Antagonistic Control of Disease Resistance Protein Stability in the Plant Immune System

See allHide authors and affiliations

Science  05 Aug 2005:
Vol. 309, Issue 5736, pp. 929-932
DOI: 10.1126/science.1109977


Pathogen recognition by the plant immune system is governed by structurally related, polymorphic products of disease resistance (R) genes. RAR1 and/or SGT1b mediate the function of many R proteins. RAR1 controls preactivation R protein accumulation by an unknown mechanism. We demonstrate that Arabidopsis SGT1b has two distinct, genetically separable functions in the plant immune system: SGT1b antagonizes RAR1 to negatively regulate R protein accumulation before infection, and SGT1b has a RAR1-independent function that regulates programmed cell death during infection. The balanced activities of RAR1 and SGT1, in concert with cytosolic HSP90, modulate preactivation R protein accumulation and signaling competence.

Specificity in the Arabidopsis immune system relies on ∼125 polymorphic disease resistance (R) genes, many of which encode NB-LRR proteins containing nucleotide binding sites and leucine-rich repeats. NB-LRR proteins “recognize” pathogen proteins that can contribute to pathogen virulence in the absence of host recognition. When recognized by the plant, these are termed avirulence (Avr) proteins. Pathogens from various kingdoms trigger similar NB-LRR-mediated defense responses. Conserved plant proteins control NB-LRR signaling (1, 2). These include RAR1, SGT1, and cytosolic HSP90, each identified by recessive mutations and/or gene silencing in barley, Arabidopsis, potato, tobacco, and tomato (38).

RAR1 plays a generic role in maintaining preactivation NB-LRR protein levels (911) (see below). However, rar1 mutants suppress the resistance function of only a subset of NB-LRR proteins. A “threshold model” can explain the discrepancy between genetic requirements for RAR1 and its apparent biochemical function (11). Thus, RAR1-“independent” NB-LRR proteins accumulate to relatively high steady-state levels and remain above a threshold required for efficient defense activation even when destabilized in a rar1 background. In contrast, RAR1-“dependent” NB-LRR proteins accumulate to relatively low levels that fall below a critical threshold in rar1 mutants. Consistent with the semidominant nature of many R-mediated responses, the threshold model predicts that NB-LRR proteins are quantitative, response-limiting regulators. Cytosolic HSP90 is an additional determinant of steady-state NB-LRR protein accumulation (12). RAR1 likely collaborates with cytosolic HSP90 as a co-chaperone maintaining signal-competent NB-LRR proteins (1316).

In yeast, SGT1 functions in kinetochore and SCF ubiquitin-ligase assembly (1719). Arabidopsis has two SGT1 paralogs, SGT1a and SGT1b (78% amino acid identity), but only sgt1b mutations suppress NB-LRR function (7, 8, 20). RAR1, SGT1, and HSP90 interact in vivo, and RAR1 and SGT1 each interact with subunits of the COP9 signalosome, a likely proteasome lid complex (5, 14, 20). Further, SGT1 interacts with SCF ubiquitin ligase components, provoking speculation that SGT1 mediates the degradation of negative regulators of plant immune function (20). Concomitant losses of RAR1 and SGT1b additively impair function of the Arabidopsis NB-LRR protein RPP5 (7), suggesting separable activities for these two genes. Accordingly, we define a RAR1-independent SGT1b function in programmed cell death. Unexpectedly, however, our data also demonstrate that SGT1b can negatively regulate NB-LRR protein accumulation, and that this activity is antagonized by both RAR1 and HSP90.

The Arabidopsis NB-LRR proteins RPM1, RPS2, and RPS5 confer resistance to Pseudomonas syringae. Each is impaired in rar1 (10, 20, 21), but unaffected in sgt1a or sgt1b (7, 22) (Fig. 1, A and B). Unexpectedly, RPS5 function, but not RPM1 or RPS2 function, was recovered in rar1 sgt1b (Fig. 1, A and B; RPS2 data not shown). None of the rar1 mutant phenotypes were recovered in rar1 sgt1a. Therefore, SGT1b mediates the loss of RPS5 function in rar1, whereas SGT1a and SGT1b may act redundantly in this process for RPM1 and RPS2 (6).

