Survival of plants at low temperature depends on mechanisms for limiting physiological damage and maintaining growth. We mapped the chs1-1 (chilling sensitive1-1) mutation in Arabidopsis accession Columbia to the TIR-NBS gene At1g17610. In chs1-1, a single amino acid exchange at the CHS1 N-terminus close to the conserved TIR domain creates a stable mutant protein that fails to protect leaves against chilling stress. The sequence of another TIR-NBS gene (At5g40090) named CHL1 (CHS1-like 1) is related to that of CHS1. Over-expression of CHS1 or CHL1 alleviates chilling damage and enhances plant growth at moderate (24°C) and chilling (13°C) temperatures, suggesting a role for both proteins in growth homeostasis. chs1-1 mutants show induced salicylic acid production and defense gene expression at 13°C, indicative of autoimmunity. Genetic analysis of chs1-1 in combination with defense pathway mutants shows that chs1-1 chilling sensitivity requires the TIR-NBS-LRR and basal resistance regulators encoded by EDS1 and PAD4 but not salicylic acid. By following the timing of metabolic, physiological and chloroplast ultrastructural changes in chs1-1 leaves during chilling, we have established that alterations in photosynthetic complexes and thylakoid membrane integrity precede leaf cell death measured by ion leakage. At 24°C, the chs1-1 mutant appears normal but produces a massive necrotic response to virulent Pseudomonas syringae pv. tomato infection, although this does not affect bacterial proliferation. Our results suggest that CHS1 acts at an intersection between temperature sensing and biotic stress pathway activation to maintain plant performance over a range of conditions.
Temperature modulates plant resistance to pathogen infection, with higher temperatures tending to reduce activation of anti-microbial defense pathways and responses triggered by isolate-specific Resistance (R) genes (Wang et al., 2009; Alcázar and Parker, 2011). Most characterized R genes encode intracellular nucleotide-binding site/leucine-rich repeat (NBS-LRR) receptors that are activated by specific pathogen effectors (Dodds and Rathjen, 2010). NBS-LRR receptors trigger local resistance associated with programmed cell death as part of a hypersensitive response, and amplify basal defenses involving the signaling hormone salicylic acid (SA), leading to systemic resistance (Dodds and Rathjen, 2010). Lowering of temperature reduces the threshold for NBS-LRR defense activation, which, if not checked, leads to stunting of plant growth (Alcázar and Parker, 2011). The actions of one class of NBS-LRR receptor that contains an N-terminal Toll interleukin-1 receptor (TIR) domain (TIR-NBS-LRRs) are especially temperature-sensitive. For example, constitutive immunity and stunting caused by auto-active TIR-NBS-LRR mutants or mutations in TIR-NBS-LRR regulators are suppressed at high temperature (Yang and Hua, 2004; Wang et al., 2009). Also, temperature-sensitive growth inhibition, defense pathway activation and necrosis caused by allelic mis-matches in Arabidopsis incompatible hybrids involve TIR-NBS-LRR genes (Bomblies et al., 2007; Alcázar et al., 2009). The Arabidopsis TIR-NBS-LRR protein SNC1 was shown to act as a temperature sensor in resistance and growth homeostasis (Zhu et al., 2010). Temperature also determines the expression of TIR-NBS-LRR regulatory genes such as EDS1 (enhanced disease susceptibility 1) and PAD4 (phytoalexin deficient 4), which function in a feed-forward defense amplification loop (Feys et al., 2001; Yang and Hua, 2004; Wang et al., 2009).
Cold-tolerant plants must suppress hypersensitive response-related cell death at low temperatures (Huang et al., 2010b). Exposure of Arabidopsis plants to cold results in induction of the transcription factor CBF (C-repeat binding factor), which activates expression of cold responsive (COR) genes (Gilmour et al., 1998). CBF/COR-dependent factors regulate metabolism, allowing plants to withstand subsequent freezing stress (cold acclimation) (McClung and Davis, 2010).
Mutagenized Arabidopsis plants of accession Columbia (Col) were screened for altered growth and survival at 13°C, resulting in isolation of 21 chilling sensitive (chs) mutants (Hugly et al., 1990). Low-temperature activation of defense and chilling sensitivity in chs2 was due to an auto-activated allele of the TIR-NBS-LRR receptor RPP4 (Resistance to Peronospora parasitica4) (Huang et al., 2010a). Another isolated gain-of-function mutant, chs3-1, mapped to a TIR-NBS-LRR gene with a predicted zinc-binding LIM (Lin-11, Isl-1 and Mec-3) domain at its C-terminus (Yang et al., 2010). Growth defects and cell death phenotypes in chs2 and chs3-1 were suppressed by eds1 (Huang et al., 2010a; Yang et al., 2010). The chs1-1 mutant, attributed to a single mutation on chromosome 1, exhibits leaf chlorosis and death at 13°C, but roots, seeds or calli are not affected (Hugly et al., 1990). Comparative gene expression, protein and lipid profiling revealed that chloroplast metabolism, integrity and intracellular lipid trafficking are disturbed in chs1-1 compared to wild-type (WT) during chilling (Hugly et al., 1990; Patterson et al., 1993; Provart et al., 2003).
To understand the molecular basis for the low-temperature sensitivity of chs1-1, we isolated the CHS1 gene and found that it belongs to a family of so far uncharacterized TIR-NBS genes. Our analysis shows that CHS1 protects leaves from chilling-induced cell death by dampening EDS1/PAD4 defenses and chloroplast-associated degradative processes. The results suggest a close relationship between temperature modulation of EDS1/PAD4 disease resistance signaling pathways and chloroplast metabolism.
