Plant defense responses are regulated by temperature. In Arabidopsis, the chilling-sensitive mutant chs2-1 (rpp4-1d) contains a gain-of-function mutation in the TIR-NB-LRR (Toll and interleukin 1 receptor-nucleotide binding-leucine-rich repeat) gene, RPP4 (RECOGNITION OF PERONOSPORA PARASITICA 4), which leads to constitutive activation of the defense response at low temperatures.
Here, we identified and characterized two suppressors of rpp4-1d from a genetic screen, hsp90.2 and hsp90.3, which carry point mutations in the cytosolic heat shock proteins HSP90.2 and HSP90.3, respectively.
The hsp90 mutants suppressed the chilling sensitivity of rpp4-1d, including seedling lethality, activation of the defense responses and cell death under chilling stress. The hsp90 mutants exhibited compromised RPM1 (RESISTANCE TO PSEUDOMONAS MACULICOLA 1)-, RPS4 (RESISTANCE TO P. SYRINGAE 4)- and RPP4-mediated pathogen resistance. The wild-type RPP4 and the mutated form rpp4 could interact with HSP90 to form a protein complex. Furthermore, RPP4 and rpp4 proteins accumulated in the cytoplasm and nucleus at normal temperatures, whereas the nuclear accumulation of the mutated rpp4 was decreased at low temperatures. Genetic analysis of the intragenic suppressors of rpp4-1d revealed the important functions of the NB-ARC and LRR domains of RPP4 in temperature-dependent defense signaling. In addition, the rpp4-1d-induced chilling sensitivity was largely independent of the WRKY70 or MOS (modifier of snc1) genes. [Correction added after online publication 11 March 2013: the expansions of TIR-NB-LRR and RPS4 were amended]
This study reveals that Arabidopsis HSP90 regulates RPP4-mediated temperature-dependent cell death and defense responses.
Temperature is one of the most important factors affecting plant growth and geographical distribution and the plant defense response to pathogens (Guy, 1990; Alcazar & Parker, 2011; Hua, 2013). High temperatures (32°C) inhibit an increase in salicylic acid (SA) levels and prevent the induction of the pathogenesis-related (PR) genes and resistance of tobacco plants when infected with Tobacco mosaic virus (TMV). The application of exogenous SA results in the recovery of PR gene expression, but it cannot overcome the inhibition of the hypersensitive response (HR) (Malamy et al., 1992). Consistently, low temperatures can induce PR gene expression. Cold stress has been shown to activate the transcription factor NAC, resulting in its translocation from the plasma membrane to the nucleus, which, in turn, induces PR genes by direct binding to their promoters independent of SA (Seo et al., 2010). Another known factor to mediate both the cold and biotic stress responses is DEAR1 (DREB and EAR motif protein 1). DEAR1 is induced by both pathogen infection and cold treatment. The overexpression of DEAR1 in Arabidopsis results in dwarf and cell death phenotypes, as well as the constitutive expression of PR genes and SA accumulation. However, the induction of CBF (C-repeat binding factor) genes by cold treatment is suppressed in transgenic plants overexpressing the DEAR1 gene, leading to a decreased freezing tolerance (Tsutsui et al., 2009). It has been reported that the mekk1 mutant grown at 22°C suffers from severe dwarfism and has a constitutively active defense response, and this phenotype was greatly suppressed when the plants were grown at 28°C (Ichimura et al., 2006). In soybean plants, different temperature regimens during seed development can affect the expression of defense-related genes, resulting in varying resistance to pathogens (Upchurch & Ramirez, 2011).
The plant Resistance (R) genes encode immune receptors that recognize, directly or indirectly, pathogen effectors (Jones & Dangl, 2006). R proteins belonging to the largest class contain nucleotide-binding (NB) and leucine-rich repeat (LRR) domains. Emerging evidence has shown that temperature can modulate the plant defense responses through the R proteins. The null mutant bon1-1 was found to display a miniature phenotype when grown at 22°C, but showed a wild-type appearance when grown at 28°C (Hua et al., 2001). Further study indicated that the NB-LRR gene SNC1 (Suppressor of npr1-1, constitutive 1) was activated in the bon1-1 mutant, leading to temperature-dependent constitutive defense responses and reduced cell growth at 22°C, but not at 28°C (Yang & Hua, 2004). It is intriguing to find that other point mutations in SNC1 retain disease resistance at 28°C. Similar mutations introduced into the tobacco R gene N conferred defense responses at an elevated temperature (Y. Zhu et al., 2010). Our previous studies have shown that mutations in the NB-LRR type of R genes, including RPP4/CHS2 (RECOGNITION OF PERONOSPORA PARASITICA 4/CHILLING-SENSITIVE 2) and CHS3, result in the activation of the defense responses at low temperatures (Huang et al., 2010; Yang et al., 2010). Moreover, hybrid necrosis and genetic incompatibility were also found to be modulated by the NB-LRR genes in a temperature-dependent manner (Bomblies et al., 2007; Alcazar et al., 2009). In addition to these R proteins, our recent study has reported that a missense mutation of a TIR-NB (Toll and interleukin 1 receptor-nucleotide binding) protein, CHS1, shows chilling-induced defense responses (Wang et al., 2013). Thus, temperature is an important factor in regulating plant growth homeostasis by controlling the activation of some R or R-like proteins.
