The crystal structure of Pseudomonas avirulence protein AvrPphB: A papain-like fold with a distinct substrate-binding site

Abstract

AvrPphB is an avirulence (Avr) protein from the plant pathogen Pseudomonas syringae that can trigger a disease-resistance response in a number of host plants including Arabidopsis. AvrPphB belongs to a novel family of cysteine proteases with the charter member of this family being the Yersinia effector protein YopT. AvrPphB has a very stringent substrate specificity, catalyzing a single proteolytic cleavage in the Arabidopsis serine/threonine kinase PBS1. We have determined the crystal structure of AvrPphB by x-ray crystallography at 1.35-Å resolution. The structure is composed of a central antiparallel β-sheet, with α-helices packing on both sides of the sheet to form a two-lobe structure. The core of this structure resembles the papain-like cysteine proteases. The similarity includes the AvrPphB active site catalytic triad of Cys-98, His-212, and Asp-227 and the oxyanion hole residue Asn-93. Based on analogy with inhibitor complexes of the papain-like proteases, we propose a model for the substrate-binding mechanism of AvrPphB. A deep and positively charged pocket (S2) and a neighboring shallow surface (S3) likely bind to aspartic acid and glycine residues in the substrate located two (P2) and three (P3) residues N terminal to the cleavage site, respectively. Further implications about the specificity of plant pathogen recognition are also discussed.

Plants activate a highly coordinated disease-resistance response upon recognition of a pathogen. Failure of this recognition event or suppression of the resistance response by the pathogen results in plant disease. Pathogen recognition requires the simultaneous presence of an avirulence (Avr) gene product from the pathogen and a cognate plant resistance (R) protein (1). The genetic interaction between an R gene and an Avr gene leads to the activation of plant disease-resistance signaling cascades, culminating in the plant hypersensitive response (HR) (1). The HR refers to the rapid, localized programmed cell death response at the site of pathogen attack that correlates with the restriction of pathogen growth. The predominant class of R proteins is known as NB-LRR proteins, because they contain a nucleotide-binding site followed by leucine-rich repeats (1, 2). Interestingly, the domain architecture of the plant NB-LRR proteins resembles that of the mammalian Nod proteins, which also play a role in the innate immune response (3). The presence of ≈175 NB-LRR proteins encoded in the Arabidopsis genome is believed to provide the specificity and diversity required for plant pathogen recognition (4, 5).

Extensive genetic studies in the past 10 years have identified >30 pairs of Avr-R genes (2). The biochemical mechanisms underlying the Avr-R genetic interaction are just beginning to be unraveled (6). The “gene-for-gene” interaction model proposed >30 years ago by Flor suggested that Avr proteins act as ligands and R proteins serve as the corresponding receptors (7). This is likely to be an oversimplified model. Instead, emerging biochemical evidence suggests that Avr proteins specifically modify host targets, thus generating signals detected by R proteins in resistant plants (8-11). According to the “guard” hypothesis, modification of the host target(s) also contributes to pathogen virulence and promotes disease in susceptible hosts that lack the corresponding R gene (1, 12). Therefore, it is expected that biochemical studies of Avr proteins, aimed at identifying their host targets and revealing the nature of target modification, will facilitate our understanding of the molecular mechanisms conferring plant disease resistance and pathogen virulence.

We have used the AvrPphB protein from the plant pathogen Pseudomonas syringae as a model to understand disease-resistance mechanisms. AvrPphB is an effector protein secreted by the Pseudomonas type III secretion system into the plant host (13, 14). AvrPphB induces the HR in a number of host plants, including Arabidopsis (15). The PBS1 and RPS5 genes of Arabidopsis are specifically required for AvrPphB-induced HR (16-18). RPS5 is a member of the NB-LRR class of plant R proteins, and PBS1 is a protein serine/threonine kinase. Our biochemical studies have established that AvrPphB belongs to an additional family of cysteine proteases (15). The charter member of this family is the virulence factor YopT from the pathogenic Yersinia species.

AvrPphB undergoes an autoproteolytic processing event, cleaving itself between Lys-62 and Gly-63 (15, 19). The catalytic triad in AvrPphB (Cys-98, His-212, and Asp-227) is required for autoproteolytic cleavage and AvrPphB-induced disease resistance (15). Recently, we have shown that PBS1 is a proteolytic substrate for AvrPphB (10). The cleavage of PBS1 appears to be required for AvrPphB/RPS5-mediated disease resistance. Therefore, we propose that AvrPphB is recognized indirectly via its enzymatic activity (cleavage of PBS1) by the RPS5 protein in Arabidopsis (10). This could represent a widespread resistance mechanism, because several Avr proteins are now thought to function as specific proteases (10, 15, 20-26).

