Resistance in tomato (Solanum lycopersicum) to infection by Pseudomonas syringae involves both detection of pathogen-associated molecular patterns (PAMPs) and recognition by the host Pto kinase of pathogen effector AvrPtoB which is translocated into the host cell and interferes with PAMP-triggered immunity (PTI). The N-terminal portion of AvrPtoB is sufficient for its virulence activity and for recognition by Pto. An amino acid substitution in AvrPtoB, F173A, abolishes these activities. To investigate the mechanisms of AvrPtoB virulence, we screened for tomato proteins that interact with AvrPtoB and identified Bti9, a LysM receptor-like kinase. Bti9 has the highest amino acid similarity to Arabidopsis CERK1 among the tomato LysM receptor-like kinases (RLKs) and belongs to a clade containing three other tomato proteins, SlLyk11, SlLyk12, and SlLyk13, all of which interact with AvrPtoB. The F173A substitution disrupts the interaction of AvrPtoB with Bti9 and SlLyk13, suggesting that these LysM-RLKs are its virulence targets. Two independent tomato lines with RNAi-mediated reduced expression of Bti9 and SlLyk13 were more susceptible to P. syringae. Bti9 kinase activity was inhibited in vitro by the N-terminal domain of AvrPtoB in an F173-dependent manner. These results indicate Bti9 and/or SlLyk13 play a role in plant immunity and the N-terminal domain of AvrPtoB may have evolved to interfere with their kinase activity. Finally, we found that Bti9 and Pto interact with AvrPtoB in a structurally similar although not identical fashion, suggesting that Pto may have evolved as a molecular mimic of LysM-RLK kinase domains.
An initial layer of induced plant defense against potential pathogens involves host perception of pathogen-associated molecular patterns (PAMPs), typically by plasma membrane-localized pattern recognition receptors (PRRs) (Gómez-Gómez and Boller, 2002; Zipfel, 2008). Pattern recognition receptors, in concert with associated proteins, then activate signaling pathways leading to PAMP-triggered immunity (PTI) (Chisholm et al., 2006; Jones and Dangl, 2006; Chinchilla et al., 2007). To date, diverse PAMPs including flagellin, elongation factor Tu (EF-Tu), chitin, and peptidoglycan (PGN), are known to be recognized by plants (Boller, 1995; Zipfel, 2008). However, the PRRs that play a role in the recognition of the majority of PAMPs remain unknown.
The two best characterized plant PRRs are from Arabi-dopsis, and both recognize bacterial PAMPs: FLS2 detects a 22-amino-acid epitope (flg22) in flagellin and EFR detects an 18-amino-acid epitope in EF-Tu (Gómez-Gómez and Boller, 2000; Chinchilla et al., 2006; Zipfel et al., 2006). Both of these proteins are leucine-rich repeat (LRR) receptor-like kinases (LRR-RLKs) and FLS2, in particular, has been studied intensively. FLS2 binds the flg22 peptide after which it associates with another LRR-RLK, the BRI1-associated kinase BAK1 (Chinchilla et al., 2006, 2007). Plants with mutations in the FLS2 gene are compromised for PTI and are more susceptible to certain bacterial pathogens (Zipfel et al., 2004; Hann and Rathjen, 2007). Recently FLS2 has been shown to also recognize Ax21, a PAMP from Xanthomonas campestris pv. campestris and, surprisingly, the Clavata3 peptide which is expressed exclusively in the shoot apical meristem (Danna et al., 2011; Lee et al., 2011).
Pathogens have evolved virulence proteins (‘effectors’) which they deliver into the plant cell to suppress PTI (Hogenhout et al., 2009). Gram-negative bacterial pathogens, for example, use a type III secretion (TTS) system to inject effector proteins directly into the cytoplasm of the plant cell (Mudgett, 2005; Cornelis, 2006; Grant et al., 2006; Guo et al., 2009). Strain DC3000 of Pseudomonas syringae pv. tomato (Pst), the causal pathogen of tomato (Solanum lycopersicum) bacterial speck disease, expresses and delivers about 30 type III effectors into host cells (Lindeberg et al., 2006).
On plants that express the resistance (R) protein Pto and the nucleotide-binding-leucine rich repeat (NB-LRR) protein Prf, either of two Pst effectors, AvrPto or AvrPtoB (HopAB2) is recognized by the Pto kinase via a physical interaction, leading to activation of effector-triggered immunity (ETI) (Pedley and Martin, 2003). On plants that lack a functional Pto/Prf pathway, however, AvrPto and AvrPtoB promote virulence of Pst in an additive manner (Lin and Martin, 2005; Kvitko et al., 2009; Cunnac et al., 2011). Both effectors suppress PTI signaling in Arabidopsis and Nicotiana benthamiana plants (He et al., 2006; Hann et al., 2010).