Fig. 1.

SGT1b antagonizes RAR1 to control RPS5-mediated disease resistance. (A) Pseudomonas syringae pv. tomato (Pto DC3000) carrying empty vector (EV) or expressing avrPphB (to trigger RPS5) or avrRpm1 (to trigger RPM1) was infiltrated into leaves at ∼1 × 105 colony forming units (CFU)/ml. Photos of disease symptoms were taken 5 days postinoculation (dpi). Plant lines, alternative alleles tested, extended protocols, and genotyping are described in (30). (B) Plants (genotypes listed at bottom; mutant loci in red) were hand inoculated (bacterial strains listed above each panel) as in (A) and bacterial growth was assessed 3 dpi. Values are mean CFU/ml ± 2SE. (C) The upper half of each leaf was infiltrated as in (A) with 1 × 108 CFU/ml. At these higher inoculum levels [compare to (A)], HR is readily observed as tissue collapse before the onset of disease symptoms. For photographic purposes, we used trypan blue, which gives dark staining in regions of the leaf undergoing cell death (representative trypan leaves shown). Numbers of leaves scored as positive for HR out of the total examined for each genotype are listed below the trypan blue–stained leaves. (D) Plants were inoculated as in (A). In addition to Col-0 rar1-21 [rar1 allele used in (A)], we tested additional Arabidopsis ecotypes and rar1 alleles. As controls for mutant lines with reduced basal resistance, we inoculated enhanced disease susceptibility (eds1) mutants.

NB-LRR activation often triggers a rapid localized programmed cell death, called the hypersensitive response (HR) (23). The HR likely limits the growth of biotrophic fungi and oomycetes (4, 21, 24, 25), although its role in resistance to bacterial pathogens is unclear. RAR1 is required for RPS5-, RPM1-, and RPS2-mediated HR (10). Of these, only the RPS5-mediated HR additionally required SGT1b (Fig. 1C; fig. S1A). Neither RPS5-, RPM1-, nor RPS2-dependent HR were restored in rar1 sgt1b. Using the oomycete parasite Peronospora parasitica, we extended these findings to two additional NB-LRR functions (RPP4 and RPP31; fig. S1, B to E). Thus, SGT1b can control the HR in a RAR1-independent manner. Further, NB-LRR–mediated disease resistance and HR are genetically separable.

Notably, rar1 mutations in different genetic backgrounds allowed enhanced growth of the virulent bacterial strain P. syringae (Pto) DC3000 (Fig. 1, B and D). These data demonstrate a role for RAR1 in basal resistance, an ostensibly R-independent response that limits pathogen spread in susceptible plants (1). This rar1 phenotype is also suppressed in rar1 sgt1b, but not rar1 sgt1a (Fig. 1D). Therefore, SGT1b also antagonizes RAR1 in the control of basal resistance. Given that the only known function for RAR1 is to promote NB-LRR protein accumulation, then NB-LRR proteins also are very likely to function in basal resistance.

Requirements for RAR1 and SGT1b have been defined for NB-LRR genes that confer resistance to different isolates of the oomycete parasite Peronospora parasitica (Pp) (table S1). RPP8 was weakly impaired by rar1, as indicated by low levels of asexual parasite sporulation (Fig. 2, A and B). We bred isogenic plants hemizygous for an RPP8 transgene (RPP8/-) in each mutant background to determine whether the small phenotypic effect of rar1 might depend on RPP8 dosage. RPP8/rar1 plants exhibited increased susceptibility as compared to homozygous controls, supporting the threshold model (11). RPP8/rar1 sgt1b plants were completely resistant, indicating that SGT1b mediates susceptibility in RPP8/rar1. As with RPP4, RPP31, and RPS5, these data are inconsistent with the hypothesis that RAR1 and SGT1 act additively in all NB-LRR–mediated disease resistance responses.