Isolation of CHS1
Previously, Arabidopsis Col chs1-1 was isolated as a recessive mutation located within a 28.6 cm interval on chromosome 1 (Hugly et al., 1990). We grew Col wild-type (WT) and chs1-1 plants for 28 days at 24°C before transferring to 13°C. The first chilling symptoms (leaf chlorosis) were visible after 2–4 days. After 8 days at 13°C, homozygous chs1-1 plants are easily distinguished from WT (Figure 1). This chilling assay was used to fine map chs1-1 in an F2 population derived from a chs1-1 ×Landsberg erecta (Ler) cross, and the CHS1 interval was narrowed to 200 kb (Figure S1a). Heterozygous chs1-1 plants displayed intermediate chlorosis between WT and homozygous chs1-1 (Figure 1), indicative of CHS1 semi-dominance or haplo-insufficiency. Cosmid clones containing WT genomic DNA spanning the CHS1 interval were isolated and transformed into chs1-1. Transformants containing cosmid clone II (line chs1-1 gCHS1) exhibited almost WT chilling resistance, indicating complementation (Figure 1 and Figure S1a). Cosmid II encompasses four predicted full-length open reading frames (ORFs) (Figure S1a). Only one gene, At1g17610, had a point mutation in chs1-1. A G→A substitution at position 28 of the ORF results in an Ala→Thr exchange at position 10 of the predicted CHS1 amino acid sequence (Figure 2). Sequencing of At1g17610 in a second allele (chs1-2) (Hugly et al., 1990) revealed that it had the same point mutation as chs1-1. The At1g17610 1263 bp ORF encodes a 421 amino acid protein. We amplified the At1g17610 sequence from Col genomic DNA by PCR, and transformed it under the control of the CaMV 35S promoter into chs1-1. All transgenic T1chs1-1 plants expressing a WT copy of At1g17610 survived at low temperature, indicating complementation (Figure 1).
CHS1 expression is not affected by chilling or the chs1-1 mutation
CHS1 is moderately expressed in leaves and flowers, as evaluated by GeneChip expression data (AtGenExpress; www.arabidopsis.org). Our analysis by quantitative RT-PCR revealed that CHS1 is expressed in all organs (Figure S2a,b), with highest expression in leaves and flowers and lower expression in stems, siliques and roots. The CHS1 transcript abundance was indistinguishable between WT and chs1-1, indicating that chs1-1 is not an mRNA null mutant. CHS1 mRNA levels did not change in WT or chs1-1 after exposure to 13°C (Figure S2b), indicating that chilling does not substantially alter CHS1 expression. The CHS1 promoter was fused to the β-glucuronidase (GUS) gene and transformed into WT. CHS1 promoter activity was observed in leaves, and, at a lower intensity, in stems. Expression was low in roots, flowers and seeds (Figure S2) and independent of chilling, in agreement with quantitative RT-PCR data.
A single insertion line (chs1-3) was identified (SAIL_95_C03) that carries a T-DNA insertion 3' to the stop codon of CHS1 (Figure 3). Homozygous chs1-3 plants showed similar CHS1 gene expression and growth at normal and low temperatures compared to WT, indicating that the T-DNA insertion does not disrupt CHS1 function (Figure 3a,b). As an alternative strategy to obtain plants with reduced CHS1 expression, we introduced a double-stranded CHS1 RNAi construct into WT. Two independent transformants with diminished CHS1 expression (CHS1 RNAi lines 1 and 8) were selected by quantitative RT-PCR (Figure 3b). Exposure to low temperature revealed that the CHS1 RNAi lines displayed chilling sensitivity, but to a lower extent than chs1-1 (Figure 3a). Taken together, the results show that CHS1 confers chilling tolerance in Arabidopsis.
CHS1 belongs to the TIR-NBS sub-family of Resistance-like proteins
BLAST searches revealed that CHS1 encodes a protein with an N-terminal TIR domain adjacent to an NBS domain, which is characteristic of TIR-NBS-LRR immune receptors (Figure 2 and Figure S1c) (Meyers et al., 2003). CHS1 lacks an LRR domain and therefore belongs to a family of 20 TIR-NBS genes and one TIR-NBS pseudogene in Arabidopsis for which functions have not yet been described (Meyers et al., 2003). A TIR-NBS gene, At5g40090, on chromosome 5 has highest similarity to CHS1 (56% amino acid sequence identity), and we therefore named it CHL1 (CHS1-like 1) (Figure 2 and Figure S1). The high similarity between CHS1 and CHL1 suggests that the two proteins may have overlapping functions. Similar to CHS1, CHL1 transcripts were present in all organs, and expression in leaves was independent of chilling (Figure S2c,d). In contrast to CHS1, expression of CHL1 was higher in roots than leaves, in agreement with GeneChip expression data (AtGenExpress) (Figure S2). We tested a line (chl1, SAIL_558_A07) that harbors a T-DNA insertion within the CHL1 ORF (Figure 3). Homozygous chl1 plants showed reduced CHL1 transcript accumulation (Figure 3), but did not differ from WT when grown at normal or chilling temperatures. Therefore, CHL1 does not contribute to chilling tolerance. Alternatively, the presence of WT CHS1 compensates for chl1 reduced function.