The N-terminal portion of AvrPtoB has two distinct domains that contribute to its virulence activity. The region consisting of amino acids 1–387 of AvrPtoB (AvrPtoB1–387) but not a shorter fragment, AvrPtoB1–307, is able to suppress FLS2-mediated PTI (He et al., 2006; Shan et al., 2008). On tomato plants lacking Prf (RG-prf3), the other domain, AvrPtoB1–307, is sufficient to confer virulence comparable to AvrPtoB in DC3000ΔavrPto/ΔavrPtoB (a strain with avrPto and avrPtoB deleted) (Xiao et al., 2007). AvrPtoB1–307 is also able to elicit Pto-mediated ETI on tomato plants having a functional Pto/Prf pathway. Interestingly, substitution in AvrPtoB1–307 of a key amino acid, phenylalanine-173 to alanine (F173A) abolishes both its virulence and avirulence activity. The F173A mutation, however, has no apparent effect on AvrPtoB PTI suppression in Arabidopsis (Xiao et al., 2007).
As with AvrPto, the molecular mechanisms used by AvrPtoB to suppress PTI are not fully understood. Previous studies showed the N-terminal region of AvrPtoB (AvrPtoB1–387 in Arabidopsis and AvrPtoB1–307 in tomato) to be sufficient for virulence, probably due to targeting of PRRs or BAK1 (He et al., 2006; Xiao et al., 2007; Shan et al., 2008). However, two recent studies reported the E3 ligase activity of AvrPtoB CTD to be required for suppression of PTI while promoting virulence of Pst in Arabidopsis (Gohre et al., 2008; Gimenez-Ibanez et al., 2009a). Despite this discrepancy, it is clear AvrPtoB targets other PRRs or adaptor proteins in addition to FLS2/EFR/BAK1, as AvrPtoB suppresses PTI triggered by PAMPs in addition to flagellin and EF-Tu (Zipfel et al., 2004; He et al., 2006; Shan et al., 2008). In this regard, AvrPtoB has recently been shown to target the Arabidopsis LysM-RLK, CERK1, which itself was shown to play a role in the immune response against Pst (Gimenez-Ibanez et al., 2009a).
To identify possible virulence target(s) of AvrPtoB in tomato, we performed a yeast two-hybrid screen using the effector as a bait against a tomato leaf cDNA library. One AvrPtoB tomato-interacting (Bti) protein identified in the screen, Bti9, was found to encode a lysin motif (LysM) receptor-like kinase (LysM-RLK) whose closest amino acid match in Arabidopsis is CERK1. Here we characterize the role of Bti9 and a closely related tomato protein, SlLyk13, in plant immunity and investigate the mechanism by which AvrPtoB may interfere with the activity of these proteins.
The N-terminal region of AvrPtoB is sufficient to suppress immunity in tomato triggered by bacterial PAMPs other than flagellin
We reported previously that either of the N-terminal fragments, AvrPtoB1–307 or AvrPtoB1–387, was able to enhance Pst virulence on tomato RG-prf3 plants to the same extent as full-length AvrPtoB when expressed in DC3000ΔavrPto/ΔavrPtoB (Xiao et al., 2007). To determine if the AvrPtoB N-terminal domain is capable of suppressing PTI triggered by Pst PAMPs other than flagellin, we inoculated RG-prf3 plants with DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC strains (DC3000 with deletion of the genes avrPto, avrPtoB, hopQ1-1, and fliC) carrying empty vector, full-length AvrPtoB, AvrPtoB1–387, or AvrPtoB1–387(F173A). The F173A substitution also abolishes AvrPtoB1–307 virulence; however, we used AvrPtoB1–387 instead of AvrPtoB1–307 in these experiments because its expression in Pst is more similar to that of full-length AvrPtoB (Xiao et al., 2007). The fliC gene encodes flagellin and this strain therefore lacks this PAMP (Kvitko et al., 2009). The hopQ1-1 gene is irrelevant to this experiment as it confers avirulence of Pst on N. benthamiana plants, but not on tomato (Wei et al., 2007). To distinguish between subtle differences in disease symptoms, we used a low titer of 3 × 104 colony-forming units (cfu) ml−1 bacteria for inoculation. Consistent with our previous findings (Xiao et al., 2007), we found that DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC strains expressing AvrPtoB or AvrPtoB1–387 caused more severe disease symptoms on tomato RG-prf3 plants compared with strains carrying an empty vector or AvrPtoB1–387(F173A) (Figure 1a).
