The Gram-negative plant pathogenic bacterium Xanthomonas campestris pv. vesicatoria translocates effector proteins via a specialized type III secretion (TTS) system into the host cell cytosol. The efficient secretion of many effector proteins depends on the global TTS chaperone HpaB. Here, we identified a novel export control protein, HpaC, which significantly contributes to bacterial pathogenicity. Deletion of hpaC leads to a severe reduction in secretion of effector proteins and the putative type III translocon proteins HrpF and XopA. By contrast, secretion of the TTS pilus protein HrpE is not affected. We provide experimental evidence that HpaC differentiates between two classes of effector proteins. Using an in vivo reporter assay, we found that HpaC specifically promotes the translocation of the effector proteins XopJ and XopF1 into the plant cell, whereas AvrBs3 and XopC are efficiently translocated even in the absence of HpaC. Similar findings were obtained for HpaB. Inhibition of protein synthesis suggests that HpaB is involved in the secretion of stored effector proteins. Furthermore, protein–protein interaction studies revealed that HpaB and HpaC form an oligomeric protein complex and that they interact with members of both effector protein classes and the conserved TTS system component HrcV. Taken together, our data indicate that HpaB and HpaC play a central role in recruiting TTS substrates to the secretion apparatus.
Many Gram-negative plant and animal pathogenic bacteria depend on a specialized type III protein secretion (TTS) system to invade successfully their eukaryotic host organisms. This highly conserved secretion apparatus spans both bacterial membranes and is associated with extracellular needle- or pilus-like appendages that presumably serve as transport channels for secreted proteins (Hueck, 1998; He et al., 2004). The TTS system mediates protein secretion into the extracellular milieu and the translocation of effector proteins into the host cell cytosol. Type III effectors presumably interfere with a variety of host cellular processes such as cytoskeletal rearrangements, vesicle trafficking or defence responses to the pathogen's benefit (Chang et al., 2004; Espinosa and Alfano, 2004; Mudgett, 2005; Navarro et al., 2005). Translocation of effector proteins requires the type III translocon, a predicted bacterial protein channel that inserts into the host plasma membrane (Büttner and Bonas, 2002; Coombes and Finlay, 2005). As mutations in type III translocon proteins result in a complete loss of bacterial pathogenicity, it has been suggested that translocated effector proteins are crucial for the outcome of the host-pathogen interaction (Büttner and Bonas, 2003).
The mechanisms underlying recruitment of effector proteins to the TTS system are far from understood. The signal that targets effector proteins for TTS and translocation presumably resides within the N-terminal 15–20 residues and is not conserved on the amino acid level (Lloyd et al., 2001a; Schechter et al., 2004). In some cases a signal in the 5′ region of the corresponding mRNA is proposed, suggesting cotranslational secretion of these proteins (Anderson and Schneewind, 1997; Anderson et al., 1999; Ramamurthi and Schneewind, 2005). In plant and animal pathogenic bacteria, the efficient secretion and/or stability of a number of TTS substrates also depends on specific cytoplasmic interaction partners, termed TTS chaperones. TTS chaperones are typically small, acidic and leucine-rich proteins that often interact with one particular effector that is encoded next to the chaperone gene (Feldman and Cornelis, 2003; Parsot et al., 2003). It has been proposed that binding of the chaperone keeps the corresponding interaction partner in a secretion-competent and thus partially unfolded conformation (Ghosh, 2004). Based on their substrate specificities, TTS chaperones were grouped into several classes: class IA chaperones interact with one or several homologous effectors whereas class IB chaperones bind to different effectors that do not share any sequence similarity. Chaperones that are specific for translocon proteins are grouped as class II (Parsot et al., 2003).
One of the model organisms to study type III-dependent protein export in plant pathogenic bacteria is Xanthomonas campestris pathovar (pv.) vesicatoria– also termed Xanthomonas axonopodis pv. vesicatoria (Vauterin et al., 2000) – the causal agent of bacterial spot disease in pepper and tomato. The TTS system of X. campestris pv. vesicatoria is essential for the bacteria to grow and cause disease symptoms in susceptible plants and to induce the hypersensitive response (HR), a rapid, programmed cell death at the infection site in resistant plants (Klement, 1982; Bonas et al., 1991). Induction of the HR is triggered by individual bacterial effector proteins – also termed avirulence (Avr) proteins – that are specifically recognized by matching plant disease resistance proteins (Dangl and Jones, 2001).
