A role for circadian evening elements in cold-regulated gene expression in Arabidopsis

Authors

  • Michael D. Mikkelsen,

    1. MSU-DOE Plant Research Laboratory, Michigan State University, East Lansing, MI 48824, USA
    Search for more papers by this author
    • Present address: University of Copenhagen, Faculty of Life Sciences, Department of Plant Biology, Plant Biochemistry Laboratory, VKR Research Centre for Pro-Active Plants, 40 Thorvaldsensvej, DK-1871 Frederiksberg C, Copenhagen, Denmark.

  • Michael F. Thomashow

    Corresponding author
    1. MSU-DOE Plant Research Laboratory, Michigan State University, East Lansing, MI 48824, USA
    2. Department of Crop and Soil Sciences, Michigan State University, East Lansing, MI 48824, USA
    Search for more papers by this author

*(fax +517 353 9168; e-mail thomash6@msu.edu).

Summary

The plant transcriptome is dramatically altered in response to low temperature. The cis-acting DNA regulatory elements and trans-acting factors that regulate the majority of cold-regulated genes are unknown. Previous bioinformatic analysis has indicated that the promoters of cold-induced genes are enriched in the Evening Element (EE), AAAATATCT, a DNA regulatory element that has a role in circadian-regulated gene expression. Here we tested the role of EE and EE-like (EEL) elements in cold-induced expression of two Arabidopsis genes, CONSTANS-like 1 (COL1; At5g54470) and a gene encoding a 27-kDa protein of unknown function that we designated COLD-REGULATED GENE 27 (COR27; At5g42900). Mutational analysis indicated that the EE/EEL elements were required for cold induction of COL1 and COR27, and that their action was amplified through coupling with ABA response element (ABRE)-like (ABREL) motifs. An artificial promoter consisting solely of four EE motifs interspersed with three ABREL motifs was sufficient to impart cold-induced gene expression. Both COL1 and COR27 were found to be regulated by the circadian clock at warm growth temperatures and cold-induction of COR27 was gated by the clock. These results suggest that cold- and clock-regulated gene expression are integrated through regulatory proteins that bind to EE and EEL elements supported by transcription factors acting at ABREL sequences. Bioinformatic analysis indicated that the coupling of EE and EEL motifs with ABREL motifs is highly enriched in cold-induced genes and thus may constitute a DNA regulatory element pair with a significant role in configuring the low-temperature transcriptome.

Introduction

Low temperature is an environmental factor that affects many aspects of plant growth and development. Two prominent examples are freezing tolerance and time to flowering. In both cases, low-temperature-induced changes in gene expression have key roles. Within minutes of exposing Arabidopsis plants to low temperature, changes in transcript levels can be detected followed by waves of changes in transcriptome composition (Fowler and Thomashow, 2002; Maruyama et al., 2004; Vogel et al., 2005). By 24 h, the transcript levels of more than 1000 genes, some of which contribute to enhanced freezing tolerance, have either increased or decreased (Chinnusamy et al., 2007). Other changes in gene expression, like repression of a key regulator of vernalization, FLC, occur after weeks of exposure to low temperature (Amasino, 2004).

A fundamental goal common to both cold acclimation and vernalization research is to determine how changes in gene expression are brought about in response to low temperature. One line of investigation focuses on identifying the cis-acting DNA regulatory elements and trans-acting protein factors that control expression of the low-temperature transcriptome. To date, the best characterized cold regulatory element/transcription factor pairs are components of the CBF cold response pathway. Within 15 min of transferring Arabidopsis to low temperature, the CBF1, -2 and -3 genes (also known as DREB1b, -1c and -1a, respectively) are induced (Stockinger et al., 1997; Gilmour et al., 1998; Jaglo-Ottosen et al., 1998; Liu et al., 1998). These genes encode transcriptional activators that are members of the AP2/ERF family of DNA-binding proteins (Riechmann and Meyerowitz, 1998). The CBF transcription factors bind to the cold- and drought-responsive CRT/DRE element, which has the core sequence CCGAC, and activate expression of about 100 genes with roles in freezing tolerance (Jaglo-Ottosen et al., 1998). It is also known that ICE1, a member of the basic helix–loop–helix (bHLH) family of transcription factors, binds to consensus MYC binding sites, CANNTG (Meshi and Iwabuchi, 1995), present in the promoter of CBF3, and acts as a positive regulator (Chinnusamy et al., 2003). Also, a member of the MYB family of transcription factors, MYB15, binds to MYB-binding sites in the promoters of CBF1–3 and has a small repressive effect on expression of these genes (Agarwal et al., 2006).