Several key components have been shown to be genetically required for R gene-dependent defense activation in Arabidopsis. The different requirements for the two important components, EDS1 (ENHANCED DISEASE SUSCEPTIBILITY 1) and NDR1 (NON-RACE-SPECIFIC DISEASE RESISTANCE 1) (Century et al., 1995; Parker et al., 1996), in pathogen defense define two major R gene-mediated signaling pathways in Arabidopsis (Aarts et al., 1998; McDowell et al., 2000; Bittner-Eddy & Beynon, 2001; Li et al., 2001; Borhan et al., 2004). Further studies have identified additional components that function in SA-dependent signaling during disease resistance in Arabidopsis. PAD4 (PHYTOALEXIN DEFICIENT 4) and SAG101 (SENESCENCE-ASSOCIATED GENE101) interact individually with EDS1 to form several spatially distinct complexes to activate SA signaling (Feys et al., 2001; Wagner et al., 2011; Zhu et al., 2011). The protein complex SGT1-RAR1-HSP90 (SALICYLIC ACID GLUCOSYLTRANSFERASE 1-REQUIRED FOR Mla12 RESISTANCE 1-HEAT SHOCK PROTEIN 90), acting as a chaperone, is required for the defense mediated by several R genes, including RPM1 (RESISTANCE TO PSEUDOMONAS MACULICOLA 1), RPS2 (RESISTANCE TO P. SYRINGAE 2) and RPS4 in Arabidopsis (Hubert et al., 2003; Takahashi et al., 2003; Zhang et al., 2004), N in Nicotiana benthamiana (Liu et al., 2004) and Rx in potato (Boter et al., 2007). [Correction added after online publication 11 March 2013: the expansions of NDR1, SGT1-RAR1-HSP90 and RPS2 were amended]
Our previous study showed that a gain-of-function mutant of a TIR-NB-LRR-type R protein RPP4, chs2-1/rpp4-1d, leads to cell death during chilling stress. The rpp4-1d -conferred chilling sensitivity required EDS1, RAR1 and SGT1b, but did not require PAD4 or SA. In this study, we report a genetic screen for rpp4-1d suppressors. We show that mutants which carry point mutations in two isoforms of the cytosolic heat shock protein HSP90, namely HSP90.2 and HSP90.3, suppress the rpp4-1d-associated phenotypes at chilling temperatures, and HSP90 interacts with RPP4/rpp4. Therefore, HSP90 acts as an important chaperone to modulate RPP4-mediated temperature-dependent cell death and defense responses.
Materials and Methods
Plant material and growth conditions
Arabidopsis thaliana (L.) Heynh. plants of the Columbia (Col-0) and Landsberg erecta (Ler) accessions were used in this study. The mutants used in this study were hsp90.3-3, mos1-5 (Li et al., 2010), mos2-2 (Zhang et al., 2005), mos3-2 (Zhang & Li, 2005), mos4-2 (Palma et al., 2007), mos6-4 (Palma et al., 2005), mos7-1 (Cheng et al., 2009) and wrky70 (Knoth et al., 2007). Plants were grown under a 16-h light (100 μmol m−2 s−1) : 8-h dark regime at 22 or 16°C in the soil or on an agar plate containing Murashige and Skoog medium (Sigma).
Genetic mapping and cloning of the SUCH1 (SUPPRESSOR OF CHS2-1) genes
The suppressors of chs2-1/rpp4-1d were screened and mapped as described previously (Huang et al., 2010). Approximately 500 homozygous or heterozygous rpp4-1d plants with wild-type morphology at 4°C were used for mapping. The candidate genes were PCR amplified and the mutations were identified from the mutant plants.
Plasmid construction and plant transformation
A 4.4-kb genomic fragment containing the HSP90.2 gene was amplified by PCR using the HSP90.2-p1F and HSP90.2-p1R primers (Supporting Information Table S1), and cloned into the pCAMBIA1300 vector to generate pHSP90.2:HSP90.2. A 3.2-kb genomic fragment containing the HSP90.3 gene was amplified using the HSP90.3-p1F and HSP90.3-p1R primers (Table S1), and cloned into pCAMBIA1300 to generate pHSP90.3:HSP90.3.
The HSP90.3 and hsp90.3 cDNAs were amplified by PCR from wild-type and hsp90.3-1 mutant plants using the primers HSP90.3-p2F and HSP90.3-p2R (Table S1). Both fragments were then cloned into pSuper1300 (Yang et al., 2010) with a Super promoter (Ni et al., 1995) to generate pSuper:HSP90.3 and pSuper:hsp90.3.
The Agrobacterium tumefaciens-mediated transformation was performed according to the floral dip method (Clough & Bent, 1998).
To generate double mutants, rpp4-1d was crossed with the wrky70, mos1-5, mos2-2, mos3-2, mos4-2, mos6-4 and mos7-1 mutant lines. The F2 progeny were specifically genotyped, and the homozygosity of the rpp4-1d mutation was identified as described by Huang et al. (2010).
Ion leakage, proline (Pro) content and SA measurement
The electrolyte leakage assay was performed as described by Lee et al. (2002). Pro content was measured as described by Bates et al. (1972). Free SA and total SA were extracted from 3-wk-old plants grown at different temperatures and measured as described previously (Huang et al., 2010).