Nine hrp gene products are conserved in plant and animal pathogenic bacteria and the corresponding genes were therefore designated hrc (for hrp conserved; He, 1998). In X. campestris pv. vesicatoria, most hrc and hrp genes analysed so far are required for type III-dependent protein secretion, suggesting that they are involved in the formation of the membrane-spanning secretion apparatus (Rossier et al., 2000). However, the precise role of individual Hrp proteins was elucidated only for a few of them. Examples are the Hrp pilus protein HrpE and the translocon protein HrpF, both of which are secreted by the TTS system (Rossier et al., 2000; Weber et al., 2005). While HrpE is crucial for secretion of all type III substrates, HrpF is essential for the translocation of effector proteins into the plant cell cytosol. HrpF presumably inserts into the host plasma membrane as part of a channel-like protein complex (Büttner et al., 2002; Szurek et al., 2002). Another putative component of the type III translocon is XopA, which is encoded in the hrp pathogenicity island and is secreted by the TTS system. xopA mutants are impaired in growth in planta and cause reduced reactions in the plant, but TTS in vitro is not affected (Noël et al., 2002).
In addition to XopA, Hrc and Hrp proteins, pathogenicity depends on the cytoplasmic HpaB protein which is encoded within the hrpE operon. HpaB belongs to the class of Hpa (hrp-associated) proteins that often contribute to pathogenicity but are not essential for the HR induction (Huguet et al., 1998; Noël et al., 2002). Deletion of hpaB leads to a complete loss of pathogenicity in susceptible plants but allows the induction of a partial HR in resistant plants (Büttner et al., 2004). HpaB is a leucine-rich protein of 162 amino acids with an acidic pI and thus displays typical features of a TTS chaperone. Recently, protein–protein interaction studies and in vitro secretion analyses revealed that HpaB binds to the effector proteins AvrBs1 and AvrBs3 and promotes secretion of several effectors, suggesting that it acts as a global class IB TTS chaperone (Büttner et al., 2004). In agreement with this hypothesis, the protein-fold recognition program 3D-PSSM (Kelley et al., 2000) predicts structural similarities between HpaB and the TTS chaperones SycE from Yersinia (E-value of 2.03), Spa15 from Shigella flexneri (E-value of 2.04) and SigE from Salmonella enterica (E-value of 2.16).
In this study, we report on a novel Hpa protein from X. campestris pv. vesicatoria, HpaC, which is encoded in the hrp gene cluster. Protein–protein interaction and in vivo translocation studies indicate that HpaC is involved in the export control of type III-dependent protein secretion.
Identification of HpaC as an important virulence factor from X. campestris pv. vesicatoria
Previously, sequence analysis of the hrpC operon from X. campestris pv. vesicatoria led to the identification of an open reading frame (ORF) downstream of hrcV and will hereafter be designated hpaC (accession number CAJ22055; O. Rossier and U. Bonas, unpubl. data; Thieme et al., 2005). The corresponding predicted gene product (213 amino acids) shares homology with putative proteins encoded by the hrp gene clusters from Xanthomonas oryzae pv. oryzae (HpaP, accession number BAD29991, 89% identity, 93% similarity), Ralstonia solanacearum (HpaP, accession number CAD18013, 27% identity, 42% similarity) and Burkholderia pseudomallei (hypothetical protein, accession number CAH38877, 26% identity, 41% similarity). Using 3D-PSSM, we found a predicted structural similarity of HpaC to the C-terminal domain of the Rob transcription factor from Escherichia coli (E-value of 0.265; Huffman and Brennan, 2002).
To investigate a possible contribution of hpaC to pathogenicity of X. campestris pv. vesicatoria, we deleted the gene from the genome of strains 85-10 and 85*. Strain 85* is a derivative of the wild-type strain 85-10 and contains hrpG*, a mutated version of the key regulatory gene hrpG. hrpG* strains express the TTS system constitutively and are key for in vitro secretion analyses (Rossier et al., 1999; Wengelnik et al., 1999). However, in planta growth of hrpG* strains is like wild type (Wengelnik et al., 1999). When bacteria were infiltrated into the susceptible pepper line Early Cal Wonder (ECW), the hpaC deletion mutant 85-10ΔhpaC displayed reduced bacterial growth (approximately 100-fold) and disease symptom formation when compared with the wild-type strain 85-10 (Fig. 1A and B). Furthermore, in the absence of hpaC the HR in the resistant line ECW-10R was reduced (Fig. 1A). ECW-10R plants express the Bs1 resistance gene and trigger the HR upon recognition of the Avr protein AvrBs1 which is delivered by strain 85-10 (Minsavage et al., 1990; Escolar et al., 2001).
