Generation of Affinity Reagents to Challenging - UIC Indigo

Generation of Affinity Reagents to Challenging

Targets Through Phage Display

BY

JENNIFER ELISE MCGINNIS B.S., University of Illinois at Urbana-Champaign, 2014

Thesis

Submitted as partial fulfillment of the requirements for the degree of Doctor of Philosophy in Biological Sciences

in the Graduate College of the University of Illinois at Chicago, 2019

Chicago, IL

Defense Committee:

Brian Kay, Advisor David Stone, Chair Teresa Orenic Yury Polikanov Andrei Karginov, Pharmacology

For my mother, Marla, who has been my mentor when I have needed guidance, my

cheerleader when I have needed encouragement, and my best friend when I have needed

companionship.

ii

ACKNOWLEDGEMENTS

I would first like to thank my advisor, Dr. Brian Kay. Over the past five years, Brian

has consistently made my education his priority. He frustratingly never gave me

immediate answers to problems in the lab (even when he already knew the answer), and

instead sat with me, and talked at length until I had expanded my mind and thought of

new, creative solutions. Earning a Ph.D. is not just about becoming extremely

knowledgeable in a very specific research area. It is also about becoming a scientist: a

person capable of independent and creative thinking and using this to develop solutions

and solve problems. Because of Brian’s expertise and mentorship, I am now a budding

scientist. I only hope one day I can be as creative, intelligent, and ambitious as he is. A

single paragraph certainly does not describe how impactful Brian has been to me. I am

so honored to have been invited into his lab, so humbled to have learned so much from

him, and so thankful to have gained a lifelong friend and mentor.

I would like to thank my committee members for the guidance they have provided

throughout my studies at UIC. Dr. Stone’s thoughtful questions and high expectations

have challenged me to think critically about my experiments. Dr. Orenic has always been

there to encourage me and celebrate my successes. It has made getting through

graduate school struggles so much more manageable. Dr. Polikanov’s and Dr. Karginov’s

expertise in protein structures and protein engineering have been instrumental in guiding

the direction of my thesis and helping me interpret my experimental results.

I would not have survived graduate school without all the members of the Kay lab,

both past and present. They have provided forever treasured guidance, friendship, and

iii

support. Specifically, I would like to acknowledge Dr. Kevin Gorman and Dr. Leon

Venegas who took time out of their busy days to train me and always answer my

questions. Sehar Khosla began this journey with me. She was there to celebrate my

successes and encourage me after my failures. Christina Miller has brought so much joy

into the lab this past year. Between all our hard work, our inside jokes and funny moments

have lessened my stress while preparing to graduate. I do not doubt that we will be friends

for life.

Finally, I want to thank my close friends and family. Stephanie Czarnik, Christina

Norman, and Grace Aldrich are always there when I need support, love, or just a good

laugh. My stepfather and stepmother have spoiled and treated me as if I was one of their

own. Most importantly, my mother, father, and sister have made sure I am surrounded by

love, happiness, and encouragement every single day of my life. From piano recitals and

gymnastics meets, to presentations and graduations, they have been there through it all,

both big and small, cheering me on. Thank you!

iv

CONTRIBUTION OF AUTHORS

Chapter 1 is a literature review of the affinity reagent generation and protein engineering

fields. Portions of section 1.5 Phage Display were previously published in:

Gorman, K., McGinnis, J. & Kay, B. (2018) Generating FN3-Based Affinity

Reagents Through Phage Display, Current Protocols in Chemical Biology. 10,

e39.

Dr. Kevin Gorman provided figures 3 and 4 and contributed to writing the protocol section

of the manuscript. The portion of the work duplicated in this thesis can be found in the

commentary section of the manuscript and was written by me and edited by my research

mentor, Dr. Brian Kay.

Chapter 2 represents the published manuscript:

McGinnis, J. E. & Kay, B. K. (2018) Generation of recombinant affinity reagents

against a two-phosphosite epitope of ATF2, New biotechnology. 45, 45-50.

for which I is was the primary author and main researcher. Dr. Kay contributed to

experimental design and editing of the manuscript.

Chapter 3 represents the published manuscript:

McGinnis, J. E., Venegas, L. A., Lopez, H. & Kay, B. K. (2018) A Recombinant Affinity

Reagent Specific for a Phosphoepitope of Akt1, Int J Mol Sci. 19, 3305.

for which I was the primary author and main researcher. Dr. Venegas contributed figures

2, 3, and 5 and wrote the methods section of the manuscript. Dr. Kay contributed to

experimental design and editing of the manuscript.

Chapter 4 represents my unpublished experimental results that contribute to improving

v

diagnostic and biochemical sandwich assays. With the addition of a few experiments, I

presume this work will soon be published as a co-authored manuscript.

Chapter 5 summarizes the work presented in this dissertation, potential future

experiments, and the overall impact of this work.

Dr. Brian Kay has been instrumental in editing each chapter and providing financial

(through federal grants) and intellectual support for this dissertation.

vi

TABLE OF CONTENTS

CHAPTER PAGE

1. INTRODUCTION .............................................................................. 1

1.2 Applications of Antibodies .......................................................... 1

1.3 Targets of Affinity Reagents ...................................................... 3

1.4 Methods for Generating Affinity Reagents ................................. 5

1.5 Phage Display ........................................................................... 9

1.6 Displayed Proteins ................................................................... 13

1.7 Thesis Goals ............................................................................ 19

1.8 Literature Cited ........................................................................ 22

2. GENERATION OF RECOMBINANT AFFINITY REAGENTS AGAINST A TWO-PHOSPHOSITE EPITOPE OF ATF2 ................ 35

2.1 Abstract ................................................................................... 35

2.2 Introduction .............................................................................. 36

2.3 Materials and Methods ............................................................ 38

2.3.1 Reagents ........................................................................ 38

2.3.2 Cloning and Bacterial Expression ................................... 39

2.3.3 Enzyme-Linked Immunosorbent Assay (ELISA) ............. 41

2.3.4 Affinity Selection of the Primary Library .......................... 41

2.3.5 Secondary Library Construction and Affinity Selection ... 43

2.3.6 Western Blot ................................................................... 43

2.4 Results and discussion ............................................................ 44

2.4.1 The FHA Domain Scaffold Can Recognize a Dual-Phosphorylated Epitope ................................................. 44

2.4.2 Characterization of the Interaction Between an FHA

Variant Selective for the Dual-Phosphorylated ATF2

Peptide ........................................................................... 44

2.4.3 Generation of FHA Affinity Reagents to Specifically

Recognize the Mono-Phosphorylated Forms of the

ATF2 Peptide ................................................................. 50

2.4.4 Using an FHA Reagent to Detect Phospho-ATF2

in a Western Blot ............................................................ 53

2.5 Conclusions ............................................................................. 55


vii

TABLE OF CONTENTS (continued)

CHAPTER PAGE

3. A RECOMBINANT AFFINITY REAGENT SPECIFIC FOR A PHOSPHOEPITOPE OF AKT1 ...................................................... 60

3.1 Abstract ................................................................................... 60

3.2 Introduction .............................................................................. 60


3.3.1 Peptides ......................................................................... 62

3.3.2 Cloning and Bacterial Expression of Proteins................. 63

3.3.3 Affinity Selections ........................................................... 63

3.3.4 Enzyme-Linked Immunosorbent Assay (ELISA) ............. 65

3.3.5 Surface Plasmon Resonance ......................................... 65

3.4 Results and Discussion ........................................................... 65

3.4.1 Directed Evolution of the FHA1 Domain Yielded ... Variants that Recognize an Akt1 Phosphopeptide ....................... 65

3.4.2 An Isolated FHA Clone binds the Akt1 Peptide with Unique Specificity ....................................................................... 70

3.5 Conclusions ............................................................................. 74


4. STREAMLING THE VALIDATION PROCESS FOR BINDING REAGENTS ISOLATED BY MEGASTAR ...................................... 79

4.1 Abstract ................................................................................... 79

4.2 Introduction .............................................................................. 80


4.3.1 Cloning and Overexpression of FN3 Fusion Proteins ..... 84

4.3.2 Isolation of binding pairs by MegaSTAR ........................ 85

4.3.3 Sandwich ELISA with FN3 fusion binding pairs .............. 85

4.3.4 Monobody labeling via sortase reaction ......................... 86

4.3.5 Sandwich ELISA with monobodies labeled ....... via sortase reaction .......................................................................... 86

4.4 Results and Discussion ........................................................... 87

4.4.1 MegaSTAR is robust and reproducible ........................... 87

4.4.2 Improving binding pairs by assay development .............. 87

4.4.3 Easy Use of Pairs for Heterogenous Assays .................. 93

4.5 Conclusions ............................................................................. 97


viii

TABLE OF CONTENTS (continued)

CHAPTER PAGE

5. CONCLUSIONS ........................................................................... 102

5.1 Thesis Summary .................................................................... 102

5.2 Future Directions ................................................................... 104

5.2.1 Incorporate Multiple Scaffolds and Linker ....... Lengths into MegaSTAR ................................................................... 104

5.2.2 Generating Reagents to Cell-Free Expressed ........ Targets ..................................................................................... 106

5.2.3 Use of Sortase-Mediated Ligation to Format ......... Pairs for Homogeneous Assays ...... Error! Bookmark not defined.

5.3 Overall Impact ....................................................................... 110

5.7 Literature cited ....................................................................... 112

Chapter 6 APPENDIX ................................................................................... 114

Chapter 7 VITA ............................................................................................. 117

ix

LIST OF TABLES

TABLE PAGE

I. IN VITRO DISPLAY TECHNOLOGIES ........................................................... 8

II. ALTERNATIVE BINDING SCAFFOLD PROTEINS FOR ...... PHAGE DISPLAY ......................................................................................................................15

III. LIST OF PRIMERS AND THEIR SEQUENCES USED IN THIS STUDY .......40

IV. ATF2 PHOSPHOPEPTIDES ........................................................................45

V. OUTPUT SEQUENCES OF CLONES ISOLATED FROM SELECTIONS ......52

VI. AFFINITY MEASUREMENTS ........................................................................69

VII. SANDWICH ELISA METRICs WITH MONOBODY FUSION .................. PAIRS ......................................................................................................................92

x

LIST OF FIGURES

FIGURE PAGE

1.1. The M13 phage particle ............................................................................. 9

1.2. The M13 phage life cycle ......................................................................... 11

1.3. Generation of affinity reagents through phage display. ............................ 12

1.4. Antibody and antibody variant structures. ................................................ 14

1.5. FHA1 interacts with phosphothreonine containing peptides through its β loops. .................................................................................. 18

1.6. PyMOL representations of the FN3 monobody ........................................ 19

2.1. FHA variants can recognize the dual-phosphorylated ATF2 ............ epitope ................................................................................................................. 46

2.2. The FHA library contains variants that recognize a pT-X-p(S/T) motifs ................................................................................... 46

2.3. Alanine-scanning of β4-β5 and β10-β11 loop residues from FHAαATF2 F6 ................................................................................. 48

2.4. Determining FHA variant F6′s affinity for its target ................................... 50

2.5. Generation of reagents that specifically recognized the mono-phosphorylated targets ............................................................................ 51

2.6. Detection of full-length ATF2 in a western blot ........................................ 54

3.1. Primary structure of the Akt1 protein ....................................................... 66

3.2. Affinity selection process and ELISA of 12 output clones ........................ 66

3.3. Amino acid sequence analysis of two loops randomized in the phage-displayed scaffold ............................................................... 68

3.4. Comparing the relative affinity of four clones ........................................... 69

3.5. Binding of an FHA variant to a set of related peptides ............................. 70

xi

LIST OF FIGURES (continued)

FIGURE PAGE

3.6. Identification of important residues on the peptide by alanine scanning. ................................................................................................................. 72

3.7. Binding of an FHA variant to corresponding peptides from Akt2 and Akt3 ................................................................................................................. 73

4.1. Sandwich ELISA ...................................................................................... 81

4.2. Megaprimer shuffling for tandem affinity reagents ................................... 83

4.3. Loop sequences of tandem clones isolated against COPS5 ................... 88

4.4. Monobody fusion proteins for improved sandwich assays ....................... 90

4.5. Sandwich ELISA with NanoLuc readout .................................................. 91

4.6. Sandwich ELISA with alkaline phosphatase (AP_ readout ...................... 92

4.7. Site-specific monobody labeling via sortase-mediated ligation ............... 94

4.8. Sortase-mediated ligation efficiency ........................................................ 95

4.9. Sandwich ELISA with monobody binding pairs labeled via ............ sortase-mediated ligation ..................................................................................... 96

4.10. Workflow for identifying and characterizing binding reagents through MegaSTAR .............................................................................................. 97

5.1. An array of tandem vectors for improved binding pairs .......................... 105

5.2. Target immobilization via HaloTag technology ...................................... 106

5.3. Homogeneous assays utilizing sortase-mediated ligation...................... 108

5.4. Generating recombinant enzyme labels for sortase-mediated ligation .. 109

xii

LIST OF ABBREVIATIONS

A Alanine

Ab Antibody

ABTS 2,2-azinobis(3-ethylbenzthiazoline-6–sulfonic acid)

Akt1 Protein kinase B

ATF2 Activating Transcription Factor 2

CB Carbenicillin

CDR Complementarity-determining region

COPS5 Cop9-signalosome subunit 5

DARPin Designed ankyrin repeat protein

DNA deoxyribonucleic acid

dsDNA double-stranded DNA

E. coli Escherichia coli

ELISA Enzyme linked immunosorbent assay

epPCR error-prone Polymerase Chain Reaction

ERK Extracellular-regulated kinase

Fab Fragment antigen-binding

FHA1 Forkhead-associated I domain

FN3 Fibronectin type III domain

G Glycine

HRP Horseradish peroxidase

IgG Immunoglobulin G

xiii

KN Kanamycin

M13KO7 Helper phage

mAb Monoclonal antibody

MegaSTAR Megaprimer Shuffling for Tandem Affinity Reagents

MEK Mitogen-activated protein kinase-kinase

Tm Melting temperature

p38 p38 mitogen-activated protein kinases

pAb Polyclonal antibody

PBS Phosphate buffered saline

PBST Phosphate buffered saline with 0.1% Tween

PCR Polymerase Chain Reaction

PDB Protein Data Bank

PEG Polyethylene glycol

PNK Polynucleotide kinase

pS Phosphoserine

pT Phosphothreonine

PTM Post-translational modification

pY Phosphotyrosine

Raf Raf protein kinase

RalGDS Ral guanine nucleotide dissociation stimulator

S Serine

scFv Single-chain variable fragment

SDS-PAGE Sodium dodecyl sulfate-polyacrylamide gel electrophoresis

xiv

SPR Surface Plasmon Resonance

ssDNA single-stranded DNA

SUMO Small ubiquitin-like modifier

T Threonine

TM13KO7 Trypsin cleavable helper phage

WT Wild-type

Y Tyrosine

xv

Summary

Antibodies are essential tools in both the research and medical communities, which

can be utilized in a variety of biochemical experiments, diagnostic assays, and therapies.

However, they sometimes are cross-reactive in specificity, have a modest affinity for their

target, and are time consuming and expensive to produce. In recent years, alternative

methods for generating recombinant affinity reagents have overcome many of the

limitations of polyclonal and monoclonal antibodies. During my Ph.D. research, I applied

the phage display technology to generate high-quality affinity reagents to a variety of

phosphopeptides and folded proteins and demonstrated their utility in biochemical

assays. Furthermore, I have designed and successfully implemented several

experimental strategies to reduce the cost needed to generate and validate these

reagents.

Antibodies have been extremely useful probes to monitor post-translational

modifications, such as protein phosphorylation in cells. Protein phosphorylation provides

a method of activating or inhibiting proteins, altering their cellular location, modulating

their stability, or facilitating protein-protein interactions. Unfortunately, many of the

antibodies generated against peptides carrying phosphoserine or phosphothreonine are

cross-reactive, which limits their usefulness.

To overcome the limitations of phosphospecific antibodies, I screened a library of

phage that displayed variants of the engineered Forkhead Associated (FHA) domain to

generate recombinant affinity reagents that bind selectively to several different

phosphopeptides of human proteins. First, I generated FHA reagents to phosphopeptides

xvi

SUMMARY (continued)

of Activating Transcription Factor 2 (ATF2), a transcription factor involved proliferation,

apoptosis, and DNA repair. Because ATF2 contains two neighboring phosphothreonines,

my goal was to isolate reagents that distinguish between its various phosphostates. I was

successful in developing through directed evolution one affinity reagent that specifically

recognized the ATF2 peptide when di-phosphorylated and another that only recognized

one of the two mono-phosphorylated states. On the contrary, a commercial antibody to

the di-phosphopeptide was cross reactive between all three phosphopeptides, illustrating

that phage display can be a better source for generating high quality, phosphospecific

reagents. Additionally, I showed that an engineered FHA domain could detect native,

phosphorylated ATF2 protein in a western blot.

I also generated FHA affinity reagents to a phosphopeptide from Akt1, a protein that

is observed to be phosphorylated in many cancer patients. Not only were these reagents

uniquely specific for the peptide - capable of distinguishing between Akt1 and its highly

conserved isoforms - but they also had high affinity for the Akt1 phosphopeptide,

illustrating their potential success in a variety of different assays.

In the last portion of my Ph.D. research, I focused on developing affinity reagents that

work as binding pairs in sandwich assays. The sandwich assay requires two separate

affinity reagents to bind an analyte simultaneously, which substantially lowers false

positive results. Using a phage library displaying engineered fibronectin type III (FN3)

monobodies, I isolated 6 different pairs with a new technique in the laboratory,

Megaprimer Shuffling for Tandem Reagents (MegaSTAR). I streamlined the validation of

xvii

SUMMARY (continued)

pairs isolated by MegaSTAR by eliminating the need to purify large quantities of each

binding reagent separately, which reduces time and cost, and allows testing of more pairs

compared to past methods. Additionally, I labeled binding reagents in vitro with various

tags through site-specific, enzymatic ligation, making these reagents suitable for

heterogenous and homogenous sandwich assays.

While assays with pairs of antibodies are very common in basic research and

diagnostics, they are difficult to develop, and they depend on a steady supply of each

antibody. The phage display technology provides an alternative method to generate high

quality binding reagents quickly and inexpensively. My work advances this technology in

generating binding reagents against challenging targets and improving their utility in

assays.

xviii

1

Chapter 1 1. INTRODUCTION

Portions of this chapter have been published in Current Protocols in Chemical Biology

[1].

This work was done in collaboration with Dr. Kevin Gorman and Dr. Brian Kay. Dr. Kevin

Gorman provided figures and helped write the protocol section of the article. Dr. Kay

helped with editing.

1.1 Antibodies and Antigens

Antibodies are plasma proteins that are naturally produced by the immune system in

response to foreign agents that have entered the body. They circulate throughout the

blood and bind to these foreign agents, termed “antigens”, and render them inactive

through neutralization, aggregation, or internalization by cells. Antibodies protect the body

from harmful pathogens, viruses, and infections [2, 3]. The immune system has the ability

to generate billions of antibodies that vary in amino acid sequence at their antigen-binding

sites, which allows them to recognize a multitude of antigens [4].

1.2 Applications of Antibodies

Probes that recognize biological targets are invaluable tools for biomedical and basic

laboratory research and have assisted in analysis of the structure and function of the cell

[5]. The most commonly used probes of proteins are antibodies due to their natural

binding properties. Antibodies are used in a variety of biochemical assays such as

western blots [6], enzyme linked immunosorbent assays (ELISA) [7], and pull-downs [8]

to detect or quantify particular proteins in complex mixtures, such as blood or cell lysates.

Additionally, scientists have employed antibodies in immunohistochemistry [9] and

immunofluorescence [10] experiments to detect particular proteins in fixed cells or tissue

samples. An antibody’s ability to selectively recognize and bind these proteins provides

2

information regarding the protein’s location within a cell or tissue and insight into the

protein’s possible function.

Antibodies also serve a crucial role in diagnostics. While doctors rely on a patient’s

symptoms and medical history to assess their illness, without confirmatory tests, the

patient could be misdiagnosed. Many diseases are caused by similar environmental and

genetic factors and/or produce similar symptoms, but require different treatments [11].

Antibodies help in such a situation because they can monitor biomarkers in a patient’s

blood, serum, urine, stool, or tissue and indicate a pathological biological process is taking

place in the body as a result of a specific disease [12]. Biomarkers can range from the

presence, absence, or change in levels of specific antigens in the body, a change in a

normal enzymatic activity, neoepitopes generated by genetic mutations, post-

translational modifications, or a change in a protein’s conformation [13, 14]. There are

currently many tests where biomarkers are detected with antibodies. These include tests

for pregnancy [15, 16], cancer [17-19], myocardial infarction [20], and bacterial and viral

infections [21, 22]. Antibodies have been instrumental in detecting and monitoring the

progression of a disease, as well as monitoring a patient’s response to treatment [11].

In recent years, antibodies have also served as therapeutics to treat various diseases,

such as inflammatory, respiratory, and cardiovascular diseases, cancer, and viral and

bacterial infections [23]. Humira (adalimumab), an anti-tumor necrosis factor alpha

(TNFα) antibody, is one of the most popular pharmaceuticals, with global sales of $18

billion in 2017. Humira has been shown to treat a variety of autoimmune diseases,

including rheumatoid arthritis [24], plaque psoriasis [25], and Crohn’s disease [26].

Antibody therapies have become a leading product in the biopharmaceutical industry,

3

with > 30 antibody therapies in the market, contributing to > $100 billion in global sales in

2017. With hundreds more in development and clinical trials, it is likely that global sales

of therapeutic antibodies will grow substantially in the future [27, 28].

Antibody therapeutics rely on several different mechanisms to combat diseases and

infections. The natural functions of antibody-mediated immune responses have been

exploited by many of these reagents. One such way is through their neutralization

capabilities; in this case, antibodies bind a hormone or receptor, blocking a cell signaling

pathway that promotes proliferation, growth, stress responses, etc. [23, 29-31]. Other

reagents utilize an antibody-dependent cell-mediated cytotoxic (ADCC) response in

which the antibody binds the antigen on the cell surface and also recruits immune effector

cells to lyse and destroy the target cell [32, 33]. Of course, researchers have not limited

themselves to these natural methods of action and have further engineered antibodies to

drive certain responses. For example, reagents conjugated to a toxin or other drug guide

their selective and efficient degradation of the antigen without harming the rest of the

body [34]. Other antibodies have been genetically engineered to bind two targets

simultaneously, such as antigens on tumor and immune cells, to drive the immune

response by forcing the antigen and immune cell into proximity [35]. Many of these

approaches have been clinically effective and will likely be exploited more in the future.

1.3 Targets of Affinity Reagents

Affinity reagents that detect and track protein post-translational modifications (PTMs)

are extremely useful in research. PTMs such as phosphorylation, methylation,

acetylation, glycosylation, ubiquitination, lipidation, and nitrosylation influence almost all

aspects of cell biology [36-41]. For example, affinity reagents that monitor protein

4

phosphorylation are of great interest because phosphorylation plays a crucial role in

regulating, activating, or inhibiting a protein, altering a protein’s cellular location, signaling

a protein for degradation, or creating or destroying specific binding interactions [39, 40].

Over 100,000 phosphosites have been mapped on > 13,000 human proteins [42, 43],

illustrating their significance in cell function, growth, and survival.

Most proteins have multiple phosphorylation sites, and the effects of these

modifications cannot be fully understood by examining only individual phosphorylation

sites for a protein [44]. Many signaling pathways depend on multi-site phosphorylation

events to maintain proper temporal regulation of cells as well as regulate a complex

system of signaling cascades. Thus, affinity reagents that selectively recognize

phosphoproteins in their various phospho-states are needed to study the triggers, timing,

and downstream effects of these events.

