The Pseudomonas syringae effector AvrB interacts with four related soybean (Glycine max) proteins (GmRIN4a–d), three (GmRIN4b, c, d) of which also interact with the cognate resistance (R) protein, Rpg1-b. Here, we investigated the specific requirements for the GmRIN4 proteins in R-mediated resistance and examined the mechanism of Rpg1-b activation.
Using virus-induced gene silencing, we show that only GmRIN4a and b are required for Rpg1-b-mediated resistance. In planta binding assays show that GmRIN4a can associate with Rpg1-b indirectly via GmRIN4b. Pathogen-delivered AvrB induces the phosphorylation of GmRIN4b alone, and prevents interactions between GmRIN4b and Rpg1-b or GmRIN4a.
Consistent with this result, a phosphomimic derivative of GmRIN4b (pm4b) fails to bind Rpg1-b and GmRIN4a. Conversely, a phosphodeficient derivative of GmRIN4b (pd4b) continues to bind the R protein and GmRIN4a, in the presence of AvrB. Coexpression of Rpg1-b with pm4b, but not GmRIN4b or pd4b, induces cell death and ion leakage in the heterologous Nicotiana benthamiana.
Our data suggest that the AvrB-induced phosphorylation of GmRIN4b, and the subsequent inhibition of interaction among GmRIN4b, GmRIN4a and Rpg1-b, might activate the R protein. Furthermore, even though GmRIN4c and d are not required for Rpg1-b-derived resistance, they do function in resistance derived from other R loci.
Plant defense to microbial pathogens involves pre-formed barriers as well as induced responses. A well-known form of induced defense involves the deployment of plant resistance (R) proteins, which confer protection against specific races of pathogens. R-mediated specificity is determined by the presence of cognate effector molecules in the pathogen, termed avirulence (Avr) proteins (Flor, 1971). Direct/indirect interactions between the R and Avr proteins are thought to activate one or more signal transduction pathways that eventually prevent pathogen colonization. Early advances in the understanding of Avr-dependent defense activation in a host genotype-specific manner were made from the analyses of soybean (Glycine max) responses to the bacterial blight pathogen Pseudomonas syringae pv. glycinea (Psg) (Staskawicz et al., 1987). Psg infection of susceptible soybean cultivars results in chlorotic and spreading lesions, whereas resistant cultivars exhibit localized cell death, resulting in necrotic lesions in the cells surrounding the infection site. The localized cell death phenotype is one of the earliest visible manifestations of induced defense reactions in resistant plants, and is termed hypersensitive response (HR; Holliday et al., 1981).
Studies with naturally occurring strains of Psg led to the identification and characterization of several functional bacterial Avr proteins (Keen & Holliday, 1982). These include AvrA1Pgyrace6 (Napoli & Staskawicz, 1987), AvrB1Pgyrace4 (commonly designated AvrB, Staskawicz et al., 1987; Lindeberg et al., 2005), and AvrB2Pgyrace4 (original designation AvrC, Tamaki et al., 1988; Lindeberg et al., 2005). In addition, AvrD1PtoPT23 from P. syringae pv. tomato (Pto), the causal agent of bacterial speck in tomato, has been shown to impart avirulence to Psg in a host cultivar-specific manner (Keen et al., 1990; Kobayashi et al., 1990a). Several Psg strains also express endogenous AvrD1 alleles; however, these do not elicit HR in soybean (Kobayashi et al., 1990b; Keith et al., 1997).
Of these different Avr proteins, resistance to AvrB has been extensively studied. These studies have been carried out in the heterologous host Arabidopsis thaliana using AvrB-expressing Pto (Innes et al., 1993; Bisgrove et al., 1994; Grant et al., 1995; Mackey et al., 2002, 2003; Chung et al., 2011; Liu et al., 2011). In Arabidopsis, the RPM1 (resistance to P. syringae pv. maculicola) gene provides specificity to Pto strains expressing AvrB or AvrRPM1 (Keen & Holliday, 1982; Debener et al., 1991; Innes et al., 1993; Bisgrove et al., 1994; Grant et al., 1995). AtRIN4 (RPM1-interacting 4) protein, which binds both RPM1 and AvrB, is essential for RPM1 resistance (Mackey et al., 2003). Extensive analysis has shown that RPM1 activation is associated with the AvrB-dependent phosphorylation of AtRIN4 (Mackey et al., 2003). This phosphorylation of AtRIN4 via a receptor-like cytoplasmic kinase (designated RIPK for RPM1-induced protein kinase) was recently shown to be essential for RPM1 activation (Mackey et al., 2002; Chung et al., 2011; Liu et al., 2011).
