The sfr6 mutant of Arabidopsis displays a deficit in freezing tolerance after cold acclimation. We previously observed that the transcripts of three cold-, ABA- and drought-inducible genes, each having a C-repeat motif or the drought-responsive element (CRT/DRE) in its promoter, failed to normally accumulate in this mutant. We now report that the effects of sfr6 upon transcript levels are reflected in the levels of the encoded proteins, confirming that the cold-inducible protein expression is affected by the sfr6 mutation. Using microarray analysis, we found not only that this effect may be general to cold-inducible genes with CRT/DRE promoter elements, but also that it extends to some other genes whose promoters lack a CRT/DRE element. The role of the CRT/DRE has been empirically tested by use of a synthetic promoter, confirming that the CRT/DRE is sufficient to confer the sfr6 effect upon expression. Tolerance of osmotic stress was also found to be reduced in sfr6, consistent with a role in osmotic stress tolerance for the cold-, ABA- and drought-inducible genes whose expression is affected by the sfr6 mutation.
Cold acclimation is the term used to describe the molecular and physiological changes that occur in some plant species in response to low positive temperatures, allowing them to better withstand the exposure to freezing temperatures (Levitt, 1980). During the acclimation period, many genes are expressed (Hughes and Dunn, 1996; Thomashow, 1999) and cellular changes are observed, including alterations in the membrane composition (Steponkus, 1984). Many of the genes expressed during cold acclimation are also inducible by drought stress, and are likely to play a role in protection against cellular dehydration, which occurs during both freezing and drought conditions (Steponkus, 1984). Arabidopsis, like many temperate species, is capable of cold acclimation and provides a useful model for the study of this phenomenon (Thomashow, 1994). Despite the discovery of increasing numbers of cold-regulated genes, few proteins encoded by such genes have been shown to affect the ability to cold acclimate (Thomashow, 1998; Wanner and Junttila, 1999). Notable exceptions include the genes encoding the DREB1/CBF family of transcription factors, expression of which has been shown to cause an increase in freezing and drought tolerance (Jaglo-Ottosen et al., 1998; Liu et al., 1998). The isolation of mutants such as the sensitive to freezing (sfr) mutants of Arabidopsis (Warren et al., 1996) can aid in understanding the genes and proteins necessary to achieve tolerance of freezing temperatures.
The sfr mutants are identified on the basis of their specific failure to gain freezing tolerance after cold acclimation treatment (Warren et al., 1996) and as such are not sensitive to low, non-freezing (i.e. chilling) temperatures. We have previously shown that the sfr6 mutant shows marked reduction in cold-induced cold-regulated (COR) gene mRNA accumulation (Knight et al., 1999). Amongst the genes affected were KIN1/2, LTI78 and COR15A, all of which contain the C-repeat motif or the drought-responsive element (CRT/DRE) promoter sequence motif (Yamaguchi-Shinozaki and Shinozaki, 1994). Expression of this group of genes is activated by binding of the CBF (DREB1) and DREB2 families of transcription factors to the CRT/DRE element (Liu et al., 1998). Two other cold-inducible genes with promoters that do not contain this motif were shown to be expressed normally in sfr6 in response to cold (Knight et al., 1999). It was thus possible that the sfr6 mutation specifically affected the expression of CRT/DRE-containing genes. The conclusion from this previous work was that the effect of sfr6 upon the expression of certain cold- and drought-regulated genes must be as a result of alterations in the transcription, the transcript processing or the transcript stability (Knight et al., 1999). It was not known whether the effect of the sfr6 lesion was limited to effects on CRT/DRE-containing promoters (i.e. whether the sfr6 lesion might not affect genes with other promoter elements also), or whether the effect was on the CRT/DRE itself or on other elements present in the promoters of the same genes (Knight et al., 1999).
It is possible that the ability to cold acclimate is reduced in sfr6 because of its failure to express certain cold-regulated genes and a consequent failure to accumulate the corresponding proteins. Previous comparisons of boiling-soluble protein profiles in cold-treated wild-type and sfr6 plants have, however, failed to reveal any differences (McKown et al., 1996). The reduced expression of genes in response to osmotic stress (Knight et al., 1999) might be expected to affect protein levels, but neither this nor the osmotic tolerance of the sfr6 mutant has been investigated to date.
In this study, we tested the hypothesis that the sfr6 mutation specifically affects the expression of genes containing the CRT/DRE motif, and examined the mechanism by which the expression of these genes is affected. We examined levels of protein products from two genes downregulated in sfr6, and tested the osmotic stress sensitivity of this mutant.