Assessment of bacterial growth in these plants revealed that 2 days after inoculation the Pst strains expressing AvrPtoB or AvrPtoB1–387 grew to a statistically significant higher level than the strain carrying an empty vector (Figure 1b). The growth of Pst expressing AvrPtoB1–387(F173A) was not significantly different from any of the other three strains (including the empty vector control; Figure 1b) although this strain caused less lower leaf necrosis. Each of the AvrPtoB proteins was expressed in Pst (Figure 1c). Importantly, the Pst strain expressing AvrPtoB1–387 had the same virulence activity as the strain expressing AvrPtoB, indicating that the N-terminal domain (lacking the CTD E3 ligase) is sufficient to suppress PTI triggered by PAMPs other than flagellin of Pst in tomato. Note also that although the abundance of AvrPtoB1–387 protein was less than AvrPtoB, this amount appeared to be sufficient for full virulence as no additional growth was observed with the strain expressing full-length AvrPtoB (Figure 1b,c). Together, these observations indicate that the N-terminal domain is sufficient for full AvrPtoB virulence on tomato and that F173 plays an important role in the formation of disease symptoms (especially necrosis of the lower leaves).
A LysM-RLK, Bti9, interacts with AvrPtoB1–307
To identify host targets of AvrPtoB we performed a yeast two-hybrid screen using the effector AvrPtoB as bait with a tomato leaf cDNA library. One interactor, Bti9, was found to interact strongly with AvrPtoB and AvrPtoB1–307 but significantly less so with the virulence-attenuated proteins AvrPtoB(F173A) and AvrPtoB1–307(F173A) (Figure 2a). Bti9 did not interact with AvrPto (not shown). All AvrPtoB proteins were expressed in yeast as confirmed by western blotting and by their interaction with Bti6, a non-specific AvrPtoB interactor (Figure 2a,b).
The Bti9 region that interacted with AvrPtoB1–307 encodes a serine/threonine protein kinase domain with 76% amino acid identity to Arabidopsis Chitin Elicitor Receptor Kinase 1 (CERK1) (AtLysM-RLK1, AtLyk1) (Miya et al., 2007; Wan et al., 2008). Generation of a full-length cDNA of Bti9 revealed that it encodes a protein with a predicted extra-cellular domain containing three lysin motifs (LysM) and the C-terminal kinase that interacts with AvrPtoB1–307 (Figure S1 in Supporting Information). The presence of a putative signal sequence and transmembrane region in Bti9 (Figure S1) suggested that the protein is localized to the plasma membrane. Indeed, we detected Bti9 at the periphery of the cell when it was expressed transiently as a GFP fusion in tomato protoplasts (Figure S2, Methods S1), supporting a possible role for the protein in extracellular pathogen recognition.
LysM domains are known to bind N-acetylglucosamine (GlcNAc)-containing glycan molecules including PGN from several bacteria and chitin from fungi (Bateman and Bycroft, 2000; Steen et al., 2003; Bielnicki et al., 2006; Buist et al., 2008; Iizasa et al., 2010; Petutschnig et al., 2010). LysM-RLKs play an important role in Lotus japonicus and Medicago truncatula in their association with Rhizobium bacteria (Limpens et al., 2003; Radutoiu et al., 2003, 2007; Smit et al., 2007). Furthermore, CERK1 was recently shown to be involved in immunity against Pst (Gimenez-Ibanez et al., 2009a). These observations raised the possibility that Bti9 plays a role in recognition of a bacterial PAMP and, as a result, has become a host target of the AvrPtoB N-terminal domain. We therefore tested whether the expression of Bti9, like FLS2 and other immunity-associated kinase genes (Zipfel et al., 2004), is induced in leaves upon challenge by PAMPs. We observed that Bti9 transcript abundance increased 4 h after exposure to DC3000 as compared with control leaves. Interestingly, at 8 h after exposure to wild-type DC3000, but not to a type III secretion mutant, this increase was no longer observed (Figure S3). These results suggest that Bti9 gene expression is induced by PAMPs and that type III effectors interfere with its expression.
Two LysM-RLKs are implicated in AvrPtoB-suppressed immunity against P. syringae
The tomato genome sequence became available during the course of this work and allowed us to search for homologs of Bti9 (http://solgenomics.net/ ). This search revealed tomato has 13 LysM-RLK genes that we have named in accordance with previous recommendations and, in part, their relationship to the five Arabidopsis LysM-RLKs (Figure S4; Zhang et al., 2007). Three of these genes, SlLyk11, SlLyk12, and SlLyk13 reside in the same clade as Bti9 (also referred to as SlLyk1) and share extensive sequence similarity (Figure S5).
To test whether Bti9 and possibly other members of the ‘Bti9 clade’ are involved in the tomato immune response we developed stable transgenic plants that carry a ‘hairpin’ Bti9 (hpBti9) construct which could potentially silence all four genes (Wesley et al., 2001; Helliwell et al., 2002). The presence and copy number of the transgene in plants of T2 transgenic lines was examined (Figure S6). Two lines, CF5 and CF26, from independent transformation events were chosen for further characterization. Quantitative RT-PCR revealed that both lines had reduced transcript abundance of Bti9, SlLyk11, SlLyk12, and SlLyk13 (Figure 3).
