Roles of CAMTA transcription factors and salicylic acid in configuring the low-temperature transcriptome and freezing tolerance of Arabidopsis

Authors


For correspondence (e-mail thomash6@msu.edu).

Summary

Previous studies in Arabidopsis thaliana established roles for CALMODULIN BINDING TRANSCRIPTION ACTIVATOR 3 (CAMTA3) in the rapid cold induction of CRT/DRE BINDING FACTOR (CBF) genes CBF1 and CBF2, and the repression of salicylic acid (SA) biosynthesis at warm temperature. Here we show that CAMTA1 and CAMTA2 work in concert with CAMTA3 at low temperature (4°C) to induce peak transcript levels of CBF1, CBF2 and CBF3 at 2 h, contribute to up-regulation of approximately 15% of the genes induced at 24 h, most of which fall outside the CBF pathway, and increase plant freezing tolerance. In addition, CAMTA1, CAMTA2 and CAMTA3 function together to inhibit SA biosynthesis at warm temperature (22°C). However, SA levels increase in Arabidopsis plants that are exposed to low temperature for more than 1 week. We show that this chilling-induced SA biosynthesis proceeds through the isochorismate synthase (ICS) pathway, with cold induction of ICS1 (which encodes ICS), and two genes encoding transcription factors that positively regulate ICS1CBP60g and SARD1 –, paralleling SA accumulation. The three CAMTA proteins effectively repress the accumulation of ICS1, CBP60g and SARD1 transcripts at warm temperature but not at low temperature. This impairment of CAMTA function may involve post-transcriptional regulation, as CAMTA transcript levels did not decrease at low temperature. Salicylic acid biosynthesis at low temperature did not contribute to freezing tolerance, but had a major role in configuring the transcriptome, including the induction of ‘defense response’ genes, suggesting the possible existence of a pre-emptive defense strategy programmed by prolonged chilling temperatures.

Introduction

Arabidopsis thaliana (hereafter referred to as Arabidopsis) and other plants that live in temperate regions are able to sense low temperature and activate both rapid and delayed responses that enable them to successfully inhabit their environment. An important rapid response is the activation of pathways that increase freezing tolerance. Exposure of Arabidopsis plants to low non-freezing temperatures for 1–2 days results in a detectible increase in freezing tolerance that reaches a maximum in approximately 1–2 weeks (Gilmour et al., 1988; Kurkela et al., 1988). A key regulatory pathway that conditions freezing tolerance in Arabidopsis and other plants is the CRT/DRE Binding Factor (CBF) pathway (Thomashow, 2010; Knight and Knight, 2012). The pathway comprises three closely related members of the AP2/ERF family of DNA-binding proteins – CBF1, CBF2 and CBF3 (also known as DREB1B, DREB1C and DREB1A, respectively) – and CBF-regulated genes, referred to as the CBF regulon (Stockinger et al., 1997; Gilmour et al., 1998; Jaglo-Ottosen et al., 1998; Liu et al., 1998; Kasuga et al., 1999). CBF1, CBF2 and CBF3 are induced within 15 min of transferring plants from warm to cold temperatures (Gilmour et al., 1998; Liu et al., 1998). The resulting CBF proteins bind to the CRT/DRE regulatory element (rCCGAC) that is present in the promoters of approximately 100 target genes and induce their expression (Fowler and Thomashow, 2002; Sakuma et al., 2002; Maruyama et al., 2004; Vogel et al., 2005). Increased transcript levels for genes that comprise the CBF regulon were detected after plants were exposed to low temperature for 2–3 h, and reached a maximum level at approximately 24 h (Gilmour et al., 1998; Liu et al., 1998). Expression of the CBF regulon activates multiple mechanisms that enhance freezing tolerance, including the biosynthesis of cryoprotective sugars and proteins (Steponkus et al., 1998; Cook et al., 2004; Kaplan et al., 2007).

Our understanding of how plants sense low temperature and process the information to rapidly induce transcription of CBF1, CBF2 and CBF3 is limited, although it is clear that this regulation is complex, involving the action of multiple transcription factors. Some of these factors act as repressors, including MYB15 (Agarwal et al., 2006) and the phytochrome-interacting factors PIF4 (Lee and Thomashow, 2012) and PIF7 (Kidokoro et al., 2009; Lee and Thomashow, 2012), while others act as activators, including the MYB transcription factor ICE1 (Chinnusamy et al., 2003; Miura et al., 2007) and the circadian regulatory factors CCA1 and LHY (Dong et al., 2011). In addition, Doherty et al. (2009) have shown that rapid induction of CBF1 and CBF2 involves the action of CALMODULIN BINDING TRANSCRIPTION ACTIVATOR 3 (CAMTA3).

There are six members of the CAMTA protein family, of which three – CAMTA1, CAMTA2 and CAMTA3 – are the most closely related (Finkler et al., 2007). Doherty et al. (2009) reported that a camta3 loss-of-function mutation resulted in an approximately 50% reduction in transcript levels for CBF1 and CBF2 in plants exposed to low temperature for 2 h. Detailed analysis of the CBF2 promoter indicated that CAMTA3 binds to a CAMTA DNA regulatory motif, vCGCGb, located in a region of the promoter that is known to positively regulate CBF2. Doherty et al. (2009) thus proposed that CAMTA3 provides the missing link between the long-known rapid spike in the cytoplasmic calcium level that occurs in response to low temperature (Knight et al., 1991), and the induction of at least some cold-regulated genes (Monroy and Dhindsa, 1995; Knight et al., 1996). The simple model proposed was that the increase in cytoplasmic calcium produced by exposing plants to low temperature results in higher levels of calcium-bound calmodulin (or calmodulin-like protein), and that the interaction of calcium-bound calmodulin with CAMTA3 modifies CAMTA3 activity, resulting in rapid induction of CBF1 and CBF2. Although the camta3 mutation did not affect freezing tolerance, a camta1 camta3 (camta1/3) double mutation did, providing evidence that CAMTA1 and CAMTA3 act together to condition freezing tolerance.