Although the CBF cold response pathway has a central role in cold acclimation, the regulatory elements and transcription factors involved in controlling expression of the large majority of cold-responsive genes are not known. Bioinformatic analysis has shown that in addition to the CRT/DRE element, the ABA-responsive element (ABRE), the G-box element and the evening element (EE) are enriched in the promoters of cold-induced genes (Kreps et al., 2003; Suzuki et al., 2005). However, the roles of these regulatory elements in cold-regulated gene expression are not well understood. The ABREs, which have a core recognition sequence of ACGT and preference for ACGTG (Suzuki et al., 2005), mediate ABA-inducible gene expression by binding members of the basic region/leucine zipper motif (bZIP) family of transcription factors (Choi et al., 2000; Jakoby et al., 2002). In Arabidopsis, exposure to low temperature results in a small transient increase in ABA levels which induces low-level expression of genes that are highly responsive to ABA, such as RAB18 (Lang et al., 1994). However, the general thought is that cold-regulated gene expression proceeds primarily through ABA-independent pathways (Shinozaki et al., 2003). The G-box, CACGTG, includes the core and preferred ABRE sequences, but also includes the core binding sequence for bHLH transcription factors. Indeed, the G-box has been implicated in several regulatory pathways including ABA-regulated gene expression, which presumably involves the binding of bZIP factors, and light-regulated gene expression, which involves binding of the bHLH phytochrome-interacting-factor (PIF) transcription factors (Duek and Fankhauser, 2005). It may also have a significant role in cold-regulated gene expression as cold-responsiveness of the ADH1 promoter requires a G-box (Dolferus et al., 1994).

The EE motif, AAAATATCT, is a key component of circadian-regulated gene expression that induces peak expression in the evening (Harmer et al., 2000). There is no evidence that the EE can induce or repress gene expression in response to low temperature. However, it has been shown that cold induction of CBF1–3 is gated by the circadian clock (Fowler et al., 2005). Rapid cold induction of CBF1–3 occurs when plants are grown under either constant light or under diurnal conditions. Furthermore, when plants are grown under a 12-h photoperiod and then transferred to constant light the level of induction of CBF2 by low temperature is greater at 4 and 28 h after dawn than it is at 16 and 40 h after transfer (Fowler et al., 2005). Whether this gating involves action of the EE is not known.

The primary objective of this study was to determine whether EE motifs have direct roles in cold-regulated gene expression. The results indicate that they do and that ABRE-like (ABREL) motifs potentiate their action. The results also provide evidence for the EE being a site of integrating cold- and circadian-regulated gene expression.

Results

COL1 and COR27 are rapidly induced in response to low temperature through a CBF-independent pathway

To assess the role of the EE in cold-regulated gene expression, we first identified genes that were induced in response to low temperature and had EE motifs in their promoters. Also, to broaden our understanding of cold response pathways, we wanted to work with genes that were independent of the CBF cold response pathway. We identified such genes among the COS (cold-standard) genes described by Vogel et al. (2005). These investigators identified a group of 302 genes, designated COS, that were consistently induced in response to low temperature. Of these, 218 were not up-regulated by constitutive overexpression of CBF2 and were thus considered CBF-independent. Of these 218 genes, 43% (93) contained EE motifs (see Table S1 in Supporting Information). We searched this gene set for candidates that did not contain CRT/DRE elements in their promoters (a 1 kb region 5′ to the ATG start codon), were highly induced by low temperature and showed little response to ABA, drought, high salinity and oxidative stress in experimental data deposited in publicly available databases. From these, we chose to investigate CONSTANS-like 1 (COL1; At5g54470) and a gene encoding a 27-kDa protein that we designated COLD-REGULATED GENE 27 (COR27; At5g42900). The function of COL1 is not known, but as a member of the CONSTANS gene family it may be a regulator of flowering (Putterill et al., 1995). The annotation for COR27 is ‘protein of unknown function’.

An inspection of the COL1 promoter sequence (see Figure S1) indicated that it has one EE motif, two EE-like (EEL) motifs, defined as AATATCT, and one CCA1-binding site (CBS), AAAAATCT; CCA1 is a MYB family transcription factor that is a key component of the circadian clock (see Discussion). It also has six ABREL motifs, one of which is part of a G-box. Similarly, the COR27 promoter (see Figure S2) was found to have two EE motifs, two EEL motifs and three ABREL motifs, two of which are part of G-boxes.

Time-course experiments indicated that both COL1 and COR27 were rapidly induced upon exposing plants to low temperature (Figure 1a). The transcript levels for both genes increased slightly at 1 h, continued to increase over the next 4 h and remained high at 24 h. This kinetic pattern was different from the rapidly induced CBF2 gene where transcript levels were near maximum at 1 h and by 4 h had already started to decline, reaching low levels at 24 h. The CBF2-targeted gene COR15a was induced after COL1 and COR27 (Figure 1a).

Figure 1.

COR27 and COL1 are cold induced through a CBF-independent pathway.
Semiquantitative RT-PCR was performed on total RNA (5 μg) as described in the Experimental procedures using ACT3 as a normalization control. Plants were grown for 10–14 days under continuous light at 24°C (W) and treated as follows:
(a) Placed at 4°C for the times indicated.
(b) Placed at 4°C (C) for 24 h either in the dark (D) or the light (L).
(c) Placed at 4°C for 24 h (C). Ws-2 (Ws) is the background for the CBF2-overexpresser (C-ox).

Cold induction of a small percentage of genes is strongly dependent on light (Soitamo et al., 2008). This, however, was not the case with either COR27 or COL1; the transcript levels attained for each gene were about the same when plants were treated at low temperature in the dark or light (Figure 1b). Neither of the genes had CRT/DRE elements in their promoter nor were they induced in response to CBF2 overexpression (Figure 1c), indicating that they were members of a CBF-independent cold response pathway. Also, neither COR27 nor COL1 were members of the ZAT12 regulon, COS genes which are either induced (9 genes) or repressed (15 genes) in response to overexpression of the ZAT12 transcription factor (Vogel et al., 2005).