Although genetic resistance to AvrB-expressing bacteria was first identified in soybean, the corresponding R gene was identified more recently. In soybean, resistance to AvrB-expressing Psg (avrB Psg) is determined by the Rpg1 (resistance to Psg) locus (Staskawicz et al., 1987; Keen et al., 1990; Keen & Buzzell, 1991). The Rpg1-b gene on this locus encodes a protein that belongs to the coiled coil-nucleotide binding-leucine rich repeat (CC-NB-LRR) class of R proteins and provides specificity against Psg avrB (Bisgrove et al., 1994; Ashfield et al., 2003; Ashfield et al., 2004). Although Rpg1-b and RPM1 are nonorthologous, they share several signaling features. For example, both R proteins require RAR1 (required for Mla12-mediated resistance, Torp & Jorgensen, 1986) for resistance signaling (Tornero et al., 2002; Fu et al., 2009), and neither R protein directly interacts with AvrB (Grant et al., 1995; Ashfield et al., 2004; Selote & Kachroo, 2010a). Furthermore, like RPM1, Rpg1-b also interacts with RIN4-like proteins; of the four related soybean proteins, designated GmRIN4a–d for their high sequence similarities to AtRIN4, GmRIN4b, c and d interact with Rpg1-b (Selote & Kachroo, 2010a,b). Like AtRIN4, the GmRIN4 proteins all bind AvrB and at least GmRIN4a and b are required for resistance derived from Rpg1-b (Selote & Kachroo, 2010a,b). Here, we further investigated the role of the four GmRIN4 isoforms in resistance derived from Rpg1-b and other R loci in soybean and examined the changes in these proteins in response to the AvrB effector.
Materials and Methods
Plant growth conditions
Soybean (Glycine max (L.) Merr.) cvs Harosoy/Merit (Rpg1-b) and Essex/Peking (rpg1-b) were grown in the glasshouse with day and night temperatures of 25 and 20°C, respectively. For silencing experiments, inoculation of recombinant BPMV vectors and confirmation of silencing were carried out as described previously(Kachroo et al., 2008).
Coimmunoprecipitation and in planta phosphorylation assays
Proteins were expressed as N-terminal epitope-tagged fusions in Nicotiana benthamiana using the pSITE vectors (Martin et al., 2009). Total extracts of plant proteins were prepared by grinding 1 g leaf tissue in buffer containing 10% glycerol, 25 mM Tris-HCl, pH 7.5, 1 mM EDTA, pH 8.0, 150 mM NaCl, 10 mM DTT, 0.15% NP-40, 2% polyvinylpolypyrrolidone (PVPP), and 1X plant protease inhibitor cocktail (Sigma). Immunoprecipitation (IP) assays were performed by incubating total protein extracts with M2 FLAG-affinity beads (unless noted otherwise), followed by extensive washing with extraction buffer lacking PVPP. Proteins were separated on 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) at 100 V for 1–2 h, and detected by Western blot analysis using protein/tag-specific antibodies. To test effects of Psg avrB, Psg avrA1, or Paraquat on GmRIN4 binding, the respective proteins were transiently expressed in 3- to 4-wk-old N. benthamiana plants using Agrobacterium. Forty hours after Agrobacterium infiltration, leaves were infiltrated with water, Paraquat (20 or 50 μM), Psg Vir, avrB, and avrA1 (2 × 107 cfu ml−1). Total proteins were extracted 12–16 h after Psg or Paraquat treatments and subject to IP. For band shift resulting from phosphorylation, protein extracts were specifically separated on 8% SDS-PAGE at 40 V for 8–10 h followed by western blotting. For in planta phosphorylation assays, MYC-tagged GmRIN4 proteins were expressed in N. benthamiana using Agrobacterium, followed by infiltration of Psg Vir or Psg avrB (2 × 107 cfu ml−1) along with 25 μCi of γ32P-dATP. Twelve hours later, total protein extracts were subject to IP using α-MYC antibodies. SDS-PAGE electrophoresis of IP extracts followed by phosphoimager analysis of the gel was used to detect 32P-labeled proteins. The approximate expected molecular weights of proteins were as follows: AvrB, 36 kD; FLAG-4a/4b, 30 kD; FLAG-4c, 29 kD; FLAG-4d, 31 kD; FLAG-Rpg1-b, 125 kD; FLAG-Rpg1-b (CC-domain), 23 kD; MYC-4a/4b, 28 kD; MYC-4c, 27 kD; MYC-4d, 29 kD; EGFP-4a/4b, 54 kD; RFP-4c, 53 kD; nEYFP-4d, 45 kD.
Construction of viral vectors, in vitro transcription and plant inoculation
Silencing of soybean sequences was carried out using virus-induced gene silencing (VIGS) with a previously described bean pod mottle virus (BPMV)-based vector (Zhang & Ghabrial, 2006). Generation of silencing vectors, in vitro transcription, and rub-inoculation of soybean leaves were carried out as described previously (Kachroo et al., 2008). The 159 bp fragment (G184-Q236) of GmRIN4a, the 243 bp fragment (T59-P139) of GmRIN4b, the 180 bp fragment (G69-I128) of GmRIN4c, or the 177 bp fragment (V73-D131) of GmRIN4d was used to generate viral vectors targeting GmRIN4a (Glyma03g19920), GmRIN4b (Glyma16g12160), GmRIN4c (Glyma18g36000), or GmRIN4d (Glyma08g46400), respectively.