In contrast to the rapid cold induction of the CBF pathway, the biosynthesis of salicylic acid (SA) in response to low temperature is a delayed response. Scott et al. (2004) first showed that SA levels increase in Arabidopsis plants at chilling temperatures (5°C), but not until plants had been exposed to low temperature for more than 1 week. To explore the role of SA biosynthesis at low temperature, the investigators compared wild-type (WT) plants with transgenic plants expressing NahG (a bacterial gene encoding a salicylate hydroxylase that converts SA to catechol), and found that SA had little, if any, effect on net carbon assimilation rates or the photochemical efficiency of photosystem II. However, they showed that SA inhibited plant growth rate and reduced cell size.

The pathway responsible for SA biosynthesis at low temperature, and its regulation by temperature, were not determined by Scott et al. (2004). However, Du et al. (2009) recently showed that CAMTA3 (referred to as AtSR1 by Du et al.) is a repressor of SA biosynthesis in Arabidopsis plants grown at warm temperature. These results raise the possibility that CAMTA3 is not only involved in rapid changes in gene expression at low temperature, but may also have roles in delayed low-temperature responses that include regulation of genes involved in SA biosynthesis. Here we test this hypothesis and further explore the roles of CAMTA transcription factors and SA in configuring the low-temperature transcriptome and conditioning freezing tolerance.

Results

SA biosynthesis at low temperature proceeds through the ICS pathway

Our first objective was to determine whether SA biosynthesis at low temperature proceeds through the isochorismate synthase (ICS) pathway, the primary pathway responsible for SA biosynthesis in response to pathogen attack (Dempsey et al., 2011). The pathway comprises two steps: chorismate is converted to isochorismate by ICS, and isochorismate is converted to SA by isochorismate pyruvate lyase. In Arabidopsis, ICS is encoded by two genes, ICS1 and ICS2, with ICS1 having the primary role in SA biosynthesis in plant–pathogen interactions (Wildermuth et al., 2001). We therefore compared the levels of SA and SA glucosides (SAGs) in WT plants with those in mutant plants carrying the sid2–1 mutation (Wildermuth et al., 2001), a loss-of-function allele of ICS1. When WT plants were exposed to low temperature (4°C), SA and SAG levels remained relatively low for the first week, followed by a marked increase between weeks 1 and 2 (Figure 1a,b). In contrast, SA and SAG levels remained low in the sid2–1 plants throughout the experiment. These results indicated that most, if not all, of the SA biosynthesis that occurs at low temperature proceeds through the ICS pathway and requires the action of ICS1.

Figure 1.

SA and SAG levels and ICS1, CBP60g and SARD1 transcript levels increase in plants exposed to low temperature for more than 1 week. (a, b) SA (a) and SAG (b) levels in WT and sid2–1 plants grown at 22°C and transferred to 4°C for the indicated times. SA and SAG levels were determined by LC/MS/MS. (c) WT plants are smaller than sid2–1 and NahG mutant plants grown at 4°C for 5 months. (d–f) ICS1 (d), CBP60g (e) and SARD1 (f) transcript levels in WT plants grown at 22°C and transferred to 4°C for the indicated times. Transcript levels were determined by quantitative RT–PCR. Error bars indicate the standard deviation for triplicate samples from a representative experiment (a, b) or triplicate samples from each of three independent experiments (d–f).

Scott et al. (2004) found that transgenic Arabidopsis plants expressing NahG were larger than WT plants after the plants had been exposed to low temperature for multiple weeks, suggesting that SA inhibited the growth of Arabidopsis at low temperature. Consistent with this interpretation was our finding that the sid2–1 plants were considerably larger than the WT plants after prolonged exposure to low temperature (Figure 1c). Plants grown at warm temperature did not differ in size.

A simple model to explain the increase in SA and SAG levels at low temperature is that ICS1 is induced in response to chilling temperatures. Indeed, ICS1 transcript levels increased dramatically between 1 and 2 weeks after the start of exposure of plants to low temperature (Figure 1d). In addition, the transcript levels for two genes encoding transcription factors that act as positive regulators of ICS1, CBP60g and SARD1 (Wang et al., 2009, 2011; Zhang et al., 2010) increased between 1 and 2 weeks after the start of cold treatment (Figure 1e,f). These results indicated that the increase in SA that occurs in cold-treated plants results from delayed induction of ICS1, and that this probably involves delayed cold induction of CBP60g and SARD1.

