Molecular Basis of Plant Cold Acclimation: Insights Gained from Studying the CBF Cold Response Pathway

For the 75th Anniversary Issue of Plant Physiology, I contributed an article (Thomashow, 2001) in which I highlighted recent advances in the identification of genes with roles in cold acclimation: the process whereby certain plants increase in freezing tolerance in response to low nonfreezing temperatures. At that time, about 20 Arabidopsis (Arabidopsis thaliana) genes had been identified as cold regulated, and it seemed that it might be relatively easy to model the low-temperature transcriptional regulatory network. Since then, largely through the development and use of microarrays and other genomic tools, we have learned that exposing plants to low temperature triggers a highly complex regulatory program that results in extensive reorganization of the transcriptome. These changes include up- and down-regulation of hundreds of genes occurring in waves with time of incubating plants at low temperature (Fowler and Thomashow, 2002; Maruyama et al., 2004; Hannah et al., 2005; Vogel et al., 2005; Oono et al., 2006; Kilian et al., 2007; Robinson and Parkin, 2008). A hierarchical cluster analysis of the transcriptome data of Kilian et al. (2007) clearly reveals these features (Fig. 1A) and provokes a number of fundamental questions. How do plants sense low temperature; that is, what is the nature of the low-temperature thermometer? How is this information processed to activate the first wave of cold-regulated genes? What is the regulatory logic that underlies the cascading pattern of the low-temperature gene network? And what biological functions can be ascribed to the genes that constitute the various circuits of the network? At present, the answers to these questions are far from complete, but significant insights have been gained. Here, I will highlight a few of the advances that speak to these issues, focusing on the CBF cold response pathway of Arabidopsis. Additional information on the topics covered here and other facets of cold acclimation can be found in other recent reviews (Chinnusamy et al., 2007; Guy et al., 2008; Penfield, 2008; Galiba et al., 2009; Hua, 2009). Waves of cold-regulated genes produced upon exposing Arabidopsis plants to low temperature. The AtGenExpress global stress expression data set (Kilian et al., 2007) was analyzed using the Bioconductor LIMMA package to identify genes that were up- or down-regulated at least 4-fold (false discovery rate < 0.01) at any time point of the cold treatment. This resulted in the identification of 1,335 up-regulated and 1,061 down-regulated genes. A, Genes were grouped using k-means cluster analysis. Log2-transformed expression levels are represented by the color gradient included at the bottom of the cluster diagram. The extreme values given are −3 to +3 or higher in expression level. B, The patterns for up-regulated genes encoding transcription factors (174 total) were identified according to Arabidopsis Gene Ontology (Berardini et al., 2004) and grouped using k-means cluster analysis. Determining the overall “wiring diagram” of the low-temperature regulatory network will not be a trivial matter. Indeed, among the more than 1,000 genes that are cold induced, there are more than 170 genes that encode transcription factors (Fig. 1B). Presumably, all of these transcription factors act in concert to reconfigure the transcriptome at low temperature. However, as has been learned in other systems, certain transcription factors can serve as major regulatory “hubs” controlling the expression of a large number of genes and contribute much to the biology under study. The Arabidopsis CBF transcription factors constitute such a regulatory hub for cold acclimation. Arabidopsis encodes three cold-inducible CBF genes, CBF1, CBF2, and CBF3 (Stockinger et al., 1997; Gilmour et al., 1998; Medina et al., 1999), also referred to as DREB1b, DREB1c, and DREB1a, respectively (Liu et al., 1998). These genes, which are located in tandem array in the genome, encode closely related transcription factors that are members of the AP2/ERF family of DNA-binding proteins. The CBF proteins bind to the CRT/DRE regulatory element present in the promoters of target genes, referred to as the CBF regulon, and stimulate their transcription. Induction of the CBF genes occurs within about 15 min of transferring plants to low temperature (4°C), followed by induction of the CBF target genes beginning at about 2 to 3 h. Vogel et al. (2005) found that 85 (28%) of the 302 genes that were determined to be cold induced in wild-type plants were also induced by constitutive overexpression of CBF2 in transgenic plants grown at warm temperature, indicating that the CBF cold response pathway plays a major role in configuring the low-temperature transcriptome. This may be an underestimate of the size of the CBF regulon, as it assumes that the CBF transcription factors by themselves are sufficient to induce the expression of target genes at warm temperature. This may not be the case, as combinatorial regulation of gene expression is a common theme in gene regulation (Priest et al., 2009). In addition, CBF2 overexpression was found to cause down-regulation of eight (4%) of the 212 genes that were found to be repressed in response to low temperature (Vogel et al., 2005). The promoters of these did not have the CRT/DRE element and thus would not appear to be direct targets of CBF, but they might be regulated by transcriptional repressors that are induced by CBF expression and thus