Activation of a Phytopathogenic Bacterial Effector Protein by a Eukaryotic Cyclophilin

See allHide authors and affiliations

Science  22 Apr 2005:
Vol. 308, Issue 5721, pp. 548-550
DOI: 10.1126/science.1108633


Innate immunity in higher plants invokes a sophisticated surveillance system capable of recognizing bacterial effector proteins. In Arabidopsis, resistance to infection by strains of Pseudomonas syringae expressing the effector AvrRpt2 requires the plant resistance protein RPS2. AvrRpt2 was identified as a putative cysteine protease that results in the elimination of the Arabidopsis protein RIN4. RIN4 cleavage serves as a signal to activate RPS2-mediated resistance. AvrRpt2 is delivered into the plant cell, where it is activated by a eukaryotic factor that we identify as cyclophilin. This activation of AvrRpt2 is necessary for protease activity. Active AvrRpt2 can then directly cleave RIN4.

Plants have evolved host defense mechanisms to combat infection against a wide variety of pathogens. A gene-for-gene hypothesis was proposed to explain the plant resistance response; the hypothesis states that for a plant to be resistant to a particular pathogen, there must be matching pairs of host resistance and pathogen effector genes (1). Although more than 50 resistance genes have been isolated, direct binding between a resistance gene product and its corresponding pathogen effector has been demonstrated in only a few cases (24). Recent experimental evidence suggests an indirect mode of bacterial pathogen recognition whereby a resistance protein recognizes the biochemical alteration of a second plant protein by a bacterial effector (58).

Delivery of the bacterial effector AvrRpt2 by Pseudomonas syringae specifically induces disease resistance in Arabidopsis plants expressing the RPS2 resistance protein. RPS2 and RIN4 physically associate in Arabidopsis, and delivery of AvrRpt2 results in RIN4 elimination during pathogenesis, which supports the hypothesis that AvrRpt2 directly cleaves RIN4, thus indirectly activating RPS2 (6, 7). AvrRpt2 encodes a 28-kD effector protein delivered into plant cells during infection, where it is cleaved near its amino terminus (9). The 21-kD carboxyterminal product of AvrRpt2 is sufficient to trigger RIN4 elimination and RPS2 activation in planta. Amino acids 1 to 71 encode a type III secretion signal, whereas amino acids 72 to 255 encode the effector domain of AvrRpt2 (9). Although AvrRpt2 does not share extensive primary sequence homology to proteins of known biochemical function, the predicted secondary sequence structure of AvrRpt280–255 matches the catalytic core of the staphopain cysteine protease, with a putative catalytic triad composed of cysteine 122, histidine 208, and aspartate 226 (10). Mutations in any of these residues abolish AvrRpt2 processing, RIN4 elimination, and RPS2 recognition in Arabidopsis (10).

AvrRpt2 is not N-terminally processed in Escherichia coli or P. syringae but is processed in all eukaryotic extracts tested to date, including Saccharomyces cerevisiae (Fig. 1, A and B) (9, 11). We hypothesize that AvrRpt2 is delivered as an inactive protease and is activated upon delivery into the plant cell (9, 11). Protein extracts were prepared from Arabidopsis and S. cerevisiae and added to purified recombinant AvrRpt2 with an N-terminal hexahistidine and a C-terminal hemagglutin (HA) tag. Both crude extracts resulted in specific AvrRpt2:HA N-terminal processing (Fig. 1A). The AvrRpt2 mutant protein AvrRpt2:HA(C122A), in which alanine is substituted for the catalytic cysteine residue, was not processed (Fig. 1B). We detected two additional cleavage products after incubation with crude protein extracts (Fig. 1, A and B). These bands are likely the result of nonspecific protease activity, because they disappear upon further biochemical purification. Extensive dialysis and protein precipitation did not eliminate processing, which suggests that the factor is proteinaceous (11, 12).

Fig. 1.