Trypan blue staining and 3,3′-diaminobenzidine (DAB) staining were performed as described previously (Bowling et al., 1997; Thordal-Christensen et al., 1997). For HSP90 inhibitor treatment, the true leaves were injected with 10 μM geldanamycin (GDA) (Sigma) and the same volume of DMSO was used as a negative control. The phenotype was observed at day 7 after treatment.
The wild-type and mutated forms of RPP4 were cloned into pSuper1300 containing the Myc tag. The wild-type and mutated forms of HSP90.3 were cloned into pSuper1300 containing the green fluorescent protein (GFP) tag. The purified plasmids were transformed into Arabidopsis mesophyll protoplasts following a previous protocol (Zhai et al., 2009). After overnight incubation, the protoplasts were lysed and the RPP4/rpp4-Myc proteins were immunoprecipitated by anti-Myc agarose conjugate (Sigma). The co-immunoprecipitation products were detected by immunoblot analysis with anti-HSP90 antibody (Abmart, Shanghai, China).
HSP90.3/hsp90.3 was fused with N-Luc in the pCAMBIA-nLUC vector, and the TIR, NB-ARC and LRR domains of the RPP4 protein were fused with C-Luc in the pCAMBIA-cLUC vector. The vectors were transformed into N. benthamiana leaves using an Agrobacterium-mediated method. After 72 h of infiltration, Luc activity was observed as described previously (Chen et al., 2008).
Protein fractionation and immunoblot assays
The protoplasts transformed with rpp4-Myc were treated at 4 or 22°C for the indicated time. Soluble and nuclear proteins from protoplasts were isolated using a Plant Nuclei Isolation/Extraction Kit (Sigma). HSP90 protein was used as a cytosolic marker, and histone H3 was used as a nuclear marker. Nuclear protein extracts (N) were 9× concentrated compared with soluble fractions (S). The rpp4-Myc fusion protein was detected using an anti-Myc antibody (Sigma). HSP90 protein was detected using an anti-HSP90 antibody. The accumulation of nuclear RPP4 protein was analyzed based on the N/S ratio of RPP4. The experiments were repeated three times, and a representative figure is shown.
Bacterial growth assay
A pathogen resistance test on Pseudomonas syringae pv tomato (Pst) strain DC3000 was performed as described previously (Kim & Delaney, 2002). The pathogen concentrations were OD600 = 0.02 for virulent strain Pst DC3000 and OD600 = 0.2 for avirulent strains Pst DC3000 with avrRpm1, avrRpt2 and avrRps4. Four-week-old plants were dipped into the bacterial suspension containing 10 mM MgCl2 and 0.025% Silwet L-77. The leaves were collected 2 h after inoculation and at 3 d post-inoculation. For each time point, three replicate samples were collected to determine the susceptibility of Pst DC3000.
A pathogen resistance test on Peronospora parasitica Emwa1 was performed as described by Yang & Hua (2004). Twelve-day-old seedlings were sprayed at a concentration of 105 spores ml−1 of water. The oomycete spores on the leaf surface were counted 10 d after inoculation.
Identification of the such1 mutants
Our previous study showed that a gain-of-function mutant of RPP4, chs2-1/rpp4-1d, confers sensitivity to low temperature (Huang et al., 2010). To identify new components involved in the RPP4-mediated pathway, the such (suppressor of chs2-1) mutants were screened from an ethane methyl sulfonate (EMS)-mutagenized pool of rpp4-1d plants, as described previously (Huang et al., 2010). The rpp4-1d plants have a phenotype of yellow, wilted leaves at chilling temperatures (Huang et al., 2010), and the suppressor screening was performed to isolate mutants that reverted to the wild-type morphology under chilling conditions. We identified two allelic mutants of such1, such1-1 and such1-2, and one such2 mutant. The such1 and such2 mutants reverted to the sensitive phenotype of the rpp4-1d mutant at both 4 and 16°C (Fig. 1a). Ion leakage is an indicator of the integrity of the plasma membrane. The increased ion leakage observed in rpp4-1d plants at 4°C was significantly suppressed by the such1 and such2 mutations (Fig. 1b). Accumulation of Pro, an important osmolyte, was dramatically reduced in the rpp4-1d mutant under cold stress (Huang et al., 2010). To determine whether the such1 or such2 mutation affected Pro accumulation in rpp4-1d, Pro was extracted from the rpp4-1d such1 and rpp4-1d such2 double mutants and measured. The Pro levels in the rpp4-1d such2 and rpp4-1d such2 double mutants were significantly increased compared with those in rpp4-1d (Fig. 1c).
The cell death-related phenotypes are suppressed by such1 and such2
A previous study showed that low temperatures induced extensive cell death in the rpp4-1d mutant (Huang et al., 2010). However, cell death was dramatically suppressed in the rpp4-1d such1 and rpp4-1d such2 double mutants following 4°C treatment (Fig. 2a). Moreover, the accumulation of hydrogen peroxide (H2O2), as revealed by DAB staining, was obviously inhibited in the rpp4-1d such1 and rpp4-1d such2 double mutants at 4°C (Fig. 2a).
In the rpp4-1d plants, the RPP4 and PR genes were up-regulated after cold treatment (Huang et al., 2010). However, expression of RPP4, PR1 and PR2 was suppressed in the rpp4-1d such1 and rpp4-1d such2 double mutants relative to rpp4-1d (Fig. 2b,c). The SA levels were elevated in the rpp4-1d mutant following cold treatment (Huang et al., 2010). As shown in Fig. 2(d), the levels of free and total SA in the double mutants were drastically reduced compared with those in rpp4-1d. Therefore, such1 and such2 largely suppressed all of the known rpp4-conferred phenotypes under chilling stress.
