Structure-Function Analysis of the Coiled-Coil and Leucine-Rich Repeat Domains of the RPS5 Disease Resistance Protein 1[W]

The Arabidopsis RPS5 disease resistance protein mediates recognition of the Pseudomonas syringae effector protein AvrPphB. RPS5 belongs to the coiled-coil-nucleotide binding site-leucine-rich repeat (CC-NBS-LRR) family and is activated by AvrPphB-mediated cleavage of the protein kinase PBS1. Here we present a structure-function analysis of the CC and LRR domains of RPS5 using transient expression assays in Nicotiana benthamiana . We found that substituting the CC domain of RPS2 for the RPS5 CC domain did not alter RPS5 specificity and only moderately reduced its ability to activate programmed cell death , suggesting that the CC domain does not play a direct role in the recognition of PBS1 cleavage. Analysis of an RPS5-super Yellow Fluorescent Protein (sYFP) fusion revealed that RPS5 localizes to the plasma membrane (PM). Alanine substitutions of predicted myristoylation (glycine 2) and palmitoylation residues (cysteine 4) affected RPS5 PM localization, protein stability, and function in an additive manner, indicating that PM localization is essential to RPS5 function. The first 20 amino acids of RPS5 were sufficient for directing sYFP to the PM. C-terminal truncations of RPS5 revealed that the first 4 LRR repeats are sufficient for inhibiting RPS5 autoactivation; however, the complete LRR domain was required for recognition of PBS1 cleavage. Substitution of the RPS2 LRR domain resulted in of RPS5, indicating that the LRR domain must co-evolve with the NBS domain. We conclude that the RPS5 LRR domain RPS5 of


