• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of pnasPNASInfo for AuthorsSubscriptionsAboutThis Article
Proc Natl Acad Sci U S A. May 5, 2009; 106(18): 7666–7671.
Published online Apr 16, 2009. doi:  10.1073/pnas.0812625106
PMCID: PMC2678593
Plant Biology

Integration of evolutionary and desolvation energy analysis identifies functional sites in a plant immunity protein


Plant immune responses often depend on leucine-rich repeat receptors that recognize microbe-associated molecular patterns or pathogen-specific virulence proteins, either directly or indirectly. When the recognition is direct, a molecular arms race takes place where plant receptors continually and rapidly evolve in response to virulence factor evolution. A useful model system to study ligand-receptor coevolution dynamics at the protein level is represented by the interaction between pathogen-derived polygalacturonases (PGs) and plant polygalacturonase-inhibiting proteins (PGIPs). We have applied codon substitution models to PGIP sequences of different eudicotyledonous families to identify putative positively selected sites and then compared these sites with the propensity of protein surface residues to interact with protein partners, based on desolvation energy calculations. The 2 approaches remarkably correlated in pinpointing several residues in the concave face of the leucine-rich repeat domain. These residues were mutated into alanine and their effect on the recognition of several PGs was tested, leading to the identification of unique hotspots for the PGIP-PG interaction. The combined approach used in this work can be of general utility in cases where structural information about a pattern-recognition receptor or resistance-gene product is available.

Keywords: adaptive evolution, molecular recognition, plant-pathogen interactions, protein-protein interactions, polygalacturonase-inhibiting protein (PGIP)

The molecular recognition between a plant and a pathogen is a key event for the activation of the plant immune response. Microbe-associated molecular patterns (MAMPs), molecular motifs common to many microbes, are recognized by plant sensors such as the receptor kinases FLS2 or EFR (1), whereas pathogen-specific virulence molecules (effectors or Avr molecules) are sensed, either directly or indirectly, by specific resistance (R) proteins (2, 3), leading to the activation of a complex array of defense responses. The evolutionary conflict between plants and pathogens reflects a continual molecular coevolution process that takes place between plant sensors and their ligands: the improvement of the sensors selects for improved pathogen molecules. This adaptive evolutionary dynamic often leads to a faster variation of specific amino acid residues indicated as positively selected sites, likely important for recognition (36).

The evolution of plant immunity sensors is complex and depends not only on the strength of the selective pressure because of the population dynamics of both hosts and pathogens, but also on the protein function, the type of recognition (direct or indirect), and the complexity of the gene family (reviewed in ref. 5). Fast evolution (arms race) characterizes R-receptors performing a direct interaction with pathogen effectors (69), whereas a selectively maintained presence or absence polymorphism (balancing selection) has been demonstrated for R-proteins mediating indirect recognition (3, 6, 10). Gene exchange among R-gene paralogs also contributes to the generation of diversity and unique resistance specificities (11). A replacement polymorphism higher than that of nondefense proteins, although much lower than in R-proteins, characterizes chitinases (12), polygalacturonase-inhibiting proteins (PGIPs) (13, 14), and β-1,3-endoglucanases (15). In these defense proteins, positive selection appears to be constrained by the need of maintaining function: most amino acid sites are subject to strong purifying selection and only a small fraction is potentially targeted by adaptive evolution, likely because of their role in the interaction with the pathogen-derived ligands.

In the vast majority of plant-immunity receptors, as well as in animal receptors that mediate MAMP perception and inflammatory processes, recognition is mediated by modular domains consisting of the tandem arrangement of leucine-rich repeats (LRRs) (1619). LRR domains are extracellular in the R-receptors Cf- of tomato and Xa21 of rice (2), in the MAMP receptors FLS2 and EFR (20), as well as in PGIP (21), and are intracellular in the NBS-LRR-type R-proteins (22). In the LRRs of these proteins, the xxLxLxx motifs are often hypervariable and exhibit the highest number of positively selected sites (2, 6, 9, 10, 19). In the crystal structure of PGIP, the only available structure of an LRR plant-defense protein, LRRs form a solenoid and the concave β-sheet surface formed by the highly variable xxLxLxx motifs is apparently responsible for ligand binding (23, 24).