Map-based cloning of SUCH1 and SUCH2
To examine the nature of the such1 mutation, rpp4-1d such1 was backcrossed to rpp4-1d, and the F1 progeny exhibited the rpp4-conferred chilling sensitivity. Of 200 F2 plants, 156 exhibited rpp4-like morphology (expected ratio is three of four). These results indicate that such1 is a single, recessive nuclear mutation.
A map-based cloning approach was used to identify the mutation in such1 that suppressed the rpp4-1d phenotypes. The rpp4-1d such1-1 double mutant in the Col ecotype was crossed with Ler. At 4°C, wild-type-looking plants that were homozygous or heterozygous at the rpp4-1d locus in the F2 progeny were used for rough mapping. The such1-1 locus was found to have linkage with markers on the bottom arm of chromosome 5. Further fine mapping using 201 F2 mutant plants that harbored the rpp4-1d mutation narrowed the location of the such1-1 mutation to an 843-kb region between the MBG8 and MHM17 markers on chromosome 5 (Fig. 3a). Sequencing analysis of this region revealed a C299T transition in At5g56010 (HSP90.3), which resulted in the S100F substitution (Fig. 3b). In such1-2, a G to A transition at nucleotide 370 resulted in the G124S mutation in the same gene. Although genetic analysis showed that such1 and such2 were at the same locus (see later for details), there was no mutation in the At5g56010 gene in the rpp4-1d such2-1 double mutants. In Arabidopsis, there are five genes encoding HSP90, and four of these, including HSP90.3, are arranged in tandem. Therefore, we sequenced the other three HSP90 genes in the rpp4-1d such2-1 double mutant and found a point mutation in the third exon of At5g56030 that resulted in the mutation of the 100th amino acid (S to F), which was previously named hsp90.2-2 (Hubert et al., 2003).
To examine whether the mutations in the hsp90 genes were responsible for the suppression of the rpp4-1d phenotypes, the HSP90 inhibitor GDA was used. Application of GDA dramatically inhibited the chilling sensitivity, cell death and H2O2 accumulation of rpp4-1d under chilling stress (Fig. 3c). Next, the wild-type fragments of At5g56010/HSP90.3 and At5956030/HSP90.2, under the control of their own promoters, were transformed into rpp4-1d such1 and rpp4-1d such2-1, respectively. All transformed plants restored the rpp4-conferred phenotypes under chilling stress, indicating that SUCH1 is HSP90.3, and SUCH2 is HSP90.2 (Fig. 3d). Furthermore, the rpp4-1d such2-1 phenotypes were also suppressed by HSP90.3 (Fig. 3d), suggesting the redundant function of HSP90.2 and HSP90.3. For conciseness and consistency, the mutant such1-1 is hereafter named hsp90.3-1, such1-2 is hsp90.3-2 and such2-1 is hsp90.2-2.
We also obtained two T-DNA insertion lines, designated hsp90.3-3 and hsp90.2-10, from ABRC. However, the hsp90.3-3 and hsp90.2-10 loss-of-function mutants could not rescue the phenotype of rpp4-1d under chilling stress (Fig. 4a). In addition, we found that the F1 progeny of rpp4-1d hsp90.3 and rpp4-1d hsp90.2-2 showed the same phenotypes as both of its parents at 4°C (Fig. 4b). These results suggest that the functional redundancy of the HSP90 isoforms and the mutations of hsp90.3-1, hsp90.3-2 and hsp90.2-2 have interfering effects, which is consistent with a hypothesis previously proposed that there are redundant functions among the HSP90 members (Hubert et al., 2003). In addition, overexpression of the mutated form hsp90.3-1 could also rescue the phenotype of rpp4-1d to the wild-type at 4°C, but the wild-type HSP90.3 could not (Fig. 4c). The expression of PR1 in the rpp4-1d mutant was also inhibited by the overexpression of the mutated hsp90.3-1 (Fig. 4d). These results indicate that the mutated hsp90.3 had dose-dependent effects on the rpp4-conferrred phenotypes.
To generate the hsp90.3 single mutant, we crossed rpp4-1d hsp90.3-1 to the wild-type Col plants. Plants homozygous for hsp90.3-1 and wild-type for RPP4 were selected as the hsp90.3-1 single mutant. The hsp90.3-2 and hsp90.2-2 single mutants were isolated using the same approach. These hsp90 mutants showed pleiotropic morphological defects, including low germination rate, short petiole and primary roots, serrated leaves and early senescence (Table S2).
The hsp90.3 mutants exhibit compromised RPM1-, RPS4- and RPP4-mediated resistance
To test whether HSP90.3 contributes to basal defense against virulent pathogens, the hsp90.3-1, hsp90.3-2 and hsp90.3-3 plants were challenged with Pst strain DC3000 at a low concentration (OD600 = 0.02), at which wild-type plants usually develop weak disease symptoms. No significant difference was observed between the hsp90.3 mutants and wild-type plants (Fig. 5a). Therefore, HSP90.3 may not contribute to basal resistance.