Reagents with this level of specificity are challenging to find, especially for multiple

phosphorylation sites in proximity to one another in a protein. Cells contain a “priming”

protein kinase that phosphorylates one residue in a protein, which leads to

phosphorylation of a nearby residue by a second protein kinase. A classic example

highlighting this mechanism involves phosphorylation of target proteins by Glycogen

Synthase Kinase 3 (GSK-3). A priming kinase phosphorylates one residue of a GSK-3

substrate, whereupon GSK-3 binds the priming phosphate group (through a positively

charged pocket adjacent to the kinase’s active site) and then phosphorylates a serine or

threonine four residues N-terminal of the priming phosphate [45]. Depending on the

priming site, GSK-3 has different affinities for its targets, leading to differences in the

efficiency and timing of how quickly a target is phosphorylated [46]. While mass

5

spectrometry is a useful tool in determining the timing of each of these phosphorylation

events, it does not provide information about the location of these events in cells and

tissue. Alternatively, affinity reagents to particular PTM epitopes provide both spatial and

temporal information regarding the phosphorylation of particular residues in proteins.

There is a considerable need for affinity reagents that recognize specific

conformational epitopes as well. Changes in a protein’s conformation can result from a

variety of factors along with genetic changes including temperature, pH, post-translational

modifications, binding of a ligand (i.e., allostery), solvent polarity, and ion concentration

[47-49]. Thus, reagents that recognize a conformational epitope are useful for monitoring

the effects of these factors on a protein’s structure or determining if a protein is denatured

or misfolded. Additionally, many diseases, such as neurodegenerative disorders and

cancers, are caused by protein misfolding, thereby rendering these proteins toxic [50, 51].

For example, in Huntington’s disease, the huntingtin protein is misfolded and interacts

with other correctly folded copies of itself to catalyze their transition to the toxic

configuration [52]. Other neurodegenerative disorders including Alzheimer’s disease,

dementia, and amyotrophic lateral sclerosis (ALS) progress through similar paths. Affinity

reagents that can distinguish between these native and misfolded proteins have been

instrumental tools in studying samples from deceased patients [53]. Additionally, affinity

reagents that recognize epitopes only found in misfolded proteins have potential

therapeutic value for sequestering the misfolded, but not the native, forms of proteins.

1.4 Methods for Generating Affinity Reagents

Traditionally, antibody generation requires immunizing an animal with an antigen,

which induces the animal’s immune system to produce immunoglobulin subtype G (IgG)

6

antibodies specific for that antigen. After waiting several weeks for the response to occur,

the animal’s serum can be harvested, and the desired antibodies purified by affinity

chromatography (via the immobilized antigen) [54-56]. This pool of antibodies, described

as polyclonal, contains a population of IgGs that can be used in various biochemical

assays. While polyclonal antibodies have been critical tools for research, they have their

limitations. Polyclonal antibodies are not renewable and are only regenerated by

repeating the immunization process. Not only is this time-consuming and expensive, but,

but the quality, specificity, and affinity of polyclonal antibodies vary from batch to batch,

sometimes leading to experimental irreproducibility [57].

Rather than harvesting the animal’s serum, scientists discovered they could isolate

antibody-producing B-cells from the spleen and fuse them to immortal, myeloma cells to

create “hybridomas” [58]. The creation of these hybrid cells eliminates some of the major

issues with polyclonal antibodies. First, because these cells are immortal, they secrete

antibodies as they grow and divide in culture flasks, eliminating the constant need for

animal immunization. Additionally, as each B-cell only secretes a single antibody, the

antibodies produced by the hybridoma will also only have a single amino acid sequence,

also known as a monoclonal antibody.

Monoclonal antibodies still have their limitations, though. Ideally, they are perpetually

secreted from hybridomas and do not vary from batch to batch as they have a single

identity. Unfortunately, this is not usually the case. After many passages, genetic drift and

mutations can occur in the monoclonal antibody or cells stop secreting the antibody [59].

Furthermore, a recent study analyzed over 185 hybridoma cell lines and discovered that

over 30% secreted two or more antibodies. While it is unclear why this happens - perhaps

7

due to multiple B-cells fusing to generate heterokaryons or innate characteristics of the

myeloma cells – a mixture of monoclonal antibodies leads to off-target binding and

decreases the binding signals for the desired target [60].

To circumvent the problems that arise from monoclonal antibody generation, in vitro

methods for antibody generation has been developed. In vitro display methods have

taken advantage of the polymerase chain reaction (PCR) to amplify and clone millions of

antibody genes (See section 1.5 for types of recombinant antibody libraries) and “display”

them recombinantly. [61-64]. These recombinant antibody libraries can then be screened

against targets of interest without animal immunization [65, 66]. Unlike, the traditional

methods of antibody generation, recombinant antibodies provide complete control over

screening conditions including antigen conformation, buffer environment, and the addition

of competitors to drive binding towards a specific epitope. Additionally, because these

libraries are recombinantly expressed, alternative, non-antibody proteins, or scaffolds,

with unique characteristics can be used in place of antibody libraries (See section 1.5 for

types of scaffolds and their advantages) [67, 68]. This increases the probability of isolating

reagents with high specificity and affinity for antigens.

Most significantly, display technologies, notably phage display [69, 70], yeast display

[71-73], mRNA display [74], and ribosome display [75] link each antibody or binding

protein variant to its corresponding DNA or RNA sequence through several genetic fusion

methods (Table 1). By directly attaching each variant’s genotype to its phenotype, the

primary structure of selected binders can be quickly identified by DNA sequencing.

Because of this advantage, the reagents are renewable and inexpensive to produce by

recombinant expression in bacteria or Chinese Hamster Ovary (CHO) cells. Furthermore,

8

the display reagents can be engineered by directed evolution to increase their specificity

and affinity [76-78]

.Table I. IN VITRO DISPLAY TECHNOLOGIES

TABLE I. IN VITRO DISPLAY TECHNOLOGIES

Name Display Methodology Recovery Average library

size

mRNA display

Recombinant protein with puromycin-mRNA fusion incorporated into C-terminal end

mRNA from selected binders is converted to cDNA and expressed as soluble protein for characterization

1013

Phage display

Recombinant protein genetically fused to bacteriophage coat protein that displays on surface of phage. Phage houses genome containing recombinant protein gene

Recovered virions infect E. coli cells and utilize their quick growth rate to propagate more bacteriophage

1010

Ribosome Display

Recombinant protein attached to ribosome-mRNA complex due to addition of a genetic spacer that prevents release factors from binding and triggering disassembly during translation

mRNA from selected binders is converted to cDNA and expressed as soluble protein for characterization

1013

Yeast Display

Recombinant protein genetically fused to yeast surface protein, and thus, displayed on the surface of yeast cells that house genome containing recombinant protein gene

Cell sorting 107

9

1.5 Phage Display

Since its inception in 1985 [65], researchers have spent a great amount of effort to

improve and expand upon the phage display technique. In fact, in 2018 George Smith

won the Nobel Prize in Chemistry for his invention of phage display. It is now a mature

tool that is used to generate binding partners (i.e., peptides, antibody fragments, scaffold

proteins, protein fragments) for a wide variety of targets. Bacteriophage M13 is the most

popular vehicle for display because its structure is well understood [79]. The virion houses

a circular, single-stranded DNA that is covered with five coat proteins: 2700 copies of

pVIII and five copies each of pIII, pVI, pVII, and pIX (Figure 1.1) [80, 81].

Figure 1.1. The M13 phage particle. Cartoon of an M13 phage particle with its capsid

proteins as well as its single stranded genome. Copy numbers are listed per virion.

10

The life cycle of the M13 bacteriophage begins with infection by protein III (pIII) of a

virion binding the F pilus of a male E. coli cell [80]. This interaction leads to a pore forming

on the bacterial cell surface that allows the virion’s DNA to enter the cell, where it is

converted into double-stranded DNA, and synthesis of the M13 proteins begins [82]. Coat

proteins, are co-translationally inserted into the periplasm, while new, single stranded

phage DNA is synthesized and coated by pV, a single-stranded DNA-binding protein [80,

82]. The protein-coated, single-stranded DNA then interacts with export machinery,

including other capsid proteins, to assemble into virions that are secreted from the cell

(Figure 1.2) [83-85].

In phage display, the coding region for a peptide, an antibody fragment, or scaffold

protein is fused to the coding region of one of the five capsid proteins by molecular

cloning, and when the chimeric gene is transcribed and translated, it is assembled into

the virion and displayed on the viral particle surface. Each of the five coat proteins has

been successfully used for display, with pIII and pVIII protein fusions being the most

commonly used [65, 86-91]. Each coat protein has its distinct advantages and

disadvantages. In a virion, there are 2700 copies of pVIII. This is useful for displaying

peptides that have weak affinity for their targets because they will be present at a high

copy number on each virion which can create an avidity effect (increased apparent

affinity). However, only small peptides can be displayed on all 2700 copies of pVIII without

steric hindrance. Conversely, in a virion there are five copies of pIII, and it tolerates display

of large peptides and proteins. This also helps in the discrimination of tight binders

because the avidity effect is greatly reduced [70].

11

Figure 1.2. The M13 phage life cycle. The M13 virion interacts with the F-pilus on the

bacterial cell membrane, allowing the virion to insert its single-stranded DNA (ssDNA)

into the bacterial cell. The M13 genome is converted to dsDNA by host cell machinery

and transcription/translation of phage proteins is initiated. Additionally, single stranded

copies of the M13 genome are synthesized through rolling circle amplification. Once a

sufficient amount of the capsid protein, pV, has been synthesized, it coats the M13

ssDNA, preventing its conversion to dsDNA. The pV-DNA complex is recognized by

phage proteins, which have already accumulated in the inner cell membrane, and it is

exported to the periplasm. Here, pV is removed and capsid proteins assemble around the

DNA genome to secrete a mature virion from the bacterial cell. Not to scale.

12

For affinity selection experiments, a phage library is first mixed with a target of interest

(Figure 1.3). Phage variants that do not bind or bind weakly are washed away, while

variants that are bound tightly are eluted and amplified for subsequent rounds of

selection. After two or three rounds of affinity selection (each increasing in stringency with

additional washes, less target, and/or negative selection steps), individual clones are

tested for binding to the target in an ELISA. Because the gene coding for the binding

element protein is encoded in the phage genome, the identity of “positive hits” is deduced

by DNA sequencing.

Figure 1.3. Generation of affinity reagents through phage display. A phage library is

incubated with immobilized target. Non-binding phage are washed away, the remaining

phage are eluted, amplified, and subjected to further rounds of selection. Individual clones

are examined for binding to the target by ELISA. It typically takes 2-3 weeks to go from

target to isolation of binding phage clones.

13

Over the years, the phage display system has been modified. In the first version of

phage display, the fusion protein or peptide was displayed from every copy of the coat

protein to which it was fused [65]. To circumvent the avidity effect caused by this type of

multivalent display, phagemid systems were developed. Phagemids are plasmids

containing the capsid fusion sequence, bacteria and phage origins of replication, and the

phage packaging signal. However, they lack coding for all other phage proteins needed

for assembly and release. Upon infection of bacterial cells with helper phage, which

supplies in trans the other viral proteins, replication and virion assembly begins [92]. The

“3+3 display system" allows for a small number of copies of the chimeric coat protein to

be displayed along with the wild-type capsid protein [93].

The phagemid display technique has been applied to a variety of directed evolution

experiments. Not only have they been utilized for affinity reagent generation, but also to

improve protein stability [94], generate inhibitors [95], provide insight into potential protein

binding interactions [96], and map transcription factor–DNA interactions [97]. As the

technology continues to develop, the opportunities involving phage display will as well.

1.6 Displayed Proteins

Antibodies are large molecules (150 kDa) that consist of two heavy and two light chains

connected by disulfide bonds. Each polypeptide chain contains constant (CH and CL)

regions, needed for cell surface receptor binding, and variable (VH and VL) regions,

responsible for binding antigens [98]. Displaying a molecule of this size on the phage

surface is challenging, and so researchers have eliminated the antibody constant regions

while retaining its variable regions, which specify antigen binding (Figure 1.4) [99]. Two

types of antibody fragments are commonly displayed on virions. The 50 kDa antigen-

14

binding fragment (Fab) only contains the heavy and light chain variable regions and a

portion of the heavy and light chain constant regions. Single-domain variable fragments

(scFv) are even smaller (i.e., 25 kDa) and consist of only the light and heavy chain

variable regions, connected by a fifteen amino acid linker [98].

Figure 1.4. Antibody and antibody fragment structures. Cartoon of a whole antibody

(IgG) (Left), Fab (Middle), and scFv (Right). Variable regions (V) are shown in green and

constant regions (C) are shown in blue. Heavy chains (H) are represented as dark green

and dark blue, and light chains (L) are shown as light green and light blue. Each number

represents a single domain. Single black lines represent amino acid linkers and red lines

indicate interdomain disulfide bonds.

15

In 1989, it was discovered that a scFv could be displayed on the surface of virions

[100, 101]. This proof-of-concept demonstration led to the generation of phage displaying

libraries of naïve scFvs to discover scFvs that bound to many different targets through

affinity selection [102]. Since then, phage libraries have incorporated a number a different

antibody formats including divalent scFvs and Fabs as display proteins [66]. Scientists

have made great strides in therapeutics, diagnostics, and research assays because of

these advances. Nevertheless, these antibody formats have several disadvantages. They

are difficult to express in bacteria due to inter- and intra-domain disulfide bonds, thus,

requiring labor intensive and expensive expression in mammalian CHO cells. Additionally,

these molecules have low thermal stability (Tm = 50-70°C), are prone to aggregation, and

are often cross-reactive [103-105].

Recombinant affinity reagents generated using alternative binding proteins, or

scaffolds, have eliminated many of the disadvantages associated with antibodies.

Typically, scaffolds are natural proteins that have inherent binding properties. Desirable

scaffolds are also small, easy to express in bacteria, thermally stable, highly soluble, and

lacking disulfide bonds [66, 103]. Their binding properties are exploited to generate a

library of variants in which amino acids that participate in the natural binding interaction

but do not participate in maintaining the molecules structural integrity are randomized.

There are numbers of different scaffold proteins that have successfully displayed as

libraries on the phage surface and produced binders to targets of interest (Table II) in

order to bind a variety of different targets with femtomolar to micromolar affinities [1, 94,

106-111].

II. ALTERN ATIVE BINDIN G AFFOLD PROTEIN S FOR PH AGE D ISPLAY

16

This work will focus specifically on generating FHA and FN3 scaffold binding reagents.

The Forkhead Associated (FHA) 1 domain, a naturally occurring phosphothreonine

binding domain, has been used as a scaffold in phage display to select for variants that

recognize phosphothreonine peptide targets. The domain is made of two β-sheets which

fold into a twisted β-sandwich. The imidazole side chain of the conserved Histidine 88

residue interacts with Serine 85, Isoleucine 104, and Glycine 108 to create a pocket to

accommodate the phosphothreonine residue. This pocket allows for discrimination of

phosphothreonine from phosphoserine because it precisely fits the γ-methyl group from

the threonine residue, and Serine 85 and Arginine 70 provide additional contact with the

phosphate group. Furthermore, the FHA domain interacts with other residues on the

phosphopeptide target through two loop regions that connect the β-strands together, the

TABLE II. ALTERNATIVE BINDING SCAFFOLD PROTEINS FOR PHAGE DISPLAY

Scaffold Origin Structure Size (kDa)

Tm (°C) Types of targets

Affibody Z domain of protein A

three alpha helices

6 >90 Proteins, peptides, post-translational modifications

Designed ankyrin repeat protein (DARPin)

Mammalian ankryin proteins

3-7 repeat motifs

14-18 >90 Proteins, peptides, post-translational modifications

Forkhead-associated domain (FHA)

FHA domain from Saccharomyces cerevisiae

β-sandwich 25 70-80 Phosphothreonine peptides

Fibronectin type-III (FN3/ monobody)

Fibronectin domain from

many animal proteins involved in ligand binding

β-sandwich; resembles immunoglobulin domains

10 >90 Conformational epitopes

17

β4-β5 and β10-β11 loops, making them good candidates for randomization (Figure 1.5)

[112-114]. A phage display library was created by Dr. Kritika Pershad in the Kay Lab by

randomizing amino acids 82-84 and 133-139, present in the β4-β5 and β10-β11 loops,

respectively, of the FHA1 domain of the yeast Rad9 protein. The library has successfully

generated FHA variants that recognize an array of mono- and dual-phosphothreonine

peptide targets at an 82% success rate. Additionally, each reagent selectively recognizes

targets containing phosphothreonine and cannot recognize the targets when that

particular residue is changed to phosphoserine, phosphotyrosine, or is unphosphorylated

[115-117]. Thus, the FHA1 library proves to be a valuable tool to generate reagents that

selectively recognize a phosphothreonine post-translational modification.

The human fibronectin 10th type III (FN3) domain has a folding pattern similar to that

of immunoglobulin variable domains (Figure 1.6) [109, 118]. However, the FN3 has

several significant biochemical advantages over antibodies: it lacks disulfide bonds, can

be easily overexpressed in E. coli, is thermally stable (Tm = 88°C), and retains binding

when absorbed onto microtiter plate wells, unlike 95% of monoclonal antibodies, which

lose functionality when adsorbed onto plastic [119, 120]. Additionally, protein engineering

experiments have shown that it is possible to randomize residues within three loops (BC,

DE, FG) on one side of the 10 kDa domain without impacting stability or folding [118,

121]. The BC, DE, and FG loops mimic the complementarity determining regions (CDR)

of the variable domains of the light and heavy chains of antibodies [109, 122]. FN3

variants, also known as “monobodies,” have been generated via phage and yeast display

to a wide variety of targets, such as Abl [123], β-catenin [124], EphA2 [110], estrogen

receptor [125, 126], Fyn [95], integrin [127], Pak1 [128], VEGF-R [129], and several other

18

human cell-signaling proteins [128]. In every case, the FN3 appears to recognize a

conformational, rather than a linear, epitope making it an ideal scaffold for current needs

in research and diagnostics.

Figure 1.5. FHA1 interacts with phosphothreonine containing peptides through its

β loops. FHA1 domain (grey) bound to the Rad9 phosphothreonine peptide (green) (PDB

1G6G). Residues randomized in the β4-5 and β10-11 loops of the FHA phage library are

shown in purple. (Left) Cartoon of the FHA1 domain interacting with the Rad9

phosphopeptide. (Right) Zoomed in surface view representation of FHA1 β loops bound

to the Rad9 phosphopeptide. The phosphate group of the phosphothreonine (pT) residue

in the peptide interacts with Ser85 and Arg70 of the FHA1 domain through electrostatic

interactions. The FHA1 domain also contains a hydrophobic pocket that accommodates

the γ-methyl group of the pT residue in the phosphopeptide. Residues C-terminal to the

pT residue of the peptide, especially +3 (The third amino acid C-terminal to the pT),

strongly interact with residues in the β4-5 and β10-11 FHA1 loops.

19

Figure 1.6. PyMOL representations of the FN3 monobody. FG, BC, and DE loops are

shown in yellow, blue, and red, respectively (PDB: 1ttg). All other residues are shown in

grey. (Left) Schematic showing β-sheets and loops. (Middle) Surface representation.

(Right) Surface representation, viewed from the top.

1.7 Thesis Goals

The overarching goal of this thesis was to develop recombinant affinity reagents to

challenging targets through phage display. I used phage-display libraries of two different

scaffolds to generate high-quality affinity reagents to phosphothreonine peptides and

conformational epitopes and reformatted some of them to work in sensitive biochemical

assays.

In Chapter 2, I successfully screened a phage library displaying variants of the

Forkhead Associated (FHA) domain and produced affinity reagents that bind

phosphopeptides corresponding to the human protein, Activation Transcription Factor 2

20

(ATF2). This segment of ATF2 contains two phosphosites and it exists in various

phosphorylated (i.e., monophosphorylated, diphosphorylated) states in the cell. Unlike a

commercial antibody, the FHA affinity reagents generated in the chapter are capable of

distinguishing between these phosphostates. I use alanine scanning to identify the

molecular recognition elements of an FHA variant that binds only to the fully

phosphorylated ATF2 peptide, and I further improve the selectivity of other FHA reagents

through mutagenesis to distinguish between ATF2’s partially phosphorylated states.

Finally, I demonstrate that an engineered FHA domain can recognize native,

phosphorylated ATF2 protein in a cell lysate analyzed by western blotting.

In Chapter 3, I generate FHA-based affinity reagents to a phosphorylated peptide of

human AKT1. These are the first FHA domains engineered to recognize a target

containing glycine at the position three residues C-terminal to the phosphothreonine

(pT+3). While the pT+3 position has previously been shown to be extremely important for

FHA binding, it was surprising to find that glycine, a small and flexible residue, could

contribute to binding at this position. Analysis by surface plasmon resonance (SPR) and

alanine scanning reveal an unusually high affinity and specificity for the phosphopeptide,

respectively. These results demonstrate the potential of developing high quality FHA-

based affinity reagents to a wide array of phosphopeptides.

In Chapter 4, I streamline the validation process of FN3 binding pairs in sandwich

ELISAs. These pairs were previously isolated through Megaprimer Shuffling for Tandem

Reagents (MegaSTAR), a technique to quickly and inexpensively generate tandem

binding reagents with potential use as sandwich assay binding pairs. My new protocols

eliminate the need to purify the affinity reagents from overexpressing bacterial cells and

21

instead use crude cell lysates directly in ELISAs. This new strategy permits testing more

pairs to discover the ones that yield the best assays. Additionally, I label FN3 proteins in

vitro with various tags through sortase-mediated ligation. Because of the simple labeling

method, researchers will be able to use these binding reagents in a variety of different

assays.

In the final chapter, I summarize my work and discuss future experiments to improve

MegaSTAR by incorporating multiple scaffolds and linkers into the selection process as

well as generate targets of selection by cell free protein synthesis. I then discuss using

sortase-mediated ligation to format binding pairs for highly sensitive homogeneous

assays. Finally, I examine the impact of my work on basic research and diagnostics.

22

1.8 Literature Cited

1. Gorman, K., McGinnis, J. & Kay, B. (2018) Generating FN3-Based Affinity Reagents Through Phage Display, Current Protocols in Chemical Biology. 10, e39.

2. Forthal, D. N. (2014) Functions of Antibodies, Microbiology spectrum. 2, 1-17.

3. Wootla, B., Denic, A. & Rodriguez, M. (2014) Polyclonal and Monoclonal Antibodies in Clinic in Human Monoclonal Antibodies: Methods and Protocols (Steinitz, M., ed) pp. 79-110, Humana Press, Totowa, NJ.

4. Litman, G. W., Rast, J. P., Shamblott, M. J., Haire, R. N., Hulst, M., Roess, W., Litman, R. T., Hinds-Frey, K. R., Zilch, A. & Amemiya, C. T. (1993) Phylogenetic diversification of immunoglobulin genes and the antibody repertoire, Molecular Biology and Evolution. 10, 60-72.

5. Kennedy, P. J., Oliveira, C., Granja, P. L. & Sarmento, B. (2018) Monoclonal antibodies: technologies for early discovery and engineering, Critical Reviews in Biotechnology. 38, 394-408.

6. Kurien, B. T. & Scofield, R. H. (2006) Western blotting, Methods. 38, 283-293.

7. Reen, D. J. (1994) Enzyme-Linked Immunosorbent Assay (ELISA) in Basic Protein and Peptide Protocols (Walker, J. M., ed) pp. 461-466, Humana Press, Totowa, NJ.

8. Dong, Q. & Li, F. (2018) Antibody Pull-Down Experiments in Fission Yeast, Methods Mol Biol. 1721, 117-123.

9. Scanziani, E. (1998) Immunohistochemical Staining of Fixed Tissues in Mycoplasma Protocols (Miles, R. & Nicholas, R., eds) pp. 133-140, Humana Press, Totowa, NJ.

10. Odell, I. D. & Cook, D. (2013) Immunofluorescence Techniques, Journal of Investigative Dermatology. 133, 1-4.

11. Leow, C. H., Fischer, K., Leow, C. Y., Cheng, Q., Chuah, C. & McCarthy, J. (2017) Single Domain Antibodies as New Biomarker Detectors, Diagnostics (Basel). 7, 52.