The hpBti9 plants and appropriate controls were inoculated with Pst strain DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC and monitored for disease symptoms. We observed the hpBti9 plants displayed more severe disease symptoms than control plants (Figures 3b and S7a). Interestingly, the hpBti9 plants developed extensive necrosis on their lower leaves, which is reminiscent of the phenotype associated with virulence activity of the AvrPtoB N-terminal region (Figures 1a and 3b; Xiao et al., 2007). Bacterial population assays showed there was a slight, but statistically significant, increase in Pst growth in leaves of the hpBti9 plants as compared with control leaves 2 days after inoculation (Figures 3c and S7b). These disease assays on two independent hpBti9 transgenic lines support a role for one or more members of the Bti9 clade in tomato immunity against Pst.
As one way of assessing which of the four Bti9 clade genes may contribute to the plant immune response, we examined whether the kinase domain encoded by SlLyk11, SlLyk12, or SlLyk13 interacts with AvrPtoB in an F173-dependent manner. Our rationale was that the F173A substitution abolishes AvrPtoB1–307 virulence activity and therefore if a SlLyk protein plays a role in the immune response suppressed by AvrPtoB its interaction with this effector would be expected to require F173. Yeast two-hybrid analyses showed that both SlLyk11 and SlLyk12 interacted with AvrPtoB1–307 but this occurred even when the F173A substitution was present (Figure 4a). SlLyk13, however, displayed an interaction profile identical to Bti9. The less-related proteins, SlLyk3 and SlLyk10, did not interact with AvrPtoB1–307. Each of the Lyk proteins was expressed in yeast and, while SlLyk3 and SlLyk10 were expressed poorly, their abundance was the same as Pto which interacts strongly with AvrPtoB1–307 in this assay (Figure 4b). We conclude from these experiments that Bti9 and SlLyk13 are the best candidates for contributing to the immunity against Pst which is suppressed by AvrPtoB1–307.
AvrPtoB inhibits Bti9 kinase activity
We focused on Bti9 as representative of both Bti9 and SlLyk13 in order to test if it is an active protein kinase. An in vitro kinase assay showed that Bti9 weakly autophosphorylates and strongly transphosphorylates myelin basic protein, an artificial substrate (Figure 5a). In common with the Pto variant Y207D (Rathjen et al., 1999), substitution at the corresponding Bti9 amino acid, Y489D, caused a complete loss of Bti9 kinase activity (autophosphorylation and transphosphorylation). Bti9 is therefore an active protein kinase.
The kinase activity of FLS2 is required for its role in innate immunity (Gómez-Gómez, et al. 2001). In addition, recent work on suppression of PTI by AvrPto showed that this effector inhibits kinase activity of PRR complexes (Shan et al., 2008; Xiang et al., 2008, 2011). We therefore examined whether the AvrPtoB N-terminal region affects Bti9 kinase activity. Standard in vitro kinase activity assays for Bti9 were conducted in the presence of increasing amounts of glutathione-S-transferase (GST)-AvrPtoB, GST-AvrPtoB1–307 or GST alone. A Bti9 reaction without of any of these proteins was used as a control. GST-AvrPtoB inhibited Bti9 kinase activity in a dose-dependent manner (Figure 5b). In contrast, GST alone had no effect on Bti9 kinase activity. GST-AvrPtoB1–307 inhibited Bti9 kinase activity even more effectively than AvrPtoB, but this may be due to the higher relative molarity of this smaller protein (Figure 5b). We next examined the ability of AvrPtoB(F173A) to inhibit Bti9 kinase activity (Figure 5c). We found that AvrPtoB(F173A) was unable to inhibit Bti9 kinase activity, even though the abundance of this variant was the same as wild-type AvrPtoB (Figure 5c). Collectively, these results provide additional evidence for interaction between AvrPtoB and Bti9 and also support a model in which AvrPtoB inhibits kinase activity of Bti9, and possibly that of SlLyk13, to block PTI signaling.
Bti9/SlLyk13 are probably not the only virulence targets of AvrPtoB1–307 in tomato
A single pathogen effector can have multiple host targets for promoting pathogen virulence (Hogenhout et al., 2009). In the case of AvrPtoB, the region between amino acids 307 and 387 has been reported to be required for interference with FLS2/BAK1-mediated PTI (Xiao et al., 2007; Shan et al., 2008). It is therefore possible that the AvrPtoB1–307 domain has targets in addition to Bti9/SlLyk13. To test this, we monitored bacterial growth of DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC either expressing or not expressing AvrPtoB1–387 on RG-prf3 and hpBti9 plants. Consistent with our earlier observations, AvrPtoB1–387 enhanced virulence of this DC3000 strain on RG-prf3 plants and silencing of Bti9/SlLyk13 rendered tomato plants more susceptible (Figure S8). Importantly, when this DC3000 strain was expressing AvrPtoB1–387 it was still more virulent on hpBti9 plants than the same strain carrying an empty vector. This suggests that the AvrPtoB N-terminal region has other host targets in addition to Bti9. However, we cannot rule out the possibility this enhanced virulence is due to AvrPtoB suppression of residual Bti9/SlLyk13 accumulation in the cell in spite of the hpBti9-mediated gene silencing.
