154 FGAIAPNTGTPKAQGDYKDT 0.999 Yes 0.79

99 AIDGGERTDRTLKNTASADV 0.999 No 1.34

76 ASAATGGNISVTCDGTDPVD 0.987 Yes 1.06

38 STGCTVGGSQTEGNMNKFGT 0.959 Yes 1.68

Protein CsuB

Position Epitope Score Surface exposed Antigenecity Fnal prediction

88 GGENLQNNTRRMKNSSSPNY 0.997 Yes 0.77 Probabble antigen

148 VDLENNNEPHAAGIYKDTVS 0.971 Yes 0.38 Probabble non antigen

64 NVINSKGSWNIRCTESLPVS 0.943 No 0.92 Probabble non antigen

37 NGCSIDNIEQNMDFGKYSAL 0.765 Yes 0.02 Probabble non antigen

Membrane protein

Position Epitope Score Surface exposed Antigencity Final prediction

93 AYPEITPPVSNISNEQSVQA 0.999 Yes 0.57 Probabble antigenic

196 GMAGGVTTTGDNTYIQLIRN 0.997 No 0.78 Probabble antigenic

406

330 SNDVVYVDATGLTRWQRIVN 0.954 No -0.03 Probabble non antigenic

293 SRIYVVRTNPNDRTTEIYHL 0.926 No 0.75 Probabble antigenic

26 QTYDIPSEGVYKTDLGTTVN 0.919 Yes 0.7 Probabble antigenic

171 SVQGSVTKGGQFYLNDQPVS 0.917 Yes 0.71 Probabble antigenic

114 GYPIDQSGYIQFPLVGRYKA 0.907 Yes -0.07 Probabble non antigenic

234 HKLLVQPNDTIYVSTRENQK 0.812 No 0.42 Probabble antigenic

150 ARFLKNPDVVVRVVSYEGQR 0.79 Yes 0.47 Probabble antigenic

407

Section: Selected B-cell Epitopes

Ligand-gated channel protein

Position Epitope Score Surface exposed Antigenicity Final prediction

153 TNAASGGWSENYYIRGFESS 0.977 Yes 0.7 Probabble antigen

489 EDRLQLTLGVRYQEVESSNL 0.941 Yes 1.11 Probabble antigen

633 GGASYIEPEVTKTAIQSNEG 0.939 Yes 0.62 Probabble antigen

449 WTTNLYDPIWGKAVPFNAPL 0.929 Yes 0.46 Probabble antigen

201 LKGPSALLNGMPPTGSIGGT 0.829 Yes 0.78 Probabble antigen

Protein CsuA/B; putative secreted protein related to type I pili

154 FGAIAPNTGTPKAQGDYKDT 0.999 Yes 0.79 Probabble antigen

76 ASAATGGNISVTCDGTDPVD 0.987 Yes 1.06 Probabble antigen

Membrane protein

Position Epitope Score Surface exposed Antigenic score Prediction

203 PVVEVAPVEPTPVTPQPQEL 1 Yes 0.66 Probabble antigenic

337 GSRTVVVQPGQEAAAPAAAQ 1 Yes 0.26 Probabble non-

antigenic

137 VNRGTRGTSEEGTLGNAGVG 0.999 No 1.96 Probabble antigenic

36 QDSQHNNGGKDGNLTNGPEL 0.995 No 2.18 Probabble non-

antigenic

73 LGFEAEYNQVKGDVDGASAG 0.971 Yes 1.69 Probabble antigenic

258 EYPNATARIEGHTDNTGPRK 0.95 No 1.18 Probabble antigenic

314 QPIADNKTKEGRAMNRRVFA 0.925 No 1.08 Probabble antigenic

293 ALVNEYNVDASRLSTQGFAW 0.883 No 0.98 Probabble antigenic

408

38 STGCTVGGSQTEGNMNKFGT 0.959 Yes 1.68 Probabble antigen

Protein CsuB

88 GGENLQNNTRRMKNSSSPNY 0.997 Yes 0.77 Probabble antigen

EpsA

93 AYPEITPPVSNISNEQSVQA 0.999 Yes 0.57 Probabble antigenic

26 QTYDIPSEGVYKTDLGTTVN 0.919 Yes 0.7 Probabble antigenic

171 SVQGSVTKGGQFYLNDQPVS 0.917 Yes 0.71 Probabble antigenic

150 ARFLKNPDVVVRVVSYEGQR 0.79 Yes 0.47 Probabble antigenic

OmpA

203 PVVEVAPVEPTPVTPQPQEL 1 Yes 0.66 Probabble antigenic

73 LGFEAEYNQVKGDVDGASAG 0.971 Yes 1.69 Probabble antigenic

409

Section: T-cell Epitopes

Proteins

Selection of epitopes fulfilling all parameters for vaccine development

B-cell epitopes T-cell MHC-I MHC-II Total Allel

binding

IC50 (nM) value of

epitopes for

DRB1*0101

(MHCPred)

Virulent

Pred Vaxijen

Ligand-gated

channel protein

TNAASGGWSENYYIRGFESS YYIRGFESS 4 2 6 121 1.06 0.39

EDRLQLTLGVRYQEVESSNL VRYQEVESS 2 38 40 95.94 1.06 1.27

GGASYIEPEVTKTAIQSNEG IEPEVTKTA 8 3 11 441.57 1.06 1.18

WTTNLYDPIWGKAVPFNAPL WGKAVPFNA 4 3 7 27.04 1.06 0.29

Protein CsuA/B;

putative

secreted

protein

related to

type I pili

FGAIAPNTGTPKAQGDYKDT FGAIAPNTG 2 5 7 1070.52 1.06 -0.12

ASAATGGNISVTCDGTDPVD VTCDGTDPV 5 5 10 591.16 1.06 -0.13

Protein CsuB GGENLQNNTRRMKNSSSPNY LQNNTRRMK 1 16 17 90.36 1.06 1.32

Membrane

protein

QTYDIPSEGVYKTDLGTTVN YKTDLGTTV 6 2 8 65.01 1.06 0.82

SVQGSVTKGGQFYLNDQPVS FYLNDQPVS 4 13 17 1.36 1.05 0.8

membrane

protein LGFEAEYNQVKGDVDGASAG FEAEYNQVK 3 3 6 467 1.05 1.25

410

Section: Final List

Proteins B-cell epitopes (BCPreds ) T-cell MHC-I MHC-II

Total

Allel

binding

IC50 (nM) value

of epitopes for

DRB1*0101

(MHCPred)

Virulent

Pred Vaxijen Allergenicity Score

Ligand-gated

channel

protein

EDRLQLTLGVRYQEVESSNL VRYQEVESS 2 38 40 95.94 1.06 1.27 Non-allergen 0.263

Protein CsuB GGENLQNNTRRMKNSSSPNY LQNNTRRMK 1 16 17 90.36 1.06 1.32 Non-allergen 0.263

Membrane

protein SVQGSVTKGGQFYLNDQPVS FYLNDQPVS 4 13 17 1.36 1.05 0.8 Non-allergen 0.261

411

S-Table.4.5. Characterization of selected B-cell epitopes of the ligated gated channel, CsuB and EpsA protein.

Section: Ligated gated channel protein

Analysis Position Residue Score Average Minimum Maximum

Emini Surface Accessibility Prediction Prediction

3 R 0.781

1 0.314 3.397

4 L 0.335

5 Q 0.393

6 L 0.314

7 T 0.66

8 L 0.66

9 G 0.594

10 V 0.712

11 R 1.158

12 Y 1.568

13 Q 3.397

14 E 1.43

4 L 1.256

5 Q -0.022

6 L -0.667

7 T -1.344

8 L 0.344

9 G 0.544

10 V 1.156

11 R 1.444

12 Y 3.189

13 Q 3.278

14 E 4.467

15 V 2.978

Chou & Fasman Beta-Turn Prediction Prediction 4 L 0.936 0.922 0.852 1.052

412

5 Q 0.909

6 L 0.852

7 T 0.873

8 L 0.917

9 G 0.89

10 V 0.88

11 R 0.856

12 Y 0.949

13 Q 0.934

14 E 1.052

15 V 1.012

Kolaskar & Tongaonkar Antigenicity Prediction

4 L 1.015

1.058 0.993 1.107

5 Q 1.074

6 L 1.075

7 T 1.107

8 L 1.081

9 G 1.063

10 V 1.078

11 R 1.071

12 Y 1.045

13 Q 1.06

14 E 0.993

15 V 1.035

Karplus & Schulz Flexibility Prediction Prediction

4 L 0.977

0.978 0.944 1.065

5 Q 0.954

6 L 0.946

7 T 0.944

8 L 0.945

9 G 0.946

10 V 0.951

11 R 0.959

413

12 Y 0.973

13 Q 0.993

14 E 1.015

15 V 1.042

16 E 1.065

Section: CsuB protein



3 E 0.325

1 0.325 2.067

4 N 0.528

5 L 0.858

6 Q 0.715

7 N 0.87

8 N 2.067

9 T 1.181

10 R 1.469

11 R 1.469

12 M 1.364

13 K 0.933

14 N 0.639

15 S 0.998

16 S 0.802

17 S 0.782

Parker Hydrophilicity Prediction Prediction

4 N 4.286

4.358 3.486 5.900

5 L 4.471

6 Q 4.4

7 N 3.886

8 N 3.486

9 T 4.2

10 R 4.157

11 R 4.157

12 M 4.086

414

13 K 4.271

14 N 4.6

15 S 4.3

16 S 5.9

17 S 4.814

Chou & Fasman Beta-Turn Prediction Prediction

4 N 1.221

1.187 1.066 1.439

5 L 1.221

6 Q 1.136

7 N 1.166

8 N 1.079

9 T 1.08

10 R 1.084

11 R 1.084

12 M 1.066

13 K 1.133

14 N 1.201

15 S 1.283

16 S 1.42

17 S 1.439


4 N 0.917

0.907 0.852 0.973

5 L 0.903

6 Q 0.908

7 N 0.911

8 N 0.925

9 T 0.864

10 R 0.852

11 R 0.852

12 M 0.886

13 K 0.9

14 N 0.92

15 S 0.947

16 S 0.94

17 S 0.973

Karplus & Schulz Flexibility Prediction Prediction 4 N 1.05 1.069 1.038 1.133

415

5 L 1.041

6 Q 1.049

7 N 1.064

8 N 1.073

9 T 1.064

10 R 1.049

11 R 1.038

12 M 1.042

13 K 1.07

14 N 1.103

15 S 1.127

16 S 1.133

Section: EpsA protein



3 Q 0.405

1 0.405 2.220

4 G 0.436

5 S 1.176

6 V 0.672

7 T 0.672

8 K 0.868

9 G 1.013

10 G 1.1

11 Q 0.453

12 F 0.737

13 Y 1.243

14 L 1.243

15 N 2.22

16 D 1.052

17 Q 1.709

416

Parker Hydrophilicity Prediction Prediction

4 G 3.214

2.322 0.400 4.443

5 S 3.1

6 V 4.443

7 T 4.4

8 K 4.443

9 G 2.2

10 G 2.457

11 Q 0.4

12 F 0.586

13 Y 1.2

14 L 1.243

15 N 0.686

16 D 1.471

17 Q 2.671

Chou & Fasman Beta-Turn Prediction Prediction

4 G 1.051

1.103 0.991 1.226

5 S 0.991

6 V 1.143

7 T 1.226

8 K 1.143

9 G 1.024

10 G 1.116

11 Q 1.063

12 F 1.141

13 Y 1.127

14 L 1.044

15 N 1.121

16 D 1.107

17 Q 1.149

417


4 G 1.084

1.025 0.979 1.084

5 S 1.072

6 V 1

7 T 0.979

8 K 1

9 G 1.011

10 G 0.979

11 Q 1.028

12 F 1.006

13 Y 1.005

14 L 1.025

15 N 1.032

16 D 1.074

17 Q 1.052

Karplus & Schulz Flexibility Prediction Prediction

4 G 1.059

1.031 0.951 1.083

5 S 1.05

6 V 1.054

7 T 1.068

8 K 1.08

9 G 1.083

10 G 1.061

11 Q 1.018

12 F 0.975

13 Y 0.951

14 L 0.966

15 N 1

16 D 1.035

418

Section: Final peptide

B-Cell epitope

Accessibility, hydrophilicity, flexibility,Beta-

turn prediction and antigencity B-Cell

epitope

T-Cell epitope Allergenicity Allergenicity Score VirulentPred

SVQGSVTKGGQFYLNDQPVS FYLNDQ FYLNDQPVS Non-allergen 0.263 1.06

GGENLQNNTRRMKNSSSPNY LQNNTRRMK LQNNTRRMK Non-allergen 0.263 1.06

EDRLQLTLGVRYQEVESSNL VRYQE VRYQEVESS Non-allergen 0.263 1.06

419

S-Table.4. 7. Structure evaluation of models generated by different online tools. The evaluation was done on the basis of Procheck,

Errat, Verify-3D, and Prosa- Z score.

Structure

Source

G Factor

ERRAT

Quality

Factor

VERIFY-

3D -3D (3D-

1D score >

0.2 (%))

ProSA-

Web

Z-

score

Number of residues

CsuB protein Most favored regions Additionally

allowed region

Generously

allowed

regions

Disallowed regions

ModWeb 98 14 5 3 -0.24 18.254 80.6 -4.67

I-TASSER 81 61 10 4 -0.76 82.716 55.23 -3.22

Phyre2 79 18 1 2 -1.16 24.52 87.72 -1.21

Swiss model 104 15 2 0 -0.38 72.034 82.35 -5.4

EpsA

ModWeb 272 23 2 2 -0.06 74.627 70.06 -7.36

I-TASSER 246 62 8 5 -0.42 87.955 67.76 -7.17

Phyre2 257 36 2 1 -0.36 72.72 70.97 -7.2

Swiss model 2114 284 18 8 -0.17 90.845 70.03 -7.38

420

S-Table. 4.8. Five potential prioritized vaccine proteins against M. pneumonia.

Gene Symbol

Locus

Tag

Localization

(Probability)

Adhesin

Probability

Trans-

membrane

helices

Similar

Human

Protein

Protein Length

sp|P75568|Y101_MYCPN MPN_101 Uncharacterized protein MPN_101

[Mycoplasma pneumoniae (strain ATCC

29342 / M129)]

Outer Membrane

(Prob.=0.949)

0.610 0 428

sp|P75267|Y131_MYCPN MPN_131

Putative adhesin P1-like protein

MPN_131 [Mycoplasma pneumoniae

(strain ATCC 29342 / M129)]

Extracellular

(Prob.=0.964) 0.745 0 221


Putative adhesin P1-like protein

MPN_132 [Mycoplasma pneumoniae

(strain ATCC 29342 / M129)]

Extracellular

(Prob.=0.964) 0.645 0 256


Uncharacterized protein MPN_205


29342 / M129)]

Extracellular

(Prob.=0.964) 0.651 0 438


Uncharacterized protein MPN_502


29342 / M129)]

Outer

Membrane

(Prob.=0.949)

0.619 0 422

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_gene_symbol&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_locus_tag&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_locus_tag&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_Final_Localization&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_Final_Score&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_SPAAN&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_SPAAN&order=DESC

http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_tmhmm_PredHel&order=DESC



http://www.violinet.org/vaxign/query/show_results.php?currPage=1&c_spaan_score_value=&c_spaan_score_check=&c_spaan_score_opt=&c_tmhmm_PredHel_opt=&c_tmhmm_PredHel_value=&c_tmhmm_PredHel_check=&c_query_id=CQNFP82XD5&c_hit_p_value=0.05&c_human_alignment=&c_mouse_alignment=&keywords=&keywords_type=&query_seq_id_type=&query_seq_ids=&have_num_orthologs=&show_ortholog=&recordsPerPage=50&orderby=c_protein_length&order=DESC

http://www.violinet.org/vaxign/query/show_one_sequence.php?c_query_id=CQNFP82XD5&c_sequence_id=4591443&c_hit_p_value=0.05&mhc_species=&mhc_allele=&epitope_length=





421

Chapter # 5

S-Fig.5.1. Structure evaluation plots for KdsA enzyme (a) Ramachandran plot, (b) Errat, (c) Verify-3D, (d) Pros-SA.

422

S-Fig.5.2. DOPE score KdsA template and KdA enzyme.

423

S-Fig.5.3. Binding pockets revealed by GHECOM for KdsA enzyme.

424

S-Fig.5.4. Potential binding cavities of KdsA revealed by fpocket. Residues of the most active pocket are shown.

425

S-Fig.5.5. Multiple sequence alignment for KdsA enzyme and its orthologues.

426

S-Fig.5.6. (A) 3D alignment of KdsA from 4 different organisms: A. baumannii KdsA (cyan), P. aeruginosa (blue), V. cholerae (Gold),

N. meningitidis (dark cyan). (B) 3D conservation of active site residues (Gln, Leu, Pro, Ala, Phe, Leu). For clarity, A. baumannii active

site residues are only shown.

427

S-Fig.5.7. (A) Superimposition of all the predicted models for KdsA: Modeller-1 (yellow), Modeller-2 (dark khaki), Modeller-3 (cyan),

Modeller-4 (forest green),Modeller-5(cornflower blue), Phyre-2 (magenta), Swiss-Model (chartreuse), Mod-web (blue), and I-Tasser

(olive drab). (B) Superimposition of active site residues critical for inhibition. For clarity, ribbon and residues number of Modeller-1 are

only shown.

428

S-Table.5.1. Number of proteins screened out at each step of subtractive proteomics.