The mutant phenotype of strain 85-10ΔhpaC could be complemented for the HR induction and disease symptom formation, respectively, by plasmid pDMhpaC, which encodes a C-terminally c-Myc epitope-tagged derivative of HpaC (Fig. 1A). When bacteria were incubated in secretion medium, HpaC-c-Myc was detected by a monoclonal anti-c-Myc antibody in total protein extracts but not in the culture supernatant of the hrpG* strain 85*, suggesting that HpaC is not secreted (Fig. 1C). We also analysed a chimera comprising the N-terminal 100 amino acids of HpaC and AvrBs3Δ2. AvrBs3Δ2 is a derivative of the effector protein AvrBs3 and is used as a reporter protein to monitor in vivo protein translocation. Because AvrBs3Δ2 lacks amino acids 2-152 and thus the secretion and translocation signal, it is not secreted by the TTS system. However, AvrBs3Δ2 contains the effector domain and thus induces the HR in the resistant pepper line ECW-30R when fused to a functional translocation signal (Minsavage et al., 1990; Szurek et al., 2002; Noël et al., 2003). When analysed for in vitro secretion, HpaC1−100-AvrBs3Δ2 was not detected in the culture supernatant (Fig. 1C). Furthermore, in the more sensitive translocation assay, strain 85* expressing HpaC1−100-AvrBs3Δ2 did not induce the HR in AvrBs3-responsive plants, suggesting that the first 100 amino acids of HpaC do not contain a functional translocation signal.
HpaC is required for the efficient secretion of effector and translocon proteins
Next, we analysed the contribution of HpaC to the type III-dependent protein secretion in vitro. For this, strains 85* and 85*ΔhpaC were incubated in secretion medium and total protein extracts and culture supernatants were analysed by immunoblotting, using antibodies against the putative translocon proteins HrpF and XopA, respectively, and the pilus protein HrpE. The secretion of these proteins was shown to be type III-dependent in previous experiments (Rossier et al., 2000; Noël et al., 2002; Weber et al., 2005). Figure 2A shows that secretion of HrpF and XopA was clearly reduced in strain 85*ΔhpaC when compared with the hpaC wild-type strain 85*. This secretion deficiency could be complemented by plasmid pDMhpaC as shown in Fig. 2B for the secretion of HrpF. In contrast to HrpF and XopA, the secretion of HrpE was not affected in the absence of HpaC (Fig. 2A). This indicates that the observed secretion deficiency in hpaC mutants is probably not due to a general defect in the TTS system. In agreement with this hypothesis, electron microscopy studies revealed that Hrp pilus formation was not affected in hpaC mutant strains (data not shown).
To investigate the influence of HpaC on the secretion of effector proteins, we analysed different AvrBs3Δ2 chimeras that harbour the N termini of AvrBs3 and the recently identified effector proteins XopC, XopJ and XopF1 respectively (Noël et al., 2003; Roden et al., 2004; D. Büttner and U. Bonas, unpubl. data). AvrBs31−200-, XopC1−200- and XopJ1−155-AvrBs3Δ2 chimeras were expressed under control of the respective promoters. To allow detection of the XopF11−200-AvrBs3Δ2 chimera, expression had to be driven by the lac promoter. AvrBs3Δ2 chimeras were introduced into strains 85* and 85*ΔhpaC and bacteria were incubated in secretion medium. Like HrpF and XopA, secretion of all AvrBs3Δ2 chimeras by strain 85*ΔhpaC was significantly reduced when compared with the wild type (Fig. 2C). These findings were confirmed for the native AvrBs3 effector protein which is naturally expressed in X. campestris pv. vesicatoria strain 82-8, but not in strain 85-10. When the hrpG* strains 82* (82-8 hrpG*) and 82*ΔhpaC (82-8 hrpG*ΔhpaC) were incubated in secretion medium, AvrBs3 was detected in the culture supernatant of the wild type but not of the hpaC mutant strain (Fig. 2D). Taken together, these data indicate that HpaC contributes to the efficient secretion of both effector and translocon proteins, but is not required for the export of the Hrp pilus protein HrpE.
Translocation studies in hpaC mutants identify two classes of effector proteins
In addition to in vitro secretion, we also investigated the influence of HpaC on the translocation of different effector proteins. For this, strains 85-10 and 85-10ΔhpaC expressing AvrBs3Δ2 chimeras as described above were inoculated into AvrBs3-responsive pepper plants. All hpaC wild-type derivatives triggered the HR as expected, indicating that in each case the reporter protein was translocated. Furthermore, strain 85-10ΔhpaC expressing AvrBs3- and XopC-AvrBs3Δ2 chimeras, respectively, induced the HR similar to the wild-type strain. In contrast, the HR induction by XopJ- and XopF1-chimeras was significantly reduced in strain 85-10ΔhpaC(Fig. 3A). As shown in Fig. 3B for XopJ1−155-AvrBs3Δ2, the mutant phenotypes could be complemented by ectopic expression of HpaC-c-Myc.