There is a great interest in unraveling the adaptive significance of gene diversity in plant-immunity proteins and a promising approach is the combination of molecular evolution analysis with functional studies (5, 25). PGIPs and their ligands polygalacturonases (PGs) have been structurally and functionally characterized and are a suitable model system for studying the basis of molecular adaptation in LRR proteins involved in immunity (14, 21). PGIPs are apoplastic proteins that counteract the action of fungal and insect PGs (13, 26, 27). They exhibit 10 imperfect LRRs of the extracellular type [consensus motif: xxLxLxxNxLt/sGxIPxxLxxLxx (24)] and are encoded by small gene families of clustered paralogs that show different patterns of gene expression and specificity against PGs (26, 2830). Single PGIP residues located in the concave surface of the LRR solenoid (i.e., the β-sheet B1) differentially affect the inhibition of fungal PGs (23, 31). PGIPs of several plant species, as well as single PGIPs, show different inhibition kinetics toward various fungal PGs (3134). Consistently, molecular-docking simulations of PG-PGIP complexes suggested that PGIP isoform 2 from Phaseolus vulgaris (PvPGIP2) binds the various PG partners in different areas of their surface (21). A molecular coevolution can be envisaged between PGIPs that continually adapt in response to variation of pathogen-derived PGs, and the enzymes that in turn evolve to escape PGIP recognition (14, 35, 36).

In this study we show that the integration of evolutionary, structural, and functional knowledge on PGIPs can provide new insight into the adaptive significance of diversification of immunity recognition proteins. We applied codon substitution models (37) to PGIP sequences of different eudicotyledonous families to identify positively selected sites, taking into account the available structural and site-directed mutagenesis data as well as previous similar studies (14, 35). Then we used an in-silico prediction of protein-protein interacting sites based on desolvation energy analysis (38) and compared the resulting sites with the previously identified sites. The two approaches remarkably correlated in identifying a number of residues in the concave face of the LRR domain. These residues were mutated into alanine and their effect on the recognition of several PGs was tested, leading to the identification of unique hotspots for the PGIP-PG interaction. The integrative approach used in this work can be of general utility to shed light into the functional significance of diversification in plant surveillance proteins, when their structure or a reliable homology model is known.


Diversification of PGIP Genes Has Been Driven by Positive Selection.

Maximum likelihood (ML) site-specific models of codon evolution were applied to 4 sets of PGIP sequences from different eudicotyledonous species (Fabaceae, Brassicaceae, Rosaceae and Rutaceae) [Supporting Information (SI) Table S1 and Figs. S1 and S2]. As detection of positive selection through the ω ratio needs information on both synonymous and nonsynonymous changes, neither too similar nor too divergent sequences are suitable for this analysis. Highly similar sequences provide little information; on the other hand, loss of power can result from extremely divergent sequences because of saturated substitution rates (39). Sequence divergence in the 4 different datasets, measured by the expected number of nucleotide substitutions per codon along the tree (S = tree length; Fabaceae: S = 2.4; Brassicaceae: S = 2.2; Rosaceae S = 1.4 and Rutaceae: S = 0.2) was considered suitable for the analysis.

The ML codon substitution models M0, M1, and M7 that do not account for positively selected sites, and M2, M3, and M8 that do (37) were used to detect positive selection among codon sites (see SI Materials and Methods for a more detailed description of the models). A likelihood ratio test (LRT) was then performed to compare the nested models M2 versus M1 and M8 versus M7, which are both tests of positive selection, and M3 versus M0, which is a test of heterogeneity of ω values among sites. The analysis provided evidence of positive selection occurring during PGIP gene diversification (Table 1). In the Fabaceae, Brassicaceae, and Rosaceae datasets, which showed a moderate level of sequence divergence (see Fig. S1), LRT was significant for the M3–M0 and M8–M7 comparisons, but not for the M2–M1, suggesting that diversification driven by positive selection has been limited to only a small number of PGIP residues. None of the LRTs was significant in the case of the Rutaceae dataset (data not shown), in which the lowest tree length was observed (S = 0.2).

Table 1.
Results from the CODEML analysis

Using the Bayes method, 1, 7, and 10 residues in the Rosaceae, Fabaceae, and Brassicaceae datasets, respectively, were found to be under positive selection with a posterior probability (PP) >0.95 (M8 model) (see Table 1); because 2 residues were shared in 2 different datasets, a total of 16 positively selected sites were identified. When a PP >0.80 was considered, a total of 2, 23, and 24 putative positively selected sites were detected in the Rosaceae, Fabaceae, and Brassicaceae datasets, respectively; 9 of them were shared in at least 2 datasets, resulting a total of 40 positively selected sites (Table 2).