INTRODUCTION
To defend themselves against pathogens, plants have evolved a two-tiered innate immune system. The first tier uses transmembrane pattern recognition receptors to detect the presence of pathogen associated molecular patterns (PAMPs), which are broadly conserved molecules produced by the majority of a given class of micro-organism (e.g. flagellin for bacteria and chitin for fungi) (Jones and Dangl, 2006). Upon detection of PAMPs, these receptors activate multiple signaling pathways, including a mitogenactivated protein-kinase (MAPK) cascade. This is called PAMP triggered immunity (PTI).
Successful pathogens can suppress PTI by delivering effector proteins into host cells to block signaling pathways at various steps (Desveaux et al., 2006;Kamoun, 2006;Dodds et al., 2009). This has led to the evolution of a second tier in the plant immune system consisting of intracellular receptors capable of recognizing the presence of effectors (Jones and Dangl, 2006). This effector triggered immunity (ETI) typically leads to a localized activation of programmed cell death called the hypersensitive response (HR).
The immune receptors of the ETI system are encoded by resistance (R) genes. Most R proteins are intracellular and are composed of a central nucleotide binding site (NBS) and a C-terminal leucine-rich repeat (LRR) domain, (McHale et al., 2006;Caplan et al., 2008;Takken and Tameling, 2009). The NBS domain of plant R-proteins shares structural and functional similarities with the metazoan apoptosis factors Apaf-1 and CED-4 (van der Biezen and Jones, 1998;Albrecht and Takken, 2006;Takken et al., 2006) and appears to function as a molecular switch regulated by nucleotide-dependent conformational changes (Tameling et al., 2002;Tameling et al., 2006;Collier and Moffett, 2009). Indeed, the NBS domain of a plant R-protein has been shown to bind and hydrolyze ATP (Tameling et al., 2002) and mutations in the NBS domain that block ATP binding also compromise function (Tameling et al., 2006).
Plant NBS-LRR proteins are divided into two classes based on their N-terminal domains. The TIR-NBS-LRR class contains an N-terminal domain with similarity to Toll and Interleukin-1 Receptors (TIR), while the non-TIR class most often contains a predicted coiled-coil (CC) domain (Meyers et al., 1999;Cannon et al., 2002). Over-  (Frost et al., 2004;Zhang et al., 2004;Swiderski et al., 2009;Krasileva et al., 2010;Bernoux et al., 2011;Collier et al., 2011;Maekawa et al., 2011). HR-induction appears to require homodimerization as amino acid substitutions that reduced dimerization reduced HR strength (Bernoux et al., 2011;Maekawa et al., 2011). However, over expression of the Arabidopsis RPS5 CC domain by itself in N. benthamiana does not induce HR, whereas over expression of the CC-NBS region does (Ade et al., 2007). In addition, over expression of the NB subdomain of the potato CC-NBS-LRR protein Rx can induce HR by itself when fused to GFP (Rairdan et al., 2008), suggesting that the NBS domain of at least some CC-NBS-LRR proteins plays a role in downstream signaling. The molecular mechanism regulating downstream signaling by CC-NBS-LRR proteins thus remains unclear.
The N-termini of plant NBS-LRR proteins are also proposed to participate in effector recognition. Many plant NBS-LRR proteins appear to recognize their corresponding effectors by sensing effector-induced modifications of other host proteins (Van der Biezen and Jones, 1998;Dangl and Jones, 2001). Several R-proteins have been shown to interact with their corresponding effector targets via their N-terminal domains (Mackey et al., 2002;Mucyn et al., 2006;Ade et al., 2007;Burch-Smith et al., 2007), indicating that N-terminal domains contribute to the specificity of effector recognition. This is supported by the observation that recognition specificities of specific alleles of the flax L locus can be swapped by swapping only the N-terminal TIR domains (Luck et al., 2000).
Generally, the LRR domain is presumed to play a central role in the recognition of pathogen effectors. In support of this, the LRR domains of many plant R proteins have been shown to be under diversifying selection (McDowell et al., 1998;Meyers et al., 1998;Ellis et al., 2000), implying that this domain in particular has been co-evolving with pathogen effectors. Indeed, the LRR domains of some R proteins are known to bind pathogen effectors directly (Deslandes et al., 2003;Dodds et al., 2006;Krasileva et al., 2010), and in the case of the L and P loci from flax, recognition specificity can be swapped by swapping the LRR domains between alleles (Ellis et al., 1999;Dodds et al., 2001 ). We therefore first tested whether the 2-5-5 chimera could induce HR by itself.
However, when the 2-5-5 protein was co-expressed with PBS1 and AvrPphB, a typical HR was observed (Fig. 1A (Fig. 1B). Interestingly, when the 2-5 (D266E) -5 chimera was co-expressed with PBS1 and AvrPphB, a strong HR was observed (Fig. 1B). This HR depended on PBS1 cleavage as co-expression with PBS1 and C98S did not trigger HR (supplemental Fig. S1B). These observations indicate that substitution of the RPS2 CC domain for the RPS5 CC domain reduces the efficiency of RPS5 activation and/or signaling. They also indicate that the activation of RPS5 caused by the D266E mutation is weaker than that triggered by PBS1 cleavage.
Surprisingly, introduction of the D266E mutation into the 2-5 chimera (2-5 (D266E) ) restored autoactivity (Fig. 1C). Therefore, deletion of the LRR and the D266E substitution have additive effects on autoactivation, and neither one alone is sufficient to overcome the loss of efficiency associated with substitution of RPS2 CC domain for the . This result is consistent with the observed reduction in HR observed for the 2-5 D266E -5 chimera in the transient system (Fig. 1B), and suggests that the 2-5-5 chimera falls below a threshold required for HR induction when expressed under native protein levels.