be part of the CBF regulatory network. The CBF regulon has a major role in freezing tolerance. The initial evidence for this was the finding that transgenic Arabidopsis plants constitutively expressing CBF1, CBF2, or CBF3 resulted in an increase in freezing tolerance without exposing plants to low temperature (Jaglo-Ottosen et al., 1998; Liu et al., 1998; Gilmour et al., 2004). Subsequently, Novillo et al. (2007) used RNA interference and antisense constructs to down-regulate the expression of CBF1 and CBF3 and found that this resulted in about a 25% to 50% decrease in freezing tolerance in cold-treated plants. In addition, Hannah et al. (2006) examined six accessions of Arabidopsis that differed in freezing tolerance and found that the levels of CBF1 and CBF2 expression, as well as genes that constitute the CBF2 regulon, positively correlated with freezing tolerance. Finally, Alonso-Blanco et al. (2005) mapped quantitative trait loci (QTLs) for freezing tolerance in recombinant inbreed lines produced from a cross between two accessions that differ in their abilities to cold acclimate, Cape Verde Islands (Cvi) and Landsberg erecta (Ler); Cvi is less freezing tolerant than Ler. A total of seven QTLs were identified, one of which, FTQ4, mapped to the CBF locus and accounted for about 20% of the variation in freezing tolerance. The recombinant inbred lines carrying the FTQ4 locus from Cvi were those that were less freezing tolerant. It was found that cold induction of the CBF2 gene (and three known CBF regulon genes) was much less in Cvi plants than in Ler plants and that the promoter region of the Cvi CBF2 gene had a 1.6-kb deletion compared with the promoter of the Ler CBF2 gene. From these and other results, it was concluded that the promoter deletion was likely to account for the poor induction of the Cvi CBF2 gene and that this was the underlying molecular basis for the freezing tolerance QTL. The studies summarized above point to the CBF regulatory hub having a major role in configuring the low-temperature transcriptome and in conditioning freezing tolerance. But, do all three CBF genes have the same function? To address this question, Gilmour et al. (2004) compared the transcriptomes of transgenic Arabidopsis plants (ecotype Wassilewskija-2) constitutively overexpressing CBF1, CBF2, or CBF3 and found that each transcription factor affected the expression of the same set of cold-regulated genes; no evidence for specialized function was obtained. In contrast, Novillo et al. (2004) concluded that the function of CBF2 is distinct from that of CBF1 and CBF3. In particular, they found that Arabidopsis plants (ecotype Columbia) carrying a T-DNA insertion that inactivated the expression of CBF2 had higher levels of CBF1 and CBF3 transcripts and greater freezing tolerance than did wild-type plants; this was true for comparisons of both nonacclimated and cold-acclimated plants. In addition, they reported that CBF1 and CBF3 were induced a little earlier than CBF2. From these and other results, the investigators proposed that CBF2 is a negative regulator of CBF1 and CBF3 expression and a negative regulator of freezing tolerance. The apparent conflict in conclusions reached in the Gilmour et al. (2004) and Novillo et al. (2004) studies may only be apparent due to the different ecotypes or approaches used. In regard to the latter, overexpression studies (in these cases using the cauliflower mosaic virus 35S promoter) not only eliminate potential tissue specificity issues but may also eliminate potential subtle differences in CRT/DRE binding affinities between the three CBF proteins. A bit more difficult to understand is why the natural variation study of Alonso-Blanco et al. (2005) indicated that down-regulation of CBF2 results in a decrease in freezing tolerance, whereas Novillo et al. (2004) concluded that down-regulation of CBF2 results in an increase in freezing tolerance; but again, the systems are different. Clearly, additional work is needed to resolve this fundamental issue about the CBF regulatory hub. Another fundamental question is how does the CBF regulon bring about an increase in freezing tolerance? A complete answer to this question will require additional study, but it is clear that multiple mechanisms are involved. As has been reviewed elsewhere (Thomashow, 1999; Sung et al., 2003; Penfield, 2008) and thus will only be mentioned here, certain CBF regulon genes encode enzymes involved in the biosynthesis of low-M r cryoprotectants such as Suc, raffinose, and Pro, and others encode hydrophilic cryoprotective polypeptides. All of these cryoprotectants appear to function, at least in part, by protecting membranes and proteins against the severe dehydration stress that occurs with freezing. The function of the CBF regulatory hub, however, involves mechanisms that go beyond the production of cryoprotectants. Arabidopsis plants constitutively overexpressing CBF1, CBF2, or CBF3 grow slowly, have a dwarf stature, and are delayed in flowering (Liu et al., 1998; Gilmour et al., 2004). It is not surprising that CBF-overexpressing transgenic plants grow poorly at normal growth temperature, as they are forced to grow with a transcriptome (Maruyama et al., 2004; Vogel et al., 2005) and metabolome (Cook et al., 2004; Kaplan et al., 2007; Usadel et al., 2008) designed for life at low temperature. However, a more profound reason for the observed effects of CBF expression was discovered by Achard et al. (2008a). What these investigators found was that constitutive overexpression of CBF1 results