Cleavage of Arabidopsis PBS1 by a Bacterial Type III Effector

See allHide authors and affiliations

Science  29 Aug 2003:
Vol. 301, Issue 5637, pp. 1230-1233
DOI: 10.1126/science.1085671


Plant disease-resistance (R) proteins are thought to function as receptors for ligands produced directly or indirectly by pathogen avirulence (Avr) proteins. The biochemical functions of most Avr proteins are unknown, and the mechanisms by which they activate R proteins have not been determined. In Arabidopsis, resistance to Pseudomonas syringae strains expressing AvrPphB requires RPS5, a member of the class of R proteins that have a predicted nucleotide-binding site and leucine-rich repeats, and PBS1, a protein kinase. AvrPphB was found to proteolytically cleave PBS1, and this cleavage was required for RPS5-mediated resistance, which indicates that AvrPphB is detected indirectly via its enzymatic activity.

The molecular mechanisms by which pathogens trigger disease resistance in plants are poorly understood. The “gene-for-gene” hypothesis was proposed more than 40 years ago by Flor, whereby plant disease resistance occurs only in the simultaneous presence of a specific resistance (R) gene in the plant and its corresponding avirulence (Avr) gene in the pathogen (1), which suggests that a direct Avr-R interaction activates disease resistance. Isolation of more than 30 R genes from different plant species has revealed that most encode proteins containing a predicted nucleotide-binding site and leucine-rich repeats (NB-LRR) (2). Although remarkable progress has been made in cloning corresponding pairs of NB-LRR genes and Avr genes (2), a direct physical interaction between NB-LRR proteins and pathogen Avr proteins has only been shown twice (3, 4). Because several Avr genes are known to enhance pathogen virulence in susceptible plants (5), an alternative model has been proposed whereby Avr proteins interact with and modify a specific target (or targets) in the plant to promote disease in the absence of their corresponding R proteins (2, 6). R proteins serve to “guard” these virulence targets and thereby activate resistance signaling pathways when these targets are modified by specific Avr factors. Although evidence supporting this guard hypothesis is emerging (79), little is known about the targets of pathogen Avr proteins or how these targets are modified.

We have shown that the AvrPphB protein of Pseudomonas syringae belongs to a novel family of cysteine proteases and that its protease activity is required for induction of a hypersensitive response (HR, a diseaseresistance response characterized by localized cell death at the point of pathogen attack) in Arabidopsis plants that carry the NB-LRR gene RPS5 (10, 11). AvrPphB is likely secreted into host cells by the P. syringae type III secretion system, because mutations in this system block AvrPphB-induced resistance in bean plants (12). Consistent with this, transgenic expression of AvrPphB in the cytoplasm of Arabidopsis induces an RPS5-mediated HR (13). In the context of the guard model, a plant substrate of the AvrPphB protease may activate RPS5 upon cleavage by AvrPphB. A candidate for this AvrPphB substrate is the Arabidopsis PBS1 protein kinase, which is specifically required for AvrPphB/RPS5-mediated resistance (13, 14).

To test this hypothesis, we coexpressed AvrPphB and a hemagglutinin (HA)-tagged form of PBS1 in Arabidopsis leaves using an Agrobacterium-mediated transient transformation system (15). Coexpression of PBS1-HA and AvrPphB(C98S), a protease inactive mutant of AvrPphB in which Cys98 is replaced by Ser (10, 16), in pbs1-1 mutant plants generated full-length PBS1-HA protein (60 kD, Fig. 1A). Expression of wild-type AvrPphB with PBS1-HA markedly reduced the 60-kD band intensity, which suggests that PBS1 is degraded when elicited with AvrPphB.

Fig. 1.