RNA extraction, northern and reverse transcription PCR (RT-PCR) analysis
RNA from leaf tissues of soybean plants at V2/V3 growth stage was extracted using the TRIzol reagent (Invitrogen) according to the manufacturer's instructions. Northern blot analysis and synthesis of random-primed probes were carried out as described previously (Kachroo et al., 2008). Reverse transcription and first-strand cDNA synthesis were carried out using Superscript II (Invitrogen). Two to three independent RNA preparations were analyzed by semiquantitative RT-PCR using gene-specific primers to evaluate the relative differences in various transcript abundances.
Paraquat treatment and trypan blue staining of cell death
Paraquat was prepared in sterile water. Leaves from 3- to 4-wk-old N. benthamiana plants were infiltrated with either water or 20 and 50 μM Paraquat solutions. Leaves were sampled 24 h post-infiltration for cell death staining. Soybean leaves were vacuum-infiltrated with trypan blue stain prepared in 10 ml acidic phenol, 10 ml glycerol, and 20 ml sterile water with 10 mg of trypan blue. The samples were placed in a heated water bath (90°C) for 2 min and incubated at room temperature for 2–12 h. The samples were destained using chloral hydrate (25 g per 10 ml sterile water; Sigma), mounted on slides and observed for cell death with a compound microscope.
For conductivity measurements, leaves of 3- to 4-wk-old N. benthamiana plants were infiltrated with Agrobacterium strains carrying the respective proteins. For ion leakage and cell death in response to AvrB expression, reduced amounts (105) of AvrB-expressing Agrobacterium cells were infiltrated. Thirty-six hours later, five leaf discs per treatment (7 mm) were removed with a cork borer, washed in distilled water for 50 min, and subsequently transferred to tubes containing 5 ml of distilled water. Conductivity of the solution was determined with a NIST traceable digital conductivity meter (Fisher Scientific, Waltham, MA, USA) at the indicated time points. Standard deviation was calculated from four replicate measurements per treatment per experiment. Results are representative of three independent experiments.
Pathogen strains and inoculations
Pseudomonas syringae pv. glycinea race 4 expressing AvrB, AvrB2, or AvrA1 via the broad host range plasmid pDSK519, and AvrD1 via the broad host range plasmid pDSK600 (Staskawicz et al., 1984; Keen et al., 1988; Tamaki et al., 1988) was used. A strain expressing the empty pDSK519 plasmid was used as the Vir control. Psg strains were grown on King's B medium at 28°C, supplemented with rifampicin 50 μg ml−1 plus kanamycin 50 μg ml−1. Psg inoculation of soybean and monitoring of bacterial proliferation were carried out as described previously (Fu et al., 2009). Mock inoculations were carried out with 10 mM MgCl2 in 0.04% Silwett L-77. Results are representative of three to four independent repeats, unless noted otherwise. Race 1 of Phytophthora sojae was grown on V8 agar at 25°C in the dark. Soybean inoculations with P. sojae were carried out as described previously (Kachroo et al., 2008). Results are representative of three to five independent experiments with 15–20 plants tested per silenced line, per experiment.
GmRIN4c and GmRIN4d are not required for Rpg1-b-derived resistance
Previously we showed that the GmRIN4a and b isoforms are required for resistance to Psg avrB in Rpg1-b plants (Selote & Kachroo, 2010a). The GmRIN4c and d isoforms are closely related to GmRIN4a and b, and bind both Rpg1-b and AvrB (Selote & Kachroo, 2010a,b). To test if GmRIN4c or d also contributed to Rpg1-b-mediated resistance, we silenced the GmRIN4a, b, c, and d genes individually in Rpg1-b plants (cv Merit; Supporting Information, Fig. S1a) using a BPMV-based vector for VIGS in soybean (Zhang & Ghabrial, 2006). Specific silencing of targeted isoforms was confirmed by RT-PCR analysis. Expression of two other related (< 20% sequence identity) sequences was not affected in the GmRIN4-silenced plants either (Fig. S1a). The GmRIN4-silenced plants were inoculated with Psg avrB or Psg expressing the plasmid vector pDSK519 (Vir). As expected, Psg avrB was less virulent on control plants (designated V, for empty silencing vector) than Psg Vir, and silencing GmRIN4a or b resulted in susceptibility to Psg avrB (Fig. 1a). By contrast, silencing GmRIN4c or d did not alter Rpg1-b resistance to Psg avrB; levels of Psg avrB were comparable in control, GmRIN4c- or d-silenced plants, while GmRIN4a or b silenced plants accumulated sevenfold more bacteria. These results, which were confirmed in another Rpg1-b cultivar, Harosoy (Fig. S2a), indicated that GmRIN4c and d are not required for resistance derived from Rpg1-b.