MD

R-T

J

LA

C-4

MD

R-

ZJ

06

AB

50

75

Ab

H1

20

-

A2

Z

W8

5-1

A1

AC

ICU

AB

30

7-

02

94

D1

27

97

79

16

56

-2

TY

TH

-1

BJ

AB

07

1

04

BJ

AB

07

1

5

BJ

AB

08

6

8

AC

29

AB

03

1

AB

03

0

62

00

NC

GM

23

7

IOM

TU

43

3

XH

38

6

Ab

04

-mff

AT

CC

17

978

-mff

D3

6

KB

N1

0P

0

21

43

YU

-R6

12

XH

86

0

XH

85

9

XH

85

7

XH

85

6

CIP

70

.10

R2

09

1

R2

09

0

XH

85

8

Total Proteome

38

32

34

99

36

42

37

65

35

13

35

43

35

73

36

00

33

63

33

47

36

52

35

45

37

05

37

09

36

50

35

55

34

35

38

70

37

10

36

44

38

45

38

73

38

44

36

60

37

76

37

45

38

14

34

69

35

61

34

85

35

57

35

50

35

66

34

85

37

93

Non Paralogs

37

21

33

97

35

24

36

63

34

45

34

76

34

87

35

10

32

98

32

72

35

26

34

55

36

07

36

06

35

73

34

61

33

50

36

75

36

34

35

13

37

17

37

32

37

22

35

75

36

86

35

69

36

56

33

80

34

64

34

10

34

46

34

73

34

85

33

78

36

51

Non Homologs

31

42

28

27

29

51

30

90

28

71

28

93

29

16

29

43

27

33

27

04

29

56

28

82

30

34

30

39

29

99

28

88

27

96

30

99

30

61

29

39

31

28

31

53

31

51

29

98

30

83

29

98

30

83

28

25

28

94

28

44

28

71

28

83

28

95

28

03

30

78

Essential Proteins 81

1

78

3

80

0

81

4

80

5

81

2

81

1

79

1

79

3

78

5

79

9

79

1

80

1

79

8

80

3

79

9

79

9

80

4

81

5

79

9

81

8

81

6

79

8

80

7

80

1

80

0

80

1

79

8

79

2

79

7

80

0

83

0

83

3

79

9

80

2

Metabolic pathways

proteins 22

9

22

9

22

4

22

5

22

3

22

9

22

6

22

1

22

3

22

5

22

1

22

2

22

4

22

4

22

4

22

4

22

5

22

6

23

0

22

4

22

7

22

5

22

6

22

4

22

3

22

4

22

3

22

4

22

4

22

5

22

4

22

7

22

7

22

2

22

5

Metabolic pathways

12

4

12

6

12

0

12

5

11

9

12

0

12

5

11

8

12

3

11

7

11

9

12

1

12

0

12

5

12

1

12

0

12

3

11

8

12

3

12

0

12

5

12

0

12

5

12

0

11

8

11

9

11

8

12

0

12

0

11

9

11

9

12

5

12

6

11

6

12

5


































429

Unique Pathways 42

42

39

41

39

40

41

38

39

37

38

39

38

41

39

39

39

39

41

39

43

39

40

39

38

38

38

39

39

39

39

39

41

36

41

Essential Proteins involved

in unique pathways 80

77

79

81

80

82

81

78

75

78

78

80

78

79

79

79

77

80

85

78

81

80

80

83

77

75

77

77

77

79

76

81

82

79

79

Essential,Unique and

Conserved Cytoplasmic

Proteins

32

Virulent Proteins 15

Physicochemicaly favored

proteins 10

Potential drug targets 10

430

S-Table.5.2. List of 15 virulent proteins extracted through VFDB analysis.

Protein identity Bit score Mw (KDa)

ompR 45 171 28.82

LpxA 100 531 28.29

RstA 90 1073 27.14

GlnG 39 306 56.23

LpxC 100 597 32.94

KdpE 39 159 27.21

UvrY 58 244 23.15

PhoB 37 103 26.65

KdsB 49 215 28.4

LpxD 100 705 38.41

KdsD 44 244 35.02

GspE 64 600 55.47

KdtA 38 273 49.07

LpxB 99 764 43.82

KdsA 81 478 31.54

431

S-Table.5.3. Molecular weight prioritization of virulent proteins.

Protein Mw (KDa)

ompR 28.82

LpxA 28.29

RstA 27.14

GlnG 56.23

LpxC 32.94

KdpE 27.21

UvrY 23.15

PhoB 26.65

KdsB 28.4

LpxD 38.41

KdsD 35.02

GspE 55.47

KdtA 49.07

LpxB 43.82

KdsA 31.54

432

S-Table.5. 4. Physicochemical characterization of virulent proteins for potential drug targets.

KdsA

Number of amino acids: 285

Molecular weight: 31549.56

Theoretical pI: 6.37

Amino acid composition:

Ala (A) 27 9.5%

Arg (R) 12 4.2%

Asn (N) 9 3.2%

Asp (D) 19 6.7%

Cys (C) 4 1.4%

Gln (Q) 12 4.2%

Glu (E) 16 5.6%

Gly (G) 20 7.0%

His (H) 10 3.5%

Ile (I) 18 6.3%

Leu (L) 33 11.6%

Lys (K) 20 7.0%

Met (M) 9 3.2%

Phe (F) 16 5.6%

Pro (P) 15 5.3%

Ser (S) 13 4.6%

Thr (T) 11 3.9%

Trp (W) 1 0.4%

Tyr (Y) 4 1.4%

Val (V) 16 5.6%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

433

(Z) 0 0.0%

(X) 0 0.0%

Total number of negatively charged residues (Asp + Glu): 35

Total number of positively charged residues (Arg + Lys): 32

Atomic composition:

Carbon C 1417

Hydrogen H 2251

Nitrogen N 383

Oxygen O 405

Sulfur S 13

Formula: C1417H2251N383O405S13

Total number of atoms: 4469

Extinction coefficients:

Extinction coefficients are in units of M-1 cm-1, at 280 nm measured in water.

Ext. coefficient 11710

Abs 0.1% (=1 g/l) 0.371, assuming all pairs of Cys residues form cystines


Abs 0.1% (=1 g/l) 0.363, assuming all Cys residues are reduced

Estimated half-life:

The N-terminal of the sequence considered is M (Met).

The estimated half-life is: 30 hours (mammalian reticulocytes, in vitro).

>20 hours (yeast, in vivo).

>10 hours (Escherichia coli, in vivo).

Instability index:

The instability index (II) is computed to be 38.54

This classifies the protein as stable.

434

GspE





Ala (A) 39 7.8%

Arg (R) 39 7.8%

Asn (N) 14 2.8%

Asp (D) 30 6.0%

Cys (C) 5 1.0%

Gln (Q) 28 5.6%

Glu (E) 38 7.6%

Gly (G) 31 6.2%

His (H) 16 3.2%

Ile (I) 30 6.0%

Leu (L) 60 12.1%

Lys (K) 18 3.6%

Met (M) 13 2.6%

Phe (F) 11 2.2%

Pro (P) 19 3.8%

Ser (S) 28 5.6%

Thr (T) 31 6.2%

Trp (W) 1 0.2%

Tyr (Y) 9 1.8%

Val (V) 37 7.4%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

435

(X) 0 0.0%



Atomic composition:

Carbon C 2422

Hydrogen H 3955

Nitrogen N 707

Oxygen O 744

Sulfur S 18














Instability index:



Aliphatic index: 100.06

Grand average of hydropathicity (GRAVY): -0.251

436

LpxD





Ala (A) 28 7.9%

Arg (R) 12 3.4%

Asn (N) 21 5.9%

Asp (D) 19 5.3%

Cys (C) 6 1.7%

Gln (Q) 13 3.7%

Glu (E) 18 5.1%

Gly (G) 33 9.3%

His (H) 16 4.5%

Ile (I) 35 9.8%

Leu (L) 30 8.4%

Lys (K) 20 5.6%

Met (M) 3 0.8%

Phe (F) 10 2.8%

Pro (P) 5 1.4%

Ser (S) 26 7.3%

Thr (T) 23 6.5%

Trp (W) 2 0.6%

Tyr (Y) 7 2.0%

Val (V) 29 8.1%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

437

(Z) 0 0.0%

(X) 0 0.0%



Atomic composition:

Carbon C 1691

Hydrogen H 2720

Nitrogen N 480

Oxygen O 521

Sulfur S 9














Instability index:





438

KdsB





Ala (A) 21 8.3%

Arg (R) 17 6.7%

Asn (N) 7 2.8%

Asp (D) 22 8.7%

Cys (C) 4 1.6%

Gln (Q) 11 4.3%

Glu (E) 19 7.5%

Gly (G) 11 4.3%

His (H) 7 2.8%

Ile (I) 14 5.5%

Leu (L) 29 11.5%

Lys (K) 12 4.7%

Met (M) 6 2.4%

Phe (F) 6 2.4%

Pro (P) 14 5.5%

Ser (S) 14 5.5%

Thr (T) 9 3.6%

Trp (W) 2 0.8%

Tyr (Y) 5 2.0%

Val (V) 23 9.1%

439

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

(X) 0 0.0%



Atomic composition:

Carbon C 1251

Hydrogen H 2014

Nitrogen N 350

Oxygen O 382

Sulfur S 10














440

Instability index:



PhoB





Ala (A) 19 8.1%

Arg (R) 22 9.3%

Asn (N) 5 2.1%

Asp (D) 22 9.3%

Cys (C) 2 0.8%

Gln (Q) 10 4.2%

Glu (E) 13 5.5%

Gly (G) 17 7.2%

His (H) 5 2.1%

Ile (I) 16 6.8%

Leu (L) 26 11.0%

Lys (K) 8 3.4%

Met (M) 8 3.4%

Phe (F) 8 3.4%

Pro (P) 9 3.8%

Ser (S) 9 3.8%

Thr (T) 12 5.1%

Trp (W) 2 0.8%

Tyr (Y) 6 2.5%

Val (V) 17 7.2%

Pyl (O) 0 0.0%

441

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

(X) 0 0.0%



Atomic composition:

Carbon C 1175

Hydrogen H 1889

Nitrogen N 337

Oxygen O 349

Sulfur S 10














Instability index:





442

UvrY





Ala (A) 19 9.0%

Arg (R) 10 4.7%

Asn (N) 4 1.9%

Asp (D) 11 5.2%

Cys (C) 2 0.9%

Gln (Q) 11 5.2%

Glu (E) 16 7.6%

Gly (G) 14 6.6%

His (H) 4 1.9%

Ile (I) 17 8.1%

Leu (L) 19 9.0%

Lys (K) 11 5.2%

Met (M) 8 3.8%

Phe (F) 4 1.9%

Pro (P) 8 3.8%

Ser (S) 13 6.2%

Thr (T) 11 5.2%

Trp (W) 0 0.0%

Tyr (Y) 7 3.3%

Val (V) 22 10.4%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

443

(X) 0 0.0%



Atomic composition:

Carbon C 1024

Hydrogen H 1671

Nitrogen N 275

Oxygen O 312

Sulfur S 10




This protein does not contain any Trp residues. Experience shows that

this could result in more than 10% error in the computed extinction coefficient.











Instability index:




Grand average of hydropathicity (GRAVY): 0.064

444

KdpE





Ala (A) 13 5.5%

Arg (R) 15 6.3%

Asn (N) 5 2.1%

Asp (D) 15 6.3%

Cys (C) 1 0.4%

Gln (Q) 19 8.0%

Glu (E) 16 6.7%

Gly (G) 13 5.5%

His (H) 3 1.3%

Ile (I) 16 6.7%

Leu (L) 36 15.1%

Lys (K) 14 5.9%

Met (M) 3 1.3%

Phe (F) 4 1.7%

Pro (P) 12 5.0%

Ser (S) 7 2.9%

Thr (T) 17 7.1%

Trp (W) 3 1.3%

Tyr (Y) 9 3.8%

Val (V) 17 7.1%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

445

(X) 0 0.0%



Atomic composition:

Carbon C 1226

Hydrogen H 1986

Nitrogen N 330

Oxygen O 358

Sulfur S 4














Instability index:





446

LpxC





Ala (A) 23 7.7%

Arg (R) 8 2.7%

Asn (N) 16 5.3%

Asp (D) 25 8.3%

Cys (C) 3 1.0%

Gln (Q) 10 3.3%

Glu (E) 15 5.0%

Gly (G) 23 7.7%

His (H) 8 2.7%

Ile (I) 26 8.7%

Leu (L) 27 9.0%

Lys (K) 17 5.7%

Met (M) 7 2.3%

Phe (F) 18 6.0%

Pro (P) 13 4.3%

Ser (S) 16 5.3%

Thr (T) 12 4.0%

Trp (W) 0 0.0%

Tyr (Y) 8 2.7%

Val (V) 25 8.3%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

447

(X) 0 0.0%



Atomic composition:

Carbon C 1484

Hydrogen H 2323

Nitrogen N 383

Oxygen O 443

Sulfur S 10




This protein does not contain any Trp residues. Experience shows that

this could result in more than 10% error in the computed extinction coefficient.











Instability index:





448

LpxA





Ala (A) 17 6.5%

Arg (R) 8 3.1%

Asn (N) 16 6.1%

Asp (D) 12 4.6%

Cys (C) 5 1.9%

Gln (Q) 12 4.6%

Glu (E) 12 4.6%

Gly (G) 28 10.7%

His (H) 15 5.7%

Ile (I) 31 11.8%

Leu (L) 18 6.9%

Lys (K) 11 4.2%

Met (M) 5 1.9%

Phe (F) 7 2.7%

Pro (P) 7 2.7%

Ser (S) 20 7.6%

Thr (T) 10 3.8%

Trp (W) 2 0.8%

Tyr (Y) 5 1.9%

Val (V) 21 8.0%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

449

(X) 0 0.0%



Atomic composition:

Carbon C 1247

Hydrogen H 1985

Nitrogen N 357

Oxygen O 374

Sulfur S 10














Instability index:





450

OmpR





Ala (A) 17 6.7%

Arg (R) 29 11.4%

Asn (N) 6 2.4%

Asp (D) 19 7.5%

Cys (C) 1 0.4%

Gln (Q) 11 4.3%

Glu (E) 18 7.1%

Gly (G) 14 5.5%

His (H) 4 1.6%

Ile (I) 12 4.7%

Leu (L) 31 12.2%

Lys (K) 5 2.0%

Met (M) 6 2.4%

Phe (F) 8 3.1%

Pro (P) 14 5.5%

Ser (S) 13 5.1%

Thr (T) 14 5.5%

Trp (W) 3 1.2%

Tyr (Y) 3 1.2%

Val (V) 26 10.2%

Pyl (O) 0 0.0%

Sec (U) 0 0.0%

(B) 0 0.0%

(Z) 0 0.0%

451

(X) 0 0.0%



Atomic composition:

Carbon C 1270

Hydrogen H 2066

Nitrogen N 374

Oxygen O 376

Sulfur S 7














Instability index:



452

S-Table.5.5. Potential ligand binding sites predicted by Meta pocket for KdsA enzyme.

The potential 2 ligand binding sites in your protein:

HEADER binding site ID: 1

ASP208 SER209 GLY211 ARG213 THR202

SER207 ALA210 GLY212 ASP241 LYS244

GLN201 GLN216 LEU200 ARG214 ARG252

SER249 ASP239 PRO240 ALA250 ILE217

LEU251 GLN255 LEU253 THR197 LYS242

HIS237 PRO238 LEU234 GLU235 SER236

LEU256 PHE259 VAL196 THR218 ALA221

LEU232 LEU175 VAL176 VAL177 LEU220

ASN174 ALA199 PRO203 ARG206 ALA215

HIS198 GLY247 PRO248 ALA243 ASP246

SER254 ASN24 MET23 VAL25 GLU27


THR197 ILE217 LEU234 GLU235 SER236

LEU253 ARG214 ARG252 HIS237 ARG213

GLY212 ALA215 GLY211 ASP239 ASP241

LEU200 LEU251 GLN255 SER249 ALA250

PRO240 LYS244 ALA210 SER209 HIS198

SER207 ASN24 GLN201 THR202 ASP208

LYS58 LYS53 ASN60 ARG166 PHE115

GLN139 SER55 ALA59 ARG61 ASP195

GLN111 ASP93 LYS136 ASP57 HIS95

ALA114 PRO113 PHE233

453

Chapter # 6

S-Fig.6.1. Gram-negative bacteria envelope.

454

S-Fig.6.2. Four sequential steps of KDO biosynthesis.

455

S-Fig.6.3. Binding cavity prediction through GHECOM (A) and Docsite Scorer (B).

456

S-Fig.6.4. Binding interactions (Left) and binding mode (Right) of the top docked drug-like compound (shown in gold color) in the

catalytic pocket of KdsB enzyme (shown in dodger blue in color).

457

S-Fig.6.5. Total, bound and free binding energy decomposed per iteration of Waterswap assay.

458

S-Table.6.1. Hydrogen bonds between KdsB enzyme and the screened inhibitor at different time scale of simulation protocol.

At zero nanosecond

Distance

in

Angstrom

At 50 ns

Distance

in

Angstrom

At 100 ns

Distance

in

Angstrom

At 150 ns

Distance

in

Angstrom

At 200ns

Distance

in

Angstrom

LIG N1 : 236 ASP OD2 4.0 LIG O1 : 184 GLY N 3.0 LIG O1 : 98 GLN HE22 3.9 LIG N1 : 236 ASP HB2 3.6 LIG N1 : 236 ASP OD2 3.4

LIG N2 : 236 ASP HB2 3.8 LIG O1 : 184 GLY H 2.6 LIG O2 : 234 GLY H 3.7 LIG N1 : 236 ASP OD1 2.8 LIG N1 : 236 ASP OD1 3.8

LIG H15 : 236 ASP OD2 3.7 LIG N1 : 184 GLY H 2.9 LIG H15 : 236 ASP OD1 1.8 LIG H15 : 236 ASP OD1 3.3

LIG H15 : 236 ASP OD1 3.4 LIG N1 : 98 GLN NE2 3.4 LIG H15 : 236 ASP OD2 3.4 LIG H15 : 236 ASP OD2 2.4

LIG O2 : 236 ASP HB2 3.4 LIG N1 : 98 GLN HE22 2.6 LIG H15 : 236 ASP HB2 3.2 LIG H15 : 14 ARG HH21 3.3

LIG O2 : 236 ASP HB3 3.3 LIG N1 : 184 GLY N 3.5 LIG H15 : 14 ARG NH1 3.5 LIG O2 : 234 GLY H 3.8

LIG O2 : 14 ARG HH21 3.6 LIG N1 : 98 GLN HE21 3.8 LIG HN : 141 SER O 3.9

LIG O2 : 14 ARG HH22 3.3 LIG N2 : 101 GLU OE1 2.9

LIG O2 : 14 ARG NH2 3.8 LIG H15 : 184 GLY H 3.9

LIG H15 : 98 GLN OE1 3.6

LIG H15 : 98 GLN NE2 3.0

LIG H15 : 98 GLN HE21 3.6

LIG H15 : 98 GLN HE22 2.0

LIG O2 : 235 VAL O 3.9

LIG O3 : 186 TYR OH 3.4

LIG O3 : 186 TYR HH 2.8

LIG HN : 101 GLU OE1 1.9

LIG HN : 101 GLU OE2 3.3

459

Chapter # 7

S-Fig.7.1. The reaction mechanism of Ddl enzyme.

460

S-Fig.7.2. DoGSiteScorer predicted binding cavity for Ddl enzyme.

461

S-Fig.7.3. PKa calculations for the compound-331.

462

S-Fig.7.4. Titration curve for the entire DDl molecule.

463

S-Fig.7.5. MM/GBSA free energy versus number of frames.

464

S-Table.7.1. The potential ligand binding sites in Ddl enzyme predicted by Metapocket.


RESI GLN105 TYR117 ARG118 ILE119 GLN108

RESI PRO116 ALA115 ILE120 VAL172 VAL101

RESI VAL173 MET174 LYS100 GLU176 ASP99

RESI LYS104 LEU267 GLU268 VAL269 ASN270

RESI GLU71 GLY97 GLU20 THR271 MET98

RESI THR103 GLY67 HIS66 ARG68 SER18

RESI ALA19


RESI LYS100 GLU176 LEU267 GLU268 GLU183

RESI MET257 LYS104 MET174 ILE179 ALA115

RESI TRP178 LYS177 TRP265 VAL101 THR180

465

S-Table.7.2. List of all computed pKs for the DDl enzyme.