To exclude that the observed HR reduction by XopJ- and XopF1-chimeras delivered by the hpaC mutant was due to intrinsic properties of the protein chimeras that interfered with the recognition of the reporter, we infiltrated serial dilutions of the wild-type strain 85-10 expressing different AvrBs3Δ2 chimeras into AvrBs3-responsive pepper plants. Figure 3C shows that all strains induced a visible HR when infiltrated up to a bacterial density of 2 × 106 cfu ml−1. The HR induction at low inoculation density was more pronounced for XopC- and XopF1-chimeras. This indicates that in both cases the reporter was recognized with similar efficiency (Fig. 3C). We also analysed the stability of AvrBs3Δ2 chimeras in wild-type and hpaC mutant strains. As shown in Fig. 2C, comparable amounts of each AvrBs3Δ2 chimera were present in hpaC wild-type and mutant strains. Thus, the observed HR reduction in hpaC mutants cannot be explained by reduced protein stabilities. It should be noted that both the weakly expressed AvrBs3- and the more abundant XopC-chimera were efficiently translocated in the absence of HpaC. We therefore speculate that HpaC plays a specific role in promoting the translocation of XopJ and XopF1.
HpaC interacts with effector proteins and XopA
The contribution of HpaC to type III-dependent protein secretion and translocation prompted us to investigate possible interactions between HpaC and different TTS substrates including effectors and the putative translocon protein XopA. For this, we performed glutathione S-transferase (GST) pull-down assays. We did not test the interaction between HpaC and HrpF because HrpF binds unspecifically to proteins (D. Büttner and U. Bonas, unpubl. data). XopF1-, AvrBs3-, XopA-GST chimeras and GST alone were expressed in E. coli, immobilized on glutathione sepharose and incubated with E. coli lysates containing HpaC-c-Myc. Figure 4A shows that HpaC-c-Myc specifically bound to GST-XopF1 and GST-AvrBs3 but not to GST alone. Furthermore, HpaC-c-Myc also co-eluted with GST-XopA (Fig. 4B).
To exclude the possibility that HpaC-c-Myc binds non-specifically to the beads, we incubated HpaC-c-Myc-containing lysates with glutathione sepharose alone. As HpaC-c-Myc was not detectable in the eluted protein fractions (data not shown), the observed interactions with GST-chimeras were specific. Taken together, we conclude that HpaC interacts with type III effector proteins and the putative TTS translocon protein XopA and promotes their secretion. Our data suggest a chaperone-like function of HpaC even if the protein does not share any sequence or predicted structural similarities with known TTS chaperones.
HpaC interacts with the global TTS chaperone HpaB
We recently described that the efficient in vitro secretion of different effector proteins depends on HpaB (Büttner et al., 2004). HpaB binds to the effector proteins AvrBs3 and AvrBs1 and presumably acts as a global class IB TTS chaperone. The fact that both HpaB and HpaC contribute to effector protein export prompted us to investigate whether both proteins interact with each other. For this, we expressed a GST-HpaB chimera in E. coli, immobilized it on glutathione sepharose and added an E. coli lysate containing HpaC-c-Myc. Figure 5A shows that HpaC-c-Myc specifically bound to GST-HpaB, but not to GST alone. These results were confirmed by the reciprocal experiment in which GST-HpaC bound to sepharose was incubated with HpaB-c-Myc (Fig. 5B). As HpaB-c-Myc specifically bound to GST-HpaC but not to GST alone, we conclude that HpaB and HpaC interact with each other.
Effector proteins differ in their HpaB-dependent translocation
The fact that HpaC promotes the translocation of one subset of effectors (see above, Fig. 3A) prompted us to investigate whether HpaB plays a similar role during effector protein translocation. We therefore performed in vivo translocation assays with hpaB wild-type and mutant strains expressing AvrBs31−200-, XopC1−200-, XopJ1−155- and XopF11−200-AvrBs3Δ2 chimeras. All chimeras induced the AvrBs3-specific HR when delivered by the wild-type strain. Also, the hpaB deletion mutant expressing the AvrBs3- and XopC-chimeras induced the HR, indicating that the reporter was still translocated in the absence of hpaB(Fig. 6A). However, the HR was reduced when compared with the respective hpaB wild-type strains, suggesting that the efficient export of both proteins depends on HpaB. Interestingly, strain 85*ΔhpaB expressing the XopJ- and XopF1-AvrBs3Δ2 chimeras, respectively, did not induce the AvrBs3-specific HR (Fig. 6A). This is reminiscent of the finding that translocation of both fusions was significantly reduced in the absence of hpaC (Fig. 3A). As shown for the XopF1-AvrBs3Δ2 chimera, the mutant phenotype could be complemented by plasmid pLhrpE which carries the hrpE region downstream of a lac promoter (Fig. 6B; Büttner et al., 2004).