Biochemical purification of AvrRpt2's eukaryotic activator. (A and B) Recombinant AvrRpt2:HA and AvrRpt2:HA(C122A) proteins were incubated with crude S. cerevisaie BY4730 or Arabidopsis Col-0 protein extract and subjected to immunoblot with monoclonal antibody to HA (anti-HA). Lane 1, nΔ71AvrRpt2:HA expressed in planta. (C) Biochemical purification of AvrRpt2's activator from S. cerevisiae BY4730 cytosol by hydrophobic interaction, anion exchange, and gel-filtration chromatography. Gel-filtration fractions surrounding activity were loaded onto a SDS-PAGE gel, stained by colloidal Coomassie blue (top panel), and subjected to anti-HA immunoblot (bottom panel) to visualize nΔ71AvrRpt2:HA. (D) Processing of AvrRpt2 inside S. cerevisiae mutants. AvrRpt2 and AvrRpt2(C122A) were expressed from the pESC vector inside S. cerevisiae BY4730 (lanes 1 and 2, wild type), KDY75.3b (lane 3, Δcpr1,6,7, Δfpr1), and KDY98.4a (lanes 4 and 5, Δcpr1-8, Δfpr1-4) and subjected to polyclonal anti-AvrRpt2 immunoblot. Cyclophilin mutants were complemented with Arabidopsis ROC1 coexpressed with AvrRpt2.

A biochemical approach was employed to identify the factor responsible for AvrRpt2 activation. Because S. cerevisiae's genome is sequenced and contains only 6,000 genes and because there are many mutant lines available, identification of the factor in S. cerevisiae would be less complicated than pursuing similar experiments in Arabidopsis. Therefore, AvrRpt2's eukaryotic activator was purified from S. cerevisiae. Proteins were sequentially fractionated by hydrophobic interaction, anion exchange, and gel-filtration chromatography. Fractions were assayed for their ability to induce AvrRpt2 processing. Those possessing activity from the final purification step were analyzed by SDS–polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 1C). A single 18-kD protein correlated with activity. Mass spectrometry identified this protein as CPR1 (P14832), a single-domain cyclophilin peptidyl-prolyl cis-trans isomerase (PPIase) (fig. S1).

PPIases are chaperones and folding catalysts with the ability to catalyze the cis-trans isomerization of prolyl bonds, a rate-limiting step in protein folding (13). PPIases span three structurally unrelated protein families: the cyclophilins, FKPBs, and paruvulins. S. cerevisiae possesses eight cyclophilins (14). CPR1 is the yeast homolog of human cyclophilin A, localized in the cytosol and nucleus, and the primary target of the immunosuppressant drug cyclosporin A (14, 15). There are seven Arabidopsis cytosolic single-domain cyclophilins with amino acid sequence homology ranging from 43 to 58% identity and 59 to 69% similarity to CPR1 (fig. S2). The closest Arabidopsis CPR1 homolog is ROC1 (At4 g38740). To verify that cyclophilin is the eukaryotic activator, AvrRpt2 processing in S. cerevisiae cyclophilin mutants was investigated. The S. cerevisiae cytosolic cyclophilin knockout did not abolish processing when AvrRpt2 was expressed in yeast, although processing was reduced in comparison with wild-type cells, which suggests that multiple cyclophilins may facilitate processing (Fig. 1D, lane 3). Yeast mutants lacking all eight cyclophilins and four FKBPs were unable to induce processing, and the inability of AvrRpt2 to process was complemented when AvrRpt2 and ROC1 were coexpressed (Fig. 1D). This verifies that the eukaryotic cofactor is cyclophilin in vivo and that ROC1 can complement CPR1 with respect to AvrRpt2 activation.

To determine whether purified CPR1 and ROC1 would enable AvrRpt2 N-terminal processing in vitro, both proteins were expressed in E. coli with hexahistidine N-terminal tags and purified by affinity and gel-filtration chromatography. Both CPR1 and ROC1 enabled AvrRpt2:HA processing in vitro, and mutant AvrRpt2:HA(C122A) protein was unable to process when incubated with cyclophilin (Fig. 2A). The addition of 10 μM cyclosporin A (cyclophilin inhibitor) to the in vitro reaction abolished cleavage, whereas the addition of 10 μM rapamycin (FKBP inhibitor) had no effect (Fig. 2B). The N- and C-terminal portions of AvrRpt2 were digested with trypsin and glutamyl endopeptidase, respectively, and subjected to tandem mass spectrometry. We were able to sequence up to G71 from the N-terminal fragment and up to G72 from the C-terminal fragment, verifying appropriate and specific AvrRpt2 cleavage (Fig. 2C and fig. S3). These results provide the first biochemical evidence that AvrRpt2 possesses in vitro cysteine protease activity and that AvrRpt2 activation by cyclophilin induces self-cleavage between G71 and G72. N-terminal processing of AvrRpt2 may be necessary for protease activity or proper subcellular localization within the plant cell (11). Previous reports have shown that AvrRpt2 processing mutants are localized to the chloroplast rather than the membrane by monitoring AvrRpt2:GFP accumulation in protoplasts, which indicates that N-terminal processing may be required for proper localization to the membrane (7, 11). However, the role of AvrRpt2's N terminus in its localization has yet to be unambiguously established.