Fig. 2.

SGT1b antagonism of RAR1 is generalizable to several NB-LRR resistance specificities. (A) Seven- to 10-day-old cotyledons of rpp8 plants expressing a stable RPP8 transgene were inoculated with the asexual spores of Peronospora parasitica (Pp) isolate Emco5 (40). Representative, trypan blue–stained leaves are shown to illustrate cell death and Pp structures (hyphae, asexual sporangiophores). (B) Asexual sporangiophores were quantified 7 dpi on at least 50 cotyledons for each of the indicated genetic backgrounds. The numbers below each tested genotype (key genotypes shown in red) represent mean sporangiophores/cotyledon (± 2 SE).

To further investigate the recovery of RPS5-mediated disease resistance in rar1 sgt1b, we constructed isogenic lines expressing hemag-glutinin (HA) epitope-tagged RPS5 driven by the native promoter in the La-er ecotype (an rps5 null) (26). RPS5:HA accumulated exclusively in the microsomal fraction of wild-type, rar1, and sgt1b, and its accumulation was greatly diminished in rar1 (Fig. 3A). These results are similar to previous observations for RPM1 and RPS2 (9, 10, 27, 28).

Fig. 3.

RAR1 and SGT1 act antagonistically to control RPS5 protein accumulation. (A) Tissue samples for protein blot analysis were taken from independent, F1 plants transformed with an HA epitope–tagged RPS5 transgene [RPS5:HA (30)]. Protein was separated into total (T), soluble (S), and membrane (M) fractions (28). Ascorbate peroxidase and RIN4 antibodies were used as controls for the cytoplasmic and membrane fractions, respectively (41, 42). Equal loading for all protein samples in Fig. 3 was ensured by protein quantification before loading and Ponceau Red staining of nitrocellulose membranes after transfer. (B) Total protein extracts were isolated from 10 independent, F1 Col-0 rps5 mutants transformed with the RPS5:HA transgene. Before protein blot analysis, four leaves per plant were visually scored for HR (as in Fig. 1C) at 12 and 20 hours (%HR@12 or 20 hrs). Mean relative RPS5:HA protein accumulation (MRA) levels were quantified using ImageJ (version 1.31) (43). All values were transformed such that the weakest RPS5:HA-expressing plants (first three lanes on blot) were equalized to MRA = 1.0. (C) La-er (rps5) ecotype plants and the rar1, sgt1b, and rar1 sgt1b mutants [also in La-er (30)] were transformed with the RPS5:HA transgene. Individual, F1 transformants were selected in each genetic background, and RPS5:HA protein accumulation was visualized by protein blot. MRA values were transformed such that pooled values from the wild-type La-er ecotype was set to 1.0. Symbols above rar1 and rar1 sgt1b lanes are explained in (E). (D) A stable RPM1:Myc transgenic line (28) was crossed to the rar1 sgt1b mutant. Indicated genotypes were selected by polymerase chain reaction from the F2 population and examined by protein blot analysis as in (C). The lane designated with an asterisk (*) represents the parental RPM1:Myc line. (E) A La-er RPS5:HA/-rar1/rar1 transformant [male; (♦) in (C)] was crossed to either La-er rar1/rar1 or La-er RAR1/RAR1 (females in each cross). Similarly, a La-er RPS5:HA/-rar1/rar1 sgt1b/sgt1b transformant [male; (⚫) in (C)] was crossed to either La-er rar1/rar1 sgt1b/sgt1b or La-er rar1/rar1 SGT1b/SGT1b (females). The resulting genotypes are shown above each lane. The first lane of each pair recapitulates the original parental genotype, and the second represents altered gene dosages of either RAR1 or SGT1b (red text). The secondary antibody reacting band further demonstrates equal loading. Relative accumulation (RA) levels were transformed such that the parental lane in each comparison equals 1.0. (F) Stable, nonsegregating rps5 rar1 sgt1b triple-mutant plants were isolated and tested for disease resistance as in Fig. 1B. (G) Leaves were infiltrated with either dimethyl sulfoxide (DMSO) alone or 10 μM geldanamycin (GDA; A.G. Scientific, San Diego, CA) dissolved in DMSO (30). Samples were collected for protein blot analysis 24 hours after inoculation (similar results were seen at 18 hours). GDA did not alter RPS2:HA accumulation (data not shown) (30).