Bti9 and Pto interact with AvrPtoB1–307 in a structurally similar but not identical way
The crystal structure of the subdomain in AvrPtoB1–307 that is sufficient for its interaction with Pto has been solved (AvrPtoB121–200; Dong et al., 2009). This structure revealed that Pto has one interface that it uses for interaction with both AvrPto and AvrPtoB and another that is unique for AvrPtoB (Figure 6a). Notably, the F173 residue, which is required for virulence of AvrPtoB1–307, is also required for recognition by Pto indicating that the tomato immune system has evolved to specifically detect this virulence determinant of AvrPtoB.
To further examine the possible similarity between the interaction of Bti9 and Pto with AvrPtoB1–307, we took advantage of a series of substitutions in AvrPtoB that were developed for analysis of the AvrPtoB121–200–Pto complex (Dong et al., 2009). The expression of each of these variants was verified and their interactions with Bti9 and Pto were tested in the yeast two-hybrid system (Figure 6b,c). The interactions of these two proteins with AvrPtoB1–307 were largely similar, with the only differences being M176D and I181D which disrupted the interaction with Pto but not with Bti9. These observations support a model in which Pto evolved as a structural mimic, albeit not a perfect one, of Bti9/SlLyk13.
By searching for host targets of AvrPtoB virulence activity we isolated a LysM-RLK gene, Bti9, and later identified three Bti9-related genes in the tomato genome. Our rationale for initially focusing on Bti9 was that it interacts with a minimal fragment of AvrPtoB (amino acids 1–307) that enhances virulence of Pst on tomato to the same degree as full-length AvrPtoB (Xiao et al., 2007). Furthermore, this interaction was dependent on phenylalanine-173, a residue that is required for virulence activity of AvrPtoB1–307. Similar to many kinase genes involved in PTI, Bti9 gene expression was induced by PAMPs and appeared to be suppressed by type III effectors. The fact that four genes exist in the Bti9 clade raised the possibility they may have some redundancy in function. Our hpBti9 construct successfully knocked down expression of all four genes and resulted in increased susceptibility to Pst. However, only Bti9 and SlLyk13 interacted with AvrPtoB in an F173-dependent manner, suggesting they are the most likely virulence targets of this effector protein. AvrPtoB also interacted with Bti9 in vitro as inferred from its ability to inhibit Bti9 kinase activity. This inhibition was dependent on F173 which supports kinase inhibition as the primary mechanism for interference in the function of Bti9, and possibly SlLyk13, and thus PTI suppression. Finally, based on the similar interaction of Bti9 and Pto with AvrPtoB our data support the possibility that Pto evolved as a molecular mimic (a decoy) of Bti9, although it does not appear to be a ‘perfect’ structural mimic as some differences in interaction were observed. Together, our data support a model in which Bti9 and/or SlLyk13 play a role in perception of a bacterial PAMP and in PTI signaling and have therefore been targeted for interference by AvrPtoB.
We have shown here and previously that the N-terminal portion of AvrPtoB is sufficient to enhance the virulence of Pst in tomato to the same degree as full-length AvrPtoB (Xiao et al., 2007). We have also reported that a naturally occurring truncated AvrPtoB homolog (HopPmaL) which lacks the E3 ligase domain retains full virulence in tomato (Lin et al., 2006). These observations are in contrast to those of two recent papers, both of which conclude the E3 ubiquitin ligase of AvrPtoB is required for virulence on Arabidopsis. It is possible there are actual differences between tomato and Arabidopsis in this regard. However, an alternative explanation is that certain experimental conditions differed between our work and these other studies. For example, the expression levels in Pst of the E3 ligase-deficient versions of AvrPtoB were very low in one of the studies and may have underestimated the virulence activity of these proteins (Gimenez-Ibanez et al., 2009a). In the other study, the AvrPtoB1–387 fragment was tested and shown to have less virulence activity but its activity was not compared with full-length AvrPtoB expressed from the same plasmid (Gohre et al., 2008). Consistent with our observations, the AvrPtoB1–387 N-terminal domain is sufficient to suppress molecular readouts associated with PTI in Arabi-dopsis, also indicating that the E3 ligase is dispensable for virulence in this species (Shan et al., 2008). In the cases where AvrPtoB has been compared with similarly expressed truncated versions, the results indicate the E3 ligase is not required for virulence activity of AvrPtoB. To date, the only well-substantiated role reported for the AvrPtoB E3 ligase is to facilitate ubiquitination-mediated degradation of the Fen kinase to overcome ETI (Rosebrock et al., 2007). The presence of Fen-like kinases in diverse plant species suggests that this mechanism has been evolutionarily important in defeating this conserved immune response.