In addition to translocation, we analysed the in vitro secretion of the effector fusions under study. Figure 6C shows that none of the AvrBs3Δ2 chimeras was detectable in the culture supernatant of hpaB mutant strains when bacteria were incubated in secretion medium. This extends our previous findings (Büttner et al., 2004). Secretion of the translocon protein HrpF was not affected in the absence of hpaB (Fig. 6C), suggesting that the lack of effector protein secretion was not due to a general defect of the TTS system.
The observation that the ΔhpaB strain expressing AvrBs3- and XopC-AvrBs3Δ2 chimeras can induce the AvrBs3-specific HR (Fig. 6A) although both proteins were undetectable in the culture supernatants (Fig. 6C) reveals the high sensitivity of the in vivo assay. Hence, the test for the HR induction in planta allows us to monitor translocation of proteins that cannot be visualized in supernatants of in vitro cultures. Taken together, our translocation and interaction studies suggest that HpaB and HpaC form a complex and specifically promote the translocation of a subset of effectors including XopJ and XopF1. We therefore propose to group effector proteins from X. campestris pv. vesicatoria into at least two classes, which hereafter we refer to as class A (XopJ and XopF1) and class B (AvrBs3 and XopC).
Translocation of presynthesized effector proteins depends on HpaB but not on HpaC
Experimental evidence suggested that TTS chaperones from animal pathogenic bacteria are involved in the secretion of presynthesized effector proteins from the bacterial cytoplasm (Lloyd et al., 2001b; Page et al., 2002). To address this question in X. campestris pv. vesicatoria, we investigated the roles of HpaB and HpaC in effector targeting to the TTS system in the presence and absence of bacterial protein synthesis. For this, we blocked protein synthesis in strains 85*, 85*ΔhpaB and 85*ΔhpaC all expressing AvrBs3Δ2 chimeras as above. We used spectinomycin, which inhibits the elongation factor G-cycle and the peptidyl tRNA translocase reaction (Carter et al., 2000). When bacteria were infiltrated into AvrBs3-responsive pepper plants in the presence of spectinomycin, strain 85* expressing AvrBs3-, XopC- and XopJ-chimeras, respectively, still induced the HR (Fig. 7A), suggesting that protein translocation per se does not depend on de novo protein synthesis. Similar results were observed for hpaC mutant strains (data not shown). By contrast, no HR induction was observed for derivatives of strain 85*ΔhpaB expressing AvrBs3- and XopC-AvrBs3Δ2 respectively (Fig. 7A). We verified that total protein amounts of all AvrBs3Δ2 chimeras were reduced in the presence of spectinomycin (data not shown). However, similar amounts of AvrBs3-, XopC- and XopJ-AvrBs3Δ2 were detected in total cell extracts of hpaB wild-type and mutant strains (Fig. 7B), suggesting that the lack of protein translocation by strain 85*ΔhpaB was not due to reduced protein stabilities but rather to conformational problems to secrete presynthesized effectors. Taken together, we conclude that in the absence of protein synthesis effector protein translocation depends on the TTS chaperone HpaB but not on HpaC.
HpaB and HpaC interact with a conserved core component of the TTS apparatus
The finding that HpaC and HpaB are not secreted by the TTS system but contribute to the secretion of different effector proteins (Fig. 2; Büttner et al., 2004) suggests that both proteins are key for recruiting TTS substrates to the secretion apparatus. We therefore investigated whether HpaB and HpaC can bind to core components of the TTS system in the inner membrane. For this, we generated a chimera comprising GST and the putative inner membrane protein HrcV. HrcV is a conserved core component of the TTS system and is essential for type III-dependent protein secretion (Plano et al., 1991; He, 1998; Rossier et al., 1999). When GST-HrcV was immobilized on glutathione sepharose and incubated with a HpaC-c-Myc-containing lysate, HpaC-c-Myc specifically bound to GST-HrcV, but not to GST alone (Fig. 8A). Similarly, HpaB-c-Myc co-eluted with GST-HrcV but not with GST (Fig. 8B). An interaction between the HpaB and HrcV homologues from X. axonopodis pv. citri was previously shown in yeast (Alegria et al., 2004). Taken together, these observations indicate that both HpaB and HpaC can interact with HrcV. It is therefore tempting to speculate that they form a linker complex between the TTS system and secreted proteins.