Table 2.
Positively selected sites in PvPGIP2a and corresponding ODAb values

The 16 positively selected sites with PP >0.95, highlighted on the 3-dimensional structure of PvPGIP2 in Fig. 1, are all solvent-exposed. Twelve are located in the β-sheet B1 corresponding to the concave surface of the protein (24). Residue 152 was previously shown to contribute to PG recognition (23, 31). Residue 224, here identified with PP >0.80, was also shown by Leckie et al. (23) to play a crucial role in recognition of PG of Fusarium moniliforme strain FC-10 [now reclassified as F. phyllophilum; (40)]. Both residues were declared to be under positive selection (with PP >0.80) in previous studies (14, 35).

Fig. 1.
Stereoview of the crystal structure of PvPGIP2 showing the 16 positively selected sites (PP >0.95). These sites are highlighted in green in the sequence; putative positively sites (PP >0.80) are highlighted in yellow; residues located ...

Desolvation Energy Analyses Identify PGIP Residues with a High Tendency to Be Involved in Interactions.

If positively selected sites of PGIP evolved to interact with protein partners, residues showing ω estimates significantly >1 should also be predicted as interacting sites by bioinformatics methods used to identify protein-protein interaction surfaces. To test this hypothesis, we used the optimal docking area (ODA) method (38), which predicts the propensity of protein surface residues to interact with protein partners. Taking into account desolvation energies, the ODA method identifies areas of a protein surface that exhibit a favorable energy change when buried upon protein-protein association: sites with low desolvation energy values (negative values) are predicted to be involved in interactions. When applied to the PvPGIP2 crystal structure, this analysis revealed that 44 out of 313 residues displayed ODA values < –6.0 kcal/mol. We chose this value as a medium-stringency threshold on the basis of previous tests performed on known protein-protein interfaces (38). Interestingly, out of the 44 residues showing ODA < –6.0 kcal/mol, 16 were also putative positively selected sites with PP >0.80 and 9 with a PP >0.95 (see Table 2). A representation of the residues characterized by ω values >1 or ODA values < –6.0 kcal/mol on the PvPGIP2 surface highlights the notable correspondence between the residues identified with the 2 methods (Fig. 2).

Fig. 2.
Representation of PvPGIP2 surface with highlighted (A) positively selected sites (ω>1) with PP >0.95 (green) and with PP >0.80 (yellow), (B) residues showing ODA values lower than –6.0 kcal/mol (red) and (C) (merged ...

The tendency of negative ODA values for residues showing higher ω values was confirmed by the correlation between the 2 values. A negative and significant correlation, calculated over the 313 PGIP residues, was indeed observed between ω and ODA values both in the Fabaceae (rF = –0.27, P = 0.000001, n = 313) and the Brassicaceae (rB = –0.26, P = 0.000005, n = 313) datasets. The correlation was also calculated by considering the 3 structurally distinct regions of the protein (24): the β-sheet B1 (xxLxLxx: rF = –0.37, P = 0.0016, n = 70; rB = –0.29, P = 0.014, n = 70), the β-sheet B2 (NxLx.Gx: rF = 0.01, P = 0.94, n = 62; rB = –0.02, P = 0.90, n = 62), and the 310-helices (IPxxLxxL: rF = 0.10, P = 0.37, n = 77; rB = 0.05, P = 0.65, n = 77) (see Fig. S3). The correlation remained negative and significant only in the β-sheet B1 region, consistent with the available structural and mutagenesis data that identify this region as the one involved in recognition of PGs (23, 24, 31).

Alanine Scanning Confirms Hotspots for the PGIP-PG Interaction Identified by ω/ODA Analyses.

To assess the validity of combining molecular evolution and desolvation energy analysis to predict hot spots in the PG-PGIP interaction, residues having both an ODA value < –6.0 kcal/mol and an ω value >1.0 (PP >0.80 in at least one dataset) were selected and subjected to alanine-scanning mutagenesis (underlined in Table 2). Residues were chosen among those belonging to the β-sheet B1, where the correlation between ω and ODA is higher (reaching the value of 0.7–0.9 in some individual B1 strands; data not shown). Residues V152, S178, and H271 were not included because mutagenesis data are already available (23, 31). G85 and P92 were excluded because their mutation into alanine would likely lead to structural instability. Conversely, residue Q224 with an ODA value of only –3.18 kcal/mol was included because its importance in PGIP specificity is well known (23, 31). In total, 10 different mutations were separately introduced in the PvPGIP2 gene (Tables 2 and and3).3). Alanine variants were successfully expressed in Pichia pastoris, with the exception of N79A. The expressed variants were purified and tested against 4 different fungal PGs [Aspergillus niger PGII (AnPG); Colletotrichum lupini PG (ClPG); Fusarium phyllophilum PG (FpPG), and Fusarium graminearum PG1 (FgPG)] (Fig. S4); their inhibitory activity, compared to that of the wild-type PvPGIP2, is shown in Table 3. The group of variants comprising H104A, Y107A, and K225A exhibited a decreased activity against all PGs examined. A second group behaved differently, depending on the PG tested: Y105A was significantly less efficient against AnPGII, showed no inhibition against FpPG, and displayed activity comparable to the wild-type inhibitor against ClPG and FgPG. In contrast, V128A, F269A, and F201A behaved normally against AnPG, ClPG, and FgPG, but were impaired in the ability to inhibit FpPG: V128A and F269A were less efficient, whereas F201A showed no inhibition. A last group of variants comprising Q224A and K268A had an inhibitory activity comparable to that of the wild-type protein. The inhibition of FpPG was influenced by as many as 6 mutations out of 9 and was completely abolished or highly compromised by mutating Y105, Y107, F201, K225, and F269. The importance of these residues as hotspots for the interaction with this enzyme was previously unknown.