To determine whether HSP90.3 is involved in resistance mediated by R proteins, including RPM1, RPS2 and RPS4, we challenged the hsp90.3 mutants with Pst DC3000 expressing avrRpm1, avrRpt2 and avrRps4, respectively, at a concentration of OD600 = 0.2. The hsp90.3-1 and hsp90.3-2 mutants obviously compromised resistance mediated by RPM1, whereas the T-DNA insertion mutant hsp90.3-3 behaved in a similar manner to the wild-type plants (Fig. 5b). Furthermore, the hsp90.3-1 and hsp90.3-2 mutants exhibited slightly reduced RPS4-dependent resistance (Fig. 5c). By contrast, no significant difference was detected between the hsp90.3 mutants and the wild-type plants in RPS2-mediated resistance (Fig. 5d). These results suggest that HSP90.3 is not required for the basal defense response, but is essential for the full RPM1- and RPS4-mediated defense responses.
To determine whether these hsp90 alleles compromise RPP4-mediated oomycete resistance, we compared the disease resistance phenotypes of Col, rpp4-r26 (an intragenic revertant of rpp4-1d), hsp90.3-1, hsp90.3-2 and hsp90.3-3 against Hyaloperonospora arabidopsidis (H. a.) Emwa1. The hsp90.3-1 and hsp90.3-2 mutants were more susceptible to H. a. Emwa1 than wild-type Col and the T-DNA insertion line hsp90.3-3 (Fig. 5e). As the controls, Col was completely resistant to H. a. Emwa1, whereas eds1-2 (Col) (Bartsch et al., 2006) and rpp4-r26, which contains a mutation in RPP4 resulting in the introduction of stop codons in front of the LRR domain (for details, see Fig. 8d), were susceptible to H. a. Emwa1 (Fig. 5e). These results suggest that HSP90.3 is required for full RPP4-mediated oomycete resistance.
HSP90 interacts with RPP4/rpp4 protein in vivo
Previous studies have demonstrated that the molecular chaperone HSP90 associates with SGT1b and RAR1, and is required for the accumulation of R proteins (Hubert et al., 2003; Liu et al., 2004; Zhang et al., 2004; Boter et al., 2007). We then asked whether HSP90 interacts with RPP4 in vivo. pSuper:RPP4-Myc/pSuper:rpp4-Myc and pSuper:HSP90.3-GFP/pSuper:hsp90.3-GFP were transiently expressed in Arabidopsis protoplasts. RPP4-Myc could precipitate HSP90-GFP. The mutation in rpp4-1d did not affect the interaction between rpp4 and HSP90 (Fig. 6a). In addition, the RPP4–HSP90 interaction was not affected by low temperature or by the mutations in hsp90 alleles (Fig. 6a). We further determined which domains of RPP4, including the TIR, NBS and LRR domains (Fig. 6b), interacted with HSP90 using the firefly LCI assay in N. benthamiana leaves (Fig. S1). As shown in Fig. 6(c), all domains could associate with HSP90. Taken together, these results suggest that RPP4 and HSP90 are in the same complex in vivo.
The rpp4 protein level in plants was not affected by low temperatures
As the rpp4-conferred phenotypes are dependent on a low temperature, we wondered whether the expression of the rpp4 protein was regulated by temperature. The Myc-tagged rpp4 protein was transiently expressed in wild-type protoplasts, which were incubated at 22°C for 16 h and then treated at 4°C for different periods of time. The rpp4 protein levels remained similar during the cold treatment (Fig. S2a). We also found that rpp4 protein levels were not obviously changed after exposure to a higher temperature (28°C) (Fig. S2b).
A previous study showed that the mutations in the ATP-binding domain of HSP90 severely reduced RPM1 accumulation and thus compromised RPM1 function (Hubert et al., 2003). Because the mutations in the hsp90.2-2, hsp90.3-1 and hsp90.3-2 mutants were all located in the conserved ATPase domain of HSP90, it is possible that these mutations may affect the accumulation of the rpp4 protein at low temperatures. To test this possibility, we analyzed the rpp4 protein level at 4°C in protoplasts prepared from the hsp90 mutants. The rpp4 protein was not affected by the hsp90.3-1, hsp90.3-2 or hsp90.2-2 mutation on cold stress (Fig. S2a). To further determine whether the hsp90.3 mutation affects rpp4 protein stability, we expressed rpp4-Myc in protoplasts of Col and hsp90.3 plants for 16 h at 22°C, followed by the addition of the protein synthesis inhibitor cycloheximide (CHX), and examined the protein levels of rpp4. As shown in Fig. S3, the stability of the rpp4 protein was not influenced by the mutation in hsp90.3. Taken together, these results indicate that the rpp4 protein level is not regulated by temperature or the hsp90 mutations.
Nuclear accumulation of rpp4 is decreased at low temperatures
To further investigate the function of HSP90 in the rpp4-mediated chilling response, we investigated the subcellular localization of rpp4 and RPP4 at 22 and 4°C. These proteins were fused to Myc at the C-terminus and were expressed in Arabidopsis protoplasts. RPP4-Myc was mainly located in the cytoplasm and nucleus at 22 and 4°C according to immunoblot analysis (Fig. 7a). The mutated rpp4 protein was also located in both the cytoplasm and the nucleus at 22°C; however, the protein level of rpp4 in the nucleus was dramatically decreased at 4°C relative to 22°C (Fig. 7a).