23

12. Hulka, B. S. & Wilcosky, T. (1988) Biological Markers in Epidemiologic Research, Archives of Environmental Health: An International Journal. 43, 83-89.

13. Aronson, J. K. & Ferner, R. E. (2017) Biomarkers—A General Review, Current Protocols in Pharmacology. 76, 9.23.1-9.23.17.

14. Zhao, X., Modur, V., Carayannopoulos, L. N. & Laterza, O. F. (2015) Biomarkers in Pharmaceutical Research, Clinical Chemistry. 61, 1343-1353.

15. Cole, L. A. (2010) Biological functions of hCG and hCG-related molecules, Reprod Biol Endocrinol. 8, 102-102.

16. Gnoth, C. & Johnson, S. (2014) Strips of Hope: Accuracy of Home Pregnancy Tests and New Developments, Geburtshilfe Frauenheilkd. 74, 661-669.

17. Kierny, M. R., Cunningham, T. D. & Kay, B. K. (2012) Detection of biomarkers using recombinant antibodies coupled to nanostructured platforms, Nano Rev. 3, 10.3402/nano.v3i0.17240.

18. Baird, C. L., Fischer, C. J., Pefaur, N. B., Miller, K. D., Kagan, J., Srivastava, S. & Rodland, K. D. (2010) Developing recombinant antibodies for biomarker detection, Cancer Biomarkers. 6, 271-279.

19. Zhu, J., Zou, N., Mao, H., Wang, P., Zhu, D., Ji, H., Cong, H., Sun, C., Wang, H., Zhang, F., Qian, J., Jin, Q. & Zhao, J. (2013) Evaluation of a modified lateral flow immunoassay for detection of high-sensitivity cardiac troponin I andmyoglobin, Biosensors and Bioelectronics. 42, 522-525.

20. Agnibh, M., Uzair, A., Martin, B., Ibrahim, A. & Michael, B. (2017) Clinically Relevant Biomarkers in Acute Heart Failure: An Update, Current Pharmaceutical Biotechnology. 18, 482-490.

21. Ji, M., Cho, B., Cho, Y. S., Park, S.-Y., Cho, S.-N., Jeon, B.-Y. & Yoon, B.-S. (2014) Development of a quantitative sandwich enzyme-linked immunosorbent assay for detecting the MPT64 antigen of Mycobacterium tuberculosis, Yonsei Med J. 55, 746-752.

24

22. Arai, H., Petchclai, B., Khupulsup, K., Kurimura, T. & Takeda, K. (1999) Evaluation of a rapid immunochromatographic test for detection of antibodies to human immunodeficiency virus, J Clin Microbiol. 37, 367-370.

23. Suzuki, M., Kato, C. & Kato, A. (2015) Therapeutic antibodies: their mechanisms of action and the pathological findings they induce in toxicity studies, J Toxicol Pathol. 28, 133-139.

24. Goodman, S. M. (2015) Rheumatoid arthritis: Perioperative management of biologics and DMARDs, Seminars in Arthritis and Rheumatism. 44, 627-632.

25. Burness, C. B. & McKeage, K. (2015) Adalimumab: A Review in Chronic Plaque Psoriasis, Drugs. 75, 2119-2130.

26. Faubion, W. A., Dubinsky, M., Ruemmele, F. M., Escher, J., Rosh, J., Hyams, J. S., Eichner, S., Li, Y., Reilly, N., Thakkar, R. B., Robinson, A. M. & Lazar, A. (2017) Long-term Efficacy and Safety of Adalimumab in Pediatric Patients with Crohn's Disease, Inflamm Bowel Dis. 23, 453-460.

27. Ecker, D. M., Jones, S. D. & Levine, H. L. (2015) The therapeutic monoclonal antibody market, MAbs. 7, 9-14.

28. Lai, Y., Wang, R., Chen, X., Tang, D., Hu, Y., Cai, J., Zhang, Q. & Hu, H. (2017) Emerging trends and new developments in monoclonal antibodies: A scientometric analysis (1980-2016), Hum Vaccin Immunother. 13, 1-10.

29. Cavallo, F., Calogero, R. A. & Forni, G. (2007) Are oncoantigens suitable targets for anti-tumour therapy?, Nature Reviews Cancer. 7, 707.

30. Ferguson, K. M. (2004) Active and inactive conformations of the epidermal growth factor receptor, Biochemical Society Transactions. 32, 742-745.

31. Albanell, J., Codony, J., Rovira, A., Mellado, B. & Gascón, P. (2003) Mechanism of Action of Anti-HER2 Monoclonal Antibodies: Scientific Update on Trastuzumab and 2C4 in New Trends in Cancer for the 21st Century: Proceedings of the International Symposium on Cancer: New Trends in Cancer for the 21st Century, held November 10–13, 2002, in Valencia, Spain (Llombart-Bosch, A. & Felipo, V., eds) pp. 253-268, Springer US, Boston, MA.

25

32. Cartron, G., Dacheux, L., Salles, G., Solal-Celigny, P., Bardos, P., Colombat, P. & Watier, H. (2002) Therapeutic activity of humanized anti-CD20 monoclonal antibody and polymorphism in IgG Fc receptor FcγRIIIa gene, Blood. 99, 754-758.

33. Zafir-Lavie, I., Michaeli, Y. & Reiter, Y. (2007) Novel antibodies as anticancer agents, Oncogene. 26, 3714.

34. de Graaf, M., Boven, E., Oosterhoff, D., van der Meulen-Muileman, I. H., Huls, G. A., Gerritsen, W. R., Haisma, H. J. & Pinedo, H. M. (2002) A fully human anti-Ep-CAM scFv-beta-glucuronidase fusion protein for selective chemotherapy with a glucuronide prodrug, British journal of cancer. 86, 811-818.

35. Sedykh, S. E., Prinz, V. V., Buneva, V. N. & Nevinsky, G. A. (2018) Bispecific antibodies: design, therapy, perspectives, Drug Des Devel Ther. 12, 195-208.

36. Blanc, R. S. & Richard, S. (2017) Arginine Methylation: The Coming of Age, Molecular cell. 65, 8-24.

37. Casey, P. J. & Seabra, M. C. (1996) Protein Prenyltransferases, Journal of Biological Chemistry. 271, 5289-5292.

38. Dalziel, M., Crispin, M., Scanlan, C. N., Zitzmann, N. & Dwek, R. A. (2014) Emerging Principles for the Therapeutic Exploitation of Glycosylation, Science. 343, 1235681.

39. Humphrey, S. J., James, D. E. & Mann, M. (2015) Protein Phosphorylation: A Major Switch Mechanism for Metabolic Regulation, Trends in Endocrinology & Metabolism. 26, 676-687.

40. L N Johnson, a. & Barford, D. (1993) The Effects of Phosphorylation on the Structure and Function of Proteins, Annual Review of Biophysics and Biomolecular Structure. 22, 199-232.

41. Zhao, S., Xu, W., Jiang, W., Yu, W., Lin, Y., Zhang, T., Yao, J., Zhou, L., Zeng, Y., Li, H., Li, Y., Shi, J., An, W., Hancock, S. M., He, F., Qin, L., Chin, J., Yang, P., Chen, X., Lei, Q., Xiong, Y. & Guan, K.-L. (2010) Regulation of cellular metabolism by protein lysine acetylation, Science (New York, NY). 327, 1000-1004.

26

42. Goel, R., Harsha, H. C., Pandey, A. & Prasad, T. S. K. (2012) Human Protein Reference Database and Human Proteinpedia as resources for phosphoproteome analysis, Mol Biosyst. 8, 453-463.

43. Khoury, G. A., Baliban, R. C. & Floudas, C. A. (2011) Proteome-wide post-translational modification statistics: frequency analysis and curation of the swiss-prot database, Scientific reports. 1, 90.

44. Suwanmajo, T. & Krishnan, J. (2015) Mixed mechanisms of multi-site phosphorylation, Journal of The Royal Society Interface. 12, 20141405.

45. Aoki, M., Yokota, T., Sugiura, I., Sasaki, C., Hasegawa, T., Okumura, C., Ishiguro, K., Kohno, T., Sugio, S. & Matsuzaki, T. (2004) Structural insight into nucleotide recognition in tau-protein kinase I/glycogen synthase kinase 3[beta], Acta Crystallographica Section D. 60, 439-446.

46. Jope, R. S., Yuskaitis, C. J. & Beurel, E. (2007) Glycogen synthase kinase-3 (GSK3): inflammation, diseases, and therapeutics, Neurochem Res. 32, 577-595.

47. van den Berg, B., Wain, R., Dobson, C. M. & Ellis, R. J. (2000) Macromolecular crowding perturbs protein refolding kinetics: implications for folding inside the cell, The EMBO journal. 19, 3870-3875.

48. Dobson, C. M. (2003) Protein folding and misfolding, Nature. 426, 884-890.

49. Hartl, F. U. (1996) Molecular chaperones in cellular protein folding, Nature. 381, 571-580.

50. Chaudhuri, T. K. & Paul, S. (2006) Protein-misfolding diseases and chaperone-based therapeutic approaches, The FEBS Journal. 273, 1331-1349.

51. Nin, D. S., Li, F., Visvanathan, S. & Khan, M. (2015) Misfolded N-CoR is Linked to the Ectopic Reactivation of CD34/Flt3-Based Stem-Cell Phenotype in Promyelocytic and Monocytic Acute Myeloid Leukemia, Frontiers in Oncology. 5.

52. Arrasate, M. & Finkbeiner, S. (2012) Protein aggregates in Huntington's disease, Exp Neurol. 238, 1-11.

27

53. Gregoire, S., Irwin, J. & Kwon, I. (2012) Techniques for Monitoring Protein Misfolding and Aggregation in Vitro and in Living Cells, Korean J Chem Eng. 29, 693-702.

54. Hanly, W. C., Artwohl, J. E. & Bennett, B. T. (1995) Review of Polyclonal Antibody Production Procedures in Mammals and Poultry, ILAR Journal. 37, 93-118.

55. Dunbar, B. S. & Schwoebel, E. D. (1990) Preparation of polyclonal antibodies in Methods in Enzymology (Deutscher, M. P., ed) pp. 663-670, Academic Press.

56. Liu, S., Li, S., Zhang, Y., Wang, Y., Zhu, Y., Wang, B. & Chen, Z.-N. (2017) Purification of a polyclonal antibody against CD147 for ELISA using antigen‑immunoaffinity chromatography, Mol Med Rep. 15, 4035-4040.

57. Nelson, P. N., Reynolds, G. M., Waldron, E. E., Ward, E., Giannopoulos, K. & Murray, P. G. (2000) Monoclonal antibodies, Mol Pathol. 53, 111-117.

58. KÖHler, G. & Milstein, C. (1975) Continuous cultures of fused cells secreting antibody of predefined specificity, Nature. 256, 495-497.

59. Corrêa, A. L., Senna, J. P. M. & de Sousa, Á. P. B. (2016) Effects of passage number on growth and productivity of hybridoma secreting MRSA anti-PBP2a monoclonal antibodies, Cytotechnology. 68, 419-427.

60. Bradbury, A. R. M., Trinklein, N. D., Thie, H., Wilkinson, I. C., Tandon, A. K., Anderson, S., Bladen, C. L., Jones, B., Aldred, S. F., Bestagno, M., Burrone, O., Maynard, J., Ferrara, F., Trimmer, J. S., Görnemann, J., Glanville, J., Wolf, P., Frenzel, A., Wong, J., Koh, X. Y., Eng, H.-Y., Lane, D., Lefranc, M.-P., Clark, M. & Dübel, S. (2018) When monoclonal antibodies are not monospecific: Hybridomas frequently express additional functional variable regions, MAbs. 10, 539-546.

61. Skerra, A. & Pluckthun, A. (1988) Assembly of a functional immunoglobulin Fv fragment in Escherichia coli, Science. 240, 1038-1041.

62. Larrick, J. W., Danielsson, L., Brenner, C. A., Abrahamson, M., Fry, K. E. & Borrebaeck, C. A. K. (1989) Rapid cloning of rearranged immunoglobulin genes from human hybridoma cells using mixed primers and the polymerase chain reaction, Biochemical and biophysical research communications. 160, 1250-1256.

28

63. Orlandi, R., Güssow, D. H., Jones, P. T. & Winter, G. (1989) Cloning immunoglobulin variable domains for expression by the polymerase chain reaction, Proceedings of the National Academy of Sciences of the United States of America. 86, 3833-3837.

64. Sastry, L., Alting-Mees, M., Huse, W. D., Short, J. M., Sorge, J. A., Hay, B. N., Janda, K. D., Benkovic, S. J. & Lerner, R. A. (1989) Cloning of the immunological repertoire in Escherichia coli for generation of monoclonal catalytic antibodies: construction of a heavy chain variable region-specific cDNA library, Proceedings of the National Academy of Sciences of the United States of America. 86, 5728-5732.

65. Smith, G. (1985) Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface, Science. 228, 1315-1317.

66. Bradbury, A. R. M., Sidhu, S., Dübel, S. & McCafferty, J. (2011) Beyond natural antibodies: the power of in vitro display technologies, Nature biotechnology. 29, 245-254.

67. Binz, H. K., Amstutz, P. & Plückthun, A. (2005) Engineering novel binding proteins from nonimmunoglobulin domains, Nature Biotechnology. 23, 1257-1268.

68. Binz, H. K. & Plückthun, A. (2005) Engineered proteins as specific binding reagents, Current Opinion in Biotechnology. 16, 459-469.

69. Paschke, M. (2006) Phage display systems and their applications, Applied Microbiology and Biotechnology. 70, 2-11.

70. Kehoe, J. W. & Kay, B. K. (2005) Filamentous Phage Display in the New Millennium, Chemical Reviews. 105, 4056-4072.

71. Boder, E. T. & Wittrup, K. D. (1997) Yeast surface display for screening combinatorial polypeptide libraries, Nature Biotechnology. 15, 553-557.

72. Feldhaus, M. J., Siegel, R. W., Opresko, L. K., Coleman, J. R., Feldhaus, J. M. W., Yeung, Y. A., Cochran, J. R., Heinzelman, P., Colby, D., Swers, J., Graff, C., Wiley, H. S. & Wittrup, K. D. (2003) Flow-cytometric isolation of human antibodies from a nonimmune Saccharomyces cerevisiae surface display library, Nature Biotechnology. 21, 163-170.

29

73. Boder, E. T., Midelfort, K. S. & Wittrup, K. D. (2000) Directed evolution of antibody fragments with monovalent femtomolar antigen-binding affinity, Proceedings of the National Academy of Sciences of the United States of America. 97, 10701-10705.

74. Seelig, B. (2011) mRNA display for the selection and evolution of enzymes from in vitro-translated protein libraries, Nature Protocols. 6, 540-552.

75. Zahnd, C., Amstutz, P. & Plückthun, A. (2007) Ribosome display: selecting and evolving proteins in vitro that specifically bind to a target, Nature Methods. 4, 269.

76. Razai, A., Garcia-Rodriguez, C., Lou, J., Geren, I. N., Forsyth, C. M., Robles, Y., Tsai, R., Smith, T. J., Smith, L. A., Siegel, R. W., Feldhaus, M. & Marks, J. D. (2005) Molecular Evolution of Antibody Affinity for Sensitive Detection of Botulinum Neurotoxin Type A, Journal of molecular biology. 351, 158-169.

77. Hanes, J., Schaffitzel, C., Knappik, A. & Plückthun, A. (2000) Picomolar affinity antibodies from a fully synthetic naive library selected and evolved by ribosome display, Nature Biotechnology. 18, 1287-1292.

78. Schier, R., McCall, A., Adams, G. P., Marshall, K. W., Merritt, H., Yim, M., Crawford, R. S., Weiner, L. M., Marks, C. & Marks, J. D. (1996) Isolation of Picomolar Affinity Anti-c-erbB-2 Single-chain Fv by Molecular Evolution of the Complementarity Determining Regions in the Center of the Antibody Binding Site, Journal of molecular biology. 263, 551-567.

79. Marvin, D. A. (1998) Filamentous phage structure, infection and assembly, Current Opinion in Structural Biology. 8, 150-158.

80. Barbas, C. F. (2001) Phage display : a laboratory manual.

81. van Wezenbeek, P. M. G. F., Hulsebos, T. J. M. & Schoenmakers, J. G. G. (1980) Nucleotide sequence of the filamentous bacteriophage M13 DNA genome: comparison with phage fd, Gene. 11, 129-148.

82. Nakamura, M., Tsumoto, K., Kumagai, I. & Ishimura, K. (2003) A morphologic study of filamentous phage infection of Escherichia coli using biotinylated phages, FEBS letters. 536, 167-172.

30

83. Feng, J. N., Russel, M. & Model, P. (1997) A permeabilized cell system that assembles filamentous bacteriophage, Proceedings of the National Academy of Sciences of the United States of America. 94, 4068-4073.

84. Marciano, D. K., Russel, M. & Simon, S. M. (1999) An Aqueous Channel for Filamentous Phage Export, Science. 284, 1516-1519.

85. Rakonjac, J. & Model, P. (1998) Roles of pIII in filamentous phage assembly11Edited by M. Gottesman, Journal of molecular biology. 282, 25-41.

86. Fuh, G. & Sidhu, S. S. (2000) Efficient phage display of polypeptides fused to the carboxy-terminus of the M13 gene-3 minor coat protein, FEBS letters. 480, 231-234.

87. Gao, C., Mao, S., Kaufmann, G., Wirsching, P., Lerner, R. A. & Janda, K. D. (2002) A method for the generation of combinatorial antibody libraries using pIX phage display, Proceedings of the National Academy of Sciences of the United States of America. 99, 12612-12616.

88. Gao, C., Mao, S., Lo, C.-H. L., Wirsching, P., Lerner, R. A. & Janda, K. D. (1999) Making artificial antibodies: A format for phage display of combinatorial heterodimeric arrays, Proceedings of the National Academy of Sciences. 96, 6025-6030.

89. Hayhurst, A. & Harris, W. J. (1999) Escherichia coliSkp Chaperone Coexpression Improves Solubility and Phage Display of Single-Chain Antibody Fragments, Protein Expression and Purification. 15, 336-343.

90. Hufton, S. E., Moerkerk, P. T., Meulemans, E. V., de Bruı̈ne, A., Arends, J.-W. & Hoogenboom, H. R. (1999) Phage display of cDNA repertoires: the pVI display system and its applications for the selection of immunogenic ligands, Journal of Immunological Methods. 231, 39-51.

91. Weiss, G. A. & Sidhu, S. S. (2000) Design and evolution of artificial M13 coat proteins11Edited by G. Von Heijne, Journal of molecular biology. 300, 213-219.

92. Zhong, G., Smith, G. P., Berry, J. & Brunham, R. C. (1994) Conformational mimicry of a chlamydial neutralization epitope on filamentous phage, Journal of Biological Chemistry. 269, 24183-24188.

31

93. Lund, P. E., Hunt, R. C., Gottesman, M. M. & Kimchi-Sarfaty, C. (2010) Pseudovirions as vehicles for the delivery of siRNA, Pharm Res. 27, 400-420.

94. Pershad, K., Wypisniak, K. & Kay, B. K. (2012) Directed evolution of the forkhead-associated domain to generate anti-phosphospecific reagents by phage display, Journal of molecular biology. 424, 88-103.

95. Huang, R., Fang, P. & Kay, B. K. (2012) Isolation of monobodies that bind specifically to the SH3 domain of the Fyn tyrosine protein kinase, New biotechnology. 29, 526-533.

96. Halperin, I., Wolfson, H. & Nussinov, R. (2003) SiteLight: binding-site prediction using phage display libraries, Protein Sci. 12, 1344-1359.

97. Freckleton, G., Lippman, S. I., Broach, J. R. & Tavazoie, S. (2009) Microarray profiling of phage-display selections for rapid mapping of transcription factor-DNA interactions, PLoS Genet. 5, e1000449-e1000449.

98. Nelson, A. L. (2010) Antibody fragments: hope and hype, MAbs. 2, 77-83.

99. Hammers, C. M. & Stanley, J. R. (2014) Antibody phage display: technique and applications, J Invest Dermatol. 134, 1-5.

100. Skerra, A. & Plückthun, A. (1988) Assembly of a Functional Immunoglobulin Fragment in Escherichia coli, Science. 240, 1038-1041.

101. Better, M., Chang, C. P., Robinson, R. R. & Horwitz, A. H. (1988) Escherichia coli Secretion of an Active Chimeric Antibody Fragment, Science. 240, 1041-1043.

102. McCafferty, J., Griffiths, A. D., Winter, G. & Chiswell, D. J. (1990) Phage antibodies: filamentous phage displaying antibody variable domains, Nature. 348, 552-554.

103. Yu, X., Yang, Y.-P., Dikici, E., Deo, S. K. & Daunert, S. (2017) Beyond Antibodies as Binding Partners: The Role of Antibody Mimetics in Bioanalysis, Annu Rev Anal Chem (Palo Alto Calif). 10, 293-320.

32

104. Bradbury, A. & Pluckthun, A. (2015) Reproducibility: Standardize antibodies used in research., Nature. 518, 27-29.

105. Voskuil, J. (2014) Commercial antibodies and their validation, F1000Res. 3, 232-232.

106. Kummer, L., Parizek, P., Rube, P., Millgramm, B., Prinz, A., Mittl, P. R., Kaufholz, M., Zimmermann, B., Herberg, F. W. & Pluckthun, A. (2012) Structural and functional analysis of phosphorylation-specific binders of the kinase ERK from designed ankyrin repeat protein libraries, Proceedings of the National Academy of Sciences of the United States of America. 109, E2248-57.

107. McGinnis, J. E., Venegas, L. A., Lopez, H. & Kay, B. K. (2018) A Recombinant Affinity Reagent Specific for a Phosphoepitope of Akt1, Int J Mol Sci. 19, 3305.

108. McGinnis, J. & Kay, B. K. (2017) Generation of recombinant affinity reagents against a two-phosphosite epitope of ATF2, New biotechnology.

109. Koide, A., Wojcik, J., Gilbreth, R. N., Hoey, R. J. & Koide, S. (2012) Teaching an old scaffold new tricks: monobodies constructed using alternative surfaces of the FN3 scaffold, Journal of molecular biology. 415, 393-405.

110. Park, S.-H., Park, S., Kim, D.-Y., Pyo, A., Kimura, R. H., Sathirachinda, A., Choy, H. E., Min, J.-J., Gambhir, S. S. & Hong, Y. (2015) Isolation and Characterization of a Monobody with a Fibronectin Domain III Scaffold That Specifically Binds EphA2, PLoS One. 10, e0132976-e0132976.

111. Friedman, M., Nordberg, E., Höidén-Guthenberg, I., Brismar, H., Adams, G. P., Nilsson, F. Y., Carlsson, J. & Ståhl, S. (2007) Phage display selection of Affibody molecules with specific binding to the extracellular domain of the epidermal growth factor receptor, Protein Engineering, Design and Selection. 20, 189-199.

112. Huang, Y.-M. M. & Chang, C.-E. A. (2011) Mechanism of PhosphoThreonine/Serine Recognition and Specificity for Modular Domains from All-atom Molecular Dynamics, BMC biophysics. 4, 12-12.

113. Huang, Y.-M. M. & Chang, C.-E. A. (2014) Achieving peptide binding specificity and promiscuity by loops: case of the forkhead-associated domain, PLoS One. 9, e98291-e98291.

33

114. Durocher, D., Taylor, I. A., Sarbassova, D., Haire, L. F., Westcott, S. L., Jackson, S. P., Smerdon, S. J. & Yaffe, M. B. (2000) The molecular basis of FHA domain:phosphopeptide binding specificity and implications for phospho-dependent signaling mechanisms, Molecular cell. 6, 1169-82.


116. Venegas, L. A., Kall, S. L., Bankole, O., Lavie, A. & Kay, B. (2018) Generating a recombinant phosphothreonine-binding domain for a phosphopeptide of the human transcription factor, c-Myc, New biotechnology.