AvrPphB induces PBS1 cleavage in planta, independent of RPS5, AtRAR1, and PBS1 kinase activity. (A) Dexamethasone-inducible constructs of PBS1-HA, AvrPphB, and a protease-inactive mutant form of AvrPphB (C98S) were transiently coexpressed in pbs1-1, rar1-20, and rps5-2 mutants of Arabidopsis accession Col-0 and in the rps5 null accession Ler (15, 30). Total protein extracts were subjected to HA-specific antibody and immunoblot analysis. (B) AvrPphB-dependent cleavage of PBS1 occurs in tobacco, irrespective of PBS1 kinase activity. PBS1-HA, PBS1(G252R)-HA, PBS1(K115N)HA, AvrPphB, and C98S were transiently expressed in tobacco leaves as indicated, and total protein extracts were analyzed as in (A).

RPS5-mediated resistance also requires the RAR1 gene (17), which has been hypothesized to function in a proteasome-dependent protein degradation pathway (18). We tested whether RAR1 and RPS5 are required for the degradation of PBS1 by coexpressing AvrPphB and PBS1-HA in rar1-20 and rps5-2 Col-0 mutants, and in the Landsberg erecta (Ler) accession, which lacks RPS5. A 28-kD HAcontaining band was detected in these lines (Fig. 1A), which indicates that AvrPphB causes PBS1 cleavage in Arabidopsis, independent of RAR1 and RPS5. The absence of this band in the pbs1-1 line is likely caused by nonspecific degradation resulting from activation of the HR. Given the susceptibility of rps5 mutants, these data indicate that cleavage of PBS1 is insufficient to induce disease resistance in the absence of RPS5, which suggests that RPS5 may function downstream or independent of the cleavage event.

If PBS1 is a direct substrate of AvrPphB in planta, cleavage of PBS1 should be independent of host cell background. To test this, we coexpressed PBS1 and AvrPphB in tobacco (Nicotiana tabacum), which belongs to a plant family (Solanaceae) different from that of Arabidopsis (Brassicaceae). PBS1 was cleaved in tobacco in the presence of wild-type AvrPphB, but not AvrPphB(C98S) (Fig. 1B). Previously, we identified a mutant PBS1 allele, pbs1-2, that blocks AvrPphB-induced resistance in Arabidopsis (14). Because the protein encoded by this allele, PBS1(G252R) (16), lacks kinase activity, we asked whether PBS1 kinase function was required for cleavage. PBS1(G252R) was cleaved by AvrPphB in tobacco (Fig. 1B), which indicated that AvrPphB cleavage of PBS1 occurs without PBS1 kinase activity. This conclusion was confirmed by the cleavage of PBS1(K115N), which is mutated in the ATP-binding site and lacks kinase activity (Fig. 1B and fig. S1).

We tested whether AvrPphB and PBS1 physically interact in planta using a coimmunoprecipitation assay (15). We immunoprecipitated total protein extracts of Nicotiana benthamiana, transiently transformed as above, with an HA-specific antibody, and we analyzed immune complexes using an AvrPphB-specific antibody. The protease inactive and wild-type forms of AvrPphB immunoprecipitated with full-length PBS1 and the C-terminal cleavage product of PBS1, respectively (Fig. 2A). This shows that AvrPphB associates with a PBS1-containing complex in N. benthamiana.

Fig. 2.

PBS1 is a substrate of AvrPphB. (A) PBS1 and AvrPphB coimmunoprecipitate in planta. The total protein extracts described in Fig. 1B were immunoprecipitated using an HA antibody affinity matrix (15). The immuno-complexes and the total extracts were subjected to immunoblotting with an HA-specific monoclonal antibody (top) and an AvrPphB-specific polyclonal antibody (bottom). (B) In vitro cleavage of PBS1 by AvrPphB. Partially purified GST-PBS1(G252R)-His6 protein was incubated with purified AvrPphB-His6, or AvrPphB(C98S)-His6, or in blank buffer. Reactions were separated in an SDS-polyacrylamide gel electrophoresis gel followed by Coomassie blue staining (left). Immunoblots from the same samples were made with GST antibodies (middle) and His antibodies (right).