Residue pKint pK_(1/2)

NTSER-1 7.601 7.553

LYS-5 10.176 >12.000000

LYS-8 10.391 >12.000000

LYS-16 10.099 11.154

GLU-19 3.601 2.720

ARG-20 11.706 >12.000000

ASP-25 3.822 1.611

ASP-32 4.533 2.832

ARG-36 11.620 >12.000000

GLU-42 4.940 2.059

ASP-45 3.425 <0.000000

ASP-48 4.638 4.201

ARG-49 12.205 >12.000000

GLU-53 4.202 3.441

TYR-57 12.897 >12.000000

ASP-58 4.668 3.233

ARG-59 13.069 >12.000000

HID-65 3.929 5.906

ARG-67 11.447 >12.000000

GLU-70 2.105 <0.000000

ASP-71 2.448 1.086

GLU-79 3.754 3.737

TYR-85 11.197 >12.000000

ASP-98 4.556 <0.000000

LYS-99 10.309 >12.000000

LYS-101 9.269 >12.000000

LYS-103 10.415 >12.000000

ASP-110 4.264 3.605

TYR-116 12.165 11.982

ARG-117 11.938 >12.000000

LYS-121 10.068 >12.000000

GLU-122 4.260 4.192

ASP-124 2.493 2.236

ASP-126 3.641 4.235

GLU-131 4.779 3.043

LYS-139 9.411 >12.000000

HID-142 5.765 7.805

GLU-143 4.293 3.145

LYS-150 9.807 >12.000000

GLU-152 4.855 2.645

LYS-153 9.900 >12.000000

GLU-155 4.507 2.822

ASP-156 3.346 3.302

GLU-162 5.213 2.858

LYS-163 9.532 10.351

HIS-167 5.485 3.890

ASP-168 3.732 <0.000000

GLU-174 8.615 0.618

LYS-175 11.558 >12.000000

ARG-180 12.595 >12.000000

GLU-181 4.291 3.888

ARG-196 13.309 >12.000000

TYR-199 10.462 9.750

CYS-203 9.032 10.516

GLU-207 4.293 2.069

GLU-209 3.440 2.599

GLU-210 3.907 3.438

LYS-211 10.105 >12.000000

LYS-212 10.151 10.483

CYS-217 11.349 >12.000000

ARG-219 12.515 >12.000000

GLU-227 4.503 4.712

ARG-231 10.536 >12.000000

ASP-233 4.744 0.657

ASP-237 2.198 1.337

GLU-238 3.781 3.970

GLU-246 6.841 <0.000000

466

HIS-256 6.113 7.029

LYS-261 11.183 >12.000000

LYS-264 10.241 11.032

TYR-268 10.229 10.916

ASP-271 3.669 2.140

GLU-272 3.321 3.239

CYS-274 11.835 >12.000000

GLU-279 5.440 2.718

CTLEU-282 4.737 4.624

Site NTSER-1

6.399998 0.934210 NTSER-1

6.599998 0.899671 NTSER-1

6.799998 0.849893 NTSER-1

6.999998 0.781422 NTSER-1

7.199997 0.692851 NTSER-1

7.399997 0.587341 NTSER-1

7.599997 0.473147 NTSER-1

7.799997 0.361693 NTSER-1

7.999997 0.263368 NTSER-1

8.199997 0.184070 NTSER-1

8.399997 0.124608 NTSER-1

8.599997 0.082454 NTSER-1

8.799996 0.053688 NTSER-1

Site LYS-5

11.799994 0.942982 LYS-5

11.999993 0.915822 LYS-5

Site LYS-8

11.599994 0.929056 LYS-8

11.799994 0.898995 LYS-8

11.999993 0.861667 LYS-8

Site LYS-16

9.999995 0.932674 LYS-16

10.199995 0.897501 LYS-16

10.399995 0.847035 LYS-16

10.599995 0.778063 LYS-16

10.799994 0.689741 LYS-16

10.999994 0.585658 LYS-16

11.199994 0.474312 LYS-16

11.399994 0.366820 LYS-16

11.599994 0.272760 LYS-16

11.799994 0.197067 LYS-16

11.999993 0.139967 LYS-16

Site GLU-19

0.600000 0.939002 GLU-19

0.800000 0.921141 GLU-19

1.000000 0.898183 GLU-19

1.200000 0.869729 GLU-19

1.400000 0.835392 GLU-19

1.600000 0.796353 GLU-19

1.800000 0.751201 GLU-19

2.000000 0.701413 GLU-19

2.200000 0.648006 GLU-19

2.400000 0.592093 GLU-19

2.600000 0.534777 GLU-19

2.800000 0.476999 GLU-19

3.000000 0.419447 GLU-19

3.200001 0.362722 GLU-19

3.400001 0.307741 GLU-19

3.600001 0.255884 GLU-19

3.800001 0.208643 GLU-19

4.000000 0.167036 GLU-19

4.200000 0.131322 GLU-19

4.400000 0.101245 GLU-19

4.600000 0.076454 GLU-19

4.800000 0.056659 GLU-19

Site ARG-20

Site ASP-25

0.200000 0.940865 ASP-25

0.400000 0.913606 ASP-25

0.600000 0.875864 ASP-25

0.800000 0.827085 ASP-25

467

1.000000 0.763862 ASP-25

1.200000 0.689313 ASP-25

1.400000 0.600839 ASP-25

1.600000 0.505401 ASP-25

1.800000 0.410266 ASP-25

2.000000 0.322555 ASP-25

2.200000 0.247014 ASP-25

2.400000 0.185336 ASP-25

2.600000 0.136856 ASP-25

2.800000 0.099673 ASP-25

3.000000 0.071618 ASP-25

3.200001 0.050741 ASP-25

Site ASP-32

1.200000 0.941716 ASP-32

1.400000 0.918048 ASP-32

1.600000 0.887766 ASP-32

1.800000 0.849645 ASP-32

2.000000 0.802905 ASP-32

2.200000 0.746376 ASP-32

2.400000 0.679045 ASP-32

2.600000 0.600993 ASP-32

2.800000 0.514528 ASP-32

3.000000 0.424526 ASP-32

3.200001 0.337311 ASP-32

3.400001 0.258643 ASP-32

3.600001 0.192113 ASP-32

3.800001 0.138801 ASP-32

4.000000 0.097827 ASP-32

4.200000 0.067426 ASP-32

Site ARG-36

Site GLU-42

0.800000 0.936873 GLU-42

1.000000 0.905421 GLU-42

1.200000 0.860371 GLU-42

1.400000 0.799768 GLU-42

1.600000 0.719980 GLU-42

1.800000 0.631413 GLU-42

2.000000 0.529767 GLU-42

2.200000 0.429390 GLU-42

2.400000 0.337339 GLU-42

2.600000 0.259364 GLU-42

2.800000 0.196815 GLU-42

3.000000 0.148278 GLU-42

3.200001 0.111120 GLU-42

3.400001 0.082667 GLU-42

3.600001 0.060790 GLU-42

Site ASP-45

0.000000 0.452939 ASP-45

0.200000 0.351925 ASP-45

0.400000 0.262801 ASP-45

0.600000 0.189371 ASP-45

0.800000 0.134176 ASP-45

1.000000 0.092867 ASP-45

1.200000 0.063401 ASP-45

Site ASP-48

2.800000 0.941890 ASP-48

3.000000 0.916744 ASP-48

3.200001 0.881790 ASP-48

3.400001 0.834601 ASP-48

3.600001 0.772884 ASP-48

3.800001 0.695320 ASP-48

4.000000 0.602963 ASP-48

4.200000 0.500577 ASP-48

4.400000 0.396395 ASP-48

4.600000 0.299635 ASP-48

4.800000 0.217226 ASP-48

5.000000 0.152114 ASP-48

5.199999 0.103677 ASP-48

5.399999 0.069260 ASP-48

Site ARG-49

Site GLU-53

468

2.000000 0.945503 GLU-53

2.200000 0.921530 GLU-53

2.400000 0.888222 GLU-53

2.600000 0.842824 GLU-53

2.800000 0.782761 GLU-53

3.000000 0.707201 GLU-53

3.200001 0.617844 GLU-53

3.400001 0.520073 GLU-53

3.600001 0.421368 GLU-53

3.800001 0.329093 GLU-53

4.000000 0.248466 GLU-53

4.200000 0.181886 GLU-53

4.400000 0.129456 GLU-53

4.600000 0.089855 GLU-53

4.800000 0.061048 GLU-53

Site TYR-57

11.399994 0.934736 TYR-57

11.599994 0.904767 TYR-57

11.799994 0.865085 TYR-57

11.999993 0.815941 TYR-57

Site ASP-58

1.800000 0.942671 ASP-58

2.000000 0.916680 ASP-58

2.200000 0.881022 ASP-58

2.400000 0.834273 ASP-58

2.600000 0.773256 ASP-58

2.800000 0.699137 ASP-58

3.000000 0.611996 ASP-58

3.200001 0.516163 ASP-58

3.400001 0.418769 ASP-58

3.600001 0.327095 ASP-58

3.800001 0.246790 ASP-58

4.000000 0.180725 ASP-58

4.200000 0.129132 ASP-58

4.400000 0.090465 ASP-58

4.600000 0.062368 ASP-58

Site ARG-59

Site HID-65

4.400000 0.947136 HID-65

4.600000 0.928527 HID-65

4.800000 0.902182 HID-65

5.000000 0.864820 HID-65

5.199999 0.812805 HID-65

5.399999 0.743316 HID-65

5.599999 0.656202 HID-65

5.799999 0.555679 HID-65

5.999999 0.450305 HID-65

6.199998 0.350412 HID-65

6.399998 0.263518 HID-65

6.599998 0.193291 HID-65

6.799998 0.139183 HID-65

6.999998 0.098614 HID-65

7.199997 0.068682 HID-65

Site ARG-67

11.599994 0.944747 ARG-67

11.799994 0.920965 ARG-67

11.999993 0.887307 ARG-67

Site GLU-70

0.000000 0.134251 GLU-70

0.200000 0.118054 GLU-70

0.400000 0.105981 GLU-70

0.600000 0.096790 GLU-70

0.800000 0.088559 GLU-70

1.000000 0.080315 GLU-70

1.200000 0.070848 GLU-70

1.400000 0.060846 GLU-70

1.600000 0.051139 GLU-70

Site ASP-71

0.000000 0.839084 ASP-71

0.200000 0.796314 ASP-71

0.400000 0.744717 ASP-71

469

0.600000 0.683833 ASP-71

0.800000 0.613727 ASP-71

1.000000 0.535684 ASP-71

1.200000 0.452926 ASP-71

1.400000 0.372780 ASP-71

1.600000 0.297316 ASP-71

1.800000 0.232060 ASP-71

2.000000 0.178498 ASP-71

2.200000 0.136018 ASP-71

2.400000 0.103010 ASP-71

2.600000 0.077666 ASP-71

2.800000 0.058325 ASP-71

Site GLU-79

2.400000 0.937556 GLU-79

2.600000 0.908621 GLU-79

2.800000 0.869115 GLU-79

3.000000 0.816930 GLU-79

3.200001 0.748518 GLU-79

3.400001 0.664835 GLU-79

3.600001 0.569220 GLU-79

3.800001 0.468453 GLU-79

4.000000 0.371025 GLU-79

4.200000 0.284178 GLU-79

4.400000 0.211822 GLU-79

4.600000 0.154499 GLU-79

4.800000 0.110651 GLU-79

5.000000 0.077927 GLU-79

5.199999 0.053977 GLU-79

Site TYR-85

Site ASP-98

0.000000 0.156680 ASP-98

0.200000 0.120431 ASP-98

0.400000 0.093587 ASP-98

0.600000 0.073623 ASP-98

0.800000 0.058455 ASP-98

Site LYS-99

Site LYS-101

10.999994 0.945217 LYS-101

11.199994 0.919058 LYS-101

11.399994 0.882341 LYS-101

11.599994 0.832857 LYS-101

11.799994 0.769540 LYS-101

11.999993 0.693174 LYS-101

Site LYS-103

Site ASP-110

2.400000 0.933329 ASP-110

2.600000 0.900329 ASP-110

2.800000 0.853622 ASP-110

3.000000 0.790442 ASP-110

3.200001 0.708933 ASP-110

3.400001 0.610969 ASP-110

3.600001 0.502671 ASP-110

3.800001 0.393714 ASP-110

4.000000 0.294174 ASP-110

4.200000 0.210968 ASP-110

4.400000 0.146395 ASP-110

4.600000 0.099085 ASP-110

4.800000 0.065795 ASP-110

Site TYR-116

10.599995 0.941830 TYR-116

10.799994 0.912382 TYR-116

10.999994 0.871118 TYR-116

11.199994 0.816239 TYR-116

11.399994 0.747743 TYR-116

11.599994 0.667993 TYR-116

11.799994 0.581198 TYR-116

11.999993 0.492045 TYR-116

Site ARG-117

Site LYS-121

11.599994 0.935790 LYS-121

11.799994 0.904018 LYS-121

470

11.999993 0.858908 LYS-121

Site GLU-122

3.000000 0.927800 GLU-122

3.200001 0.894309 GLU-122

3.400001 0.847344 GLU-122

3.600001 0.783765 GLU-122

3.800001 0.702007 GLU-122

4.000000 0.603903 GLU-122

4.200000 0.495773 GLU-122

4.400000 0.387294 GLU-122

4.600000 0.288328 GLU-122

4.800000 0.205655 GLU-122

5.000000 0.141985 GLU-122

5.199999 0.095123 GLU-122

5.399999 0.062528 GLU-122

Site ASP-124

1.000000 0.930602 ASP-124

1.200000 0.895397 ASP-124

1.400000 0.846945 ASP-124

1.600000 0.783119 ASP-124

1.800000 0.702189 ASP-124

2.000000 0.614634 ASP-124

2.200000 0.516993 ASP-124

2.400000 0.422094 ASP-124

2.600000 0.336710 ASP-124

2.800000 0.266827 ASP-124

3.000000 0.208845 ASP-124

3.200001 0.163298 ASP-124

3.400001 0.128087 ASP-124

3.600001 0.100983 ASP-124

3.800001 0.079929 ASP-124

4.000000 0.063186 ASP-124

Site ASP-126

2.400000 0.946125 ASP-126

2.600000 0.931256 ASP-126

2.800000 0.912689 ASP-126

3.000000 0.888935 ASP-126

3.200001 0.857857 ASP-126

3.400001 0.816873 ASP-126

3.600001 0.763286 ASP-126

3.800001 0.695103 ASP-126

4.000000 0.612227 ASP-126

4.200000 0.517633 ASP-126

4.400000 0.417697 ASP-126

4.600000 0.320928 ASP-126

4.800000 0.235263 ASP-126

5.000000 0.167090 ASP-126

5.199999 0.113430 ASP-126

5.399999 0.075192 ASP-126

Site GLU-131

1.600000 0.949171 GLU-131

1.800000 0.925466 GLU-131

2.000000 0.892366 GLU-131

2.200000 0.848100 GLU-131

2.400000 0.788828 GLU-131

2.600000 0.713336 GLU-131

2.800000 0.623471 GLU-131

3.000000 0.522132 GLU-131

3.200001 0.418098 GLU-131

3.400001 0.320704 GLU-131

3.600001 0.237128 GLU-131

3.800001 0.170452 GLU-131

4.000000 0.120035 GLU-131

4.200000 0.083239 GLU-131

4.400000 0.056886 GLU-131

Site LYS-139

Site HID-142

6.399998 0.949268 HID-142

6.599998 0.925848 HID-142

6.799998 0.892803 HID-142

6.999998 0.847130 HID-142

471

7.199997 0.785724 HID-142

7.399997 0.706517 HID-142

7.599997 0.610586 HID-142

7.799997 0.502932 HID-142

7.999997 0.393447 HID-142

8.199997 0.292733 HID-142

8.399997 0.208404 HID-142

8.599997 0.143233 HID-142

8.799996 0.095965 HID-142

8.999996 0.063176 HID-142

Site GLU-143

0.200000 0.941872 GLU-143

0.400000 0.928604 GLU-143

0.600000 0.913521 GLU-143

0.800000 0.896987 GLU-143

1.000000 0.877080 GLU-143

1.200000 0.853444 GLU-143

1.400000 0.825478 GLU-143

1.600000 0.793135 GLU-143

1.800000 0.757036 GLU-143

2.000000 0.718260 GLU-143

2.200000 0.678041 GLU-143

2.400000 0.637602 GLU-143

2.600000 0.598046 GLU-143

2.800000 0.560217 GLU-143

3.000000 0.524507 GLU-143

3.200001 0.490680 GLU-143

3.400001 0.457871 GLU-143

3.600001 0.424760 GLU-143

3.800001 0.389795 GLU-143

4.000000 0.351461 GLU-143

4.200000 0.308789 GLU-143

4.400000 0.262162 GLU-143

4.600000 0.213835 GLU-143

4.800000 0.167441 GLU-143

5.000000 0.126487 GLU-143

5.199999 0.093089 GLU-143

5.399999 0.067418 GLU-143

Site LYS-150

11.199994 0.930727 LYS-150

11.399994 0.898410 LYS-150

11.599994 0.854235 LYS-150

11.799994 0.796117 LYS-150

11.999993 0.724076 LYS-150

Site GLU-152

1.200000 0.933516 GLU-152

1.400000 0.903850 GLU-152

1.600000 0.864130 GLU-152

1.800000 0.813712 GLU-152

2.000000 0.751779 GLU-152

2.200000 0.680466 GLU-152

2.400000 0.601442 GLU-152

2.600000 0.518633 GLU-152

2.800000 0.436367 GLU-152

3.000000 0.359036 GLU-152

3.200001 0.289967 GLU-152

3.400001 0.230600 GLU-152

3.600001 0.180676 GLU-152

3.800001 0.139108 GLU-152

4.000000 0.104785 GLU-152

4.200000 0.076897 GLU-152

4.400000 0.054851 GLU-152

Site LYS-153

11.199994 0.942685 LYS-153

11.399994 0.913166 LYS-153

11.599994 0.871037 LYS-153

11.799994 0.813473 LYS-153

11.999993 0.738764 LYS-153

Site GLU-155

