Many plants, including Arabidopsis, increase in freezing tolerance in response to low non-freezing temperatures, a phenomenon known as cold acclimation. A fundamental goal of cold acclimation research is to identify genes that have roles in this increase in freezing tolerance. In recent years, it has been established that the expression of hundreds of genes is altered in response to low temperature and that some of these cold-responsive genes contribute to freezing tolerance. The CBF transcription factors were the first transcriptional activators demonstrated to have a role in controlling the expression of cold-responsive genes with a role in cold acclimation. Here, we review what is known about the CBF cold-response pathway, including several recent developments, and we discuss emerging evidence in support of several additional transcription factors having roles in freezing tolerance pathways.
Many plants, including Arabidopsis, increase in freezing tolerance in response to low non-freezing temperatures, a phenomenon known as cold acclimation (Smallwood and Bowles 2002, Thomashow 1999). A fundamental goal of cold acclimation research is to understand the mechanisms responsible for bringing about the increase in freezing tolerance in response to low temperature. Early on, it was established that many biochemical and physiological changes occur with cold acclimation including changes in composition of lipid membranes, increases in total soluble protein content and increases in levels of molecules that can serve as cryoprotectants such as sugars and proline (Guy 1990, Steponkus et al. 1993, Thomashow 1999). More recently, it was shown that cold acclimation is associated with changes in gene expression. This observation led to the hypothesis that changes in gene expression were probably responsible for some of the biochemical and physiological changes that occurred in response to low temperature and were likely to contribute to an increase in freezing tolerance. In recent years, the validity of this hypothesis has been demonstrated (Gilmour et al. 2000, Gilmour et al., 2004, Jaglo-Ottosen et al. 1998, Kasuga et al. 1999, Liu et al. 1998). This was accomplished largely through the identification of transcription factors having roles in cold acclimation. Here, we present a brief overview of these studies and direct the reader to other articles for coverage of other aspects of cold acclimation (Guy 1990, Knight 2002, Smallwood and Bowles 2002, Sung et al. 2003, Thomashow 1999).
The CBF cold-response pathway
Identification of the CBF transcription factors
Guy et al. (1985) presented the first compelling evidence that changes in gene expression occur during cold acclimation. This study, which was with spinach, was quickly followed by similar studies in other plants that cold acclimate including Arabidopsis (reviewed in Thomashow 1999). The Arabidopsis genes that were found to be upregulated in response to low temperature were given a variety of names including COR (cold-regulated), LTI (low-temperature induced), KIN (cold -induced), RD (responsive to dehydration) and ERD (early responsive to dehydration) genes. The transcript levels for these genes were found to increase greatly within a few hours of plants being exposed to low temperature, and remain elevated for the duration of the cold exposure, but quickly return to pretreatment levels upon returning the plants to warm temperatures. Reporter gene fusion studies established that the promoters of multiple COR genes were induced by low temperature and defined a core sequence, alternatively called the C-repeat (CRT), the dehydration-responsive element (DRE) and low temperature response element (LTRE), that could impart both cold and drought responsiveness.
An important step in understanding cold-regulated gene expression came with the discovery of the CBF (CRT/DRE binding factor)/DREB1 (DRE-binding factor 1) genes. CBF1, 2 and 3 (corresponding to DREB1b, DREB1c and DREB1a, respectively) are present in tandem array on chromosome 4 and are rapidly induced (within 15 min) in response to low temperature, reaching peak expression about 2 h after exposure to cold (Gilmour et al. 1998, Liu et al., 1998, Stockinger et al. 1997). The proteins encoded by these genes are transcriptional activators which contain AP2/ERF DNA-binding domains that recognize the CRT/DRE element present in the promoters of COR and other cold-regulated genes (Fig. 1). Constitutive overexpression of CBF1, 2 or 3 in transgenic Arabidopsis plants results in constitutive expression of the COR and other CRT/DRE-containing target genes and an increase in whole plant-freezing tolerance in the absence of a cold stimulus (Gilmour et al. 2000, Gilmour et al. 2004, Jaglo-Ottosen et al. 1998, Kasuga et al. 1999, Liu et al. 1998). Thus, the CBF genes were shown to be ‘master switches’ that control the expression of a group of genes, the CBF regulon, that has roles in freezing tolerance.