The novel export control protein HpaC is involved in TTS
In this study, we identified HpaC as a novel key factor from X. campestris pv. vesicatoria in TTS. Mutant analyses revealed that hpaC contributes to bacterial growth and disease symptom formation in susceptible plants (Fig. 1). Furthermore, hpaC is required for the efficient HR induction in resistant plants. The finding that HpaC is conserved among xanthomonads and the plant pathogenic bacteria R. solanacearum and B. pseudomallei suggests that it plays an important role for bacterial pathogenicity. In agreement with this hypothesis, deletion of the hpaC homologue hpaP from R. solanacearum leads to a reduction of symptom formation in both susceptible and resistant plants (Van Gijsegem et al., 2002).
We found that HpaC from X. campestris pv. vesicatoria is essential not only for the efficient secretion of effectors but also of the putative translocon proteins HrpF and XopA (Fig. 2). Intriguingly, secretion of the Hrp pilus subunit HrpE and Hrp pilus formation as such are not affected in hpaC mutants. These observations indicate that secretion of Hrp pilus subunits and translocon proteins is regulated by different mechanisms. The specific contribution of a TTS-associated protein to effector and translocon protein secretion but not to pilus formation has so far not been observed in any plant pathogenic bacterium and identifies HpaC as a new type of export control protein.
HpaC and HpaB promote the translocation of a subset of effectors
In vivo translocation studies with different effectors revealed that HpaC specifically promotes the translocation of a subset of effector proteins including XopJ and XopF1. Similar data were obtained for the class IB chaperone HpaB. Translocation of XopJ- and XopF1-AvrBs3Δ2 chimeras was severely reduced in the hpaC mutant and completely abolished in the absence of HpaB (Fig. 3A and 6A). By contrast, other effector proteins such as AvrBs3 and XopC are translocated even in the absence of HpaB (Fig. 6A). We also analysed six additional effector proteins in hpaB and hpaC mutant strains (Thieme et al. unpubl. data). Preliminary studies revealed that four Xop-AvrBs3Δ2 chimeras are efficiently translocated by both mutants. In contrast, delivery of two Xop-chimeras is significantly affected in the absence of hpaC and completely abolished in the hpaB mutant background (J. Stuttmann, C. Lorenz, U. Bonas and D. Büttner, unpubl. data). Taken together, these data suggest that HpaB and HpaC promote the translocation of the same set of effectors. We therefore propose to group effectors into at least two classes: class A containing XopJ and XopF1 and class B containing AvrBs3 and XopC. To our knowledge, this is the first experimental evidence that suggests a classification of type III effector proteins based on their different translocation behaviour.
Protein–protein interaction studies revealed that HpaC binds to AvrBs3 and XopF1, which represent the two classes of effector proteins (Fig. 4). Similar data were obtained for HpaB (Büttner et al., 2004; data not shown), suggesting that it is not the binding of HpaC or HpaB per se which determines the specific need of class A effectors for both proteins. It remains to be investigated how HpaB and HpaC differentiate between class A and class B effector proteins. Comparative sequence analyses of the known effector proteins did not uncover any N-terminal sequence similarities indicative of a conserved binding site or a common targeting signal (F. Thieme and U.B., unpubl. data).
Given the importance of HpaB (Büttner et al., 2004) and HpaC for bacterial pathogenicity (Fig. 1) it is conceivable that the plethora of class A effectors plays a key role in the establishment of a pathogenic interaction with the host. A temporal regulation of effector protein translocation is an open question in plant pathogenic bacteria. Based on the data presented here, it might very well be that class A effectors are translocated first during the initial phase of the plant–pathogen interaction. In the animal pathogen Yersinia spp., experimental evidence suggested that the secretion of chaperoned effectors precedes the export of proteins that lack a cognate chaperone (Cheng and Schneewind, 1999; Boyd et al., 2000; Lloyd et al., 2001b). Furthermore, in S. flexneri different regulatory mechanisms that control effector protein expression imply that effectors are required and thus translocated at different stages of the infection process (Le Gall et al., 2005).
HpaC is part of a chaperone-dependent protein complex
Using GST pull-down assays, we could show that HpaC and HpaB interact with each other suggesting that they are part of a heterooligomeric protein complex that controls type III-dependent protein secretion and translocation. Our protein–protein interaction studies also revealed that HpaB and HpaC self-interact (Fig. 5). Similar findings were reported for a number of known TTS chaperones (Birtalan and Ghosh, 2001; van Eerde et al., 2004; Büttner et al., 2005; Kabisch et al., 2005).