To understand disease-resistance mechanisms at the molecular level and provide a rationale for engineering this process for the benefit of crop protection, a structural understanding of the proteins mediating pathogen recognition is needed. In the present study, we determined the crystal structure of the autoprocessed form of AvrPphB. The overall structure of AvrPphB, along with a detailed analysis of its catalytic site, is described. Our work confirms the previous prediction that AvrPphB belongs to the papain superfamily of cysteine proteases, with the structurally conserved active site catalytic triad and the oxyanion hole (15). Comparison of the AvrPphB structure with other cysteine protease family members provides insight into the interactions that most likely occur between AvrPphB and its substrate, PBS1.

Materials and Methods

Protein Expression and Purification. Full-length AvrPphB from P. syringae pv. phaseolicola was cloned into the NdeI/SalI sites of pET21a (Novagen) and transformed into Escherichia coli strain BL21 (DE3) to be expressed as a recombinant protein with a hexahistidine tag at its C terminus. The E. coli strain harboring this plasmid was grown in M9 minimal media and induced by 0.4 mM isopropyl β-d-thiogalactoside at 20°C for 20 h. AvrPphB was purified by Ni²⁺-NTA affinity chromatography (Qiagen, Chatsworth, CA) with a subsequent size-exclusion Superdex 200 column (Amersham Pharmacia). A selenomethionine form of the protein was expressed and purified following the same protocol with E. coli strain B834 (DE3).

Crystallization and Data Collection. Purified proteins were buffer-exchanged into 20 mM Tris·HCl (pH 7.5), 10 mM NaCl, and 1 mM Tri(2-carboxyethyl)phosphine hydrochloride by using microfiltration (Centricon-10, Amicon) and further concentrated to 17 mg/ml. Both AvrPphB and selenomethionine AvrPphB crystals were grown by the sitting drop vapor diffusion method under the optimal condition of 1.95 M (NH₄)₂SO₄, 100 mM Pipes (pH 6.7) at room temperature. Crystals appeared 3-5 days after initial set-up and grew to optimal size (0.2 × 0.3 × 0.3 mm³) in 7-10 days. Crystals were transferred in three steps into a cryo-protecting solution containing 2.1 M (NH₄)₂SO₄, 100 mM Pipes (pH 6.7), and 30% (wt/vol) glucose and then flash-cooled with liquid nitrogen. Cryo-protected crystals were used for data collection under a nitrogen stream at 110 K. Both native and four-wavelength multiple-wavelength anomalous diffraction data were collected at the Structural Biology Center Beamline ID-19 at the Advanced Photon Source and processed with the HKL2000 package (27). Crystals belong to the orthorhombic space group P2₁2₁2₁ with unit cell constants a = 43.6 Å, b = 49.6 Å, c = 75.7 Å (Table 1, which is published as supporting information on the PNAS web site). There is one AvrPphB molecule in the crystallographic asymmetric unit.

Structure Determination and Refinement. Experimental phases were obtained by the multiple anomalous diffraction method (28) (Table 2, which is published as supporting information on the PNAS web site). All structure determination and refinement procedures were carried out with the CNS program (29) and model building used the O program (30). Two selenium sites expected for one AvrPphB molecule were found by using the Patterson method. Subsequently, heavy atom parameters were refined, and multiple-wavelength anomalous diffraction phases were calculated and improved by solvent flipping to 1.85-Å resolution. The high quality of the experimental electron density maps and the known selenium sites allowed unambiguous tracing of the protein main chain and most of the side chains. A random selection of 10% of the reflections was set aside as the test set for cross-validation during the refinement (Table 3, which is published as supporting information on the PNAS web site). Initial refinement consisted of several iterations of conjugated gradient minimization, torsion angle dynamics-simulated annealing by the maximum-likelihood target function with experimental phases as a prior phase distribution, grouped B-factor refinement, and model rebuilding (29). These were performed against the native data set initially at 2.0-Å resolution and gradually extended to 1.35 Å. Later rounds of refinement also included selecting chemically reasonable water molecules and model rebuilding in phase-combined σ_A-weighted 2|F_o| - |F_c| and |F_o| - |F_c| maps and individual atomic B-factor refinement. Water molecules were modeled into isolated electron density that appear in both σ_A-weighted 2|F_o| - |F_c| and |F_o| - |F_c| maps and form hydrogen bonds with AvrPphB or other water molecules. The final model includes residues 81-267 and a total of 209 water molecules.