To address whether any other plant proteins are required for cleavage of PBS1 by AvrPphB, we simultaneously overexpressed epitope-tagged forms of AvrPphB and PBS1 in mammalian HEK 293T cells. AvrPphB, but not AvrPphB(C98S), cleaved PBS1 into a 28-kD C-terminal product and a 32-kD N-terminal product (fig. S2). The same PBS1 cleavage products were obtained when purified recombinant AvrPphB protein was incubated with PBS1 synthesized in a wheat germ-derived in vitro transcription and translation system (fig. S3). To confirm that PBS1 is a direct substrate of AvrPphB and that no other proteins are required for cleavage, we overexpressed recombinant proteins of PBS1 [GST-PBS1(G252R)-His6] and AvrPphB (AvrPphB-His6) in E. coli, and mixed the purified proteins in vitro. In the presence of wild-type AvrPphB, but not AvrPphB(C98S) or buffer alone, PBS1 was cleaved into two fragments (asterisks in Fig. 2B).

AvrPphB encodes a 35-kD protein that cleaves itself between Lys62 and Gly63, uncovering a myristoylation motif (10, 19, 20). To identify potential cleavage sites in PBS1, we compared the PBS1 sequence to the AvrPphB sequence flanking the autocleavage site (Fig. 3A). A stretch of three amino acids immediately preceding the AvrPphB autocleavage site was found in PBS1 at a position consistent with the size of the PBS1 cleavage products (G241D242K243), which suggests that AvrPphB may cleave PBS1 after K243 (Fig. 3A). This was confirmed by N-terminal Edman sequencing of the C-terminal cleavage product of PBS1 obtained in vitro (Fig. 2B). By aligning both sequences around the GDK motif, we identified seven identical amino acids potentially required for AvrPphB cleavage of PBS1 (Fig. 3A). We tested whether each of these amino acids was required for cleavage by individually mutating each residue in PBS1 to alanine. Each PBS1 mutant was assayed in wheat-germ extracts. G241A, D242A, and K243A mutations reduced PBS1 cleavage by about 90, 75, and 15%, respectively, without affecting kinase activity (fig. S4); the double mutant PBS1(G241A/D242A) was completely resistant to cleavage by AvrPphB (21). All other mutations tested either had no effect on cleavage by AvrPphB, or they had an effect but disrupted PBS1 kinase activity, which may indicate gross structural changes (fig. S4). The importance of the G241D242K243 triad for cleavage by AvrPphB was corroborated in planta, as a PBS1(G241S/D242A/K243A) (PBS1-GDK) mutant was completely resistant to cleavage by AvrPphB when expressed in N. benthamiana (Fig. 3B).

Fig. 3.

Cleavage-site specificity of the AvrPphB protease and the role of cleavage in RPS5-mediated disease resistance. (A) Alignment of the AvrPphB and PBS1 protein sequences in the region containing the AvrPphB autoprocessing site. Identical residues are indicated in red, and residues individually mutated to alanine in fig. S4 are underlined. The cleavage sites are indicated by an arrow. (B) Mutation of the GDK motif of PBS1 blocks cleavage in planta. PBS1-GDK-HA and PBS1-HA were transiently expressed with AvrPphB or AvrPphB(C98S) in N. benthamiana. Protein samples were processed as in Fig. 1A (C). PBS1 cleavage and PBS1 kinase activity are independently required for RPS5-mediated resistance. Leaves of 4-week-old pbs1-1 transgenic plants expressing the indicated genes were photographed 20 hours after injection with P. syringae expressing AvrPphB. Collapsed sections indicate an HR, thus complementation of pbs1-1. (D) AvrPphB cleaves neither RPS5 nor two PBS1-homologous kinases from Arabidopsis. The Arabidopsis RPS5, At5g02800, and At3g17410 genes, along with PBS1 control, were in vitro transcribed and translated in the wheat-germ extract system and treated with recombinant AvrPphB (15).

