Theoretical Model of the Three-dimensional Structure of a Disease Resistance Gene Homolog Encoding

Resistance Protein in Vigna mungo

http://www.jbsdonline.com

Abstract

Plant disease resistance (R) genes, the key players of innate immunity system in plants en-code ‘R’ proteins. ‘R’ protein recognizes product of avirulance gene from the pathogen and activate downstream signaling responses leading to disease resistance. No three dimensional (3D) structural information of any ‘R’ proteins is available as yet. We have reported a ‘R’ gene homolog, the ‘VMYR1’, encoding ‘R’ protein in Vigna mungo. Here, we describe the homology modeling of the ‘VMYR1’ protein. The model was created by using the 3D struc-ture of an ATP-binding cassette transporter protein from Vibrio cholerae as a template. The strategy for homology modeling was based on the high structural conservation in the super-family of P-loop containing nucleoside triphosphate hydrolase in which target and template proteins belong. This is the first report of theoretical model structure of any ‘R’ proteins.

Key words: Vigna mungo; Resistance genes; ‘R’ protein; ABC transporter; Homodimer; and Homology modeling.

Introduction

Plants have an innate immune system based on dominant resistance (R) genes in which the disease resistance is elicited by the products of avirulance (avr) genes from the pathogen (1). It is generally considered that the ‘R’ proteins mediate elici-tor recognition and activate downstream signaling responses leading to disease re-sistance (2). These responses are frequently associated with a type of programmed cell death termed the hypersensitive response.

Most plant ‘R’ genes known to date encode proteins with a central nucleotide-bind-ing domain (NBD) and a C-terminal Leucine-rich repeat (LRR) domain (3, 4). The NBD contains three ATP/GTP binding motifs known as the Kinase-1a or P-Loop, Kinase-2, and Kinase-3 motifs. It has already been reported that the NBD of ‘R’ proteins form a functional nucleotide binding site and specifically binds ATP rather than other nucleoside triphosphates (5). NBD-LRR proteins can be differentiated into two subclasses on the basis of their amino-terminal sequence. One contains the TIR domain with homology to Toll and interleukin receptor-like genes (2), and the other subclass contains a putative coiled coil (CC) structure (6). These NBD-LRR structures form homo- or hetero-oligomeric associations facilitating interactions between proteins and possibly impart a role in the interaction of ‘R’ proteins with molecules downstream in the signal transduction pathway (7).

Basak et al., (8) have recently reported a Yellow Mosaic Virus resistance linked ‘R’ gene homolog (‘VMYR1’, NCBI accession number AY297425) from Vigna mungo. The 445bp nucleotide sequence of the ‘VMYR1’ represents an ORF, has high homol-ogy with other plant ‘R’ genes and also with soybean expressed sequence tagged

Journal of Biomolecular Structure &Dynamics, ISSN 0739-1102Volume 24, Issue Number 2, (2006)©Adenine Press (2006)

Jolly Basak1,§,*

Ranjit P. Bahadur2,†,*

1Plant Molecular and Cellular Genetics SectionBose InstituteP-1/12, CIT Scheme VII MKolkata 700054. India.2Department of BiochemistryBose InstituteP-1/12, CIT Scheme VII MKolkata 700054. India

§Present address:Ecologie Systématique EvolutionUMR 8079, Bât. 360 & 362Université de Paris-Sud91405 Orsay cedex, France.†Present address:Institut de Biochimie et de Biophysique Moleculaire et CellulaireUMR 8619, Bât. 430Université de Paris-Sud91405 Orsay Cedex, France

123

*Phone: +33 1 69 15 79 66Fax: +33 1 69 85 37 15Email: ranjit.bahadur@ibbmc.u-psud.frEmail: jolly.basak@ese.u-psud.fr

124

Basak and Bahadur

sites (ESTs). The translated 148 amino acid sequence of the ‘VMYR1’ (NCBI ac-cession number AAQ63168) showed high sequence similarity with other plant ‘R’ proteins. The knowledge on the structural features of ‘R’ proteins is necessary to understand the mechanism of R-Avr interaction and also to predict pathogen speci-ficity of a given ‘R’ protein. Inspite of this imminent need, attention is lacking and there is no three dimensional (3D) structure of plant ‘R’ proteins available in the Protein Data Bank (9). Therefore, it is prerequisite to know the 3D structure of the ‘VMYR1’, a protein conferring YMV-resistance, to understand the structure func-tion relationship for the YMV-reaction in V. mungo.