Fig. 2.

Cleavage site specificity of the AvrRpt2 protease. (A) Recombinant Arabidopsis and S. cerevisiae cyclophilins induce AvrRpt2 processing in vitro. Recombinant AvrRpt2:HA or AvrRpt2:HA(C122A) were incubated with ROC1 and CPR1 and subjected to anti-HA immunoblot. Lane 1, nΔ71AvrRpt2:HA expressed in planta. (B) Cyclosporin A inhibits AvrRpt2:HA activation by ROC1. Recombinant AvrRpt2:HA was incubated with ROC1, cyclosporin A (Cys A), and rapamycin and subjected to anti-HA immunoblot. Lane 1, nΔ71AvrRpt2:HA expressed in planta. (C) AvrRpt2 self-processes between G71 and G72. Recombinant AvrRpt2:HA, ROC1, ROC7, and CPR1 were incubated in various combinations and visualized by Coomassie blue. Full-length AvrRpt2(1) was cleaved in the presence of cyclophilin, generating the C terminus of AvrRpt2(2) and the N terminus of AvrRpt2(3). (D) AvrRpt2 and ROC1 directly interact. Recombinant AvrRpt2:HA proteins were incubated with GST:ROC1 and GST alone. GST interacting proteins were recovered by incubation with G beads. Input controls (lanes 1 to 4) and GST binding proteins (IPs) were detected by anti-HA (top panel) and anti-GST (bottom panel) immunoblots.

To verify that cyclophilin and AvrRpt2 directly interact, ROC1 was expressed in E. coli as a glutathione S-transferase (GST) fusion protein. GST:ROC1 was incubated with AvrRpt2:HA and mutant AvrRpt2:HA(C122A) protein. Western blot analysis demonstrated that AvrRpt2:HA and AvrRpt2:HA(C122A) specifically bind to GST:ROC1 but not to GST alone (Fig. 2D). Only full-length AvrRpt2 interacts with GST:ROC1 in vitro, which suggests that this interaction is transient and that after AvrRpt2 is properly folded, the two proteins may disassociate in vitro.

Elimination of the Arabidopsis protein RIN4 during infection with P. syringae that express AvrRpt2 led to the hypothesis that RIN4 is a direct substrate for the AvrRpt2 protease (6, 7). A stretch of seven amino acids surrounding AvrRpt2's cleavage site is homologous to two regions within the RIN4 protein, which suggests that AvrRpt2 may cleave RIN4 at two sites (Fig. 3A) (16). To confirm that RIN4 is a direct substrate of AvrRpt2 and that no other plant proteins are required for cleavage, we mixed purified recombinant AvrRpt2:HA, AvrRpt2:HA(C122A), ROC1, and RIN4 proteins in vitro. RIN4 was only cleaved in the presence of wild-type AvrRpt2:HA and ROC1, verifying that AvrRpt2 directly cleaves RIN4 after it is activated by cyclophilin (Fig. 3, B and C). The two RIN4 fragments were identified by mass spectrometric peptide mapping and tandem mass spectrometry as the N terminus of RIN4 and an internal fragment of RIN4 between sites I and II (Fig. 3C and fig. S4B). The C-terminal fragment of RIN4 was undetectable by immunoblot analysis or Coomassie staining. In addition, full-length RIN4 protein was analyzed by mass spectrometry, verifying that the C terminus was intact prior to cleavage by AvrRpt2 (fig. S4A). Recombinant RIN4 mutant proteins [RIN4(F9A), RIN4(F151A), and RIN4(F9A F151A)] were incubated with AvrRpt2 and ROC1, and RIN4 cleavage products were identified by mass spectrometry. RIN4(F9A) was cleaved only at site II, RIN4(F151A) was cleaved only at site I, and RIN4(F9A F151A) was not cleaved by AvrRpt2, indicating that AvrRpt2 directly cleaves RIN4 at two positions in vitro (Fig. 3B and fig. S4, C and D). We were able to identify RIN4 site I as being cleaved between G10 and N11 by tandem mass spectrometry (fig. S5). AvrRpt2 also cleaves itself at a similar position in its recognition site (fig. S3). Taken together, these data suggest that AvrRpt2 cleaves RIN4 site II between G152 and D153.