Microarray analysis of gene expression in the sfr6 mutant
In order to examine the extent of the effect of the sfr6 mutation upon the gene expression, we performed a microarray analysis, comparing samples from cold-treated sfr6 and wild-type plants. Taking a cut-off of an apparent downregulation of twofold or more, it was found that 50 different genes were downregulated in the sfr6 mutant. Because all the genes that had previously been observed to be downregulated in sfr6 contained a CRT/DRE element in their promoters (Knight et al., 1999), 600 bp of upstream sequence of each of the downregulated genes were scanned for the presence of the 8 bp motif (A/G)CCGACAT, allowing one or zero mismatch outside the core CCGAC motif: CRT/DRE elements were recognised in 12 of the 50 promoters (Table 1;Figure 1). Table 2 lists the genes which do not contain a CRT/DRE element in their promoters, but which are downregulated in sfr6. Ten of these 38 genes contained an ABRE element in their promoter with up to a single mismatch, compared to 4 out of 12 amongst the CRT/DRE-containing genes (Table 1). According to the microarray data, downregulation of the CRT/DRE-containing genes affected by sfr6 ranged from 2–8.5-folds. Equivalent analysis was performed on the promoters of 50 genes taken at random from the large set shown to be unaffected by the sfr6 mutation: only two contained a recognised CRT/DRE (Table 1). A Chi-square test performed with this data indicates a 99% confidence level for rejecting the null hypothesis (i.e. equal distribution of CRT/DRE between the two sets of promoters). This indicated that the sfr6 mutation is, although not exclusively (as 76% of the downregulated genes did not contain this motif), affecting the expression of the genes containing the CRT/DRE motif.
Table 1. Genes containing CRT/DRE elements in the promoter that are downregulated in sfr6
The table lists 12 CRT/DRE-containing genes downregulated in sfr6 and two such genes not downregulated in sfr6, as determined by microarray analysis. ‘Position’ relates to the position relative to the ATG start codon, measured from the first C residue in the CCGAC core motif. ‘Identity’ refers to the number out of eight nucleotides matching exactly the (G/A)CCGACAT consensus. ‘Sequence’ is the actual sequence of the motif in each case (the CCGAC core sequence is in bold, any mismatches are shown in lowercase). ‘Spots’ relate to the number of individual spots on the microarray corresponding to each gene per individual duplicate microarray print.
Expression reduced in sfr6
Putative 3-methylcrotonyl-CoA carboxylase
Cold-regulated dehydrin-like protein (RD17)
Cold-regulated dehydrin-like protein (ERD10)
Putative glycine-rich protein
Similar to cold acclimation protein WCOR413 [Triticum aestivum]
Putative membrane channel protein
Putative serine carboxypeptidase precursor
Cold-regulated protein COR6.6 (KIN2)
Expression unaffected in sfr6
Chlorophyll a/b-binding protein CP29
Table 2. Genes not containing CRT/DRE elements in the promoter, that are downregulated in sfr6
Table lists the 38 genes downregulated in sfr6 as determined by microarray analysis. ‘Spots’ relate to the number of individual spots on the microarray corresponding to each gene per individual duplicate microarray print.
Ribulose bisphosphate carboxylase small chain 3b precursor
Outer membrane lipoprotein-like
Photosystem I reaction centre subunit psaN precursor (PSI-N)
Direct test of CRT/DRE element regulation in sfr6
We constructed a CRT/DRE reporter by transcriptional fusion of the luciferase (LUC) coding region to a promoter containing a fourfold repeat of the CRT/DRE element. This was genetically transformed into wild-type plants, which were then crossed with sfr6. We compared expression in the F2 homozygous sfr6 and the F2 wild-type siblings, both offsprings of the same original F1 parent. As can be seen in Figure 2, expression of this construct in response to both cold (Figure 2a) and osmotic stress (Figure 2b) was significantly reduced in the sfr6 mutant. As the promoter here is a minimal construct with no possibility of containing other elements that may be present in the native promoters affected by sfr6, we concluded that the CRT/DRE element is indeed directly involved in the interaction(s) modulated by the SFR6 protein.
The effect of low temperature treatment on COR protein levels
The effect of the sfr6 mutation on transcript levels implies that the expression of cold-induced genes is reduced in the mutant, an implication that is pivotal to discussions of how the mutation produces its freezing-sensitive phenotype. However, this has previously not been tested by observations of the protein levels. Therefore, we tested protein extracts from the stress-treated plants with a primary polyclonal antibody specific to COR6.6 and one to COR78 protein. The DNA probe and polyclonal antibody used for detection of COR6.6 mRNA and COR6.6 protein, respectively, did not distinguish between COR6.6 mRNA/COR6.6 protein (also known as KIN2) and the closely related KIN1 mRNA/KIN1 protein (Sarah Gilmour, personal communication). Both the COR6.6 and KIN1 promoters contained the CRT/DRE element. We have referred to COR6.6 mRNA and COR6.6 protein throughout for simplicity. For protein measurements, parallel gels, using aliquots from the same protein preparation, were stained with Coomassie blue to show the total protein. In the case of RNA gel blots, the levels of β-tubulin transcripts were employed as the loading controls.