117. Venegas, L. A., Pershad, K., Bankole, O., Shah, N. & Kay, B. K. (2016) A comparison of phosphospecific affinity reagents reveals the utility of recombinant Forkhead-associated domains in recognizing phosphothreonine-containing peptides, New biotechnology. 33, 537-43.

118. Koide, A., Bailey, C. W., Huang, X. & Koide, S. (1998) The fibronectin type III domain as a scaffold for novel binding proteins11Edited by J. Wells, Journal of molecular biology. 284, 1141-1151.

119. Butler, J. E., Ni, L., Brown, W. R., Joshi, K. S., Chang, J., Rosenberg, B. & Voss, E. W. (1993) The immunochemistry of sandwich elisas—VI. Greater than 90% of monoclonal and 75% of polyclonal anti-fluorescyl capture antibodies (CAbs) are denatured by passive adsorption, Molecular Immunology. 30, 1165-1175.

120. Jacobs, S. A., Diem, M. D., Luo, J., Teplyakov, A., Obmolova, G., Malia, T., Gilliland, G. L. & O'Neil, K. T. (2012) Design of novel FN3 domains with high stability by a consensus sequence approach, Protein Engineering, Design and Selection. 25, 107-117.

121. Batori, V., Koide, A. & Koide, S. (2002) Exploring the potential of the monobody scaffold: effects of loop elongation on the stability of a fibronectin type III domain, Protein Engineering, Design and Selection. 15, 1015-1020.

122. De Genst, E., Silence, K., Decanniere, K., Conrath, K., Loris, R., Kinne, J., Muyldermans, S. & Wyns, L. (2006) Molecular basis for the preferential cleft recognition by dromedary heavy-chain antibodies, Proceedings of the National Academy of Sciences of the United States of America. 103, 4586-4591.

34

123. Wojcik, J., Hantschel, O., Grebien, F., Kaupe, I., Bennett, K. L., Barkinge, J., Jones, R. B., Koide, A., Superti-Furga, G. & Koide, S. (2010) A potent and highly specific FN3 monobody inhibitor of the Abl SH2 domain, Nat Struct Mol Biol. 17, 519-527.

124. Yeh, J. T. H., Binari, R., Gocha, T., Dasgupta, R. & Perrimon, N. (2013) PAPTi: a peptide aptamer interference toolkit for perturbation of protein-protein interaction networks, Scientific reports. 3, 1156-1156.

125. Koide, A., Abbatiello, S., Rothgery, L. & Koide, S. (2002) Probing protein conformational changes in living cells by using designer binding proteins: application to the estrogen receptor, Proceedings of the National Academy of Sciences of the United States of America. 99, 1253-1258.

126. Huang, J., Koide, A., Nettle, K. W., Greene, G. L. & Koide, S. (2006) Conformation-specific affinity purification of proteins using engineered binding proteins: Application to the estrogen receptor, Protein Expression and Purification. 47, 348-354.

127. Richards, J., Miller, M., Abend, J., Koide, A., Koide, S. & Dewhurst, S. (2003) Engineered Fibronectin Type III Domain with a RGDWXE Sequence Binds with Enhanced Affinity and Specificity to Human αvβ3 Integrin, Journal of molecular biology. 326, 1475-1488.

128. Huang, R., Gorman, K. T., Vinci, C. R., Dobrovetsky, E., Gräslund, S. & Kay, B. K. (2015) Streamlining the Pipeline for Generation of Recombinant Affinity Reagents by Integrating the Affinity Maturation Step, Int J Mol Sci. 16, 23587-23603.

129. Fellouse, F. A., Esaki, K., Birtalan, S., Raptis, D., Cancasci, V. J., Koide, A., Jhurani, P., Vasser, M., Wiesmann, C., Kossiakoff, A. A., Koide, S. & Sidhu, S. S. (2007) High-throughput Generation of Synthetic Antibodies from Highly Functional Minimalist Phage-displayed Libraries, Journal of molecular biology. 373, 924-940.

35

Chapter 2 2. GENERATION OF RECOMBINANT AFFINITY REAGENTS AGAINST A TWO-

PHOSPHOSITE EPITOPE OF ATF2

A portion of this work was previously published in New Biotechnology [1]

This work was done in collaboration with Dr. Brian Kay. Dr. Kay helped with experimental

design and editing.

2.1 Abstract

Activating Transcription Factor 2 (ATF2) plays an important role in mammalian cell

proliferation, apoptosis and DNA repair. Its activation is dependent on the sequential

phosphorylation of residue threonine 71 (T71) followed by threonine 69 (T69) in its

transactivation domain. While these modifications can be directed by a variety of kinases,

the time to reach full phosphorylation is dependent on which signaling pathway has been

activated, which is thought to be important for proper temporal regulation. To explore this

phenomenon further, there have been ongoing efforts to generate affinity reagents for

monitoring phosphorylation events in cellular assays. While phospho-specific antibodies

have been valuable tools for monitoring cell signaling events, those raised against a

peptide containing two or more adjacent phosphosites tend to cross-react with that

peptide’s various phospho-states, rendering such reagents unusable for studying

sequential phosphorylation. As an alternative, we have employed the N-terminal

Forkhead-associated 1 (FHA1) domain of yeast Rad53p as a scaffold to generate

recombinant affinity reagents via phage display and were successful in generating a set

of reagents that can distinguish between the dual-phosphorylated epitope, 63-

IVADQpTPpTPTRFLK-77, and the mono-phosphorylated epitope, 63-

IVADQpTPTPTRFLK-77, in the human ATF2 transactivation domain.

36

2.2 Introduction

Protein phosphorylation is an important post-translational modification that allows cell

signaling pathways to activate or inhibit proteins, alter protein cellular locations, change

a protein’s stability or create new protein binding interactions [2]. Over the years, mass

spectrometry has mapped >100,000 phosphosites on >13,000 human proteins [3],

illustrating the prevalence of protein phosphorylation. However, in recent years it has

become apparent that phosphorylation is not always used as a simple switch to alter

protein function. Due to extensive regulatory networks, cells utilize multisite protein

phosphorylation to coordinate signal timing and strength, which is imperative for proper

pathway crosstalk and gene expression [4]. Sequential phosphorylation, in which multiple

residues on a protein are modified over time, provides a mechanism to delay or develop

a graded response to maintain temporal regulation [5-7].

One well-studied example illustrating the potential importance of sequential

phosphorylation is the human protein, Activating Transcription Factor 2 (ATF2). In

unstimulated conditions, ATF2 is ubiquitously expressed and maintained in an inactive

state due to intramolecular interactions between its N-terminal and C-terminal domains

[8]. Upon signaling, ATF2 is phosphorylated on threonine 71 (T71) and threonine 69 (T69)

of the protein, which disrupts an intramolecular interaction between the N- and C-termini,

allowing co-activating transcription factors, such as c-Jun, to bind the ATF2 B-Zip domain

and create a transcriptionally active heterodimer [9]. Multiple signaling pathways are

responsible for these ATF2 phosphorylation events [10], most notably stress and growth

activated pathways. When the MAPK pathway is stimulated upon stress signals, T71 is

phosphorylated by c-Jun N-terminal kinase (JNK), and, shortly thereafter, T69 is

37

phosphorylated by p38 mitogen activated protein kinase. This near simultaneous, two-

step mechanism activates ATF2 as an “early response” transcription factor [11].

Conversely, signaling that is initiated by growth factors involves a longer time to fully

phosphorylate this ATF2 epitope. This delay is caused by a first requirement for

phosphorylation of T71 through the Raf–MEK–ERK pathway. Once this modification

occurs, it takes several minutes for T69 to become phosphorylated in the Ral–RalGDS–

Src-p38 pathway [12]. While it is apparent that these two-step phosphorylation

mechanisms are utilized to alter the timing of full ATF2 activation depending on which

pathway is activated, it is unclear how the timing differences between pathways affect the

cell’s response. Monitoring these phosphorylation events in the cell would be highly useful

in exploring this further.

While mass spectrometry is a powerful method for monitoring protein phosphorylation,

antibodies that recognize phospho-epitopes have also been instrumental in tracking

individual phosphorylation events in cells [13, 14]. Unfortunately, antibodies that

recognize two or more adjacent phosphosites tend to cross-react with the singly

phosphorylated states of that epitope, and vice versa [15], limiting their usefulness in

studying dual-phosphorylation events. Consequently, there has been growing interest in

utilizing engineered protein scaffolds as alternatives to antibodies, because their target

specificity can be altered through protein engineering techniques [16]. Several protein

domains have been employed as scaffolds to engineer phospho-specific recombinant

affinity reagents, among them the Src Homology 2 domain (SH2) [17], the 10th fibronectin

type III domain (10FnIII) [18], antigen binding fragments (Fab) [19], single chain variable

fragments (scFv) [20], designed ankyrin repeat proteins (DARPins) [21] and the

38

Forkhead-associated 1 (FHA1) domain [22]. As we have previously succeeded in

generating FHA domains to a variety of pT-containing peptides, we investigated in this

study whether we could identify FHA domain variants that discriminate between mono-

and dual-phosphorylated ATF2. Given that a functionally and structurally similar FHA

domain, Dun1-FHA, requires a dual-phosphorylated target for recognition [23], we

hypothesized that we could evolve the FHA1 scaffold to behave in a similar manner. Our

FHA phage-display library, which displayed FHA domains randomized at residues 82–84

in the β4-β5 loop and residues 133–139 in the β10-β11 loop [24] was screened by affinity

selection to isolate binders to ATF2 phospho-peptides. Through phage-display, we

isolated FHA domain variants that recognize the ATF2 dual-phosphosite epitope,

containing pT at only position 69, or at both positions 69 and 71, with greater selectivity

than a commercially available antibody.

2.3 Materials and Methods

2.3.1 Reagents

The peptides were synthesized at the University of Illinois at Chicago's Research

Resource Center with >90% purity. Each peptide contained a biotin molecule at their N-

terminus and an amine group at their C-terminus. The peptide targets used for affinity

selection and reagent characterization included IVADQpTPpTPTRFLKY (pT69-pT71),

IVADQTPTPTRFLKY (T69-T71), IVADQpTPTPTRFLKY (pT69-T71), and

IVADQTPpTPTRFLKY (T69-pT71). Peptide targets used for the phosphoserine

substitution study included IVADQpSPpTPTRFLKY (pS69-pT71),

IVADQpTPpSPTRFLKY (pT69-pS71), and IVADQpSPpSPTRFLKY (pS69-pS71).

39

The commercial antibody used to compare against the FHA variants was a mouse

monoclonal antibody (mAb) generated against the dual-phosphorylated form of an ATF2

peptide (Millipore Sigma, catalog# 05-891). A goat anti-mouse IgG, conjugated to

horseradish peroxidase (HRP; ThermoFisher Scientific, catalog# 62–6520), was used to

detect the mAb retained in washed microtiter plate wells. The anti-Flag IgG, clone M2,

conjugated to HRP (Sigma-Aldrich, catalog# A8592), was used to detect the FHA variants

retained in microtiter plate wells. An anti-ATF2 IgG, conjugated to Alexa Fluor 790 (Santa-

Cruz, catalog #sc-8398) was used to detect both unphosphorylated and phosphorylated

ATF2 on a membrane blot and an anti-FLAG IgG, conjugated to DyLight™ 800 (Rockland

Immunochemicals, catalog #600-445-383) was used to detect the FHA clone F6 bound

to the membrane blot.

2.3.2 Cloning and Bacterial Expression

The phagemid DNA isolated from affinity selection against the ATF2 dual-

phosphorylated peptide was digested with NcoI and NotI restriction endonucleases and

the FHA coding region was subcloned into the pKP600 phagemid vector [22]. This

construct included a C-terminally fused gene III coding sequence, flanked by AscI

restriction sites, an N-terminally-fused DsbA leader sequence, and a FLAG epitope tag.

Upon digestion with the AscI restriction endonuclease, the gene III coding region dropped

out, and when the plasmid was ligated back together, an in-frame His6-tag at the C-

terminal (pKP600ΔIII) was generated. All vectors were sequenced using the DsbA-Fw

oligonucleotide primer. All primers and oligonucleotides were ordered from Integrated

DNA Technologies, and their sequences can be found in Table III.

40

To generate each alanine-scanning mutant, the Kunkel mutagenesis protocol [25] was

followed using the F6 clone coding region in the pKP600 vector as the template and one

of ten oligonucleotides. Each oligonucleotide converted one of the ten amino acids in the

β4-5 and β10-11 loops to alanine.

TABLE III. LIST OF PRIMERS AND THEIR SEQUENCES USED IN THIS STUDY

Primer Sequence

DsbA-Fw 5'- CGC TGG CTG GTT TAG TTT TAG CGT -3'

P82A 5'- GAC AAA TAC GCA ATG TTA CCT AAG TG -3'

Y83A 5'- GTG TTT ATT AGA CAA CGC AGG AAT GTT ACC -3'

L84A 5'- CCC AGG AGG ATT TGA AAG TGT TTA TTA GAC GCA TAA GGA ATG -3'

A133G 5'- GAT CAT TAG GCT GCC CTA CCG TAA TTT CG -3'

Q134A 5'- TGA TCA TTA GGC GCC GCT ACC GTA ATT TCG -3'

P135A 5'- GAT CAT TCG CCT GCG CTA CCG TAA TTT CGT C -3'

N136A 5'- AAA TCC GAT GAT CCG CAG GCT GCG CTA C -3'

D137A 5' GCT TAA AAT CCG ATG CGC ATT AGG CTG CG -3'

H138A 5'- GCG CAG CCT AAT GAT GCC CGG ATT TTA AGC -3'

R139A 5'- GAC TAA GCT TAA AAT CGC ATG ATC ATT AGG CTG C -3'

epβ4-β5-FW 5'- GCG CCA TGG CGA TGG AAA ATA TTA -3'

epβ4-β5-RV 5'- CCA TTT GTT GAG ATG TCG TTG -3'

epβ10-β11-FW 5'- GGC TCA ACG GTC AAA AAG TAG -3'

epβ10-β11-RV 5'- TGC GGC CGC GGT ATT TTT AAG -3'

Protein expression and purification of the FHA variants was performed as described

previously [26]. Protein purity was assessed by sodium dodecyl sulfate-polyacrylamide

gel electrophoresis (SDS-PAGE) and the protein concentration was determined with a

NanoDrop spectrophotometer.

41

2.3.3 Enzyme-Linked Immunosorbent Assay (ELISA)

Biotinylated peptides (100 μL, 5 μg/μL) were immobilized on Nunc Maxisorp™ flat-

bottom 96 well plates coated with NeutrAvidin (100 μL, 5 μg/μL) and blocked with 2%

skim milk in Phosphate Buffered Saline (PBS; 137 mM NaCl, 3 mM KCl, 8 mM Na2HPO4,

1.5 mM KH2PO4). After washing the wells with PBS with Tween 20 (PBST; 137 mM NaCl,

3 mM KCl, 8 mM Na2HPO4, 1.5 mM KH2PO4, 0.1% Tween 20), purified FHA variants in

PBST (at a range of concentrations from 0.01 μM to 1 μM) or the anti-phospho-ATF2

(Thr69/71) antibody (Millipore Sigma) in PBST (1:1000 dilution) were added and

incubated for 1 h (h). The plates were washed three times with PBST and incubated with

the anti-FLAG M2-HRP antibody or goat-anti-mouse IgG (HRP) to monitor binding of the

FHA variants and anti-ATF2 antibody, respectively. Wells were washed three times with

PBST, and 100 μL of ABTS in 50 mM sodium citrate (pH 4), 0.03% hydrogen peroxide

was added to each well. The absorbance of the color change was measured at 405 nm

on a POLARstar OPTIMA microtiter plate reader (BMG Labtech)

2.3.4 Affinity Selection of the Primary Library All selection steps were performed at room temperature. One hundred μL of

Dynabeads® MyOne™ Streptavidin T1 magnetic beads (Invitrogen) were washed three

times with PBS and incubated with 1 μg of the biotinylated phosphopeptide for 30 min.

The beads were washed once with PBS to remove unbound target and incubated with

blocking buffer (2% skim milk in PBS. with 1 μM free biotin) for 1 h. The phage library

[22], with 2 × 109 unique members, was diluted in an equal volume of 4% skim milk in

PBS and incubated with the blocked beads for 1 h. The beads were washed three times

with PBST followed by two washes with PBS to remove weak-binding or non-binding

42

phage particles. Virions were eluted from the beads by incubating with 40 μg of TPCK-

treated trypsin (Sigma-Aldrich), diluted in 400 μL of 50 mM Tris-HCl (pH 8) and 1 mM

CaCl2, and used to infect 800 μL of TG1 cells (at mid-logarithmic growth phase) for 1 h at

37 °C. The cells were plated on one 15 cm 2 × YT/Carbenicillin (CB) agar plate and

incubated overnight at 30 °C. The next day, cells were scraped with 5 mL of 2xYT/CB

medium and approximately 109 cells were inoculated into 60 mL of 2xYT/CB medium.

Once the cells had grown to mid-logarithmic phase, 30 mL was infected with trypsin

cleavable helper phage (TM13KO7) [22]. Infected cells were centrifuged and

resuspended in 30 mL of 2xYT/CB/Kanamycin (Kan) medium and amplified for 24 h at

25°C/200 rpm, and precipitated with a polyethylene glycol (PEG)/NaCl mixture to

concentrate the virions 30-fold.

The second and third rounds of selection were completed in a similar manner,

however, 0.75 μg and 0.5 μg of immobilized target peptide was incubated with the beads,

respectively. Additionally, the number of washes before phage elution were increased,

and in the third round of selection, the phage/immobilized target mixture was incubated

with free dual-phosphorylated and mono-phosphorylated peptides at a 10-fold increase

as compared to the immobilized peptide concentration. After third round phage had

infected TG1 cells at mid-logarithmic phase, serial dilutions were made and plated on

2xYT/CB containing petri plates to produce single colonies. Ninety-six clones were

propagated as phage and used as probes in a phage ELISA as previously described [22].

Positive clones were sequenced and further characterized for specificity by protein

ELISA.

43

2.3.5 Secondary Library Construction and Affinity Selection

Phagemid DNA was recovered from the third round of selection. The pooled clones

were used as a template and epβ4-β5-Fw, epβ4-β5-RV, epβ10-β11-Fw, epβ10-β11-RV

as primers for PCR. The error-prone PCR method and construction of the secondary FHA

libraries were completed as described [27]. The amplified secondary libraries were

screened through three more rounds of selection following the primary selection scheme.

Additionally, a 100-fold and 1000-fold increase of non-biotinylated mono-phosphorylated

and dual-phosphorylated peptide (as compared to the immobilized peptide target

concentration) was incubated with the phage/target mixtures for the 2nd and 3rd rounds

of secondary library selection, respectively. Individual clones were propagated and tested

for binding in an ELISA.

2.3.6 Western Blot

25 μg of Tumor necrosis factor alpha (TNFα) stimulated and 25 μg of unstimulated

HeLa cell lysates (Santa Cruz) were run on an 8-16% gradient SDS-PAGE gel and the

resolved proteins were transferred to a polyvinylidene fluoride (PVDF) membrane. After

the transfer, the blot was blocked in 5% skim milk for 1 hr at room temperature with

rocking. The blot was then probed with the anti-ATF2 IgG conjugated to Alexa Fluor 790

(1:200 in PBST) or the FHA clone F6 (1μM in PBST) for 1 hr at room temperature with

rocking. The blot probed with the FHA was washed three times with PBST and then

incubated with anti-FLAG IgG, conjugated to DyLight™ 800 (1:2000 in PBST) for 1 hr at

room temperature. Each blot was washed three times with PBST and then imaged with

the Odyssey Fc imaging system (LI-COR).

44

2.4 Results and discussion

2.4.1 The FHA Domain Scaffold Can Recognize a Dual-Phosphorylated Epitope Discovering the importance of multiple phosphorylation states in single protein

epitopes has created a need for affinity reagents to monitor such states in cells.

Considering the inability of some antibodies to distinguish between the various

phosphorylation states, we utilized our FHA domain library to search for specific binders.

To determine if the phage library, displaying 2×109 FHA domain variants, contained

clones capable of recognizing a dual-phosphorylated epitope, we performed three rounds

of affinity selection with a biotinylated synthetic phosphopeptide whose sequence was

identical to a portion of the human ATF2 sequence, including phosphorylated threonine

69 (pT69) and phosphorylated threonine 71 (pT71) (Table IV). Three FHA variants were

isolated from affinity selection with the dual-phosphorylated peptide, which is consistent

with FHA reagent generation against other mono-phosphorylated targets [22]. Although

two of the clones and a commercial monoclonal antibody, generated against the same

peptide sequence, showed cross-reactive binding to the two mono-phosphorylated forms

of the peptide, clone F6 uniquely recognized the dual-phosphorylated peptide with little

or no binding to the other peptides (Figure 2.1). These results were consistent at varying

FHA concentrations (data not shown).

2.4.2 Characterization of the Interaction Between an FHA Variant Selective for the

Dual-Phosphorylated ATF2 Peptide

To examine the interaction between clone F6 and the target ATF2 peptide further, we

utilized ATF2 dual-phosphopeptide targets that contained pS substitutions at positions 69

(pS69-pT71), 71 (pT69-pS71), or both 69 and 71 (pS69-pS71) to determine recognition

45

TABLE IV. ATF2 PHOSPHOPEPTIDES

Nomenclature and respective sequences of the ATF2 peptide targets used in the study.

Sequences were taken from residues 64-77 of human ATF2 sequence. Note an additional

Tyrosine added to the C-terminus to determine concentration by absorbance. A “p”

preceding a “T” indicates phosphothreonine.

patterns for these residues (Figure 2.2). This FHA variant had a strong preference for the

pT69-pT71 and pT69-pS71 peptides, with almost a complete loss in signal when pT69

was replaced by pS. Because all known FHA domains and the FHA1 variants, which we

have previously generated, differentiate between pS and pT residues [7, 28], it is likely

that clone F6 recognizes pT69 of the ATF2 peptide in the same manner. This is unlike

clone F6′s recognition for pT71, which requires the presence of a phosphate group but

appears to not discriminate between phospho-threonine or phospho-serine. This is

surprising given that the Dun1-FHA domain requires the second phosphorylated residue

to also be pT for binding [23]; thus, it appears that FHA domains are capable of

recognizing a broad range of dual-phosphorylated epitopes through more than one

mechanism.

46

Figure 2.1. FHA variants can recognize the dual-phosphorylated ATF2 epitope.

Biotinylated peptides were captured via NeutrAvidin-coated microtiter plates and probed

with FHA variants and an anti-phospho-ATF2 (T69/71) monoclonal mouse antibody

(mAb). Complexes were detected by the M2-HRP and goat α-rabbit-HRP antibodies,

respectively. Error bars represent standard deviation of triplicate measurements.

Figure 2.2. The FHA library contains variants that recognize a pT-X-p(S/T) motifs.

ATF2 peptides, pT69-pT71, T69-T71, and peptides in which one or both threonine

residues were replaced with pS (pT69-pS71, pS69-pT71, pS69-pS71), were used as

targets in an ELISA. Clone F6 was used as a probe. Complexes were detected with the

M2-HRP antibody. The binding signal of clone F6 for the pT69-pT71 peptide was

47

normalized to 100% and the percentage binding of all other circumstances was calculated

accordingly. Error bars represent standard deviation of triplicate measurements.

To further unpack clone F6′s interaction with the peptide, we created a series of

mutations in clone F6, in which the β4-β5 and β10-β11 loop residues were systematically

replaced with alanine, in order to determine which residues in the loops contribute to

binding the dual-phosphorylated peptide [29]. The binding of each alanine-scanned

variant was assessed in an enzyme linked immunosorbent assay (ELISA; Figure 2.3).