Results and Discussion

Overall Structure of AvrPphB. AvrPphB overexpressed and purified from E. coli undergoes autoproteolytic processing (15, 19), thus our purified AvrPphB protein was 28 kDa and began with Gly-63, a potential myristoylation site (15, 31). The first 18 residues (63-80) as well as the C-terminal hexa-histidine tag were not visible in the electron density map, and thus were not included in our structure. The structure was determined to near-atomic resolution of 1.35 Å with an R factor of 20.3% and an R_free of 21.8%. The rms deviations from ideal values in bond lengths and bond angles are 0.004 Å and 1.3°, respectively. All of the residues are located in favored regions of the Ramachandran plot.

The structure of AvrPphB has an α/β fold, with seven β-strands (β1-β7) and six α-helices (αA-αF) (Fig. 1). It is organized around a central seven-stranded antiparallel β-sheet, with α-helices packing on both sides of the sheet to form a two-lobe structure. The top lobe, as viewed in Fig. 1, contains the β-sheet and two helices. The order of the strands in the sheet follows β1, β6, β5, β4, β3, β7, and β2. His-212 and Asp-227, which have both been implicated in catalysis, are located at the N-terminal and C-terminal ends of strand β4 and strand β5, respectively. The bottom lobe consists of helices αA-αD. The active site nucleophile Cys-98 is located at the N terminus of helix αA. Right above the cysteine slightly to the left in the top lobe is His-212. A potential substrate-binding site of AvrPphB is located between the molecule's two lobes, within a crevice that is open and shallow on the left side but narrow and deep on the right side. As discussed below, the catalytic nucleophile Cys-98, the general base His-212, and Asp-227 form a catalytic triad that, along with other structural features, resembles the well-known superfamily of papain-like cysteine proteases.

Fig. 1.

A stereo ribbon diagram of AvrPphB. α-Helices are drawn as magenta coils and labeled A-F; β-strands are drawn as green arrows and labeled 1-7; and other structural elements are drawn as thick gray lines. The catalytic triad residues at the enzyme active site (Cys-98, His-212, and Asp-227) and the residue forming the oxyanion hole (Asn-93) are drawn as ball-and-stick models. Image was prepared with the program molscript (51).

The catalytic cysteine residue is solvent exposed and appears to have extra electron density surrounding its sulfur atom. The extra density could correspond to sulfinic acid, which results from oxidation of the catalytic cysteine. This is possibly caused by the higher reactivity and corresponding lowered pK_a for Cys-98, because the other two cysteines in the AvrPphB structure are free of such modification. This finding was confirmed by treatment with 10 mM DTT, which resulted in the disappearance of this extra density. Similar oxidation of active site sulfhydryl group has been observed for other papain family members (32).

Despite the relatively low sequence identity among members of the YopT family of cysteine proteases, all members have the same predicted secondary structure assignment as determined by the neural network program of the predictprotein server (33) (Fig. 2). This finding suggests that all members of the YopT family are likely to adopt a common 3D structure represented by that of AvrPphB. There are a number of highly conserved residues within the family in addition to the invariant catalytic triad residues as highlighted in Fig. 2. They are clustered mostly around the catalytic triad, in particular, Asp-227. Three conserved residues, Trp-105, Phe-226, and Pro-228, stack on each other via Van del Waals interactions (Fig. 3). Trp-105 also forms a hydrogen bond with the conserved Glu-232. These interactions presumably bring helix αA near strands β4 and β5 to help form and stabilize the catalytic triad. Consistent with this analysis, replacement of the corresponding tryptophan (Trp-146) in YopT with alanine abolished the cytotoxic effect of YopT in yeast (15) and the cytoskeleton disruption phenotype in transfected mammalian cells (data not shown).

Fig. 2.