Genes that comprise the CBF regulon
In recent years, attention has been focused on identifying genes that are controlled by the CBF proteins. Using both cDNA microarrays and Affymetrix GeneChips, transcriptome-profiling experiments have been performed to identify sets of genes that are induced both in response to low temperature and in response to CBF overexpression (Choi et al. 2002, Fowler and Thomashow 2002, Gilmour et al. 2004, Kreps et al. 2002, Maruyama et al. 2004, Seki et al. 2001, Seki et al. 2002, Vogel et al. 2005). Genes that are upregulated under both of these conditions are considered to be members of the CBF regulon. These studies have resulted in assigning a little more than 100 genes to the CBF regulon. While the majority of these genes were upregulated in response to low temperature and CBF overexpression, several genes were repressed in response to cold and CBF overexpression. This observation was somewhat surprising given that the CBF proteins have only been demonstrated to function as transcriptional activators. Thus, the downregulated genes may be controlled by other regulatory genes that fall within the CBF regulon. A study by Gilmour et al. (2004) using Affymetrix GeneChips-containing probe sets corresponding to approximately 8000 Arabidopsis genes provided no evidence for differences in target gene specificity for CBF1, 2 or 3 indicating that the proteins are likely to control the expression of nearly identical regulons.
In a recent transcript-profiling study by Vogel et al. (2005), the authors first identified a core set of 514 cold-responsive genes, referred to as COS genes (cold standard), which were reliably cold-responsive in plants grown under two different laboratory conditions. Through additional microarray experiments, 93 of the 514 COS genes were then assigned to the CBF regulon. The COS genes that were assigned to the CBF regulon were found to comprise the majority of the genes that were most highly upregulated in response to cold; the large majority of genes that were upregulated only slightly in response to cold (<five-fold) were not assigned to the CBF regulon.
The genes assigned to the CBF regulon encode proteins that fall into a wide range of functional categories, including transcription factors, signal transduction pathway components, biosynthetic proteins, cryoprotectant proteins and other stress-related proteins, as well as a large number of genes encoding proteins of unknown function (Fowler and Thomashow 2002, Maruyama et al. 2004, Vogel et al. 2005). Several of the CBF regulon members encode proteins that have previously been shown to contribute to increased freezing tolerance, such as the cryoprotectant protein COR15a, an enzyme involved in regulating proline levels (P5CS2) and galactinol synthase, the enzyme that catalyses the first committed step in the synthesis of raffinose. However, a challenge for future research will be to determine the functions of the other CBF regulon members and to establish whether they also play roles in increasing freezing tolerance or in other aspects of life at low temperatures. Interestingly, several of the genes in the CBF regulon encode known or putative transcription factors, raising the possibility that these proteins may control subregulons within the CBF regulon. This possibility could explain in part the fact that approximately 20% of the cold-induced CBF regulon genes do not contain the core CRT/DRE sequence CCGAC within 1 kb of their start codons (Vogel et al. 2005).
CBF proteins and their mechanism of transcriptional activation
As previously mentioned, the CBF proteins belong to the AP2/ERF superfamily of transcription factors, as they have an AP2 DNA-binding domain (Riechmann and Meyerowitz 1998). This domain is present in over 140 proteins encoded in the Arabidopsis genome. These proteins are classified into five subfamilies based on the amino acid sequences of the AP2 domains (Sakuma et al. 2002). The CBF/DREB1 proteins fall into the 56-member DREB subfamily and are further set apart from other members of this subfamily by the presence of two specific amino acid ‘signature sequences’ that flank the AP2 DNA-binding domain (Jaglo et al. 2001). The CBF signature sequences consist of the consensus sequence PKK/RPAGRxKFxETRHP immediately upstream of the AP2 domain and the sequence DSAWR immediately downstream of the AP2 domain. While the precise functions of the signature sequences are unknown, their conservation between CBF proteins from diverse plant species suggests that they play an important role in CBF function (Jaglo et al. 2001).