Cold induction of COL1 and COR27 proceeds through an ABA-independent pathway

Low-level cold induction of some genes, such as RAB18, proceeds through an ABA-dependent pathway (Mantyla et al., 1995). The presence of ABREL motifs in the promoters of COL1 and COR27 raised the possibility that their cold induction also proceeded through an ABA-dependent pathway. To test this, we first determined whether the genes were induced in response to ABA treatment and found that they were not (not shown). Next, we compared cold-induction of COL1 and COR27 in wild-type plants and in mutant plants carrying either the aba1 or abi1 mutations which render the plants ABA deficient or ABA insensitive, respectively (Koornneef et al., 1982, 1984). The results indicated that the ABA mutations had little if any effect on the cold-induced expression of COL1 or COR27 (Figure 2). We thus concluded that cold induction of these two genes proceeded through an ABA-independent pathway.

Figure 2.

COL1 and COR27 are cold induced through an ABA-independent pathway.
Plants were grown for 10–14 days under continuous light at 24°C and then transferred to 0°C for the times indicated. Plants were harvested and total RNA was isolated for semiquantitative RT-PCR using ACT3 as a normalization control as described in the Experimental procedures. The number of PCR cycles was: COR15a, 26; COL1, 33; COR27, 28; ACT3, 27. Wt refers to the background ecotype of the mutants, Ler-0.

The promoter regions of COL1 and COR27 include cold-responsive cis-acting DNA regulatory elements

Gene fusion experiments were conducted to determine whether the promoters of COL1 and COR27 were able to impart cold-regulated gene expression. The 5′ sequences for COL1 and COR27, ranging from the putative ATG translational start site upstream to −1140 bp and −1069 bp, respectively, were fused to the luciferase (luc) reporter gene and the constructs were transformed into Arabidopsis and tested for cold induction. The results indicated that both the COL1p-1140:luc (Figure 3a, top construct) and the COR27p-1069:luc (Figure 4a, top construct) constructs were highly induced in response to low temperature, indicating the presence of cold-responsive cis-acting elements in both promoters.

Figure 3.

 The COL1 promoter region includes cold-responsive DNA regulatory elements.
(a) Deletions of the COL1 promoter with conserved elements shown as colored boxes. The bent arrow indicates the putative start of transcription (−174 from ATG). Stable T2 lines were either grown at 24°C for 10–14 days (red bars) or grown at 24°C then transferred to 0°C for 5 days (blue bars). Luciferase activity was measured as described in the Experimental procedures. Error bars indicate SE. Reading from top to bottom, the number of independent lines and percentage showing luminescence are: = 16, 94%; = 9, 89%; = 11, 96%; = 13, 38%; = 19, 11%; = 15, 0%; = 11, 0%; = 36, 86%; = 26, 88%; = 10, 90%.
(b) Mutagenized elements in the −646 COL1 promoter. Boxes (1, 2, 3, 4 and 5) that were mutagenized are indicated by a red X. The bent arrow indicates the putative start of transcription (−174 from ATG). Treatment and luciferase activity determination is as in (a). Error bars indicate SE. Reading from top to bottom, the number of independent lines and percentage showing luminescence are: = 22, 96%; = 14, 85%; = 11, 92%; = 15, 67%; = 15, 27%; = 18, 17%; = 20, 0%; = 11, 36%; = 12, 25%; = 18, 17%.
(c) Sequences of the conserved elements shown as colored boxes in (a) and (b).

Figure 4.

 The COR27 promoter region includes cold-responsive DNA regulatory elements.
(a) Deletions of the COR27 promoter with conserved elements shown as colored boxes. The bent arrow indicates the putative start of transcription (−81 to ATG). Stable T2 lines were either grown at 24°C for 10–14 days (red bars) or grown at 24°C then transferred to 0°C for 5 days (blue bars). Luciferase activity was measured as described in the Experimental procedures. Error bars indicate SE. Reading from top to bottom, the number of independent lines and percentage showing luminescence are: = 24, 92%; = 15, 100%; = 15, 93%; = 10, 100%; = 31, 87%; = 8, 50%; = 19, 0%; = 37, 81%; = 20, 95%; = 39, 0%.
(b) Boxes (1, 2 and 3) that were mutagenized are indicated by a red X. The bent arrow indicates the putative start of transcription (−81 to ATG). Treatment and luciferase activity determination is as in (a). Error bars indicate SE. Reading from top to bottom, the number of independent lines and percentage showing luminescence are: = 12, 100%; = 11, 91%; = 12, 92%; = 39, 0%.
(c) Sequence of the conserved elements shown as colored boxes in (a) and (b).

Promoter deletion experiments were conducted to identify regions of the COL1 and COR27 promoters that contained cold-responsive regulatory elements. The 5′ deletions of the COL1 promoter indicated that trimming the promoter fragment from −1140 to −823 or −646 had little if any effect on promoter activity (Figure 3a). This eliminated a need for three ABREL motifs, one of which was a G-box, in cold induction of the promoter. Deletion to −533, however, resulted in a complete loss of cold induction, indicating the presence of one or more cold-responsive elements. In this region of the promoter (−646 to −533) there were two ABREL motifs, an EE motif, two EEL motifs and a CBS motif. Deletions of the promoter sequence to −440, −306 and −216 did not result in recovery of cold responsiveness. The 3′ deletions to −131, −187 or −350 had no effect on the activity of the promoter, indicating a lack of essential cold-regulatory elements in these regions of the promoter, which includes one ABREL motif.