Absence of any known 3D structure of plant ‘R’ proteins impelled us to build a 3D structure of the ‘VMYR1’ using homology modeling. The strategy undertaken for homology modeling was to identify a protein template that has not only sequence similarity with the ‘VMYR1’ but also has functional analogy. The ATP-binding cas-sette (ABC) transporter proteins are one such class of proteins found in the PDB, which fulfills the required criteria. The ABC-transporters possesses four conserved domains, two of which are transmembrane domains (TMD), while the other two are highly conserved nuclear binding domains (NBD) (10). Higgins and Linton (10) have documented regions responsible for ATP-binding within these two NBDs of ABC transporters. They have postulated that the energy generated through the hy-drolysis of ATP, subsequent to the binding, is utilized in drug and/or substrate trans-portation. Likewise, the homologous nuclear binding sites of ‘R’ proteins utilize the energy (generated by ATP binding and hydrolysis) and activate down-stream signaling cascades; conferring plant disease resistance. The supporting evidence of the ATP-binding, coupled with hydrolysis and energy release, has already been demonstrated in ‘R’ gene product of tomato (5). This functional homology between ABC transporters and ‘R’ proteins prompted us to consider the ABC transporter protein of Vibrio cholerae (PDB code 1PF4) as a template for homology modeling. By integrating all information available on sequence, structure and key functions, we have proposed a theoretical model for the ‘VMYR1’ during this investigation.

Materials and Methods

The nucleotide sequence of the ‘VMYR1’ was translated in silico using the transla-tion tool at the Swiss bioinformatics server (http://us.expasy.org/). The 148 amino acid sequence obtained thus far represents a part of the NBD containing the Kinase-2 and Kinase-3 motifs. The program BLAST (11) was employed for detecting sequence similarities between the ‘VMYR1’, other ‘R’ proteins, and 1PF4. Align-ment of all these sequences was carried out by ClustalW (12) with default param-eters. The secondary structure prediction of the ‘VMYR1’ was carried out using the PHDsec method in the PredictProtein server (http://www.predictprotein.org/). ‘Superfamily’ server (http://supfam.mrc-lmb.cam.ac.uk/) was employed to predict the superfamily in which the ‘VMYR1’ belongs.

The 3D model of the ‘VMYR1’ was constructed using the crystal structure coordi-nates of the 1PF4 obtained at a resolution of 3.8 Å (13). The NBD of 1PF4 starts at the residue number 344 to the end of the peptide. The 150 amino acids (residue num-ber 433-582 within the NBD domain) at the C-terminal of the 1PF4 were used as the template sequence, which has the sequence similarity and aligns with the ‘VMYR1’ in the ClustalW alignment method. Automated homology model building was per-formed using protein structure modeling program MODELLER (14), which model protein tertiary structure by satisfaction of spatial restraints. The input for MOD-ELLER consisted of the aligned sequences of IPF4 and the ‘VMYR1’, a steering file that gives all the necessary commands to the MODELLER to produce homology model of the target on the basis of its alignment with the template. Energy mini-mization was performed by the steepest descent followed by the conjugate gradient method using 20 Å non-bonded cutoff and a constant dielectric of 1.0. Evaluation of the predicted model involved analyses of the geometry and the stereochemistry of

125Theoretical Modeling

of a Plant Disease Resistance Protein

the model. The reliability of the model structure was tested using the ENERGY com-mands of the MODELLER. Program PROCHECK (15) has performed assessment of the predicted model of the ‘VMYR1’ and Ramachandran plots (16) were drawn.

The dimer of the predicted structure was generated by applying the crystal sym-metry of the template structure given in the PDB file. The dimeric structure was visualized on RASMOL 2.6 (17). Deviation of the modeled structure from the template was calculated by superposition of Cα traces and backbone atoms of the model onto the template and the RMSD values for positional differences between equivalent atoms were calculated.