Because of the high degree of sequence identity, we considered the possibility that CHL1 expression was also down-regulated in the CHS1 RNAi lines. The 205 bp CHS1 RNAi sequence showed 81% identity with CHL1, but was not similar to other sequences in the Arabidopsis genome (BLASTN). Indeed, CHL1 expression was reduced in CHS1 RNAi lines 1 and 8 (Figure 3b). As reduced CHL1 expression in the chl1 mutant plants is not associated with chlorosis during chilling (Figure 3a), we concluded that the chilling phenotype of the CHS1 RNAi line is due mainly to reduced CHS1 expression.
The CHS1 gene on chromosome 1 is flanked by a second TIR-NBS gene (At1g17615) and a predicted full length TIR-NBS-LRR gene (At1g17600) (Figure S1b). Another locus on chromosome 5 (At5g40100) adjacent to CHL1 encodes a TIR-NBS-LRR protein with highest similarity to At1g17600 (Figure S1b,c) (Meyers et al., 2003). No functions have been assigned to At1g17615, At1g17600 or At5g40100, and no other TIR-NBS-LRR genes lie in these two clusters. At1g17600 and At1g17610 (CHS1) on chromosome 1, and At5g40100 and At5g40090 (CHL1) on chromosome 5 are presumably derived from a segmental chromosomal duplication (Figure S1). Many genes of the TIR-NBS sub-class possess an intron between the TIR and NBS sequences, but four genes including CHS1 and CHL1 lack introns (Figure S1) (Meyers et al., 2003).
The chs1-1 mutation lies at the N-terminus of the TIR domain
The Ala→Thr exchange in chs1-1 is close to the N-terminus of CHS1 (Figure 2). CHL1 has a similar N-terminal extension and carries a Leu residue at the equivalent position (Figure 2). This short sequence precedes the start of the TIR domain of TIR-NBS-LRR receptors such as flax L6 and Arabidopsis RPP5 (Figure 2) (Bernoux et al., 2011; Maekawa et al., 2011). Four TIR motifs (TIR1–4) were defined based on amino acid sequence similarity between TIR proteins (Meyers et al., 2003), and more recently the crystal structures of AtTIR (At1g72930) (Chan et al., 2010) and the flax L6 TIR-NBS-LRR immune receptor (Bernoux et al., 2011). All four TIR sub-domains are present in CHS1 and CHL1, as determined by sequence homologies (Figure 2). Therefore, the two proteins probably contain functional TIR domains.
The NBS of NBS-LRR proteins is part of a larger NBS-ARC domain defined in APAF-1, plant R proteins and CED-4 (van Ooijen et al., 2008; Maekawa et al., 2011). Crystallization and structural analysis of APAF-1 and CED-4 revealed various NBS-ARC sub-domains: NBS (α/β fold), ARC1 (helical domain I), ARC2 (winged-helix domain) and ARC3 (helical domain II). The latter is missing in CED-4 and plant NBS-LRR proteins (Riedl et al., 2005; van Ooijen et al., 2008; Qi et al., 2010). ADP is bound by amino acids in the NBS, ARC1 and ARC2 sub-domains of APAF-1, and the NBS and ARC2 sub-domains are predicted to participate in nucleotide hydrolysis (Qi et al., 2010; Williams et al., 2011). Motifs defining the NBS and ARC1 sub-domains are conserved in CHS1, CHL1, L6 and RPP5 (Figure 2). However, the RNBS-D and MHD motifs of ARC2 are not present in CHS1 or CHL1. While the RNBS-D motif is also not conserved in APAF-1 or CED-4 and does not contribute to ADP binding, the MHD domain is present in APAF-1, CED-4 and many R proteins (van Ooijen et al., 2008). His438 of the MHD domain binds to the β-phosphate of ADP and is involved in ADP–ATP exchange (Riedl et al., 2005; van Ooijen et al., 2008; Williams et al., 2011). In several NBS-LRR proteins, mutations in the MHD sequence result in receptor auto-activation, probably by affecting nucleotide hydrolysis (van Ooijen et al., 2008). We concluded that CHS1 and CHL1 contain sequences with striking similarity to functional TIR-NBS domains, especially within the TIR portions, but are unlikely to autonomously bind and hydrolyze nucleotides.
Over-expression of CHS1 or CHL1 increases plant biomass
We tested whether CHS1 or CHL1 over-expression (OE) affects plant growth at moderate or low temperature by transforming Col with either gene driven by the 35S promoter. Two independent CHS1 OE lines (lines 1 and 6) and CHL1 OE lines (lines 1 and 5) were selected as high expressors by quantitative RT-PCR (Figure 4a). When grown at 24°C, the CHS1 OE and CHL1 OE lines developed normally but had higher fresh weights than WT after 28 days (one-way anova, P <0.002) (Figure 4b,c). This growth differential with WT was even more pronounced at 13°C, at which temperature chs1-1 deteriorated within 8 days while CHS1 RNAi lines survived but were stunted (Figure 3a). Therefore, high levels of CHS1 or CHL1 expression result in enhanced growth, which is most obvious during chilling (Figure 4). These data also suggest that CHL1 produces a functional protein with a CHS1-related activity.