Analysis of the COR27 promoter fusions indicated that a 5′ deletion to −758, which removed a potential G-box, had no effect on cold induction (Figure 4a). Deletions to −642, −499 and −377 – the last removing an EEL sequence – resulted in a decrease of between 25 and 50% in cold-induction as compared to the full-length promoter. Further deletion to −254 resulted in a decrease of about 85% in promoter activity and deletion to −162 resulted in complete loss of activity. The region between −377 and −254 has a sequence resembling an EE, AAAATATCA, and five potential Dof transcription factor recognition sites, AAAG (Yanagisawa, 2002). Dof transcription factors are involved in many aspects of plant growth and development, including regulation of gene expression in response to photoperiod (Sawa et al., 2007). The 3′ deletions indicated that cutting back to −79 had no effect on cold induction; that deletion to −115, which removed a potential G-box, reduced activity by about 40%; and that deletion to −234, which removed one ABRE, one EEL and two EE motifs, resulted in complete loss of low-temperature responsiveness.

EE and EEL motifs are directly involved in cold-induction of the COL1 and COR27 promoters

Taken together, the deletion analyses indicated that there are cis-acting elements in the promoters of both COL1 and COR27 that impart cold-induced gene expression, and implicated EE, EEL and ABREL/G-box elements as having a role in this regulation. This was tested further by site-directed mutagenesis. Mutagenesis of the COL1 promoter was conducted in the context of the COL1p-646:luc construct (the region −646 to the ATG) (Figure 3b). This construct contains the cluster of one EE, two EEL, one CBS and two ABREL motifs that were shown to be essential for cold induction of the promoter (Figure 3a). Mutation of all four A-rich sequences, the EE (box 4), the two EELs (boxes 1 and 5) and the CBS (box 3) resulted in complete elimination of cold responsiveness, indicating a critical role for one or more of these motifs. Mutation of three of the four elements (boxes 1, 3, 4 or 3, 4, 5) also resulted in near complete loss of cold induction, suggesting that the sequences common to these deletions, the EE (box 4) and CBS (box 3) elements, might have a critical role in cold induction. This simple scenario was not the case, as the promoter construct with the double EE/CBS mutation (boxes 3 and 4) was still strongly responsive to low temperature. However, this double mutation did result in a decrease of about 50% in induction level, indicating that one or both of the elements contributed to cold induction. Finally, mutation of either EEL sequence (boxes 1 and 5) alone resulted in about a 25% decrease in cold induction. Taken together, these results indicated that the A-rich elements in the promoter of COL1 function together to bring about cold induction.

Similar results were obtained with the COR27 promoter (Figure 4b). In this case, mutational analysis was conducted in the context of the COR27p-1069/-79:luc construct. The results from the 5′ and 3′ deletion analyses (Figure 4a) indicated that the region between −234 and −162 was required for cold induction. This region included two EE motifs and one EEL motif. Site-specific mutagenesis indicated that a triple mutation removing all three sequences (boxes 1, 2 and 3) resulted in complete elimination of cold responsiveness of the promoter, whereas single mutations leaving either two EE (boxes 1 and 2) or an EE and EEL intact (boxes 2 and 3) had no significant effect on cold induction (Figure 4b). These results are consistent with a model in which the A-rich sequences act together to impart cold-regulated gene expression, though the experiments do not rule out the possibility that the middle EE (box 2) is essential for cold induction.

A role for the ABREL motif in cold induction of COL1

The region of the COL1 promoter that includes the cluster of EE, EEL and CBS elements (−646 to −533) required for cold induction also includes an ABREL motif (Figure 3). To test whether this element is active in cold-induced gene expression, we mutagenized the sequence within the context of the COL1p-646:luc reporter construct and determined the cold responsiveness of the promoter (Figure 3b). The results indicated that elimination of the ABREL motif (box 2) resulted in a reduction of 80% in the activity of the promoter fragment, with little effect on fold induction; i.e. the cold to warm ratio was approximately the same for the wild-type construct (22.9 ± 4.0) and the ABREL mutant construct (19.8 ± 7.8). Thus, the ABREL motif appeared to have a supportive role in the cold responsiveness brought about by the EE/EEL/CBS motif cluster. Adding a CBS mutation (box 3) to the ABREL mutation (box 2) had little effect on cold induction of the promoter fragment, but adding a mutation in one of the EELs (box 5) did, decreasing the cold-induced expression level by another 80%. These results suggest that the ABREL potentiates the activities of the A-rich elements.

Coupling of an EE with an ABREL is sufficient to impart cold-induced gene expression

The results presented above indicated that the A-rich elements play a critical role in cold activation of the COL1 and COR27 promoters and that their activity is amplified by interaction with an ABREL motif. From these results, we could not conclude whether these elements are sufficient for cold induction. To test this, we made an artificial EE/ABREL promoter:luc fusion containing four copies of the EE interspersed with three copies of the ABREL motif, transformed the construct into Arabidopsis and assayed for cold induction. The results indicated that the artificial promoter could impart cold-induced gene expression and that both the EE and ABREL are required for this activity (Figure 5).

Figure 5.