SA biosynthesis and expression of ICS1 are repressed by CAMTA1, CAMTA2 and CAMTA3 at warm temperature

Du et al. (2009) found that Arabidopsis plants carrying a camta3 loss-of-function mutation had higher levels of SA and SAG than did WT plants. Our results confirmed this finding: the levels of SA and SAG were higher in camta3 plants than in WT plants when plants were grown at warm temperature (22°C; Figure 2a,b; see 'Experimental procedures' for a description of the camta T–DNA insertion mutants used in this study). Given the close phylogenetic relationships between CAMTA1, CAMTA2 and CAMTA3 (Finkler et al., 2007), we hypothesized that these three CAMTA proteins may function together to repress SA biosynthesis at warm temperature. Indeed, we found that combining either a camta1 or camta2 mutation with the camta3 mutation resulted in plants that had higher SA and SAG levels (Figure 2a,b), with the camta2 camta3 (camta2/3) combination having a greater effect than the camta1 camta3 (camta1/3) combination, and the camta1 camta2 camta3 (camta1/2/3) triple mutation having the greatest effect.

Figure 2.

The SA and SAG levels and ICS1, CBP60g and SARD1 transcript levels are higher in camta mutant plants than in WT plants when they are grown at warm temperature. (a, b) SA (a) and SAG (b) levels in WT and camta mutant plants grown at 22°C. SA and SAG levels were determined by LC/MS/MS. (c–e) ICS1 (c), CBP60g (d) and SARD1 (e) transcript levels in WT and camta mutants grown at 22°C. Transcript levels were determined by quantitative RT–PCR. (f) WT and camta mutants grown at 22°C. The white arrow indicates the camta1/2/3 triple mutant plant. (g) Mean fresh weight of the aerial portion of plants shown in (f). Error bars indicate the standard deviation for triplicate samples from a representative experiment (a, b), triplicate samples from three independent experiments (c–e), or samples from four or five individual plants (g). Data were subjected to statistical analysis using a two-tailed (a, b, g) or one-tailed (c–e) Student's t test (< 0.05). Different letters indicate statistically significant differences when compared to WT and each other.

Our finding that SA biosynthesis at low temperature occurred through the ICS pathway, and that ICS1, CBP60g and SARD1 were induced at low temperature, prompted us to determine whether these three genes were repressed by the three CAMTA proteins in plants grown at warm temperature (22°C). Our results indicated that they were, as the transcript levels for all three genes were greater in the camta mutants than they were in WT plants (Figure 2c–e). It is possible that this repression was direct, as the promoter regions of ICS1, CBP60g and SARD1, respectively, were found to have five, nine and five CAMTA DNA-binding motifs (vCGCGb and vCGTGb; Galon et al., 2010b; Figure S1).

Taken together, our results imply that repression of ICS1, CBP60g and SARD1 by CAMTA1, CAMTA2 and CAMTA3 is impaired in plants that had been exposed to low temperature for more than 1 week. This decrease in CAMTA repressive activity may have resulted from the CAMTA genes being down-regulated by low temperature. However, the transcript levels for CAMTA1 and CAMTA2 did not change appreciably during exposure of plants to low temperature for 5 weeks, and the transcript levels for CAMTA3 increased (Figure 3). Thus, the decrease in CAMTA repression activity at low temperature appears to involve post-transcriptional regulation.

Figure 3.

CAMTA1, CAMTA2 and CAMTA3 are expressed in WT plants exposed to low temperature for up to 5 weeks. Plants were grown at 22°C, transferred to 4°C for the indicated times, and the CAMTA transcript levels were determined by quantitative RT–PCR. Error bars represent the standard error for triplicate samples from each of three independent experiments.

Du et al. (2009) reported that camta3 mutant plants were smaller than WT plants and that this was due to the elevated SA levels in the camta3 mutant plants. Our results are consistent with these findings. Salicylic acid levels were at similar low levels in WT plants, camta1 and camta2 single mutant plants, and camta1 camta2 (camta1/2) double mutant plants when grown at warm temperature (Figure 2a), and the sizes and fresh weights of these plants were very similar (Figure 2f,g). The SA and SAG levels were progressively greater in the camta3, camta1/3, camta2/3 and camta1/2/3 mutant plants (Figure 2a,b), and their sizes and fresh weights were progressively smaller (Figure 2f,g).

SA has a major role in configuring the low-temperature transcriptome

Salicylic acid signaling affects the expression of hundreds of genes during plant–pathogen interactions (Katagiri, 2004; Wang et al., 2006; Blanco et al., 2009). Thus, it is possible that many of the genes that were up-regulated in plants exposed to low temperature for more than 1 week were actually induced in response to SA signaling rather than by low temperature per se. To test this, we compared the changes in gene expression that occurred in WT and sid2–1 plants upon exposure to low temperature for 3 weeks (Figure 4a and Table S1). We found that 989 genes were induced twofold or greater (FDR ≤ 0.05) in cold-treated WT plants, and that, of these genes, 264 (27%) showed impaired cold induction in sid2–1 plants (induction was reduced between 1.5- and 90-fold; Figure 4a, blue bar, and Table S2). These SA-induced genes were strongly up-regulated in camta1/2/3 plants grown at warm temperature (Figure 4a), a finding that was consistent with the high levels of SA in the triple mutant plants (the effects of the camta mutations on global gene expression are detailed below). As expected for SA-induced genes, the group of 264 genes was highly enriched in genes associated with SA signaling, including ‘defense response’, ‘innate immune response’ and ‘response to salicylic acid stimulus’ (Figure 4b). In addition, the promoters of these genes were highly enriched in the TGA transcription factor-binding site, TGACG (Figure 4c), a result that was consistent with the known role of TGA transcription factors in regulating the expression of genes downstream of SA (An and Mou, 2011; Moore et al., 2011). The promoters of these genes were also enriched in the vCGCGb CAMTA-binding site, suggesting that the CAMTA proteins may directly repress some of the SA-induced genes at warm temperature. The promoters were not enriched in the CBF-binding site, rCCGAC, suggesting that they are not part of the CBF pathway.