CHS1 fractionates as a soluble, cytoplasmic protein
We monitored the expression and intracellular localization of native CHS1 protein in Arabidopsis using a polyclonal anti-CHS1 antiserum raised against a synthetic peptide unique to CHS1. To confirm the specificity of the anti-CHS1 antiserum, we performed a Western blot experiment with recombinant CHS1 protein expressed in yeast. A specific band migrating at approximately 50 kDa was detected in yeast expressing CHS1, in agreement with a calculated CHS1 mass of 47.6 kDa (Figure S3). As chs1-1 is not an mRNA null mutant (Figure 3), we tested the antibody specificity on Western blots of leaf extracts from chs1-1 and the CHS1 RNAi and OE lines. A single band migrated at approximately 50 kDa that was stronger in the CHS1 OE lines and weaker in CHS1 RNAi lines compared to WT (Figure 5a), consistent with the antibody specifically recognizing CHS1. The similar signal intensities in chs1-1 and Col suggest that the chs1-1 mutation creates a stable but defective protein.
Fractionation of leaf extracts from WT plants grown at 24°C followed by Western blot analysis with anti-CHS1 antibody and antibodies recognizing cell compartment-specific markers showed that CHS1 protein is mainly soluble and cytosolic (Figure 5b), although we cannot exclude the existence of a nuclear CHS1 pool that becomes depleted upon fractionation (García et al., 2010). Transfer of Col plants to 13°C for 4 days did not alter CHS1 localization. Similarly, fractionation of chs1-1 leaves revealed cytoplasmic CHS1 at 24 and 13°C (Figure 5c). Thus, CHS1 amounts and intracellular distribution do not appear to be substantially altered by chilling or reduced CHS1 function.
Chilling activates immune responses in chs1-1
The relatedness of CHS1 to the N-terminal portions of TIR-NBS-LRRs, together with the known temperature dependence of basal and TIR-NBS-LRR triggered immunity (Alcázar and Parker, 2011), raised the possibility that the physiological role of CHS1 is to limit disease resistance responses. We therefore tested whether chs1-1 displays hallmarks of activated defenses. Accumulation of SA is an important basal resistance output (Feys et al., 2001; Zhang et al., 2003). We observed a small but significant increase in total SA in chs1-1 compared to WT at 24°C (Figure 6a). After transfer to 13°C, SA levels increased further by approximately tenfold in chs1-1 at 4 days, but remained unchanged in WT over 10 days (Figure 6a). Transcripts of genes associated with SA accumulation or the hypersensitive response (ICS1, isochorismate synthase1; EDS5, enhanced disease susceptibility5; PR1, 2, 5, pathogenesis related genes 1, 2, 5; EDS1 and PAD4) were not detected on RNA gel blots of WT or chs1-1 plants grown at 24°C (0 h; Figure 6b), but were strongly up-regulated in chs1-1 but not WT over a period of 24 h after transfer to 13°C (Figure 6b). Therefore, the chs1-1 mutation causes SA pathway activation at low temperature.
We crossed chs1-1 with various defense mutants: eds1-2 to disable all TIR-NBS-LRR triggered defenses (Aarts et al., 1998), pad4-1 to abolish SA-dependent and SA-independent pathways but not TIR-NBS-LRR-triggered cell death (Jirage et al., 1999; Feys et al., 2001), ndr1 to suppress resistance by many coiled-coil-NBS-LRR receptors (Aarts et al., 1998; Knepper et al., 2011) and sid2-1 (gene ICS1) to prevent pathogen-induced SA accumulation (Wildermuth et al., 2001). Homozygous double mutants of each chs1-1 combination were tested for chilling sensitivity. Whereas the eds1-2 chs1-1 and pad4-1 chs1-1 mutants were indistinguishable from WT after exposure to 13°C, ndr1 chs1-1 and sid2-1 chs1-1 mutants developed chlorosis (Figure 6c). These trends were reflected in chlorophyll measurements. Total leaf chlorophyll amounts were similar in all lines except chs1-1, ndr1 chs1-1 and sid2-1 chs1-1, in which chlorophyll contents decreased to approximately 50% of WT 4 days after moving plants to 13°C (Figure 6d). The requirement for EDS1 and PAD4 but the dispensability of SA and NDR1 in chs1-1 leaf deterioration suggests that activation of a basal or TIR-NBS-LRR-triggered, SA-independent resistance branch underlies chs1-1 chilling sensitivity.
Changes in chloroplast ultrastructure and thylakoid lipid composition during chs1-1 chilling injury
The appearance of leaf chlorosis as the first macroscopic change in chs1-1 suggests that chloroplast malfunctions may underpin the deterioration of chs1-1 plants at 13°C. We examined the leaf lipid composition and chloroplast ultrastructure of mesophyll cells during chilling. Monogalactosyldiacylglycerol (MGDG) and phosphatidylglycerol (PG) are abundant thylakoid lipids. MGDG in Arabidopsis characteristically has a high proportion of triunsaturated fatty acids (16:3 and 18:3). Therefore, changes in the amounts of MGDG, PG and tri-unsaturated fatty acids are indicative of thylakoid lipid degradation. The levels of MGDG and PG and the relative proportions of 16:3 and 18:3 decreased in chs1-1 at approximately 4 days during chilling (Figure S4). Electron microscopy revealed that chs1-1 chloroplasts maintained the same size and ultrastructure as those of WT at 24°C (Figure 7). After 2–4 days of chilling, thylakoid membranes lost integrity and plastoglobule numbers increased in chs1-1 but not WT (Figure 7). Plastoglobules are protein–lipid particles in chloroplasts that take up lipid breakdown products, such as fatty acid phytyl esters derived from chlorophyll and galactolipid degradation (Lippold et al., 2012). We concluded that chilling causes the degradation of chlorophyll, thylakoid lipids and fatty acids, resulting in disintegration of thylakoid membranes in chs1-1.