in the accumulation of DELLA proteins, a small family of growth-restraining proteins that are part of the gibberellin (GA) signaling pathway (Harberd et al., 2009). In plants with normal levels of biologically active GAs, the DELLA proteins are degraded through the ubiquitin-proteasome pathway. However, when biologically active GA levels are low, DELLA proteins accumulate, restrain growth, and cause a dwarf stature and delayed flowering. Achard et al. (2008a) found that plants overexpressing CBF1 had reduced levels of biologically active GAs and that this was due to increased expression of two genes encoding GA 2-oxidases, enzymes that convert biologically active GAs to inactive GAs. The decrease in active GAs results in an increase in DELLA proteins, which in turn causes the dwarf and delayed flowering phenotypes. Indeed, constitutive overexpression of CBF1 did not result in dwarf and delayed flowering phenotypes in plants that carried the gai-t6 and rga-24 mutations, which inactivate two major DELLA proteins, GAI and RGA, respectively. In addition to affecting plant growth, DELLA protein activity was found to be required for plants to attain full freezing tolerance (Achard et al., 2008a). The increase in freezing tolerance brought about by constitutive overexpression of CBF1 in transgenic plants grown at warm temperature was reduced about 50% by introduction of the gai-t6 and rga-24 mutations. Moreover, the increase in freezing tolerance observed in cold-treated wild-type plants was reduced about 50% by introduction of the gai-t6 and rga-24 mutations. These results, along with others reported by Achard et al. (2008a), indicate that a function of the CBF regulatory hub is to restrain plant growth through affecting the activity of the GA signaling pathway, and this contributes to the ability of plants to survive freezing. The mechanism by which the DELLA proteins contribute to freezing tolerance is not known but may involve reducing the levels of reactive oxygen species (Achard et al., 2008b). In addition, DELLA-mediated change in plant architecture could potentially contribute to survival in the field. For instance, it has been observed (Roberts, 1990) that varieties of wheat (Triticum aestivum) with greater winter survival have a more prostrate growth habit, which results in a greater chance of the plants being covered by snow and insulated against harsh air temperatures. Finally, DELLA-induced restraint of growth may have a broad role in abiotic stress tolerance, as mutations inactivating the DELLA genes also reduce tolerance to high salinity stress (Achard et al., 2006). As alluded to earlier, a fundamental question regarding cold acclimation is how do plants sense low temperature and initiate the regulatory program that results in enhanced freezing tolerance; that is, what is that nature of the low temperature thermometer? One approach to address this issue is to identify the transcription factors involved in expression of the first wave of cold-regulated genes and to work back upstream into the temperature-sensing system by understanding how the activities of the transcription factors are affected by low temperature. Such an approach has provided clues about the nature of two cold-sensing pathways that control CBF expression, one of which involves posttranslational protein modification and the other calcium signaling. A major positive regulator of CBF3 is ICE1, a basic helix-loop-helix transcription factor that binds to multiple Myc DNA regulatory elements present in the CBF3 promoter and stimulates CBF3 transcription (Chinnusamy et al., 2003). The role of ICE1 has been extensively reviewed (Chinnusamy et al., 2007; Hua, 2009), so only key points regarding two forms of posttranscriptional regulation will be mentioned here. The ICE1 gene is constitutively expressed, suggesting that cold induction of CBF3 involves low temperature-induced activation of the ICE1 protein. The results of Miura et al. (2007) provide strong evidence that this is the case and that this is accomplished through low temperature-induced sumoylation of ICE1 mediated by the SIZ1 protein, a SUMO E3 ligase. This activation process, however, is countered by HOS1, a RING finger E3 ligase that mediates ubiquitination and degradation of ICE1 (Lee et al., 2001; Dong et al., 2006). At normal warm growth temperature, the HOS1 gene is expressed and the HOS1 protein is located in the cytoplasm, but upon exposing plants to low temperature, HOS1 locates to the nucleus. Thus, low temperature appears to initiate a cycle of activation and inactivation of ICE1: a rapid activation of ICE1 due to SIZ1-mediated sumoylation followed by ICE1 inactivation due to HOS1-mediated ubiquitination and degradation. These events are consistent with the “transient” nature of CBF expression; low temperature induces a rapid dramatic increase (greater than 10-fold) in CBF transcript levels, reaching a maximum at about 2 to 3 h followed by a decrease to levels that are only a few fold over those found in warm-grown plants. Thus, the SIZ1-HOS1 system appears to contribute to a fine tuning of CBF gene expression that enables up-regulation of the system but also prevents “runaway” induction with potential deleterious effects. Such fine tuning, however, includes factors in addition to the SIZ1-HOS1 system, as the loss-of-function hos1 mutation does not eliminate the transient nature of CBF induction; peak levels of CBF3 transcripts are greater and remain higher for a longer period of time in plants carrying the hos1 mutation but still return