Table 3.
Inhibitory activities of PvPGIP2 and 9 alanine variants against 4 fungal polygalacturonases

Although tested against a limited set of PGs, 7 out of 8 PvPGIP2 alanine variants mutated in residues predicted as positive by the integration of ODA, and ω analysis had a significant lower inhibitory activity compared to the wild type against at least one PG. Another variation (i.e., V152G) previously chosen as one of those that distinguish PvPGIP2 from PvPGIP1, also displayed a differential activity against a Botrytis cinerea PG (23, 31). Instead, the variant S178A showed no difference against AnPG and FpPG (23), but its change of inhibition against other PGs cannot be ruled out. Overall, we obtained a positive predictive value of 8 out of 10 (80%). This result is remarkable if we consider that replacement of many surface residues that do not fulfill the ODA < –6 kcal/mol/ω >1 (PP >0.80) criterion does not affect the affinity of PvPGIP2 for its partners. This is the case of PGIP2 residues L60, Q291, A297, and A311, which were mutated into H, S, K, and S, respectively, because of their occurrence in the corresponding sites of PvPGIP1 (23). It is also the case of the conserved residue D157 that contributes to form a negative pocket on the PGIP surface and that was mutated into A (41), as well as residue A213, mutated into T, present in an allelic variant of PvPGIP2 (42). Furthermore, the 5, 10, and 8 variations that distinguish PGIP2 of Phaseolus coccineus, P. acutifolius, and P. lunatus, respectively, from PvPGIP2 (a total of 15 sites, 13 of which are silent based on our criterion) also do not significantly affect inhibitory activity against FpPG, ClPG, AnPG, and BcPG (42).


The LRR domain of both resistance and defense proteins is a fast-evolving region, with variable sites involved in direct or indirect recognition of pathogen effectors (7, 8, 43, 44). PGIP, with its adaptable extracellular LRR domain, is a good example of a defense protein coevolving with its pathogen-derived ligand and was chosen for an integrated approach combining evolutionary, structural, and functional studies aimed at identifying residues with an adaptive significance. Previous analyses of orthologous and paralogous PGIP sequences had already suggested adaptive evolution (14, 35), and some of the residues declared as positively selected sites play an important role in PG recognition. An example is residue 224, which is responsible for the different specificity of PvPGIP1 and PvPGIP2 against FpPG (23, 31).

In this work, we have first predicted PGIP residues under positive selection by applying a method that has been previously used by other authors (14) on 4 different datasets selected on the basis of sequence divergence. By doing so, we improved power and reliability of codon-substitution models and identified 16 positively selected sites with a PP >0.95. Only one of these residues (position 178) (see Table 3) was also detected as positively selected in the previous studies (14, 35), likely because different datasets and codon-evolution models were used; however, mutagenesis of this residue does not affect recognition of FpPG and AnPG (23). Overall, the positively selected sites identified in this study are mostly distributed in the concave face of the PGIP and all of them are solvent-exposed. Only 4 are located outside of the concave surface of the protein, but in the convex face (residues 92 and 232) or in the C terminus (residues 310 and 311). It is tempting to speculate a role other than recognition of PG for these residues, because PGIP also interacts with other partners (45) and possibly with still unknown extracytosolic proteins.

Despite the risk of false positives, we identified additional sites that may be subjected to positive selection by using a less stringent threshold (PP >0.80), as already shown by other authors (46), and searching for sites declared as positively selected in different and independent datasets. Among the 40 sites identified, a functional role has been shown for residue 224 (23) and residue 152, which contributes to the recognition of B. cinerea PG1 (31). Notably, none of the residues involved in the interaction of PGIP with homogalacturonan (45) was identified as positively selected, in agreement with the structural conservation of the polysaccharide during evolution.