Chloroplast malfunctions precede cell death in chs1-1
We examined more precisely the relationship between chloroplast changes and cell death. Leaves were harvested from WT, chs1-1, the complemented line chs1-1 gCHS1, CHS1 RNAi lines 1 and 8 and CHS1 OE lines 1 and 6 at daily intervals after transfer from 24 to 13°C. We then determined leaf chlorophyll content and photosynthetic quantum yield (FV/FM) as measures of chloroplast function, and ion leakage as a measure of cell death. There were no changes in these parameters in WT, line chs1-1 gCHS1 or CHS1 OE lines (Figure 8). In chs1-1, and to a lesser extent CHS1 RNAi lines, chlorophyll contents declined as early as 1 day after chilling (Figure 8a). A lower quantum yield was apparent in chs1-1 and CHS1 RNAi lines compared to WT at approximately 5 days of chilling. The extent of ion leakage in chs1-1 and the CHS1 RNAi lines was similar to WT at 1–2 days of chilling, but between 4 and 8 days, ion leakage increased dramatically in chs1-1 and to a lesser extent in the CHS1 RNAi lines (Figure 8c). These data show that certain chloroplast malfunctions precede leaf cell death in chs1-1 plants at low temperature but that a decline in photosynthetic performance occurs quite late together with a massive increase in cell death.
Chilling leads to xanthophyll cycle activation and accumulation of tocopherol in chs1-1
The hypersensitive response is accompanied by accumulation of reactive oxygen species (ROS) (Rustérucci et al., 2001), and various anti-oxidative mechanisms, including the xanthophyll cycle and tocopherol accumulation, respond to oxidative stress (Havaux and Niyogi, 1999; Havaux et al., 2005). We found that violaxanthin levels were reduced in chs1-1, while antheraxanthin and zeaxanthin accumulated as early as 1–2 days after the start of chilling, indicating xanthophyll cycle activity. The levels of β-carotene, lutein and neoxanthin decreased in chs1-1 during chilling (Figure S5), consistent with a general disintegration of thylakoid membranes (Figure 7).
Tocopherol (vitamin E) also acts as a ROS quencher (Havaux et al., 2005). Total tocopherol levels rose in chs1-1 more rapidly than in WT during chilling (Figure S6). Unusual forms (γ-tocopherol and δ-tocopherol, differing by the number of methyl groups) accumulated to high levels in chs1-1 leaves, while amounts of α-tocopherol, the common form in WT leaves, did not change (Figure S6). Notably, δ-tocopherol provides stronger antioxidant activity than α-tocopherol (Kamal-Eldin and Appelqvist, 1996). These metabolic changes indicate that chs1-1 responds acutely to photosynthesis-related ROS stress during chilling.
CHS1 limits leaf necrosis associated with virulent bacterial infection
We also investigated the role of CHS1 in disease resistance responses by spray-inoculating chs1-1, CHS1 RNAi lines 1 and 8, CHS1 OE lines 1 and 6, CHL1 OE lines 1 and 5, WT and eds1-2 with virulent Pseudomonas syringae pv. tomato DC3000 (Pst) bacteria. Plants were grown and infected at 24°C to avoid chilling damage in chs1-1 and CHS1 RNAi lines, and Pst titers (colony forming units, cfu) were measured in leaves at 0 and 3 days post-infection (dpi). While Pst multiplied to high levels (107–108 cfu cm−2) in all plants at 3 dpi, chs1-1 and the CHS1 RNAi lines exhibited a more severe necrotic response than WT at 14 dpi (Figure 9). By contrast, the CHS1 OE and CHL1 OE lines showed reduced lesion development and were larger and greener than all other genotypes at 14 dpi (Figure 9b). These dramatic differences suggest a role for CHS1 and CHL1 in restricting host disease-associated damage and cell death, but not in limiting bacterial growth. We also spray-inoculated plants at 24°C with Pst expressing the effector AvrRps4 (Pst/AvrRsp4) triggering EDS1-dependent RPS4/RRS1 TIR-NBS-LRR receptor resistance (Birker et al., 2009; Narusaka et al., 2009). All lines except the susceptible eds1-2 mutant showed restricted Pst/AvrRps4 growth at 3 dpi and necrotic symptoms at 14 dpi (Figure S7a). Therefore, CHS1 and CHL1 have little impact on AvrRps4-triggered bacterial resistance or macroscopic cell death (Figure S7b).
Survival over prolonged periods of cold requires adaptation to chilling temperatures and mechanisms for dampening the deleterious consequences of activated stress pathways (McClung and Davis, 2010; Alcázar and Parker, 2011). Here we identify a role for one member of an as yet uncharacterized family of TIR-NBS genes, CHS1, in protection of leaves against chilling damage. Analysis of the chs1-1 mutant and lines with altered CHS1 expression shows that CHS1 restricts the effects of cold temperature (13°C) by limiting chloroplast damage and cell death. At low temperature, CHS1 helps to maintain photosynthetic performance and growth. At moderate temperature (24°C), CHS1 limits necrosis associated with Pst infection. This is not accompanied by enhanced basal resistance to bacterial growth, thus uncoupling control of cell death from containment of virulent pathogens. We propose that CHS1 is necessary for maintaining biotic stress and growth homeostasis over a range of temperatures.