In animal pathogenic bacteria, TTS chaperones were proposed to promote the export of pre-synthesized effector proteins (Ghosh, 2004). This is also the case for HpaB from X. campestris pv. vesicatoria. Using in vivo translocation assays, we found that HpaB is crucial for effector protein translocation in the absence of de novo protein synthesis. The need for HpaB can be explained by the fact that the proposed TTS channel, the Hrp pilus, is too narrow to allow folded proteins to pass through (Ghosh, 2004; Weber et al., 2005). The binding of HpaB therefore probably helps effectors to acquire or retain a secretion-competent conformation. Our observations are reminiscent of the finding that the Yersinia class IA chaperone SycE is required for the secretion of its cognate effector YopE when translation is inhibited (Lloyd et al., 2001b).
In contrast to HpaB, HpaC is not required for protein synthesis-independent effector protein translocation. HpaC lacks typical features of classical TTS chaperones, and blast and 3D-PSSM analyses did not reveal any similarity of HpaC with known TTS chaperones. Our protein–protein interaction studies revealed that HpaC interacts with effector proteins and the putative translocon protein XopA (Fig. 4). This is in contrast to HpaB, which does not bind to XopA. Because XopA secretion is not affected in hpaB mutants, but is significantly reduced in the absence of HpaC (Fig. 2A; Büttner et al., 2004), we speculate that it is the binding of HpaC that facilitates the export of XopA. These findings suggest a general role of HpaC in recruitment of effector and translocon proteins to the TTS machine which was not observed before for any TTS chaperone. TTS chaperones of plant and animal pathogenic bacteria have been divided into separate classes based on their specificity for either effector or translocon proteins (Parsot et al., 2003). It is therefore tempting to speculate that HpaC is the first member of a novel class of TTS chaperones.
It should be emphasized that in contrast to other plant and animal pathogenic bacteria (Ghosh, 2004), there are no TTS chaperones known in X. campestris pv. vesicatoria that are specific for one effector. Furthermore, inspection of the recently finished genomic sequence of X. campestris pv. vesicatoria strain 85-10 did not reveal genes for typical TTS chaperones in the vicinity of most effector genes (Thieme et al., 2005). It might therefore very well be that TTS in X. campestris pv. vesicatoria is mediated by a protein complex that contains the global class IB chaperone HpaB and HpaC.
HpaC and HpaB act as linkers between the TTS apparatus and its substrates
Notably, both HpaC and HpaB do not only bind to TTS substrates but also interact with the conserved HrcV protein, which is a predicted component of the TTS machinery in the inner membrane. This finding is corroborated by a recent report on the interaction of HrcV and HpaB homologues from X. axonopodis pv. citri that was observed in a yeast two-hybrid screen (Alegria et al., 2004). HpaB and HpaC are presumably not secreted by the TTS system (Fig. 1C; Büttner et al., 2004) and thus might act as linkers between the TTS system and its protein substrates. A similar role was proposed for chaperones of the TTS system and the evolutionary related flagellar export apparatus in animal pathogenic bacteria. The chaperones CesT and FlgN from enteropathogenic E. coli and Salmonella typhimurium, respectively, were shown to interact with the conserved ATPase of the membrane-spanning secretion apparatus (Gauthier and Finlay, 2003; Thomas et al., 2004; 2005). It has therefore been suggested that TTS chaperones guide their cognate interaction partners to the secretion apparatus.
In conclusion, our data indicate that type III-dependent protein secretion in X. campestris pv. vesicatoria is regulated by a chaperone-dependent protein complex that recruits TTS substrates to the secretion system and specifically promotes the translocation of a subset of effectors.
Bacterial strains and growth conditions
Bacterial strains and plasmids used in this study are listed in Table 1. E. coli cells were cultivated at 37°C in Luria–Bertani (LB) or Super medium (Qiagen, Hilden, Germany). X. campestris pv. vesicatoria strains were grown at 30°C in NYG medium (Daniels et al., 1984) or in minimal medium A (Ausubel et al., 1996) supplemented with sucrose (10 mM) and casamino acids (0.3%). Plasmids were introduced into E. coli by electroporation and into X. campestris pv. vesicatoria by conjugation, using pRK2013 as a helper plasmid in triparental matings (Figurski and Helinski, 1979). Antibiotics were added to the media at the following final concentrations: ampicillin, 100 µg ml−1; kanamycin, 25 µg ml−1; rifampicin, 100 µg ml−1; spectinomycin, 100 µg ml−1; tetracycline, 10 µg ml−1.
Table 1. Bacterial strains and plasmids used in this study.