The solvent accessible surface area (ASA) of the predicted dimer was computed using the program ACCESS (18) that is an implementation of the algorithm of Lee and Richards (19). The interface area between the two subunits was estimated as:

IA = ASA (subunit 1) + ASA (subunit 2) - ASA (dimer)

All atoms or amino acid residues in the monomer that lose more than 0.1Å2 ASA in the dimer were counted as interface atoms or residues (20, 21, 22). The contribu-tion of polar (N, O, S) and non-polar (all carbon containing groups) area on the total interface area was also estimated according to the method described by Bahadur et al. (22). Hydrogen bond interactions between dimeric subunits were detected using the program HBPLUS (23) with default parameters.

Surface electrostatic potential was calculated by the Poisson-Boltzmann equation as implemented in the program DELPHI (24) and running in the program GRASP (25). Default values for charge (full charge), salt concentration (0.0), interior di-electric (2.0), and exterior dielectric (80.0) were used.

Results and Discussion

Knowledge of the three-dimensional structure of a protein is a necessary prereq-uisite for understanding its molecular function. To create a 3D structure of the ‘VMYR1’, BLAST search was first performed in protein databases for proteins with similar sequence and known 3D structure. The ‘VMYR1’ has sequence simi-larity with more than 500 GenBank accessions in the NCBI non-redundant data-base and the highest sequence identity of 95% and E-value of 1e-62 was obtained with the disease resistance protein of Vigna radiata. All of these accessions were either plant ‘R’ proteins or partial ‘R’ protein sequences. However, 3D structure of none of these ‘R’ proteins is available in the PDB database. A PSI-BLAST search among the protein sequences available in the PDB database has revealed that the ABC transporter proteins have sequence similarity with the ‘VMYR1’, of which, the protein, Msba from Escherichia coli (PDB code 1JSQ) (26) showed maximum sequence similarity (identities 32%) with an E-value of 8e-16 (after the second iteration of PSI-BLAST search). However, this could not be used as a template for homology modeling as only the Cα coordinates of this protein structure is available in the PDB. Whereas, another ABC transporter protein (PDB code 1PF4) from V. cholerae, which has sequence identity 32% (E-value 8e-16, after the second iteration of PSI-BLAST search) with the target protein and existing crystal struc-ture, was considered as a template. The structure of 1PF4 is a close conformation whereas the structure of 1JSQ is an open conformation. Vibrio cholereae-MsbA (1PF4) shares 68% sequence identity with the MsbA of E. coli (1JSQ), suggesting that the two function by a common mechanism.

Characteristic features of the primary structures of ABC transporter protein include a set of motifs, most notably the Walker A, Walker B, ABC signature motif, and switch II. Of these motifs, the Walker A or P-loop, Walker B, and switch II resemble the Kinase-1a or P-Loop, Kinase-2, and Kinase-3 motifs of majority of the plant ‘R’

126

Basak and Bahadur

proteins, respectively, in which the ‘VMYR1’ belongs. This highlights the sequence identity between the conserved motifs of the target and template proteins.

Additionally, secondary structure of the ‘VMYR1’ as predicted by several second-ary structure prediction servers (PHDsec: www.predictprotein.org/; PSIPRED: http://bioinf.cs.ucl.ac.uk/psipred/) is quite similar to the secondary structure of 1PF4. The PHDsec program predict all the secondary structures according to the assigned repetitive structures in 1PF4 except 3 helices Met 111-Glu 119, Gln 240-Ala 247, and Gln 468-Gly 478. PSIPRED server also predicts all the secondary structures of the 1PF4 according to the assigned repetitive structure except one helix Gln468-Gly478. With regard to the ‘VMYR1’ both the method predicted the similar secondary structures. This finding supports results obtained from several fold-recognition servers (FUGUE: www-cryst.bioc.cam.ac.uk/~fugue/; 3DPSSM: www.sbg.bio.ic.ac.uk/) suggesting that the ‘VMYR1’ has the fold of P-loop contain-ing nucleoside triphosphate hydrolase superfamily like that of 1PF4. The region of 1PF4 used as the template contains six α-helices and three β-sheets, whereas, the predicted model structure of the ‘VMYR1’ contains six α-helices, one 310 helix, and two β-sheets (Fig. 1). Absence of one β-sheet (corresponding to β1 in 1PF4) is due to the inaccessibility of the amino acid residues at the N-terminal of the ‘VMYR1’ that are required for the formation of the β-sheet.