Expression of a hairpin Bti9 construct in tomato resulted in increased susceptibility to Pst infection with no other morphological or developmental defects being observed in these plants. We used a DC3000 strain that lacks flagellin in order to exclude the strong FLS2-mediated immune response and uncover other potential PTI responses, and observed enhanced disease symptoms in the hpBti9 plants, particularly on the lower leaves. This lower-leaf necrosis is remarkably similar to the phenotype associated with AvrPtoB virulence activity and, along with our yeast two-hybrid data, further supports Bti9/SlLyk13 as targets of AvrPtoB that may play an especially important role in older leaves. Despite the effective silencing of Bti9/SlLyk13 we still observed virulence activity of AvrPtoB on hpBti9 plants. The most likely explanation for this observation is probably that AvrPtoB1–307 has additional virulence targets in tomato. However, it is also possible that AvrPtoB1–307 interferes with residual Bti9/SlLyk13 protein still present in our hpBti9 plants allowing enhanced disease and bacterial growth.
The Bti9 clade contains four closely related LysM-RLK genes, all of which were silenced to some degree in our hpBti9 plants. However, we present data suggesting that Bti9 and/or SlLyk13 are the most likely targets of AvrPtoB as these proteins interact in a biologically relevant way with AvrPtoB. SlLyk11 and SlLyk12 even interact with virulence-attenuated AvrPtoB(F173A). They are probably not responsible for the PTI response that we see suppressed by AvrPtoB, but our data cannot exclude the possibility they may contribute to the tomato immune response.
We found that both full-length AvrPtoB and AvrPtoB1–307 inhibited Bti9 kinase activity in a dose-dependent manner, and this inhibition was dependent on the virulence-associated F173 residue. Although we have not demonstrated that kinase activity is required for Bti9 signaling, that is the case for other immunity-associated kinases such as FLS2 and TPK1b and it seems likely it is the case for Bti9 (Gómez-Gómez et al., 2001; Abuqamar et al., 2008). The fact that AvrPtoB inhibits this activity would therefore support a model in which this is the primary mechanism of AvrPtoB activity directed against Bti9/SlLyk13. In the future, it will be important to identify substrates of these proteins that act in the tomato immune response as that will allow further testing of the effectiveness of AvrPtoB to interfere with natural Bti9 phosphorylation events.
Bti9 and Pto interacted with AvrPtoB1–307 using mostly the same effector residues, thereby supporting the hypothesis that Pto might have evolved as a molecular mimic of the Bti9 kinase domain. However, two aspartate substitutions, at AvrPtoB residues M176 and I181, disrupted the interaction with Pto but not with Bti9. These residues lie in a loop of AvrPtoB that interacts with the AvrPtoB-specific contact site of Pto. These observations raise the possibility that AvrPtoB might be able to acquire mutations that would alter M176 and I181 allowing the effector to escape Pto detection while at the same time retaining its ability to interfere with Bti9. The fact that such mutations are not present in over 20 AvrPtoB (HopAB) homologs suggests that such mutations might have other detrimental effects on the structure or function of AvrPtoB (http://pseudomonas-syringae.org/).
CERK1, which based on amino acid sequence similarity is the closest Arabidopsis protein to Bti9, has also been reported to play a role in immunity against Pst and to be targeted by AvrPtoB (Gimenez-Ibanez et al., 2009a). Our results are largely consistent with this previous work while providing additional insights into the AvrPtoB–Bti9 interaction and also supporting an alternative model for how AvrPtoB interferes with Bti9 function. In common with this earlier paper, we discovered that the AvrPtoB1–307 domain is sufficient for interaction with Bti9 and we extended this to show that the virulence-associated residue F173 is required for this interaction. The compromised immunity to Pst observed with a cerk1 Arabidopsis mutant and our hpBti9 tomato plants indicates that, in addition to playing a role in chitin perception (Miya et al., 2007; Wan et al., 2008), these kinases may also play a role in detection of a bacterial PAMP. LysM domains are known to bind certain N-acetylglucosamine (GlcNAc)-containing glycan molecules including PGN from bacteria (Petutschnig et al., 2010), and it is an attractive hypothesis that Bti9/SlLyk13 and CERK1 may bind PGN from Pst to activate PTI. However, preliminary experiments indicate that this is not the case (Gimenez-Ibanez et al., 2009b) and the identity of the putative cognate PAMP remains unknown.