Figure 3 shows that in wild-type plants, COR6.6 protein levels are detectable after 12 h at 4°C, reaching a maximum after 48 h and declining slightly at 72 h. In sfr6, however, expression of this protein was noticeable only after 48 h, and this was at a low level compared to the wild type (Figure 3). At 20°C, the COR6.6 protein was barely detectable in the wild-type plants after 72 h and was undetectable in sfr6 at this temperature. Similarly, COR78 protein expression in the wild type was faintly detectable after 6 h and then it increased, reaching a maximum after 24 h and decreasing at 48 and 72 h. Again, in sfr6, expression levels did not increase noticeably until after 24 h of cold treatment and were considerably lower than those observed in wild-type plants (Figure 3). No expression of the COR78 protein was detected in either wild-type or sfr6 plants grown at 20°C. Thus, the known effects of sfr6 upon transcript levels were closely reflected by changes in the levels of the cognate proteins, at least for the two COR proteins tested.
Effect of ABA treatment on COR transcript levels and protein expression
Levels of mRNA for COR78 and COR6.6 have been shown to increase in response to ABA and drought as well as to cold treatments (Baker et al., 1994; Kurkela and Borg Franck, 1992; Mäntyläet al., 1995). We have previously shown that sfr6 substantially lowers the accumulation of COR78 and COR6.6 mRNA in response to a brief (3 h) exposure to ABA or mannitol (Knight et al., 1999). However, this could be the result of a delayed response rather than an overall reduction in the amount of expression in the long term, thus having little physiological relevance. In the present study, we investigated the effect of more prolonged treatments with mannitol (producing osmotic stress) and ABA on mRNA and protein levels in the sfr6 mutant.
The pattern of transcript levels of COR6.6 and COR78 is shown in Figure 4. In 7-day-old wild-type plants treated with 100 µm ABA, high levels of COR6.6 expression were detected after 6 h of treatment. Levels of expression fell slightly over the following period up to or until 48 h. A small amount of expression was detected after 6 h in 0.1% ethanol (control), but by 48 h, this had fallen to an undetectable level. In sfr6, expression of COR6.6 exhibited similar kinetics, peaking at 6 h, but at each time-point, COR6.6 mRNA levels were much lower in sfr6 than in the wild type (Figure 4). In wild-type plants, COR78 gene expression followed a similar pattern to COR6.6, peaking at 6 h and thereafter reducing gradually over time. Expression was not detected in the control samples taken at 6 and 48 h. In sfr6, COR78 transcript expression did not reach detectable levels at any time-point (Figure 4).
Figure 5 shows that in the wild type, COR6.6 protein was just detectable after 6 h of ABA treatment, but at a level similar to that in the controls. COR6.6 protein levels increased progressively up to 48 h in mannitol (Figure 5a). Detectable COR6.6 protein was just visible in extracts from sfr6 plants treated with ABA for 12 or 24 h, with slightly higher levels observed after 48 h of treatment. No COR6.6 protein was detectable in either of the control samples for sfr6 (Figure 5a). At each time-point, COR6.6 protein levels in sfr6 were significantly lower than those seen in the wild type. COR78 protein was strongly expressed after 6 h of ABA treatment in wild-type plants, attaining maximal levels at 12 h and declining thereafter (Figure 5b). COR78 was detectable in the control samples, but at very low levels (Figure 5b). In sfr6, expression of COR78 protein was very low at all time-points, and no expression was detectable in control treatments (Figure 5b).
Effect of mannitol treatment on COR transcript levels and protein expression
In 7-day-old wild-type plants, a very strong accumulation of COR6.6 transcripts was detected in response to treatment with 0.33 m mannitol, and this was maximal after 6 h (Figure 6). COR6.6 mRNA levels had fallen considerably by 12 h and continued to decrease slightly at 24 and 48 h. COR6.6 expression was detected after 6 h of the control treatment (although at much lower levels than in response to the 6 h mannitol treatment), but the levels fell after 48 h in water (Figure 6). In sfr6, the kinetics of mannitol-induced COR6.6 mRNA expression were similar to those observed in wild type, but absolute levels of transcripts were always much lower in sfr6, and no expression was detected in either of the control treatments (Figure 6). COR78 mRNA levels also peaked at 6 h in wild-type plants, with no expression visible in controls, but transcripts accumulated only to barely detectable levels in sfr6 after 6 and 12 h of mannitol treatment (Figure 6). Levels of β-tubulin transcripts were monitored to control, as before, for the amount of RNA per sample (Figure 6).