It is now well established that proteins belonging to the same superfamily have their folds well conserved even if their sequence identity is low (27). Thus, in presence of well-defined structural information, comparative modeling can suc-cessfully applied by assessing with care the sequence alignment, even when the sequence identity between the target and template proteins is lower than 40% (28). Based on this background information, we assume that the ‘VMYR1’ may be fold-ed like that of the 1PF4. Consequently, the ‘A’ subunit of the 1PF4 was used as a template to predict the 3D structure of the ‘VMYR1’ by applying the comparative modeling strategy as described in the Materials and Methods. The result obtained upon energy minimization showed no major changes between the predicted model structure and the minimized structure.

The 1PF4 template crystallized as a dimer and each of the subunit composed of a TMD fused with a NBD; and both the TMDs and NBDs make intramolecular con-tacts that stabilize the dimer. Accordingly, the homodimeric structure of the ‘VMYR1’ was created to obtain the stable form (Fig. 2). Beside the sequence based analysis we also analyzed different physicochemical properties of the homodimer interface of the ‘VMYR1’ from the 3D structure to justify the dimeric form of the predicted model.

Figure 1: Amino acid sequence alignment of the ‘VMYR1’ with the template 1PF4 and also with 1JSQ. Identical amino acids are highlighted in black and con-servative replacements of amino acids are highlighted in gray. Amino acids in italics were not modeled because of the absence of corresponding regions in the template structure. Secondary structural elements of both the template (black) and target (gray) are shown as helices (curved line) and sheets (arrow). Dotted regions in both the template and target are disordered in the structure.

127Theoretical Modeling

of a Plant Disease Resistance Protein

Figure 2: Schematic view of the three-dimensional struc-ture of the dimeric ‘VMYR1’. Subunits are colored in red and blue showing the N-terminal ends with arrows.

N-terminal

Figure 3: Representation of the molecular surface and electrostatic potential of the monomeric (A) and dimeric (B) structure of the ‘VMYR1’, with a color scale that varies from blue to red, representing positive and nega-tive potential, respectively. One subunit of 3B is col-ored according to the electrostatic potential, while other subunit is being imaged as a backbone tube.

A

B

Surface Potential -6.517 -3.309 0.000 3.300 6.601

Surface Potential -6.592 -3.296 0.000 3.373 6.745

128

Basak and Bahadur

Validation of the Model

PROCHECK analysis (Table I) on the stereochemical quality of the 3D model struc-ture of the ‘VMYR1’ revealed that 67.2% of residues are in the most favored region in the Ramachandran plot. Moreover, the percentages of residues in the additional allowed regions and generously allowed regions are 24.6% and 6.6%, respectively. However, 1.6% of residues (Leu46 and Leu124) remain on the disallowed region, of which, Leu46 is on the borderline position. Overall 3D structure of the target protein is not affected by these two leucine residues, as both of these are part of the loop region. PROCHECK analysis of the truncated 1PF4 containing region that has been used as template revealed that an isoleucine residue, Ile556 is on the disal-lowed region. Table I summarizes the fact that the 3D model of the ‘VMYR1’ and 1PF4 has nearly the same PROCHECK distribution. We superpose the Cα atoms of the template and the target protein to calculate the root mean square deviation between the model structure of the ‘VMYR1’ and the truncated 1PF4-containing region that has been used as template and the deviation was found to be 0.12 Å.

Features of the Dimeric Interface

The physicochemical features of the interface formed in the homodimer have been analyzed. In the homodimeric structure of the ‘VMYR1’, the two protein com-ponents contribute equally to the interface area. The total interface area of the ‘VMYR1’ dimer obtained is 1383 Å2 (Table II), which is nearly one fourth com-pared to the size of the total dimeric interface of 1PF4. However in 1PF4, the NBD (the region used as template) contributes around 1600 Å2 to the total dimeric interface area (13). This value is nearly equivalent in size to the dimeric interface of the ‘VMYR1’. The non-polar interface area (contributed by the C atoms only) is 855 and 3525 Å2 in the ‘VMYR1’ and in 1PF4, respectively, which is 62% and 64%, respectively, compared to their total interface area. Bahadur et al., (21) showed that the average value of the non-polar interface area in homodimers is around 65%. This indicates that the hydrophobic effect plays an important role in stabilizing both the dimeric interfaces of the ‘VMYR1’ and ‘1PF4’, which is the general features of the homodimer interfaces (20, 21).