1.400000 0.927747 GLU-155

1.600000 0.895745 GLU-155

472

1.800000 0.853767 GLU-155

2.000000 0.802306 GLU-155

2.200000 0.740138 GLU-155

2.400000 0.668889 GLU-155

2.600000 0.590680 GLU-155

2.800000 0.508958 GLU-155

3.000000 0.428034 GLU-155

3.200001 0.351908 GLU-155

3.400001 0.283181 GLU-155

3.600001 0.222907 GLU-155

3.800001 0.171232 GLU-155

4.000000 0.128040 GLU-155

4.200000 0.093078 GLU-155

4.400000 0.065811 GLU-155

Site ASP-156

1.200000 0.939347 ASP-156

1.400000 0.913272 ASP-156

1.600000 0.879740 ASP-156

1.800000 0.839359 ASP-156

2.000000 0.794504 ASP-156

2.200000 0.748964 ASP-156

2.400000 0.704496 ASP-156

2.600000 0.662345 ASP-156

2.800000 0.621115 ASP-156

3.000000 0.577783 ASP-156

3.200001 0.528920 ASP-156

3.400001 0.472158 ASP-156

3.600001 0.407493 ASP-156

3.800001 0.337832 ASP-156

4.000000 0.268224 ASP-156

4.200000 0.204060 ASP-156

4.400000 0.149279 ASP-156

4.600000 0.105565 ASP-156

4.800000 0.072673 ASP-156

Site GLU-162

1.600000 0.925807 GLU-162

1.800000 0.892245 GLU-162

2.000000 0.846041 GLU-162

2.200000 0.785685 GLU-162

2.400000 0.710988 GLU-162

2.600000 0.623229 GLU-162

2.800000 0.528323 GLU-162

3.000000 0.431104 GLU-162

3.200001 0.339756 GLU-162

3.400001 0.258926 GLU-162

3.600001 0.191873 GLU-162

3.800001 0.138972 GLU-162

4.000000 0.098874 GLU-162

4.200000 0.069387 GLU-162

Site LYS-163

9.199996 0.932859 LYS-163

9.399996 0.898566 LYS-163

9.599996 0.848500 LYS-163

9.799995 0.779698 LYS-163

9.999995 0.690999 LYS-163

10.199995 0.585606 LYS-163

10.399995 0.471857 LYS-163

10.599995 0.361240 LYS-163

10.799994 0.263805 LYS-163

10.999994 0.185390 LYS-163

11.199994 0.126665 LYS-163

11.399994 0.084992 LYS-163

11.599994 0.056505 LYS-163

Site HIS-167

1.800000 0.941722 HIS-167

2.000000 0.923820 HIS-167

2.200000 0.901874 HIS-167

2.400000 0.875487 HIS-167

2.600000 0.844250 HIS-167

2.800000 0.807638 HIS-167

3.000000 0.764929 HIS-167

473

3.200001 0.715475 HIS-167

3.400001 0.659168 HIS-167

3.600001 0.596906 HIS-167

3.800001 0.530578 HIS-167

4.000000 0.462479 HIS-167

4.200000 0.394535 HIS-167

4.400000 0.328037 HIS-167

4.600000 0.264259 HIS-167

4.800000 0.205189 HIS-167

5.000000 0.153651 HIS-167

5.199999 0.111016 HIS-167

5.399999 0.078057 HIS-167

5.599999 0.053864 HIS-167

Site ASP-168

0.000000 0.480919 ASP-168

0.200000 0.397512 ASP-168

0.400000 0.320043 ASP-168

0.600000 0.252390 ASP-168

0.800000 0.198786 ASP-168

1.000000 0.153946 ASP-168

1.200000 0.120081 ASP-168

1.400000 0.093810 ASP-168

1.600000 0.074167 ASP-168

1.800000 0.059309 ASP-168

Site GLU-174

0.000000 0.693363 GLU-174

0.200000 0.634151 GLU-174

0.400000 0.569880 GLU-174

0.600000 0.505606 GLU-174

0.800000 0.443556 GLU-174

1.000000 0.385809 GLU-174

1.200000 0.334169 GLU-174

1.400000 0.288141 GLU-174

1.600000 0.247727 GLU-174

1.800000 0.211958 GLU-174

2.000000 0.179767 GLU-174

2.200000 0.150446 GLU-174

2.400000 0.123819 GLU-174

2.600000 0.100043 GLU-174

2.800000 0.079291 GLU-174

3.000000 0.061635 GLU-174

Site LYS-175

11.199994 0.929973 LYS-175

11.399994 0.898253 LYS-175

11.599994 0.855059 LYS-175

11.799994 0.798500 LYS-175

11.999993 0.728651 LYS-175

Site ARG-180

Site GLU-181

2.200000 0.942283 GLU-181

2.400000 0.923630 GLU-181

2.600000 0.899069 GLU-181

2.800000 0.866776 GLU-181

3.000000 0.824693 GLU-181

3.200001 0.770945 GLU-181

3.400001 0.704576 GLU-181

3.600001 0.626425 GLU-181

3.800001 0.539627 GLU-181

4.000000 0.449127 GLU-181

4.200000 0.360491 GLU-181

4.400000 0.278775 GLU-181

4.600000 0.207810 GLU-181

4.800000 0.149754 GLU-181

5.000000 0.104973 GLU-181

5.199999 0.071963 GLU-181

Site ARG-196

Site TYR-199

8.599997 0.924788 TYR-199

8.799996 0.886763 TYR-199

8.999996 0.833356 TYR-199

9.199996 0.762450 TYR-199

474

9.399996 0.674934 TYR-199

9.599996 0.576226 TYR-199

9.799995 0.474772 TYR-199

9.999995 0.380104 TYR-199

10.199995 0.298512 TYR-199

10.399995 0.231954 TYR-199

10.599995 0.179157 TYR-199

10.799994 0.137582 TYR-199

10.999994 0.104855 TYR-199

11.199994 0.079258 TYR-199

11.399994 0.059568 TYR-199

Site CYS-203

9.199996 0.936986 CYS-203

9.399996 0.906597 CYS-203

9.599996 0.864377 CYS-203

9.799995 0.808060 CYS-203

9.999995 0.736724 CYS-203

10.199995 0.651727 CYS-203

10.399995 0.557031 CYS-203

10.599995 0.458679 CYS-203

10.799994 0.363919 CYS-203

10.999994 0.278480 CYS-203

11.199994 0.206653 CYS-203

11.399994 0.150207 CYS-203

11.599994 0.107306 CYS-203

11.799994 0.076091 CYS-203

11.999993 0.053955 CYS-203

Site GLU-207

0.800000 0.932013 GLU-207

1.000000 0.899690 GLU-207

1.200000 0.852755 GLU-207

1.400000 0.793277 GLU-207

1.600000 0.716841 GLU-207

1.800000 0.628086 GLU-207

2.000000 0.532859 GLU-207

2.200000 0.438215 GLU-207

2.400000 0.350265 GLU-207

2.600000 0.272920 GLU-207

2.800000 0.207882 GLU-207

3.000000 0.155258 GLU-207

3.200001 0.114034 GLU-207

3.400001 0.082507 GLU-207

3.600001 0.058784 GLU-207

Site GLU-209

1.000000 0.944103 GLU-209

1.200000 0.918973 GLU-209

1.400000 0.887180 GLU-209

1.600000 0.845318 GLU-209

1.800000 0.793808 GLU-209

2.000000 0.732690 GLU-209

2.200000 0.662012 GLU-209

2.400000 0.583239 GLU-209

2.600000 0.499475 GLU-209

2.800000 0.415584 GLU-209

3.000000 0.336852 GLU-209

3.200001 0.267135 GLU-209

3.400001 0.207972 GLU-209

3.600001 0.159051 GLU-209

3.800001 0.119258 GLU-209

4.000000 0.087438 GLU-209

4.200000 0.062585 GLU-209

Site GLU-210

1.800000 0.931936 GLU-210

2.000000 0.908070 GLU-210

2.200000 0.878068 GLU-210

2.400000 0.840881 GLU-210

2.600000 0.795446 GLU-210

2.800000 0.740795 GLU-210

3.000000 0.676119 GLU-210

3.200001 0.601157 GLU-210

3.400001 0.517134 GLU-210

475

3.600001 0.427664 GLU-210

3.800001 0.338583 GLU-210

4.000000 0.256421 GLU-210

4.200000 0.186340 GLU-210

4.400000 0.130802 GLU-210

4.600000 0.089336 GLU-210

4.800000 0.059867 GLU-210

Site LYS-211

11.799994 0.930461 LYS-211

11.999993 0.898826 LYS-211

Site LYS-212

9.199996 0.945754 LYS-212

9.399996 0.917047 LYS-212

9.599996 0.875395 LYS-212

9.799995 0.817478 LYS-212

9.999995 0.741440 LYS-212

10.199995 0.648567 LYS-212

10.399995 0.544245 LYS-212

10.599995 0.437043 LYS-212

10.799994 0.336115 LYS-212

10.999994 0.248840 LYS-212

11.199994 0.178145 LYS-212

11.399994 0.124534 LYS-212

11.599994 0.085899 LYS-212

11.799994 0.059097 LYS-212

Site CYS-217

11.199994 0.947409 CYS-217

11.399994 0.930891 CYS-217

11.599994 0.908929 CYS-217

11.799994 0.879762 CYS-217

11.999993 0.841466 CYS-217

Site ARG-219

Site GLU-227

3.200001 0.944341 GLU-227

3.400001 0.920678 GLU-227

3.600001 0.888611 GLU-227

3.800001 0.846180 GLU-227

4.000000 0.791597 GLU-227

4.200000 0.723837 GLU-227

4.400000 0.643485 GLU-227

4.600000 0.553327 GLU-227

4.800000 0.458287 GLU-227

5.000000 0.364624 GLU-227

5.199999 0.278527 GLU-227

5.399999 0.204617 GLU-227

5.599999 0.145124 GLU-227

5.799999 0.099799 GLU-227

5.999999 0.066946 GLU-227

Site ARG-231

Site ASP-233

0.000000 0.742699 ASP-233

0.200000 0.682398 ASP-233

0.400000 0.608940 ASP-233

0.600000 0.525089 ASP-233

0.800000 0.437386 ASP-233

1.000000 0.354650 ASP-233

1.200000 0.281241 ASP-233

1.400000 0.221550 ASP-233

1.600000 0.174902 ASP-233

1.800000 0.138988 ASP-233

2.000000 0.111067 ASP-233

2.200000 0.090307 ASP-233

2.400000 0.072933 ASP-233

2.600000 0.059143 ASP-233

Site ASP-237

0.000000 0.937037 ASP-237

0.200000 0.907802 ASP-237

0.400000 0.867498 ASP-237

0.600000 0.813538 ASP-237

0.800000 0.744161 ASP-237

1.000000 0.660086 ASP-237

476

1.200000 0.567747 ASP-237

1.400000 0.469192 ASP-237

1.600000 0.373583 ASP-237

1.800000 0.292797 ASP-237

2.000000 0.221217 ASP-237

2.200000 0.163586 ASP-237

2.400000 0.119569 ASP-237

2.600000 0.087114 ASP-237

2.800000 0.063724 ASP-237

Site GLU-238

2.400000 0.946282 GLU-238

2.600000 0.926769 GLU-238

2.800000 0.899710 GLU-238

3.000000 0.862739 GLU-238

3.200001 0.813511 GLU-238

3.400001 0.750312 GLU-238

3.600001 0.672854 GLU-238

3.800001 0.583030 GLU-238

4.000000 0.485275 GLU-238

4.200000 0.386239 GLU-238

4.400000 0.293467 GLU-238

4.600000 0.213315 GLU-238

4.800000 0.149173 GLU-238

5.000000 0.101131 GLU-238

5.199999 0.067069 GLU-238

Site GLU-246

0.000000 0.096725 GLU-246

0.200000 0.097489 GLU-246

0.400000 0.100674 GLU-246

0.600000 0.105984 GLU-246

0.800000 0.112824 GLU-246

1.000000 0.120340 GLU-246

1.200000 0.127460 GLU-246

1.400000 0.133073 GLU-246

1.600000 0.136189 GLU-246

1.800000 0.136128 GLU-246

2.000000 0.132557 GLU-246

2.200000 0.125472 GLU-246

2.400000 0.115196 GLU-246

2.600000 0.102407 GLU-246

2.800000 0.088141 GLU-246

3.000000 0.073685 GLU-246

3.200001 0.060260 GLU-246

Site HIS-256

5.599999 0.946358 HIS-256

5.799999 0.922147 HIS-256

5.999999 0.888906 HIS-256

6.199998 0.844254 HIS-256

6.399998 0.785917 HIS-256

6.599998 0.710831 HIS-256

6.799998 0.619407 HIS-256

6.999998 0.515695 HIS-256

7.199997 0.408269 HIS-256

7.399997 0.307033 HIS-256

7.599997 0.220457 HIS-256

7.799997 0.152974 HIS-256

7.999997 0.103148 HIS-256

8.199997 0.068305 HIS-256

Site LYS-261

10.599995 0.945473 LYS-261

10.799994 0.931511 LYS-261

10.999994 0.913421 LYS-261

11.199994 0.889256 LYS-261

11.399994 0.856287 LYS-261

11.599994 0.811393 LYS-261

11.799994 0.751637 LYS-261

11.999993 0.676211 LYS-261

Site LYS-264

9.399996 0.938253 LYS-264

9.599996 0.914214 LYS-264

9.799995 0.883668 LYS-264

477

9.999995 0.845493 LYS-264

10.199995 0.798310 LYS-264

10.399995 0.740968 LYS-264

10.599995 0.673200 LYS-264

10.799994 0.596257 LYS-264

10.999994 0.513353 LYS-264

11.199994 0.429525 LYS-264

11.399994 0.350446 LYS-264

11.599994 0.280498 LYS-264

11.799994 0.221428 LYS-264

11.999993 0.172694 LYS-264

Site TYR-268

9.599996 0.939900 TYR-268

9.799995 0.910237 TYR-268

9.999995 0.868915 TYR-268

10.199995 0.813551 TYR-268

10.399995 0.742714 TYR-268

10.599995 0.656872 TYR-268

10.799994 0.559481 TYR-268

10.999994 0.457177 TYR-268

11.199994 0.358294 TYR-268

11.399994 0.270329 TYR-268

11.599994 0.197457 TYR-268

11.799994 0.140731 TYR-268

11.999993 0.098648 TYR-268

Site ASP-271

0.600000 0.949766 ASP-271

0.800000 0.923743 ASP-271

1.000000 0.886159 ASP-271

1.200000 0.838657 ASP-271

1.400000 0.783387 ASP-271

1.600000 0.711982 ASP-271

1.800000 0.633259 ASP-271

2.000000 0.553001 ASP-271

2.200000 0.477063 ASP-271

2.400000 0.409378 ASP-271

2.600000 0.350981 ASP-271

2.800000 0.300497 ASP-271

3.000000 0.255484 ASP-271

3.200001 0.213738 ASP-271

3.400001 0.174423 ASP-271

3.600001 0.137441 ASP-271

3.800001 0.104289 ASP-271

4.000000 0.076302 ASP-271

4.200000 0.054158 ASP-271

Site GLU-272

1.600000 0.931088 GLU-272

1.800000 0.905345 GLU-272

2.000000 0.873240 GLU-272

2.200000 0.834186 GLU-272

2.400000 0.787816 GLU-272

2.600000 0.732654 GLU-272

2.800000 0.668557 GLU-272

3.000000 0.596123 GLU-272

3.200001 0.516343 GLU-272

3.400001 0.432341 GLU-272

3.600001 0.348523 GLU-272

3.800001 0.270003 GLU-272

4.000000 0.201224 GLU-272

4.200000 0.144769 GLU-272

4.400000 0.101041 GLU-272

4.600000 0.068783 GLU-272

Site CYS-274

Site GLU-279

1.400000 0.936756 GLU-279

1.600000 0.906790 GLU-279

1.800000 0.864910 GLU-279

2.000000 0.808228 GLU-279

2.200000 0.736153 GLU-279

2.400000 0.650948 GLU-279

2.600000 0.555293 GLU-279

478

2.800000 0.461626 GLU-279

3.000000 0.371095 GLU-279

3.200001 0.289806 GLU-279

3.400001 0.221430 GLU-279

3.600001 0.166142 GLU-279

3.800001 0.122785 GLU-279

4.000000 0.089651 GLU-279

4.200000 0.064860 GLU-279

Site CTLEU-282

3.200001 0.931909 CTLEU-282

3.400001 0.905938 CTLEU-282

3.600001 0.871047 CTLEU-282

3.800001 0.824888 CTLEU-282

4.000000 0.765449 CTLEU-282

4.200000 0.691975 CTLEU-282

4.400000 0.605947 CTLEU-282

4.600000 0.511492 CTLEU-282

4.800000 0.414766 CTLEU-282

5.000000 0.322542 CTLEU-282

5.199999 0.240620 CTLEU-282

5.399999 0.172662 CTLEU-282

5.599999 0.119756 CTLEU-282

5.799999 0.080791 CTLEU-282

5.999999 0.053486 CTLEU-282

479

S-Table.7.3. Significant contributions to pK of each site from neighbouring groups.

Significant contributions to pK of each site from neighbouring groups.

The site of interest is listed first, list of neighbours with strong contributions follow.

The first number is the background contribution from non-titratable groups;

The last one is the interaction with other titratable groups.