COR6.6 protein was detectable in all mannitol-treated wild-type samples, peaking after 12 h and declining slightly at 48 h of treatment (Figure 7). Very low amounts of protein were detectable in the control (uninduced) treatments after 6 or 48 h (Figure 7). COR6.6 protein was barely detectable in any of the sfr6 samples, but was most noticeable after 12 h. Accumulation of COR78 protein in wild-type plants reached a peak at 12 h and decreased slightly after 24 and 48 h. The protein was detected in the wild-type control sample after 6 h of the treatment, but it had declined to undetectable levels after 48 h of the treatment (Figure 7). In sfr6, levels of COR78 protein were much lower than in the wild type, and just detectable from 12 h onwards.
Survival of osmotic stress
The effect of increasing concentrations of mannitol on plants was used to assess the ability of the sfr6 mutant to tolerate osmotic stress and to give a possible indication of its tolerance towards drought stress. Wild-type and sfr6 seedlings floating on water or solutions of mannitol were examined after 3 days for visual symptoms of drought stress. Whereas both wild-type and sfr6 plants appeared green and healthy after 3 days in water, plants of both the types became increasingly chlorotic in increasing concentrations of mannitol (Figure S1). In both cases, the cotyledons became chlorotic at lower mannitol concentrations than the true leaves. The differences in symptoms between wild type and sfr6 were marked most at concentrations of 440 and 550 mm mannitol (Figure S1). At 440 mm mannitol, sfr6 cotyledons were highly chlorotic, leaves were partially chlorotic and growth had been inhibited. In contrast, at this concentration of mannitol, wild-type cotyledons were only slightly chlorotic and leaves were unaffected. At 550 mm mannitol, sfr6 leaves were even more severely affected. In contrast, although wild-type cotyledons now showed severe chlorosis, the leaves appeared to tolerate this treatment.
The chlorosis assay described above could not be easily quantified as sfr6 had already altered pigmentation when compared to wild type before stress treatment; therefore, we also used a germination test to determine the physiological osmotic tolerance of sfr6. Therefore, the effect of increasing concentrations of mannitol upon seed germination was used to assess the ability of the sfr6 mutant to tolerate the osmotic stress. Wild-type and sfr6 seeds were sown and germinated on agar plates containing 0, 200, 300 and 400 mm mannitol. After 7 days, these plates were scored for germination (successful germination being defined as visible emergence of the radicle). As can be seen in Figures 8 and 9, in the wild type, germination was close to 100% without mannitol and did not decrease appreciably when the mannitol concentration was raised to 300 mm. At 400 mm mannitol, germination of wild type was reduced to around 70%. With sfr6, the situation was different: germination on 0 mm mannitol was almost complete, but severely reduced at all concentrations of mannitol, to around 40% at 200 mm and down to less than 10% at 400 mm (Figures 8 and 9).
The sfr6 mutant of Arabidopsis was isolated on the basis of its inability to survive freezing after cold acclimation (Warren et al., 1996). We have previously shown that the low-temperature accumulation of transcripts from three genes, including COR78 and COR6.6, is strongly suppressed in sfr6 (Knight et al., 1999). This reduced transcript accumulation is clearly seen after 3 h of cold treatment and still obvious after 48 h, indicating that the mutation does not simply retard the initial accumulation of COR transcripts (Knight et al., 1999).
Data presented in this paper (Table 1; Figures 1 and 2) demonstrate that the sfr6 mutation affects the stress-induced expression of genes regulated by the CRT/DRE (Yamaguchi-Shinozaki and Shinozaki, 1994). The sfr6 mutation affects the activation of expression via the CRT/DRE element itself (Figure 2). As this occurs in response to both cold and osmotic stress (Figure 2), sfr6 must act to affect the action of both DREB1/CBF and DREB2 transcription factors. This study revealed 12 CRT/DRE-containing genes that are downregulated in sfr6 (Table 1). They included COR6.6, which we have previously shown to be downregulated in sfr6 (Knight et al., 1999). Thus, in addition to COR15A and COR78 (not present on our microarray), it seems that at least 14 CRT/DRE-containing genes are downregulated in sfr6. Seki et al. (2002) reported microarray data in which cold, drought and salt expression of 7000 full-length cDNAs were compared, including 6 out of the 12 CRT/DRE-containing genes that we identified as being downregulated in sfr6 (At1g03090, At1g20440, At1g20450, At2g15970, At2g23120 and AT5g15970). All six of these genes were shown to be upregulated in the wild type in response to the cold and osmotic stress (Seki et al., 2002). This suggests that the CRT/DRE elements predicted in these 12 genes (Table 1) are indeed functional. Of the two CRT/DRE-containing genes that were not downregulated in sfr6, one was present in this microarray data of Seki et al. (AT5g01530; Seki et al., 2002). This gene was shown not to be upregulated either by cold or by osmotic stress. It is possible that in this case, and in the case of the other CRT/DRE-containing gene that is not downregulated in sfr6, the other strong elements in the promoter may mask the CRT/DRE function and give constitutive expression.