Most residues when mutated to alanine appeared to have little effect on binding, however,

mutating tyrosine at position 83 to alanine (Y83A) led to a complete loss of binding of

clone F6 to the dual-phosphorylated ATF2 peptide. Previous studies have determined

that the FHA1 domain and other FHA variants are highly dependent on two characteristics

of the target for recognition: the presence of pT and a specific residue C-terminally

adjacent to pT. Position 83 on the FHA1 domain is important for its interaction with the

pT + 3 residue (the third residue C-terminal to pT) on the Rad9 target [24]. Our result is

consistent with this finding, and suggests that Y83 is important for maintaining the FHA-

target complex through interaction with a C-terminally adjacent residue, such as pT71, as

well.

Interestingly, mutating L84, N136, or H138 to alanine in clone F6 led to an increase in

binding to the pT69-T71 mono-phosphorylated peptide. Considering that asparagine

contains an amide group that often plays a role in hydrogen bonding and histidine is

positively charged at physiological pH, we speculate that N136 and H138 maintain a

selective dual-phosphosite recognition of the target through hydrogen bonding and

electrostatic interactions with the negatively charged phosphate group of pT71. When

48

these residues are altered, this selectivity is lost. This would be consistent with our finding

that the clone F6′s recognition of the pT71 residue is highly dependent on the presence

of the phosphate group, serving as the pT + 2 residue, and suggests that we can

manipulate the β4-β5 and β10-β11 loop residues to alter phospho-state specificity of other

FHA reagents.

Figure 2.3. Alanine-scanning of β4-β5 and β10-β11 loop residues from FHAαATF2

F6. The ATF2 peptides were probed with clone F6 (WT) or (A) β4-β5 or (B) β10-β11 loop

alanine mutants in an ELISA. Each mutant is denoted by the identity of the residue that

has been changed, followed by its position number in the domain, and its mutation to

alanine. Complexes were detected by the M2-HRP antibody. The binding signal of WT

for the pT69-pT71 peptide was normalized to 100% and the percentage binding of all

49

other mutants was calculated accordingly. Error bars represent standard deviation of

triplicate measurements.

A competition assay was performed to estimate the affinity of the F6 clone for its

target peptide (Figure 2.4) The IC50 value of the F6 clone for the dual-phosphorylated

target was ∼1.3 μM, which is consistent with IC50 values measured for other FHA variants

isolated from our library as well as the FHA1/Rad9 complex [9, 22]. While the affinity of

F6 for its target is concerning for use as a binding reagent, its consistency with other FHA

binders suggests it would be worthwhile to explore increasing their apparent affinity

through avidity, rather than attempting to isolate high affinity binders immediately from

selection.

50

Figure 2.4. Determining FHA variant F6′s affinity for its target. Competition binding

of clone F6 to immobilized dual-phosphorylated ATF2 peptide in the presence of free

dual-phosphorylated peptide. The dashed lines indicate an IC50 of 1.3 μM for clone F6

to the dual-phosphorylated peptide.

2.4.3 Generation of FHA Affinity Reagents to Specifically Recognize the Mono-

Phosphorylated Forms of the ATF2 Peptide

Given that alanine scanning of clone F6 revealed residues in the β4-β5 and β10-β11

loops play an important role in phospho-state specific recognition of the ATF2 peptide,

we hypothesized that we could screen the FHA library for complementary clones that

specifically recognized the two mono-phosphorylated forms of the ATF2 peptides. To our

surprise, no binders specific for either of the mono-phosphorylated peptides were

isolated, which prompted us to modify our affinity selection protocol. Instead, the library

was affinity selected over three rounds (with each peptide), secondary libraries were

generated through error-prone polymerase chain reaction (PCR), and the secondary

libraries, containing ∼2×107 variants, were affinity selected for three additional rounds

(Figure 2.5A). During the second and third rounds of affinity selection, beads were

incubated with a 100-fold and 1000-fold excess of the other phosphorylated ATF2

peptides, respectively [27]. By this method, two binders were isolated against the mono-

phosphorylated peptide, pT69-T71, of which one variant, G7, appeared to be specific

(Figure 2.5B). Aside from position 82 in the β4-β5 loop, which previously did not appear

to be a contributor to phospho-state specificity, the sequences of G7 and the other

isolated clone, C1, were identical with one notable difference: G7 carries a R164C

mutation (Table V). While we observed no change in binding pattern of clone G7 in the

51

presence of the reducing reagent, dithiothreitol (DTT), upon reverting residue 164 back

to arginine, selectivity for pT69-T71 peptide was lost and the binding pattern mimicked

that of the C1 clone (not shown). The importance of this residue is surprising given that

position 164 is located on the opposing surface of the β4-β5 and β10-β11 loops, however,

it is possible that this change slightly alters FHA folding and consequently its binding

surface.

Figure 2.5. Generation of reagents that specifically recognized the mono-

phosphorylated targets. (A) Workflow used to isolate FHA reagents against pT69-T71

peptide. (B) Reagents isolated after round 6 against pT69-T71 were characterized in an

ELISA in which immobilized ATF2 peptides were used as targets. The FHA variants were

used as probes. Complexes were detected by the M2-HRP antibody. The binding signal

of the variants to their cognate target peptide were normalized to 100% and the

52

percentage binding to all other peptides was calculated accordingly. Error bars represent

standard deviation of triplicate measurements.

TABLE V. OUTPUT SEQUENCES OF CLONES ISOLATED FROM SELECTIONS

β4-β5 and β10-β11 loops are the variable regions. Framework mutations were generated

by error-prone PCR.

Our new scheme (Figure. 2.5A) was not successful in yielding unique binders for the

second mono-phosphorylated ATF2 peptide, T69-pT71. While two FHA domain variants

were isolated and demonstrated capable of binding this peptide, they also bound the dual-

phosphorylated peptide to some degree (data not shown). The crystal structure of the

FHA1 domain in complex with the phosphorylated Rad9 peptide shows that the residues

N-terminally adjacent to the pT point away from the domain and have little interaction with

it [24]. We conclude that the FHA1 scaffold cannot be used to generate variants that can

discriminate between peptides with variations that are N-terminally adjacent to the pT

53

residue and that other alternatives must be explored. First, expanding the β4-β5 loop to

increase the domain’s proximity to the N-terminally adjacent residues or using another

naturally occurring FHA domain as a binding scaffold deserves examination. Secondly,

“affinity clamping”, in which two different affinity reagents are linked together in a

polypeptide, and can bind both sides of a target peptide, is another experimental option

[30]. Another scaffold, such as the fibronectin type III (FN3) monobody, could be linked

to the FHA variant and serve to bind the exposed N-terminal residues. Thirdly, it may also

be worthwhile to explore alternative scaffolds for generating recombinant affinity reagents

to proteins with and without various post-translational modifications. For example,

DARPins have been generated that can distinguish between non-phosphorylated and

dual-phosphorylated ERK [21].

2.4.4 Using an FHA Reagent to Detect Phospho-ATF2 in a Western Blot

Our FHA reagents were screened against phosphopeptides to ensure they were

targeting the desired phosphoresidues on ATF2. However, to have value in the scientific

community, we needed to demonstate that FHA variants are capable of recognizing full-

length phosphorylated cognate proteins. We performed a western blot and probed for

phosphorylated ATF2 (expected to resolve as a 60 kDa species in an SDS-PAGE gel) in

tumor necrosis factor alpha (TNFα) stimulated and unstimulated HeLa cell lysates. ATF2

remains unphosphorylated in unstimulated HeLa cells, and is rapidly di-phosphorylated

in stimulated cells [31].

A commercial antibody that recognizes both unphosphorylated and phosphorylated

ATF2 was used as a positive control; it detected the full-length protein in both stimulated

and unstimulated cell lysates as expected (Figure 2.6A). However, a commercial

54

antibody, which was generated against a phospho-ATF2 peptide, failed to bind ATF2 in

either stimulated or unstimulated cell lysates (data not shown). Conversely, when the blot

was probed with the FHA variant, clone F6, a reactive species was detected at ATF2’s

expected size in only the lysate prepared from TNFα stimulated cells. This was the only

band detected, illustrating that clone F6 is highly specific for full-length, dully-

phosphorylated ATF2. This further indicates that FHA recombinant affinity reagents are

not only comparable but can yield more accurate results than some commercial

antibodies.

Figure 2.6. Detection of full-length ATF2 in a western blot. TNFα stimulated and

unstimulated HeLa cell lysates were probed with (A) a commercial anti-ATF2 IgG

antibody conjugated to Alexa Fluor 790 or (B) the FHA clone F6, followed by an anti-

FLAG IgG antibody conjugated to DyLight 800. Blots were imaged with the Odyssey Fc

imaging system (LI-COR).

55

2.5 Conclusions

We demonstrate here that the FHA domain can distinguish between two of three

phospho-states of a peptide segment of the human transcription factor, ATF2 and can

recognize full-length phospho-ATF2 in a western blot. Due to our ability to manipulate the

FHA sequence to increase specificity, this domain is an appealing alternative to traditional

antibodies that lack such specificity and often cannot be further engineered because there

is limited knowledge about their sequence [32]. We anticipate that the FHA scaffold may

also prove useful for generating phospho-state specific reagents against other targets.

Proteins such as Extracellular signal-regulated kinase1/2 (ERK1/2) [33] and Checkpoint

kinase 2 (CHK2) [34] make intriguing candidate targets because they are also activated

by nearby sequential phosphorylation events containing phosphothreonine. Thus,

engineered FHA domains have the potential to complement commercial antibodies in

monitoring mono- and dual-phosphorylation of nearby threonines in eukaryotic proteins.

The field of recombinant affinity reagent generation continues to grow at an increasing

rate with unique and exciting ideas to overcome existing specificity and affinity limitations

of antibodies. Our FHA reagents have potential value as phospho-state specific probes.

By combining the promising possibilities that are developing in the field and our success

thus far, we hope to utilize a set of recombinant affinity reagents in cellular assays to

detect phosphorylated ATF2 protein and provide a tool for cell biologists to study

sequential phosphorylation.

56


1. McGinnis, J. E. & Kay, B. K. (2018) Generation of recombinant affinity reagents against a two-phosphosite epitope of ATF2, New biotechnology. 45, 45-50.

2. Humphrey, S. J., James, D. E. & Mann, M. (2015) Protein Phosphorylation: A Major Switch Mechanism for Metabolic Regulation, Trends in Endocrinology & Metabolism. 26, 676-687.

3. Goel, R., Harsha, H. C., Pandey, A. & Prasad, T. S. K. (2012) Human Protein Reference Database and Human Proteinpedia as resources for phosphoproteome analysis, Mol Biosyst. 8, 453-463.

4. Nishi, H., Demir, E. & Panchenko, A. R. (2015) Crosstalk between Signaling Pathways Provided by Single and Multiple Protein Phosphorylation Sites, Journal of molecular biology. 427, 511-520.

5. Ferreon, J. C., Lee, C. W., Arai, M., Martinez-Yamout, M. A., Dyson, H. J. & Wright, P. E. (2009) Cooperative regulation of p53 by modulation of ternary complex formation with CBP/p300 and HDM2, Proceedings of the National Academy of Sciences. 106, 6591-6596.

6. Patwardhan, P. & Miller, W. T. (2007) Processive phosphorylation: Mechanism and biological importance, Cellular signalling. 19, 2218-2226.

7. Kholodenko, B. N., Hancock, J. F. & Kolch, W. (2010) Signalling ballet in space and time, Nature Reviews Molecular Cell Biology. 11, 414.

8. Liu, H., Deng, X., Shyu, Y. J., Li, J. J., Taparowsky, E. J. & Hu, C.-D. (2006) Mutual regulation of c-Jun and ATF2 by transcriptional activation and subcellular localization, The EMBO journal. 25, 1058-1069.

9. Durocher, D., Henckel, J., Fersht, A. R. & Jackson, S. P. (1999) The FHA Domain Is a Modular Phosphopeptide Recognition Motif, Molecular cell. 4, 387-394.

10. Bhoumik, A. & Ronai, Z. e. (2008) ATF2: A transcription factor that elicits oncogenic or tumor suppressor activities, Cell cycle. 7, 2341-2345.

57

11. Zhang, J.-Y., Jiang, H., Gao, W., Wu, J., Peng, K., Shi, Y.-F. & Zhang, X.-J. (2008) The JNK/AP1/ATF2 pathway is involved in H2O2-induced acetylcholinesterase expression during apoptosis, Cellular and Molecular Life Sciences. 65, 1435-1445.

12. Ouwens, D. M., de Ruiter, N. D., van der Zon, G. C. M., Carter, A. P., Schouten, J., van der Burgt, C., Kooistra, K., Bos, J. L., Maassen, J. A. & van Dam, H. (2002) Growth factors can activate ATF2 via a two-step mechanism: phosphorylation of Thr71 through the Ras–MEK–ERK pathway and of Thr69 through RalGDS–Src–p38, The EMBO journal. 21, 3782-3793.

13. Nagata, K.-i., Izawa, I. & Inagaki, M. (2001) A decade of site- and phosphorylation state-specific antibodies: recent advances in studies of spatiotemporal protein phosphorylation, Genes to Cells. 6, 653-664.

14. Jczernik, A., Girault, J.-A., Nairn, A. C., Chen, J., Snyder, G., Kebabian, J. & Greengard, P. (1991) [23] Production of phosphorylation state-specific antibodies in Methods in Enzymology pp. 264-283, Academic Press.

15. Mandell, J. W. (2003) Phosphorylation State-Specific Antibodies: Applications in Investigative and Diagnostic Pathology, The American Journal of Pathology. 163, 1687-1698.

16. Bradbury, A. R. M., Sidhu, S., Dübel, S. & McCafferty, J. (2011) Beyond natural antibodies: the power of in vitro display technologies, Nature biotechnology. 29, 245-254.

17. Kaneko, T., Huang, H., Cao, X., Li, X., Li, C., Voss, C., Sidhu, S. S. & Li, S. S. (2012) Superbinder SH2 domains act as antagonists of cell signaling, Science signaling. 5, ra68.

18. Olson, C. A., Liao, H.-I., Sun, R. & Roberts, R. W. (2008) mRNA Display Selection of a High-Affinity, Modification-Specific Phospho-IκBα-Binding Fibronectin, ACS chemical biology. 3, 480-485.

19. Shih, H. H., Tu, C., Cao, W., Klein, A., Ramsey, R., Fennell, B. J., Lambert, M., Ni Shuilleabhain, D., Autin, B., Kouranova, E., Laxmanan, S., Braithwaite, S., Wu, L., Ait-Zahra, M., Milici, A. J., Dumin, J. A., LaVallie, E. R., Arai, M., Corcoran, C., Paulsen, J. E., Gill, D., Cunningham, O., Bard, J., Mosyak, L. & Finlay, W. J. (2012) An ultra-specific avian antibody to phosphorylated tau protein reveals a unique mechanism for phosphoepitope recognition, The Journal of biological chemistry. 287, 44425-34.

58

20. Vielemeyer, O., Yuan, H., Moutel, S., Saint-Fort, R., Tang, D., Nizak, C., Goud, B., Wang, Y. & Perez, F. (2009) Direct Selection of Monoclonal Phosphospecific Antibodies without Prior Phosphoamino Acid Mapping, Journal of Biological Chemistry. 284, 20791-20795.



23. Lee, H., Yuan, C., Hammet, A., Mahajan, A., Chen, E. S., Wu, M. R., Su, M. I., Heierhorst, J. & Tsai, M. D. (2008) Diphosphothreonine-specific interaction between an SQ/TQ cluster and an FHA domain in the Rad53-Dun1 kinase cascade, Molecular cell. 30, 767-78.

24. Durocher, D., Taylor, I. A., Sarbassova, D., Haire, L. F., Westcott, S. L., Jackson, S. P., Smerdon, S. J. & Yaffe, M. B. (2000) The Molecular Basis of FHA Domain:Phosphopeptide Binding Specificity and Implications for Phospho-Dependent Signaling Mechanisms, Molecular cell. 6, 1169-1182.

25. Kunkel, T. A. (1985) Rapid and efficient site-specific mutagenesis without phenotypic selection, Proceedings of the National Academy of Sciences. 82, 488-492.

26. Kierny, M. R., Cunningham, T. D., Bouhenni, R. A., Edward, D. P. & Kay, B. K. (2015) Generating Recombinant Antibodies against Putative Biomarkers of Retinal Injury, PLoS One. 10, e0124492-e0124492.



59

29. Morrison, K. L. & Weiss, G. A. (2001) Combinatorial alanine-scanning, Current Opinion in Chemical Biology. 5, 302-307.

30. Huang, J., Makabe, K., Biancalana, M., Koide, A. & Koide, S. (2009) Structural Basis for Exquisite Specificity of Affinity Clamps, Synthetic Binding Proteins Generated through Directed Domain-interface Evolution, Journal of molecular biology. 392, 1221-1231.

31. Winston, B. W., Lange-Carter, C. A., Gardner, A. M., Johnson, G. L. & Riches, D. W. (1995) Tumor necrosis factor alpha rapidly activates the mitogen-activated protein kinase (MAPK) cascade in a MAPK kinase kinase-dependent, c-Raf-1-independent fashion in mouse macrophages, Proceedings of the National Academy of Sciences. 92, 1614-1618.

32. Venegas, L. (2017) Reagents for detecting phosphosites within proteins, Journal of Bioinformatics, Genomics and Proteomics.

33. Ahmed, S., Grant, K. G., Edwards, L. E., Rahman, A., Cirit, M., Goshe, M. B. & Haugh, J. M. (2014) Data-driven modeling reconciles kinetics of ERK phosphorylation, localization, and activity states, Mol Syst Biol. 10, 718-718.

34. Guo, X., Ward, M. D., Tiedebohl, J. B., Oden, Y. M., Nyalwidhe, J. O. & Semmes, O. J. (2010) Interdependent Phosphorylation within the Kinase Domain T-loop Regulates CHK2 Activity, Journal of Biological Chemistry. 285, 33348-33357.

60

Chapter 3 3. A RECOMBINANT AFFINITY REAGENT SPECIFIC FOR A

PHOSPHOEPITOPE OF AKT1

This research has previously been published in International journal of molecular

sciences [1].

This work was done in collaboration with Dr. Leon Venegas and Dr. Brian Kay. Dr.

Venegas helped to generate binding reagents to Akt1 and write the methods section and

Dr. Kay helped with experimental design and editing.

3.1 Abstract

The serine/threonine-protein kinase, Akt1, plays an important part in mammalian cell

growth, proliferation, migration and angiogenesis, and becomes activated through

phosphorylation. To monitor phosphorylation of threonine 308 in Akt1, we developed a

recombinant phosphothreonine-binding domain that is highly selective for the Akt1

phosphopeptide. A phage-display library of variants of the Forkhead-associated 1 (FHA1)

domain of yeast Rad53p was screened by affinity selection to the phosphopeptide, 301-

KDGATMKpTFCGTPEY-315, and yielded 12 binding clones. The strongest binders have

equilibrium dissociation constants of 160-180 nanomolar and are phosphothreonine-

specific in binding. The specificity of one FHA variant was compared to commercially

available polyclonal antibodies (pAbs) generated against the same phosphopeptide. The

clone was either as equal to or better than three pAbs, in detecting the Akt1 pT308

phosphopeptide in ELISAs.

3.2 Introduction

The serine/threonine-protein kinase, Akt1, is responsible for regulating a range of

biochemical pathways involved in cell proliferation and survival. It contains an N-terminal

61

pleckstrin homology (PH) domain, a serine/threonine kinase catalytic domain, and a C-

terminal regulatory domain [2, 3]. Activation of Akt1 is dependent on recruitment of the

protein, through its PH domain [4], to the inner side of the plasma membrane, which

causes a conformational change [5], allowing PDK1 to phosphorylate threonine 308

(T308) in Akt1’s catalytic domain and mTORC2 to phosphorylate serine 473 (S473) in

Akt1’s regulatory domain [6, 7]. Once these two residues are phosphorylated, Akt1 is fully

active and phosphorylates a range of intracellular proteins involved in cell survival,

growth, proliferation, cell migration and angiogenesis [8].

Given the critical roles that Akt1 serves in the cell, biologists are extremely interested

in understanding its involvement in cancer. Mass spectrometry and phospho-specific

antibodies have been essential tools in pursuing this question by tracking Akt1’s

phosphorylation state and levels in cells and tissues [9, 10]. Such methods have shown

a strong link between the hyperactivation of Akt1 through increased phosphorylation

levels in breast, prostate [11], ovarian [12], and pancreatic cancer [13]. Additionally,

studying phosphorylation of specific residues within a protein can provide valuable

information as some diseases are marked by the excessive phosphorylation of only one

or a few of these residues. For example, the phosphorylation of T308, but not of S473,

has been characterized as a marker of lung cancer [14]. Thus, antibodies that recognize

specific phosphorylated residues as part of their epitopes serve as valuable diagnostic

tools to distinguish between diseases caused by Akt1 deregulation.

Unfortunately, mass spectrometry is not well suited to monitoring protein

phosphorylation at the cytological level, and antibodies are often poorly validated,

unsequenced, and not amenable to protein engineering [15]. To circumvent these

62

limitations, current efforts have been focused on generating engineered protein scaffolds

that recognize phosphoepitopes, such as the 10th fibronectin type III domain (10FnIII)

[16], designed ankyrin repeat proteins (DARPins) [17], the Src Homology 2 domain (SH2)

[18], single chain variable fragments (scFv) [19], antigen binding fragments (Fab) [20],

and the Forkhead-associated (FHA) domain [21]. Unlike other scaffolds and most

antibodies, FHA domains are selective for phosphothreonine (pT)-containing targets due

to a pocket on the domain that interacts with phosphate and γ-methyl group of

phosphothreonine (pT) [22, 23]. Because of this unique characteristic, a phage library

displaying FHA1 variants randomized at residues 82-84 in the β4-β5 loop and residues

133-139 in the β10-β11 loop has been employed to generate affinity reagents to a variety

of targets [21, 22, 24]. Here, we describe the isolation and characterization of Akt1

phosphothreonine 308 (pT308)-binding reagents. We show that these reagents are pT-

dependent, bind with high affinity, and recognize the phosphopeptide with comparable or

better specificity than commercially made antibodies.


3.3.1 Peptides

Peptides were synthesized at University of Illinois at Chicago’s Research Resource

Center with >90% purity. All peptides are biotinylated at their N-terminus and amidated at

their C-termini. A phosphopeptide corresponding to human Akt1 was

KDGATMKpTFCGTPEY (Akt1-pT308) used as the target in phage display affinity

selection. In addition, a number of related peptides were synthesized:

KDGATMKTFCGTPEY (T308), KDGATMKpSFCGTPEY (pT308pS),

KDGATMKpYFCGTPEY (pT308pY), KDGATMKDFCGTPEY (pT308D), and

63

KDGATMKEFCGTPEY (pT308E), KDGAAMKpTACGTPEY (T305A),

KDGATAKpTFCGTPEY (M306A), KDGATMApTFCGTPEY (K307A),

KDGATMKpTACGTPEY (F309A), KDGATMKpTFAGTPEY (C310A),

KDGATMKpTFCATPEY (G311A), KDGATMKpTFCGAPEY (T312A), and

KDGAAAAApTAAAAAPEY (Ala).

Polyclonal antibodies (pAbs) to the Akt1-pT308 peptide were purchased from Abcam,

Cell Signaling Technology, Millipore, and ThermoFisher Scientific. The secondary

reagent for Akt1-pTBD detection was the anti-Flag epitope mAb, M2, conjugated to

horseradish peroxidase (HRP), was purchased from Sigma–Aldrich. The secondary

reagent for the pAbs was a goat anti-rabbit immunoglobulin G (IgG), conjugated to HRP

(Abcam).

3.3.2 Cloning and Bacterial Expression of Proteins

The phagemid DNA isolated from affinity selection against the Akt1-pT308 peptide

were subcloned into the pKP600ΔIII vector [25], and expressed and purified as described

elsewhere [23]. Protein purity was assessed by sodium dodecyl sulfate-polyacrylamide

gel electrophoresis (SDS-PAGE), and protein concentrations were measured with a

NanoDrop A280 spectrophotometer.