While much has been learned about the gene regulon controlled by CBF proteins, less is known about the mechanism by which CBF proteins stimulate transcription of their target genes. A recent domain-swap study demonstrated that the N-terminal 115 amino acids are sufficient to both target CBF1 to COR gene promoters and enable binding to CRT/DRE elements, while the C-terminal 98 amino acids are sufficient for transcriptional activation (Wang et al. 2005). Extensive mutational analysis of the C-terminal domain revealed that the CBF-activation domain consists of four hydrophobic motifs that contribute functional redundancy to the trans activation function. While each of the four motifs appeared to contribute to the overall CBF trans-activation capacity, mutation of any single motif was not sufficient to noticeably hamper trans-activation in plants. The authors suggested that maintaining this level of functional redundancy may be important to ensure that the CBF pathway is activated in response to low temperature.
There is also evidence to suggest that the CBF proteins function at least in part by recruiting chromatin modification complexes to the promoters of downstream target genes. CBF proteins expressed in yeast were able to drive transcription of reporter genes containing the CRT/DRE sequence as an upstream regulatory element, demonstrating the ability of CBF to interact with the yeast transcriptional machinery and activate transcription (Gilmour et al. 1998, Stockinger et al. 1997). Several well-studied transcription factors from yeast and other eukaryotic systems, including Gcn4, Pho4 and VP16, contain transcriptional activation domains that are rich in acidic amino acids and whose function involves the recruitment of histone acetyltransferase (HAT)-containing transcriptional adaptor complexes, such as the Ada or SAGA complex, to the promoters of their target genes (Gregory et al. 1998, Ikeda et al. 1999, Kuo et al. 1998). Stockinger et al. (2001) observed that the C-terminal half of the CBF protein, which functions as a transcriptional activation domain in yeast and in plants, is also rich in acidic amino acids and hypothesized that its function may also involve the recruitment of HAT complexes to promoters. In a test of this hypothesis, they demonstrated that the ability of CBF proteins to stimulate transcription in yeast was dependent upon components of the Ada and SAGA transcriptional adaptor complexes. Ada and SAGA are large multiprotein complexes that contain the transcriptional adaptor proteins Ada2 and Ada3, the HAT protein Gcn5, as well as several other proteins (Grant et al. 1997). It has been demonstrated that recruitment of the HAT activity in the yeast SAGA complex to promoters leads to the acetylation of specific lysine residues in the N-terminal tails of nucleosomal histones at the promoter region (Sterner and Berger 2000, Utley et al. 1998). It is thought that these chromatin modifications facilitate the recruitment of RNA polymerase II and other components of the core transcriptional machinery to the target gene promoter. Stockinger et al. (2001) showed that the yeast transcriptional adaptor complex proteins Ada2, Ada3 and Gcn5 were required for CBF's ability to stimulate transcription in yeast and went on to demonstrate that Arabidopsis contains homologs for components of the SAGA and Ada complexes (GCN5, ADA2a and ADA2b), that the Arabidopsis GCN5 protein exhibits HAT activity and that CBF1 physically interacts with the Arabidopsis GCN5 and ADA2 proteins.
The results of this work suggest that CBF proteins may stimulate transcription by recruiting chromatin-modifying transcriptional adaptor complexes containing GCN5 or ADA2 proteins to the promoters of their target genes. Consistent with this hypothesis, Vlachonasios et al. (2003) demonstrated that cold-induced expression of several CBF-regulated COR genes was both reduced and delayed in Arabidopsis mutants containing T-DNA inserts in either the ADA2b or the GCN5 gene. Surprisingly, non-acclimated ada2b-1 mutant plants were more freezing tolerant than non-acclimated wildtype plants, although they did not show accumulation of COR gene transcripts, suggesting that the ADA2b protein may be involved in repressing a CBF-independent freezing tolerance pathway at warm temperatures.
Another protein that might have a role in CBF activity is SFR6 (sensitive to freezing). In a screen for Arabidopsis mutants that are deficient in the ability to cold acclimate, Warren et al. (1996) identified a gene designated SFR6; sfr6 mutants failed to increase in freezing tolerance in response to low temperature. The inability of the sfr6 mutants to cold acclimate appears to be due to a severe reduction in the induction of CRT/DRE-containing COR genes by low temperatures (Knight et al. 1999). However, in the sfr6 mutants, CBF1-3 transcripts accumulate to normal levels in response to low temperature suggesting that the defect in COR gene induction results from impaired CBF function. Perhaps the SFR6 gene product, like the GCN5 and ADA2 proteins, is a component of a multiprotein complex required for CBF to function in transcriptional activation. Other possibilities include the SFR6 protein having a role in the translation or stability of CBF protein.