 Evening element (EE) and ABA response element-like (ABREL) motifs act together to impart cold-induced gene expression.
(a) Luciferase activity of an artificial promoter composed of EE and ABREL sequences. Multimerized wild-type and mutant EE and ABREL sequences fused to LUC were transformed into plants. Stable T2 lines were either grown at 24°C for 10–14 days (gray bars) or grown at 24°C then transferred to 0°C for 5 days (black bars). Luciferase activity was measured as described in the Experimental Procedures. Error bars indicate SE. Reading from left to right, the number of independent lines and percentage showing luminescence are: = 16, 31%, = 20, 0%; = 23, 0%. mABREL indicates that all ABRELs were mutagenized and mEE indicates that all EEs were mutagenized.
(b) Multimerized EE and ABREL motifs in artificial promoters. The sequence in parentheses was repeated three times.

Promoters of cold-induced genes are enriched in coupled EE and ABREL motifs

The results above indicated that an EE motif coupled with an ABREL is sufficient to impart cold-induced gene expression. A question raised was whether a coupling of these two elements occurs at a higher frequency in the promoters of cold-regulated genes than it does randomly among all promoters. Bioinformatic analysis indicated that it does (Table 1). The promoters of the 218 CBF-independent cold-induced COS genes were highly enriched for an EE motif combined with an ABREL motif (4.51-fold; P < 0.001). They were also enriched for an EEL motif with an ABREL motif (2.59-fold; P < 0.001) and a CBS element with an ABREL motif (2.18-fold; P < 0.005). The enrichment for these combinations of motifs was even greater in the promoters of the 84 CBF regulon cold-induced COS genes, i.e. an EE motif combined with an ABREL motif (7.11-fold; P < 0.001), an EEL motif with an ABREL motif (3.33-fold; P < 0.001) and a CBS motif with an ABREL motif (4.26-fold; P < 0.001). In both sets of promoters, there was also enrichment for an EE motif with a G-box and an EEL motif with a G-box. Taken together, these results suggest that coupling of an EE or EEL with an ABREL and coupling of an EE or EEL with a G-box motif comprise cis-acting regulatory element pairs that may be responsible for the cold induction of a considerable number of the CBF-independent COS genes: 70 (32%) in the case of an EE/EEL coupled with an ABREL and 31(14%) in the case of an EE/EEL coupled with a G-box. Likewise, these couplings would appear to contribute to the cold induction of a number of CBF regulon COS genes: 36 (43%) in the case of an EE/EEL coupled with an ABREL and 14 (17%) in the case of an EE/EEL coupled with a G-box.

Table 1.   Motifa enrichment in promoters of cold-induced genes
MotifTotal (33 282)CBF independent (218)CBF regulon (84)
GenesEnrichmentχ2P valueGenesEnrichmentχ2P value
  1. aMotif sequences are: ABREL (ABA response element-like), ACGTG; G-box, CACGTG; EE (Evening Element), AAAATATCT; EEL (EE-like), AATATCT; CBS (CCA1-binding site), AAAAATCT.

ABREL18 0961391.176.8<0.01631.3313.38<0.001
G-box5011601.8324.7<0.001272.0617.65<0.001
EE2317422.7747.6<0.1193.1429.05<0.001
EEL103111051.5528.3<0.001431.6014.91<0.001
CBS3313261.200.7<0.5171.968.69<0.005
ABREL + EE 915274.5169.4<0.001177.1187.76<0.001
ABREL + EEL4132702.5973.7<0.001363.3367.98<0.001
ABREL + CBS125872.1810.5<0.005144.2634.38<0.001
G-box + EE377112.2725.4<0.00177.1032.01<0.001
G-box + EEL1592312.973.7<0.001143.3623.19<0.001
G-box + CBS470184.4539.3<0.00164.8815.64<0.001

COL1 and COR27 are regulated by the circadian clock

The EE and CBS elements have roles in circadian-regulated gene expression (Harmer et al., 2000; Michael and McClung, 2002). Thus, we were interested in knowing whether the COL1 and COR27 genes are subject to circadian regulation. To test this, plants that had been grown under a 12-h photoperiod at 24°C were moved to constant light and the COL1 and COR27 transcript levels were determined at various times after dawn, Zeitgeber Time 0 (ZT0). The results indicated circadian regulation of both COL1 (Figure 6a) and COR27 (Figure 6b). However, the phases for the genes were about 12 h out of sync with each other: whereas peak expression of COL1 was at ZT16 and ZT40 COR27 transcript levels peaked at ZT4 and ZT28.

Figure 6.

 Cold induction of COR27 is gated by the circadian clock.
Plants were grown at 24°C under a 12-h photoperiod for 10–14 days and then placed under continuous light at dawn (ZT0). At the indicated time points, plant tissue was harvested, RNA was isolated and the transcript levels for the indicated genes were determined; the resulting values are plotted at the corresponding time point (closed triangles). Also at the indicated times, plants were transferred to 0°C for 6 h under continuous light and then plant tissue was harvested, RNA isolated and transcript levels were determined; the values are plotted at the time point at which the plants were transferred to low temperature (open squares). Transcript levels were measured on total RNA by semiquantitative RT-PCR as described in the Experimental Procedures. Error bars indicate SE. The slashed boxes represent subjective night.
(a, d) COL1 transcript levels in wild-type plants.
(b, e) COR27 transcript levels in wild-type plants.
(c) Luciferase transcript levels in COR27p-1069:luc.
Note: the data in (a) and (b) are also presented in (d) and (e), respectively, to more easily compare the relative transcript levels in the warm- and cold-treated plants.