We tested the ability of the PBS1-GDK mutant gene to complement pbs1-1 plants. In contrast to plants transformed with the PBS1 wild-type gene, transgenic pbs1-1 plants expressing PBS1-GDK were unable to induce an HR in response to AvrPphB (Fig. 3C). The kinase inactive PBS1(K115N) mutant also failed to complement pbs1-1. Because the PBS1-GDK protein retains kinase function (fig. S1) and PBS1(K115N) is cleavable (Fig. 1B), we conclude that PBS1 kinase activity and PBS1 cleavage are independently required for RPS5-mediated resistance.

AvrPphB cleaves PBS1 in the activation segment of the kinase domain, a regulatory region conserved in most protein kinases. To assess whether AvrPphB could cleave related protein kinases, we tested the cleavage of At5g02800, an Arabidopsis kinase similar to PBS1 (82% amino acid identity in the kinase domain), which contains all the amino acids shown to be important for cleavage of PBS1 by AvrPphB. We also tested At3g17410, which is less similar to PBS1 (49% identity in the kinase domain), but a possible ortholog of Pti1, a tomato kinase implicated in the recognition of a different bacterial effector (AvrPto) (22). Neither kinase was processed by AvrPphB in wheat-germ extracts (Fig. 3D). This shows that AvrPphB specifically targets PBS1, but not its homologs, and that the G241D242K243 triad in PBS1 is necessary but not sufficient for its cleavage by AvrPphB. On the basis of genetic analyses (14), RPS5 could also be a substrate for AvrPphB. However, RPS5 was resistant to AvrPphB protease activity when assayed in wheat-germ extracts (Fig. 3D).

Together, these results suggest a new model for the specific activation of the RPS5-mediated resistance by AvrPphB (fig. S5). We propose that it is the cleavage of PBS1 that activates RPS5. Because pbs1 null mutants are susceptible to P. syringae carrying AvrPphB (13, 14), the elimination of PBS1 is insufficient to induce resistance. Therefore, we propose that one of the cleavage products of PBS1, possibly in a complex with AvrPphB, binds to and activates RPS5. We suggest this cleavage product must be phosphorylated via PBS1 auto-phosphorylation activity to account for the fact that PBS1 kinase activity is required for RPS5 resistance, but not for cleavage by AvrPphB. Our model for RPS5 activation differs from other models recently proposed under the guard hypothesis (79, 23), which suggests that different R proteins are activated by different mechanisms or that intermediate signaling steps are still to be identified that would lead to a similar mode of activation.

A key common point in these models is that pathogen virulence proteins are recognized as a consequence of their virulence function, rather than by direct interaction with a plant R protein. Such indirect recognition would be expected to significantly constrain the coevolution of pathogens and their plant hosts, as evasion of detection would require modification of virulence functions. Furthermore, by recognizing pathogen virulence proteins based on their enzymatic activity rather than their shape, plants could likely detect a large number of pathogen effectors with a limited number of R proteins. Although the extent of this mechanism is unknown [only three R genes are known to have dual specificity (2426)], this seems an attractive strategy for the plant to maximize its surveillance capacity against the multitude of potential pathogen effectors. This may be a critical aspect of the plant immune system, as, unlike vertebrates, plants cannot generate a diversity of antibodies via somatic recombination.

AvrPphB belongs to a family of cysteine proteases found in both animal and plant pathogens (10), at least one of which, AvrPpiC2, is also known to induce resistance responses in specific plant genotypes (27). Several other bacterial Avr proteins display structural similarity to a second family of cysteine proteases (the YopJ family), and mutations in the putative catalytic residues of one of these, AvrBsT, abolish avirulence activity (28). Similarly, the fungal Avr protein AvrPita is homologous to known metalloproteases, and mutations in its putative catalytic residues also abolish avirulence activity (29). In light of our data, these observations suggest that proteolysis of host target proteins may be a common trigger for many plant R gene pathways.

Supporting Online Material

Materials and Methods

Figs. S1 to S5


References and Notes

Stay Connected to Science

Navigate This Article