Regulation of CBF transcription
A fundamental goal regarding the CBF cold-response pathway is determining how the CBF genes are upregulated in response to low temperature. Transcripts for CBF1, CBF2 and CBF3 accumulate to detectable levels within 15 min of exposing plants to low temperature (Gilmour et al. 1998, Jaglo-Ottosen et al. 1998, Liu et al. 1998, Medina et al. 1999). This rapid accumulation suggests that the components of the signal transduction and transcriptional regulatory pathways involved in CBF induction are already present at warm temperatures and that these pathways become activated by a post-transcriptional mechanism(s) in response to low temperature. The fact that the transcript levels of the CBF genes increase in response to the inhibition of protein synthesis (Zarka et al. 2003) further supports this notion.
One strategy being used to better understand the regulation of CBF transcription is to identify cis-acting DNA regulatory sequences involved in CBF cold responsiveness and to identify the trans-acting factors that bind to these elements. Using promoter–reporter gene fusions, Shinwari et al. (1998) demonstrated that the promoters of CBF1, 2 and 3 can drive cold-responsive gene expression. Sequence alignment of the three cold-responsive CBF promoters revealed several blocks of conserved sequence, raising the possibility that these conserved regions (Boxes I-VI) may be important for cold-induced transcription of the three promoters (Fig. 2A). To test this, Zarka et al. (2003) performed extensive mutational analysis of the CBF2 promoter. This work led to the identification of a 125 base-pair (bp) region of the CBF2 promoter which is sufficient to drive cold-induced expression of a GUS reporter gene. Within this 125-bp region, they further identified two smaller regions, designated ICEr1 and ICEr2 (Induction of CBF Expression region 1 or 2), that contribute to the cold responsiveness of this promoter fragment (Fig. 2B). ICEr1 and ICEr2 include conserved promoter sequences from Box IV and Box VI, respectively. The ICEr1 region contains a CANNTG, or E-box sequence, which is a consensus binding site for the basic helix-loop-helix (bHLH) class of transcription factors. A goal now is to identify the proteins that bind to ICEr1 and ICEr2.
Chinnusamy et al. (2003) have used an alternative strategy to identify components of the signalling pathway upstream of CBF cold induction. In particular, they used a mutagenized population of transgenic Arabidopsis harbouring the CBF3 promoter fused to the firefly luciferase (LUC) reporter gene to screen for mutants showing altered expression of the CBF3::LUC reporter gene. These efforts resulted in the identification of a gene, ICE1 (Induction of CBF Expression), which encodes a MYC-like bHLH transcriptional activator that binds to MYC recognition sequences in the CBF3 promoter. ICE1 is constitutively expressed, suggesting that some form of cold-induced post-translational modification is required for it to activate transcription from the CBF3 promoter. Overexpression of ICE1 enhances the expression of the CBF regulon in response to low temperature, resulting in plants that are more freezing tolerant than wildtype plants following cold acclimation. Together, these results indicate that ICE1 functions as an upstream regulatory protein that positively controls the transcription of CBF3 in the cold. ICE1, however, does not appear to be involved in the cold induction of CBF1 or CBF2. However, given that the Arabidopsis genome encodes more than 160 bHLH transcription factor proteins, it is possible that MYC-like proteins other than ICE1 may contribute to cold-induced expression of CBF1 and CBF2.
While ICE1 is the only transcription factor that has been identified to act directly on a CBF promoter, several other proteins have been suggested to play a role in CBF cold-induction. As discussed below, overexpression of the ZAT12 transcription factor leads to a dampening of CBF1, 2 and 3 transcript accumulation in response to cold, suggesting that either ZAT12 itself, or another factor affected by expression of the ZAT12 regulon, functions as a negative regulator of CBF expression (see below). In addition, Novillo et al. (2004) have proposed that CBF2 acts as a negative regulator of CBF1 and CBF3. This suggestion was based on their finding that a T-DNA insertion in the CBF2 gene resulted in constitutive expression of CBF1 and CBF3. As would be predicted, it was found that cbf2 mutant plants also constitutively expressed CBF target genes and were more freezing tolerant than wildtype plants. The model put forth by the authors is that CBF1 and CBF3 are quickly induced in response to low temperature followed closely by the induction of CBF2 which, in turn, leads to the suppression of CBF1 and CBF3 expression.