SITE: NTser- 1

1 SER 0.481322 -0

4 THR 0.214986 0

5 LYS -0.123216 -0.245364

279 GLU 0.0160218 0.107086

SITE: LYS- 5

1 SER 0.24604 -0.245364

2 ASN 2.0827 0

6 PHE 0.338186 0

7 GLY 0.12449 0

39 VAL -0.144581 0

58 ASP -0.0452792 0.138124

59 ARG -0.0372275 -0.316502

268 TYR -0.0589558 0.174988

271 ASP 0.0557484 0.107514

272 GLU 0.0320458 0.145611

274 CYS 0.0369966 0.248511

275 VAL 0.177382 0

278 LEU -0.143915 0

279 GLU 0.233567 2.38323

282 LEU -0.0680309 0.171468

SITE: LYS- 8

9 VAL -0.0514029 0

40 GLN 0.347579 0

41 ALA -0.174343 0

42 GLU 0.836326 2.07556

49 ARG 0.0449271 -0.466546

53 GLU 0.144854 0.217728

54 LEU 0.136452 0

56 ASN 1.28149 0

57 TYR -0.23282 1.79292

58 ASP -0.142527 0.45773

59 ARG 0.0389807 -0.208248

SITE: LYS- 16

14 GLY 0.0954619 0

15 GLY -0.20165 0

19 GLU -0.0391764 0.154516

20 ARG -0.114121 -0.185256

45 ASP -0.0494022 0.310224

46 PRO -0.0419594 0

48 ASP -0.0398519 0.115654

65 HID 0.164118 -0.197774

66 GLY 0.160052 0

67 ARG 0.109317 -0.333364

68 GLY 0.328242 0

69 GLY 0.0573501 0

70 GLU -0.0816239 0.216067

71 ASP 0.108714 0.576359

73 GLN 1.41275 0

74 ILE -0.0103906 0

85 TYR -0.0195371 0.105528

143 GLU -0.0062489 0.116375

SITE: GLU- 19

14 GLY 0.159981 0

16 LYS -0.116759 -0.154516

17 SER -0.839014 0

18 ALA -0.441368 0

20 ARG -0.00858567 -0.353076

22 VAL -0.239184 0

23 SER -0.258028 0

25 ASP -0.0741952 0.171726

480

45 ASP -0.0205376 0.135459

65 HID 0.209404 -0.670018

66 GLY -0.200205 0

67 ARG -0.0363824 -0.208007

70 GLU -0.0390305 0.582858

71 ASP -0.021403 0.277831

98 ASP -0.0404421 0.122088

99 LYS -0.0193743 -0.359393

103 LYS -0.0123096 -0.114598

139 LYS -0.0432912 -0.176008

142 HID 0.0477576 -0.148036

143 GLU 0.0464366 0.698784

144 GLY -0.0203044 0

163 LYS 0.0338846 -0.156711

167 HIS 0.00630903 -0.326063

168 ASP -0.0235486 0.182311

174 GLU 0.014572 0.105396

181 GLU -0.0130486 0.112029

231 ARG -0.0165033 -0.329402

233 ASP 0.0176538 0.273

246 GLU -0.0683963 0.259782

256 HIS -0.0445566 -0.133864

SITE: ARG- 20

13 LEU 0.109907 0

14 GLY -0.528996 0

15 GLY 0.718011 0

16 LYS 0.0317683 -0.185256

19 GLU -0.264434 0.353076

23 SER -0.173879 0

25 ASP -0.0629228 0.133169

45 ASP 0.823712 2.33631

46 PRO -0.18564 0

47 GLN -0.0237143 0

48 ASP 0.483432 1.30835

49 ARG 0.0419242 -0.166887

65 HID 0.120378 -0.294149

70 GLU -0.0231718 0.123829

71 ASP 0.00138778 0.100603

SITE: ASP- 25

19 GLU 0.0924816 0.171726

20 ARG 0.165381 -0.133169

21 ALA 0.344309 0

22 VAL 0.553868 0

23 SER 0.127351 0

24 LEU 0.113831 0

26 SER -0.661364 0

27 GLY -0.132605 0

28 GLN -0.339244 0

29 ALA -0.250003 0

32 ASP -0.088611 0.160778

36 ARG 0.0138839 -0.155459

45 ASP -0.0182216 0.118772

63 VAL 0.073018 0

65 HID -0.0187384 -0.574341

70 GLU 0.00853843 0.152669

231 ARG -0.0118183 -0.143732

252 GLY -0.156622 0

253 MET -0.282326 0

254 THR -1.14086 0

255 SER -0.159754 0

256 HIS -0.406528 -1.21419

257 SER -0.22914 0

261 LYS -0.00962846 -0.184674

264 LYS -0.00931194 -0.123102

271 ASP 0.0325906 0.103902

274 CYS 0.026879 0.242671

SITE: ASP- 32

25 ASP 0.0935976 0.160778

27 GLY 0.152819 0

28 GLN 0.240626 0

29 ALA 0.457068 0

481

30 VAL 0.14449 0

31 LEU 0.0650615 0

33 ALA -0.284567 0

35 LEU -0.33058 0

36 ARG -0.19606 -1.7419

37 SER -0.168754 0

264 LYS -0.00237927 -0.10904

271 ASP 0.0332704 0.374802

274 CYS -0.0155296 0.412635

SITE: ARG- 36

25 ASP 0.0843946 0.155459

28 GLN 0.00114847 0

29 ALA 0.257289 0

30 VAL 0.102468 0

32 ASP -0.0965586 1.7419

37 SER -0.247574 0

261 LYS 0.0109233 -0.109305

264 LYS -0.00495021 -0.192214

271 ASP 0.26671 0.909549

272 GLU -0.0193584 0.101641

274 CYS -0.0386775 0.395467

SITE: GLU- 42

8 LYS -0.258814 -2.07556

9 VAL 0.334282 0

10 ALA -0.164916 0

40 GLN 0.210292 0

41 ALA -0.331002 0

45 ASP -0.0250104 0.142453

49 ARG 0.179243 -0.685769

53 GLU 0.0789478 0.211074

56 ASN 0.124561 0

57 TYR -1.56982 2.70561

58 ASP -0.0634181 0.171652

59 ARG 0.0287517 -0.105776

274 CYS 0.0166608 0.106667

SITE: ASP- 45

11 VAL 0.158947 0

12 LEU -0.227136 0

13 LEU 1.02348 0

14 GLY -0.81246 0

15 GLY -0.756754 0

16 LYS 0.0369616 -0.310224

19 GLU -0.00448845 0.135459

20 ARG -1.25026 -2.33631

25 ASP -0.0261943 0.118772

42 GLU 0.0479534 0.142453

43 ALA -0.167736 0

46 PRO -0.94895 0

47 GLN -1.67625 0

48 ASP -0.649936 1.1848

49 ARG 0.0553803 -0.450477

53 GLU -0.0301605 0.205435

57 TYR 0.0366734 0.239384

65 HID 0.119225 -0.287966

68 GLY 0.0878662 0

70 GLU -0.0326716 0.138128

71 ASP 0.00912599 0.139078

274 CYS 0.0198314 0.101717

SITE: ASP- 48

15 GLY 0.048904 0

16 LYS 7.59054e-05 -0.115654

20 ARG 0.4889 -1.30835

45 ASP 0.2628 1.1848

46 PRO -0.184534 0

47 GLN -0.4831 0

49 ARG 0.141163 -0.401049

53 GLU 0.0158156 0.152381

57 TYR 0.00802662 0.104384

SITE: ARG- 49

8 LYS -0.0135379 -0.466546

10 ALA -0.221982 0

482

11 VAL 0.148602 0

20 ARG 0.039826 -0.166887

42 GLU 0.118634 0.685769

43 ALA 1.42839 0

44 PHE -0.14304 0

45 ASP -0.177983 0.450477

46 PRO -0.0403478 0

48 ASP -0.510455 0.401049

50 SER 0.127918 0

53 GLU 0.275846 0.813114

56 ASN -0.190718 0

57 TYR 1.05084 2.23999

SITE: GLU- 53

8 LYS 0.00721811 -0.217728

42 GLU 0.0541855 0.211074

45 ASP 0.0403733 0.205435

48 ASP -0.0714912 0.152381

49 ARG -2.03704 -0.813114

50 SER 0.29049 0

51 VAL 0.20926 0

54 LEU -0.211594 0

56 ASN -0.463114 0

57 TYR 0.0352963 0.480171

SITE: TYR- 57

8 LYS -0.0358682 -1.79292

9 VAL 0.133158 0

10 ALA -0.290946 0

11 VAL 0.0676582 0

40 GLN 0.20619 0

41 ALA -0.313612 0

42 GLU 1.60105 2.70561

43 ALA 0.306338 0

44 PHE -0.208166 0

45 ASP -0.0436355 0.239384

48 ASP -0.0683131 0.104384

49 ARG 0.411231 -2.23999

50 SER 0.160929 0

53 GLU 0.27273 0.480171

54 LEU 0.155464 0

56 ASN -0.219065 0

58 ASP -0.138474 0.23648

59 ARG 0.0861211 -0.148156

274 CYS 0.023728 0.110074

SITE: ASP- 58

3 ALA 0.149785 0

5 LYS 0.0248994 -0.138124

6 PHE 0.513186 0

7 GLY -0.644904 0

8 LYS 0.158184 -0.45773

9 VAL -0.0127171 0

40 GLN -0.0483663 0

42 GLU 0.0579368 0.171652

54 LEU 0.149948 0

57 TYR -0.349658 0.23648

59 ARG -0.15828 -0.961689

84 PRO 0.159984 0

282 LEU 0.00916376 0.108369

SITE: ARG- 59

2 ASN 0.179708 0

5 LYS 0.145011 -0.316502

6 PHE 0.666708 0

7 GLY 0.101827 0

8 LYS 0.416862 -0.208248

9 VAL -0.191212 0

42 GLU 0.0558674 0.105776

57 TYR -0.122915 0.148156

58 ASP -0.827455 0.961689

79 GLU 0.0260627 0.114606

84 PRO 0.0827118 0

85 TYR -0.0388046 0.15456

274 CYS 0.0728764 0.160953

483

278 LEU 0.291867 0

279 GLU -0.0277968 0.243994

281 THR 2.25574 0

282 LEU 0.168068 0.42072

SITE: HID- 65

12 LEU 0.294695 0

13 LEU -0.251844 0

14 GLY 0.107284 0

15 GLY -0.0727178 0

16 LYS -0.16858 -0.197774

17 SER -0.27405 0

19 GLU 0.167713 0.670018

20 ARG 0.316377 -0.294149

22 VAL 0.300552 0

23 SER -0.551099 0

24 LEU 0.12646 0

25 ASP -0.364122 0.574341

26 SER -0.193567 0

27 GLY -0.0831921 0

29 ALA -0.165752 0

30 VAL -0.19199 0

45 ASP -0.0630895 0.287966

46 PRO -0.0917149 0

62 ILE 0.0568881 0

63 VAL 0.808309 0

64 LEU 0.0131602 0

66 GLY 0.0973296 0

67 ARG 0.185732 -0.161632

69 GLY -0.156755 0

70 GLU 0.0414114 1.24453

71 ASP -0.0488938 0.368757

75 GLN -0.168771 0

85 TYR -0.110735 0.153529

86 THR 0.149083 0

87 GLY -0.128597 0

88 THR 0.103747 0

98 ASP -0.0266548 0.11971

99 LYS -0.0116112 -0.31437

103 LYS -0.0100863 -0.130056

139 LYS -0.0245498 -0.178973

142 HID 0.04491 -0.1013

143 GLU 7.25571e-05 0.295996

144 GLY -0.0838493 0

167 HIS -0.00240773 -0.114745

174 GLU 0.0167593 0.112036

181 GLU -0.0135309 0.126331

217 CYS 0.0254612 0.134643

230 GLY -0.0689349 0

231 ARG -0.0692822 -0.532891

233 ASP 0.028769 0.369895

246 GLU -0.082874 0.283906

248 ASN -0.268154 0

250 VAL 1.39069 0

251 PRO 0.888266 0

252 GLY -0.478493 0

253 MET -0.608588 0

256 HIS -0.103018 -0.260836

259 VAL -0.15944 0

260 PRO -0.0858997 0

261 LYS -0.0455853 -0.187766

264 LYS -0.0130335 -0.117352

271 ASP 0.0478458 0.100219

274 CYS 0.0943932 0.459252

SITE: ARG- 67

16 LYS 0.0291279 -0.333364

19 GLU -0.024365 0.208007

65 HID 0.0891408 -0.161632

66 GLY -0.0223504 0

68 GLY -0.196203 0

70 GLU -0.178764 0.246511

71 ASP 0.0540452 0.59809

484

98 ASP -0.035825 0.118462

99 LYS -0.0114073 -0.137457

101 LYS -0.0148183 -0.129564

142 HID 0.0315648 -0.234196

143 GLU -0.0205392 0.287021

167 HIS 0.0498704 -0.132703

168 ASP -0.0478032 0.248555

SITE: GLU- 70

12 LEU 0.104818 0

14 GLY 0.102495 0

16 LYS -0.140503 -0.216067

17 SER -0.26028 0

19 GLU -0.203742 0.582858

20 ARG 0.0315343 -0.123829

25 ASP -0.0448644 0.152669

45 ASP -0.0330011 0.138128

62 ILE 0.0112747 0

64 LEU 0.492189 0

65 HID -0.864205 -1.24453

66 GLY -0.381992 0

67 ARG 0.0856634 -0.246511

68 GLY -0.20609 0

69 GLY -0.305316 0

71 ASP -0.477054 0.935014

72 GLY -0.344145 0

73 GLN -0.21706 0

74 ILE -0.20601 0

75 GLN -0.290072 0

85 TYR -0.0884646 0.224464

86 THR 0.132582 0

87 GLY -0.239674 0

88 THR -0.00438621 0

89 GLY 0.175814 0

90 VAL 0.148394 0

92 GLY 0.121029 0

93 SER -0.172582 0

96 GLY -0.0571924 0

97 MET 0.419898 0

98 ASP -0.0595746 0.359498

99 LYS -0.0162619 -0.678281

101 LYS -0.08551 -0.23037

103 LYS -0.0280245 -0.281407

139 LYS -0.0464898 -0.33804

140 PRO 0.0823849 0

141 VAL 0.146624 0

142 HID 0.144229 -0.245079

143 GLU -0.071868 0.497046

144 GLY -0.517498 0

167 HIS 5.03132e-05 -0.150158

168 ASP -0.00618192 0.191808

174 GLU 0.0311281 0.21864

181 GLU -0.0455244 0.174802

217 CYS 0.0700708 0.242581

226 ALA -0.190081 0

227 GLU 0.123597 0.144741

229 TRP 0.146966 0

230 GLY -0.216022 0

231 ARG -0.015965 -0.687436

233 ASP 0.0226234 0.620753

246 GLU -0.231858 0.582799

247 VAL 0.142562 0

248 ASN -3.1591 0

249 THR -1.62054 0

250 VAL -0.0259899 0

251 PRO 0.115816 0

252 GLY 0.0319508 0

256 HIS -0.0368612 -0.131133

261 LYS -0.03818 -0.120941

274 CYS 0.0419926 0.191767

SITE: ASP- 71

14 GLY 0.114873 0

485

16 LYS -0.192838 -0.576359

19 GLU -0.0639162 0.277831

20 ARG -0.0341946 -0.100603

45 ASP -0.0293769 0.139078

64 LEU 0.199747 0

65 HID 0.101671 -0.368757

66 GLY 0.818877 0

67 ARG -1.45181 -0.59809

68 GLY -1.9746 0

69 GLY -0.695398 0

70 GLU -0.796705 0.935014

72 GLY -0.550202 0

73 GLN -1.05464 0

74 ILE -0.15434 0

75 GLN 0.120388 0

85 TYR -0.0143796 0.171962

90 VAL 0.139282 0

93 SER -0.126026 0

97 MET -0.217863 0

98 ASP -0.0808848 0.279652

99 LYS -0.00482755 -0.265821

101 LYS -0.0374139 -0.326036

139 LYS -0.0181718 -0.105248

142 HID 0.0981869 -0.392703

143 GLU -0.0582828 0.356005

167 HIS 0.0497973 -0.110664

168 ASP -0.0219908 0.271232

231 ARG 0.00643459 -0.13983

233 ASP 0.00501268 0.131894

246 GLU -0.0558955 0.149932

SITE: GLU- 79

59 ARG -0.0663411 -0.114606

75 GLN 0.414118 0

76 GLY 0.0261026 0

80 TRP -0.30854 0

81 LEU -0.200248 0

82 ASN -0.530048 0

83 ILE 0.213542 0

85 TYR -1.40688 2.52378

89 GLY -0.191128 0

90 VAL -0.178006 0

227 GLU -0.0385559 0.158545

282 LEU 0.0195158 0.105509

SITE: TYR- 85

16 LYS -0.0216752 -0.105528

59 ARG -0.141384 -0.15456

65 HID -0.00800269 -0.153529

69 GLY 0.132256 0

70 GLU 0.0198366 0.224464

71 ASP -0.05151 0.171962

72 GLY 0.0065536 0

75 GLN -0.431276 0

76 GLY 0.0508141 0

77 VAL -0.166919 0

78 LEU -0.471082 0

79 GLU 0.665128 2.52378

82 ASN -0.214328 0

83 ILE 0.296474 0

84 PRO 0.00477427 0

87 GLY 0.209378 0

88 THR 0.300312 0

89 GLY -0.717002 0

90 VAL -0.486678 0

227 GLU -0.0363382 0.283802

249 THR 0.211304 0

274 CYS 0.0389735 0.102992

282 LEU 0.0267176 0.139093

SITE: ASP- 98

19 GLU -0.0307732 0.122088

65 HID -0.0276723 -0.11971

67 ARG -0.0573927 -0.118462

486

70 GLU 0.00504048 0.359498

71 ASP -0.110941 0.279652

92 GLY 0.164477 0

93 SER 0.193968 0

94 ALA 0.227776 0

95 ILE 0.870622 0

96 GLY 0.230288 0

99 LYS -0.64881 -1.14064

100 VAL -1.28591 0

101 LYS -1.53434 -2.44519

102 THR -0.917868 0

103 LYS -0.275504 -0.598695

116 TYR -0.0657996 0.446377

139 LYS 0.0263214 -0.416857

140 PRO 0.0763982 0

141 VAL 0.252271 0

142 HID 0.36466 -1.23462

143 GLU -0.088932 0.300709

144 GLY 0.0944746 0

168 ASP 0.10475 0.538931

172 MET -0.10825 0

174 GLU 0.0318456 0.365401

217 CYS 0.0326005 0.125991

231 ARG 0.0228863 -0.169573

233 ASP 0.00949733 0.222884

245 LEU 0.131942 0

246 GLU -0.161759 0.514309

SITE: LYS- 99

19 GLU -0.110085 0.359393

65 HID -0.0999254 -0.31437

66 GLY -0.094641 0

67 ARG -0.0393836 -0.137457

70 GLU -0.00904614 0.678281

71 ASP -0.0738854 0.265821

93 SER 0.0858194 0

95 ILE 0.211684 0

96 GLY 0.310719 0

97 MET 0.7373 0

98 ASP -0.586111 1.14064

100 VAL -0.22078 0

101 LYS -0.135036 -0.389321

102 THR -0.402806 0

103 LYS -0.249743 -0.965276

116 TYR -0.0874361 0.307243

139 LYS -0.371582 -1.41593

140 PRO 2.10548 0

141 VAL 1.59073 0

142 HID -0.176113 -0.664225

143 GLU 0.325434 1.07274

144 GLY 2.43472 0

146 VAL 0.440627 0

147 GLY -0.0194557 0

163 LYS 0.0224791 -0.226493

167 HIS -0.126347 -0.306922

168 ASP 0.0166003 0.692157

170 VAL 0.2288 0

171 VAL -0.177314 0

172 MET 0.123913 0

174 GLU 0.124598 0.790102

181 GLU -0.114683 0.281463