Multiple sequence alignment of the representative members of the YopT cysteine protease family. Y4zC and Q9amw4 are putative type III effectors from plant symbiotic bacteria Rhizobium and Bradyrhizobium, respectively; Q9rbw5 is from P. syringae pv. phaseolicola and is also known as ORF4; AvrPphB is an Avr protein from P. syringae pv. phaseolicola; and YopT is an effector protein from human pathogen Yersinia pestis. All of the sequences were aligned with the program clustalw (52). Invariant residues are shaded, and the catalytic triad residues are indicated with bold red letters. Secondary structure elements are indicated underneath the sequence: α-helices, rectangles; β-strands, arrows; other elements, solid lines; structurally unobserved residues at the N terminus, dashed lines. Color assignment is the same as in Fig. 1. For the purpose of clarity, sequences in non-AvrPphB proteins, where no homology is identified with AvrPphB, have been replaced by bracketed numbers indicating the number of residues omitted from the display.

Fig. 3.

Electron density map of AvrPphB. Stereo pair of a σ_A-weighed 2|F_o| - |F_c| simulated-annealing omit electron density map (1.35 Å, contoured at 1.2 σ) calculated with the final refined coordinates. Shown here is a region near the catalytic triad. The region includes highly conserved residues Trp-105, Asp-227 (one of the catalytic triad residues), Pro-228, Asn-229, Gly-231, Glu-232, and Phe-233. Notably, the rings of Trp-105, Phe-226, and Pro-228 are stacked on each other. The image was prepared with the program molscript.

Structural Homologs of AvrPphB and the Mechanism of Catalysis. The overall architecture of AvrPphB bears similarity to that seen for members of papain-like cysteine protease superfamily (34), including papain (35), actinidin (36), glycyl endopeptidase (37), cruzain (38), cathepsin B (39), bleomycin hydrolase (40), staphopain,¶ and ubiquitin C-terminal hydrolase (41). The structural features of the papain superfamily consist of the common core of one α-helix and four strands of antiparallel β-sheet, which are also present in the AvrPphB structure. Members differ by insertion into and circular permutation of the catalytic core. Of the papain-like proteases, staphopain has the highest overall structural similarity to AvrPphB, as indicated by a search performed with the Dali algorithm (42). The rms deviation value between AvrPphB and Staphopain for 110 selected C^α atoms is 3.7 Å.

Secondary structure elements of AvrPphB that have structural equivalents in papain-like proteases include most of the central antiparallel β-sheet (β2-β7) and helix αA. The major differences among these structures lie in the connections between secondary structure elements (Fig. 4A). For example, AvrPphB contains three helices between αA and β2 with short loops connecting the helices. This gives AvrPphB a somewhat more compact and rigid bottom lobe. Other members including papain and cathepsin B have a long winding loop at the bottom of the lobe. On the top lobe, AvrPphB has a tight β-turn between β3 and β4 (which contains the active site residue His-212). At the corresponding location, papain has a 23-residue insertion that loops around the top part of the molecule whereas staphopain has an eight-residue loop (Fig. 4A). This region contributes to the formation of the substrate-binding S2 site in papain-like proteases and has the highest B factor in the AvrPphB structure (average B factor 28.7 Ų as compared with average B factor 15.5 Ų for the entire protein). As discussed below, based on the location and structure flexibility, we hypothesize that this region plays an important role in shaping the substrate-binding S2 site in AvrPphB.

Fig. 4.

Structure comparison of AvrPphB with papain-like cysteine proteases. (A) Ribbon diagrams of AvrPphB (Left), papain (PDB ID code 1ppn) (Center), and staphopain (PDB ID code 1cv8) (Right). The structures are aligned based on least-squares superimposition on the active site residue C^α atoms. The active site residues are shown as ball-and-stick models. The structurally conserved helix αA and central β-sheets are colored magenta and green, respectively. The loop between strands β3 and β4 are colored red. (B) Comparison of AvrPphB and papain-like active sites. Active site residues of AvrPphB, Asn-93, Cys-98, His-212, and Asp-227 are shown as green stick models. A representative collection of five papain-like cysteine protease active sites are shown as gray stick models after least-squares superimposition on the active site residue C^α atoms. The papain-like structures shown have PDB ID codes 2act, 1cv8, 1gec, 1me4, and 1ppn. Other papain-like structures used in structural comparisons in this article are PDB ID codes 1meg, 1huc, 1yal, 1cqd, 1k3b, 8pch, 1atk, 1mhw, 1gho, 1fh0, 1ef7, 2cb5, and 1dki. Refer to the Protein Data Bank for primary references to these structures, which are not included here because of space limits. Images were prepared with the program molscript.