Fashion to RPS5 Localization and Function
RPS5 contains both a predicted myristoylation motif at glycine 2 and a palmitoylation motif at cysteine 4. It also contains a glycine at position 3 that could be potentially myristoylated if G2 were removed. To investigate the importance of acylation in RPS5 function, we mutated the glycine residues at positions 2 and 3 to alanine (G2/3A) and the cysteine at position 4 to alanine (C4A), both as separate mutations and combined. G2/3A-RPS5 was still able to activate HR in response to PBS1 cleavage by AvrPphB when transiently overexpressed in N. benthamiana ( Fig. 2A). However, the G2/3A substitution nearly eliminated the autoactivity of the full-length RPS5 mutant D266E, without affecting protein accumulation at four hours post induction (Fig. S2B). Like the 2- However, the G2/3AC4A mutation completely abolished RPS5-induced electrolyte leakage. Interestingly, the G2/3A-RPS5 mutant induced only a moderate conductivity increase at 8 hours post DEX induction, but an electrolyte leakage comparable to WT RPS5 at 24 hours post DEX induction (Fig. 2D). This is consistent with the macroscopic HR assay. Taken together, these results suggest that N-terminal myristoylation and palmitoylation play an additive and partially redundant role in RPS5 function.
A possible explanation for the functional defect of RPS5 acylation mutants is that these modifications compromise interactions of the CC domain with PBS1 and/or with the NBS domain. We tested this possibility using co-immunoprecipitation assays, but we observed no significant reduction in the ability of the RSP5 CC domain containing these substitutions to co-immunoprecipitate either PBS1 or the RPS5 NBS domain (Fig. 3).
Likewise, the G2/3AC4A-RPS5 full-length protein retained the ability to interact with PBS1 (supplemental Fig. S3B). Note, however, that these interactions may be occurring in solution following cell lysis, and do not necessarily indicate that the G2/3AC4A derivatives co-localize with PBS1 inside the cell.

2
The requirement for myristoylation and palmitoylation motifs in RPS5 suggests that RPS5 must localize to a membrane to be functional. We therefore investigated the subcellular localization of RPS5. A C-terminal sYFP fusion of RPS5 was transiently expressed in N. benthamiana and visualized by laser scanning confocal microscopy. The   accumulation is similar between wild-type and mutant constructs. We thus infer that reduction in HR caused by the acylation mutations is due to the reduction in PM localization rather than a reduction in protein levels.
Since the 2-5 (D266E) -5 chimera only triggered HR when co-expressed with PBS1 and AvrPphB, but not by itself, and the RPS2 CC domain contains a predicted palmitoylation residue, but no myristoylation residue, we hypothesized that the 2-5-5 chimera might be only partially associated with the PM, thus reducing signaling activity. To test this, we analyzed the subcellular localization of the 2-5-5:sYFP; however, the 2-5-5:sYFP was exclusively localized to the PM (supplemental Fig. S4B), indicating that the absence of autoactivity in the 2-5 (D266E) -5 background cannot be attributed to dissociation from the PM.  Co-IP assays.

The RPS5 LRR Domain Is Required for Recognition of PBS1 Cleavage
A positive role for the LRR in NBS-LRR protein activation has been reported for the R-  (supplemental Fig. S2D). Unlike the full-length G2/3-RPS5 and G2/3-D266E (Fig. 2), co-expression of G2/3A-CC-NBS or G2/3-CC-NBS (D266E) with PBS1 and AvrPphB failed to induce an HR (Fig. 6A). These observations indicate that the RPS5 LRR domain plays a positive regulatory role in RPS5 activity in addition to its known negative regulatory role in preventing RPS5 CC-NBS autoactivation.
To test whether the positive regulatory role of the RPS5 LRR is sequence specific, we swapped the RPS2 LRR (amino acids 508-919) for the RPS5 LRR (amino acids 513-889) creating a 5-5-2 chimera. Interestingly, this 5-5-2 chimera proved to be autoactive ( Fig. 6B). This observation is consistent with findings from other NBS-LRR proteins (Rairdan and Moffett, 2006;van Ooijen et al., 2008), which suggests that the NBS and LRR domains co-evolve, and that negative regulation requires precise interactions between the NBS and LRR domains.

Four LRR Repeats
To identify the minimum RPS5 LRR region required to inhibit RPS5 autoactivation, five C-terminal truncation mutants of different LRR lengths were created (Fig. 7A).  (Fig. S2E). There was no obvious tissue collapse in the CC-NBS-4LRR injected area and only some mild isolated cell death was observed (Fig. 7C). Constructs with longer LRR fragments displayed similar phenotypes (Fig. 7C). Consistent with the macroscopic observations, electrolyte leakage measurements demonstrated a strong conductivity increase for the CC-NBS-3LRR construct, but only moderate or mild conductivity increase for other RPS5 C-terminal truncation mutants (Fig. 7B). Hence, the first 4 LRRs appear to be the minimum region required to suppress RPS5 autoactivation.