Binding of the transcription factor, C-box binding factor (CBF1), to the CRT/DRE element in the COR gene promoter (Liu et al., 1998; Stockinger et al., 1997) can activate cold-induced expression of this group of genes. Gilmour et al. (1998) have demonstrated that CBF1 is a member of a gene family that includes two other homologues, CBF2 and CBF3. The CBF genes have also been described as the DREB1 family (Shinwari et al., 1998). Drought-inducible expression of CRT/DRE-regulated genes is mediated via the same element through interaction with another family (DREB2) of the transcription factors (Liu et al., 1998). Levels of the DREB1/CBF and DREB2 transcription factor transcripts are unaffected in sfr6 (Knight et al., 1999; Openshaw, M., Knight, M.R. and Knight, H., unpublished). This strongly indicates that the effect of the sfr6 mutation on DREB1/CBF and DREB2 action is via a post-transcriptional (to DREB1/DREB2) mechanism. Our microarray analysis shows that suppression of gene expression is not limited to CRT/DRE-containing genes; therefore, the phenotype of sfr6 may be because of suppression of not only CRT/DRE genes but also other genes.
We tested whether suppression of cold-, ABA- and mannitol-induced COR gene expression at the transcript level was reflected at the level of COR protein accumulation. An early attempt to compare protein levels in cold-treated wild type and sfr6 revealed no gross differences in expression patterns between the mutant and the wild type (McKown et al., 1996). Here, we have shown, using antibodies specific to two types of COR protein, COR78 and the COR6.6/KIN1 family, that marked differences do occur in the amount of accumulation of these proteins (Figure 3). This closely mirrors the deficiency of expression of KIN1/COR6.6 transcripts that we have shown previously (Knight et al., 1999) and also in this study. These results show that in response to mannitol (therefore likely also drought), cold and ABA, sfr6 is defective in expression of COR gene transcripts and that this deficiency results in a failure to accumulate COR proteins. To cause these effects on both low-temperature and osmotic stress responses, the wild-type SFR6 protein must be a component involved in the signalling pathways operating via both CBF/DREB1 and DREB2. Our results are consistent with the freezing phenotype of sfr6 being as a result of its deficit in cold-induced proteins, and may indicate therefore that COR proteins are required for freezing tolerance. In all three cases, levels of transcripts and proteins, although much reduced in sfr6 compared to the wild type, were still increased in response to cold/ABA/mannitol treatment, as observed previously for the cold treatment (Knight et al., 1999). This indicates that the mechanism likely affected by the sfr6 mutation is not an effect upon transcriptional initiation itself, but upon the levels of transcriptional activity subsequent to activation. In other words, the kinetics of expression are unaffected, but the absolute magnitude of expression is altered. In terms of the role of SFR6 in signalling, it seems that SFR6 does not regulate the ‘switch’ which induces stress gene expression, but rather controls the ‘flow’ through the pathway once initiated.
The sfr6 mutant is impaired in the accumulation of cold-induced proteins in response to cold and fails to acclimate to freezing temperatures. In view of this fact and the inability of sfr6 to induce COR gene expression and protein accumulation in response to osmotic stress as well as cold, it seemed possible that the mutant might be impaired in its ability to tolerate osmotic stress also, and this was tested in the present study (see Supplementary Material and Figure 9). Challenging seedlings of sfr6 and wild type with increasing concentrations of mannitol (see Supplementary Material) shows that sfr6 is more sensitive to such stress, as determined by greater levels of chlorosis. Wild-type and mutant seeds were sown and germinated on a range of concentrations of mannitol, and germination levels were monitored after 7 days (Figures 8 and 9). A much greater effect of increasing mannitol concentration upon germination and development was seen in the sfr6 mutant than in the wild type (Figures 8 and 9), indicating a reduced tolerance of osmotic stress in sfr6. This data, together with the deficiency in osmotic stress-induced COR protein accumulation, might suggest that COR proteins were necessary for drought tolerance as well as cold acclimation.