Alternatively, GmRIN4c and d were functionally redundant. The high sequence similarities amongst the GmRIN4 isoforms precluded the simultaneous silencing of GmRIN4c and d without affecting the expression of GmRIN4a or b. We addressed the potential functional redundancy of GmRIN4c and d by testing their roles in defense to virulent pathogens, as silencing GmRIN4a or b enhanced resistance to virulent strains of Psg and the oomycete P. sojae, as well as resistance to Psg avrB in plants lacking Rpg1-b (Selote & Kachroo, 2010a). The rpg1-b plants (cv Essex, Fig. S1b) silenced for GmRIN4c and d were infected with Psg Vir, Psg avrB, or P. sojae (race 1). Psg avrB was twofold (P <0.0001) more virulent on rpg1-b plants (Fig. 1b). The GmRIN4c or d silenced plants showed a small (twofold), but significant (P <0.001) increase in Psg Vir growth (Fig. 1b) as compared with control plants (V). This indicated that, unlike GmRIN4a or b, silencing GmRIN4c or d did not enhance resistance to Psg Vir. Furthermore, the GmRIN4c or d silenced plants accumulated more Psg avrB than control plants. However, the relative difference in growth of Psg avrB vs Psg Vir in GmRIN4c or d silenced plants were comparable to those in control plants (compare gray and black bars for the respective line). The increased accumulation of Psg avrB was likely the result of overall increased susceptibility in GmRIN4c- or d-silenced plants since these plants were also more susceptible to P. sojae. At 4 d postinoculation, > 40% GmRIN4c- or d-silenced plants succumbed to disease in comparison to only 20% of control plants (Fig. S2b). The increased susceptibility to P. sojae was also observed in the Harosoy cultivar. However, unlike in Essex, silencing GmRIN4c- or d did not enhance susceptibility to Psg Vir in Harosoy plants (Figs 1b, 2a). This indicated that while GmRIN4c and d were generally required for basal resistance to P. sojae, their role in basal defense to Psg might be cultivar-specific, possibly influenced by the genetic variability amongst different cultivars. Importantly, these results demonstrated that the lack of either GmRIN4c or d could affect basal defense to virulent pathogens in soybean.
GmRIN4c and d function nonredundantly in resistance derived from other R loci
To further clarify the functional nonredundancy of GmRIN4c and d in R-mediated defense, we tested their requirements for resistance derived from other R loci. Besides Rpg1-b, three other known soybean R loci mediate race-specific resistance to Psg. These include Rpg2, Rpg3, and Rpg4, which specify resistance against Psg expressing AvrA1, AvrB2 and AvrD1, respectively (Napoli & Staskawicz, 1987; Keen et al., 1990; Keen & Buzzell, 1991). We tested if any of the GmRIN4 isoforms contributed to resistance derived from these loci. The GmRIN4a, b, c and d were silenced in Rpg2 (cv Merit) and Rpg3 Rpg4 (cv Flambeau) plants (Fig. S1b) and analyzed for their responses to the respective avr strains of Psg. As expected, Rpg2 plants accumulated fewer (13-fold) Psg avrA1 (Fig. 2a), while the Rpg3 Rpg4 plants accumulated fewer Psg avrB2 and Psg avrD1 (Fig. 2b), in comparison to Psg Vir. The GmRIN4a- or b-silenced Rpg2 plants were not altered in resistance to Psg avrA1. However, Rpg2 plants silenced for GmRIN4c or d accumulated significantly (P <0.001) more (eightfold) Psg avrA1, at levels comparable to Psg Vir growth in control plants (Fig. 2a). This indicated that GmRIN4c and d, but not GmRIN4a or b, were required for Rpg2-derived resistance.
In contrast to Rpg2, Rpg3-derived resistance required GmRIN4b since the GmRIN4b-silenced Rpg3 Rpg4 plants accumulated sixfold more Psg avrB2 than control plants (Fig. 2b). Interestingly, plants silenced for GmRIN4a, c, or d accumulated significantly (P <0.0001) fewer (> 13-fold) Psg avrB2 than the control plants (Fig. 2b). This was consistent with increased Psg avrB2-responsive visible and microscopic cell death in these plants (Fig. S3). This suggested that, whereas GmRIN4b was required for Rpg3-derived resistance, the loss of GmRIN4a, c, or d enhanced resistance in this background. We tested the specificity of this against Psg avrB2, by inoculating the GmRIN4a-, c-, or d-silenced Rpg3 Rpg4 plants with Psg Vir or Psg avrD1. Similar to Psg avrB2, resistance to Psg Vir was also significantly (P <0.001) enhanced in these plants (Fig. 2b, grey bars). By contrast, the GmRIN4a-, c- or d-silenced Rpg3 Rpg4 plants did not exhibit enhanced resistance to Psg avrD1; Psg avrD1 levels in the GmRIN4-silenced Rpg3 Rpg4 plants were comparable (GmRIN4d-silenced) or slightly higher (two- to threefold in GmRIN4a-, b-, or c-silenced, P <0.002) than those in control plants (Fig. 2b). Notably, only the GmRIN4a-silenced plants accumulated more Psg avrD1 than they did Psg Vir. However, Psg avrD1 levels in the Rpg3 Rpg4 GmRIN4a-silenced plants were significantly lower than those of Psg Vir in control plants. Likewise, even though the GmRIN4c-silenced plants accumulated similar levels of Psg avrD1 as Psg Vir, these levels were lower than those of Psg Vir in control plants. This suggested that absence of GmRIN4a or GmRIN4c might only partially hamper resistance derived from the Rpg4 locus. Alternatively, the heightened background resistance in the respective silenced plants might mask the requirements for GmRIN4a or GmRIN4c by Rpg4. Importantly, these results demonstrated the functional nonredundancy of the GmRIN4c and d isoforms in R-mediated resistance. At this point, it is unclear how silencing GmRIN4a, c, or d enhances resistance in the Flambeau background. At least one possibility is the ectopic activation of R/R-like proteins as a result of silencing these GmRIN4 isoforms, similar to that in Arabidopsis (ecotype Col-0). For example, the Atrin4 mutation and the Pto AvrRpt2-derived degradation of AtRIN4 activate the R proteins RPM1 and RPS2, respectively (Axtell & Staskawicz, 2003; Mackey et al., 2003; Belkhadir et al., 2004).