Region
Because the C-terminal truncation constructs with four or more LRR repeats were not autoactive, we were able to test whether they could be activated by cleavage of PBS1.
None of these constructs were capable of inducing HR in the presence of PBS1 and AvrPphB (Fig. 7D), indicating that they are unable to recognize PBS1 cleavage.
To further assess the role of different LRRs in the recognition of PBS1 cleavage, we constructed a second series of mutants by introducing internal deletions of 3 LRRs at different sites (RPS5 : deletion of amino acids 820-889). These mutants did not display an auto-active phenotype (data not shown). Next, we tested whether these mutants were still functional in recognizing cleavage of PBS1 by AvrPphB.
None of these mutants induced HR in response to PBS1 cleavage (Figure 8), although protein accumulated to levels similar to wild-type RPS5 (Fig. S2F). To exclude the possibility that these mutants were actually defective in signaling rather than in PBS1 cleavage recognition, the same mutations were introduced into the auto-active RPS5 mutant D266E, i.e. D266E indicating that these internal LRR deletions compromised recognition of PBS1 cleavage, but not RPS5 signaling ability. Collectively, these results indicate that an intact RPS5 LRR domain is required for the recognition of PBS1 cleavage. Conversely, suppression of auto-activation is dependent on a minimum length of the RPS5 LRR region.

The Role of the RPS5 CC Domain in Downstream Signaling
The role of the CC domain in the function of CC-NBS-LRR proteins is poorly understood and controversial.  , 2008). This is unlikely to be the case for the RPS5 NB subdomain, however, as we demonstrated that the RPS5 mediated HR requires PM localization mediated by the N-terminus ( Fig. 2 and Fig. 4). In addition, over-expression of the RPS5 CC-NB fragment (without the ARC subdomain) failed to cause HR as well, even when fused to GFP to enhance stability (data not shown). We also expressed the full NBS domain of RPS5 fused to the first 20 amino acids of RPS5 to direct it to the plasma membrane, but this was also unable to induce HR (Fig. S1E). Thus, the minimal region of RPS5 required to induce HR in N. benthamiana is the CC plus full NBS domain.
One of the more surprising results presented in our study is that the RPS2 CC domain can partially substitute for the RPS5 CC domain in transient overexpression assays (Fig. 1A). However, this finding is consistent with a phylogenetic analysis of CC-  (Fig. 2 and Fig. 4)

Plasma Membrane Association Is Required for RPS5 Function
Our results demonstrate that RPS5 localizes to the plasma membrane and strongly suggest that this localization is required for RPS5 function. Mutation of N-terminal myristoylation and palmitoylation motifs abrogated both PM localization and RPS5induced HR in transient expression assays, despite high initial levels of protein accumulation (Fig. 4) localization and function. Mutation of either motif alone reduced the initial electrolyte leakage (Fig. 2) and the accumulation of RPS5 on the PM (Fig. 4), while the double mutation completely abolished electrolyte leakage/HR induction (Fig. 2) and PM accumulation (Figure 4). This is consistent with the previously reported dosage effect of RPS5 on HR and resistance in which the speed of HR in response to inoculation with P.
syringae expressing AvrPphB, and the level of resistance, was proportional to protein levels in transgenic Arabidopsis plants expressing RPS5:HA (Holt et al., 2005).
Furthermore, our data indicate that this dosage effect is correlated with RPS5 abundance on the PM. Interestingly, PM localization appears to be essential for stabilization of the RPS5 protein (Fig. 4B). This likely explains why a G2/3A-RPS5 mutant gene driven by the RPS5 promoter in stable transgenic plants fails to complement an rps5 mutation (data not shown), as the amount of G2/3A-RPS5 protein on the PM may not reach the threshold for robust downstream signaling. Although single acylation mutations did not block HR induction by PBS1cleavage in N. benthamiana transient assays (Fig.2), they did block induction of HR by the D266E autoactive form of RPS5 (Fig.2). Significantly, these single mutant forms of RPS5 (i.e. G2/3A-D266E and C4A-D266E) could be activated by PBS1 cleavage. This observation indicates that PBS1 cleavage has a stronger effect on RPS5 activation than the D266E substitution. The latter is believed to reduce the ATP hydrolysis rate of RPS5 (Ade et al., 2007), thus increasing the proportion of RPS5 in the ATP-bound state, which is inferred to be the active state for signaling (Takken et al., 2006). Thus, we speculate that activation of RPS5 by PBS1 cleavage involves more than simple nucleotide exchange (i.e. release of ADP and binding of ATP), with the LRR possibly inducing further conformational changes that stabilize the ATP-bound form of RPS5.
Although our data indicate that RPS5 must be at the PM to initiate signaling, it is a formal possibility that once activated, RPS5 may relocalize. Several NBS-LRR proteins have been shown to partially relocalize to the nucleus following activation, with such relocalization appearing to be required to confer resistance (Burch-Smith et al., 2007;Shen et al., 2007;Wirthmueller et al., 2007). We feel this is unlikely for RPS5, however, as the autoactive derivatives of RPS5 (i.e. D266E and CC-NBS) localized primarily to the PM (Fig. S4D and data not shown). Furthermore, addition of a nuclear localization  (Fig. S5).