Loss of CHS1 activates the EDS1–PAD4 resistance pathway
The predicted protein sequence of CHS1 and its paralog CHL1 have features of a conserved, functional TIR domain (Figure 2) that defines a large family of TIR-NBS-LRR immune receptors (Meyers et al., 2003). The LRR domain is important for receptor auto-inhibition and activation upon pathogen effector recognition (Maekawa et al., 2011). The fact that the conserved TIR domains of CHS1 and CHL1 do not behave as defense auto-activating or cell suicide molecules, even during chilling, may be due to the presence of the NBS-ARC domains, as these domains suppress auto-necrosis elicited by the TIR portion of flax L6 (Bernoux et al., 2011). However, unlike L6 and other functional TIR-NBS-LRRs, the CHS1 and CHL1 NBS-ARC domains lack essential features (most conspicuously a conserved MHD motif) that are necessary for autonomously binding and hydrolyzing ATP (Figure 2) (van Ooijen et al., 2008; Bernoux et al., 2011). Therefore, CHS1 and CHL1 probably behave as non-canonical signaling or signal-inhibiting scaffolds.
The chilling sensitivity of chs1-1 has genetic hallmarks of a TIR-NBS-LRR-triggered disease resistance response because it requires EDS1 and PAD4 but not NDR1 or SA. In effector-triggered immunity, TIR-NBS-LRRs recruit an EDS1/PAD4 basal resistance signaling function (Feys et al., 2001; Rietz et al., 2011). An EDS1/PAD4 driven, SA-independent branch in the TIR-NBS-LRR/basal resistance signaling pathways has been identified but remains poorly defined (Zhang et al., 2003; Bartsch et al., 2006; Tsuda et al., 2009). TIR-NBS-LRR-triggered and basal resistance pathways are acutely sensitive to temperature (Yang and Hua, 2004; Bomblies et al., 2007; Alcázar et al., 2009; Wang et al., 2009). Consistent with the temperature dependence of auto-activated TIR-NBS-LRR mutants, the chilling sensitivity of chs2 and chs3-1 is caused by gain-of-function amino acid exchanges in TIR-NBS-LRR genes (Huang et al., 2010a; Yang et al., 2010). Also, introduction of sid2-1 into chs2 or chs3-1 did not suppress their chilling phenotypes (Huang et al., 2010a; Yang et al., 2010). In contrast to chs2 and chs3-1, the chilling sensitivity of chs1-1 is inherited as a semi-dominant trait (Figure 1) (Hugly et al., 1990). Thus, the CHS1 WT protein may molecularly restrain TIR-NBS-LRR receptor activation or signaling during exposure to cold. As steady-state expression of CHS1 itself is not strongly regulated by temperature (Figure S2), CHS1 may only become engaged once TIR-NBS-LRR pathways are activated. A possible scenario is CHS1 homotypic dimerization with an exposed TIR domain of one or more activated TIR-NBS-LRR proteins. In mammalian innate immunity, intracellular signal relay from membrane-bound Toll-like receptors depends on homotypic TIR–TIR interactions between the receptor and TIR domain signaling adaptors (Kawai and Akira, 2007). In plants, TIR domain homo-dimerization of the flax L6 rust resistance receptor was necessary for cell death signaling (Bernoux et al., 2011). Thus, a TIR-NBS protein may interfere with signaling-productive TIR–TIR domain interactions. Alternatively, the CHS1 TIR-NBS moiety may compete with activated TIR-NBS-LRRs for association with signaling components such as EDS1 or PAD4, as these regulators were found to reside in complexes with several TIR-NBS-LRR receptors (Bhattacharjee et al., 2011; Heidrich et al., 2011; Kim et al., 2012b).
Importance of the CHS1 N-terminus in controlling chilling stress
An Ala→Thr exchange in chs1-1 lies at the extreme N-terminus of CHS1 close to the TIR1 motif (Figure 2) (Chan et al., 2010; Bernoux et al., 2011). This mutation gives rise to an apparently stable CHS1 protein with reduced or no function (Figures 1 and 2), and may therefore define an important N-terminal region of CHS1 and CHL1. Related serine-rich sequences at the N-termini of AtTIR1 and L6 did not reveal clear electron density, and therefore may be disordered (Chan et al., 2010; Bernoux et al., 2011). Some TIR-NBS-LRR proteins extend further N-terminally to include membrane-association domains that are not present in CHS1, CHL1 or RPP5 (Takemoto et al., 2012). The exchange in chs1-1 results in replacement of a non-polar amino acid (Ala) with a polar amino acid (Thr) in a motif of four non-polar residues that are highly conserved between CHS1 and CHL1 (LLAG and LLLG, respectively) but absent from AtTIR, L6 and RPP5 (Figure 2). It remains to be seen whether this short hydrophobic sequence is critical for regulation of biotic stress responses and growth homeostasis by CHS1 and CHL1.