The previous paper on CERK1 presented evidence that the AvrPtoB E3 ligase is required for effector virulence and that it played a role in proteasome-independent degradation of CERK1 via a vacuolar-mediated process (Gimenez-Ibanez et al., 2009a). In contrast, we observed the E3 ligase domain to be dispensable for full virulence of AvrPtoB in tomato. Instead, we discovered that AvrPtoB inhibits Bti9 kinase activity supporting this inhibition as the primary mechanism of AvrPtoB virulence. A possible synthesis of these results would be that Bti9 kinase activity is inhibited and the inactivated Bti9 is turned over by a vacuolar-dependent receptor recycling pathway. Down-regulation of receptors by internalization and subsequent lysosomal degradation typically involves monoubiquitination or K63-linked polyubiquitination of the substrates (Mukhopadhyay and Riezman, 2007; Lauwers et al., 2009) and the AvrPtoB E3 ligase would probably not play a biologically significant role in this process.
We propose a model in which Bti9/SlLyk13 are localized to the plasma membrane and play a direct or indirect role in recognizing a bacterial PAMP upon pathogen infection (Figure S9). This recognition activates Bti9/SlLyk13 kinase activity that in turn phosphorylates an unknown substrate(s) leading to downstream signaling culminating in PTI. To subvert this PTI response, the N-terminal region of AvrPtoB has evolved to inhibit Bti9/SlLyk13 kinase activity, which blocks PTI signaling and signals the cell to recycle these proteins via a vacuolar-mediated process (Gimenez-Ibanez et al., 2009a). To counter this interference, tomato has evolved Pto, probably as a molecular mimic of the Bti9/SlLyk13 kinase domain. Pto interacts with the AvrPtoB N-terminal region, which elicits ETI against the pathogen. Examination of this model in the future will benefit from a crystal structure of the AvrPtoB1–307–Bti9 (or AvrPtoB1-307-SlLyk13) complex, identification of the cognate PAMP and possible Bti9/SlLyk13 adaptor proteins, and importantly an understanding of substrates and the pathways that these LysM-RLK activate.
Generation of full-length cDNA for the Bti9 gene
The C-terminal domain of Bti9 (nucleotide 814 to the end) in prey vector pJG4-5 was recovered from yeast and sequenced. 5′-Rapid amplification of cDNA ends (RACE) using gene-specific primer LRZ01 (Table S1) and the 5′-RACE System (Invitrogen, http://www.invitrogen.com/) was performed to obtain the full-length sequence of the gene.
Generation of hpBti9 transgenic plants
Gateway cloning (Hartley et al., 2000) was used to generate a Bti9‘hairpin’ construct (hpBti9) consisting of the C-terminal region of the Bti9 gene (nucleotides 1206–1807). A primer pair LRZ02 and LRZ03 (Table S1) was used to amplify the C-terminal fragment with Easy-A High-Fidelity PCR Cloning Enzyme (Stratagene, http://www.stratagene.com/) and cloned into pCR8/GW/TOPO vector (Invitrogen). The fragment was transferred into the pHellsgate8 binary vector (Wesley et al., 2001; Helliwell et al., 2002) by Gateway LR reactions. The pHellsgate8 plasmid containing the C-terminal region of Bti9 was transformed into tomato RG-prf3 (Pto/Pto, prf/prf) using Agrobacterium tumefaciens-mediated gene transfer technique (Hoekema et al., 1983). The expression level of Bti9 in T1 transformants was initially assessed by reverse transcriptase-PCR using primer pair LRZ04 and LRZ05 (Table S1). Those T1 transformants that showed reduced Bti9 expression were advanced to the T2 generation. The copy number of the hpBti9 construct in plants of T2 lines was examined by DNA gel blots as shown in Figure S4. T2 plants derived from T1 lines 9CF-5 and 9CF-26, both of which had a single copy of the hpBti9 transgene in RG-prf3 background, were used for pathogen assays.
Assessment of pathogen virulence in tomato leaves
RG-prf3 (Pto/Pto, prf/prf) and hpBti9 plants were grown in the greenhouse. Plasmids carrying avrPtoB or its variants expressed from a hrp promoter were developed using the vector pCPP45 and transformed into Pst DC3000ΔavrPtoΔavrPtoBΔhopQ1-1ΔfliC (Lin and Martin, 2005). Inoculation of tomato plants was performed as described (Anderson et al., 2006). In brief, 3- to 4-week-old plants were vacuum-infiltrated with the Pst strains using an inoculum of 3 × 104 cfu ml−1. The inoculated plants were maintained in a growth chamber at 24/20°C day/night temperature, 50% humidity, and a 16-/8-h light/dark cycle. Bacterial populations were recovered and measured at 0 and 2 days after inoculation. Photos of the plants were taken at 4 days after inoculation. Plants of the same size were used for each experiment, and lower leaves from approximately the same position on each plant were sampled. As described (Anderson et al., 2006), consistency in handling of bacterial strains and plants is critical for reproducible virulence assays. The presence of the hpBti9 transgene in each T2 plant was examined by PCR of genomic DNA using primers LRZ06 and LRZ07 (Table S1) (T-DNA right border) and primers LRZ08 and LRZ09 (T-DNA left border). Expression of Bti9 was analyzed by semi-quantitative PCR as described above.