Fig. 3.

RIN4 is a direct substrate for AvrRpt2. (A) RIN4 possesses two sites with homology to AvrRpt2's processing site. RIN4 site I is located between V6 and W12, and site II is between V148 and W154. (B) Diagram of the RIN4 recombinant protein. FG I, site I; FG II, site II. (C) RIN4 is cleaved by activated AvrRpt2 in vitro. Recombinant AvrRpt2:HA, AvrRpt2(C122A):HA, ROC1, and RIN4 proteins were incubated in combinations in vitro, and cleavage products were visualized by Coomassie blue. I indicates RIN4(F9A), II indicates RIN4(F151A), and I&II indicates RIN4(F9A F151A) mutant proteins. RIN4 cleavage products were identified by mass spectrometry. Fragment 1 results from a single cleavage event at either site I or site II, fragment 2 is the internal portion between sites I and II, and fragment 3 is the N-terminal piece of RIN4. The C-terminal fragment of RIN4 is undetectable after AvrRpt2 cleavage at site II in vitro.

Previous experiments indicate that the RIN4 double mutants are also not cleaved in planta after delivery of AvrRpt2 (17, 18). The RIN4(F9A F151A) double mutant is not eliminated when transiently expressed with AvrRpt2 in Nicotiana benthamiana, whereas RIN4 single mutants are undetectable by immunoblot, which suggests that RIN4 is cleaved at two sites in planta (17). Furthermore, the RIN4 site I and II mutant is unable to be eliminated in Arabidopsis upon infection with P. syringae expressing AvrRpt2 (18). Because RIN4 negatively regulates RPS2 (6, 7), RPS2-mediated resistance does not occur in Arabidopsis plants expressing the RIN4 double mutant upon infection with Pseudomonas expressing AvrRpt2 (18).

Together, these results suggest a novel mechanism to explain AvrRpt2's activation upon delivery into the plant cell. Spontaneous peptidyl-prolyl isomerization is a slow reaction and can constitute rate-limiting steps in protein folding. The discovery that eukaryotic cyclophilin activates AvrRpt2 not only provides a mechanism for effector activation upon delivery into the plant cell but also implicates Arabidopsis single-domain cyclophilins as general protein-folding catalysts.

We propose that the PPIase activity of cyclophilin activates AvrRpt2, enabling it to cleave its N terminus, localize to the membrane where RPS2 and RIN4 reside, and directly cleave RIN4 (5, 6). RIN4 is a negative regulator of RPS2 (6, 7). It is likely that the cleavage of RIN4 activates RPS2, because Arabidopsis rin4 null mutants are lethal in an RPS2 background (6) and because the overexpression of RIN4 delays RPS2 activation (6, 7). Growth of Pseudomonas expressing AvrRpt2 has been shown to be significantly higher than that in catalytically inactive mutants on Arabidopsis rin4/rps2 plants (17). These data suggest that, in addition to RIN4, AvrRpt2 likely has additional protein targets inside the plant cell that account for its virulence activity. A set of Arabidopsis proteins possessing homology to AvrRpt2's processing site was identified that are cleaved by AvrRpt2 when both proteins are transiently expressed in N. benthamiana, and these proteins may be virulence targets of the AvrRpt2 protease (17). Targeting of multiple host proteins may be a common strategy employed by a number of effectors, and future studies will investigate cleavage of this set of proteins in Arabidopsis and their potential roles in susceptibility and resistance.

A common theme in host and pathogen proteinases is that the expression of proteolytic activity is tightly regulated, both at the level of expression and secretion and in the processing of an inactive secreted precursor to its active form (19, 20). There are 6 proline residues in the prodomain and 12 residues within the effector domain of AvrRpt2 that may be important for interacting with cyclophilin. A number of bacterial effectors delivered into plant and animal cells during pathogenesis are structurally similar to cysteine proteases (21, 22). Furthermore, enzyme activity of several putative bacterial effector proteases cannot be detected in vitro, which indicates that the effectors YopJ, AvrBsT, and AvrXv4 may also require host factors for activation (22). In light of our data, folding of bacterial effector proteases by eukaryotic protein factors may be a common mechanism during pathogenesis.

Supporting Online Material

Materials and Methods

Figs. S1 to S5

References and Notes

References and Notes

View Abstract

Navigate This Article