A Role for the RPS5 LRR Domain in the Recognition of PBS1 Cleavage
As shown in Figure 7, the LRR domain of RPS5 functions in part to keep RPS5 in the off state. The inhibitory function of LRRs is thought to be mediated by physical interactions between the LRR domain and the NBS domain (specifically the ARC2 subdomain) (Rairdan and Moffett, 2006). Consistent with this model, swapping the RPS5 LRR domain with the RPS2 LRR domain resulted in the induction of HR in the absence of PBS1 cleavage (Fig. 6B). The prevailing model for NBS-LRR protein activation is that switching between the 'off' state and 'on' state is mediated by exchange of ADP for ATP (Takken and Tameling, 2009). Autoinhibition by the LRR domain is thus thought to be a consequence of inhibiting nucleotide exchange. In this study, we found that LRRmediated inhibition requires as little as the first four LRR repeats (out of 13 total) (Fig.   7A). If correct, then the first 4 LRRs of RPS5 seem to constitute the minimal NBSinteracting surface required to prevent nucleotide exchange. The failure of the RPS2 LRR domain to substitute for the RPS5 LRR in terms of autoinhibition implies that there are key differences between the first four LRRs of RPS5 and RPS2 in how they interact with the RPS5 NBS domain. Consistent with this expectation, alignment of the RPS5 and RPS2 LRRs shows that the level of identity between them is relatively low (27%, 50%, 52% and 30% for LRRs number 1, 2, 3 and 4 respectively; Warren et al., 1998) and would be unlikely to form proper associations with the RPS5 NBS.
It is also informative that deletion of the LRR domain has an additive effect when combined with the D266E substitution in terms of autoactivation (Fig. 1C). This makes sense if the former promotes exchange of ADP for ATP and the latter slows ATP hydrolysis. Combined, these mutations should greatly enhance the percentage of RPS5 molecules in the ATP-bound state, and thus strengthen signaling.
Although the LRR domain clearly plays an autoinhibitory role, our data also establish that it performs a separate positive role in activation of RPS5 and a positive role in pathogen recognition. We were able to identify these positive roles by employing a partial loss of function mutation (G2/3A) that reduced RPS5 association with the plasma membrane. This acylation mutation suppressed cell death induced by the LRR truncation mutation, even when combined with the D266E mutation (Fig. 6). In contrast, full-length G2/3A-D266E still induced HR when co-expressed with PBS1 and AvrPphB ( Fig. 2A), indicating that PBS1 cleavage activates RPS5 to a higher level than does deletion of the LRR domain and/or the D266E mutation. Furthermore, G2/3A-CC-NBS or G2/3A-CC-NBS (D266E) did not respond to PBS1 cleavage by AvrPphB (Fig. 6), indicating that the full-strength activation of RPS5 by PBS1 cleavage requires the presence of the RPS5 LRR domain. We thus infer that PBS1 cleavage alters LRR domain conformation in a manner that enhances RPS5 activation by promoting nucleotide exchange and/or stabilizing the ATP bound form of RPS5.
Significantly, the small internal deletions of RPS5 LRR domain completely disabled its activation by AvrPphB-mediated PBS1 cleavage but not the auto-activity of the D266E mutant (Fig. 8). This finding is consistent with our earlier identification of two different missense mutations in the RPS5 LRR domain that blocked recognition of AvrPphB, but did not cause autoactivation (Warren et al., 1998). These results suggest that the PBS1 cleavage recognition surface and the consequent activation of nucleotide exchange require a precise structure that cannot tolerate large deletions and substitutions.
In summary, the data described above enable us to significantly refine the model for how RPS5 activity is regulated, assigning multiple functions to both the CC and LRR domains (Fig. 9). Under this refined model, RPS5 localizes to the plasma membrane (due to N-terminal acylation), where the CC domain associates with PBS1, forming a ready-to-fire preactivation complex. When AvrPphB is injected by P. syringae into Arabidopsis, it self-processes and becomes myristoylated, and is thus targeted to the plasma membrane, where it cleaves PBS1 (Shao et al., 2002;Shao et al., 2003).
Cleavage of PBS1 causes a conformation change that then enables PBS1 to bind to the  NBS-containing proteins;Riedl et al., 2005) leading to the activation of a robust HR.  , 1997). All constructs were sequenced for verification. All primers used to generate the above constructs are listed in Table S1.