A number of other genes that affect cold induction of CBF1-3 have been identified using LUC-based genetic screens. By screening mutated populations of Arabidopsis plants containing the cold-responsive promoter of RD29A (a CBF target gene) fused to the LUC reporter gene, several mutants were isolated that displayed aberrant patterns of CBF cold induction. Recessive mutations in HOS1 were found to result in an increase in the cold-induced accumulation of CBF transcripts and of several CBF target gene transcripts, suggesting that the HOS1 protein functions to negatively regulate CBF expression (Ishitani et al. 1998, Lee et al. 2001). HOS1 encodes a novel protein with a region of limited homology to the RING-finger domain found in several proteins that act as inhibitors of apoptosis in animal systems. In these proteins, the RING-finger domain functions as an E3 ubiquitin ligase that is involved in targeting specific regulatory proteins for degradation by the 26S proteasome (Yang et al. 2000). It has therefore been hypothesized that HOS1 negatively regulates CBF transcription by targeting a positive regulator of CBF, such as ICE1, for degradation. HOS1 protein accumulates in the nucleus in response to low temperature, raising the possibility that this protein serves as a means of communicating the ‘cold signal’ from the cytoplasm into the nucleus.
The los4-1 mutation was also identified by screening for mutants that aberrantly expressed the RD29A::LUC reporter gene in response to low temperature (Gong et al. 2002). In this case, the mutation resulted in decreased expression of the reporter gene in response to cold. Analysis of this mutant revealed that it was impaired in its cold induction of CBF1-3 and their downstream target genes. Not only was this mutant unable to properly cold acclimate, it was also chilling sensitive. The LOS4 locus was cloned and found to encode a DEAD-box RNA helicase. Surprisingly, a second allele of the same gene was recently characterized and found to have the opposite effect on the CBF cold-response pathway (Gong et al. 2005). This mutant, los4-2/cryophyte, showed enhanced cold induction of the CBF genes and their downstream target genes and was more resistant to chilling stress than the wildtype counterpart. The authors demonstrated that the los4-2 mutation resulted in a disruption of mRNA export at warm or high temperatures but not at low temperatures, whereas the los4-1 mutation inhibited mRNA export at both warm and cold temperatures. It was suggested (Gong et al. 2005) that this defect in mRNA export may affect the cold acclimation response by acting on the export of either CBF mRNAs directly or of mRNA of an early component of the cold-signalling pathway.
Additional cold-regulatory pathways involved in cold acclimation
The results reported to date provide strong evidence that the CBF cold-response pathway has a prominent role in cold acclimation in Arabidopsis. However, this does not rule out the possibility that additional pathways are activated in response to low temperature which contribute to an increase in freezing tolerance. In fact, several lines of evidence suggest that this is likely to be the case. Xin and Browse (1998) isolated a mutant, eskimo1, that is constitutively freezing tolerant in the absence of cold acclimation and yet does not display constitutive expression of the COR genes. This suggests that the esk1 mutation results in the activation of a freezing tolerance pathway other than the CBF pathway. Similarly, mutation of the ADA2 transcriptional adaptor gene results in constitutive freezing tolerance without constitutive COR gene expression, again suggesting the existence of a CBF-independent freezing tolerance pathway (Vlachonasios et al. 2003). Furthermore, transcript-profiling experiments have indicated that while the majority of the genes that are most highly upregulated in response to cold are regulated by the CBF proteins, a number of cold-responsive genes fall outside of the CBF regulon (Vogel et al. 2005). Interestingly, several of these CBF-independent genes encode known or putative transcription factors that are rapidly induced in response to low temperature, raising the possibility that they may be responsible for controlling the cold-responsive expression of CBF-independent regulons.