The combination of evolutionary information with structural, biochemical, and physiological data greatly facilitates the identification of protein-protein interaction “hotspots” (i.e., essential residues that if mutated impede or severely affect protein-protein interactions). Recently, mutational, phylogenetic, and structural modeling approaches were used to identify the most conserved residues responsible for flagellin binding on the FLS2 receptor (47). Another recent report showed that the combination of a method mainly relying on evolutionary conservation with the analysis of structural features more accurately predicts protein-protein interaction hotspots (48). In this work, we show that the prediction of fast-evolving sites, which shape the capability of PGIP to recognize multiple and coevolving PGs, when combined with the ODA method for the prediction of interaction sites (38) and available structural data (24), leads to the identification of unique hotspots for the recognition of PGs. The striking correspondence between the ODA and the ω analysis in the prediction of interacting sites in the β-sheet B1 of PGIP is a cross-validation of both methods per se, and allowed us to pinpoint positively selected residues with a high tendency to be involved in protein-protein interactions. Among the 40 putative positively selected sites, we chose 10 solvent-exposed residues showing the lowest negative-ODA values. Both ODA and ω analysis pointed to an important role of the third LRR in the B1 strand. Therefore, we were not surprised that alanine substitution of all 3 residues of this LRR (H104A, Y105A, and Y107A) affected the inhibitory capabilities of PvPGIP2, suggesting that these residues are likely responsible for adaptive changes during the evolution of the PGIP family, with positive Darwinian selection promoting fixation of advantageous variants.

A second important spot for recognition is K225, for which mutation into alanine affects inhibition of all PGs tested. This was unexpected because previous data indicated that an important residue might be at position 224 rather than at position 225. Substitution of K224 of PvPGIP1 with a Q is indeed sufficient to confer the ability to inhibit FpPG that is normally not recognized by PvPGIP1 (23). Although identified by the CODEML analysis performed on the leguminosae PGIP dataset as a positively selected (PP >0.80), residue 224 is not identified as a putative major contributor to the interaction by our combined approach. Indeed, its negative ODA value (–3,18 kcal/mol) is higher than the value chosen as a medium-stringency threshold. In agreement with this prediction, mutation Q224A has no considerable effect on the inhibition of FpPG as well as of the other PGs tested. The apparent discrepancy with previous data can be explained if this site, rather than representing a contact point, negatively affects the interaction when occupied by a lysine, because of charge repulsion with the nearby lysine at 225 and possibly with positive charges in the partner PG. The requirement for the interaction with FpPG might therefore be the absence of a lysine at position 224, rather than presence of a glutamine. We also identified a number of residues, such as Y105, V128, F201, and F269, which may be classified as hotspots or not depending on the PG tested. Residue V152, analyzed in a previous work (31), also falls into this category.

PGs are extremely variable in terms of shape and electrostatic surfaces as well as desolvation energy surfaces, although maintaining the same overall fold (31, 33) and single PGIPs (i.e., PvPGIP2) can inhibit different PGs with different mechanisms, either competitive or noncompetitive. These features lead us to propose that in the diverse PvPGIP2-PG pairs different sets of contacts take place, possibly involving different surface areas and residues of the PG ligands (21). By identifying 4 previously unrecorded residues playing a differential role depending on the PG tested, we further support this hypothesis that may also explain why not all of the tested residues contribute to the interaction. Indeed, while evolutionary analyses were made on PGIPs of different plant families, only one particular PGIP (PvPGIP2) was mutagenized and its inhibitory capability tested against a limited number of fungal PGs. Furthermore, because the in vitro assay conditions are unlikely to mimic the physiological microenvironment of the apoplast and the interaction between the variants and PGs might differ in planta from that detected in vitro (49), the effects of the variants obtained in this work on plant response against pathogens therefore cannot be easily predicted and remain to be determined.

In conclusion, although, like in other proteins (50, 51), many surface residues of PvPGIP2 do not significantly contribute to the interaction with PGs (23, 31, 41, 42, 45), we have shown that 7 residues with a strong impact on binding and inhibitory activity could be identified among 8 selected on the basis of ODA and ω analyses. Our study demonstrates that the integration of evolutionary, structural, and functional analyses can substantially reduce the number of putative protein-protein interaction sites that have to be mutagenized for validation of their functional role. The combined approach used in this work can be of general utility in cases where the structure or a reliable homology model of a pattern recognition receptor or resistance gene product is known.

Materials and Methods

Selection of PGIP Sequences for Testing Positive Selection.