To examine whether HSP90 mediates the decreased nuclear accumulation of rpp4 at low temperatures, we treated the protoplasts expressing rpp4-Myc with 10 μM GDA. The decreased nuclear accumulation of rpp4 at 4°C was not affected by the application of GDA (Fig. 7b). In addition, rpp4-Myc was expressed in the protoplasts of wild-type Col and hsp90.3-2 plants for 16 h at 22°C which were then treated with cold stress. As shown in Fig. 7(c), the decrease in nuclear rpp4 was not compromised by the mutation in hsp90. Together, these results suggest that the suppression of the rpp4-conferrred phenotype by hsp90.3 does not result from the altered localization of rpp4.
To further dissect which domains are responsible for the entrance of rpp4 into the nucleus, we examined the subcellular localization of the TIR and NB-ARC domains of rpp4 in N. benthamiana leaves transiently expressing different domains of RPP4. Immunoblot analysis showed that the TIR domain was localized in both the cytoplasm and the nucleus, whereas both wild-type NB-ARC and mutated NB-ARC were only detected in the nucleus (Fig. 7d). The expression of the LRR domain was too low to be detected. These results suggest that the TIR and NB-ARC domains may function by coordinately regulating the ratio of RPP4 in the cytoplasm and the nucleus.
Identification and characterization of intragenic suppressors of rpp4-1d
We also identified 25 intragenic revertants of rpp4-1d (rpp4-r) from the M2 population. The chilling sensitivity of rpp4-1d was suppressed in these rpp4-1d revertants when the temperature was shifted from 22 to 4°C (Fig. S4a). The ion leakage of the rpp4-1d plants was also dramatically inhibited by these mutations at 4°C (Fig. S4b). Sequencing analyses showed that most mutations of these suppressors were located in the NB-ARC domain, and a few mutations were located in the LRR domain (Fig. S5). These results suggest that these amino acids in the NB-ARC and LRR domains are required for the function of RPP4.
In addition, we isolated intragenic suppressors resulting from the introduction of stop codons into RPP4, which we used to study the functions of the different domains of RPP4. We found that the deletion mutation in the N-terminus of the NB-ARC domain of rpp4 repressed the chilling sensitivity of rpp4-1d; however, this early terminated mutation activated the defense response. For example, in rpp4-r25, the substitution of G to A at position 933 resulted in the formation of a truncated protein containing only the TIR domain and part of the NB-ARC domain (Fig. S3c). This mutation suppressed the chilling lethality of rpp4-1d at 4°C (Fig. 8a). Intriguingly, this mutation resulted in extensive cell death and H2O2 accumulation at 22°C (Fig. 8b). However, the difference between rpp4-r25 and rpp4-1d was that reactive oxygen species (ROS) accumulation and cell death in the rpp4-r25 mutation was constitutive and independent of the temperature, and the mutant was not lethal, whereas rpp4-1d was very sensitive to temperature and exhibited lethality at 4°C (Fig. 8a).
To further explore the function of the TIR domain of RPP4, we generated transgenic plants overexpressing the TIR domain in the wild-type Col background (TIR-OE) (Fig. S6). The transgenic plants exhibited a dwarf stature, obvious cell death and H2O2 accumulation, as well as up-regulated PR expression at both 22 and 4°C, which is consistent with the phenotypes of rpp4-r25 (Fig. 8a–c). Taken together, these results suggest that the complete TIR domain is sufficient to trigger weak HR.
The mutations in the rpp4-r26 and rpp4-r27 revertants resulted in the introduction of stop codons in front of the LRR domain (Fig. S5). Deletion of the LRR domain failed to activate rpp4 at low temperatures, thus suppressing the chilling sensitivity of rpp4-1d (Fig. 8d). These results suggest that the LRR domain is required for the function of rpp4.
The rpp4-induced chilling sensitivity is largely independent of the MOS (modifier of snc1) genes
Because RPP4 is highly similar to SNC1 (62%) (Zhang et al., 2003), we examined whether the MOS genes were necessary for the rpp4-conferred phenotypes by generating rpp4-1d mos double mutants (Palma et al., 2005; Zhang & Li, 2005; Zhang et al., 2005; Cheng et al., 2009; Monaghan et al., 2009; Li et al., 2010). MOS3, MOS6 and MOS7 all localize to the nuclear envelope (Palma et al., 2005; Zhang & Li, 2005; Cheng et al., 2009). In addition, MOS3 is also required for RPP4-mediated resistance to H. a. Emoy2 (Zhang & Li, 2005). The loss of function of MOS3, MOS6 or MOS7 in rpp4 did not abrogate the rpp4-mediated chilling sensitivity, cell death, H2O2 accumulation or PR gene expression (Fig. 9a), indicating that MOS3, MOS6 and MOS7 are not required for rpp4-mediated signaling. MOS1, MOS2 and MOS4 function in the transcriptional control of the snc1-mediated defense responses (Zhang et al., 2005; Monaghan et al., 2009; Li et al., 2010). In the mos1 mutants, the expression of RPP4 was modestly reduced, but resistance to H. a. Emwa1, which is mediated by RPP4, was not affected by the mos1 mutations (Li et al., 2010). The double mutants of rpp4-1d with mos1, mos2 or mos4 all nearly resembled the rpp4-1d chilling-sensitive phenotypes when grown at 4 or 16°C (Figs 9a, S7); however, the cell death and PR gene expression in rpp4-1d was partially inhibited by the mos1 and mos2 mutations (Fig. 9a). Taken together, these data suggest that, although RPP4 and SNC1 are highly similar homologs, their modes of action are quite different.