Cold induction of COR27 is gated by the clock

Cold induction of the CBF1–3 genes is gated by the circadian clock (Fowler et al., 2005). To determine whether cold induction of COL1 or COR27 is gated, plants were grown at 24°C under a 12-h photoperiod and then exposed to constant light starting at dawn (ZT0). At various times, plants were moved to 0°C in the light for 6 h and the transcript levels for COL1 (Figure 6d) and COR27 (Figure 6e) were determined. The results did not reveal gating of cold induction for COL1, but strong gating was observed with COR27. The peaks of cold induction for COR27 were at about ZT4 and ZT28 and the troughs at about ZT16 and ZT40, the same times for peaks and troughs for circadian regulation of COR27 at warm temperature (the peak levels attained upon cold treatment were more than 10-fold higher than at warm temperature) (Figure 6e). To determine whether this regulation occurred at the transcriptional level, the COR27 promoter:luc fusion (COR27p-1069:luc) was tested for gating (Figure 6c). As with the endogenous transcripts, cold-induced accumulation of the reporter transcripts was gated, with the peaks of cold responsiveness at ZT4 and ZT28 and troughs at ZT16 and ZT40.

Overexpression of CCA1 down-regulates cold induction of COR27

The CCA1 transcription factor is an important component of the circadian clock (Wang and Tobin, 1998). It binds to EE elements in the promoter of the TOC1 gene and negatively regulates its expression (Alabadi et al., 2001). In turn, TOC1 is a positive regulator of CCA1. This CCA1/TOC1 regulatory loop is a key component of the central oscillator. Given that both the COL1 and COR27 promoters have EE motifs, it is possible that both might be affected by constitutive overexpression of CCA1. This was not the case for COL1 (not shown), but was for COR27 (Figure 7); the levels of COR27 transcripts at 8, 24 and 48 h after transfer to low temperature were reduced by about 40% in plants overexpressing CCA1.

Figure 7.

 Cold induction of COR27 is down-regulated by CCA1 overexpression.
Plants were grown at 24°C for 10–14 days under continuous light and placed at 0°C for the times indicated. Semiquantitative RT-PCR was used to measure COR27 transcript levels in total RNA. Wild-type Col-0 is shown by gray bars and CCA1 overexpresser by black bars. Error bars indicate SE.

Discussion

With the exception of the CBF cold response pathway, little is known about the regulatory networks that control gene expression in response to low temperature. To further an understanding of cold response regulatory pathways, we studied the expression of COL1 and COR27, two genes that we show are induced in response to low temperature through a pathway that is independent of both CBF and ABA. Mutational analysis indicated that cold induction of these genes requires the action of EE/EEL motifs, establishing that these elements are not only important in circadian regulation but also cold induction. Further, we show that the action of these elements is potentiated by the ABREL motif and that these two elements, coupled together, are sufficient to impart cold-regulated gene expression. This study also establishes that both COL1 and COR27 are subject to circadian regulation at warm temperatures and that cold induction of COR27 is gated by the clock. Bioinformatic analysis indicated that the coupling of EE/EEL motifs with the ABREL motif is highly enriched in cold-induced genes, suggesting that an EE–ABREL regulatory pathway may have a substantial role in configuring the low-temperature transcriptome.

The EE–ABREL cold response pathway

The rapidly induced COL1 and COR27 genes have multiple EE and EEL elements in their promoters. The results presented indicate that these elements work in concert with each other to mediate cold induction and that their action is potentiated by the ABRE motif. Thus, we propose that cold-regulated gene expression in Arabidopsis includes action of an EE–ABREL cold response pathway. A challenge now is to determine which transcription factors bind to these elements and how they interact to impart cold-induced gene expression. In the case of the EE/EEL motifs, they are likely to bind one or more MYB transcription factors. Indeed, the MYB transcription factors CCA1 and LHY bind to the EE present in the TOC1 promoter and repress TOC1 expression (Alabadi et al., 2001). The observation that constitutive overexpression of CCA1 reduces cold induction of COR27 is consistent with the CCA1 protein binding to the EE/EEL sequences in the COR27 promoter and down-regulating expression. However, the mutational analyses indicate that the EE/EEL motifs in COL1 and COR27 impart positive regulation. Indeed, there is evidence that the EE motif and the CCA1 and LHY transcription factors can elicit both positive and negative effects on gene expression (Harmer and Kay, 2005). Protein microarray analysis has led to the identification of 41 transcription factors that can bind to the EE, 10 of which are regulated by the clock (Gong et al., 2008). A reasonable hypothesis is that one or more of these participate in cold induction of COL1 and COR27.

The ABRE imparts ABA-regulated gene expression by binding members of the ABF family of bZIP transcription factors (Choi et al., 2000; Jakoby et al., 2002). We therefore considered the possibility that the ABREL motifs present in the promoters of COL1 and COR27 might involve action of ABA and ABFs. However, the data presented indicate that cold induction of COL1 and COR27 proceeds through an ABA-independent pathway. Thus, the ABRELs in the promoters of COL1 and COR27 are not likely to involve the action of ABFs. However, there are least 75 bZIP proteins, factors that preferentially bind to either the A-box (TACGTA), C-box (GACGTC) or G-box (CACGTG), all of which have the ACGT core sequence (Jakoby et al., 2002). This family of transcription factors is active in many processes including development, pathogen defense and abiotic stress tolerance (Jakoby et al., 2002), and may also participate in our proposed EE–ABREL cold regulatory pathway.