Figure 4.

SA has a major role in configuring the transcriptome in plants exposed to low temperature for 3 weeks. (a) Hierarchical clustering analysis of genes that were differentially expressed in WT and sid2–1 mutant plants exposed to 4°C for 3 weeks (cold), and in camta1/2/3 and sid2–1 plants grown at 22°C (warm). Each sample was compared to WT plants grown at 22°C. The blue bar indicates SA-induced genes. The color scale represents log2 fold change. (b) Enrichment of GO categories in the SA-induced genes. P values represent significant differences between the percentage of all genes (‘ATH1’) assigned to the GO category and the percentage of SA-induced genes (‘Target’) assigned to the GO category. A cut-off value of ≤ 10−4 was used. (c) Enrichment of DNA regulatory motifs (cis-elements) in the promoters of the SA-induced genes. The motifs for CAMTA (vCGCGb and vCGTGb), TGA (TGACG) and CBF (CCGAC and rCCGAC) transcription factors were tested. Z scores and P values are based on over-representation of the motif in the promoters of the SA-induced genes (occurrence per gene) compared to its occurrence in all genes (expected per gene).

SA biosynthesis at low temperature does not contribute to freezing tolerance

To determine whether the biosynthesis of SA at low temperature contributes to freezing tolerance, we exposed WT and sid2–1 plants to low temperature for 3 weeks and compared their freezing tolerance using the electrolyte leakage assay. The results indicated that the EL50 temperature (the temperature at which freezing damage results in leakage of 50% of the total cellular electrolytes) for both WT and sid2–1 plants grown at warm temperature (22°C) was approximately −4°C, and that it decreased to approximately −9°C for both WT and sid2–1 plants exposed to low temperature (4°C) for 3 weeks (Figure 5). Thus, SA biosynthesis at low temperature did not contribute to the increase in freezing tolerance under the conditions tested.

Figure 5.

Salicylic acid biosynthesis does not contribute to the increase in freezing tolerance that occurs in response to low temperature. WT and sid2–1 mutant plants were grown at 22°C (non-acclimated, NA) and then transferred to 4°C (cold-acclimated, CA) for 3 weeks. Freezing tolerance of the NA and CA plants was determined using the electrolyte leakage assay. Error bars represent the standard deviation of three replicate samples from a representative experiment.

CAMTA1, CAMTA2 and CAMTA3 function together to regulate nearly two thousand genes at warm temperature

The results presented above indicate that CAMTA1, CAMTA2 and CAMTA3 work in concert to inhibit SA biosynthesis at warm temperature by repressing the expression of genes associated with SA biosynthesis. To further explore the roles of CAMTA1, CAMTA2 and CAMTA3 in regulating gene expression at warm temperature (22°C), we determined the effects of the camta mutations on global gene expression. Previous studies showed that single camta1 (Galon et al., 2010a) and camta3 (Galon et al., 2008) mutations affected the expression of 128 and 105 genes, respectively, in plants grown at warm temperature. Our results indicated that the camta1/2/3 triple mutation affected the expression of 1915 genes (≥ twofold change; FDR < 0.01; Figure 6a and Table S3). Of these, 1086 were up-regulated in the camta triple mutant, indicating that they were repressed by action of the CAMTA transcription factors at warm temperature. In addition, 829 genes were down-regulated in the camta triple mutant, indicating that they were induced by the CAMTA transcription factors at warm temperature. With the exception of only a few genes, each camta double mutant affected expression of a subset of the genes affected by the camta triple mutant, with the camta2/3 double mutation having the greatest effect, followed by the camta1/3 mutation and then the camta1/2 mutation, results that paralleled the relative effects of the double mutations on SA levels, and plant size and weight (Figure 2). However, a heatmap prepared from these data indicated that the patterns of gene expression for all of the mutant lines were very similar except for the degree of fold change (Figure 6b). Thus, the ‘subsets’ of genes that were defined for each camta double mutation were primarily a function of the arbitrary cut-offs used to determine whether a gene was differentially regulated. Therefore, we concluded that CAMTA1, CAMTA2 and CAMTA3 act largely in a redundant fashion to regulate the expression of almost 2000 genes in plants grown at warm temperature.

Figure 6.

CAMTA1, CAMTA2 and CAMTA3 function together to regulate the expression of almost 2000 genes in plants grown at warm temperature. (a) Venn diagrams indicating the number of genes up-regulated or down-regulated at least twofold in camta double and triple mutant plants grown at 22°C compared to WT plants grown at 22°C. (b) Hierarchical clustering analysis of the results presented in (a). (c) Enrichment of GO categories for the genes that were differentially regulated in the camta1/2/3 triple mutant plants grown at 22°C. P values represent significant differences between the percentage of all genes (‘ATH1’) assigned to the GO category compared to the percentage of camta1/2/3-regulated genes (‘Target’) assigned to the GO category. A cut-off value of ≤ 10−4 was used. (d) Enrichment of DNA regulatory motifs (cis-elements) in the promoters of the camta1/2/3-regulated genes. The motifs for CAMTA (vCGCGb and vCGTGb), TGA (TGACG) and CBF (CCGAC and rCCGAC) transcription factors were tested. Z scores and P values are based on over-representation of the motif in the promoters for the camta1/2/3-regulated genes (occurrence per gene) compared to the occurrence in all genes (expected per gene).