Forty-five full-length PGIP sequences of eudicotyledonous species, belonging to 4 different families (Fabaceae, Brassicaceae, Rosaceae, and Rutaceae) (see Table S1), were chosen from the National Center for Biotechnology Information database (http://www.ncbi.nlm.nih.gov/) for inferring positive selection using comparative DNA sequence methods.

ML Models of Codon Evolution.

Comparative methods were used to test whether positive selection has driven the evolutionary divergence of orthologous and paralogous PGIP genes. Codon-substitution models implemented in the ML framework (37, 52) were applied to PGIP sequences using the CODEML module in the PAML 3.13 package (53). A detailed description of the procedure is found in the SI Materials and Methods.

Calculation of ODA Values.

ODA values for each site of PvPGIP2 crystal structure (24, 31) were calculated using a variation of the previously reported method (38), with the starting points for the generation of optimal surface patches now located at the geometrical center of every surface residue. The ODA method analyzes a protein surface in search of areas with favorable energy change when buried upon protein-protein association. Low desolvation-energy values (ODA negative values) correspond to regions, over the protein surface, predicted to be involved in protein-protein interactions.

Correlation Between ω and ODA Values.

For each site of the PvPGIP2, the ML estimation of the ω value derived from the model M8 of CODEML was considered. The Pearson linear correlation coefficient between ω and ODA values was calculated over the 313 protein sites for the Fabaceae (rF), Brassicaceae (rB), and Rosaceae (rR) datasets. The same correlation was also calculated taking into account only residues belonging to 3 structurally distinct regions: the β-sheet B1 (xxLxLxx, n = 70), the β-sheet B2 (NxLx.Gx, n = 62), and the 310-helices (IPxxLxxL, n = 77). Statistical significance for the correlation coefficients was estimated by the STATISTICA software (StatSoft Italia 1995, STATISTICA for Windows) and by permutations.

Site-Directed Mutagenesis of PvPGIP2 and Inhibitory Activity Assay.

A detailed description of the procedure is provided in the SI Materials and Methods.

Supplementary Material

Supporting Information:


We thank Lucia Tufano and Gianni Salvi for the valuable technical assistance, Manuel Benedetti for helpful assistance with fungal PG preparation, and Renato D'Ovidio and Benedetta Mattei for helpful discussions. This work was supported by the Italian Ministry of Education, University and Research Grants ERA-PG RBER063SN4 (to G.D.L.) and PRIN 2005052297 and 2007K7KY8Y (to G.D.L. and L.F.), by a European Research Council Advanced grant (to F.C.), and by the Spanish Ministry of Science Plan Nacional I+D+I Grant BIO2005–06753 (to J. F.-R).


The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

This article contains supporting information online at www.pnas.org/cgi/content/full/0812625106/DCSupplemental.