Unregulated NB-LRR expression can be lethal, suggesting that R protein accumulation must be fine tuned to provide rapid responses to infection while minimizing aberrant signaling. Dose dependence of MLA1 (11) and RPP8 (Fig. 2) suggested that NB-LRR–mediated responses should be proportional to their steady-state protein accumulation levels. To test this hypothesis, we used the inherent variability of RPS5:HA accumulation in 10 independent transgenic lines. After Pto DC3000 inoculation, random RPS5:HA rps5 transgenic plants were ordered according to HR timing, from no HR to rapid HR. Protein samples from this phenotypically ordered set of plants demonstrated that increasing RPS5:HA protein levels correlated with faster HR (Fig. 3B). Thus, the levels of RPS5, and presumably other NB-LRR proteins, can be rate limiting for response rapidity. These data further support the RAR1-mediated threshold model for NB-LRR function (11).

We quantified RPS5:HA accumulation in individual, first-generation transgenic plants of each relevant genotype (Fig. 3C). RPS5:HA accumulated to readily detectable, equivalent mean levels in La-er wild type and sgt1b, but to only 13% of wild-type levels in rar1. RPS5:HA accumulation was restored to ∼60% of wild-type levels in rar1 sgt1b. By contrast, and as expected from the lack of RPM1 functional recovery (Fig. 1B), RPM1:Myc did not reaccumulate in rar1 sgt1b (Fig. 3D).

We created genetic controls to confirm the antagonistic roles of RAR1 and SGT1b in RPS5 accumulation. A rar1/rar1 transgenic parental line expressing low, but measurable RPS5:HA was used to generate RPS5:HA RAR1/rar1 and sibling control F1 plants (Fig. 3E, first two columns). RPS5:HA accumulation was restored more than sevenfold in the RAR1/rar1 heterozygote. Similarly, a rar1 sgt1b transgenic parent that accumulated high levels of RPS5:HA was used to generate RPS5:HA rar1/rar1 SGT1b/sgt1b and sibling control F1 plants (Fig. 3E, third and fourth columns). The presence of a single copy of wild-type SGT1b resulted in 2.5 fold less RPS5:HA than in sibling controls. Importantly, disease resistance observed in RPS5 rar1 sgt1b (Fig. 1, A and B) was lost in an rps5 rar1 sgt1b triple mutant (Fig. 3F), demonstrating a direct link between restoration of RPS5 function and RPS5 protein levels. Collectively, these data demonstrate that RAR1 is a positive regulator, and SGT1b a negative regulator, of RPS5 accumulation. We envision that the recovery we observed for other NB-LRR functions in rar1 sgt1b (Fig. 2 and fig. S1, B to E) follows the same mechanism.