The 1086 genes that were up-regulated in the camta triple mutant plants were highly enriched in GO terms relating to SA signaling and pathogen defense, including ‘defense response’, ‘innate immune response’ and ‘cell death’ (Figure 6c), a finding that was consistent with the high levels of SA in these plants. Also consistent with high SA levels was the induction of known positive regulators of SA biosynthesis, e.g. EDS1, EDS5, FMO1, PAD4, PBS3, WRKY28, CBP60g and SARD1, and genes that are known to be induced downstream of SA signaling, e.g. NIMIN1, NPR1, NPR3, NPR4, PR1 and TGA5 (Dempsey et al., 2011; Moore et al., 2011; Spoel and Dong, 2012; Table S3). The promoters of the 1086 up-regulated genes were highly enriched in the TGA transcription factor binding site, TGACG (Figure 6d), and were moderately enriched for the vCGTGb CAMTA DNA-binding site, suggesting that the CAMTA proteins directly repress transcription of some of these genes.

The 829 genes that were down-regulated in the camta triple mutant plants were highly enriched in GO terms relating to a wide range of biosynthetic processes, including those for carboxylic acids, fatty acids, amines, glycosinolates and carbohydrates (Figure 6c). In addition, the group of 829 genes was enriched in genes associated with jasmonic acid (JA) signaling (i.e. ‘response to jasmonic acid’ and ‘response to wounding’), consistent with the known antagonistic relationship between SA and JA signaling (Thaler et al., 2012). The promoters of the 829 down-regulated genes were highly enriched in the CAMTA-binding site vCGTGb and the CBF-binding site rCCGAC (Figure 6d), suggesting that the CAMTA transcription factors up-regulate the CBF regulatory pathway at warm temperature.

CAMTA1, CAMTA2 and CAMTA3 function together to positively regulate CBF1, CBF2 and CBF3 and freezing tolerance

Doherty et al. (2009) established a role for CAMTA3 in the rapid cold induction of both CBF1 and CBF2. However, they unexpectedly found that, whereas the camta1/3 double mutation resulted in impaired cold induction of CBF1, it had little effect on cold induction of CBF2. Here, we further explored the roles of CAMTA1, CAMTA2 and CAMTA3 in cold induction of CBF1, CBF2 and CBF3. The results indicated that each camta double mutation reduced the induction of all three CBF genes in plants exposed to low temperature for 2 h, and that the camta2/3 mutation had the largest effect, reducing induction levels by 80–90% (Figure 7a–c). We do not know the basis for the difference between our results and those of Doherty et al. (2009) regarding the effects of the camta1/3 double mutation on CBF2 induction, but it may be due to the different growth conditions used; whereas Doherty et al. (2009) grew plants on solidified growth medium under constant light, we grew plants in soil under a 12 h photoperiod. Regardless, our results indicate that, under the growth conditions tested here, CAMTA1, CAMTA2 and CAMTA3 act additively to rapidly induce the expression of CBF1, CBF2 and CBF3 in response to low temperature.

Figure 7.

CAMTA1, CAMTA2 and CAMTA3 contribute to rapid cold induction of CBF1, CBF2 and CBF3 and freezing tolerance. (a–c) Induction of CBF1 (a), CBF2 (b) and CBF3 (c) in WT and camta double mutant plants grown at 22°C and exposed to 4°C for 2 h. Transcript levels were measured by quantitative RT–PCR. (d, e) WT and camta double mutant plants were grown at 22°C (NA) (d) and transferred to 4°C (CA) (e) for 3 weeks, and their freezing tolerance was determined using the electrolyte leakage assay. Error bars indicate the standard deviation of triplicate samples for each of two independent experiments (a–c) or three replicate samples from a representative experiment (d, e). Data were subjected to statistical analysis using a two-tailed Student's t test (< 0.05). Different letters indicate statistically significant differences when compared to WT and each other.

Doherty et al. (2009) showed that camta1/3 double mutant plants were impaired in their ability to acclimate to cold. Our results confirmed this result; whereas both non-acclimated WT plants and camta1/3 double mutant plants had EL50 values of approximately −4.5°C (Figure 7d), the EL50 values of these plants after 3 weeks of cold acclimation were −11°C and −8°C, respectively (Figure 7e). In addition, we found that both the camta1/2 and camta2/3 double mutant plants were also impaired in chilling-induced freezing tolerance. The camta2/3 double mutation also appeared to slightly reduce the freezing tolerance of non-acclimated plants, a result that is consistent with the finding presented above indicating that the CBF pathway was down-regulated in camta1/2/3 triple mutant plants grown at warm temperature.