In summary, the sfr6 mutant is deficient in CRT/DRE-regulated COR gene expression in response to cold and osmotic stress, resulting in a failure to accumulate COR proteins. The effect of the sfr6 mutation, however, is not exclusively upon genes containing the CRT/DRE motif (Figure 1). The failure to express CRT/DRE-regulated genes correctly probably involves the interaction of CBF1/DREB1 and DREB2 transcription factors with the CRT/DRE promoter element (Figure 2). This is not as a result of reduced DREB1/CBF or DREB2 transcription (Knight et al., 1999; Openshaw, M., Knight, M.R. and Knight, H., unpublished) and thus occurs at the level of a post-transcriptional mechanism. The mutant also shows a reduced tolerance of osmotic stress as well as an inability to cold acclimate (Knight et al., 1999; Figures 8 and 9 and Supplementary Material), suggesting that COR proteins may be required for full freezing and drought tolerance. In future, it will be very interesting to determine the post-translational mechanism by which the SFR6 protein modulates expression in the wild type.
Plant materials and chemicals
Wild-type Columbia (Col-0) and sfr6 (Col-0 background) plants were grown as described previously by Knight et al. (1999) on full-strength Murashige and Skoog medium (Murashige and Skoog, 1962) containing 0.8% (w/v) agar. A 16 h light/8 h dark cycle was used, and the growth chambers were maintained at a constant temperature of 20°C unless stated otherwise.
Microarray production, target labelling, immobilisation, hybridisation and data acquisition/analysis
Expressed sequence tag (EST) libraries were obtained from three sources and used as templates in standard PCR reactions with vector-specific primers to generate probes for microarray production. We selected 6236 probes from the Arabidopsis Biological Resource Center's Minimized Collection of MSU ESTs (Newman et al., 1994), based on their appearance on electrophoretic gels. We also amplified 2976 probes from an NaCl-treated Arabidopsis EST library generated in the laboratory Hans Bohnert (http://www.stress-genomics.org) and four human ESTs as negative controls. Of these combined ESTs, 7898 were deemed to have sequences of sufficient quality for contig assembly using faktory software (Myers and Miller, 1999), which returned 6040 (77%) unique contigs and singletons. The PCR amplificons were purified on glass fibre plates (Millipore, Bedford, MA), eluted in 0.1× 10 mm Tris; 1 m MEDTA pH 8.0 (TE), and dehydrated in a vacuum concentrator. These probes were subsequently dissolved in 2× SSC buffer and transferred to 384 well polypropylene plates. Probes were deposited at a nominal center-to-center spacing of 180 µm on aminoalkylsilane-coated glass microscope slides (Sigma, St. Louis, MO) using slotted, quill-type printing pins (Majer Precision, Tempe, AZ) installed in an Omnigrid arrayer (GeneMachines, San Carlos, CA). Each probe was printed once per array, with two duplicate arrays on each slide. We incorporated Cy3- or Cy5-labelled nucleotides in the first strand reverse transcription product by combining the following reagents and incubating for 90 min at 42°C: 100 µg total RNA template, 1 µl of Cy3- or Cy5-labelled dUTPs (Amersham Pharmacia, Piscataway, NJ), 0.5 µg of anchored-oligodT(23) primer, 0.5 mm of each of dCTP, dGTP, dATP, 0.1 mm dTTP, 1 U of AMV-reverse transcriptase (plus supplied RT buffer; Sigma), and 0.5 U of RNase inhibitor (Sigma, St Louis, MO, USA). The labelling reaction was stopped by the addition of EDTA to 50 mm, and the remaining RNA template was hydrolysed by adding NaOH to 0.2 m and incubating for 5 min at 60°C. This reaction mixture was then cooled on ice, and neutralised with 0.5 volumes of 1 m Tris–HCl (pH 8.0). Unincorporated dye molecules and other reagents were separated from the labelled target by diluting the labelling reaction in 400 µl of TE, then spinning this volume through a YM-30 microconcentrator (Millipore, Bedford, MA, USA).The labelled target was recovered from the YM-30 filter by centrifuging the inverted column in a fresh microfuge tube. After printing, spotted DNA was briefly rehydrated over a 65°C water bath, then snap-dried on a 65°C heat block. These slides were then exposed to 65 mJ of light in a UV-crosslinker. Following a 2 min wash in 1% SDS, the slides were denatured in boiling water for 2 min, and then dehydrated in 100% ethanol for 1 min and dried by centrifugation at 1000 g. For hybridisation, a combined 15 µl volume of labelled targets was added to 2 µl Liquid Block (Amersham Pharmacia, Piscataway, NJ, USA), 1 µl SDS, and 4 µl 20× SSC. This mixture was boiled for 2 min and then cooled on ice before transfer to the surface of a pre-warmed microarray. A hydrophobic coverslip (Z36,591–2, Sigma) was positioned over the microarray, and the microarray was placed in a humidified, pre-warmed, 50 ml screw top tube containing paper towel saturated in 2× SSC. The hybridisation assembly was incubated overnight in air at 65°C. At the completion of hybridisation, we washed the microarrays in 2× SSC, 0.5% SDS for 5 min at 65°C, then in 0.5× SSC at room temperature for 2 min and finally in 0.05× SSC at room temperature for 2 min. Washed slides were dried by centrifugation. We scanned microarrays in a GSI Lumonics ScanArray 3000, and acquired images as TIFF files. Spot boundaries were defined using imagene 4.1 software (BioDiscovery, Los Angeles, CA). For each spot, the local background was defined as the median intensity of all pixels in a region surrounding the spot. Local background was subtracted from the median intensity of all pixels within the spot boundary to yield the net signal. In both channels, spots that had been previously flagged for anomalies and those with low net signals (<256 U) were discarded. All remaining net signal values were transformed into log2 space. Net signals in each channel were normalised relative to each other by first calculating a least sum of squares regression line (y = mx + b) through the full set of filtered, transformed net signals and then by discarding the spots whose residuals were greater than twice the standard error of regression (Finkelstein et al., 2000). This process was reiterated until the removal of outliers resulted in a change of less than 0.001 in the r2 correlation coefficient. Normalisation was completed by applying the final, best-fit regression formula to the original set of filtered, transformed intensity values. Expression ratios, defined as the fold-change in relative signal intensity values between two channels, were calculated as the difference in normalised intensity values in the stress and control channels.