1. Zipfel C. Pattern-recognition receptors in plant innate immunity. Curr Opin Immunol. 2008;20:10–16. [PubMed]
2. Jones JD, Dangl JL. The plant immune system. Nature. 2006;444:323–329. [PubMed]
3. Van der Hoorn RA, Kamoun S. From guard to decoy: a new model for perception of plant pathogen effectors. Plant Cell. 2008;20:2009–2017. [PMC free article] [PubMed]
4. Stahl EA, Bishop JG. Plant-pathogen arms races at the molecular level. Curr Opin Plant Biol. 2000;3:299–304. [PubMed]
5. Tiffin P, Moeller DA. Molecular evolution of plant immune system genes. Trends Genet. 2006;22:662–670. [PubMed]
6. Ellis JG, Dodds PN, Lawrence GJ. Flax rust resistance gene specificity is based on direct resistance-avirulence protein interactions. Annu Rev Phytopathol. 2007;45:289–306. [PubMed]
7. Allen RL, et al. Host-parasite coevolutionary conflict between Arabidopsis and downy mildew. Science. 2004;306:1957–1960. [PubMed]
8. Rose LE, et al. The maintenance of extreme amino acid diversity at the disease resistance genes, RPP13, in Arabidopsis thaliana. Genetics. 2004;166:1517–1527. [PMC free article] [PubMed]
9. McHale L, Tan X, Koehl P, Michelmore RW. Plant NBS-LRR proteins: adaptable guards. Genome Biol. 2006;7:212. [PMC free article] [PubMed]
10. Bergelson J, Kreitman M, Stahl EA, Tian D. Evolutionary dynamics of plant R-genes. Science. 2001;292:2281–2285. [PubMed]
11. Parniske M, et al. Novel disease resistance specificities result from sequence exchange between tandemly repeated genes at the Cf-4/9 locus of tomato. Cell. 1997;91:821–832. [PubMed]
12. Bishop JG, Dean AM, Mitchell-Olds T. Rapid evolution in plant chitinases: molecular targets of selection in plant-pathogen coevolution. Proc Natl Acad Sci USA. 2000;97:5322–5327. [PMC free article] [PubMed]
13. De Lorenzo G, D'Ovidio R, Cervone F. The role of polygacturonase-inhibiting proteins (PGIPs) in defense against pathogenic fungi. Annu Rev Phytopathol. 2001;39:313–335. [PubMed]
14. Bishop JG. Directed mutagenesis confirms the functional importance of positively selected sites in polygalacturonase inhibitor protein. Mol Biol Evol. 2005;22:1531–1534. [PubMed]
15. Bishop JG, et al. Selection on Glycine beta-1,3-endoglucanase genes differentially inhibited by a Phytophthora glucanase inhibitor protein. Genetics. 2005;169:1009–1019. [PMC free article] [PubMed]
16. Kobe B, Kajava AV. The leucine-rich repeat as a protein recognition motif. Curr Opin Struct Biol. 2001;11:725–732. [PubMed]
17. Pancer Z, Cooper MD. The evolution of adaptive immunity. Annu Rev Immunol. 2006;24:497–518. [PubMed]
18. Trinchieri G, Sher A. Cooperation of Toll-like receptor signals in innate immune defence. Nat Rev Immunol. 2007;7:179–190. [PubMed]
19. Bent AF, Mackey D. Elicitors, effectors, and R genes: the new paradigm and a lifetime supply of questions. Annu Rev Phytopathol. 2007;45:399–436. [PubMed]
20. Schwessinger B, Zipfel C. News from the frontline: recent insights into PAMP-triggered immunity in plants. Curr Opin Plant Biol. 2008;11:389–395. [PubMed]
21. Federici L, et al. Polygalacturonase inhibiting proteins: players in plant innate immunity? Trends Plants Sci. 2006;11:65–70. [PubMed]
22. Chisholm ST, Coaker G, Day B, Staskawicz BJ. Host-microbe interactions: shaping the evolution of the plant immune response. Cell. 2006;124:803–814. [PubMed]
23. Leckie F, et al. The specificity of polygalacturonase-inhibiting protein (PGIP): a single amino acid substitution in the solvent-exposed β- strand/β-turn region of the leucine-rich repeats (LRRs) confers a new recognition capability. EMBO J. 1999;18:2352–2363. [PMC free article] [PubMed]
24. Di Matteo A, et al. The crystal structure of PGIP (polygalacturonase-inhibiting protein), a leucine-rich repeat protein involved in plant defense. Proc Natl Acad Sci USA. 2003;100:10124–10128. [PMC free article] [PubMed]
25. Golding GB, Dean AM. The structural basis of molecular adaptation. Mol Biol Evol. 1998;15:355–369. [PubMed]
26. D'Ovidio R, et al. Characterization of the complex locus of bean encoding polygalacturonase-inhibiting proteins reveals subfunctionalization for defense against fungi and insects. Plant Physiol. 2004;135:2424–2435. [PMC free article] [PubMed]
27. De Lorenzo G, Ferrari S. Polygalacturonase-inhibiting proteins in defense against phytopathogenic fungi. Curr Opin Plant Biol. 2002;5:295–299. [PubMed]
28. Ferrari S, et al. Tandemly duplicated Arabidopsis genes that encode polygalacturonase-inhibiting proteins are regulated coordinately by different signal transduction pathways in response to fungal infection. Plant Cell. 2003;15:93–106. [PMC free article] [PubMed]
29. Li R, et al. Two Brassica napus polygalacturonase inhibitory protein genes are expressed at different levels in response to biotic and abiotic stresses. Planta. 2003;217:299–308. [PubMed]