Reduction of cytosolic HSP90 function negatively affects steady-state accumulation of NB-LRR proteins (12, 14). We used the HSP90-specific inhibitor geldanamycin (GDA) (29) to examine RPS5:HA and RPM1:Myc protein accumulation in wild-type and sgt1b plants. GDA infiltration into wild-type leaves typically resulted in reduced RPS5:HA and RPM1:Myc protein accumulation, but did not eliminate disease resistance function (Fig. 3G) (30). GDA did not affect accumulation of either NB-LRR protein in sgt1b. Thus, elimination of RAR1 or inhibition of HSP90 activity is sufficient to lower NB-LRR protein accumulation through an unknown mechanism. In both cases, SGT1b can mediate this outcome. Notably, RPM1:Myc destabilization mediated by GDA is SGT1b dependent, whereas its destabilization in rar1 is not. This contrasts with RPS5:HA, suggesting that antagonism between RAR1-HSP90 and SGT1b is fine tuned for different NB-LRR proteins.

Our findings challenge suggestions of signaling functions for RAR1 and SGT1b in NB-LRR–mediated disease resistance. Restoration of RPS5-, RPP4-, RPP8-, and RPP31-mediated functions in rar1 sgt1b prove that RAR1 and SGT1b are not required for disease resistance signaling per se. Additionally, we show that SGT1b has a RAR1-independent function as a positive regulator of RPP4-, RPP31-, and RPS5-mediated HR. A general role for SGT1b in HR is now well established (6, 31), and we speculate that an efficient HR requires SGT1b-dependent elimination of an unidentified negative regulator. This SGT1b function would be particularly relevant in cases where HR plays a key role in limiting pathogen spread, explaining why some NB-LRR proteins exhibit additive requirements for RAR1 and SGT1b. In such cases, the lack of NB-LRR accumulation in rar1 sgt1b coupled to an inefficient HR would result in enhanced pathogen growth.

RAR1 and HSP90 are positive regulators of NB-LRR protein steady-state accumulation [(912, 14) and this work]. As such, RAR1 and HSP90 may determine whether NB-LRR proteins are functional in disease resistance or marked for degradation. Cytosolic HSP90 transiently binds nonnative “client” proteins to assist in proper folding (32, 33). Active folding of HSP90 client proteins is regulated by cycles of adenosine 5′-triphosphate (ATP) binding and hydrolysis that are, in turn, modulated by co-chaperones. In addition to modulating ATP hydrolysis, co-chaperones also guide HSP90 client specificity. Therefore, HSP90 apparently processes and/or maintains NB-LRR proteins to a signal-competent conformational state, with RAR1 acting as a co-chaperone.

Yeast SGT1 transiently links HSP90 to the inner kinetochore complex (CBF3), balancing CBF3 assembly and turnover (34). Specific mutations that “trap” SGT1 in CBF3 complexes result in reduced CBF3 accumulation. This is consistent with our finding that elimination of SGT1b can reduce NB-LRR turnover. We speculate that RAR1 defines a regulatory checkpoint protecting HSP90-associated NB-LRR proteins from SGT1b-mediated degradation. In rar1 mutants, this degradation pathway becomes the default, perhaps through direct interaction of HSP90-associated NB-LRR proteins with an SCF-bound SGT1 (11, 35, 36).

Coupling of folding and degradation fates has previously been demonstrated for the HSP90 clients glucocorticoid hormone receptor (GR) and cystic fibrosis transmembrane conductance regulator (CFTR) (37, 38). GR or CFTR, in complex with HSP70/HSP90, are degraded when these complexes associate with CHIP (carboxy-terminus of HSP70 interacting protein), a member of the U-box family of ubiquitin ligases. Mutations in CHIP that eliminate ubiquitin ligase function dominantly interfere with ubiquitination and subsequent GR/CFTR degradation. Like SGT1, CHIP has several tetratricopeptide repeats (TPRs) that are required for HSP70/HSP90 association (15, 19, 37). Therefore, like CHIP, SGT1-SCF complexes might couple NB-LRR proteins to the cellular degradation machinery (39). It remains unclear whether changes in NB-LRR accumulation are due to proteasome-dependent degradation or an alternative protein turnover mechanism such as endocytosis. Nevertheless, we anticipate that our genetic results will inform subsequent biochemical experiments.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Table S1

References and Notes

References and Notes

View Abstract

Navigate This Article