Quantitative real time reverse transcription PCR (qRT-PCR)
Quantitative RT-PCR was performed as described (Nguyen et al., 2010) but with qRT-PCR cycling conditions as follows: 50°C for 2 min, 95°C for 10 min and 40 cycles of 95°C for 30 sec, 55°C for 30 sec and 72°C for 30 sec. Quantitative RT-PCR expression data were analyzed using a Tukey–Kramer honestly significant difference (HSD) test (α = 0.05). Primers and their concentrations are given in Table S1.
Yeast two-hybrid assay
A LexA yeast two-hybrid system was used to test protein interactions (Golemis et al., 2008). The vector pEG202 was used for bait and pJG4-5 for prey. Pto in bait and prey vector, AvrPtoB and its variants in bait vector, and AvrPtoB1–307 and its variants in prey vector were described previously (Lin et al., 2006; Xiao et al., 2007) while SlLyks were in bait vector. Screening for AvrPtoB tomato-interacting (Bti) proteins was described previously (Abramovitch et al., 2006) using a tomato cDNA prey library (Zhou et al., 1995).
Protein expression and in vitro kinase assay
Mutation of Bti9 Y489D was generated by site-directed mutagenesis using the primers LRZ10 and LRZ11 (Table S1). Bti9 and Bti9(Y489D) were expressed as fusions with MBP (pMAL-c2; New England Biolabs, http://www.neb.com/) in BL21 (DE3) Escherichia coli (Invitrogen) and purified using amylose resin (New England Biolabs). AvrPtoB(F173A) was prepared using site-directed mutagenesis (Xiao et al., 2007). AvrPtoB, AvrPtoB(F173A) and AvrPtoB1-307 were expressed as N-terminal GST-fused proteins (pDEST15, Invitrogen) in BL21 (DE3) E. coli cells and purified using glutathione Sepharose 4 fast flow resin (GE Healthcare, http://www.gehealthcare.com/). The GST-fused proteins were eluted from Sepharose 4 resin and concentrated using an Amicon Ultra Centrifugal Filter Unit (Millipore, http://www.millipore.com/). In vitro kinase assays were done in 30 μl reactions containing 200 ng of Bti9 or Bti9(Y489D), 7.5 μg myelin basic protein (M1891; Sigma-Aldrich, http://www.sigmaaldrich.com/), 5 μCi of [γ-32P]-ATP, 20 μl kinase buffer [25 mm 2-amino-2-(hydroxymethyl)-1,3-propanediol (Tris)–HCl, pH 7.5, 10 mm MgCl2, 10 mm NaCl, 1 mm DTT, 5 μm ATP, and1 mm EDTA] at room temperature for 30 min. The reactions were stopped with SDS-loading buffer and heated at 100°C for 5 min before electrophoresis on 12.5% SDS polyacrylamide gel. The gel was dried and radioactivity visualized using a PhosphorImager (Molecular Dynamics, http://www.gelifesciences.com/). For kinase inhibition assays, purified proteins of AvrPtoB or its derivatives were added to the reactions before they were incubated at room temperature (24°C). For visualization of protein loading, the dried gel was stained with Coomassie Brilliant Blue R250 and photographed after de-staining.
Standard methods were used for western blotting. Rat anti-hemagglutinin primary antibody (clone 3F10) was from Roche Applied Science (http://www.roche-applied-science.com/). Rabbit anti-LexA primary antibody was from Invitrogen. Horseradish peroxidase-conjugated secondary antibodies and an ECL Plus detection system (GE Healthcare) were used for detection. For detection of AvrPtoB, the rabbit polyclonal anti-AvrPtoB antibody was used with a similar dilution as described (Lin et al., 2006).
Sequence data have been deposited in GenBank. See Table 2 for details.
We thank Kent Loeffler from the Department of Plant Pathology and Plant-Microbe Biology, Cornell University for photography, Mamta Srivastava at the Plant Cell Imaging Center of the Boyce Thompson Institute for assistance with confocal microscopy, and Heather McLane for Figure S3. This work was supported, in part, by the National Science Foundation (IOS-0841807) and the National Institutes of Health (R01-GM078021) to GBM.