The interface comprises of 148 atoms contributed by 30 interface residues in the ‘VMYR1’ dimer, whereas these values are 588 and 180, respectively, for the dimeric interface of 1PF4. The non-polar carbon atoms contribute 68% of the total number of interface atoms in the ‘VMYR1’, whereas for the 1PF4 it is 63%. Recently, Ba-hadur et al. (22) have demonstrated that an interface atom may be fully buried (zero ASA) after dimer formation, or it may remain partly accessible to the solvent mol-ecule. The fraction of fully buried atoms for the dimeric interface of the ‘VMYR1’ is 0.19, which is thrice larger compared to 1PF4 interface (0.06). The residue pro-pensity score, which is derived from the amino acid composition at the homodimer interface, is 2.6 for the ‘VMYR1’ interface, indicating that the structure should exist as a homodimer as it has already been reported that homodimer interfaces generally have a positive residue propensity score (22). For the 1PF4 interface, which is a large dimeric one, this value is 13.2.

Table I Statistical data of the 3D structure of the ‘VMYR1’ and 1PF4a.

Distribution of residues in the Ramachandran plotb ‘VMYR1’ 1PF4

Residues in most favoured regions 82 67.2 % 71% Residues in additional allowed regions 30 24.6% 22.6% Residues in generously allowed regions 8 6.6% 5.6%

%8.0 %6.1 2 snoiger dewollasid ni seudiseR %001 %001 231 latoT

aThe portion of 1PF4 used as template. bRamachandran plot statistics were calculated by PROCHECK.

Table II Properties of homodimer interface.

Interface parametersa VMYR1 1PF4

Interface area (A2) 1383.2 5509.5 Number of atoms 148 588 Number of residues 30 180 Fraction of non-polar atoms 0.68 0.63 Non-polar interface area (A2) 854.6 3524.9 Fraction of fully buried atoms 0.19 0.06 Residue propensity score 2.56 13.18 Number of hydrogen bonds 6 8

aCalculated after the method described by Bahadur et al.(22).

129Theoretical Modeling

of a Plant Disease Resistance Protein

The homodimer interface of the ‘VMYR1’ is stabilized by the three hydrogen bonds per subunit (total six at the interface due to the twofold symmetry), as detected by the program HBPLUS (23). The hydrogen bonds occur between the nitrogen atom ‘N’ of GLN-77 in A subunit and the oxygen atom of ARG-75 in B subunit, between the ‘NE2’ of GLN-77 in A subunit and the oxygen atom of ARG-73 in B subunit, and also between the nitrogen atom ‘NH2’ of ARG-84 in A subunit and the oxy-gen atom of LEU-104 in B subunit. The residues involved in the interface of the ‘VMYR1’ are shown in Table III. In the ‘VMYR1’, the hydrophobic residues (Ile, Leu, Phe) contributes 40% of the total interface residues, and plays an important role to stabilize the interface by the hydrophobic interactions. The calculation of the surface electrostatic potential of the predicted structure showed that upon dimer formation the nature of the electrostatic surface changes drastically (Fig. 3). Such changes are also observed at places away from the interface region. This indicates that the dimerization might be functionally relevant.

Coordinates

Coordinates for the structure of the ‘VMYR1’ have been deposited with the code 1W71 in the Protein Data Bank.

The modeling of 3D structure of the ‘VMYR1’ is a challenging task due to the ab-sence of any available 3D structure of plant ‘R’ proteins. This theoretical model of the ‘VMYR1’ may pave the way to understand the molecular functioning of ‘R’ proteins in plants. Further improvement of the model is possible by deciphering the full-length gene and this work is underway. Hopefully, the refined model will enable us to predict the molecular mechanism of resistance against YMV after elu-cidation of coat-protein structure of the virus. To the best of our knowledge this is the first report of a model structure of any plant disease resistance protein.

Acknowledgments

J. B. is thankful to the Department of Science and Technology, Govt. of India for providing her research fellowship. R. P. B. acknowledges his gratitude to the Coun-cil of Scientific and Industrial Research, India, for the research fellowship. We extend our thanks to Dr. Chaitali Mukhopadhyay, Dr. Tarun Kanti Mondal, and Dr. Rajasri Bhattacharya for valuable suggestions during this work.