3.3.3 Affinity Selections

To isolate FHA variants, the FHA1G2 library [21] was screened against the Akt1-

pT308 peptide through three rounds of affinity selection using a modified version of a

previously described protocol [21]. Affinity selection was performed at room temperature

with the KingFisher™ mL device Purification System (ThermoFisher Scientific). The

biotinylated peptide (3 ng/μL, 400 μL) was immobilized with Dynabeads™ M-270

64

Streptavidin (ThermoFisher Scientific) and blocked with 2% skim milk in phosphate

buffered saline (PBS; 137 mM NaCl, 3 mM KCl, 8 mM Na2HPO4, 1.5 mM KH2PO4). A

ten-fold excess of the phage library containing 2×109 members was incubated with the

blocked target for 1 hour. Weak or non-binding phage variants were removed by washing

the mixture three times with PBS plus 0.1% Tween 20 (PBST) followed by three times

with PBS. Virions, which contain a trypsin-cleavable site in the helper M13 viral coat

protein III [26], were eluted from the beads with 40 µg of tosyl phenylalanyl chloromethyl

ketone-treated trypsin (Sigma-Aldrich), diluted in 400 µL of 50 mM Tris-HCl (pH 8) and 1

mM CaCl2, and used to infect 800 μL of Escherichia coli TG1 cells (at mid-logarithmic

growth phase) for 1 h at 37°C. The cells were plated onto one 15 cm petri plates

containing 2xYT medium (per liter 16 g tryptone, 10 g yeast extract, 5 g NaCl), 1.4% agar,

and 0.5 µg/µL carbenicillin, incubated overnight at 37ºC, scraped the next day, and phage

amplified. Phage particles were precipitated with 24% polyethylene glycol 8000, 3 M NaCl

and the pellet was resuspended in 0.6 mL of PBS. and virions concentrated 30-fold. The

second and third rounds of affinity selection were performed in a similar manner; however,

the Akt1-pT308 peptide concentration for rounds two and three were reduced to 750 nM

and 500 nM in 400 μL of PBS, respectively. Additionally, in the third round of affinity

selection, a 10-fold excess of non-biotinylated Akt1-pT308 peptide was added during the

wash steps. After the third round of affinity selection, 96 individual clones were

propagated as phage, and used in an ELISA to identify clones that bind to the peptide

target. The DNA inserts of positive, binding clones were sequenced.

65

3.3.4 Enzyme-Linked Immunosorbent Assay (ELISA)

Biotinylated peptides diluted in PBS were incubated overnight in 0.5 mM DTT at 4°C.

ELISAs were performed as previously described [22, 24] using the peptide targets

incubated with dithiothreitol (DTT) at a concentration of 2.5 µM in 100 μL and FHA

variants at concentrations varying from 0.01 to 10 μM in PBST. The absorbance was read

at 405 nm.

3.3.5 Surface Plasmon Resonance

The affinity of FHA variants E12 and H11 to the pT308 and T308 biotinylated peptides

was measured as previously described [27].

3.4 Results and Discussion

3.4.1 Directed Evolution of the FHA1 Domain Yielded Variants that Recognize an

Akt1 Phosphopeptide

To generate recombinant affinity reagents that recognize Akt1-pT308, a peptide with

a phosphothreonine (pT) residue at position 308 in a 13-mer peptide (positions 302 to

314) was synthesized with a C-terminal biotin (Figure 3.1). A phage-display FHA library,

containing 2 × 109 members [21], was affinity selected with the Akt1-pT308 peptide. After

three rounds of affinity selection, 96 recovered clones were analyzed by ELISA and 12

were confirmed to bind to the Akt1-pT308 peptide (Figure 3.2). None of the 12 binding

clones demonstrated any binding to the non-phosphorylated form of the 302–314 Akt1

peptide.

66

Figure 3.1. Primary structure of the Akt1 protein. The kinase consists of an N-terminal

pleckstrin homology (PH) domain (blue), a serine/threonine kinase catalytic domain

(yellow), and a C-terminal regulatory domain (pink). The amino acid sequence from

residues 301 to 314, including pT308, was used in peptide form for affinity selection

experiments.

Figure 3.2. Affinity selection process and ELISA of 12 output clones. (A) An M13

bacteriophage library, displaying variants of a thermostable Forkhead-associated 1

(FHA1) H domain, was screened through three rounds (R1–R3) of affinity selection.

67

During R3, excess phosphopeptide was added for off-rate competition, and recovered

clones were examined for binding (B) Biotinylated Akt1-pT308 peptide was captured on

a neutravidin-coated plate and probed with FHA variants that were detected by an M2-

HRP conjugated antibody. Error bars represent standard deviation of triplicate

measurements.

DNA sequencing of 12 selected FHA domains revealed that all were unique. A

comparison of their coding sequences demonstrated that many of them shared amino

acid sequences in the two regions that were randomized in the FHA domain scaffold

(Figure 3.3A). The consensus motif in the β4-β5 and β10-β11 loops determined by a

WebLogo plot [28, 29] is (S/A)Y(Y/R) and (S/T)(P/A)x(R/I)(P/E)(S/D)(H/A), respectively

(Figure 3.3B). We infer from this finding that the semiconserved positions contribute to

binding of the Akt1-pT308 peptide. Correlated with this observation, clones E12 and

binding of the Akt1-pT308 peptide. Correlated with this observation, clones E12 and H11

closely resemble the consensus sequence, and are the strongest binders, whereas

clones E1 and B3 are more diverged and are the weakest binders. Among the residues

shared most often among the binders is tyrosine at position 83, consistent with previous

conclusions that position 83 is extremely important for FHA domain interactions with their

phosphopeptide targets [24, 27, 30].

Two of the affinity selected FHA domains bind the Akt1 phosphopeptide with high

affinity. To estimate the binding strength of the affinity selected FHA domain clones for

the pT308 peptide, we performed competition binding assays. Half maximal inhibitory

concentration (IC50) values for clones B3, E1, E12 and H11 were 100 μM, 1.95 μM, 45

nM and 90 nM, respectively. (Figure 3.4A). These relative values correlated well with

68

ELISA results: clone B3 was the weakest, clones E12 and H11 were the strongest

binders, and clone E1 was an intermediate binder (Figure 3.4B). We then performed

surface plasmon resonance (SPR) for two clones, E12 and H11, and determined

equilibrium dissociation constants (KD) of 162 ± 12 nM and 178 ± 8 nM, respectively

(Table VI). These clones are among the tightest binders that we have isolated [19].

Figure 3.3. Amino acid sequence analysis of two loops randomized in the phage-

displayed scaffold. (A) Primary sequences of the wild-type (WT) form of the scaffold

and 12 variants that bind the Akt1 phosphopeptide. Three and seven residues in the β4-

β5 or β10-β11 loops, respectively, were randomized with NNK codons, where N is an

equimolar mixture of A, C, G and T and K is an equimolar mixture of G and T. Residues

that differ from the wild-type sequences are show in red. (B) WebLogo plots of the

frequency of particular residues at each position (82–84 or 133–139). The height of a

residue refers to probability of the residue at the given position. Hydrophobic, polar, and

charged residues are shown in black, green, and blue, respectively.

69

Figure 3.4. Comparing the relative affinity of four clones. (A) Competition binding of

clone B3, E1, H11 and E12 to immobilized phosphorylated Akt1 peptide in the presence

of free phosphorylated peptide. Error bars represent standard deviation of triplicate

measurements. (B) IC50 values for each clone to the Akt1 phosphorylated peptide.

TABLE VI. AFFINITY MEASUREMENTS

Affinity measurements of FHA clone E12 and H11 to the phosphorylated (Akt1-pT308)

and unphosphorylated (Akt1-T308) peptides determined by SPR.

FHA Variant

E12 H11

Peptide Target Ka (M-1s-1) Kd (s-1) KD (nM) Ka (M-1s-1) Kd (s-1) KD (nM)

Akt1-pT308 4.830*104 7.821*10-3 162±12 4.157*104 7.401*10-3 178±8

Akt1-T308 2.302 2.144*10-3 9.31*105 2.165 1.01083*10-2 5.002*106

70

3.4.2 An Isolated FHA Clone binds the Akt1 Peptide with Unique Specificity

Due to its unexpected high affinity for its phosphopeptide ligand, we chose to evaluate

clone E12 further. The specificity of this FHA variant was examined by testing for binding

to peptides that substituted phosphothreonine (pT) in the phosphopeptide with

phosphoserine (pS) or phosphotyrosine (pY). As seen in Figure 3.5, E12 only binds the

peptide when position 308 is pT. On the other hand, three commercially made polyclonal

antibodies, which were generated against the same or a similar Akt1 phosphopeptide,

varied in their specificity. While one antibody (pAb 1) demonstrated a similar level of

selectivity as clone E12, another (pAb 2) lacked specificity, and a third did not bind any

of the peptides (data not shown).

Figure 3.5. Binding of an FHA variant to a set of related peptides. A set of biotinylated

Akt1 phosphopeptides were synthesized with threonine (Akt-T308), pS (Akt-pS308) or pY

(Akt-pY308) substituted at the pT position and immobilized by NeutrAvidin. Binding of the

E12 variant and two commercially produced polyclonal antibodies were determined by an

ELISA. Error bars represent standard deviation of triplicate measurements.

71

To learn about what residues in the peptide ligand contributed to binding to clone E12,

we tested a set of alanine scanned phosphopeptides. Seven biotinylated peptides, with

each position replaced one at a time by alanine, were captured in neutravidin-coated

microtiter plate wells and probed separately with a pAb and clone E12 (Figure 3.6). The

pAb bound most alanine-scanned peptides as well as or better than the wild-type

sequence, except for alanine replacement at positions −3 and +2 in the phosphopeptide

sequence. By convention the phosphorylated amino acid is defined as 0 and resides N-

terminal and C-terminal to are numbered − and +, respectively. The pAb did not bind,

however, to the non-phosphorylated peptide or to a peptide with alanine flanking the

central pT position. Conversely, the E12 clone demonstrated very little or no binding to

all of the test peptides. Additionally, it had a reduced ability to bind corresponding peptides

from isoforms Akt2 or Akt3 whose sequences differ by only one or two amino acids in

positions not tested by the alanine scan (Figure 3.7). This was surprising since previous

observations show that FHA domains recognize only a few nearby residues in addition to

the central pT. In this case, though, it appears that many of the residues that flank the

central pT in the Akt1 phosphopeptide contribute directly or indirectly to binding.

Preliminary molecular dynamic studies of the peptide and the solved structure of

activated Akt [31] suggest hairpin loops form on both sides of the pT residue, which could

explain why so many adjacent residues are important directly or indirectly for the binding

interaction. While more biophysical experiments are needed to dissect further details

regarding the binding interaction, the apparent contribution of multiple flanking residues

to the interaction correlates with the unusually tight binding observed for the E12 variant

and suggests further mutagenesis could increase its specificity for the Akt1 isoform.

72

Figure 3.6. Identification of important residues on the peptide by alanine scanning.

Alanine (A) was substituted one at a time at the −3, −2, −1, +1, +2, +3 and +4 positions

of the Akt1 peptide. Binding of E12 and a polyclonal antibody (pAb) to the wild type Akt1

peptide target was set to 100%, and the phosphopeptide variants were normalized

against it in an ELISA. Error bars represent standard deviation of triplicate measurements.

73

Figure 3.7. Binding of an FHA variant to corresponding peptides from Akt2 and

Akt3. A set of biotinylated phosphopeptides containing the Akt1 sequence, the Akt2

corresponding sequence (SDGATMKpTFCGTPE) and the Akt3 corresponding sequence

(TDAATMKpTFCGTPE) were immobilized by neutravidin. Binding of the E12 variant and

a commercially produced polyclonal antibody (pAb) were determined by an ELISA. Error

bars represent standard deviation of triplicate measurements.

Published results demonstrate that the three residues C-terminally adjacent to the pT

residue, most notably the +3 residue, contribute significantly to FHA domain interactions

with peptides [21, 22, 24, 27]. While many of the 20 amino acids occur among

phosphopeptide ligands for FHA domains, the Akt1 pT308 FHA variant is the first one

observed to have glycine at the +3 position. Glycine’s small side group and

conformational flexibility together may explain why it is not commonly positioned at

protein-protein interfaces [32]. Thus, it seems likely that these FHA variants are

compensating by interacting more strongly with the remaining peptide residues, and

therefore, have a higher specificity and affinity for Akt1 peptide target than other reported

FHA binding domains. Given this, clone E12 may serve as an attractive reagent for

probing full length phosphorylated Akt1 in western blots, fixed cells, or tissue sections.

Such experiments will provide further insight into clone E12’s potential as a diagnostic

reagent.

74

3.5 Conclusions

Herein, we demonstrate the directed evolution of the FHA domain to bind a

phosphorylated peptide that corresponds to a segment of the phosphorylated,

oncoprotein, Akt1. The work represented in this paper bolsters the utility of the

recombinant FHA domain to serve as an alternative to traditional antibodies for detecting

phosphothreonine in peptide sequences. Among the 12 variants isolated from a phage-

display FHA domain library, we discovered two that bind the Akt1 phosphopeptide,

KDGATMKpTFCGTPEY, with 160–180 nM affinity. This is the strongest interaction

between a peptide target and an FHA variant isolated from our library to date.

While these FHA variants have yet to be employed as binding reagents against full

length protein targets, this FHA probe has the potential to detect phosphorylated Akt1

protein in vivo and/or in vitro due to this high affinity. Thus, we have achieved an early

milestone in our goal to replace anti-phosphothreonine antibodies with engineered

recombinant affinity reagents. Further work to optimize assay detection methods will

prove this FHA variant’s potential as a detection and diagnostic reagent.

75


1. McGinnis, J. E., Venegas, L. A., Lopez, H. & Kay, B. K. (2018) A Recombinant Affinity Reagent Specific for a Phosphoepitope of Akt1, Int J Mol Sci. 19, 3305.

2. Liu, P., Wang, Z. & Wei, W. (2014) Phosphorylation of Akt at the C-terminal tail triggers Akt activation, Cell cycle. 13, 2162-4.

3. Manning, B. D. & Toker, A. (2017) AKT/PKB Signaling: Navigating the Network, Cell. 169, 381-405.

4. Stephens, L., Anderson, K., Stokoe, D., Erdjument-Bromage, H., Painter, G. F., Holmes, A. B., Gaffney, P. R., Reese, C. B., McCormick, F., Tempst, P., Coadwell, J. & Hawkins, P. T. (1998) Protein kinase B kinases that mediate phosphatidylinositol 3,4,5-trisphosphate-dependent activation of protein kinase B, Science. 279, 710-4.

5. Huang, B. X. & Kim, H. Y. (2006) Interdomain conformational changes in Akt activation revealed by chemical cross-linking and tandem mass spectrometry, Molecular & cellular proteomics : MCP. 5, 1045-53.

6. Alessi, D. R., Andjelkovic, M., Caudwell, B., Cron, P., Morrice, N., Cohen, P. & Hemmings, B. A. (1996) Mechanism of activation of protein kinase B by insulin and IGF-1, The EMBO journal. 15, 6541-51.

7. Sarbassov, D. D., Guertin, D. A., Ali, S. M. & Sabatini, D. M. (2005) Phosphorylation and regulation of Akt/PKB by the rictor-mTOR complex, Science. 307, 1098-101.

8. Manning, B. D. & Cantley, L. C. (2007) AKT/PKB signaling: navigating downstream, Cell. 129, 1261-74.

9. Kuo, Y. C., Huang, K. Y., Yang, C. H., Yang, Y. S., Lee, W. Y. & Chiang, C. W. (2008) Regulation of phosphorylation of Thr-308 of Akt, cell proliferation, and survival by the B55alpha regulatory subunit targeting of the protein phosphatase 2A holoenzyme to Akt, The Journal of biological chemistry. 283, 1882-92.

10. Wang, F., Zhang, W., Guo, L., Bao, W., Jin, N., Liu, R., Liu, P., Wang, Y., Guo, Q. & Chen, B. (2014) Gambogic acid suppresses hypoxia-induced hypoxia-inducible factor-1alpha/vascular endothelial growth factor expression via inhibiting phosphatidylinositol 3-kinase/Akt/mammalian target protein of rapamycin pathway in multiple myeloma cells, Cancer science. 105, 1063-70.

76

11. Nakatani, K., Thompson, D. A., Barthel, A., Sakaue, H., Liu, W., Weigel, R. J. & Roth, R. A. (1999) Up-regulation of Akt3 in estrogen receptor-deficient breast cancers and androgen-independent prostate cancer lines, The Journal of biological chemistry. 274, 21528-32.

12. Yuan, Z. Q., Sun, M., Feldman, R. I., Wang, G., Ma, X., Jiang, C., Coppola, D., Nicosia, S. V. & Cheng, J. Q. (2000) Frequent activation of AKT2 and induction of apoptosis by inhibition of phosphoinositide-3-OH kinase/Akt pathway in human ovarian cancer, Oncogene. 19, 2324-30.

13. Altomare, D. A., Tanno, S., De Rienzo, A., Klein-Szanto, A. J., Tanno, S., Skele, K. L., Hoffman, J. P. & Testa, J. R. (2002) Frequent activation of AKT2 kinase in human pancreatic carcinomas, Journal of Cellular Biochemistry. 87, 470-476.

14. Vincent, E. E., Elder, D. J., Thomas, E. C., Phillips, L., Morgan, C., Pawade, J., Sohail, M., May, M. T., Hetzel, M. R. & Tavare, J. M. (2011) Akt phosphorylation on Thr308 but not on Ser473 correlates with Akt protein kinase activity in human non-small cell lung cancer, British journal of cancer. 104, 1755-61.

15. Bradbury, A. & Pluckthun, A. (2015) Reproducibility: Standardize antibodies used in research., Nature. 518, 27-29.

16. Olson, C. A., Liao, H. I., Sun, R. & Roberts, R. W. (2008) mRNA display selection of a high-affinity, modification-specific phospho-IkappaBalpha-binding fibronectin, ACS chemical biology. 3, 480-5.


18. Kaneko, T., Huang, H., Cao, X., Li, X., Li, C., Voss, C., Sidhu, S. S. & Li, S. S. (2012) Superbinder SH2 domains act as antagonists of cell signaling, Science signaling. 5, ra68.

19. Vielemeyer, O., Yuan, H., Moutel, S., Saint-Fort, R., Tang, D., Nizak, C., Goud, B., Wang, Y. & Perez, F. (2009) Direct selection of monoclonal phosphospecific antibodies without prior phosphoamino acid mapping, The Journal of biological chemistry. 284, 20791-5.

77

20. Shih, H. H., Tu, C., Cao, W., Klein, A., Ramsey, R., Fennell, B. J., Lambert, M., Ni Shuilleabhain, D., Autin, B., Kouranova, E., Laxmanan, S., Braithwaite, S., Wu, L., Ait-Zahra, M., Milici, A. J., Dumin, J. A., LaVallie, E. R., Arai, M., Corcoran, C., Paulsen, J. E., Gill, D., Cunningham, O., Bard, J., Mosyak, L. & Finlay, W. J. (2012) An ultra-specific avian antibody to phosphorylated tau protein reveals a unique mechanism for phosphoepitope recognition, The Journal of biological chemistry. 287, 44425-34.



23. Huang, Y. M. & Chang, C. E. (2011) Mechanism of PhosphoThreonine/Serine Recognition and Specificity for Modular Domains from All-atom Molecular Dynamics, BMC biophysics. 4, 12.

24. McGinnis, J. & Kay, B. K. (2017) Generation of recombinant affinity reagents against a two-phosphosite epitope of ATF2, New biotechnology.

25. Pershad, K., Sullivan, M. A. & Kay, B. K. (2011) Drop-out phagemid vector for switching from phage displayed affinity reagents to expression formats, Analytical biochemistry. 412, 210-6.

26. Goletz, S., Christensen, P. A., Kristensen, P., Blohm, D., Tomlinson, I., Winter, G. & Karsten, U. (2002) Selection of large diversities of antiidiotypic antibody fragments by phage display, Journal of molecular biology. 315, 1087-97.

27. Venegas, L. A., Kall, S. L., Bankole, O., Lavie, A. & Kay, B. (2018) Generating a recombinant phosphothreonine-binding domain for a phosphopeptide of the human transcription factor, c-Myc, New biotechnology.

28. Crooks, G. E., Hon, G., Chandonia, J.-M. & Brenner, S. E. (2004) WebLogo: A Sequence Logo Generator, Genome Research. 14, 1188-1190.

29. Schneider, T. D. & Stephens, R. M. (1990) Sequence logos: a new way to display consensus sequences, Nucleic acids research. 18, 6097-6100.

78

30. Durocher, D., Taylor, I. A., Sarbassova, D., Haire, L. F., Westcott, S. L., Jackson, S. P., Smerdon, S. J. & Yaffe, M. B. (2000) The Molecular Basis of FHA Domain:Phosphopeptide Binding Specificity and Implications for Phospho-Dependent Signaling Mechanisms, Molecular cell. 6, 1169-1182.

31. Yang, J., Cron, P., Good, V. M., Thompson, V., Hemmings, B. A. & Barford, D. (2002) Crystal structure of an activated Akt/Protein Kinase B ternary complex with GSK3-peptide and AMP-PNP, Nature Structural Biology. 9, 940.

32. Peri, C., Morra, G. & Colombo, G. (2016) Surface energetics and protein-protein interactions: analysis and mechanistic implications, Scientific reports. 6, 24035.

79

Chapter 4 4. STREAMLING THE VALIDATION PROCESS FOR BINDING REAGENTS

ISOLATED BY MEGASTAR

4.1 Abstract

The “sandwich” assay, in which the antigen is sandwiched between two binding

reagents, is considered the “gold standard” for diagnostics and other biochemical tests.

However, pairs are difficult to find. A technique invented in the lab, Megaprimer Shuffling

for Tandem Affinity Reagents (MegaSTAR), minimizes the cost and time required to find

non-overlapping binding pairs by isolating simultaneous binders during the phage

display affinity selection process. In MegaSTAR, two binding reagents are displayed in

tandem on the surface of an M13 bacteriophage particle, and clones in which both

reagents simultaneously contribute to binding, causing increased apparent affinity

(avidity), are selected under stringent conditions. To demonstrate that a tandem pair can

work in a sandwich assay format, one subclones each coding region, overexpresses

each in Escherichia coli, purifies them, and tests them pair-wise in an assay. To

streamline this validation process, we shortened the time it takes to test reagents in a

sandwich assay format by eliminating the need to purify the binding pairs, and at the

same time, maintain performance metrics of a 6-log dynamic range and a 10 picomolar

limit of detection. Additionally, we in vitro labeled FN3 proteins with various tags through

sortase-mediated ligation, which allows binders to quickly be reformatted for a variety of

different sandwich assays.

80

4.2 Introduction

The concentrations of proteins, hormones, and signaling molecules vary in cells and

organisms in response to environmental stimuli. However, their concentrations can also

change drastically due to a condition or disease, making such molecules excellent

biomarkers [1-3]. Evaluating this change in concentration of a biomarker in a patient

sample is a useful and quick diagnostic method. There are many such assays on the

market, including those that test for pregnancy [4, 5], cancer [6, 7], myocardial infarction

[8], and bacterial and viral infections [9, 10] that rely on a specific molecule’s deviation

from a normal concentration for diagnosis. For example, the cardiac troponin I and T

proteins exist at very low quantities in the blood of healthy patients. However, both are

released into the blood when there is heart damage (i.e., myocardial infarction) in

patients. The levels of troponin I or T directly in the blood correlate with the amount of

damage to the heart, making them excellent biomarkers for monitoring heart disease in

patients [11].

Currently, the most common diagnostic assay format is the enzyme-linked

immunosorbent assay (ELISA) [12]. This is a microtiter plate-based assay in which the

antigen is immobilized on the bottom of wells and detected by an antibody conjugated

to an enzyme. The antigen’s concentration is quantified by the conjugated enzyme’s

conversion of a substrate to a measurable product [13]. More specifically, the sandwich

ELISA, in which the antigen is “ sandwiched" between two antibodies, is considered the

gold standard in diagnostics because it requires two separate antibody binding events

to occur in order to produce a signal (Figure 4.1) [14, 15]. Requiring that two different

antibodies bind to an analyte increases the assay’s specificity and minimizes false

81

positive results. The sandwich assay is a valuable tool in basic and clinical research in

which it can be used to monitor changes in a protein’s levels upon addition of various

hormones or drugs to cells, tissues, or patients [16, 17].