217 CYS 0.0506969 0.232564

231 ARG 0.0691668 -0.576411

233 ASP -0.0317939 0.72084

246 GLU -0.215363 1.73951

247 VAL 0.175492 0

249 THR -0.162919 0

SITE: LYS- 101

67 ARG -0.0773855 -0.129564

70 GLU 0.00171047 0.23037

71 ASP -0.116426 0.326036

92 GLY 0.116894 0

487

93 SER 0.118051 0

94 ALA 0.516956 0

95 ILE 0.422734 0

96 GLY 0.0486434 0

97 MET -0.203475 0

98 ASP -0.148082 2.44519

99 LYS 0.0375603 -0.389321

102 THR -0.403251 0

103 LYS -0.0814616 -0.2684

104 GLN -0.162097 0

116 TYR -0.0263708 0.225398

139 LYS 0.0103903 -0.166303

142 HID 0.21238 -1.39874

143 GLU -0.0456731 0.174476

168 ASP 0.0448564 0.313983

174 GLU 0.0125651 0.149192

233 ASP 0.00831465 0.104483

246 GLU -0.0762866 0.210422

SITE: LYS- 103

19 GLU -0.0432644 0.114598

65 HID -0.0370497 -0.130056

70 GLU 0.0268314 0.281407

96 GLY 0.137322 0

98 ASP 0.0731519 0.598695

99 LYS 0.388902 -0.965276

100 VAL 0.358217 0

101 LYS 0.0724803 -0.2684

102 THR -0.448014 0

104 GLN -0.195167 0

105 ILE -0.241152 0

106 TRP -0.198847 0

107 GLN -0.24248 0

112 PRO 0.231648 0

113 THR 0.271876 0

114 ALA 0.729832 0

115 PRO 0.380204 0

116 TYR -0.0780183 0.768386

117 ARG 0.159842 -0.183042

137 ILE 0.175262 0

139 LYS -0.00985965 -2.21719

140 PRO 0.225759 0

141 VAL 0.183542 0

142 HID 0.00584306 -0.197532

143 GLU -0.0049962 0.208592

144 GLY 0.183179 0

150 LYS -0.0668476 -0.112861

168 ASP 0.013806 0.203165

174 GLU 1.2187 4.44154

175 LYS 0.295378 -0.250267

181 GLU -0.170932 0.306767

217 CYS 0.0919515 0.470178

219 ARG -0.00869228 -0.110479

231 ARG 0.050249 -0.366594

233 ASP 0.1217 0.651558

235 MET -0.0385606 0

237 ASP -0.066291 0.181128

244 LEU 0.922744 0

245 LEU 2.038 0

246 GLU 1.07702 2.43419

247 VAL 0.0481169 0

SITE: ASP- 110

106 TRP 0.221946 0

219 ARG -0.0277784 -0.576093

SITE: TYR- 116

98 ASP 0.107954 0.446377

99 LYS 0.168104 -0.307243

100 VAL 0.185601 0

101 LYS 0.166705 -0.225398

102 THR -0.175848 0

103 LYS -0.487124 -0.768386

104 GLN 1.60331 0

488

107 GLN 0.239466 0

114 ALA 0.294178 0

117 ARG -0.153006 -0.373334

131 GLU -0.0148023 0.147733

139 LYS 0.027995 -0.316917

142 HID 0.00212911 -0.145547

168 ASP 0.0147193 0.124074

172 MET -0.326491 0

174 GLU -0.0778508 0.48341

245 LEU 0.141914 0

246 GLU 0.0273936 0.284013

SITE: ARG- 117

103 LYS -0.026923 -0.183042

116 TYR -0.176193 0.373334

124 ASP 0.107383 0.140038

127 SER 0.600762 0

128 VAL 0.114396 0

131 GLU 0.015343 1.70553

139 LYS 0.0184092 -0.119247

174 GLU -0.0934762 0.221189

175 LYS -0.0445424 -0.156388

SITE: LYS- 121

120 THR -0.186546 0

122 GLU -0.392814 0.216162

123 THR 0.166266 0

161 ILE 0.108168 0

162 GLU 1.09653 2.05709

163 LYS -0.00334706 -0.172236

164 ALA -0.202983 0

165 THR 0.65746 0

167 HIS -0.0767605 -0.1208

168 ASP 0.138747 0.213239

169 ALA 0.165259 0

171 VAL -0.0736879 0

SITE: GLU- 122

121 LYS 0.265686 -0.216162

123 THR -0.149988 0

124 ASP 0.0108324 0.11517

SITE: ASP- 124

117 ARG -0.115463 -0.140038

122 GLU 0.0124348 0.11517

123 THR -0.190468 0

125 LEU -0.521956 0

126 ASP -1.12859 0.610341

127 SER -2.37629 0

128 VAL -0.183661 0

131 GLU -0.108911 0.14979

SITE: ASP- 126

124 ASP 0.105688 0.610341

125 LEU 0.136625 0

127 SER -0.441204 0

128 VAL -0.320278 0

129 ILE -0.373846 0

130 ALA -0.198655 0

131 GLU -0.153252 0.133359

153 LYS -0.00882049 -0.127041

155 GLU 0.0229474 0.115964

SITE: GLU- 131

116 TYR 0.0411694 0.147733

117 ARG -0.267507 -1.70553

124 ASP 0.0705872 0.14979

126 ASP 0.125401 0.133359

127 SER 0.270234 0

128 VAL 0.263288 0

130 ALA -0.144654 0

132 LEU -0.230237 0

175 LYS -0.0401523 -0.158716

SITE: LYS- 139

19 GLU -0.0597166 0.176008

65 HID -0.040021 -0.178973

70 GLU 0.0255068 0.33804

489

71 ASP -0.0290584 0.105248

96 GLY 0.141821 0

98 ASP -0.110231 0.416857

99 LYS -0.34061 -1.41593

100 VAL 0.0412747 0

101 LYS -0.0317496 -0.166303

102 THR -0.200942 0

103 LYS -0.145975 -2.21719

112 PRO 0.101339 0

113 THR -0.157205 0

115 PRO 0.279258 0

116 TYR -0.150877 0.316917

117 ARG 0.164363 -0.119247

136 VAL -0.178692 0

137 ILE 0.25789 0

138 ILE 0.00786079 0

140 PRO 0.305011 0

141 VAL 0.23345 0

142 HID -0.00522458 -0.219993

143 GLU -0.00786681 0.35306

144 GLY 0.256438 0

146 VAL 0.257139 0

147 GLY 0.0223507 0

148 MET -0.367549 0

149 SER 0.0663839 0

150 LYS -0.0833378 -0.193572

152 GLU 0.00379805 0.102079

163 LYS -0.0156579 -0.219811

167 HIS -0.0953052 -0.19818

168 ASP 0.00355185 0.315567

171 VAL -0.23039 0

172 MET 0.681885 0

173 ALA -0.171729 0

174 GLU 1.04299 3.31371

175 LYS 0.367912 -0.233735

176 TRP -0.0409808 0

177 ILE 0.0952 0

181 GLU -0.226801 0.500142

217 CYS 0.0517444 0.287406

231 ARG 0.0506289 -0.480804

233 ASP -0.0978371 0.734458

235 MET -0.103483 0

237 ASP -0.0643396 0.147738

244 LEU 0.116328 0

245 LEU -0.117334 0

246 GLU 1.66396 3.69875

247 VAL 0.0893703 0

SITE: HID- 142

19 GLU -0.0206198 0.148036

65 HID -0.0109921 -0.1013

67 ARG -0.109474 -0.234196

70 GLU -0.0352256 0.245079

71 ASP -0.102247 0.392703

95 ILE 0.154224 0

97 MET 0.165161 0

98 ASP -0.430522 1.23462

99 LYS -0.124831 -0.664225

101 LYS 0.0949301 -1.39874

103 LYS -0.0575204 -0.197532

116 TYR -0.0447728 0.145547

139 LYS 0.0149692 -0.219993

140 PRO -0.158649 0

141 VAL -0.289076 0

143 GLU -0.585338 0.554836

144 GLY 0.0637906 0

167 HIS -0.00163445 -0.154693

168 ASP 0.67883 1.79055

170 VAL 0.218822 0

174 GLU 0.0169845 0.158374

231 ARG 0.00667833 -0.103436

233 ASP -0.00023166 0.119225

490

246 GLU -0.0493066 0.233837

SITE: GLU- 143

16 LYS -0.00351154 -0.116375

17 SER -0.240502 0

18 ALA -0.163376 0

19 GLU -0.0526953 0.698784

65 HID -0.0275916 -0.295996

66 GLY -0.200496 0

67 ARG -0.181079 -0.287021

70 GLU -0.0624862 0.497046

71 ASP -0.0478965 0.356005

97 MET 0.262054 0

98 ASP -0.148819 0.300709

99 LYS -0.0588234 -1.07274

101 LYS -0.0365108 -0.174476

103 LYS -0.0354368 -0.208592

139 LYS -0.132058 -0.35306

140 PRO -0.148272 0

141 VAL 0.111724 0

142 HID 0.207374 -0.554836

144 GLY 0.0550365 0

146 VAL -0.0271118 0

147 GLY -0.010888 0

163 LYS 0.109598 -0.297603

164 ALA 0.163124 0

167 HIS -0.273477 -0.773864

168 ASP -0.0908148 0.699149

170 VAL 0.137066 0

174 GLU 0.0244321 0.194437

181 GLU -0.0218283 0.115939

231 ARG -0.0116984 -0.278069

233 ASP -0.0070879 0.28186

246 GLU -0.0905987 0.411945

SITE: LYS- 150

103 LYS -0.00736123 -0.112861

135 PRO -0.3895 0

136 VAL 0.244086 0

139 LYS 0.0105249 -0.193572

151 VAL -0.0084199 0

152 GLU 0.878578 2.06744

156 ASP -0.0189605 0.284809

174 GLU 0.0573808 0.201016

175 LYS -0.0275023 -0.172624

176 TRP 0.297104 0

181 GLU -0.0389812 0.175016

238 GLU 0.0214429 0.155836

246 GLU 0.022664 0.126847

SITE: GLU- 152

135 PRO -0.33912 0

136 VAL 0.143788 0

139 LYS 0.00911504 -0.102079

150 LYS -0.182221 -2.06744

151 VAL -0.0213759 0

153 LYS 0.0400686 -0.17936

155 GLU -0.0302042 0.104001

156 ASP 0.0232931 0.521922

174 GLU 0.039037 0.112196

175 LYS -0.028372 -0.134214

SITE: LYS- 153

126 ASP 0.036756 0.127041

152 GLU -0.246746 0.17936

154 ALA -0.299968 0

155 GLU 0.838943 1.85594

156 ASP -0.105898 0.372485

SITE: GLU- 155

126 ASP 0.0249752 0.115964

152 GLU -0.0976294 0.104001

153 LYS -0.397532 -1.85594

156 ASP -0.452336 0.52582

SITE: ASP- 156

150 LYS 0.0665345 -0.284809

491

151 VAL -0.744414 0

152 GLU -1.48946 0.521922

153 LYS 0.31088 -0.372485

154 ALA 0.173445 0

155 GLU -0.378207 0.52582

157 PHE -0.146717 0

159 ALA -0.203009 0

160 ALA -0.228863 0

SITE: GLU- 162

121 LYS -0.170621 -2.05709

157 PHE 0.169777 0

158 ALA 0.421348 0

159 ALA 0.196577 0

161 ILE -0.0293715 0

163 LYS -0.213774 -0.301208

164 ALA -0.19861 0

165 THR -0.454526 0

167 HIS -0.0535166 -0.148506

168 ASP 0.0477824 0.127896

SITE: LYS- 163

19 GLU 0.0064485 0.156711

99 LYS -0.0244577 -0.226493

121 LYS 0.00327416 -0.172236

139 LYS -0.100125 -0.219811

140 PRO -0.0848904 0

143 GLU -0.0422568 0.297603

146 VAL -0.216328 0

147 GLY 0.142548 0

148 MET 0.316517 0

149 SER -0.0743754 0

159 ALA 0.18827 0

160 ALA 0.36829 0

161 ILE 0.238074 0

162 GLU -0.00922544 0.301208

164 ALA -0.102636 0

165 THR -0.274808 0

166 GLN -0.733492 0

167 HIS 0.0634215 -2.41797

168 ASP 0.00816273 0.220546

174 GLU 0.0127484 0.124841

246 GLU 0.00350231 0.152035

SITE: HIS- 167

19 GLU 0.0278494 0.326063

65 HID -0.00051024 -0.114745

67 ARG -0.0504219 -0.132703

70 GLU -0.0146966 0.150158

71 ASP -0.00914312 0.110664

99 LYS -0.0268754 -0.306922

121 LYS -0.000938699 -0.1208

139 LYS -0.122749 -0.19818

140 PRO -0.189906 0

142 HID -0.0320387 -0.154693

143 GLU -0.00335802 0.773864

146 VAL -0.3409 0

147 GLY -0.369016 0

161 ILE 0.144705 0

162 GLU 0.161114 0.148506

163 LYS 1.05623 -2.41797

164 ALA 0.509239 0

166 GLN -0.731218 0

168 ASP -0.379965 0.537361

174 GLU 0.0102313 0.106231

231 ARG 0.00196302 -0.112491

233 ASP -0.00346673 0.113878

246 GLU -0.0132893 0.166225

SITE: ASP- 168

19 GLU -0.00624201 0.182311

67 ARG -0.0727405 -0.248555

70 GLU -0.035707 0.191808

71 ASP -0.0526629 0.271232

97 MET 0.118327 0

492

98 ASP -0.253388 0.538931

99 LYS -0.157168 -0.692157

101 LYS -0.0407868 -0.313983

103 LYS -0.0505419 -0.203165

116 TYR -0.0590106 0.124074

121 LYS -0.00203174 -0.213239

139 LYS 0.0163995 -0.315567

140 PRO -0.809736 0

141 VAL -0.992682 0

142 HID -1.3595 -1.79055

143 GLU -0.417554 0.699149

144 GLY 0.102342 0

147 GLY -0.00119962 0

161 ILE 0.122502 0

162 GLU 0.0863404 0.127896

163 LYS 0.0746957 -0.220546

164 ALA 0.3424 0

166 GLN -0.178949 0

167 HIS -0.0833276 -0.537361

169 ALA -0.599272 0

170 VAL 0.656655 0

171 VAL 0.0307277 0

174 GLU 0.0155505 0.197214

231 ARG 0.00866627 -0.105337

233 ASP -0.0065284 0.121509

246 GLU -0.0176405 0.255128

SITE: GLU- 174

19 GLU -0.0304881 0.105396

65 HID -0.0267043 -0.112036

70 GLU 0.0202984 0.21864

96 GLY 0.0946479 0

98 ASP -0.026272 0.365401

99 LYS -0.0336876 -0.790102

100 VAL 0.0839867 0

101 LYS -0.00490724 -0.149192

102 THR -0.206058 0

103 LYS 0.0409347 -4.44154

107 GLN -0.164167 0

112 PRO 0.122231 0

113 THR -0.331582 0

114 ALA 0.2728 0

115 PRO 0.834612 0

116 TYR -0.279623 0.48341

117 ARG 0.227936 -0.221189

133 GLY 0.177746 0

135 PRO 0.112604 0

136 VAL -0.413718 0

137 ILE 0.38817 0

139 LYS 0.0475762 -3.31371

140 PRO 0.217288 0

141 VAL 0.120649 0

142 HID 0.00494011 -0.158374

143 GLU -0.0010932 0.194437

144 GLY 0.122986 0

146 VAL 0.143476 0

148 MET -0.231416 0

149 SER 0.0937024 0

150 LYS -0.120785 -0.201016

152 GLU 0.00769776 0.112196

163 LYS -0.0253186 -0.124841

167 HIS -0.0430317 -0.106231

168 ASP 0.013541 0.197214

172 MET 0.464506 0

173 ALA -0.240564 0

175 LYS 0.495572 -0.513087

177 ILE 0.105948 0

181 GLU -0.181904 0.357071

217 CYS 0.0532922 0.311654

231 ARG 0.0312055 -0.320044

233 ASP -0.00302918 0.529328

235 MET -0.131326 0

493

237 ASP -0.086004 0.227652

244 LEU 0.51473 0

245 LEU 0.226068 0

246 GLU 0.761976 1.80319

247 VAL 0.0478532 0

SITE: LYS- 175

103 LYS -0.012955 -0.250267

115 PRO 0.329972 0

117 ARG 0.101617 -0.156388

128 VAL 0.119601 0

129 ILE 0.149983 0

131 GLU -0.0443563 0.158716

132 LEU 1.23221 0

133 GLY 2.00222 0

134 LEU 0.219295 0

135 PRO 0.869186 0

136 VAL -0.0932717 0

139 LYS 0.0151887 -0.233735

150 LYS -0.0809224 -0.172624

151 VAL 0.111007 0

152 GLU 0.018895 0.134214

174 GLU -0.244354 0.513087

176 TRP 0.231565 0

237 ASP -0.0965006 0.248893

238 GLU -0.0380266 0.12692

246 GLU 0.039027 0.144056

SITE: ARG- 180

179 GLY -0.29804 0

181 GLU -0.0871843 0.21004

196 ARG -0.134756 -0.246314

203 CYS 0.0537029 0.416376

204 GLY 0.949394 0

209 GLU -0.0907769 0.489195

210 GLU -0.0129768 0.199945

212 LYS -0.056081 -0.103684

217 CYS -0.00800574 0.15727

233 ASP -0.0280416 0.107829

236 GLN 2.01932 0

237 ASP 0.121102 0.263636

238 GLU -0.0819145 0.246969

SITE: GLU- 181

19 GLU -0.0342142 0.112029

65 HID -0.0244368 -0.126331

70 GLU 0.0185286 0.174802

99 LYS -0.0139673 -0.281463

103 LYS -0.0099919 -0.306767

139 LYS -0.0057803 -0.500142

143 GLU 0.0312304 0.115939

150 LYS -0.0325541 -0.175016

174 GLU 0.0318232 0.357071

177 ILE 0.11114 0

180 ARG 0.0140218 -0.21004

195 ILE 0.133895 0

196 ARG -0.422363 -0.260149

197 LEU -0.718953 0

198 GLN -0.801049 0

203 CYS 0.0079182 0.161228

217 CYS 0.022562 0.235059

231 ARG 0.0643543 -0.457962

233 ASP -0.186127 0.590521

234 ALA 0.236281 0

237 ASP -0.0357732 0.134615

238 GLU 0.00574055 0.139741

246 GLU 0.0290897 0.488936

261 LYS -0.033746 -0.111987

SITE: ARG- 196

180 ARG -0.121815 -0.246314

181 GLU 0.29082 0.260149

182 PHE -0.156589 0

197 LEU 0.294956 0

198 GLN 0.566581 0

494

199 TYR 1.12377 0.117527

200 GLY -0.128276 0

202 PRO 1.70992 0

203 CYS 0.361221 0.584559

204 GLY 0.0493112 0

210 GLU 0.0304554 0.203975

217 CYS 0.000756536 0.139248

231 ARG -0.018485 -0.158061

233 ASP -0.0225824 0.174495

261 LYS -0.0523783 -0.129344

SITE: TYR- 199

196 ARG 0.0701634 -0.117527

256 HIS 0.0812324 -0.13786

260 PRO 0.0473608 0

261 LYS -0.225924 -0.592007

264 LYS -0.104862 -0.383476

SITE: CYS- 203