One additional point is that the promoters of the cold-induced CBF regulon genes and the CBF-independent cold-induced genes are not only enriched in the EE/EEL–ABREL motif pair but are also enriched in the EE/EEL–G-box pair. Our data provide no evidence that any of the G-boxes present in the promoters of COL1 or COR27 have a role in cold-regulated gene expression. However, a G-box in the ADH1 gene has been shown to be important in cold-regulated gene expression (Dolferus et al., 1994). The G-box sequence, CACGTG, includes both the core consensus bZIP-binding site, ACGT, and the core consensus binding site for bHLH proteins, CANNTG (Meshi and Iwabuchi, 1995). Thus, it is possible that there is also an EE–G-box cold response pathway that combines the action of MYB factors binding at EE/EEL elements interacting with bHLH factors binding at G-boxes.

Circadian regulation of COL1 and COR27

In Arabidopsis, the central circadian oscillator is composed of a regulatory feedback loop involving three proteins: CCA1, LHY and TOC1 (Gardner et al., 2006; McClung, 2006). CCA1 and LHY are negative regulators of TOC1 and TOC1 is a positive regulator of CCA1 and LHY. Both the CCA1 and LHY genes are induced in the early morning and down-regulate TOC1, which subsequently results in down-regulation of CCA1 and LHY, and consequently up-regulation of TOC1. This cycling has a period of about 24 h and involves a number of additional feedback loops that influence the period and stability of the clock (Gardner et al., 2006; McClung, 2006).

As noted above, CCA1 and LHY are MYB transcription factors that bind to the EE and CBS motifs. The presence of the EE, EEL and CBS sequences in the promoters of COL1 and COR27 suggested that they might be subject to circadian regulation at warm temperature. This was shown to be true. However, the phases of the two genes were opposite to each other; whereas peak expression of COL1 was at ZT16, peak expression of COR27 was at ZT4. The molecular basis for this difference may involve different outputs from interactions of the transcription factors binding the EE, EEL, CBS, ABREL and G-box sequences. Indeed, both TOC1 and the pseudo-response regulator PRR9 have EE motifs in their promoters, but TOC1 has peak expression in the evening, whereas PRR9 expression peaks in the morning (Matsushika et al., 2000). Such differences in motif interactions may also account for the fact that cold-induced expression of COR27 is gated by the clock, but cold-induced expression of COL1 is not.

Integrating cold- and clock-regulated gene expression

Our results indicate that the EE/EEL and ABREL (and possibly G-box) motifs present in the promoters of COL1 and COR27 function together to induce gene expression in response to low temperature. From precedent, the EE, EEL and CBS motifs presumably mediate the circadian regulation of these genes in warm temperatures. In the case of COL1, cold- and circadian-regulated expression would appear to be independent of each other as cold induction of COL1 is not gated by the clock. In contrast, cold and circadian regulation of COR27 are integrated; cold induction of the gene is gated by the clock.

How might this integration be accomplished? One relatively simple model would be as follows (Figure 8). There is a MYB family transcription factor that binds to the EE/EEL motifs in the promoter of COR27 and acts as a positive regulator. This transcription factor is under circadian regulation, having a peak protein level at ZT4 and a trough at ZT16; the same as for circadian-regulated expression of COR27. The cycling of the level of this transcription factor at warm temperatures would account for the low-level circadian regulation observed with COR27 at warm temperatures. Further, the action of this MYB factor is potentiated by a bZIP transcription factor that binds to the nearby ABREL. This factor is inactive at warm temperatures, but is active at cold temperatures due to a post-translational modification; for instance, phosphorylation due to cold activation of a MAP kinase pathway (Nakagami et al., 2005). At ZT4 in the cold, when the MYB protein is at high levels and the bZIP transcription factor is active, there is robust expression of COR27. In contrast, at ZT16 in the cold, the MYB factor is at low levels and the bZIP protein alone is able to elicit only low-level induction of COR27. When plants are grown under constant light, the MYB factor is present at elevated levels so that when the temperature drops and the bZIP factor is activated there is robust induction of COR27. A test of this and alternative models should provide significant new insight into how plants integrate two of the most important inputs into plant growth and development – low temperature and the circadian clock.

Figure 8.

 A model integrating cold and clock regulation of COR27.
MYB (red sphere) and basic region/leucine zipper motif (bZIP; blue sphere) transcription factors bind to the Evening Element (EE)/Evening Element-like (EEL) (yellow/orange boxes) and ABA response element-like (ABREL) (green box) regulatory elements, respectively, in the promoter of COR27. The MYB transcription factor (TF) is active at both warm and cold temperatures, but the amount of protein is regulated by the clock: at ZT4 it is high; at ZT16 it is low. In contrast, the level of the bZIP TF is not subject to circadian regulation, and thus remains constant, but the activity of the protein is affected by temperature. In this model, the protein is inactive at warm temperatures, but becomes active in the cold due to phosphorylation (red sphere on bZIP factor) by a cold-activated protein kinase. Integrating the temperature and the clock leads to high-level transcription of the COR27 gene in response to low temperature at ZT4 (depicted by large arrow) and low-level transcription under the other three combinations of temperature and time.