Overlap of the AvrPphB catalytic triad (Cys-98, His-212, and Asp-227) with that of the papain-like enzymes (Cys, His, and Asn) yields rms deviation values on the three C^α atoms of between 0.18 and 0.36 Å for 17 papain-like structures listed in the SCOP database (43). In the superimposed structures of AvrPphB and papain, the central helix and the catalytic cysteine and histidine align very well, with the third catalytic residue Asp-227 in AvrPphB corresponding to Asn-175 in papain (Fig. 4B). In papain, the main-chain amide group of Cys-25 and the side-chain amide group of Gln-19 define the so-called oxyanion hole, which binds to the main-chain carbonyl group of the P1 residue of the substrate (defined as the first residue on the N-terminal side of the cleavage site). In AvrPphB, Asn-93 occupies an analogous position to Gln-19 in papain, which indicates that it may also participate in the formation of the oxyanion hole (44, 45). Overlap of all four of these AvrPphB active site residues on papain-like proteases yields rms deviation values that range from 0.60 to 0.88 Å for C^α atoms as illustrated in Fig. 4B.

These structural similarities suggest that the catalytic mechanism of AvrPphB will mirror that of the papain-like proteases (46). Thus, it is likely that AvrPphB Cys-98 and His-212 form a thiolate-imidazolium ion pair, Asp-227 functions to orient the enzyme's active site and perhaps to stabilize the protonated form of His-212, and Asn-93 contributes to formation of the oxyanion hole. These essential roles in catalysis are supported by experimental data, which have shown that mutation of any of the Cys, His, and Asp residues in both AvrPphB and YopT completely abrogate their proteolytic activities as well as biological functions (15, 47). E-64 (N-[N-(L-3-trans-carboxirane-2-carbonyl)-l-leucyl]-agmatine), a well-studied inhibitor for papain-like cysteine proteases, is known to specifically recognize the catalytic cysteine and covalently modify its sulfhydrl group (48). In the case of AvrPphB, proteolytic cleavage of its substrate PBS1 protein is largely abolished when the enzyme is mixed with E-64 (data not shown). In addition, E-64 is a potent and efficient inhibitor of the YopT cleavage of its substrate prenylated Rho GTPases (15). These data suggest that AvrPphB and all other members of the YopT family share a common catalytic mechanism with that of papain protease.

Insights into Substrate Binding Specificity. AvrPphB cleaves itself between Lys-62 and Gly-63, and it cleaves PBS1 between Lys-243 and Ser-244 (10, 15). A common Gly-Asp-Lys motif was found to occupy the P3, P2, and P1 sites in both substrates (Fig. 5A). A likely mode of substrate binding to AvrPphB is suggested by analogy with complexes of papain-like proteases, in which bound inhibitors occupy the substrate binding sites S or S′. Although the loops connected by a disulfide bond near the active site in the bottom lobe of papain are not present in AvrPphB, the corresponding space is filled by the aromatic side chain of Tyr-138. This is likely to be the S1 substrate-binding site in AvrPphB (Fig. 5 B and C). In papain, the P1 residue main-chain carbonyl group is placed into the oxyanion hole, whereas its side chain points upward into bulk solvent. This renders little substrate specificity at the P1 residue. Mutational analysis of AvrPphB and PBS1 suggested that AvrPphB has strong preferences for the P3 and P2 residues (see below), but less selectivity for the P1 residue. Although a Lys residue is present at the P1 site in both the AvrPphB precursor and PBS1 proteins (Fig. 5A), substitution of Lys-62 with alanine at the P1 position in AvrPphB had no effect on AvrPphB autoproteolytic cleavage (data not shown). Similarly, mutation of the corresponding Lys-243 in PBS1 showed only a minor effect on its cleavage by AvrPphB (10). Therefore, it is quite possible that the P1 residue for AvrPphB may also have its main chain located in the S1 site in a similar fashion to that of papain, whereas its side chain points out toward the solvent.

Fig. 5.

The structural basis of AvrPphB substrate specificity. (A) Sequence comparison of AvrPphB cleavage sites in its precursor and substrate PBS1 protein. A common Gly-Asp-Lys motif preceding the cleavage sites is highlighted in red, and the arrow indicates the cleavage sites. (B and C) Active site clefts of papain-like enzyme and AvrPphB. Orientation is the same as in Fig. 4A. B shows the molecular surface of cruzain (PDB ID code 2aim), and C shows that of AvrPphB. The structure of cruzain was determined with the inhibitor benzoyl-Arg-Ala-fluoromethyl ketone, which occupies the S3, S2, and S1 sites and is shown in B Left as a CPK representation. C Right is a zoom-in view of the proposed active site of AvrPphB. The proposed S2 site residue (Arg-205) and four catalytically important residues are drawn underneath the molecular surface. Note the positive character of S2 and shallowness of S3 at the substrate-binding site. All surfaces are colored based on the electrostatic potential of the molecule (ranging from -23 to +23 kT). Images were prepared with the program grasp (53).