180 ARG 0.120772 -0.416376

181 GLU 0.128651 0.161228

182 PHE -0.217 0

183 THR 0.137819 0

193 PRO -0.157917 0

194 VAL 0.307226 0

196 ARG -0.19249 -0.584559

200 GLY 0.71368 0

201 ILE -0.213272 0

202 PRO -0.26391 0

204 GLY -0.74844 0

205 LEU -0.672342 0

206 SER 0.388964 0

207 GLU -0.0242579 0.208025

209 GLU -0.158188 0.317486

210 GLU 0.565971 1.96831

211 LYS -0.00200377 -0.174046

212 LYS -0.0996259 -0.101547

213 LEU -0.217086 0

214 GLN -0.166347 0

217 CYS -0.0292398 0.329158

231 ARG -0.0476641 -0.151013

233 ASP -0.0178216 0.183768

SITE: GLU- 207

203 CYS -0.0346528 0.208025

206 SER 0.320466 0

208 THR -1.4331 0

209 GLU -0.0598298 0.270544

210 GLU -0.198451 0.289472

211 LYS -0.0192305 -2.0548

212 LYS -0.0484197 -0.130302

SITE: GLU- 209

180 ARG 0.211414 -0.489195

203 CYS -0.0372143 0.317486

205 LEU -0.717484 0

206 SER -1.9242 0

207 GLU -0.0321024 0.270544

208 THR -0.532181 0

210 GLU -0.376844 0.39391

211 LYS -0.0683674 -0.271501

212 LYS -0.267396 -0.338993

213 LEU -0.187306 0

217 CYS -0.0284082 0.166647

237 ASP 0.0901051 0.143582

240 GLY 0.0779674 0

SITE: GLU- 210

180 ARG 0.0859591 -0.199945

183 THR 0.153022 0

192 LEU 0.270502 0

193 PRO -0.69605 0

194 VAL -0.659464 0

195 ILE 0.0771709 0

196 ARG 0.0231113 -0.203975

200 GLY 0.142259 0

495

202 PRO -0.2001 0

203 CYS -0.991015 1.96831

205 LEU -0.21353 0

206 SER 0.449362 0

207 GLU 0.0461843 0.289472

209 GLU -0.229518 0.39391

211 LYS -0.235076 -0.374904

212 LYS -0.189258 -0.180903

213 LEU -0.364756 0

214 GLN -0.607723 0

217 CYS -0.099775 0.466942

231 ARG -0.0573132 -0.15308

233 ASP 0.00779591 0.183163

SITE: LYS- 211

203 CYS -0.0444674 0.174046

206 SER 0.231849 0

207 GLU 1.74183 2.0548

208 THR 0.259402 0

209 GLU 0.06958 0.271501

210 GLU -0.181644 0.374904

212 LYS -0.219922 -0.350611

217 CYS -0.046883 0.153682

SITE: LYS- 212

180 ARG 0.0481993 -0.103684

203 CYS -0.0259488 0.101547

207 GLU 0.0639833 0.130302

208 THR 0.428244 0

209 GLU 0.132458 0.338993

210 GLU 0.110061 0.180903

211 LYS 0.194373 -0.350611

213 LEU -0.165865 0

217 CYS -0.0751558 0.260064

219 ARG -0.0203268 -0.155834

237 ASP 0.0299318 0.105785

SITE: CYS- 217

65 HID -0.0313642 -0.134643

70 GLU 0.0272604 0.242581

98 ASP 0.023798 0.125991

99 LYS 0.0386688 -0.232564

103 LYS 0.0338712 -0.470178

111 LEU -0.197048 0

112 PRO 0.342612 0

113 THR -0.17635 0

139 LYS -0.00263434 -0.287406

174 GLU 0.0493047 0.311654

180 ARG 0.15255 -0.15727

181 GLU -0.133851 0.235059

183 THR -0.313809 0

184 ILE -0.118282 0

185 SER 0.17535 0

186 PHE -0.193657 0

189 GLY 0.0974722 0

191 PRO -0.446912 0

192 LEU 0.26294 0

193 PRO 0.118762 0

194 VAL -0.35324 0

195 ILE 0.14738 0

196 ARG -0.020878 -0.139248

203 CYS -0.130343 0.329158

209 GLU 0.174732 0.166647

210 GLU 0.337764 0.466942

211 LYS 0.113773 -0.153682

212 LYS 0.145911 -0.260064

213 LEU 1.75031 0

214 GLN 0.527366 0

216 LEU -0.619038 0

219 ARG -0.329724 -0.376022

220 ALA -0.816252 0

221 PHE -0.277423 0

223 ALA -0.20994 0

227 GLU 0.053091 0.138254

496

230 GLY -0.0246131 0

231 ARG -0.414175 -0.562605

232 ILE -0.127887 0

233 ASP 0.288198 1.02802

234 ALA -0.202674 0

235 MET 0.101403 0

237 ASP 0.000945878 0.183573

242 PHE -0.126608 0

244 LEU -0.414187 0

246 GLU 0.268212 0.482606

247 VAL -0.122658 0

251 PRO -0.0526047 0

SITE: ARG- 219

103 LYS 0.0381004 -0.110479

109 SER 0.628668 0

110 ASP 0.943756 0.576093

111 LEU -0.0165534 0

112 PRO 0.0522442 0

212 LYS 0.0732821 -0.155834

216 LEU 0.148715 0

217 CYS 0.0371078 0.376022

220 ALA -0.159894 0

222 GLN -0.163786 0

SITE: GLU- 227

70 GLU 0.0121711 0.144741

79 GLU 0.02196 0.158545

85 TYR 0.0776222 0.283802

87 GLY -0.131014 0

188 ASN -0.132396 0

189 GLY -0.0825195 0

217 CYS 0.0477537 0.138254

226 ALA -0.182319 0

228 GLY -0.279858 0

282 LEU 0.0113612 0.161717

SITE: ARG- 231

19 GLU -0.104225 0.329402

25 ASP -0.0214913 0.143732

65 HID -0.105837 -0.532891

70 GLU 0.0900425 0.687436

71 ASP -0.0346461 0.13983

96 GLY 0.0709424 0

98 ASP -0.0288293 0.169573

99 LYS -0.0280198 -0.576411

103 LYS -0.018477 -0.366594

139 LYS -0.0422639 -0.480804

140 PRO 0.0909185 0

142 HID 0.0361009 -0.103436

143 GLU 0.0594059 0.278069

144 GLY -0.0187213 0

167 HIS -0.0118822 -0.112491

168 ASP -0.00305078 0.105337

174 GLU 0.0461648 0.320044

181 GLU -0.126482 0.457962

183 THR -0.154592 0

184 ILE 0.196311 0

193 PRO 0.180958 0

194 VAL -0.18957 0

195 ILE 0.149181 0

196 ARG -0.00658743 -0.158061

197 LEU -0.184555 0

203 CYS -0.045035 0.151013

210 GLU 0.0574238 0.15308

217 CYS 0.0735068 0.562605

230 GLY -0.0951507 0

232 ILE 0.0577409 0

233 ASP 1.01281 3.01964

246 GLU -0.0731426 0.76912

247 VAL -0.102282 0

248 ASN 0.681042 0

251 PRO -0.299071 0

252 GLY 0.120639 0

497

256 HIS -0.0120254 -0.186397

258 LEU -0.371556 0

259 VAL -0.13519 0

260 PRO -0.0630448 0

261 LYS -0.110959 -0.268605

262 ALA -0.16274 0

264 LYS -0.0374512 -0.111756

274 CYS 0.0290716 0.176629

SITE: ASP- 233

19 GLU -0.087903 0.273

65 HID -0.0905827 -0.369895

70 GLU 0.0739925 0.620753

71 ASP -0.0384341 0.131894

96 GLY 0.101659 0

98 ASP -0.0341872 0.222884

99 LYS -0.0261153 -0.72084

101 LYS -0.0291317 -0.104483

103 LYS -0.0275841 -0.651558

112 PRO 0.142047 0

139 LYS -0.0415966 -0.734458

140 PRO 0.126891 0

141 VAL 0.130695 0

142 HID 0.0313977 -0.119225

143 GLU 0.0897953 0.28186

144 GLY 0.0210858 0

146 VAL 0.108405 0

167 HIS -0.0119157 -0.113878

168 ASP 0.00150821 0.121509

174 GLU 0.0751716 0.529328

180 ARG 0.0791754 -0.107829

181 GLU -0.28398 0.590521

182 PHE 0.585682 0

183 THR -2.44732 0

184 ILE 0.0962876 0

193 PRO 0.129324 0

194 VAL -0.225692 0

195 ILE 0.22731 0

196 ARG -0.0475421 -0.174495

197 LEU -0.20452 0

203 CYS -0.061477 0.183768

210 GLU 0.0753929 0.183163

217 CYS 0.155015 1.02802

230 GLY -0.0973979 0

231 ARG 0.720794 -3.01964

232 ILE 0.128376 0

237 ASP -0.0229559 0.118392

243 TRP 0.144602 0

244 LEU -0.307874 0

245 LEU -0.50291 0

246 GLU 0.168955 1.23989

247 VAL -0.092754 0

251 PRO -0.13612 0

252 GLY 0.0509994 0

256 HIS -0.00412576 -0.128138

258 LEU -0.174478 0

261 LYS -0.078392 -0.182497

274 CYS 0.0183263 0.108237

SITE: ASP- 237

103 LYS 0.00269353 -0.181128

139 LYS 0.00310051 -0.147738

174 GLU -0.0121262 0.227652

175 LYS 0.0234447 -0.248893

177 ILE -0.201492 0

178 THR 0.676162 0

179 GLY -0.273792 0

180 ARG 0.0899626 -0.263636

181 GLU -0.0353127 0.134615

209 GLU -0.0292359 0.143582

212 LYS -0.0295694 -0.105785

217 CYS 0.00174304 0.183573

233 ASP -0.0152595 0.118392

498

236 GLN -1.02795 0

238 GLU -0.913116 0.549505

239 GLN -1.59868 0

240 GLY -1.1285 0

241 ASN -1.87227 0

243 TRP -1.43024 0

246 GLU 0.0353286 0.126168

SITE: GLU- 238

150 LYS -0.0181039 -0.155836

175 LYS 0.00911725 -0.12692

178 THR -1.9273 0

179 GLY 0.183595 0

180 ARG 0.0784435 -0.246969

181 GLU -0.0175029 0.139741

236 GLN -0.0430009 0

237 ASP -0.451318 0.549505

239 GLN 0.0234675 0

SITE: GLU- 246

19 GLU -0.0886348 0.259782

65 HID -0.0899918 -0.283906

70 GLU 0.0381189 0.582799

71 ASP -0.0450241 0.149932

96 GLY 0.224909 0

97 MET 0.203508 0

98 ASP -0.104268 0.514309

99 LYS -0.432519 -1.73951

100 VAL 0.093271 0

101 LYS -0.0576473 -0.210422

102 THR -0.265253 0

103 LYS -0.130125 -2.43419

112 PRO 0.151442 0

113 THR -0.176891 0

115 PRO 0.168133 0

116 TYR -0.064696 0.284013

137 ILE 0.0996739 0

139 LYS -0.221986 -3.69875

140 PRO 0.386545 0

141 VAL 0.308809 0

142 HID 0.0163509 -0.233837

143 GLU 0.104181 0.411945

144 GLY 0.377658 0

146 VAL 0.20976 0

150 LYS -0.0485042 -0.126847

163 LYS -0.01258 -0.152035

167 HIS -0.049947 -0.166225

168 ASP 0.00577844 0.255128

172 MET 0.224062 0

174 GLU 0.533324 1.80319

175 LYS 0.204536 -0.144056

181 GLU -0.213572 0.488936

182 PHE 0.142748 0

183 THR -0.163132 0

217 CYS 0.105341 0.482606

231 ARG 0.121225 -0.76912

232 ILE -0.200634 0

233 ASP -0.131067 1.23989

237 ASP -0.0486586 0.126168

245 LEU -0.159537 0

247 VAL 0.00142593 0

251 PRO -0.0726077 0

SITE: HIS- 256

19 GLU 0.00504395 0.133864

25 ASP 0.51448 1.21419

65 HID -0.0169466 -0.260836

70 GLU 0.0132947 0.131133

199 TYR -0.0488952 0.13786

231 ARG -0.000117894 -0.186397

233 ASP 0.00512991 0.128138

254 THR -0.387223 0

255 SER 0.857489 0

257 SER -0.327294 0

499

261 LYS 0.00963593 -0.498314

264 LYS -0.0183181 -0.126705

274 CYS 0.00860871 0.126557

SITE: LYS- 261

25 ASP 0.0289112 0.184674

36 ARG -0.00978886 -0.109305

65 HID -0.0222313 -0.187766

70 GLU 0.0132313 0.120941

181 GLU 0.0198469 0.111987

193 PRO 0.147027 0

196 ARG 0.0964058 -0.129344

199 TYR -0.139212 0.592007

231 ARG -0.0315636 -0.268605

233 ASP 0.00982261 0.182497

252 GLY 0.10519 0

254 THR 0.646637 0

255 SER 1.27741 0

256 HIS 0.466554 -0.498314

257 SER 1.95488 0

258 LEU 0.530373 0

259 VAL 0.111412 0

260 PRO -0.179399 0

262 ALA -0.207408 0

264 LYS -0.160675 -0.514775

271 ASP 0.0350055 0.136515

274 CYS -0.00160197 0.23508

SITE: LYS- 264

25 ASP 0.0385995 0.123102

32 ASP -0.0424952 0.10904

36 ARG -0.0700863 -0.192214

65 HID -0.0174299 -0.117352

193 PRO 0.0744799 0

199 TYR 0.00726387 0.383476

231 ARG -0.0248486 -0.111756

254 THR 0.202578 0

256 HIS 0.0225918 -0.126705

258 LEU 0.173527 0

259 VAL 0.232214 0

260 PRO 1.43758 0

261 LYS 0.265034 -0.514775

263 ALA -0.252977 0

265 ALA -0.29544 0

266 VAL -0.193464 0

268 TYR 0.846568 0.161427

270 PHE 0.0695724 0

271 ASP 0.0544605 0.392799

272 GLU -0.103322 0.257074

273 LEU -0.279428 0

274 CYS -0.0695206 0.499774

SITE: TYR- 268

5 LYS 0.00234446 -0.174988

264 LYS 0.0218222 -0.161427

269 SER 0.14618 0

271 ASP 0.0398864 0.130982

272 GLU 0.0178168 0.341497

273 LEU 0.0815573 0

274 CYS 0.00751675 0.282848

276 ALA -0.219892 0

279 GLU -0.133347 0.294165

SITE: ASP- 271

5 LYS -0.0117152 -0.107514

25 ASP 0.0486134 0.103902

29 ALA 0.210496 0

30 VAL 0.147283 0

32 ASP -0.0800832 0.374802

33 ALA -0.172114 0

36 ARG -1.72636 -0.909549

37 SER -0.261712 0

65 HID -0.0118533 -0.100219

261 LYS 0.0162236 -0.136515

264 LYS -0.0102575 -0.392799

500

268 TYR -0.0682913 0.130982

269 SER 0.247172 0

272 GLU -0.54871 0.530156

273 LEU -0.275478 0

274 CYS -0.490358 1.2507

279 GLU -0.0446378 0.110528

SITE: GLU- 272

5 LYS -0.00915421 -0.145611

36 ARG -0.122563 -0.101641

264 LYS -0.00656423 -0.257074

268 TYR -0.553805 0.341497

269 SER -1.93726 0

271 ASP -0.264784 0.530156

273 LEU -0.230838 0

274 CYS -0.177721 0.448105

275 VAL -0.170037 0

276 ALA -0.17072 0

279 GLU -0.0445979 0.167236

SITE: CYS- 274

5 LYS -0.0353268 -0.248511

11 VAL -0.175367 0

23 SER 0.108656 0

25 ASP 0.122401 0.242671

26 SER 0.456582 0

27 GLY 0.298589 0

28 GLN 0.0675826 0

29 ALA 0.150551 0

30 VAL 0.0694087 0

31 LEU 0.2365 0

32 ASP -0.277012 0.412635

33 ALA -0.87274 0

34 LEU -0.29165 0

36 ARG -0.273556 -0.395467

37 SER -0.335265 0

41 ALA -0.15369 0

42 GLU 0.0498062 0.106667

45 ASP -0.0301214 0.101717

57 TYR -0.0251246 0.110074

59 ARG -0.0741233 -0.160953

62 ILE 0.18556 0

63 VAL -0.0305907 0

65 HID -0.068654 -0.459252

70 GLU 0.0128679 0.191767

85 TYR -0.116806 0.102992

186 PHE 0.141478 0

188 ASN 0.122011 0

230 GLY 0.0578948 0

231 ARG -0.0267741 -0.176629

233 ASP 0.0144385 0.108237

251 PRO 0.246415 0

252 GLY 0.0242972 0

253 MET -0.734202 0

254 THR 0.24254 0

256 HIS -0.0243819 -0.126557

259 VAL -0.103606 0

260 PRO -0.0957519 0

261 LYS -0.00562214 -0.23508

263 ALA -0.18211 0

264 LYS -0.0274027 -0.499774

268 TYR -0.0329369 0.282848

269 SER 0.49963 0

270 PHE 3.13472 0

271 ASP 0.849684 1.2507

272 GLU -0.0521936 0.448105

273 LEU -0.512946 0

275 VAL -0.0617454 0

276 ALA -0.29426 0

277 ILE -0.695118 0

278 LEU -0.339612 0

279 GLU -0.201806 0.291613

280 GLN -0.169801 0

501

SITE: GLU- 279

1 SER 0.0450938 -0.107086

2 ASN 0.323002 0

5 LYS -0.137065 -2.38323

59 ARG -0.0604065 -0.243994

268 TYR -0.0952314 0.294165

271 ASP 0.0679707 0.110528

272 GLU 0.0737092 0.167236

273 LEU 0.11972 0

274 CYS 0.107054 0.291613

275 VAL 0.5785 0

276 ALA 0.214888 0

277 ILE 0.0715424 0

278 LEU -0.110609 0

280 GLN -0.365568 0

281 THR -0.299164 0

282 LEU -0.136758 0.25692

SITE: CTleu- 282

5 LYS 0.0514814 -0.171468

58 ASP 0.00306754 0.108369

59 ARG 0.0628078 -0.42072

79 GLU 0.0237638 0.105509

85 TYR 0.0547911 0.139093

227 GLU -0.0168212 0.161717

279 GLU 0.305984 0.25692

280 GLN 0.203912 0

282 LEU 0.171911 0

502

S-Table.7.4. MM/GBSA free energy decomposition into most active residue of the enzyme.

Residue van der

Waals Electrostatic

Polar

Solvation

Non-Polar

Solvation TOTAL

LYS 100 -1.5 0.4 -1.0 -0.1 -2.2

ILE 138 -1.0 -0.1 0.2 -0.1 -1.0

GLY 145 -0.9 -0.6 0.5 -0.1 -1.1

GLU 175 -0.7 -8.0 7.0 -0.1 -1.7

TRP 177 -1.7 -1.5 0.7 -0.2 -2.7

ILE 178 -1.1 -1.2 0.5 -0.1 -1.9

LEU 245 -2.2 -0.5 0.9 -0.3 -2.1

ASN 248 -1.0 0.2 -0.3 -0.2 -1.3

503

Chapter # 8

S-Fig.8.1. A. Secondary structure of the modeled MurC and B. Ramachandran plot for the predicted MurC enzyme.

504

S-Fig.8.2. Extended RMSD of 150-ns for MurC enzyme.