Chloroplast damage precedes cell death in chs1-1 chilling sensitivity
We determined the sequence of events during deterioration of chs1-1 leaves after moving plants to 13°C, and found that SA accumulation, expression of defense-related genes (Figure 6b) and a decrease in chlorophyll content (Figure 8a) occurred within 1 day. The induction of certain defense genes was detected as early as 2–4 h (Figure 6b), and therefore nuclear transcriptional reprogramming represents an early change in chs1-1 chilling sensitivity, perhaps contributing to an initial pertubation of chloroplast metabolism. Another early change in chs1-1 was accumulation of zeaxanthin and antherxanthin at 1–2 days (Figure S5). Activation of the photoprotective xanthophyll cycle leads to dissipation of excess light energy. The product, zeaxanthin, is itself a quencher of 1O2 (Triantaphylidès and Havaux, 2009). Conversion of violaxanthin into zeaxanthin is catalyzed by violaxanthin de-epoxidase which is rapidly activated by acidification of the thylakoid lumen under conditions of photosystem II over-reduction (Arnoux et al., 2009). Hence, a detectable xanthophyll cycle in chs1-1 after only 1–2 days of chilling indicates early changes in photosynthetic complexes. Accumulation of tocopherols (in particular δ-tocopherol) and fatty acid phytyl esters (Patterson et al., 1993), thylakoid membrane disintegration and galactolipid breakdown were detected later (2–4 days) during chs1-1 chilling (Figure 7 and Figures S4 and S6). The quantum yield (FV/FM) remained unchanged in chs1-1 and then decreased after 4–5 days at 13°C. Therefore, photo-inhibition of photosystem II in chs1-1 is a late event in chilling damage, suggesting that chlorophyll degradation (at 1 day) is not due to photo-oxidation. This agrees with previous work showing that chlorosis in chs1-1 was independent of light intensities (Hugly et al., 1990). Because ion leakage increased in chs1-1 only after 3–4 days of chilling (Figure 7c), certain chloroplast changes precede cell death. Disturbance of chloroplast homeostasis and lipid metabolism may stimulate immune and stress responses (Kachroo and Kachroo, 2009). Also, 1O2 produced inside chloroplasts causes photo-inhibition of photosystem II and loss of chloroplast integrity leading to cell death initiation (Wagner et al., 2004; Kim et al., 2012a). A requirement for EDS1 in the spread of 1O2-elicited cell death in Arabidopsis flu mutant plants supports a broader role for EDS1, and possibly also TIR-NBS-LRRs, in adaptation to stress (Ochsenbein et al., 2006).
CHS1 and CHL1 as regulators of growth homeostasis
Strikingly, over-expression of CHS1 or CHL1 enhanced plant growth and health at 13°C, and to a lesser degree at 24°C (Figure 4), suggesting that both proteins have a role in maintaining growth against stress or defense pathway activation. An important trade-off for induced anti-microbial defenses is stunting, which affects survival and competitiveness in nature (Tian et al., 2003; Alcázar and Parker, 2011). We propose that CHS1 and CHL1 are important components of growth homeostasis. After Pst infection, chs1-1 produced a massive necrotic response (Figure 9b). By contrast, CHS1 and CHL1 OE lines were less damaged than WT (Figure 9). Therefore, CHS1 and CHL1 limit pathogen-induced necrosis. These CHS1 and CHL1 actions were not coupled to bacterial growth (Figure 9a). In contrast to virulent Pst infection, altering CHS1 and CHL1 had no impact on AvrRps4-triggered RPS4/RRS1 immunity (Figure S7a). A lack of macroscopic cell death in Pst/AvrRps4-inoculated leaves presumably reflects the ability of RPS4/RRS1 in Col WT to restrict bacterial growth without a strong cell death response (Birker et al., 2009; Heidrich et al., 2011). Thus, CHS1 and CHL1 may operate only once a certain defense threshold has been reached to dampen EDS1/PAD4 signaling, limit cellular damage and thus maintain photosynthetic performance and growth.
Plant growth and chilling conditions
All Arabidopsis lines (in accession Col) were grown in phytotrons at 24°C (with 120 μmol m−2 sec−1 light intensity, a 16 h light period and 65% relative humidity). Seeds of chs1-1 and a second allele (chs1-2) were obtained from the Arabidopsis Biological Resource Center (https://abrc.osu.edu/). T-DNA insertion mutants for CHS1 (SAIL_95_C03) and CHL1 (SAIL_558_A07) were obtained from the European Arabidopsis Stock Center (http://arabidopsis.info/). In chilling experiments, plants were grown under standard conditions (24°C) for 4 weeks before moving to low temperature (13°C, 100 μmol m−2 sec−1, 16 h light, 55% relative humidity). The Arabidopsis mutants used were eds1-2 (Bartsch et al., 2006), ndr1 (Century et al., 1995), pad4-1 (Glazebrook et al., 1997) and sid2-1 (Nawrath and Métraux, 1999), all in the Col background.
Positional cloning of CHS1
The chs1-1 locus was previously mapped to chromosome 1 of Arabidopsis (Hugly et al., 1990). One hundred recombinants with crossovers between the markers F21M12 and ciw12 spanning the chs1-1 interval were selected from >1000 Landsberg erecta (Ler) × chs1-1 F2 plants. Fine mapping established the CHS1 location between markers T13M22-NheI and F2H15-SacI. Cosmid clones containing Col genomic DNA covering the CHS1 locus were isolated (Meyer et al., 1994), aligned after restriction analysis using HindIII and EcoRI (according to Arabidopsis genomic sequence: Arabidopsis Information Resource database; www.arabidopsis.org), and transferred into chs1-1 via Agrobacterium transformation.