The rpp4-induced chilling sensitivity is independent of WRKY70
It has been reported previously that WRKY70 is required for full RPP4-mediated resistance to H. parasitica. However, the induction of RPP4-mediated HR was not fully blocked in wrky70 mutants (Knoth et al., 2007). To assess whether WRKY70 is involved in the rpp4-mediated chilling signaling, we generated rpp4-1d wrky70 double mutants. The rpp4-1d wrky70 double mutant behaved in a similar manner to the rpp4-1d mutant in terms of chilling sensitivity (Figs 9a, S6). H2O2 accumulation in the rpp4-1d wrky70 double mutant was also comparable with that in rpp4-1d (Fig. 9a). However, cell death and PR gene expression in the rpp4-1d mutant were partially suppressed by the wkry70 mutant (Fig. 9b), which was not sufficient to inhibit the rpp4-induced chilling sensitivity. Therefore, unlike RPP4-mediated resistance to pathogens, the rpp4-conferred chilling sensitivity was independent of WRKY70.
In this study, suppressors of chs2-1/rpp4-1d were isolated using a genetic screen. We found that point mutations in HSP90.2 and HSP90.3 could partially suppress the chilling sensitivity of rpp4-1d. However, when exposed to cold stress for > 10 d, the plants exhibited cell death, accumulated H2O2 and eventually died, suggesting that the chilling sensitivity of rpp4-1d is a cumulative effect of damage.
The Arabidopsis genome contains four genes for cytosolic HSP90: HSP90.1, HSP90.2, HSP90.3 and HSP90.4 (Krishna & Gloor, 2001). These genes are arranged in tandem at the bottom of chromosome 5. Although the Arabidopsis HSP90 isoforms are highly related, they have been reported to show different expression profiles (Prasinos et al., 2005). In addition, their loss-of-function mutants display developmental defects with some variation (Samakovli et al., 2007). Consistently, we noticed that point mutations in HSP90.2 and HSP90.3 also showed different, mild morphological alterations, indicating that they played both redundant and distinct roles in plant development. In terms of the stress responses, HSP90.1 is the only cytosolic HSP90 in Arabidopsis to be significantly induced by Pst DC3000. HSP90.1 interacts with RAR1 and SGT1 and is critical for RPS2-mediated disease resistance (Takahashi et al., 2003). In this study, we showed that at least HSP90.2 and HSP90.3 are involved in rpp4-mediated chilling sensitivity, because their mutations could compromise the rpp4-induced chilling sensitivity. However, the T-DNA insertion lines of hsp90.2 and hsp90.3 could not rescue the rpp4-1d phenotype, suggesting the functional redundancy of HSP90. The genetic analysis showed that the F1 progeny of hsp90.2-2 crossed to hsp90.3-1 or hsp90.3-2 suppressed the rpp4-1d phenotype, which is indicative of dosage dependence. That is, the inhibition of the rpp4-1d phenotype only occurred in the presence of at least two copies of the mutated HSP90 genes in the rpp4-1d mutant. This was further confirmed in transgenic plants overexpressing a mutated form of hsp90.3.
A previous study has shown that virus-induced gene silencing of HSP90.1 affects cell death in tobacco caused by RPS4 overexpression (Zhang et al., 2004). Consistent with this, we found that hsp90.3-1 and hsp90.3-2 exhibited weak susceptibility to Pst DC3000 harboring avrRPS4 and avrRPM1. Furthermore, HSP90.3 was shown to be essential for full RPP4-dependent pathogen resistance. These data suggest that RPS4-, RPM1- and RPP4-mediated pathogen resistance is dependent on different HSP90 isoforms. It is also possible that different R proteins may require different quantities of HSP90s.
R proteins are located in different subcellular locations, including the cytoplasm, nucleus, plasma membrane, endoplasmic reticulum and chloroplasts, and their localization is tightly linked to their functions. For instance, the nuclear accumulation of R proteins, such as N, MLA (mildew-resistance locus A) and RPS4, is essential for the triggering of pathogen resistance (Burch-Smith et al., 2007; Shen et al., 2007; Wirthmueller et al., 2007). In addition, snc1-mediated temperature-dependent cell death is associated with the nuclear accumulation of SNC1 (Y. Zhu et al., 2010). A previous study has reported that nuclear processes of avrRps4 are essential for bacterial growth restriction. The programmed cell death and transcriptional defense amplification require nucleocytoplasmic coordination of avrRps4. EDS1 could interact with RPS4 and avrRps4 in the nucleus, respectively. However, when the defense response is activated, the interactions between EDS1 and RPS4 or avrRps4 are only detected in the cytoplasm. This study thus proposes that the effector triggers distinct, but coordinated, subcellular defense branches through an RPS4–EDS1 complex that can accumulate in the cytoplasm and nucleus (Heidrich et al., 2011). A study of the barley MLA protein showed that the nuclear MLA10 pool alone cannot induce cell death, but is sufficient to trigger disease resistance. Enhanced accumulation of MLA10 in the cytoplasm promotes cell death activity, but compromises MLA10-mediated disease resistance (Bai et al., 2012). In this study, we found that the rpp4-mediated temperature-dependent cell death was accompanied by a decrease in the nuclear accumulation of the rpp4 protein. It is possible that the decrease in the nuclear pool of rpp4 may account for this cell death phenotype. It is also possible that the distribution of rpp4 protein between the cytoplasm and nucleus changes, when chilling activates the defense response in the rpp4-1d mutant, similar to that which occurs when the effector is recognized by the R protein. However, attempts to generate transgenic plants expressing rpp4 protein fused to a nuclear localization sequence (NLS) or a nuclear export sequence (NES) were unsuccessful. Therefore, we could not test whether the decrease in nuclear rpp4 was the reason for the rpp4-conferred chilling sensitivity.