The ZAT12 cold-response pathway
With the aim of identifying additional cold-response pathways, Vogel et al. (2005) performed transcript-profiling experiments on plants overexpressing one of several CBF-independent cold-responsive transcription factors that were induced in parallel with CBF1-3. This work led to the discovery of the ZAT12 cold-response pathway. ZAT12 is a zinc-finger protein that is predicted to encode a transcriptional repressor (Hiratsu et al. 2002). Overexpression of ZAT12 resulted in the repression of 15 genes that are downregulated in response to low temperature and induced the expression of nine genes that are induced in response to low temperature. Overexpression of ZAT12 at warm temperatures also led to a small but reproducible increase in overall freezing tolerance, indicating that the ZAT12 regulon plays a role in cold acclimation. It was also found that the ZAT12 cold-response pathway and the CBF cold-response pathway interact in at least two ways. First, there is overlap in the ZAT12 and CBF2 regulons. Specifically, four genes that are upregulated and three genes that are downregulated in response to low temperature are members of both the CBF and ZAT12 regulons. Secondly, constitutive overexpression of ZAT12 resulted in a dampening of the induction of CBF1-3 in response to cold, suggesting that ZAT12 or a member of the ZAT12 regulon plays a negative role in regulating the CBF cold-response pathway.
The HOS9 and HOS10 ‘pathways’
Recently, Zhu et al. (2004, 2005) identified two additional genes, HOS9 and HOS10, that might have roles in controlling regulons that contribute to freezing tolerance. HOS9 and HOS10, which are constitutively expressed, encode, respectively, a putative homeodomain transcription factor with similarity to the Arabidopsis proteins WUSCHEL (WUS) and PRESSED FLOWER (PRS) and a putative R2R3-type MYB transcription factor. Both genes were identified in genetic screens for mutants that displayed altered regulation of the RD29A::LUC reporter gene. The hos9-1 and hos10-1 mutations both resulted in enhanced induction of RD29A and several other cold-responsive genes but had no effect on cold induction of CBF1-3. Thus, it would appear that HOS9 and HOS10 act as negative regulators of certain cold-responsive genes either independently of CBF1-3 or at a point downstream of CBF1-3 action. A microarray analysis of hos9-1 plants grown at warm temperatures indicated that HOS9 is also involved in controlling the expression of constitutively expressed genes. Finally, HOS9 and HOS10 have been shown to have roles in freezing tolerance. However, the results of these experiments are somewhat counter-intuitive. That is, in both hos9–1 and hos10–1 plants, known cold-responsive genes that have been associated with cold acclimation are expressed at higher levels than in wildtype plants. Thus, it might be anticipated that the hos9-1 and hos10-1 plants would be more freezing tolerant than wildtype plants. However, the opposite is the case: both hos9-1 and hos10-1 mutant plants have been shown to be less freezing tolerant than wildtype plants both before and after cold acclimation treatments. An explanation for these findings put forward by the authors is that HOS9 and HOS10 might regulate the expression of genes that are essential for cold tolerance and that the loss of these activities in hos9-1 and hos10-1 plants might result in increased expression of other cold-induced genes in a response that attempts to compensate for the increased cold sensitivity caused by the mutations. Taken together, current results indicate that HOS9 and HOS10 have important roles in Arabidopsis freezing tolerance, but additional investigation will be required to clearly establish the nature of those roles.
Conclusions and future prospects
Recent large-scale microarray analyses have established that the expression of hundreds of genes is altered in Arabidopsis in response to low temperature. A major goal now is to organize these genes into regulons, determining the transcription factors that control their expression and the roles that the regulons have in freezing tolerance and other aspects of growth and survival at low temperature. To date, two partially overlapping regulatory pathways with roles in cold acclimation have been described, the CBF and ZAT12 cold-response pathways. Current evidence indicates that the CBF pathway has a much greater effect on freezing tolerance than the ZAT12 pathway, although their relative contributions to chilling tolerance and other stresses associated with low temperature remain to be determined. In addition, there is strong evidence to indicate that Arabidopsis has additional cold-response pathways that may contribute to freezing tolerance, including the ones involving action of the HOS9 and HOS10 transcription factors and the Eskimo1 protein. Since the first studies demonstrating that changes in gene expression occur in response to low temperature, significant progress has been made in identifying genes that have roles in cold acclimation. Continuing to incorporate the powerful and expanding ‘omic’ technologies into the study of cold acclimation makes it likely that the next few years will bring even greater advances in our understanding of genes involved in freezing tolerance.