The near-isogenic pepper cultivars ECW, ECW-10R and ECW-30R (Minsavage et al., 1990) were grown and inoculated with X. campestris pv. vesicatoria as described previously (Bonas et al., 1991). Generally, bacteria were hand-infiltrated into the intercellular spaces of leaves at concentrations of 2 × 108 cfu ml−1 in 1 mM MgCl2. The appearance of disease symptoms and the HR were scored over a period of 3 days after inoculation. For better visualization of the HR, leaves were bleached in 100% ethanol. For the inhibition of protein synthesis, bacteria were resuspended at concentrations of 4 × 108 cfu ml−1 in 1 mM MgCl2 and incubated 30 min before infiltration in the presence of spectinomycin. For in planta growth curves, bacteria were inoculated at a density of 104 cfu ml−1 into leaves of pepper cultivar ECW. Bacterial growth was determined as described previously (Bonas et al., 1991). Experiments were repeated at least three times.
To create a 207 bp in-frame deletion of hpaC (resulting in deletion of amino acids 14–83 of HpaC), we amplified the flanking regions of hpaC including the first 39 and the last 390 bp of the gene by PCR and cloned the amplification products into the BamHI/XbaI sites of the suicide plasmid pOK1. The resulting construct pOKΔhpaC was conjugated into X. campestris pv. vesicatoria strains 85-10, 85E* and 85*ΔhpaB. Homologous recombination events resulted in hpaC mutants which were selected as described previously (Huguet et al., 1998).
Generation of expression plasmids
For detection of HpaC by immunoblotting, hpaC was amplified by PCR and cloned into the EcoRI/SacI sites of pC3003, in frame with a triple-c-myc epitope-encoding sequence, giving pChpaC. The resulting insert was then ligated into the EcoRI/HindIII sites of plasmid pDSK602, giving pDMhpaC. To generate HpaC1−100-AvrBs3Δ2, the first 100 codons of hpaC were ligated into the EcoRI site of plasmid pDS356F, in frame with avrBs3Δ2, resulting in pDhpaC356. Similarly, the first 200 codons of xopF1 were ligated into plasmid pDS356F, giving pDxopF1356.
For the generation of GST fusion constructs, hrcV, hpaB and hpaC were amplified by PCR and cloned into the EcoRI/XhoI sites of pGEX-2TKM, downstream and in frame with the GST-encoding sequence. To create GST-XopF1, xopF1 was amplified and ligated into the EcoRI/SacI site of pGEX-2TKM.
Secretion experiments and protein analysis
For the analysis of type III-dependent protein secretion, bacteria were incubated in secretion medium as described previously (Büttner et al., 2002). Equal amounts of bacterial total cell extracts and culture supernatants were analysed by SDS-PAGE and immunoblotting. We used polyclonal antibodies specific for HrpF (Büttner et al., 2002), HrpE (Weber et al., 2005) and AvrBs3 (Knoop et al., 1991), respectively, and monoclonal anti-c-Myc and anti-GST antibodies (Amersham Pharmacia Biotech, Freiburg, Germany). Horseradish peroxidase-labelled anti-rabbit, anti-mouse and anti-goat antibodies (Amersham Pharmacia Biotech) were used as secondary antibodies. Antibody reactions were visualized by enhanced chemiluminescence (Amersham Pharmacia Biotech). To ensure that no bacterial lysis had occurred, blots were routinely reacted with an antibody specific for the intracellular protein HrcN (data not shown; Rossier et al., 2000).
Glutathione S-transferase pull-down assays
Glutathione S-transferase and GST chimeras were expressed in E. coli BL21(DE3). Bacterial pellets from 50 ml of cultures were resuspended in phosphate-buffered saline (PBS), broken with a French press and immobilized on glutathione sepharose as described (Büttner et al., 2004). Unbound proteins were removed by washing twice with PBS. GST and GST chimeras were incubated for 1 h at room temperature with 600 µl of BL21 lysates containing HpaB-c-Myc or HpaC-c-Myc. The resin was washed four times with PBS, and bound proteins were eluted for 2 h with 10 mM reduced glutathione at room temperature. Five microlitres of total protein lysates and 20 µl of eluted proteins were analysed by SDS-PAGE and immunoblotting (see above). Blots were first incubated with an anti-c-Myc and then with an anti-GST antibody.
We are grateful to Ai-Jiuan Wu for critical reading of the manuscript. This work was funded by grants from the Deutsche Forschungsgemeinschaft (BO 790/7-1) and the Bundesministerium für Bildung und Forschung to U.B. (GenoMik) and the Fonds der Chemischen Industrie to D.B.