CAMTA1, CAMTA2 and CAMTA3 positively regulate many early cold-induced genes that fall outside the CBF pathway

To further explore the role of CAMTA1, CAMTA2 and CAMTA3 in the cold induction of early cold-responsive genes, we exposed WT plants and camta1/2/3 triple mutant plants to low temperature for 24 h and compared their transcriptomes at a global level. The results indicate that, in WT plants, 849 genes were induced at least twofold (FDR < 0.01) at 24 h (Figure 8a and Table S4), and that, of these, 128 (15%) showed noticeably reduced (1.6–17-fold) induction in the camta triple mutant plants (Figure 8a, bar B1, and Table S5). The promoters of these CAMTA-induced genes were highly enriched in the vCGTGb CAMTA-binding site and moderately enriched in the vCGCGb CAMTA-binding site (Figure 8c), indicating that many of these genes may be direct targets of the CAMTA proteins. These promoters were also highly enriched in the CBF-binding site, rCCGAC, but only nine of these genes (Table S6) were defined as members of the CBF regulon by Vogel et al. (2005). Thus, the vast majority of the early cold-induced genes that are CAMTA-regulated appear to fall outside the CBF response pathway. GO enrichment analysis indicated that the group of 128 CAMTA-induced genes was only slightly enriched in genes involved in ‘response to gibberellin stimulus’ (Figure 8b), suggesting that they vary considerably in functional class.

Figure 8.

CAMTA1, CAMTA2 and CAMTA3 have a substantial role in the regulation of genes exposed to low temperature for 24 h. (a) Hierarchical clustering of genes that were differentially expressed in WT and camta triple mutant plants treated for 24 h at 4°C compared to WT grown at 22°C. Genes that were impaired in cold induction (blue bar B1) or cold repression (blue bar B2) in the camta1/2/3 mutant plants are indicated. (b) Enrichment of GO categories for the genes that were impaired in cold regulation in the camta1/2/3 triple mutant plants. P values represent a significant difference between the percentage of all genes (‘ATH1’) assigned to the GO category compared to the percentage of B1 and B2 cluster genes (‘Target’) assigned to the GO category. A cut-off value of ≤ 10−2 was used. (c) Enrichment of DNA regulatory motifs (cis-elements) in the promoters of the genes impaired in cold regulation. The motifs for CAMTA (vCGCGb and vCGTGb), TGA (TGACG) and CBF (CCGAC and rCCGAC) transcription factors were tested. Z scores and P values are based on over-representation of the motif in the promoters for the impaired B1 and B2 class genes (occurrence per gene) compared to the occurrence in all genes (expected per gene).

In regard to early cold-repressed genes, 1025 were down-regulated at least twofold (FDR < 0.01) at 24 h in WT plants (Figure 8a and Table S4), and, of these, 136 (13%) showed substantially reduced (1.7–30-fold) repression in the camta triple mutant plants (Figure 8a, bar B2, and Table S5). The promoters of the 136 CAMTA-repressed genes were not enriched in CAMTA DNA-binding sites (Figure 8c), and thus are not likely to be directly repressed by the CAMTA proteins. However, these genes were enriched in the GO functional categories ‘defense response’ and ‘cell death’ (Figure 8b), and thus may be repressed by the high levels of SA in the camta triple mutant plants.

Discussion

Exposure of Arabidopsis to low non-freezing temperatures initiates a regulatory program that produces changes in gene expression that affect multiple aspects of plant growth and development. Some of these changes in gene expression occur within minutes of exposing plants to low temperature, whereas others are delayed for days or weeks. Here we show that CAMTA1, CAMTA2 and CAMTA3 function together to regulate both rapid and delayed changes in gene expression at low temperature. Doherty et al. (2009) established a role for CAMTA3 in regulating the CBF pathway by contributing to the rapid cold induction of CBF1 and CBF2. Here we extend these findings to show that CAMTA1 and CAMTA2 function in concert with CAMTA3 to rapidly induce expression of all three CBF genes (Figure 7). Moreover, we show that the three CAMTA proteins contribute to induction of approximately 15% of the genes that are cold-induced at 24 h (Figure 8). The vCGTGb and vCGCGb CAMTA-binding sites are highly enriched in the promoters of these early cold-induced genes, suggesting a direct role for the CAMTA transcription factors in their regulation. In addition, the promoters of these genes are highly enriched in the CBF-binding site, rCCGAC, a result that is consistent with the CAMTA proteins playing an important role in regulating the CBF pathway and conditioning freezing tolerance. Indeed, we found that plants carrying each of the three combinations of camta double mutations were impaired in freezing tolerance (Figure 7). However, of the 128 early cold-induced genes that were CAMTA-regulated, only nine were identified as members of the CBF regulon by Vogel et al. (2005). Thus, most of these genes appear to fall outside the CBF pathway. From these results, we conclude that the CAMTA transcription factors, presumably in conjunction with calcium signaling, have an appreciable role in rapidly reconfiguring the transcriptome in Arabidopsis plants in response to low temperature.

Our results indicate that CAMTA1, CAMTA2 and CAMTA3 also have roles in ‘delayed’ cold-regulated gene expression. In this case, the effect is largely mediated through regulation of SA biosynthesis. Du et al. (2009) established that CAMTA3 represses the biosynthesis of SA in non-stressed plants grown at warm temperature. Our results confirm this finding and extend it by showing that the repression of SA biosynthesis involves the action of all three CAMTA proteins (Figure 2). However, this negative regulation is at least partially overcome when plants are exposed to low temperature for more than 1 week (Figure 1). At that time point, SA levels increase, SA signaling is engaged, and the transcriptome is substantially modified; approximately 27% of the genes that are up-regulated in plants exposed to low temperature for 3 weeks involve SA signaling (Figure 4). As expected, the SA-induced genes were highly enriched in GO terms related to SA signaling and defense response (Figure 4). In addition, the promoters of these genes were highly enriched in the TGA transcription factor binding site (Figure 4), a finding that is consistent with the known roles of TGA transcription factors in SA-regulated gene expression (An and Mou, 2011; Moore et al., 2011).