Figure 4.1. Sandwich ELISA. The capture antibody (Ab, blue) is adsorbed to the

microtiter plate’s surface. The well is then incubated with a complex mixture containing

the analyte (yellow), washed, and incubated with the detection antibody (black). The

amount of analyte retained in the well is quantified by the addition of a substrate that is

converted to a measurable product by the enzyme. For this assay to work, the capture

and detection antibodies must be able to bind simultaneously to the analyte (i.e., at non-

overlapping epitopes). Analytes can be proteins or small molecules.

82

Often, non-overlapping binding pairs for a sandwich assay are identified by generating

antibodies and then many are tested in pair-wise combinations. For example, to build an

assay to a particular analyte, one would buy ten antibodies (> $3000) to that protein, and

test in 45 different pairwise combinations. If lucky, one working pair may be discovered.

This trial and error strategy is time consuming, expensive, and while the best single

binding reagents are isolated in this process, these are not necessarily the best

reagents. To improve this process, our lab has invented Megaprimer Shuffling for

Tandem Reagents (MegaSTAR) [18]. This new technology involves four steps. First,

target binders are isolated by phage display. Second, binder coding regions are

amplified by PCR. These “megaprimers” are annealed to a new, single-stranded,

uricilated DNA “tandem display” phagemid vector that contains two places for the binding

variant genes to anneal, which will be done at random, with a linker between the two

binding spots. Third, the megaprimers are extended, ligated, and the heteroduplex DNA

is transformed into E. coli cells, where the uricilated parental DNA strand is degraded by

cellular enzymes [19]. Fourth, the new, tandem display library is amplified and screened

against the analyte under stringent conditions. Pairs binding simultaneously to the

analyte are selected due to their bivalent binding, which increases their apparent affinity,

or avidity. Recovered clones are confirmed for binding by ELISA and are DNA

sequenced [18].

The MegaSTAR process eliminates much of the time and cost required to generate

pairs that can bind simultaneously to an analyte. However, there remains a downstream

bottleneck - confirming and identifying the best pairs for sandwich assays - which can

83

take a considerable amount of effort. In this Chapter, I describe my improvements to the

validation steps downstream of MegaSTAR.

Figure 4.2. Megaprimer shuffling for tandem affinity reagents. ① The primary

phage library is screened against an analyte and ② the coding regions of the selection

output are PCR amplified and used as “megaprimers" to anneal to a single-stranded,

uracilated, tandem vector. ③ The primers are extended, ligated, and the DNA is

transformed into E. coli, where the parental strand is preferentially degraded. ④ The

new tandem library is amplified and screened for binding to the analyte under stringent

conditions. Virions displaying pairs that can bind simultaneously to the analyte are affinity

selected by avidity.

84


4.3.1 Cloning and Overexpression of FN3 Fusion Proteins

To generate the FN3-AviTag fusion construct, the coding region of one of the FN3

variants from the tandem clone was PCR-amplified and subcloned into a modified

version of the pET-14b vector (Novagen) that contains the AviTag sequence in frame.

The vector also includes a hexahistidine tag for purification [20] and a small ubiquitin-

like modifier (SUMO) tag for improved expression [21]. This was repeated with the other

FN3 variant from the tandem dimer, but instead, it was subcloned into a modified pET-

14b vector to generate a C-terminal NanoLuc fusion [22]. This FN3 was also subcloned

into pKP300ΔIII [23] which contains a DsbA signal sequence to direct the C-terminal

alkaline phosphatase fusion protein to the E. coli periplasm [24].

For expression of the various FN3 fusion proteins, the FN3-Avitag fusion construct

was transformed in AVB101 E. coli cells (Avidity, LLC), which promote in vivo

biotinylation during protein expression [25], and the FN3-Nanoluc fusion construct was

transformed into BL21-CodonPlus cells. Transformed cells were inoculated in 5 mL 2xYT

medium (16 g tryptone, 10 g yeast extract, 5 g NaCl per liter) cultures, supplemented

with carbenicillin (50 µg/mL), and Overnight Express Autoinduction System 1 (Novagen).

Cultures were grown for 24 hr at 30°C, 300 revolutions per min (rpm). The FN3-Alkaline

fusion construct was transformed in TG1 cells (genotype: F' [traD36 proAB+ lacIq

lacZΔM15] supE thi-1 Δ(mcrB- hsdSM)5(rK-mK-) Δ(lac-proAB) and expressed for 24 hr

in 5 mL of 2xYT supplemented with carbenicillin (50 µg/mL) at 30°C, 300 rpm.

After 24 hours of overexpression, cells were spun down at 10,000 xg for 10 minutes.

Pellets were resuspended in 1 mL of BugBuster Protein Extraction Reagent (Millipore)

85

and incubated at room temperature for 15 min, while rocking. Lysates were spun down

again and the supernatant was used immediately in a sandwich ELISA.

4.3.2 Isolation of binding pairs by MegaSTAR

Soluble COP9 Signalosome Subunit 5 (COPS5) target protein was a gift from Dr.

Susanne Gräslund’s group from the Structural Genomics Consortium in Toronto,

Canada. The FN3 phage library was subjected to two rounds of affinity selection against

the COPS5 target, as previously described [26]. The output pool of clones was used to

generate a secondary, tandem dimer phage-display library. The COPS5 target was then

screened against the secondary, tandem library through two more rounds of affinity

selection. Generation of the secondary library and selection of tandem dimers was

performed as previously described [18].

Ninety-six output clones were screened in an ELISA as previously described [18].

DNA was prepared from the ten clones that yielded the strongest ELISA signal with the

Wizard DNA Miniprep kit (Promega) and sequenced using the oligonucleotide primer,

5’-CGCTGGCTGGTTTAGTTTTAGCGT-3’. Sequences were then compared to those

previously generated [18].

4.3.3 Sandwich ELISA with FN3 fusion binding pairs

Lysates of bacteria expressing FN3-AviTag fusion proteins were added to a Nunc

Maxisorp™ flat-bottom 96 well plate, which was coated with NeutrAvidin (100 μL, 5

μg/μL) and blocked with 2% skim milk in phosphate buffered saline (PBS; 137 mM NaCl,

3 mM KCl, 8 mM Na2HPO4, 1.5 mM KH2PO4). After washing the wells with PBS+0.1%

Tween 20 (PBST), bacterial lysates spiked with recombinant COPS5 protein were added

to the plate for a 1 hr incubation. The plates were washed three times with PBST and

86

incubated with lysates overexpressing FN3-NanoLuc or FN3-alkaline phosphatase

fusions for 1 hr. Wells were again washed three times with PBST. To quantify the amount

of NanoLuc present, Nano-Glo® Luciferase substrate (Promega) was added to the plate

and luminescence was measured on the POLARstar OPTIMA microtiter plate reader

(BMG Labtech). Alkaline phosphatase was quantified by addition of p-Nitrophenyl

Phosphate substrate and the absorbance of the wells was measured at 405 nm with the

POLARstar OPTIMA (BMG Labtech) plate reader.

4.3.4 Monobody labeling via sortase reaction

The Sortase-Tag, LPxTG, was added C-terminal of FN3 coding regions by PCR and

the amplified DNA products were subcloned into the pET-14b vector. These fusion

proteins were then expressed and purified as previously described [27].

The FN3-Sortase-Tag fusion proteins (3.75 mM) were incubated with 1 unit of Sortase

A5 enzyme (Active Motif), and 125 μM of (Glycine)5-Biotin label or 25 μM of (Glycine)5-

HRP label (Active Motif) for 1 hr at 30°C, 1000 rpm. After incubation, the reactions were

applied to Amicon® Ultra Centrifugal Filters (Millipore) with 10 and 50 kDa thresholds,

respectively. Unconjugated (Glycine)5-Biotin label and (Glycine)5-HRP label (Active

Motif) were removed from the reaction product with three PBS washes. Purity of the

monobodies was analyzed by sodium dodecyl sulfate-polyacrylamide gel

electrophoresis (SDS-PAGE).

4.3.5 Sandwich ELISA with monobodies labeled via sortase reaction

Purified, FN3-(Glycine)5-Biotin (100 μL, 10 μg/μL) was added to a Nunc Maxisorp™

flat-bottom 96 well plate coated with NeutrAvidin (100 μL, 5 μg/μL) and blocked with 2%

skim milk in PBS. After washing the wells with PBST, bacterial lysates spiked with

87

recombinant COPS5 protein were added to the plate for a 1 hr incubation. The plates

were washed three times with PBST and incubated with 0.1 μM of purified, FN3-

(Glycine)5-HRP for 1 h. Wells were again washed three times with PBST, and 100 μL of

2’,2’-Azino-Bis 3-Ethylbenzothiazoline-6-Sulfonic Acid (ABTS) in 50 mM sodium citrate

(pH 4), 0.03% hydrogen peroxide was added to each well. The absorbance of the color

change was measured at 405 nm on a POLARstar OPTIMA microtiter plate reader.

4.4 Results and Discussion

4.4.1 MegaSTAR is robust and reproducible

Binding pairs to the human proteins Cop9 signalosome subunit 5 (COPS5), p21-

associated kinase 1 (PAK1), and Rho GTPase-activating protein 32 (RICS) have

successfully been generated through MegaSTAR [18]. To determine that this technique

is robust and reproducible, we repeated the entire selection and MegaSTAR process

with the human target protein, COPS5, a subunit of an important regulator in multiple

signaling pathways. Both FN3 motifs and frequency of the tandem clones were highly

conserved with the previous results (Figure 4.3). Because of this reproducibility, we are

likely isolating the strongest tandem binders from the library.

4.4.2 Improving binding pairs by assay development

MegaSTAR minimizes the time and cost involved in generating binding pairs that

work in sandwich assays. However, these binding pairs require improvements to work

in diagnostic tests that target analytes existing in complex mixtures and low, endogenous

concentrations. We, therefore, wanted to improve the sandwich binding reagents’

sensitivity and specificity for the analyte.

88

Figure 4.3. Loop sequences of tandem clones isolated against COPS5. Comparison

of FN3 clones isolated previously (Left) [18] to repeated results (Right). Each motif

determined by the randomized BC and FG loops are represented by a different color.

The frequency with which each tandem clone appears and number of total clones

analyzed are listed. FN3a and FN3b do not distinguish between orientation for each pair.

Conserved positions are noted by their single-letter amino acid abbreviations while

positions that varied are noted with an “X.”

To improve the sensitivity and specificity of binding pairs, we wanted to reformat

many tandem clones as two separate binding reagents and test each pair in the

sandwich assay set up. The current validation process after MegaSTAR and secondary

selection requires testing clones for analyte binding as tandem dimers in an indirect

ELISA. Only the few clones producing strong positive signals are sequenced. The clones

whose sequences appear at high frequency are subcloned, expressed, and purified as

separate, soluble proteins and finally tested in the sandwich ELISA setup. This

89

problematic strategy assumes high affinity and high frequency tandem dimers produce

the best separate binding pairs for a sensitive sandwich assay with a large dynamic

range. It would be highly desirable to test many more clones by sandwich ELISA.

However, the bottleneck for testing hundreds of pairs is due to the extensive time and

cost required to sequence many tandem dimers, reformat as separate binders, and

purify each monobody.

To lessen this bottleneck, we devised a strategy to eliminate the need to purify each

monobody, and instead, use them directly in crude cell lysates for assays. After isolating

tandem dimers, we cloned the FN3 coding regions into separate capture and detection

fusion vectors (Figure 4.4). The capture FN3 fusion vector contains an AviTag, allowing

for protein biotinylation in vivo, when expressed in E. coli cells that also contain an

isopropyl-β-D-thiogalactopyranoside (IPTG) inducible plasmid for BirA biotin ligase

expression [25]. Biotinylating the capture FN3 monobody as it is expressed in vivo

makes its capture by NeutrAvidin possible while in cell lysates. The detection vectors

genetically fuse NanoLuc or alkaline phosphatase enzymes, which express well in E.

coli cells, directly to the monobody. These enzymes convert substrates to measurable

products to make the monobody fusions directly detectable [22, 28]. Often in ELISAs,

the primary detection antibody or binding reagent is not conjugated to a read-out enzyme

and requires a secondary antibody for detection. Genetically fusing the enzyme directly

to the primary detection binder eliminates the need for a secondary reagent, which

reduces assay time and the number of wash steps. These improvements facilitate better

binding between the analyte and the detection monobody such that they can also be

used directly from cell lysates.

90

Figure 4.4. Monobody fusion proteins for improved sandwich assays. (A) Flow

chart outlining the generation and characterization of binding pairs isolated by

MegaSTAR. (B) Cartoons of the sub-elements in the capture and detection monobody

fusion proteins. Not to scale.

Once the FN3 coding regions were transferred to the capture and detection fusion

vectors, expressed in small volume cultures, and lysed, they were used immediately in

a sandwich ELISA (Figures 4.5A, 4.6A). The sandwich ELISAs measured the amount

of COPS5 analyte spiked over a range of concentrations into a bacterial cell lysate

(Figures 4.5B, 4.6B). Both sandwich assays successfully determined the analyte

concentration over approximately a 6-log dynamic range, which is comparable, if not

better, to other sandwich ELISAs [29-31].

91

Figure 4.5. Sandwich ELISA with NanoLuc readout. (A) Set up of the sandwich ELISA

with detection by NanoLuc. Plates are coated with NeutrAvidin, which then bind

biotinylated, capture FN3 from bacterial cell lysates. The immobilized complex is then

incubated with the analyte, COPS5, whose presence is detected by the FN3 fused to

NanoLuc and the addition of substrate. As the NanoLuc enzyme converts the furimazine

to furimamide, light is released. (B) Detection of the COPS5 analyte spiked over a range

of concentrations into a bacterial cell lysate. “Cognate pair” indicates both FN3 binding

reagents used in the assay originated from a tandem clone isolated against COPS5,

while “unrelated capture reagent” indicates use of a capture FN3 to an unrelated target

(i.e., negative control). Error bars represent standard deviation of triplicate

measurements. Data analyzed with OriginLab software.

92

Figure 4.6. Sandwich ELISA with alkaline phosphatase (AP) readout. (A) Set up of

the sandwich ELISA with AP detection. Plates are coated with NeutrAvidin, which then

bind biotinylated, capture FN3 from bacterial cell lysates. The complex is then incubated

with the analyte, COPS5, whose presence is detected by the FN3-AP fusion binder and

addition of substrate. The AP enzyme converts the p-Nitrophenyl Phosphate substrate

to p-nitrophenol, a chromogenic product. (B) Sandwich ELISA detecting COPS5 over a

range of concentrations with FN3 fusion reagents. “Cognate pair” indicates both FN3

binders originated from a tandem clone isolated against COPS5, while “unrelated

capture reagent” indicates use of a capture FN3 to an unrelated target. Error bars denote

standard deviation of triplicate measurements. Data analyzed with OriginLab software.

Additionally, both assays were able to detect the analyte at concentrations as low as

5-20 pM (Table VII), which is physiologically relevant. Many proteins that circulate

throughout the blood exist at similar, picomolar to nanomolar concentrations such as C-

reactive protein (CRP), β2 microglobulin, and Cardiac troponin I [32]. Therefore, this

assay format has the potential to be applied to a variety of different plasma proteins.

93

TABLE VII. SANDWICH ELISA METRICS WITH MONOBODY FUSION PAIRS

Metrics AP Readout NanoLuc Readout

Limit of detection 20 pM 5 pM

Dynamic range 6 logs 6 logs

R2 0.974 0.982

4.4.3 Easy Use of Pairs for Heterogenous Assays To reduce assay development time once binding pairs are confirmed, we wanted to

implement a quick and easy system to conjugate the binding pairs to a variety of labels

for sandwich assays. Rather than cloning the FN3 coding regions into multiple different

expression vectors with different genetic fusions, and expressing and purifying each

monobody fusion separately, we decided to conjugate the tags enzymatically to purified

monobodies via sortase-mediated ligation. The Sortase A enzyme, naturally found in

gram positive bacteria for cell wall biosynthesis, catalyzes the conjugation of proteins

carrying an LPxTG motif, where x is any amino acid, by cleaving between threonine and

glycine and then joining threonine to a poly-glycine peptidoglycan through a trans-

peptide ligation reaction [33]. However, this reaction can be applied to site-specific

protein labeling in vitro through the joining of a protein with a pentapeptide tag, LPxTG,

to a protein or peptide carrying poly-glycine at its N-terminus [34].

To test the sortase-catalyzed protein ligation method, we first evaluated its efficiency

in tagging monobodies for a sandwich ELISA. We genetically fused each FN3 from the

tandem MegaSTAR clone to a LPxTG peptide, and ligated the tagged proteins to an N-

terminated-(glycine)5 biotin-conjugated peptide, to create a capture reagent, and an N-

terminated-(glycine)5 horseradish peroxidase (HRP) enzyme, to create a detection

reagent, via the sortase-catalyzed ligation reaction (Figure 4.7). Almost all of the

94

detection FN3 was converted to an HRP-fusion protein and almost all unlabeled FN3

was efficiently removed by centrifugal filtration (Figure 4.8A). The removal of

unconjugated FN3 detection reagent is extremely important to eliminate its competitive

binding for the analyte. Additionally, excess (G)5-biotin carrying peptide was efficiently

removed after tagging (Figure 4.8B), which is important for eliminating its competitive

binding to NeutrAvidin for immobilization.

Figure 4.7. Site-specific monobody labeling via sortase-mediated ligation. (A)

Tandem vector open reading frame. FN3a and FN3b denote the FN3 N- and C-terminal

to the linker in MegaSTAR constructs, respectively. (B) The Sortase A enzyme from

Gram-positive bacteria catalyze the trans-peptization of a poly-Glycine (G)5 tag to the

LPxTG sortase recognition sequence. FN3a is converted into the capture FN3 by

conjugating it to a biotin-labeled peptide while FN3b is converted into the detection FN3

by conjugating it to HRP. (C) FN3 ligation is reversed. Cartoon is not to scale.

95

Figure 4.8. Sortase-mediated ligation efficiency. (A). SDS-PAGE gel showing

unlabeled FN3-LPxTG (~27 kDa) conversion to FN3-HRP (~67 kDa) fusion via sortase-

mediated ligation. Unconjugated HRP is approximately 40 kDa. (B) FN3-LPxTG

incubated with (G)5-Biotin peptide in the presence or absence of sortase enzyme and

passively fused to a microtiter plate. Biotin detected by HRP-conjugated Streptavidin.

The labeled monobody pairs were used in sandwich ELISAs where the FN3-(G)5-

Biotin peptide conjugate was captured on NeutrAvidin coated plates to then capture

COPS5. The FN3b-(G)5-HRP conjugate was then added to the complex to detect

COPS5 at the second epitope (Figure 4.9AC). Both FN3 orientations were tested to

determine if there was any difference in assay performance (Figure 4.9BD) The two

ELISA formats varied slightly in dynamic range and limit of detection (Figure 4.9E),

suggesting that both orientations should be tested to identify the best format for a robust

assay.

Overall, the sandwich assays relying on monobody labeled binding pairs via sortase-

mediated ligation were very comparable to the sandwich ELISAs with genetically fused

monobody pairs. This opens the door to quickly testing monobody pairs with many other

labels, such as IR dyes, oligonucleotides, NanoLuc, and alkaline phosphotase to further

improve assay metrics or provide researchers with their desired measuring system.

96

Figure 4.9. Sandwich ELISA with binding pairs labeled by sortase-mediated

ligation. (A and C) Set up of the sandwich ELISA with FN3 reagents labeled by sortase-

mediated ligation. Plates are coated with NeutrAvidin, which complexes with biotinylated,

capture FN3. The plate is washed and then incubated with the analyte, COPS5, which

has been spiked at various amounts in bacteria cell lysates. The analyte is detected with

a second FN3, conjugated to HRP, and addition of the ABTS substrate. (B and D)

Sandwich ELISA detecting COPS5 over a range of concentrations with capture reagent

[FN3-(G)5-Biotin] and the detection reagent [FN3-(G)5)-HRP]. Error bars denote standard

deviation of triplicate measurements. (E) Performance metrics for set up 1 and 2.

97

4.5 Conclusions

MegaSTAR is a useful and innovative technique to quickly and inexpensively

generate tandem binding reagents with potential use as sandwich assay binding pairs.

By separating and genetically fusing each binding reagent to a tag for capture and

detection, we have further streamlined this process (Figure 4.10) through improving the

sandwich assay format to validate the binding pairs. This new format promotes testing

many possible pairs simultaneously to achieve an assay with a broad dynamic range

and low limit of detection in complex mixtures. Furthermore, we have introduced site-

specific, sortase-mediated enzymatic labeling to easily prepare high performance

binding pairs for a variety of heterogeneous assays. Overall, the process reduces the

time required to isolate and validate pairs, improves performance of pairs, and offers a

user-friendly assay system.

Figure 4.10. Workflow for identifying and characterizing binding reagents through

MegaSTAR. An estimated timeline is shown for a specific number of targets and their

corresponding pairs.

98


1. Zini, R., Riant, P., Barré, J. & Tillement, J.-P. (1990) Disease-Induced Variations in Plasma Protein Levels Implications for Drug Dosage Regimens (Part I), Clinical Pharmacokinetics. 19, 147-159.

2. Rifai, N., Gillette, M. A. & Carr, S. A. (2006) Protein biomarker discovery and validation: the long and uncertain path to clinical utility, Nature Biotechnology. 24, 971-983.

3. Fu, Y., Wang, N., Yang, A., Law, H. K.-w., Li, L. & Yan, F. (2017) Highly Sensitive Detection of Protein Biomarkers with Organic Electrochemical Transistors, Advanced Materials. 29, 1703787.

4. Cole, L. A. (2010) Biological functions of hCG and hCG-related molecules, Reprod Biol Endocrinol. 8, 102-102.

5. Gnoth, C. & Johnson, S. (2014) Strips of Hope: Accuracy of Home Pregnancy Tests and New Developments, Geburtshilfe Frauenheilkd. 74, 661-669.

6. Kierny, M. R., Cunningham, T. D. & Kay, B. K. (2012) Detection of biomarkers using recombinant antibodies coupled to nanostructured platforms, Nano Rev. 3, 10.3402/nano.v3i0.17240.

7. Baird, C. L., Fischer, C. J., Pefaur, N. B., Miller, K. D., Kagan, J., Srivastava, S. & Rodland, K. D. (2010) Developing recombinant antibodies for biomarker detection, Cancer Biomarkers. 6, 271-279.

8. Agnibh, M., Uzair, A., Martin, B., Ibrahim, A. & Michael, B. (2017) Clinically Relevant Biomarkers in Acute Heart Failure: An Update, Current Pharmaceutical Biotechnology. 18, 482-490.

9. Ji, M., Cho, B., Cho, Y. S., Park, S.-Y., Cho, S.-N., Jeon, B.-Y. & Yoon, B.-S. (2014) Development of a quantitative sandwich enzyme-linked immunosorbent assay for detecting the MPT64 antigen of Mycobacterium tuberculosis, Yonsei Med J. 55, 746-752.

10. Arai, H., Petchclai, B., Khupulsup, K., Kurimura, T. & Takeda, K. (1999) Evaluation of a rapid immunochromatographic test for detection of antibodies to human immunodeficiency virus, J Clin Microbiol. 37, 367-370.

99

11. Zhu, J., Zou, N., Mao, H., Wang, P., Zhu, D., Ji, H., Cong, H., Sun, C., Wang, H., Zhang, F., Qian, J., Jin, Q. & Zhao, J. (2013) Evaluation of a modified lateral flow immunoassay for detection of high-sensitivity cardiac troponin I andmyoglobin, Biosensors and Bioelectronics. 42, 522-525.