to a few fold over those found in nonacclimated plants (Lee et al., 2001). Major questions that remain to be answered regarding the SIZ1-HOS1 system and relate to low-temperature thermometers are how does low temperature control the cellular location of HOS1, how does it affect the ability of SIZ1 to mediate sumoylation of ICE1, and how does sumoylation increase the activity of ICE1? Another cold-sensing pathway involves calcium. Nearly 20 years ago, Knight et al. (1991) established that exposure of plants to low temperature results in a rapid transient spike in cytoplasmic calcium levels. Subsequent studies linked the cold-induced calcium spikes to cold-regulated gene expression in Arabidopsis (Knight et al., 1996; Tahtiharju et al., 1997; Henriksson and Trewavas, 2003) and alfalfa (Medicago et al., and For instance, the studies with Arabidopsis that that reduce calcium into the cold induction of certain CBF regulon genes. The calcium involved in cold-induced calcium has not been but a mechanism whereby calcium levels may be and the information processed to induce CBF gene expression is by the of et al. These investigators identified a region of the CBF2 promoter that can cold-induced gene expression and that this activity was on a DNA known as the the binding for the six transcription proteins by Arabidopsis et al., Moreover, they that the activity of the promoter required a gene. The is that low temperature induces a rapid increase in calcium that to the of a complex that stimulates the transcription of target genes. Arabidopsis plants carrying a mutation were found to be in the cold induction of CBF1, CBF2, and transcript accumulation is reduced by about all of which have the DNA-binding in their suggesting that these genes are direct targets of carrying a mutation were not in freezing tolerance, but those carrying a mutation indicating an role for the transcription factors in cold acclimation. Additional study will be required to the between calcium and changes in gene expression, a of the to which cold-regulated gene expression involves the regulatory proteins. studies on cold acclimation in of those in have used plants grown under The was that low temperature induces changes in gene expression and freezing tolerance under and that the nature of the low-temperature regulatory network that for these changes would be difficult without in potentially effects by and these could be However, such a approach can to reduced for the of the biological process under study, and factors to the system can be This is the case with cold-regulated gene The first that were potentially of cold-regulated gene expression was the finding that CBF3 and certain CBF regulon genes were to regulation at normal growth et al., This the question of low-temperature into the CBF regulatory hub might be by the it Arabidopsis plants are grown under a and to the cold induction of CBF1, CBF2, and CBF3 is much greater the temperature is at h than at (Fowler et al., 2005; cold-regulated expression of the CBF genes is by the of CBF expression also occurs in et al., suggesting that this is a highly of regulation and of fundamental Determining how the and low-temperature regulatory pathways are to affect CBF gene expression and how this the ability of plants to with cold be in this regard was by et al. that the basic helix-loop-helix protein is a transcriptional involved in regulation of CBF1 and CBF2 and that this regulation involves of with a of the and a In addition, it be that the of calcium the under control of the et al., 2005) and that the spike in calcium produced in response to low temperature is by the et al., 2006). As mentioned the regulation of CBF1 and CBF2 involves of the transcription factors et al., 2009). the proteins in the and low-temperature of calcium and contribute to CBF gene regulation and freezing tolerance. A major into the of low temperature and the and the effects that this has on cold-regulated genes has from the work of et al. These investigators found that low temperature the function of the in Arabidopsis and that this for much of the variation in the of the low-temperature transcriptome reported in different For instance, when plants that are grown under a are to low temperature, the transcript levels for certain genes that are of the such as and certain genes, such as are in the are or upon transferring plants to low Thus, one the transcript for one of these genes at a given time point of cold with the transcript of the gene at the same time point in under which the transcript for the gene is the gene will be as cold induced, cold or on the time the of certain genes can be affected by low temperature was not a as normal of certain genes had been observed to be in cold-treated and and et al., 2005). However, the to which gene expression is by low temperature and the that this has on the low-temperature regulatory network were not to the study by et al. In the that I contributed to the 75th Anniversary Issue of Plant (Thomashow, I concluded with a that additional study the approaches of would likely a set of that would the of plants with increased freezing tolerance. The years has not produced the required to convert plants into so on how one might be as being However, what the years has produced is a of work that has understanding of the cold acclimation and has the fundamental questions at the of this The along with the development and of and will bring an understanding of the cold acclimation that may well include at a will the first I for of and Gilmour and Dong for on how to the