What is the nature of the regulatory mechanism that underlies the delayed biosynthesis of SA? Our results provide a partial answer to this question. We show that SA biosynthesis at low temperature proceeds through the ICS pathway, that this requires action of the ICS1 gene, and that the ICS1 gene is induced in plants that are exposed to low temperature for more than 1 week (Figure 1). We also found that cold induction of ICS1 is paralleled by cold induction of CBP60g and SARD1 (Figure 1), genes that encode closely related transcription factors that bind to the ICS1 promoter and act as positive regulators (Wang et al., 2009, 2011; Zhang et al., 2010). Presumably, cold induction of ICS1 results from cold induction of SARD1 and CBP60g. But how are SARD1 and CBP60g induced in response to low temperature? Our results indicate that this involves impaired repression by CAMTA1, CAMTA2 and CAMTA3 at low temperature, as the transcript levels for CBP60g and SARD1 at warm temperature increased approximately 6- and 14-fold respectively, in response to the camta1/2/3 triple mutation (Figure 2). Given that the promoters for SARD1 and CBP60g have five and nine CAMTA DNA-binding sites, respectively, it is possible that the CAMTA proteins directly repress transcription of these genes at warm temperature.

How is CAMTA repression of CBP60g and SARD1 reduced or eliminated at low temperature? Our results indicate that it does not result from repression of the CAMTA genes (Figure 3). Thus, it appears that the repressive activities of CAMTA1, CAMTA2 and CAMTA3 are impaired by a post-transcriptional mechanism. This mechanism remains to be determined, but the results of Du et al. (2009) are instructive in this regard. Du et al. (2009) reported that the EDS1 gene, which encodes a positive regulator of SA biosynthesis (An and Mou, 2011; Moore et al., 2011), is repressed by CAMTA3 at warm temperature in non-stressed plants. It was shown that CAMTA3 binds directly to the EDS1 promoter, and that the repression requires interaction of CAMTA3 with calmodulin. The investigators proposed that repression of EDS1 in non-stressed plants is overcome by transient changes in intracellular calcium that are known to occur in response to plant–pathogen interactions (Boudsocq et al., 2010; Moore et al., 2011). Perhaps a similar calcium signature is established upon exposing plants to low temperature for more than 1 week, resulting in reduced CAMTA1, CAMTA2 and CAMTA3 repression of EDS1, as well as CBP60g and SARD1. However, if this is true, the signature must be different from the rapid spike in intracellular calcium that occurs in response to low temperature (Knight et al., 1991, 1996), as EDS1, CBP60g and SARD1 remain repressed during the early stages of exposing plants to low temperature.

There is a final point to make regarding the function of SA biosynthesis at low temperature. Our results indicate that SA biosynthesis does not contribute to an increase in freezing tolerance (Figure 5). However, the group of genes that were induced by SA in plants exposed to low temperature for 3 weeks was highly enriched in genes with roles in ‘defense response’ and ‘innate immune response’ (Figure 4). Thus, as occurs with SA signaling at warm temperature, it is probable that the genes induced by SA at low temperature increase resistance to pathogen attack. Consistent with this notion is that an increase in disease resistance has been shown to occur in certain plants during cold acclimation (Kuwabara and Imai, 2009). The selective advantage of having plant defenses that are programmed to become activated in response to prolonged chilling temperatures remains open for speculation, but an interesting possibility is that it is a pre-emptive response akin to the increase in freezing tolerance that rapidly occurs in response to chilling temperatures. Plants that acclimate to cold ‘interpret’ chilling temperatures as a forerunner of potential freezing temperatures, and pre-emptively activate mechanisms that increase freezing tolerance. This increase helps plants to survive freezing. However, freeze-induced injury may still occur, which may increase the susceptibility of the plants to pathogen infection. Perhaps waiting to activate defense mechanisms until there is a pathogen attack, at a time when plants are at low temperature and potentially recovering from injury, is a ‘too little, too late’ strategy that has been out-competed by a pre-emptive strategy. Indeed, Todesco et al. (2010) found that certain accessions of Arabidopsis carry alleles of the ACCELERATED CELL DEATH6 (ACD6) gene that result in constitutively elevated levels of SA and enhanced disease resistance. An analysis of worldwide Arabidopsis populations led these investigators to conclude that these alleles of ACD6 are likely to provide substantial fitness benefits despite the decrease in growth rate and biomass that is associated with SA signaling. It will be of interest to determine the extent to which induction of ICS1 and SA biosynthesis at low temperature enhances plant disease resistance and is conserved among Arabidopsis ecotypes.

Experimental procedures

Plant material and growth conditions

We used Arabidopsis thaliana ecotype Col–0 as the wild-type in this study. The camta mutants are T–DNA insertion mutants in Col–0 (Alonso et al., 2003). The single camta1 (SALK_008187), camta2 (SALK_007027) and camta3 (SALK_001152) mutants and the camta1/3 double mutant were described previously (Doherty et al., 2009). The camta1/2, camta2/3 and camta1/2/3 mutants were generated in this study by genetic crossing. The position of the T–DNA inserts and transcript levels for the CAMTA genes in the camta1/2/3 triple mutant are shown in Figure S2. The NahG transgenic line (Gaffney et al., 1993) and sid2–1 mutant (Wildermuth et al., 2001) in the Col–0 background were obtained from Dean DellaPenna (Department of Biochemistry and Molecular Biology, Michigan State University, East Lansing, MI). Plants were grown for all experiments on potting soil in environmental chambers set at 22°C under a 12 h photoperiod, with a light intensity of approximately 120 μmol m−2 per sec. Cold treatment was performed at 4°C under a 12 h photoperiod, with a light intensity of approximately 35 μmol m−2 per sec. Plants were grown for 25–32 days prior to use in experiments. Leaves were harvested at ZT4 (Zeitgeber Time 4; 4 h after dawn) for SA determinations, and at ZT23 for electrolyte leakage experiments. For microarray experiments, plants were grown for 23–30 days on soil at 22°C, treated at low temperature for the indicated times, and plant tissue was harvested at ZT4.