Experimental procedures

Plant growth and treatments

Seeds of Arabidopsis thaliana (ecotype Col-0 unless otherwise stated) were sterilized in 40% bleach with 0.02% Triton X-100 and stratified at 4°C for 4 days prior to plating on Gamborg’s B-5 medium (Caisson Laboratories, http://www.caissonlabs.com/) containing 2% (w/v) sucrose and solidified with 0.8% Phytoblend agar (Caisson Laboratories). Plants were grown in constant light (110–130 μmol m−2 sec−1) at 24°C unless otherwise stated. For circadian experiments, plants were entrained for 10–14 days under a 12-h photoperiod (110–130 μmol m−2 sec−1) at 24°C prior to transfer to constant light at subjective dawn (ZT0). Low-temperature treatment was performed by placing plates at either 0°C or 4°C at 20–40 μmol m−2 sec−1. Plant transformation was performed using the floral dip method with Agrobacterium tumefaciens GV3101 (Clough and Bent, 1998). Transformants were selected on plates containing 50 μg ml−1 kanamycin.

RNA extraction and RT-PCR

RNA was extracted from 10–14-day-old seedlings using the RNeasy Plant Mini Kit (Qiagen, http://www.qiagen.com/) according to the manufacturer’s instructions. Complementary DNA was produced by use of the Reverse Transcription System (Promega, http://www.promega.com/) with random hexamers according to the manufacturer’s instructions using 5 μg total RNA. The cDNA samples were normalized to Actin3. Polymerase chain reaction was performed under the following conditions: initial denaturation of 2 min at 95°C, 24–36 cycles of 15 sec denaturation (95°C) and annealing (57°C) and 35 sec elongation (72°C). The primers used are shown in Table S2. All primers were designed to work at an annealing temperature of 57°C and, with the exception of the CBF2 primers, designed with one intron-spanning primer to prevent amplification of possible contaminating genomic DNA. The number of cycles was optimized to be in the linear range of detection in every experiment but was usually 27, 29, 25, 28 and 33 for ACT3, CBF2, COR15a, COR27 and COL1, respectively. Quantification was done as described previously (Mikkelsen et al., 2003). Briefly, PCR products were separated by agarose gel electrophoresis in an ethidium bromide containing gel. Ultraviolet images were obtained using a Gel Doc system (Bio-Rad, http://www.bio-rad.com/) at subsaturation settings. Bands were quantified using Quantity One (Bio-Rad). The SE bars were derived from analyses of three biological replicates.

Analysis of luciferase activity

T2 lines transformed with promoter–luciferase constructs were sterilized and plated out on selective medium. About 25 seedlings from each transgenic line (there were approximately 12 independent lines for each construct) were plated out on agar media and sprayed with luciferin as described by Chinnusamy et al. (2003) to measure luciferase activity at 24°C and after 5 days at 0°C, a time period that was found to give the highest levels of reporter gene expression (Figure S3). Imaging of bioluminescence was performed as recommended by Chinnusamy et al. (2003) using a Berthold NightOWL low-level luminescence imaging camera (EG&G Wallac, http://www.bertholdtech.com/). Luciferase activity was measured for each independent line and the value averaged per seedling (i.e. total activity per plate was divided by the number of seedlings on the plate). The luciferase activity reported for a given construct is the average luminescence for each transformed line using only those plates with seedlings that produced luminescence. The error bars show the variation between independent lines.

Promoter constructs

Approximately 1-kb promoter:luciferase (luc) constructs were made in a modified pBI101 vector (pBI101luc), in which the GUS gene had been replaced by the firefly luciferase coding sequence containing one intron. A modified version of this vector (pBI101luc-35Smin) was produced, in which a minimal CaMV35S promoter was inserted into the XbaI and XhoI sites of pBI101luc. The 1140 bp and 1069 bp fragments of the COL1 (At5g54770) and the COR27 (At5g42900) promoters, respectively, ending in the 5′ untranslated region (UTR) right before the start ATG codon, were amplified from genomic DNA with the primers C1 and C2 for the COL1 promoter and with primers P1 and P2 for the COR27 promoter (see Table S3 for primer sequences). These fragments were digested with XhoI and BamHI and directionally cloned into a similarly digested pBI101luc vector to produce the COL1:luc and COR27:luc constructs, respectively. The 5′ and 3′ deletions of the COL1 and COR27 promoters were generated by PCR as described in Appendix S1. Site-directed mutagenesis was performed on the 646-bp 5′ deletion and the 79-bp 3′ deletion of the COL1 and COR27 promoters, respectively, as described in Appendix S1. The synthetic EE–ABRE promoter constructs were made by annealing the pairs of oligonucleotides shown in Table S5, which generated HindIII and XbaI overhangs, and then by ligating these fragments into pBI101luc35Smin digested with HindIII/XbaI.

Acknowledgements

The authors would like to thank Sarah Gilmour, Marlene Cameron and Karen Bird for their help in preparing the manuscript; Amanda Erica Harris and Justyne Eliza Matheny for excellent technical assistance; and Eva Farré for a critical read of the manuscript. The authors would also like to thank Diane Constan for the vector used to make the promoter constructs. This work was supported by an FP6 Marie Curie Outgoing International Fellowship (514432) to MDM and grant support to MFT from the NSF Plant Genome Project (DBI 0110124 and DBI 0701709), the Department of Energy (DE-FG02-91ER20021) and the Michigan Agricultural Experiment Station.

Ancillary