The compromised R-mediated resistance in the various silenced lines was specific because silencing the GmRIN4 isoforms did not alter resistance derived from Rsv1 (specifies resistance to the G5 strain of soybean mosaic virus (SMV); Buzzell & Tu, 1984; Lim, 1985), or Rps1a and Rps1k (specify resistance to P. sojae race 1 and 3, respectively; Whisson et al., 1995; Gijzen et al., 1996) loci. Rsv1 plants (cv Essex-Rsv1) silenced for GmRIN4a, b, c or d were resistant to SMV (G5), while Rps1a (cv Harosoy 63) and Rps1k (cv Williams 82) plants silenced for GmRIN4a, b, c or d were resistant to P. sojae races 1 and 3, respectively (Fig. S4a,b).
AvrB induces the phosphorylation of GmRIN4b
Our results thus far showed that GmRIN4a and b were required for Rpg1-b derived resistance, whereas GmRIN4c and d were not (at least not the individual isoforms). We next attempted to understand how GmRIN4a and/or b contribute to Rpg1-b derived resistance. In Arabidopsis, the structurally related AtRIN4 activates the R protein RPM1 upon phosphorylation in the presence of AvrB (Mackey et al., 2002; Chung et al., 2011; Liu et al., 2011). We considered the possibility that AvrB-dependent phosphorylation of GmRIN4 proteins might likewise be associated with the activation of Rpg1-b. Sequence comparisons showed that all GmRIN4 isoforms contain the T residues corresponding to the phosphorylated T21 and T166 of AtRIN4 (Fig. S5). However, only GmRIN4a contains the S residue (S192) corresponding to the phosphorylated S160 of AtRIN4. We tested AvrB-dependent phosphorylation of the GmRIN4 isoforms by analyzing these proteins when transiently expressed in N. benthamiana leaves inoculated with Psg avrB or Vir. Interestingly, only GmRIN4b showed reduced mobility in response to Psg avrB, but not Psg Vir (Fig. 3a). Similar results were obtained in response to the Pto avrB strain, suggesting that the change in the mobility of GmRIN4b was AvrB-dependent. Incubation of protein extracts with calf intestinal phosphatase (CIP) restored the mobility of GmRIN4b (Fig. 3b) indicating that phosphorylation contributed to the mobility shift of GmRIN4b. AvrB-derived phosphorylation of GmRIN4b was further confirmed using an in planta phosphorylation assay (Fig. 3a, middle panel). In contrast to Psg avrB, Psg avrA1 did not alter the mobility or induce phosphorylation of GmRIN4b, even though like Psg avrB, this strain also induced a cell death response in N. benthamiana (Fig. S6). These results indicated that Psg avrB-induced phosphorylation of GmRIN4b was specific to the AvrB effector, and not the result of Psg infection or the induced cell death response.
Our results thus far showed that the requirements for GmRIN4a and b in Rpg1-b-derived resistance correlated neither with their abilities to interact with the R protein nor with their phosphorylation in response to AvrB. Although both GmRIN4a and b were required for Rpg1-b function, only GmRIN4b directly interacted with the R protein and was phosphorylated in an AvrB-dependant manner (Selote & Kachroo, 2010a). To test if GmRIN4a was present in the Rpg1-b-GmRIN4b complex, we coexpressed epitope-tagged GmRIN4a, b, and Rpg1-b in N. benthamiana and subjected total protein extracts from these plants to immunoprecipitation (IP) using antibodies specific to the tag on Rpg1-b. Interestingly, both GmRIN4a and b were detected in this IP (Fig. 3c), indicating that GmRIN4a was able to indirectly associate with Rpg1-b via GmRIN4b.