DRE::LUC construct, transformation and luciferase assays
A chimaeric gene was produced, consisting of four copies of the CRT/DRE motif fused to a minimal −70 CaMV 35S promoter to drive a modified firefly LUC coding sequence (Mankin et al., 1997) and ending with a nopaline synthase (NOS) transcriptional terminator sequence. A pair of oligonucleotide primers 5′-GCGCAAGCTTACCGACATTACCGACATTACCGACATTACCGACATGATATCTCCACTGACGTAAG-3′ and 5′-GCTGCAGGAATTCCCGATCT-3′ was used to amplify this construct from the plasmid pLuK07 (Mankin et al., 1997) by the polymerase chain reaction and to introduce HindIII and EcoRI restriction sites at the 5′ and 3′ ends of the gene, respectively. The whole chimaeric construct was then transferred to the Agrobacterium binary vector, pBIN19 (Bevan, 1984), as a HindIII–EcoRI insert, and the resultant plasmid purified from Escherichia coli and then transferred to the Agrobacterium strain C58C1 by a freeze–thaw method (Holsters et al., 1978). Col-0 wild-type plants were transformed with Agrobacterium by the floral dip method (Clough and Bent, 1998), and primary transformants were selected on kanamycin (50 µg ml−1) and timentin (200 µg ml−1) selection plates. Lines expressing the construct were identified using in vitro assays of LUC activity using the LUC assay system (Promega, Madison, USA), and cold and osmotic-inducibility were confirmed (data not shown). Lines expressing the construct were then crossed into sfr6, and LUC activity was monitored in response to cold and osmotic stress in the mutant and wild-type siblings, from the same parent in each case, in the F2 generation.
Plant treatment with cold, ABA or osmotic stress
For RNA gel blot and protein analyses, cold treatments were performed on 7–8-day-old seedlings growing on agar plates, which were transferred to a growth cabinet at 4°C with a 16 h light/8 h dark cycle. After 6, 12, 24, 48 and 72 h, seedlings were harvested in a cold room (at 4°C), and the tissue was transferred to microcentrifuge tubes, frozen in liquid nitrogen and stored at −80°C. Control plates were retained at 20°C in conditions as described above, and tissue was harvested after 0 and 72 h. For the microarray experiment, 7-day-old seedlings were treated at 5°C in tubes of water in a cooled waterbath. For the LUC experiments, 10-day-old plants were grown on plates and transferred to 5°C for 3 days.
ABA treatments were applied by floating seedlings in 5 ml of ABA or control solutions in multiwell culture dishes (Greiner Labortechnik Ltd, Stonehouse, UK) and incubated under normal growth-chamber conditions. A 100 mm stock solution of ABA in ethanol was diluted 1 : 1000 in distilled water to give a working concentration of 100 µm ABA used for the experimental treatments. 0.1% (v/v) ethanol was used as the control treatment.
Osmotic stress treatments for transcript and protein analysis were performed by floating seedlings in 5 ml of water or 0.33 m mannitol (BDH, Poole, UK) contained in transparent perspex multiwell plates (Greiner, UK). These dishes were then returned to the growth chamber until harvest. Mannitol-treated tissue was harvested after 6, 12, 24 or 48 h by removing seedlings from the solution and blotting briefly on tissue paper before transferring it to microcentrifuge tubes and freezing immediately in liquid nitrogen. Control treatment samples were harvested at 6 and 48 h. Approximately 40–50 seven-day-old seedlings were used per treatment in the osmotic-stress experiment. For the LUC experiments, the seedlings were treated by floating them on 0.333 m mannitol for 6 h.