505

S-Fig.8.3

506

S-Fig.8.4

507

S-Fig.8.5

508

S-Fig.8.6

509

S-Fig.8.7

S-Fig.8.3-S-Fig.8.7. Superimposition of complex at frequency of 5-ns.

510

S-Fig.8.8. Hot Spot residues in MM/GBSA (A) and MM/PBSA (B). Energy values are in kcal/mol.

511

S-Fig.8.9. Binding free energy decomposition of hot spot amino acids into its components.

512

S-Table.8.1. Structure evaluation of the predict MurC models.

R.M.F.R, Residues in most favored region, R.A.A.R, Residues in additionally allowed region, R.G.A.R, Residues in generously

allowed region, R.D.R, Residues in disallowed region.

Structure

Source R.M.F.R R.A.A.R R.G.A.R R.D.R G Factor

ERRAT

Quality

Factor

VERIFY-3D

(3D-1D score

> 0.2 (%))

ProSA-

Web Z-

score

ModWeb 382 30 1 1 -0.07 83.76 90.97 -11.13

Phyre2 308 37 5 1 0.14 86.45 91.57 -4.53

Swiss-model 364 30 3 2 -0.1 91.98 96.31 -11.79

Modeller-1 361 39 3 5 0.33 50 29.57 0.87

Modeller-2 364 32 10 2 -0.32 54.68 28.72 0.87

Modeller-3 365 36 4 3 -0.24 55.09 27.66 0.77

Modeller-4 365 36 4 3 -0.24 50.32 24.47 0.77

Modeller-5 369 34 5 0 -0.27 47.71 27.87 0.61

513

S-Table.8.2. Structure and docking scores for the shortlisted Top-10 inhibitors.

Structure GOLD

score

Chem

score

Chem

score

PLP

ASP

Auto-

Dock/Vina

(kcal/mol)

1'-((2H-imidazol-2-yl)methyl)-N-(pyridin-2-yl)-1',2'-dihydro-[4,4'-

bipyridin]-2-amine

72.05 18.61 62.58 28.00 -9.5

(R)-2-(2-(3-hydroxy-4-(4H-1,4-oxazin-4-yl)-1H-pyrrole-1-carbonyl)buta-

2,3-dien-1-yl)-3-methyloxazolidin-5-one

69.95 18.38 53.18 20.16 -9.5

514

N-((5-((1-(4,5-dihydro-1H-pyrazole-3-carbonyl)-1,2-dihydropyridin-4-

yl)methyl)-5,6-dihydropyrazin-2-yl)methyl)acetamide

68.47 25.90 66.19 31.78 -9.1

6-((1-((7aH-indol-6-yl)methyl)-1,2-dihydropyridin-3-yl)methyl)-N,N-

dimethylpyrimidine-4-carboxamide

67.06 23.30 61.43 20.88 -8.4

515

2-(2-(1-(2-(dimethylcarbamoyl)-2H-1,4-oxazin-4(3H)-yl)buta-2,3-dien-2-

yl)cyclopropyl)cyclopenta-1,4-dienecarboxylic acid

66.10 32.23 64.04 25.92 -8.0

5-(3-hydroxy-3-(phenoxymethyl)-2,3-dihydro-1H-pyrrole-1-

carbonyl)pyrimidin-4(3H)-one

65.85 32.23 60.97 31.78 -7.8

516

6-(1-((6-(tert-butyl)pyridin-2-yl)methyl)-1,6-dihydropyridin-3-yl)-2-methyl-

2H-pyrazolo[3,4-b]pyridine

65.37 27.63 60.75 26.74 -7.1

(1-(tert-butyl)-1,2-dihydropyridin-2-yl)(5-(2-(4-(tert-butyl)-1H-pyrazol-5-

yl)propan-2-yl)pyridin-1(2H)-yl)methanone

65.37 22.98 54.72 21.39 -7.2

517

3-(N-(4-(4-(tert-butyl)-2,5-dihydrothiophene-2-carbonyl)-2,5-

dimethylhexan-2-yl)-2-methylpropan-2-ylsulfonamido)-N,N,2,3-

tetramethylbutanamide

65.04 14.11 59.83 24.69 -6.5

236

N-(tert-butyl)-1-(3-(5-(tert-butyl)-2-oxobenzo[d]oxazol-3(2H)-yl)-2,2,3-

trimethylbutanoyl)-N-methyl-1,6-dihydropyrazine-2-carboxamide

64.92 16.41 44.54 20.76 -6.4

518

Control 1 (Aztreonam)

42.12 17.22 45.22 25.21 -6.1

Control 2 (Cefclidin)

45.23 19.31 44.71 26.22 -6.5


519

S-Table.8.3. Hydrogen bond analysis for complex at frequency of 5-ns.

Snapshot Number of

hydrogen bonds

Protein residue:

Atom

Inhibitor atom Distance

0-ns 1 Asp334:OD2 HN 1.8 Å

5-ns 1 Asp334:OD2 HH 1.9 Å

10-ns 3 Arg315:HH21

Asp334:OD2

Thr345:OG1

N23

NH

H47

2.1 Å

1.8 Å

2.3 Å

15-ns 4 Arg314: O

Asp334:OD2

Asp334:OD1

Thr345:OG1

H32

HN

HN

H47

2.3 Å

1.7 Å

2.2 Å

1.8 Å

20-ns 4 Arg314: O

Asp334:OD2

Asp334:OD1

Glu341: O

H32

HN

HN

H47

2.0 Å

1.8 Å

2.2 Å

2.2 Å

25-ns 2 Asp334:OD2

Thr345:OG1

HN

H47

1.9 Å

1.9 Å

30-ns 2 Asp334:OD2

Glu341: OE2

HN

H47

1.7 Å

1.8 Å

35-ns 3 Arg314: O

Asp334:OD2

Asp334:OD1

H32

HN

HN

2.4 Å

2.1 Å

2.1 Å

40-ns 4 Arg315:N

Asp334:OD2

Asp334:OD2

Asp334:OD1

H32

H47

HN

HN

2.2 Å

2.0 Å

2.0 Å

2.0 Å

45-ns 3 Arg314: O

Asp334:OD2

Asp334:OD2

H32

HN

H47

2.2 Å

1.8 Å

2.2 Å

520

50-ns 4 Arg314: O

Asp334:OD2

Asp334:OD2

Asp334:OD1

H32

H47

HN

HN

2.0 Å

2.1 Å

2.1 Å

2.0 Å

55-ns 2 Arg314: O

Asp334:OD2

H32

HN

2.3 Å

1.7 Å

60-ns 4 Arg314: O

Asp334:OD2

Asp334:OD2

Asp334:OD1

H32

H47

HN

HN

2.2 Å

1.9 Å

2.1 Å

1.8 Å

65-ns 4 Arg314: O

Asp334:OD2

Asp334:OD2

Asp334:OD1

H32

H47

HN

HN

1.9 Å

2.0 Å

1.9 Å

2.2 Å

70-ns 4 Arg314: O

Asp334:OD2

Asp334:OD2

Asp334:OD1

H32

H47

HN

HN

2.2 Å

1.9 Å

1.9 Å

2.1 Å

75-ns 4 Arg314: O

Asp334:OD2

Asp334:OD2

Asp334:OD1

H32

H47

HN

HN

2.1 Å

2.0 Å

1.8 Å

2.2 Å

80-ns 3 Asp334:OD2

Asp334:OD2

Asp334:OD1

H47

HN

HN

2.1 Å

2.1 Å

2.2 Å

85-ns 3 Arg314: O

Asp334:OD2

Asp334:OD1

H32

HN

HN

2.4 Å

2.0 Å

2.1 Å

90-ns 3 Arg314: O

Asp334:OD2

Asp334:OD1

H32

HN

HN

2.0 Å

2.0 Å

1.9 Å

521

95-ns 3 Arg314: O

Asp334:OD2

Asp334:OD2

H32

HN

HN

2.2 Å

1.9 Å

1.9 Å

100-ns 3 Arg314: O

Asp334:OD2

Asp334:OD2

H32

H47

HN

2.4 Å

2.1 Å

1.9 Å

522

Chapter # 9

S-Fig.9.1. Structure comparison of MurF ligase enzyme from E.coli (red), S. pneumoniae (magenta) and T. maritima (light green)

with AbMurF (red).

https://en.wikipedia.org/wiki/Thermotoga_maritima

523

S-Fig.9.2.(A) Docked pose of compound 114 in binding cavity of AbMurF. B. Ligplot highlighting residues of a protein involved in

interaction with the ligand.

524

S-Fig.9.3. Decomposition of MM/GBSA free energy for pair-wise residues.

525

S-Fig.9.4. Multiple sequence alignment of MurF ligase enzyme from different bacterial species. Thr42 and Asp43 are indicated by a

red box.

526

S-Table.9.1. Structures of anti-MurF compounds used in the current study.

Ligand Compound Structure

Compound 1

Compound 2

Compound 3

Compound 4

Compound 5

527

Compound 6

Compound 7

Compound 8

Compound 9

528

Compound

10

Compound

11

Compound

12

Compound

13

Compound

14

529

Compound

15

Compound

16

Compound

17

Compound

18

530

Compound

19

Compound

20

Compound

21

Compound

22

531

Compound

23

Compound

24

Compound

25

Compound

26

532

Compound

27

Compound

28

Compound

29

Compound

30

533

Compound

31

Compound

32

Compound

33

Compound

34

534

Compound

35

Compound

36

Compound

37

Compound

38

Compound

39

535

Compound

40

Compound

41

Compound

42

Compound

43

Compound

44

536

Compound

45

Compound

46

Compound

47

Compound

48

537

Compound

49

Compound

50

Compound

51

Compound

52

538

Compound

53

Compound

54

Compound

55

Compound

56

539

Compound

57

Compound

58

Compound

59

Compound

60

540

Compound

61

Compound

62

Compound

63

Compound

64

541

Compound

65

Compound

66

Compound

67

542

Compound

68

Compound

69

Compound

70

Compound

71

543

Compound

72

Compound

73

Compound

74

Compound

75

544

Compound

76

Compound

77

Compound

78

Compound

79

Compound

80

Compound

81

545

Compound

82

Compound

83

Compound

84

Compound

85

Compound

86

546

Compound

87

Compound

88

Compound

89

Compound

90

547

Compound

91

Compound

92

Compound

93

548

Compound

94

Compound

95

Compound

96

549

Compound

97

Compound

98

Compound

99

550

Compound

100

Compound

101

Compound

102

551

Compound

103

Compound

104

Compound

105

552

Compound

106

Compound

107

Compound

108

553

Compound

109

Compound

110

Compound

111

554

Compound

112

Compound

113

Compound

114

Compound

115

555

Compound

116

Compound

117

Compound

118

556

Compound

119

Compound

120

Compound

121

557

Compound

122

Compound

123

Compound

124

558

Compound

125

Compound

126

Compound

127

559

Compound

128

Compound

129

Compound

130

Compound

131

560

Compound

132

561

S-Table.9.2. GOLD fitness score and binding energies of compounds used in the current study.

S. No Ligands GOLD Score

Autodock Vina Binding

Energy (Kcal/mol)

6. Compound 114 69.5355 -9.6

7. Compound 132 68.474 -7.4

8. Compound 116 65.71 -6.5

9. Compound 8 65.3466 -6.3

10. Compound 68 65.2684 -6.9

11. Compound 122 64.6783 -6.8

12. Compound 22 62.9076 -7.6

13. Compound 10 62.8059 -6.9

14. Compound 5 62.7293 -9.3

15. Compound 115 62.5291 -5.9

16. Compound 126 61.7061 -5.3

17. Compound 15 61.6653 -7.3

18. Compound 54 61.6119 -5.2

19. Compound 55 61.2978 -5.9

20. Compound 9 61.2601 -5.1

21. Compound 56 61.2538 -4.5

22. Compound 111 61.2173 -7.7

23. Compound 13 61.0316 -6.4

24. Compound 127 60.5037 -5.5

562

25. Compound 129 59.9456 -5.8

26. Compound 21 59.8513 -5.4

27. Compound 50 59.7818 -5.9

28. Compound 99 59.6408 -5.7

29. Compound 113 59.5794 -5.2

30. Compound 76 59.4734 -5.5

31. Compound 75 58.935 -5.4

32. Compound 62 56.4926 -5.0

33. Compound 128 57.9934 -5.9

34. Compound 34 57.7646 -5.6

35. Compound 123 43.7961 -3.1

36. Compound 42 57.5335 -5.7

37. Compound 25 57.2994 -5.7

38. Compound 52 57.2294 -6.0

39. Compound 61 57.1212 -6.7

40. Compound 51 56.9733 -7.3

41. Compound 14 56.8889 -6.9

42. Compound 1 56.8811 -5.1

43. Compound 100 56.8427 -5.3

44. Compound 120 56.763 -5.1

45. Compound 63 56.7227 -5.7

46. Compound 105 56.5316 -5.5

563

47. Compound 69 56.3841 -5.9

48. Compound 64 56.338 -5.8

49. Compound 104 56.3191 -5.4

50. Compound 24 56.2459 -5.2

51. Compound 102 55.8242 -5.5

52. Compound 26 55.6502 -5.8

53. Compound 74 54.2166 -5.1

54. Compound 59 55.3455 -5.1

55. Compound 101 55.2803 -5.0

56. Compound 103 55.251 -6.6

57. Compound 45 55.1611 -5.7

58. Compound 109 55.1296 -5.2

59. Compound 112 54.5055 -5.1

60. Compound 16 54.1135 -5.3

61. Compound 89 54.07 -5.3

62. Compound 53 53.9429 -5.2

63. Compound 18 53.9288 -5.9

64. Compound 93 53.9031 -5.6

65. Compound 23 53.8108 -5.2

66. Compound 125 53.8164 -5.0

67. Compound 38 53.349 -5.1

68. Compound 73 53.3784 -5.6

564

69. Compound 17 53.3279 -5.7

70. Compound 27 53.2173 -5.0

71. Compound 7 52.7032 -8.4

72. Compound 72 52.5912 -4.7

73. Compound 108 52.5795 -5.1

74. Compound 12 51.9717 -5.6

75. Compound 67 51.4864 -5.7

76. Compound 71 51.4864 -4.9

77. Compound 124 51.4065 -4.8

78. Compound 33 51.3709 -4.2

79. Compound 97 51.2445 -4.8

80. Compound 94 51.244 -4.5

81. Compound 130 51.1426 -5.4

82. Compound 60 50.9463 -4.2

83. Compound 32 50.9341 -4.1

84. Compound 19 50.7152 -4.8

85. Compound 6 50.5601 -4.7

86. Compound 43 50.4629 -4.4

87. Compound 3 49.8621 -4.6

88. Compound 110 49.5265 -4.7

89. Compound 107 49.4318 -4.6

90. Compound 28 49.3134 -4.5

565

91. Compound 106 49.2053 -4.7

92. Compound 121 49.0808 -4.1

93. Compound 46 48.9425 -4.7

94. Compound 57 48.6857 -4.9

95. Compound 4 48.6906 -4.3

96. Compound 96 48.594 -4.7

97. Compound 30 48.3602 -4.9

98. Compound 65 47.7837 -4.6

99. Compound 90 47.7651 -4.6

100. Compound 88 47.5862 -4.5

101. Compound 29 47.3966 -4.6

102. Compound 131 47.3304 -4.7

103. Compound 58 47.1383 -4.4

104. Compound 66 46.9303 -4.3

105. Compound 92 46.6977 -4.2

106. Compound 2 46.5858 -4.2

107. Compound 36 46.5486 -4.1

108. Compound 118 46.4225 -4.1

109. Compound 41 45.9334 -4.5

110. Compound 20 45.3649 -4.6

111. Compound 40 44.2707 -4.3

112. Compound 95 43.9455 -4.2

566

113. Compound 47 43.8178 -4.1

114. Compound 119 43.5164 -4.1

115. Compound 117 43.3578 -4.5

116. Compound 11 42.775 -4.8

117. Compound 44 42.4766 -4.2

118. Compound 31 42.2968 -4.2

119. Compound 35 41.913 -4.1

120. Compound 39 41.646 -4.2

121. Compound 79 38.6907 -4.0

122. Compound 37 38.482 -4.1

123. Compound 78 38.3558 -3.9

124. Compound 70 37.9465 -3.9

125. Compound 81 37.889 -3.5

126. Compound 91 37.2578 -4.1

127. Compound 82 37.1651 -3.4

128. Compound 87 36.0794 -3.5

129. Compound 49 35.4694 -3.5

130. Compound 86 35.3061 -3.5

131. Compound 85 35.0172 -3.4

132. Compound 84 34.6583 -3.3

133. Compound 83 32.5083 -3.2

134. Compound 80 32.3247 -3.3

567

135. Compound 48 32.2099 -3.7

136. Compound 98 30.8428 -3.2

137. Compound 77 29.9341 -3.3

568

Table.9.3. Free energy decomposition into active site residues of the protein

Residue Internal van der Waals Electrostatic Polar Solvation Non-Polar Solv. TOTAL

GLU12 19.74659 -4.082909091 -13.022 -72.43890909 0.342858764 -69.4544

ARG39 27.61627 -6.073863636 -191.5718182 48.90013636 0.532664836 -120.597

ILE40 23.7495 -3.863727273 -11.65436364 -0.220409091 0.021242945 8.032243

LEU41 21.62005 -6.236318182 -6.230727273 -0.645727273 0.032617309 8.53989

THR42 19.45759 -4.817272727 -49.916 -2.817954545 0.005991709 -38.0876

ASP43 16.94409 -6.438818182 21.16368182 -86.41636364 0.152829164 -54.5946

SER44 16.40705 -4.173318182 -1.166136364 -7.016545455 0.113879455 4.164925

ARG45 26.905 -0.725545455 -177.587 33.82095455 1.642520618 -115.944

LEU56 23.51836 -7.230181818 -10.53222727 -0.463954545 0.003114327 5.295114

ARG60 28.86468 -1.232590909 -174.5409091 32.51059091 1.625656582 -112.773

PHE61 21.78923 -5.241272727 -6.781272727 -1.710772727 0.207440182 8.263349

ARG97 29.19195 -4.9955 -206.2568182 64.05777273 0.712923055 -117.29

ARG107 29.81918 -10.40254545 -187.4490909 52.87195455 0.062290473 -115.098

ARG146 26.58891 -3.127772727 -168.1810909 25.52077273 0.771503564 -118.428

MET157 19.76214 -9.476363636 -15.03927273 0.648681818 0.040118073 -4.0647

LEU160 24.25145 -6.836727273 -16.74663636 1.491227273 0.004464327 2.163783

GLU161 20.94441 -5.365681818 -41.95654545 -52.77818182 0.152786618 -79.0032

569

Chapter # 10

S-Fig.10.1. Mean RMSD estimated for the enzyme (red line) and for the bounded enzyme (blue line).

Sajjad ahmad bioinformatics 2020 qau isb prr.pdf

Documents

Transcript of Sajjad ahmad bioinformatics 2020 qau isb prr.pdf