Phosphorylation of GmRIN4b inhibits its interaction with Rpg1-b
We tested if the AvrB-dependent phosphorylation of GmRIN4b affected its ability to bind Rpg1-b by analyzing the GmRIN4- Rpg1-b interactions in N. benthamiana leaves inoculated with Psg avrB or Psg Vir. As in healthy leaves (Selote & Kachroo, 2010a), GmRIN4a did not interact with Rpg1-b in Psg Vir -infected leaves, whereas GmRIN4b, c and d did (Fig. 4a). Interestingly, GmRIN4b was unable to bind Rpg1-b in the presence of Psg avrB (Fig. 4a). No change in the interaction of Rpg1-b with GmRIN4a, c or d was observed in the presence of Psg avrB; GmRIN4c and d bound Rpg1-b, while GmRIN4a did not. Furthermore, the GmRIN4b-Rpg1-b interaction was unaltered in the presence of Psg avrA1 (Fig. 4b). This suggested that the inhibition of GmRIN4b-Rpg1-b binding was specific to Psg avrB and likely not the result of signaling induced in response to bacterial infection. Furthermore, this suggests that AvrB specifically inhibited the binding of GmRIN4b to Rpg1-b independent of pathogen-induced cell death response.
To test if phosphorylation of GmRIN4b contributed to the inhibition of its interaction with Rpg1-b, we generated phosphomimic (T22D/T198D, pm4b) or phosphodeficient (T22A/T198A, pd4b) derivatives of GmRIN4b by mutagenizing T22 and T198 (corresponding to T21 and T166 in AtRIN4) of GmRIN4b. The pm4b and pd4b derivatives were then tested for binding to Rpg1-b. As predicted, pd4b was not phosphorylated in the presence of Psg avrB; the mobility of pd4b was not altered in the presence of Psg avrB (Fig. S7a). IP assays showed that both pm4b and pd4b bound MYC-tagged AvrB, which was detected in the membrane fractions of N. benthamiana leaf extracts (Fig. S7b,c). Interestingly, only pd4b was able to bind Rpg1-b, suggesting that AvrB-dependent phosphorylation of GmRIN4b inhibited its binding with Rpg1-b (Fig. 4c). This was consistent with the Psg avrB-dependent inhibition of the GmRIN4b–Rpg1-b interaction. We tested this further by assaying binding between Rpg1-b and pd4b or pm4b in the presence of Psg avrB. Indeed, Psg avrB did not inhibit the pd4b–Rpg1-b interaction (Fig. 4d), suggesting that phosphorylation of GmRIN4b was essential for the AvrB-dependent inhibition of the GmRIN4b–Rpg1-b interaction.
Phosphorylation of GmRIN4b inhibits its interaction with self and GmRIN4a
The phosphorylation-dependent inhibition of the GmRIN4b-Rpg1-b interaction, the detection of the GmRIN4a–GmRIN4b–Rpg1-b complex, and the ability of GmRIN4b to interact with self and GmRIN4a directly (Selote & Kachroo, 2010a,b) prompted us to test whether the Psg avrB-dependent phosphorylation of GmRIN4b inhibited its interaction with self and/or GmRIN4a. As observed in healthy leaves (Selote & Kachroo, 2010a), GmRIN4b continued to bind itself and GmRIN4a in the Psg Vir-infiltrated leaves (Fig. 5a). By contrast, Psg avrB drastically reduced the self-binding of GmRIN4b, as well as its binding to GmRIN4a (Fig. 5a). To confirm that the inhibition of GmRIN4b interactions were not the result of cell death induced in response to Psg avrB (Selote & Kachroo, 2010a), we tested these interactions in leaves treated with Paraquat (1,1′-dimethyl-4,4′-bypiridilium). Paraquat, which promotes the formation of reactive oxygen species by inhibiting electron transport during photosynthesis (Farrington et al., 1973; Hiyama et al., 1993), did not alter the binding of GmRIN4b to self or GmRIN4a, when applied at two different cell death-inducing concentrations (Fig. S8; data shown for self-interaction of GmRIN4b).
These results suggest that similar to its binding with Rpg1-b, the AvrB-dependent phosphorylation of GmRIN4b also inhibited its binding to self and GmRIN4a. We tested this further by analyzing binding of pm4b or pd4b with GmRIN4a and b. IP assays showed that pd4b was able to bind both GmRIN4a and b, whereas pm4b did not bind GmRIN4b, and showed reduced binding to GmRIN4a (Fig. 5b). Moreover, as in the absence of pathogen or in the presence of Psg Vir, pd4b associated with GmRIN4a and b in the presence of Psg avrB (Fig. 5c). Together, these results suggest that the phosphorylation of GmRIN4b is responsible and essential for its inability to bind Rpg1-b, GmRIN4a and self in Psg avrB-infected leaves.