For members of the papain family, an acidic residue is often found within the S2 site, which interacts with basic residues in the P2 position of substrates. For example, a glutamic acid residue occupies the S2 site in cathepsin B and cruzain (two papain family members), and its negative side chain interacts with a positively charged arginine at the P2 site of their respective substrates (38, 49) (Fig. 5B). When the AvrPphB structure is superimposed onto the structure of cruzain bound to its substrate-mimicking inhibitor (benzoyl-Arg-Ala-fluoromethyl ketone), Arg-205 is found to occupy the S2 site (Fig. 5C). The putative S2 site in AvrPphB is rather deep and has an overall positive charge. This finding is consistent with our observation that AvrPphB prefers an aspartic acid at the substrate P2 site; mutation of Asp-242 in PBS1 (the P2 residue) drastically increases its resistance to cleavage by AvrPphB (10).

The S3 site of AvrPphB has a strong preference for a glycine residue as mutation of Gly-241 in the cleavage site of PBS1 (the P3 residue) to alanine almost completely blocks its cleavage (10). The structural comparison between AvrPphB and cruzain, as shown in Fig. 5 B and C, suggests that Leu-142, Tyr-138, and Phe-154 form the S3 site. All three residues have bulky hydrophobic side chains, which makes the S3 site a very shallow surface. Such a shallow and hydrophobic surface may not allow binding of any amino acids other than glycine.

Although the proteolytic triad is invariant in the entire YopT family of proteases, residues corresponding to the substrate-binding sites in AvrPphB, particularly the S2 and S3 sites, appear to be divergent among the family members (Fig. 2). This finding suggests that different YopT family member proteases have different substrate specificities and thus target different host proteins. Indeed, YopT from Yersinia specifically cleaves the prenylated Rho GTPases. A polybasic sequence upstream of the cleavage site, as well as an isoprenoid moiety in the substrate, plays a crucial role in dictating the cleavage specificity (47). Moreover, Y4zC from Rhizobium fails to cleave PBS1, although it also undergoes autoproteolytic cleavage (F.S. and J.E.D., unpublished data). Consistent with this observation, the site of autoproteolytic cleavage in Y4zC differs from AvrPphB, with the P1, P2, and P3 sites being occupied by Met, Lys, and Asp, respectively, rather than Lys, Asp, and Gly (F.S. and J.E.D., unpublished data).

A recent psi-blast search identifies seven Avr genes of the YopT family of cysteine proteases from the plant pathogens P. syringae and Ralstonia and three from plant symbiotic bacteria Rhizobium and Bradyrhizobium. The list will continue to grow with the ongoing efforts of microbial genome sequencing. In addition, multiple Avr proteases of the YopT family can be found in a single Pseudomonas strain. For example, HopPtoN and HopPtoC are both YopT family members and are found in P. syringae pv. tomato strain DC3000 (50). Our structural analysis suggests that these putative Avr proteases will likely target different substrates in the plant host, or possibly cleave the same substrates at different positions, generating different cleavage signals or products. Detection of these different proteases would thus be expected to require distinct R proteins. The large number of YopT-like proteins found in plant pathogens may reflect coevolutionary pressures in which the evolution of new R proteins in the host that detect the cleavage products of a given protease selects for pathogens with new protease variants.

In summary, we have shown that AvrPphB has an overall protein fold that places it within the papain-like cysteine protease superfamily, although it clearly differs from the typical papain fold. This structural similarity includes the three catalytically important residues with a structural arrangement that is superimposable to that of papain-like proteases. It is thus expected that AvrPphB uses a similar mechanism in catalyzing the proteolysis. The similarity between AvrPphB and other papain-like proteases also suggests a likely substrate-binding site that might explain the distinct substrate specificity of AvrPphB among the YopT family members. Our structural analysis of the specificity of Avr proteases provides a partial explanation for the one-to-one relationship between Avr genes and corresponding R genes, a hallmark of the plant disease resistance mechanism. Our study also begins to advance our understanding of plant disease resistance mechanisms to the structural level.