Emerging evidence supports the notion that the activation of R proteins is modulated by temperature (Alcazar & Parker, 2011; Hua, 2013). Our previous study showed that one amino acid change in rpp4-1d caused its sensitivity to chilling temperatures (Huang et al., 2010). Genetic analysis of intragenic suppressors revealed that the TIR domain of RPP4 alone could activate defense signaling, although this defense response was not as strong as the rpp4-activated response. The phenotype of the rpp4-1d mutant is blocked when only the TIR-NB-ARC domains are present, implying that NB-ARC may inhibit the activation of the TIR domain. This also suggests that the LRR domain is required to maintain the chilling response of rpp4-1d. The N-terminal TIR or coiled coil (CC) domain of the R protein is thought to be a defense signaling domain, whose activity is controlled by other domains (Swiderski et al., 2009; Bernoux et al., 2011; Maekawa et al., 2011). Consistent with our study, overexpression of the TIR domain of RPS4 in N. benthamiana leaves can trigger cell death (Zhang et al., 2004; Swiderski et al., 2009). Transient expression of the CC domain of MLA10 displays auto-active cell death (Bai et al., 2012). Previous studies have reported that the LRR domain has dual functions in the regulation of the Rx protein through its interaction with CC-NB-ARC. Interference with the intramolecular interactions by the introduction of point mutations in the NB-ARC or LRR domain can activate the Rx protein (Moffett et al., 2002; Rairdan & Moffett, 2006). This indicates that NB-ARC and LRR may play a negative role in controlling the activity of the Rx protein. However, deletion of the LRR domain fails to induce HR although the elicitor exists, suggesting that the LRR domain is necessary for the full activation of the integrated Rx protein. Furthermore, mutations in the LRR of SNC1 confer activity at high temperature, whereas further mutations in NB-ARC abolish the high-temperature but not low-temperature activity (Y. Zhu et al., 2010). Together with our analysis on the intragenic suppressors of rpp4-1d, these results suggest that specific interaction between NB-ARC and LRR is important for temperature-modulated activity.
The wrky70 loss-of-function mutant failed to induce resistance to H. a. Emoy2, and could not fully block the induction of RPP4-mediated HR (Knoth et al., 2007), suggesting that WRKY70 may be a component that is mainly required in RPP4-mediated pathogen resistance, but not in the cell death pathway. Here, we showed that, although the expression of PR genes was partially repressed, the rpp4-conferred chilling sensitivity was not suppressed by the wrky70 mutant, which might be a result of the functional redundancy of WRKY genes. It is conceivable that the rpp4-mediated chilling sensitivity is similar to HR induced by pathogen infection. HR induced by pathogen infection only occurs in a few of the cells that directly contact the pathogen. Different from pathogen-induced HR, rpp4-induced HR presumably happens in the whole plant, thus leading to extensive cell death of whole plants.
Extensive studies of the mos mutants revealed that SNC1-mediated signaling requires nucleocytoplasmic trafficking, transcriptional regulation, RNA processing and protein modification (Palma et al., 2005; Zhang et al., 2005; Zhang & Li, 2005; Goritschnig et al., 2008; Palma et al., 2007; Cheng et al., 2009; Li et al., 2010; Z. Zhu et al., 2010; Xu et al., 2011). RPP4 and SNC1 are close homologs located at the RPP5 locus and are coordinately regulated (Yi & Richards, 2007). Nevertheless, our genetic analysis demonstrated that the rpp4-induced chilling sensitivity was not dependent on these MOS genes. Moreover, SGT1b and RAR1 are not required for snc1 (Goritschnig et al., 2007), but are required for rpp4. These results suggest that rpp4 and snc1 modulate temperature-dependent defense responses through distinct mechanisms.
In summary, we propose a working model for RPP4-mediated temperature-dependent cell death and defense responses controlled by HSP90 (Fig. 10). RPP4 protein maintains a resting state as a result of its intramolecular interactions at normal temperature. However, pathogen attack and mutations in the NB-ARC domain of RPP4 (rpp4-1d) could activate the RPP4 protein at normal and chilling temperatures, respectively, via a change in its intramolecular interactions facilitated by the SGT1b-RAR1-HSP90 complex, which, in turn, leads to EDS1- and WRKY70-dependent cell death and the defense response. As a result, continuous chilling stress-induced HR causes chilling sensitivity of plants. Although the mutated hsp90 can form a complex with RPP4/rpp4, it has a dose-dependent interfering effect on the wild-type HSP90, which consequently prevents RPP4/rpp4 from moving from the resting state to the active state. TIR-NB-ARC fails to activate defense responses, whereas the TIR domain alone released from the inhibition of NB-ARC can activate cell death and the defense response at both normal and chilling temperatures.
This work was supported by China National Funds for Distinguished Young Scientists (31225003) and the National Natural Science Foundation of China (31121002 and 31330006).