AvrB-dependent phosphorylation of GmRIN4b activates Rpg1-b
AvrB-dependent phosphorylation of AtRIN4 was recently shown to activate RPM1 (Chung et al., 2011; Liu et al., 2011). To test if the AvrB-dependent phosphorylation of GmRIN4b similarly activated Rpg1-b, we used a bioassay based on the induction of cell death and ion leakage resulting from the activation of Rpg1-b. Transient coexpression (via Agrobacterium) of FLAG-Rpg1-b with empty plasmid vector, and with either, all, or various combinations of the MYC-GmRIN4 isoforms did not induce cell death or ion leakage in N. benthamiana (Figs 6, S9a). By contrast, coexpression of Rpg1-b with AvrB induced significant ion leakage and cell death (Figs 6c, S9a). This was not significantly altered in the presence of the GmRIN4 proteins (Fig. 6d). Interestingly, coexpression of Rpg1-b with pm4b induced cell death and ion leakage in the absence of AvrB, and these levels were comparable to when Rpg1-b was coexpressed with AvrB (Figs 6, S9a). This was not the case when pm4b was coexpressed with the empty plasmid vector, or when Rpg1-b was coexpressed with pd4b, even though all proteins were expressed at similar levels (Fig. 6b). Notably, coexpression of Rpg1-b with GmRIN4b derivatives carrying phosphomimic substitutions in any one of the two T residues (T22D4b or T198D4b) also failed to induce ion leakage (Fig. 6c). These results support the notion that phosphorylation of GmRIN4b contributes to the activation of Rpg1-b, and suggest that phosphorylation of GmRIN4b at both T22 and T198 is essential for this activation.
The soybean GmRIN4 family comprises four related isoforms, which bind the P. syringae avirulence effector, AvrB. We show that only two (GmRIN4a and b) of these isoforms are essential for resistance derived from Rgp1-b, the soybean R protein that specifies resistance to Psg avrB. The Psg avrB-dependent phosphorylation of GmRIN4b not only inhibits its own association (and, thereby, that of GmRIN4a) with Rpg1-b, but also activates Rpg1-b. In addition to Rpg1-b, the four GmRIN4 proteins contribute to bacterial resistance derived from three other R loci, including Rpg2, 3, and 4.
Our previous work showing that GmRIN4b directly binds Rpg1-b, but that GmRIN4a does not, posed the dilemma of how the Rpg1-b noninteracting GmRIN4a might contribute to resistance derived from this R protein. The data presented here support the possibility of a GmRIN4a–GmRIN4b–Rpg1-b complex in planta, and inhibition of these associations by Psg avrB. This raised the possibility that the dissociation of the GmRIN4a-b complex from Rpg1-b may contribute to the activation of resistance signaling. Supporting this notion are the facts that GmRIN4b directly interacts with Rpg1-b and GmRIN4a (Selote & Kachroo, 2010a) and that the three proteins were detected in a single complex when expressed heterologously in planta. Interestingly, the presence of Psg avrB inhibited binding of GmRIN4b with GmRIN4a or Rpg1-b. Psg avrB also specifically induced the phosphorylation of GmRIN4b. Clearly, the AvrB-dependent phosphorylation of GmRIN4b was necessary and sufficient to inhibit its binding with GmRIN4a or Rpg1-b; a phosphomimic derivative of GmRIN4b was unable to bind either GmRIN4a or Rpg1-b. Moreover, a phosphodeficient GmRIN4b derivative continued to bind both GmRIN4a and Rpg1-b even in the presence of Psg avrB. This raises the possibility that the dissociation of Rpg1-b from GmRIN4a and GmRIN4b in the presence of Psg avrB might be associated with Rpg1-b activation. However, such a possibility can be discounted, unless N. benthamiana encodes functionally orthologous proteins that can associate with Rpg1-b, because expression of Rpg1-b alone did not induce cell death or ion leakage in N. benthamiana. Furthermore, coexpression of Rpg1-b with the phosphomimic GmRIN4b derivative induced ion leakage and cell death, in the absence of GmRIN4a. This suggests that the AvrB-dependent phosphorylation of GmRIN4b is sufficient to activate Rpg1-b, at least in the heterologous N. benthamiana. Thus, it is possible that, whereas GmRIN4b contributes to the activation of Rpg1-b, GmRIN4a might contribute to resistance signaling, downstream of Rpg1-b activation (Fig. 7). Another possibility is that N. benthamiana ortholog(s) of RIN4 facilitated the pm4b-mediated activation of Rpg1-b in the absence of GmRIN4a. Notably, BLAST analysis of the draft N. benthamiana genome (solgenomics.net, Gomez et al., 2012) identified at least three sequences with significant identities to the GmRIN4a protein (Fig. S10a,c). Furthermore, western blot analysis of total protein extracts from N. benthamiana leaves detected protein(s) of similar molecular weight as GmRIN4, which cross-react with GmRIN4a antibodies (Fig. S10b). Although these data support the possibility of such proteins facilitating pm4b-mediated activation of Rpg1-b in N. benthamiana, conclusive evidence will require the analyses of their binding affinities for Rpg1-b, AvrB, and GmRIN4b, and the effects of AvrB on such associations. Therefore, while our findings clarify the putative mechanism of Rpg1-b activation in the heterologous N. benthamiana, it is quite possible that activation of resistance signaling in the native soybean may involve additional unidentified events.