Measurement of gene expression using RNA gel blot hybridisation
Total RNA was prepared from seedling tissue using RNeasy plant RNA minipreps (Qiagen, Dorking, UK), and 10 µg per sample was electrophoresed through 1.0% agarose (Life Technologies, Paisley, UK) formaldehyde gels (Sambrook et al., 1989). RNA was transferred to nylon membranes (Böehringer Mannheim, Mannheim, Germany) by capillary action. Blots were pre-hybridised and hybridised in 50% formamide at 42°C as described previously by Knight et al. (1999). Blots were successively washed twice in each of the following: 2× SSC (1× SSC is 0.15 m sodium chloride, 0.15 m sodium citrate, pH 7), 0.1% SDS followed by 1× SSC, 0.1% SDS and finally 0.1× SSC, 0.1% SDS at 42°C. Probes for COR6.6, COR78 and β-TUBULIN were prepared from the products of PCR using specific primers as described previously (Knight et al., 1996, 1997, 1999). Probes were labelled using 32P-CTP and DNA ‘Ready-to-go’ labelling beads (Pharmacia Biotech, St. Alban's, UK).
Approximately 30 seedlings per sample were homogenised in 100 µl 2× SDS sample buffer (100 mm Tris, pH 6.8; 4% SDS; 20% glycerol and 0.02% bromophenol blue). Phenyl methyl sulphonyl fluoride and dithiothreitol (Melford Laboratories, Ipswich, UK) were added just prior to use to final concentrations of 2 mm and 0.1 m, respectively, and heated to 95°C for 5 min. After centrifugation at 14 000 g for 20 min at 4°C, the supernatant was aliquoted into 10 µl samples, frozen in liquid N2 and stored at −80°C in microfuge tubes. Proteins were separated by discontinuous SDS–PAGE according to the methods of Laemmli (1970) and Schägger and von Jagow (1987) for COR78 and COR6.6 detections, respectively. Separated proteins were transferred to nitrocellulose membrane (Biorad, Hemel Hempstead, UK) by semi-dry blotting (model Trans-Blot SD, Biorad, UK), according to the manufacturer's instructions. The filter was washed in Tris-buffered saline with Tween 20 (TBST) (10 mm Tris–HCl, pH 8.0; 150 mm NaCl; 0.1% Tween 20) to remove any fragments of gel and was blocked for 90 min in TBST containing 5% milk powder (Marvel, Stafford, UK). After rinsing in TBST, the filter was incubated with a primary antibody (a kind gift of S.J. Gilmour and M.F. Thomashow, Michigan State University, USA) in TBST for up to 12 h. The solution was removed and the filter was washed four times in TBST for 10 min each. A goat antirabbit IgG peroxidase conjugate secondary antibody (Sigma, Poole, UK) was dissolved in TBST, according to the manufacturer's instructions, and applied to the filter for 1 h. The filter was washed four times for 10 min each in TBST.
Bands were visualised by chemiluminescence using a Renaissance kit (NEN, Boston, USA) and were recorded using a photon-counting CCD (charge-coupled device) camera (model EDC-02) and image acquisition and processing software (ifs216), all from Photek (St. Leonards-on-Sea, UK).
Osmotic stress and germination experiments
For mannitol survival experiments, seedlings grown as described above were placed in mannitol solutions at 0.22, 0.33, 0.44, 0.55 and 0.66 m or water in individual wells of a 24-well culture dish. Five seedlings were used per 2 ml of liquid in each well with duplicate treatments, and the dishes were then returned to the growth chambers until visual examination was carried out 3 days later. For germination tests, seedlings were germinated and grown on agar media plates (as described above) supplemented with 200, 300 and 400 mm mannitol. After 7 days, germination was assessed and scored (successful germination being taken as emergence of the radicle), and photographs were taken.
We would like to thank Sarah Gilmour and Michael Thomashow of Michigan State University for the kind gift of COR78 and COR6.6 antibodies, and we are also grateful to Sarah Gilmour for helpful discussions regarding their use. This work was funded by grants to MRK, HK and GW from the Biotechnology and Biological Sciences Research Council of the UK and by a grant to DWG from the National Science Foundation (DBI 98–13360, H. Bohnert, P.I.).
(a) Wild-type (Col-0) or sfr6 plants were removed from agar Petri plates and placed in multiwell culture dishes in water, 0.11, 0.22, 0.33, 0.44, 0.55 or 0.66 m mannitol (duplicate treatments) and maintained in a growth chamber at 20°C for 72 h, and then they were photographed.
(b) Enlargement showing central portion of multiwell dish from (a). Comparison of the effects of one set of replicate treatments of 330, 440 and 550 mm mannitol on wild type and sfr6.