Agrobacterium Mediated Transient Expression in N. benthamiana
Agrobacterium GV3101 (pMP90) strains carrying pTA7002 DEX-inducible constructs were grown and prepared for transient expression as previously described (Ade et al., 2007). Agrobacterium cultures were resuspended in water to an OD 600 of 1.0. Immunoprecipitations and Immunoblotting For total protein extraction, 6 infiltrated leaves were ground in lysis buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 0.1% NP-40 and Plant Proteinase Inhibitor Cocktail (Sigma)).
Homogenates were centrifuged twice at 13,000 rpm at 4°C for 10 min and supernatants were transferred to new tubes. For immunoblots, crude extracts were mixed with 4x SDS loading buffer (50 mM Tris-HCl pH 6.8, 2% SDS, 10% glycerol, 1% β -mercaptoethanol, 12.5 mM EDTA, 0.02 % bromophenol blue) at a ratio of 3:1 and boiled for 10 min before resolving by SDS-PAGE. Immunoprecipitations were performed as previously described (Ade et al., 2007) using an anti-c-Myc monoclonal antibody matrix (Clontech). The immunocomplexes were resuspended in 50 μ L of 1x SDS loading buffer and boiled for 10 min. All protein samples were resolved on 4-20% gradient Tris-HEPES-SDS polyacrylamide gels (Thermo Scientific) and transferred to a nitrocellulose membrane for probing with Anti-c-Myc-peroxidase (Roche) or anti-HA-peroxidase (Sigma). The ImmunoStar TM HRP Substrate Kit (Bio-Rad) was used for detecting antibody complexes.

Confocal Laser Scanning Microscopy
Confocal laser scanning microscopy was performed on a Leica SP5 AOBS inverted confocal microscope (Leica Microsystems) equipped with a 63X NA1.2 water objective. sYFP fusions were excited with a 514 nm Argon laser and detected using a 522 to 545 nm band-pass emission filter. mCherry was excited using a 561 nm He-Ne laser and detected using a custom 595 to 620 nm band-pass emission filter. To obtain threedimensional images, a series of Z-stack images were collected and then combined and processed using the three-dimensional image-analysis software IMARIS 7.0 (Bitplane Scientific Software; http://www.bitplane.com).

Supplemental Data
The following materials are available in the online version of this article.         www.plantphysiol.org on September 5, 2017 -Published by