12. Lequin, R. M. (2005) Enzyme Immunoassay (EIA)/Enzyme-Linked Immunosorbent Assay (ELISA), Clinical Chemistry. 51, 2415-2418.

13. Reen, D. J. (1994) Enzyme-Linked Immunosorbent Assay (ELISA) in Basic Protein and Peptide Protocols (Walker, J. M., ed) pp. 461-466, Humana Press, Totowa, NJ.

14. Kragstrup, T. W., Vorup-Jensen, T., Deleuran, B. & Hvid, M. (2013) A simple set of validation steps identifies and removes false results in a sandwich enzyme-linked immunosorbent assay caused by anti-animal IgG antibodies in plasma from arthritis patients, Springerplus. 2, 263-263.

15. Kohl, T. O. & Ascoli, C. A. (2017) Immunometric Double-Antibody Sandwich Enzyme-Linked Immunosorbent Assay, Cold Spring Harbor Protocols. 2017, pdb.prot093724.

16. Ayoub, M. A., Trebaux, J., Vallaghe, J., Charrier-Savournin, F., Al-Hosaini, K., Gonzalez Moya, A., Pin, J.-P., Pfleger, K. D. G. & Trinquet, E. (2014) Homogeneous time-resolved fluorescence-based assay to monitor extracellular signal-regulated kinase signaling in a high-throughput format, Front Endocrinol (Lausanne). 5, 94-94.

17. Li, C., Tao, Y., Yang, Y., Feng, C., Xiang, Y. & Li, G. (2017) Dynamic sandwich-type electrochemical assay for protein quantification and protein–protein interaction, Analyst. 142, 4399-4404.

18. Gorman, K. T., Roby, L. C., Giuffre, A., Huang, R. & Kay, B. K. (2017) Tandem phage-display for the identification of non-overlapping binding pairs of recombinant affinity reagents, Nucleic acids research. 45, e158.

19. Kunkel, T. A. (1985) Rapid and efficient site-specific mutagenesis without phenotypic selection, Proceedings of the National Academy of Sciences of the United States of America. 82, 488-492.

20. Porath, J. (1992) Immobilized metal ion affinity chromatography, Protein Expression and Purification. 3, 263-281.

100

21. Panavas, T., Sanders, C. & Butt, T. R. (2009) SUMO Fusion Technology for Enhanced Protein Production in Prokaryotic and Eukaryotic Expression Systems in SUMO Protocols (Ulrich, H. D., ed) pp. 303-317, Humana Press, Totowa, NJ.

22. England, C. G., Ehlerding, E. B. & Cai, W. (2016) NanoLuc: A Small Luciferase Is Brightening Up the Field of Bioluminescence, Bioconjug Chem. 27, 1175-1187.

23. Pershad, K., Sullivan, M. A. & Kay, B. K. (2011) Drop-out phagemid vector for switching from phage displayed affinity reagents to expression formats, Analytical biochemistry. 412, 210-6.

24. Schierle, C. F., Berkmen, M., Huber, D., Kumamoto, C., Boyd, D. & Beckwith, J. (2003) The DsbA signal sequence directs efficient, cotranslational export of passenger proteins to the Escherichia coli periplasm via the signal recognition particle pathway, J Bacteriol. 185, 5706-5713.

25. Ashraf, S. S., Benson, R. E., Payne, E. S., Halbleib, C. M. & Grøn, H. (2004) A novel multi-affinity tag system to produce high levels of soluble and biotinylated proteins in Escherichia coli, Protein Expression and Purification. 33, 238-245.


27. Gorman, K., McGinnis, J. & Kay, B. (2018) Generating FN3-Based Affinity Reagents Through Phage Display, Current Protocols in Chemical Biology. 10, e39.

28. Neumann, H. & Van Vreedendaal, M. (1967) An improved alkaline phosphatase determination with p-nitrophenyl phosphate, Clinica Chimica Acta. 17, 183-187.

29. Ebai, T., Souza de Oliveira, F. M., Löf, L., Wik, L., Schweiger, C., Larsson, A., Keilholtz, U., Haybaeck, J., Landegren, U. & Kamali-Moghaddam, M. (2017) Analytically Sensitive Protein Detection in Microtiter Plates by Proximity Ligation with Rolling Circle Amplification, Clinical Chemistry. 63, 1497-1505.

30. Vashist, S. K., Schneider, E. M. & Luong, J. H. T. (2015) Rapid sandwich ELISA-based in vitro diagnostic procedure for the highly-sensitive detection of human fetuin A, Biosensors and Bioelectronics. 67, 73-78.

101

31. Tang, X., Liang, Q., Liu, L., Sheng, X., Xing, J. & Zhan, W. (2018) An optimized double-antibody sandwich ELISA for quantitative detection of WSSV in artificially infected crayfish, Journal of Virological Methods. 251, 133-138.

32. Hortin, G. L. & Sviridov, D. (2010) The dynamic range problem in the analysis of the plasma proteome, Journal of Proteomics. 73, 629-636.

33. Mazmanian, S. K., Liu, G., Ton-That, H. & Schneewind, O. (1999) Staphylococcus aureus Sortase, an Enzyme that Anchors Surface Proteins to the Cell Wall, Science. 285, 760-763.

34. Mao, H., Hart, S. A., Schink, A. & Pollok, B. A. (2004) Sortase-Mediated Protein Ligation: A New Method for Protein Engineering, Journal of the American Chemical Society. 126, 2670-2671.

102

Chapter 5 5. CONCLUSIONS

5.1 Thesis Summary

The central goals of this thesis were to develop recombinant affinity reagents to

challenging targets through phage display and demonstrate their utility in a biochemical

assay. These goals were achieved through two projects. The first was to generate affinity

reagents that had high selectivity and affinity for phosphothreonine peptide targets using

a phage library developed by our lab’s former member, Dr. Kritika Pershad. The second

was to format and utilize pairs of fibronectin type III (FN3) monobodies, isolated through

a new technique, Megaprimer Shuffling for Tandem Reagents (MegaSTAR), in sandwich

assays.

In Chapter 2, I isolated FHA affinity reagents that had unique specificities for the

Activating Transcription Factor 2 (ATF2) phosphopeptide targets. ATF2 is a challenging

target because it contains two neighboring phosphothreonine residues and can exist as

an unphosphorylated, mono-phosphorylated, or di-phosphorylated protein in cells.

Because ATF2 has multiple phosphorylation states that are biologically relevant, I was

motivated to generate Forkhead Associated (FHA) domain variants that would distinguish

between the fully and partially phosphorylated states of ATF2 peptides. I successfully

isolated affinity reagents that selectively recognized the di-phosphorylated peptide and

one of the two mono-phosphorylated forms of the peptide. Additionally, I showed that an

FHA binding reagent could detect full-length, di-phosphorylated ATF2 in a western blot.

In another aspect of this work, reported in Chapter 3, I generated FHA affinity reagents

to a mono-phosphothreonine peptide of human Akt1, an oncogenic protein that is often

phosphorylated in cancer patients. These affinity reagents are noteworthy because of

103

their low nanomolar affinity for the Akt1 phosphopeptide. They also selectively recognize

the Akt1 phosphopeptide, with little to no binding to sequences in the highly conserved

isoforms, Akt2 and Akt3.

Given the desirable characteristics the binding reagents discussed in Chapters 2 and

3 provide, we found it important to investigate their molecular recognition properties

through alanine scanning and DNA sequencing. As more FHA affinity reagents are

generated to phosphopeptides in the future, the compilation of what works and what does

not as peptide ligands will help elucidate the recognition range of the engineered FHA1

domain.

Many diagnostic assays detect molecules or proteins in blood or urine samples.

Because analytes typically remain folded in such mixtures, I focused in Chapter 4 on FN3

affinity reagents that recognize conformational epitopes of protein targets. Specifically,

this chapter concentrated on developing FN3 affinity reagents that work as pairs in

sandwich assays. The sandwich assay format is ideal because it requires two separate

binding events to occur to produce a signal, which significantly limits false positive results.

FN3 pairs were isolated through MegaSTAR, a technique developed by a previous

member of our lab, Dr. Kevin Gorman, that modifies the phage-display process to

generate tandem binding reagents with potential use in sandwich assays [1]. I streamlined

the validation process for these binding pairs. In this new process, FN3 binders are

expressed at small scale and used directly in crude cell lysates, eliminating the need to

express and purify large quantities of the binding reagents. By greatly reducing the time,

cost, and labor of validating pairs, more pairs can be tested.

104

Often, high quality binding reagents are not used to their full capacity because they

must be reformatted to work in each assay. To remove this bottleneck, I genetically fused

FN3 binders to the pentapeptide motif, LPxTG, allowing researchers to quickly conjugate

labels of their choice to the binding reagents through enzymatic, sortase-mediated

ligation. This simple labeling method makes these binding reagents accessible for a

variety of assays.

5.2 Future Directions

5.2.1 Incorporate Multiple Scaffolds and Linker Lengths into MegaSTAR

MegaSTAR has shown promising results in generating binding pairs of monobodies

that can bind targets simultaneously. In its original version, pairs that were generated

recognized conformational epitopes. To expand on early efforts with MegaSTAR, we

propose to generate a suite of tandem vectors that incorporate other scaffolds and binding

reagents, such as the FHA domain, a designed ankyrin repeat protein (DARPin), and an

scFv (Figure 5.1). These new vectors would be constructed through standard cloning

methods.

First, we would perform primary selection on the target with each of the individual

libraries. Because of the variety of binding capabilities associated with each scaffold, we

can pursue a range of difficult targets, such as post translational modifications and

membrane receptor proteins. The DNA from the output clones would be PCR amplified

and annealed to the appropriate tandem vectors. After synthesizing the heteroduplex

DNA and transforming into bacterial cells, the clones would be pooled together to create

a large secondary library with tandem scaffolds. This pool would be amplified for a

combined selection, in which the best possible pairs would be isolated.

105

Figure 5.1. An array of tandem vectors for improved binding pairs. Schematic

showing (A) the tandem vector coding region and corresponding phage particle and (B)

possible tandem scaffold combinations that could be used in MegaSTAR. Not to scale.

To improve the success rate for generating binding pairs to difficult targets, we

propose to also incorporate an array of linkers between the tandem binders. Currently,

the tandem vector contains a 25 amino acid linker with the repeated GGGGS motif. If the

epitopes that two binding reagents recognize are far apart, this short linker may hinder

their ability to bind the target simultaneously. We would generate two additional tandem

vectors that incorporate 50 amino acid and 75 amino acid linkers to eliminate this issue.

Additionally, we anticipate difficulties with expressing clones that contain a long, highly

repetitive GGGGS linker due to the inherent instability of repetitive sequences in bacteria

cells. The clones may incur mutations caused by recombination or replication slippage

[2]. We propose to test alternative linkers such as the 95 amino acid disordered region of

Calcineurin [3].

106

5.2.2 Generating Reagents to Cell-Free Expressed Targets

Many eukaryotic proteins require chaperones for proper folding in the cells. Such

protein chaperons are absent in bacteria, which necessitates overexpression of human

proteins in mammalian cell lines. This is laborious, expensive, and requires technical

competency. As an alternative, we propose to use cell-free protein synthesis to generate

these target proteins in vitro and use them directly in phage-display affinity selection

experiments. HaloTagged fusion proteins would be generated through coupled in vitro

transcription-translation in HeLa cell extracts [4]. HaloTag is a modified dehalogenase

that forms a covalent bond with suicide substrates. We would capture the HaloTagged

targets directly from the protein synthesis reaction by covalently conjugating them to resin

or magnetic beads coated with suicide substrate for HaloTag (Figure 5.2) [5]. Because

no purification step is required, the production of correctly folded target proteins can be

miniaturized. Additionally, the targets would be prepared fresh before each selection,

eliminating harmful freeze-thaw cycles.

Figure 5.2. Target immobilization via HaloTag technology. Cartoon showing nucleophilic

displacement of the resin’s terminal chloride by an aspartic acid residue on HaloTag leading to a

covalent, irreversible alkyl-enzyme intermediate. Not to scale.

107

‘

We have shown that the FN3 binding reagents can be enzymatically labeled by site-

specific sortase-mediated ligation and utilized in sandwich ELISAs. Additionally, these

binding pairs can be applied to homogeneous assays to further improve sensitivity and

robustness while greatly decreasing assay time, as no washing steps are required. I

describe two popular homogenous assays below.

The proximity ligation assay (PLA) is unmatched in sensitivity and is capable of

detecting zeptomole levels of an analyte [6]. Two binding reagents are conjugated to DNA

oligonucleotides, and when they are brought into proximity to each other (due to binding

same analyte or two analytes interacting with each other) and the required substrates and

enzymes are added, the DNA oligonucleotides will anneal, be extended by DNA

polymerase, and permit rolling circle DNA amplification. Fluorescent, complementary

oligonucleotides are annealed to the DNA once it is amplified, resulting in a robust

fluorescent signal.

DNA can be randomly conjugated to binding reagents quite easily [7], But it is

imperative that oligonucleotides conjugated to the FN3 binders do not interfere with the

FN3’s binding mechanism. Because sortase-mediated ligation promotes site-specific

labeling, we propose incorporating this labeling technique to tag FN3 binding reagents

with peptide-DNA oligonucleotide conjugates and perform proximity ligation assays to

detect analytes that exist at extremely low concentrations in human samples (Figure

5.3A).

Another popular homogenous assay format is the bioluminescence resonance energy

transfer (BRET) assay (Figure 5.3B) [8]. The assay requires that one of the FN3 binding

108

Figure 5.3. Homogeneous assays utilizing sortase-mediated ligation. Overview of

homogeneous assays. FN3 binders are labeled by sortase-mediated ligation with (A)

oligonucleotides and used in a proximity ligation assay or (B) NanoLuc and the 618 ligand

(via HaloTag) and used in a BRET assay. Not to scale.

109

reagents be labeled with the NanoLuc enzyme (the energy donor) and the other be

labeled with a ligand that carries 618 nanometer wavelength emission Alexa dye (618

Ligand; the energy acceptor). The 618 ligand is formatted to covalently bond to the

HaloTag enzyme [9]. We would express and purify NanoLuc and HaloTag with N-terminal

SUMO fusions followed by a poly-glycine tag. Cleavage by site-specific SUMO protease

will yield NanoLuc and HaloTag proteins with N-terminal poly-glycine tags (Figure 5.4).

These glycine-terminated enzymes can then be ligated to FN3 monobodies by sortase-

mediated ligation. The FN3-HaloTag binder would further be altered to also be

enzymatically labeled with the 618 ligand through HaloTag enzymatic conjugation (as

discussed in section 5.2.2). Like PLA, the two binders are brought into proximity by

binding the same analyte. Upon adding the NanoLuc substrate, NanoLuc emits a signal

at 460 nM which excites the Alexa dye conjugated to the ligand, and in turn, emits a signal

at 618 nM that can be measured with a luminometer [8].

Figure 5.4. Generating recombinant enzyme labels for sortase-mediated ligation.

Cartoon illustrating the removal of a SUMO tag by site-specific cleavage to produce

recombinant NanoLuc and HaloTag enzymes with N-terminal poly-glycine tags. Not to

scale.

110

5.3 Overall Impact

While biomarker discovery has exploded in recent years, it has not translated to the

production of many sensitive and specific diagnostic and biochemical tests. This is in part

due to the technical challenge of producing robust assays with low false positive and low

false negative results [10]. Antibody binding reagents needed for these assays are difficult

to obtain as animal immunizations, antibody collection and purification, and validation

require extensive time and labor and often still result in non-specific, off-target, and/or low

affinity binders. In this work we demonstrate that by deploying phage display and

MegaSTAR to generate recombinant affinity reagents, we can reduce the time needed to

select binding pairs and improve their specificity and affinity through directed evolution,

resulting in robust, sensitive sandwich assays.

Recombinant affinity reagent generation has demonstrated success in helping

quantify the amount of an analyte present in a complex mixture. However, this is only the

beginning of what these improvements can provide. By incorporating many scaffolds into

the selection and MegaSTAR process, they will be helpful in developing a variety of other

diagnostic and biochemical tests. Not only is it useful to determine the absence or

presence of a molecule, but detecting biomarkers such as protein mutants, changes in a

protein’s folding through allosteric interactions, or post-translational modifications would

be useful in quickly diagnosing many diseases. For example, the RAS GTPase is part of

a signaling cascade and is activated by binding GTP, which ultimately leads to increased

expression of genes involved in cell growth and proliferation. Upon hydrolysis of GTP to

GDP, RAS is inactivated and the signaling pathway is turned off. However, RAS mutants

that are constitutively active are commonly observed in cancer patients [11, 12]. Through

111

use of many scaffold libraries, it is possible to distinguish between GTP and GDP bound

RAS by detecting conformational changes, and between wild-type and mutant RAS by

detecting sequence changes. These reagents could then be used in a multiplex sandwich

assay that diagnoses a patient with RAS driven cancer.

As genomic and biomarker testing becomes more reliable and routine, diagnostics will

radically reshape the healthcare landscape. Cancer, heart disease, and other disorders

will be detected in their early stages which will both reduce cost of treatment and save

more lives. Additionally, diagnostics provide personalized healthcare. Biomarkers and

gene mutations offer information about how a patient will respond to a therapeutic,

allowing doctors to tailor treatments to each patient. My work has the potential to improve

preventative and personalized health care, and ultimately enhance the lives of many.

112

5.7 Literature cited

1. Gorman, K. T., Roby, L. C., Giuffre, A., Huang, R. & Kay, B. K. (2017) Tandem phage-display for the identification of non-overlapping binding pairs of recombinant affinity reagents, Nucleic acids research. 45, e158.

2. Bzymek, M. & Lovett, S. T. (2001) Instability of repetitive DNA sequences: The role of replication in multiple mechanisms, Proceedings of the National Academy of Sciences. 98, 8319-8325.

3. Creamer, T. P. (2013) Transient disorder: Calcineurin as an example, Intrinsically Disord Proteins. 1, e26412-e26412.

4. Gagoski, D., Polinkovsky, M. E., Mureev, S., Kunert, A., Johnston, W., Gambin, Y. & Alexandrov, K. (2016) Performance benchmarking of four cell-free protein expression systems, Biotechnology and Bioengineering. 113, 292-300.

5. Ohana, R. F., Hurst, R., Vidugiriene, J., Slater, M. R., Wood, K. V. & Urh, M. (2011) HaloTag-based purification of functional human kinases from mammalian cells, Protein Expression and Purification. 76, 154-164.

6. Fredriksson, S., Gullberg, M., Jarvius, J., Olsson, C., Pietras, K., Gústafsdóttir, S. M., Östman, A. & Landegren, U. (2002) Protein detection using proximity-dependent DNA ligation assays, Nature Biotechnology. 20, 473-477.

7. Gullberg, M., Gústafsdóttir, S. M., Schallmeiner, E., Jarvius, J., Bjarnegård, M., Betsholtz, C., Landegren, U. & Fredriksson, S. (2004) Cytokine detection by antibody-based proximity ligation, Proceedings of the National Academy of Sciences of the United States of America. 101, 8420-8424.

8. Machleidt, T., Woodroofe, C. C., Schwinn, M. K., Méndez, J., Robers, M. B., Zimmerman, K., Otto, P., Daniels, D. L., Kirkland, T. A. & Wood, K. V. (2015) NanoBRET—A Novel BRET Platform for the Analysis of Protein–Protein Interactions, ACS chemical biology. 10, 1797-1804.

9. Benink, H. A. & Urh, M. (2015) HaloTag Technology for Specific and Covalent Labeling of Fusion Proteins in Site-Specific Protein Labeling: Methods and Protocols (Gautier, A. & Hinner, M. J., eds) pp. 119-128, Springer New York, New York, NY.

113

10. Frangogiannis, N. G. (2012) Biomarkers: hopes and challenges in the path from discovery to clinical practice, Transl Res. 159, 197-204.

11. Zhou, B., Der, C. J. & Cox, A. D. (2016) The role of wild type RAS isoforms in cancer, Semin Cell Dev Biol. 58, 60-69.

12. Zeitouni, D., Pylayeva-Gupta, Y., Der, C. J. & Bryant, K. L. (2016) KRAS Mutant Pancreatic Cancer: No Lone Path to an Effective Treatment, Cancers (Basel). 8, 45.

114

Chapter 6 APPENDIX

GLOBAL JOURNAL PERMISSIONS TO REUSE PREVIOUSLY PUBLISHED

MATERIALS

Permission for reuse of article, Generating FN3-based Affinity Reagents through Phage

Display, originally published by Current Protocols in Chemical Biology (Wiley publication)

Copyright policy

The Article, including all figures, illustrations, and tabular and other supplementary material, shall be considered a work made for hire for Current Protocols, and we shall own the copyright and all of the rights comprised in the copyright. If the Article or any such material does not qualify as a work made for hire, then you hereby transfer to us during the full term of the copyright and all extensions thereof the full and exclusive rights comprised in the copyright and all other rights in and to the Article, and in any such material, in all media, worldwide.

115

Permission for reuse of article, Generation of Recombinant Affinity Reagents against a

Two-phosphosite epitope of ATF2, originally published by New Biotechnology

116

The article, A Recombinant Affinity Reagent Specific for a Phosphoepitope of Akt1,

originally published in International journal of molecular sciences is an open access article

distributed under the Creative Commons Attribution License which permits unrestricted

use, distribution, and reproduction in any medium, provided the original work is properly

cited (CC BY 4.0).

117

Chapter 7 VITA

NAME: Jennifer Elise McGinnis

EDUCATION: Ph.D., Biological Sciences, University of Illinois at Chicago, Chicago,

IL, 2019

B.S., Molecular and Cellular Biology, University of Illinois at Urbana-

Champaign, Urbana, IL, 2014

TEACHING Department of Biological Sciences, University of Illinois at Chicago,

Chicago, Illinois: Genetics Laboratory, Teaching Assistant, 2014-

2017

RESEARCH Department of Biological Sciences, University of Illinois at Chicago,

Chicago IL: Laboratory of Dr. Brian Kay, Graduate Research

Assistant, 2014-2019

Downstream Process Development, Bristol-Myers Squibb, Devens,

Massachusetts: Co-Op, 2018.

Department of Molecular and Cellular Biology, University of Illinois at

Urbana-Champaign, Urbana, IL: Laboratory of Dr. Isaac Cann,

Undergraduate Research Assistant, 2012-2014.

POSTERS Jennifer E. McGinnis, Kevin T. Gorman, and Brian K. Kay. (2019)

Direct Recovery of Pairs by Phage Display. Affinity 2019. Stockholm,

Sweden.

Jennifer E. McGinnis and Brian K. Kay. (2016) Generation of an

Affinity Reagent that Recognizes a Dually Phosphorylated Epitope in

a Human Transcription Factor. Gordon Research Conference:

Chemistry and Biology of Peptides. Ventura, CA.

PUBLICATIONS McGinnis, J. E., Venegas, L. A., Lopez, H. & Kay, B. K. (2018) A

Recombinant Affinity Reagent Specific for a Phosphoepitope of Akt1,

Int J Mol Sci. 19, 3305.

Gorman, K. T., McGinnis, J. E. & Kay, B. K. (2018) Generating FN3-Based Affinity Reagents Through Phage Display, Current Protocols in Chemical Biology. 10, e39.

Kokoszka, M. E., Kall, S. L., Khosla, S., McGinnis, J. E., Lavie, A. &

Kay, B. K. (2018) Identification of two distinct peptide-binding

118

pockets in the SH3 domain of human mixed-lineage kinase 3, The

Journal of biological chemistry. 293, 13553-13565.

McGinnis, J. E. & Kay, B. K. (2017) Generation of recombinant

affinity reagents against a two-phosphosite epitope of ATF2, New

biotechnology.

AWARDS Department of Biological Sciences, University of Illinois at Chicago,

Chicago, Chicago, IL: Excellence in Teaching, 2017.

Affinity 2019 Conference, Stockholm, Sweden: Affinity Poster Award

(2nd place), 2019

Generation of Affinity Reagents to Challenging - UIC Indigo

Documents

Transcript of Generation of Affinity Reagents to Challenging - UIC Indigo