30. D'Ovidio R, et al. The characterization of the soybean polygalacturonase-inhibiting proteins (Pgip) gene family reveals that a single member is responsible for the activity detected in soybean tissues. Planta. 2006;224:633–645. [PubMed]
31. Sicilia F, et al. The polygalacturonase-inhibiting protein PGIP2 of Phaseolus vulgaris has evolved a mixed mode of inhibition of endopolygalacturonase PG1 of Botrytis cinerea. Plant Physiol. 2005;139:1380–1388. [PMC free article] [PubMed]
32. King D, et al. Use of amide exchange mass spectrometry to study conformational changes within the endopolygalacturonase II-homogalacturonan-polygalacturonase inhibiting protein system. Biochemistry. 2002;41:10225–10233. [PubMed]
33. Bonivento D, et al. Crystal structure of the endopolygalacturonase from the phytopathogenic fungus Colletotrichum lupini and its interaction with polygalacturonase-inhibiting proteins. Proteins. 2008;70:294–299. [PubMed]
34. Federici L, et al. Structural requirements of endopolygalacturonase for the interaction with PGIP (polygalacturonase-inhibiting protein) Proc Natl Acad Sci USA. 2001;98:13425–13430. [PMC free article] [PubMed]
35. Stotz HU, et al. Identification of target amino acids that effect interactions of fungal polygalacturonases and their plant inhibitors. Physiol Mol Plant Pathol. 2000;56:117–130.
36. Misas-Villamil JC, Van der Hoorn RA. Enzyme-inhibitor interactions at the plant-pathogen interface. Curr Opin Plant Biol. 2008;11:380–388. [PubMed]
37. Yang Z, Bielawski JP. Statistical methods for detecting molecular adaptation. Trends Ecol Evol. 2000;15:496–503. [PubMed]
38. Fernandez-Recio J, Totrov M, Skorodumov C, Abagyan R. Optimal docking area: a new method for predicting protein-protein interaction sites. Proteins. 2005;58:134–143. [PubMed]
39. Anisimova M, Bielawski JP, Yang Z. Accuracy and power of the likelihood ratio test in detecting adaptive molecular evolution. Mol Biol Evol. 2001;18:1585–1592. [PubMed]
40. Mariotti L, et al. Reclassification of Fusarium verticillioides (syn. F. moniliforme) strain FC-10 as F phyllophilum. Mycol Res. 2008;112:1010–1011. [PubMed]
41. Spinelli F, et al. Three aspartic acid residues of polygalacturonase inhibiting protein (PGIP) from Phaseolus vulgaris are critical for inhibition of Fusarium phyllophilum PG. Plant Biol. 2009 doi: 10.1111/i.1438-8677.2008.00175.x. [PubMed] [Cross Ref]
42. Farina A, et al. The bean polygalacturonase inhibiting protein 2 (PvPGIP2) is highly conserved in common bean (Phaseolus vulgaris L.) germplasm and related species. Theor Appl Genet. 2009 doi: 10.1007/s00122-009-0987-4. [PubMed] [Cross Ref]
43. Michelmore RW, Meyers BC. Clusters of resistance genes in plants evolve by divergent selection and a birth-and-death process. Genome Res. 1998;8:1113–1130. [PubMed]
44. Dodds PN, et al. Direct protein interaction underlies gene-for-gene specificity and coevolution of the flax resistance genes and flax rust avirulence genes. Proc Natl Acad Sci USA. 2006;103:8888–8893. [PMC free article] [PubMed]
45. Spadoni S, et al. Polygalacturonase-inhibiting protein (PGIP) interacts with pectin through a binding site formed by four clustered residues of arginine and lysine. Plant Physiol. 2006;141:557–564. [PMC free article] [PubMed]
46. Benderoth M, et al. Positive selection driving diversification in plant secondary metabolism. Proc Natl Acad Sci USA. 2006;103:9118–9123. [PMC free article] [PubMed]
47. Dunning FM, et al. Identification and mutational analysis of Arabidopsis FLS2 leucine-rich repeat domain residues that contribute to flagellin perception. Plant Cell. 2007;19:3297–3313. [PMC free article] [PubMed]
48. Ofran Y, Rost B. Protein-protein interaction hotspots carved into sequences. PLoS Comput Biol. 2007;3:e119. [PMC free article] [PubMed]
49. Joubert DA, et al. A polygalacturonase-inhibiting protein from grapevine reduces the symptoms of the endopolygalacturonase BcPG2 from Botrytis cinerea in Nicotiana benthamiana leaves without any evidence for in vitro interaction. Mol Plant-Microbe Interact. 2007;20:392–402. [PubMed]
50. Clackson T, Wells JA. A hot spot of binding energy in a hormone-receptor interface. Science. 1995;267:383–386. [PubMed]
51. Thorn KS, Bogan AA. ASEdb: a database of alanine mutations and their effects on the free energy of binding in protein interactions. Bioinformatics. 2001;17:284–285. [PubMed]
52. Yang Z. Likelihood ratio tests for detecting positive selection and application to primate lysozyme evolution. Mol Biol Evol. 1998;15:568–573. [PubMed]
53. Yang Z. London: University College; 2000. Analysis by Maximum Likelihood (PAML), Version 3.13.

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences
PubReader format: click here to try


Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...


  • MedGen
    Related information in MedGen
  • PubMed
    PubMed citations for these articles
  • Substance
    PubChem Substance links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...