References and Footnotes

Table III Residues involved at homodimeric interface.

Interface residuesa

Asp 73, Asn 74, Arg 75, Lys 76, Gln 77, Leu 78, Phe 81, Arg 84, Phe 88, Leu 103, Leu104, Lys 105, His 107, Ile 109, Glu 110

aResidues of only one subunit of the homodimer is presented.

1.2.

3.4.5.

6.7.8.9.

10.11.

12.13.14.15.

16.17.18.

H. H. Flor. Annu Rev. Phytopathol. 9, 275-296 (1971).K. E. Hammond-Kosack, J. D. G. Jones. Annu. Rev. Plant Physiol. Plant Mol. Biol. 48, 575-607 (1997).M. Saraste, P. R. Sibbald, A. Wittinghoffer. Trends Biotechnol. 15, 430-435 (1990).D. A. Jones, J. D. G. Jones. Adv. Bot. Res. Incorp. Adv. Pl. Pathol. 24, 89-167 (1997).W. I. Tameling, S. D. Elzinga, P. S. Darmin, J. H. Vossen, F. L. Takken, M. A. Haring, B. J. Cornelissen. Plant Cell 11, 2929-2939 (2002).Q. Pan, J. Wendel, R. Fluhr. J. Mol. Evol. 50, 203-213 (2000).K. U. Torii, T. W. McNellis, X. W. Deng. EMBO J. 17, 5577-5587 (1998).J. Basak, S. Kundagrami, T. K. Ghose, A. Pal. Mol. Breed. 14, 375-383 (2004).H. M. Berman, J. Westbrook, Z. Feng, G. Gilliland, T. N. Bhat, H. Weissig, I. N. Shindyalov, P. E. Bourne. Nucleic Acids Res. 28, 235-242 (2000).C. F. Higgins, K. J. Linton. Science 293, 1782-1784 (2001).S. F. Altschul, T. L. Madden, A. A. Schäffer, J. Zhang, Z. Zhang, W. Miller, D. J. Lipman. Nucleic Acids Res. 25, 3389-3402 (1997).J. D. Thompson, D. G. Higgins, T. J. Gibson. Nucleic Acids Res. 22, 4673-4680 (1994).G. Chang. J. Mol. Biol. 330, 419-430 (2003).A. Sali, T. Blundell. J. Mol. Biol. 234, 779-815 (1993).R. A. Laskowski, M. W. MacArthur, D. S. Moss, J. M. Thornton. J. Applied Crystallogr. 26, 283-291 (1993).G. N. Ramachandran, V. Sashisekharan. Adv. Prot. Chem. 23, 238-283 (1968).R. A. Sayle, E. J. Milner-White. Trends Biochem. Sci. 20, 374-375 (1995).S. J. Hubbard. Department of Biochemistry and Molecular Biology, University College of London (1992).

130

Basak and Bahadur

19.20.21.22.23.24.25.26.27.28.

B. Lee, F. M. Richards. J. Mol. Biol. 55, 379-400 (1971).S. Jones, J. M. Thornton. Prog. Biophys. Mol. Biol. 63, 31-65 (1995).R. P. Bahadur, P. Chakrabarti, F. Rodier. J. Janin, Proteins 53, 708-719 (2003).R. P. Bahadur, P. Chakrabarti, F. Rodier. J. Janin, J. Mol. Biol. 336, 943-955 (2004).I. K. McDonald, J. M. Thornton. J. Mol. Biol. 238, 777-793 (1994).B. Honig, A. Nicholls. Science 268, 1144-1149 (1995).A. Nicholls, K. A. Sharp, B. Honig. Proteins 11, 281-296 (1991).G. Chang, C. B. Roth. Science 293, 1793-1800 (2001).A. Tramontano. Methods 14, 293-300 (1998).A. M. Facchiano, P. Stiuso, M. L. Chiusano, M. Caraglia, G. Giuberti, M. Marra, A. Abbru-zzese, G. Colonna. Protein Eng. 14, 881-890 (2001).

Date Received: January 16, 2006

Communicated by the Editor Krystyna Zakrzewska