Quantification of SA and SAG levels

Approximately 50–100 mg fresh weight of leaf tissue (approximately 6–8 mature leaves) was frozen in liquid nitrogen, weighed, and then ground using a Fast Prep–24 tissue homogenizer (MP Biomedicals, www.mpbio.com). The tissue was extracted for 24 h at 4°C using 1 ml methanol/water (4:1 v/v) containing 0.1 g l−1 butylated hydroxytoluene and 180 ng ml−1 propyl 4–hydroxybenzoate as an internal standard. The extracts were filtered through 0.2 μm polytetrafluoroethylene membranes (Millipore, www.millipore.com), and analyzed on a Waters Quattro Premier XE UPLC/MS/MS System (Waters, www.waters.com) as described previously (Zheng et al., 2011). Transitions from deprotonated molecules to characteristic product ions were monitored for SA (m/z 137 > 93), SAGs (m/z 299 > 137) and propyl 4–hydroxybenzoate (m/z 179 > 137).

Quantification of transcript levels

Total RNA was extracted using an RNeasy plant mini kit (Qiagen, www.qiagen.com) with on-column DNase treatment. Total RNA (200 ng) was used to synthesize cDNA with random oligonucleotides using a reverse transcription system (Promega, www.promega.com). Quantitative real-time RT–PCR was performed using a Fast 7500 system with Fast SYBR Green Master Mix (Applied Biosystems, www.appliedbiosystems.com). UBQ10, IPP2, PP2A and YLS8 were used for normalization. Primer sequences are provided in Table S7.

Transcriptome analysis

Expression of approximately 22 600 genes was analyzed in WT and mutant plants using the Affymetrix (www.affymetrix.com) ATH1 GeneChip, and the resulting raw data, which have been deposited in the National Center for Biotechnology Information Gene Expression Omnibus (GEO), are accessible under accession numbers GSE43818 and GSE43819. Total RNA was prepared using an RNeasy plant mini kit (Qiagen), and the concentration and quality of extracted RNA was determined using a NanoDrop spectrophotometer (Thermo Scientific, www.thermoscientific.com) and a Z100 Bioanalyzer (Agilent, www.agilent.com). The Ambion MessageAmp Premier RNA amplification and purification system was used to prepare and purify cDNA from total RNA (100 ng) according to the manufacturer's instructions (Ambion, www.invitrogen.com/ambion). Fragmentation and hybridization of cRNA to Affymetrix ATH1 GeneChips were performed at the Research Technology Support Facility at Michigan State University.

Signals from CEL files were adjusted for background and normalized using the Bioconductor gcrma package (www.bioconductor.org). Statistically significant changes (FDR < 0.01 or FDR < 0.05 and a twofold change) in gene expression between WT and mutant plants and between treatments were determined using the Bioconductor limma package (Wettenhall and Smyth, 2004). Determination of the k–means and hierarchical clustering were performed using cluster (Eisen et al., 1998) on log2-normalized data. The resulting clusters were visualized using treeview (http://rana.lbl.gov/EisenSoftware.htm).

Gene ontology term enrichment was performed using the david program (Dennis et al., 2003) with the entire gene set on the ATH1 GeneChip as a background. P values were calculated using a modified Fisher's exact test as described by Dennis et al. (2003). For cis-element enrichment analysis, DNA motifs were counted within the region 1 kb upstream of the translation start site based on the tair version 10 annotation (Arabidopsis Information Resource, http://www.arabidopsis.org/). The significance of the motif enrichment in a group of genes was tested against a background distribution generated by 1000 random samplings from all genes in the ATH1 GeneChip.

Freezing tolerance tests

Electrolyte leakage assays for freezing tolerance were performed as described previously (Gilmour et al., 2004), except that plants were grown and cold-treated under a 12 h photoperiod.

Acknowledgements

We are grateful to A. Daniel Jones and Lijun Chen of the Mass Spectrometry Facility at Michigan State University (Department of Biochemistry and Molecular Biology) for help with the SA and SAG measurements, Kristen Schotts for help in growing plants and generating the camta mutants, Laurent Mene-Saffrane and Dean DellaPenna (Department of Biochemistry and Molecular Biology, Michigan State University) for kindly providing the sid2–1 and NahG mutants, and Sheng Yang He and Gregg Howe for discussions about SA and plant defense. SALK lines were obtained from the Arabidopsis Biological Resource Center at Ohio State University (http://abrc.osu.edu/). This research was supported by a grant from the Chemical Sciences, Geosciences and Biosciences Division, Office of Basic Energy Sciences, US Department of Energy (DE-FG02-91ER20021) and institutional support from the Michigan Agricultural Experiment Station.

Ancillary

Advertisement