Coexpression of Rpg1-b with AvrB induced significant ion leakage in N. benthamiana, in the absence of GmRIN4b. Furthermore, presence of GmRIN4b or the other GmRIN4 isoforms did not significantly alter this response. This suggests that N. benthamiana-encoded RIN4 orthologs likely enable the activation of Rpg1b in the presence of AvrB. Apparently these factors do not interfere with the Avr-independent activation of Rpg1-b by pm4b. Notably, this Avr-independent activation of Rpg1-b required phosphomimic substitutions at both T22 and T198 residues of GmRIN4b, because unlike pm4b (T22/198D), the T22D or T198D derivatives were unable to induce significant ion leakage when coexpressed with Rpg1-b. By contrast, a phosphomimic substitution at T166 of AtRIN4 was sufficient to activate RPM1 (Chung et al., 2011). Furthermore, unlike the GmRIN4b–Rpg1-b interaction, phosphorylation of AtRIN4 does not inhibit its interaction with RPM1. Also, the AtRIN4–RPM1 interaction is thought to be essential for RPM1 activation in the absence of pathogen effector (Chung et al., 2010). Unlike phosphomimic AtRIN4, which did not bind AvrB in yeast (Liu et al., 2009), pm4b was able to associate with AvrB in IP assays. These differences could perhaps be attributed to the fact that multiple isoforms participate in Rpg1-b signaling in soybean, whereas RPM1 activation and derived signaling have been studied in the context of the single AtRIN4. Notably, Arabidopsis encodes several other AtRIN4-related proteins whose functions in RPM1 activation and/or signaling have not yet been examined.
The Psg avrB-dependent phosphorylation of GmRIN4b is not surprising, given the fact that this protein shares very high sequence similarity to AtRIN4, whose AvrB-dependent phosphorylation is well documented. Surprisingly, AvrB-dependent phosphorylation was not detected for GmRIN4a, c and d, even though these isoforms share equally high sequence similarities to AtRIN4. Like GmRIN4b, GmRIN4a, c and d also contain the T residues corresponding to the phosphorylated residues of AtRIN4 (Chung et al., 2011; Liu et al., 2011). Moreover, GmRIN4a, but not GmRIN4b, contains the corresponding S residue, which is phosphorylated in AtRIN4. One untested possibility is that only the GmRIN4b isoform may serve as a substrate for RIPK-like protein kinase(s). Testing the substrate specificities of RIPK-like proteins from soybean during pathogen presence might help to address this. The fact that the AvrB-dependent phosphorylation of GmRIN4b was detectable in the heterologous N. benthamiana suggests that N. benthamiana also encodes RIPK-like proteins, which phosphorylate AtRIN4. This is supported by the observation that related proteins can possibly phosphorylate AtRIN4 in Arabidopsis plants lacking RIPK (Liu et al., 2011). The AvrB-dependent phosphorylation of GmRIN4b, but not GmRIN4a, does correlate with our previous report demonstrating the ability of GmRIN4b, but not GmRIN4a, to complement RPM1 function in the atrin4 mutant (Selote & Kachroo, 2010a) as well as the fact that RPM1 activation requires the phosphorylation of AtRIN4 (Chung et al., 2011; Liu et al., 2011). Thus, it is possible that GmRIN4a was unable to complement RPM1 function because it was not phosphorylated in an AvrB-dependent manner.
Both GmRIN4c and d bind Rpg1-b and AvrB (Selote & Kachroo, 2010a), yet silencing the genes encoding these proteins did not alter Rpg1-b-derived resistance. One possibility is that GmRIN4c and d function redundantly in resistance derived from this R protein. Testing this would require silencing GmRIN4c and d simultaneously, without affecting the expression of GmRIN4a or b, which we were unable to achieve because of the high sequence similarities between these isoforms. However, the GmRIN4c and d isoforms do function independently in resistance derived from other R loci, as well as in basal defense to virulent pathogens. Interestingly, the absence of GmRIN4a, c, or d resulted in the activation of Rpg3 and/or other associated R loci in Flambeau plants. This is similar to the inappropriate activation of RPM1, and the ectopic induction of RPS2 in Arabidopsis plants lacking AtRIN4 (Mackey et al., 2002, 2003; Belkhadir et al., 2004). Interestingly, Rpg1-b does not appear to be activated by the absence of any of the GmRIN4 proteins. Understanding the precise functions of each of the GmRIN4 isoforms and how they regulate the activation and/or signaling derived from multiple R proteins could provide important insights into how plants perceive and activate defenses in response to diverse pathogen effectors.
This work was made possible by the generous help of Said Ghabrial (BPMV vector), Adam Bogdanove and Massimo Delledone (Psg strains), Alan Collmer (AvrB antibodies), David Zaitlin (identifying N. benthamiana RIN4-like sequences), Pradeep Kachroo (critical review of the manuscript), and Amy Crume (management of plant growth facilities). This work was supported by a grant from the United Soybean Board (project 1291) to A.K.