CHS1 complementation, over-expression and RNAi lines
Full-length open reading frames for CHS1 and CHL1 were amplified from genomic DNA by PCR using primers PD1057 and PD1058 (for CHS1) or Bn361 and PD1061 (for CHL1) (Table S1). The PCR products were ligated into the BamHI and EcoRI sites of p35OCS containing the 35S promoter. Expression cassettes were introduced into the SfiI sites of binary vector pLH9000. The vectors p35OCS and pLH9000 were obtained from DNA Cloning Service (www.DNA-cloning.com; Hamburg, Germany). A 205 bp fragment of the CHS1 ORF was amplified from genomic DNA by PCR using primers PD1253 and PD1254, and cloned into pENTR/SD/D-TOPO (Invitrogen, www.invitrogen.com). The insert was transferred in two orientations into pK7GWIWG2(II) (Invitrogen) to generate a CHS1 RNAi construct under the control of the 35S promoter. Constructs were transformed into Col and transgenic lines were screened by quantitative RT-PCR.
Analysis of gene expression and sequence comparisons
Quantitative RT-PCR was performed on total RNA from leaves or roots using the following oligonucleotides: CHS1, PD1270/PD1271; CHL1, PD1272/PD1273; UBI10, UBI-F/UBI-R (Table S1). The promoter region (1500 bp 5' of the start codon) of CHS1 was amplified by PCR from Col genomic DNA using the primers PD1098/PD1099, and cloned into pENTR/SD/D-TOPO (Invitrogen). Total leaf RNA was separated by agarose gel electrophoresis and blotted onto nylon membranes. Expression of defense-related genes was monitored by Northern blot hybridization using 32P-dCTP labeled cDNAs. The CHS1 promoter fragment was transferred into pKGWFS7 (Invitrogen) as a transcriptional fusion with the β-glucuronidase (GUS) gene. After transformation into Col plants, GUS activity was visualized using 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid cyclohexylammonium (Bowling et al., 1994). The sequence comparison was done with ClustalW (Chenna et al., 2003) of the Lasergene software package (DNASTAR, www.dnastar.com). BLAST (Basic Local Alignment Search Tool) searches were done with the protein database of the National Center for Biotechnology Information (NCBI) according to Altschul et al. (1990).
Western blot analysis
Various cellular fractions were prepared from leaf tissues (Price et al., 1994; Kinkema et al., 2000). Proteins were separated by SDS–PAGE and immunoblotted on nitrocellulose membranes with antibodies. Polyclonal antiserum against the CHS1 polypeptide sequence CGIKAFKSESWKESS was obtained from rabbits (BioGenes, www.biogenes.de). To monitor cell fractions, anti-histone 3 (H3) (nuclei), anti-24-sterol C-methyltransferase 1 (SMT1) (microsomes), anti-UDP-glucose pyrophosphorylase (UGPase) (cytosol) and anti-D1 protein (PsbA) (chloroplasts) antibodies were used (Agrisera, www.agrisera.com). Immunoreactions were detected by chemiluminescence using peroxidase-labeled secondary antibodies. The CHS1 coding sequence was amplified by PCR using primers PD1057 and PD1058 (Table S1), and ligated into the BamHI and EcoRI sites of pYES2 (Invitrogen), and the CHS1 construct was introduced into Saccharomyces cerevisiae H1246 for expression in yeast. CHS1 expression was induced by galactose, and protein extracts were used for Western blot analysis.
Bacterial growth assays
Pseudomonas syringae strain DC3000 pv. tomato (Pst) and Pst/AvrRps4 strains (Gassmann et al., 1999) were spray-inoculated onto 3-week-old plants, and bacterial growth assays were performed at 0 dpi (3 h after spraying) and 3 dpi (Birker et al., 2009).
Biochemical measurements, determination of chlorophyll fluorescence and electrolyte leakage
Total SA was measured after hydrolysis with HCl by fluorescence HPLC (Nawrath and Métraux, 1999). Chlorophyll was measured photometrically. Carotenoid contents were determined by HPLC using a UV detector (Thayer and Björkman, 1990). Tocopherol was quantified by fluorescence HPLC using tocol as an internal standard (Zbierzak et al., 2010). Chlorophyll fluorescence was recorded on leaves of plants that were dark-adapted for at least 30 min prior to measurements using a pulse amplitude modulation fluorometer (Junior PAM, Heinz Walz, www.walz.com), and the quantum yield was calculated (Schreiber et al., 1986). Electrolyte leakage from leaf discs was measured as described by Griebel and Zeier (2010). Lipids were extracted from leaves to separated by thin-layer chromatography (Dörmann et al., 1995). Fatty acid methyl esters were prepared from lipids in whole leaves and quantified by gas chromatography (Browse et al., 1986).
Leaf ultra-thin sections were analyzed by electron microscopy on a Tecnai Sphera transmission electron microscope (FEI, www.fei.com) (Hölzl et al., 2009).
We thank Regina Wendenburg (Max Planck Institute of Molecular Plant Physiology, Potsdam, Germany), Le Thi Mai Trang (Institute of Molecular Physiology and Biotechnology of Plants, University of Bonn, Germany) and Jaqueline Bautor (Max Planck Institute for Plant Breeding Research, Cologne, Germany) for physiological measurements. This work was supported by the University of Bonn (P.D. and A.M.Z.), the Max Planck Society (T.G. and J.E.P.), Deutsche Forschungsgemeinschaft SPP1212 ‘Plant–Microbe’ grants (P.D. and J.E.P.) and a Deutsche Forschungsgemeinschaft SFB670 grant (J.E.P.).