| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
First published online July 14, 2006; 10.1104/pp.106.081141 Plant Physiology 142:98-112 (2006) © 2006 American Society of Plant Biologists OPEN ACCESS ARTICLE
Natural Genetic Variation of Freezing Tolerance in Arabidopsis[W],[OA]Max-Planck-Institut für Molekulare Pflanzenphysiologie, D–14424 Potsdam, Germany (M.A.H., D.W., S.F., O.F., D.K.H.); and Biologisches Institut, Abteilung Botanik, Universität Stuttgart, D–70569 Stuttgart, Germany (A.G.H.)
Low temperature is a primary determinant of plant growth and survival. Using accessions of Arabidopsis (Arabidopsis thaliana) originating from Scandinavia to the Cape Verde Islands, we show that freezing tolerance of natural accessions correlates with habitat winter temperatures, identifying low temperature as an important selective pressure for Arabidopsis. Combined metabolite and transcript profiling show that during cold exposure, global changes of transcripts, but not of metabolites, correlate with the ability of Arabidopsis to cold acclimate. There are, however, metabolites and transcripts, including several transcription factors, that correlate with freezing tolerance, indicating regulatory pathways that may be of primary importance for this trait. These data identify that enhanced freezing tolerance is associated with the down-regulation of photosynthesis and hormonal responses and the induction of flavonoid metabolism, provide evidence for naturally increased nonacclimated freezing tolerance due to the constitutive activation of the C-repeat binding factors pathway, and identify candidate transcriptional regulators that correlate with freezing tolerance.
Arabidopsis (Arabidopsis thaliana) is a widely spread small annual weed of the mustard family (Cruciferae) that is native to Europe and central Asia (for review, see Koornneef et al., 2004  ). Its exposure to a wide range of climatic conditions across its latitudinal range of 68°N to 0°, and its adoption as the major model plant, make it ideal for studying natural variation of adaptive traits (Koornneef et al., 2004). Low temperature defines a climatic range boundary for Arabidopsis (Hoffmann, 2002), making temperature an interesting candidate for a habitat parameter that influences the genetic structure of local populations through natural selection. Such selection should be manifest by a relationship between freezing tolerance and the minimum habitat temperature during the growing season.
Freezing tolerance in plants can be afforded at different levels depending on the anticipation of the freezing event. Nonacclimated freezing tolerance measures the plant's capacity to survive a rapid and severe temperature drop to below zero. Acclimated tolerance is higher and reflects the ability of many plants, including Arabidopsis, to increase their freezing tolerance in response to low but nonfreezing temperatures. The mechanism of nonacclimated tolerance is poorly understood, although it has been reported that it may be genetically determined by loci independent of acclimated tolerance in potato (Solanum tuberosum; Stone et al., 1993
Forward and reverse genetics have defined some of the key regulatory components of cold acclimation. A prominent role has been demonstrated for the C-repeat binding factors (CBF) 1, 2, and 3 (Gilmour et al., 1998
Transcript profiling indicated the induction of multiple transcriptional pathways in addition to CBF during cold acclimation (Fowler and Thomashow, 2002
Recently, interest in profiling the system response to low temperature has revealed thousands of transcript (Fowler and Thomashow, 2002
Natural Variation of Freezing Tolerance
To cover a wide array of phenotypic variation, we selected three laboratory and six natural accessions of Arabidopsis representing habitats from 16° to 66° northern latitude. These were C24, Canary Islands (Can), Coimbra (Co), Columbia-0 (Col-0), Cape Verde Islands (Cvi), Landsberg erecta (Ler), Niederzenz (Nd), Rschew (Rsch), and Tenela (Te). For the correlation of phenotypic variance with habitat temperatures, we assigned the origin of C24, Col-0, and Ler to their closest natural relatives Co, Gückingen (Gü), and Landsberg (La), respectively (Schmid et al., 2003
To obtain a quantitative measure of freezing tolerance in these accessions, we determined the temperature at which 50% electrolyte leakage occurred from detached leaves (Rohde et al., 2004 ). Freezing tolerance varied by 5.9°C in acclimated and 3.7°C in nonacclimated leaves (Fig. 1
). The effect of genotype, treatment, and the genotype x treatment interaction on freezing tolerance, as determined by an ANOVA analysis, were all highly significant (P = 4 x 10–14, 3 x 10–19, and 3 x 10–5, respectively). Acclimated tolerance showed continuous variation and correlated significantly (P = 0.007; R2 = 0.67) with the average minimum habitat temperature recorded during the coldest month of the growing season (Fig. 2
). This correlation was not dependent on our classification of accessions as summer or winter annuals, as it was significant even when all accessions were considered either as summer or winter annuals (P = 0.02 and 0.03; R2 = 0.54 and 0.51, respectively). Correlations were also significant (P < 0.03; R2 > 0.6) when Ler, which differs from its natural relative due to the erecta mutation, was excluded from these correlations. These data indicate the robust nature of this correlation. Nonacclimated tolerance was generally similar (–4.75°C ± 0.85°C) except for Te, which was >2°C more tolerant than all other accessions (Fig. 1). The ability to increase freezing tolerance during cold acclimation (acclimation capacity) also varied between accessions. Col-0, Rsch, and Te had high acclimation capacities of 4.5°C to 5.2°C, and C24, Can, Co, Cvi, and Ler had low capacities of 1.7°C to 2.5°C, with Nd intermediate (3.6°C).
Plants were cold acclimated for 14 d, several days past the initial increase in freezing tolerance (Wanner and Junttila, 1999 ) yet before any new leaves had been formed. This time point should represent steady-state levels of transcripts and metabolites required to maintain the cold acclimated state. Plants were either harvested for nonacclimated measurements or cold acclimated when the inflorescence was 0.5 to 1 cm long (bolting stage). Using fully developed leaves from mature rosettes provided a defined system for which freezing tolerance, transcripts, and metabolites could be directly compared. As there are leaf order-dependent effects on freezing tolerance (Takagi et al., 2003) using fully developed leaves should reduce both within plant variation and between accession differences that may be associated with developmental stages. Except for Can and Te, the mean age at bolting was 42 to 45 d after sowing (DAS; Table I). Te had a mean age of 58 DAS and Can had a mean age of 77 DAS. Only Can showed signs of senescence, i.e. anthocyanin accumulation and yellowing of the oldest leaves. Because of the large differences Can showed in comparison to the other accessions, it was excluded from analyses that required direct comparisons of expression or metabolite levels; its presence as an outlier would influence results. However, its consideration in the overall analyses does provide important insights. For clarity, the remaining eight accessions are referred to as the core accessions.
Using nine accessions, we were able to investigate how the massive transcriptional changes that occur during cold acclimation (Fowler and Thomashow, 2002
In contrast to transcript data, there are fewer reports on cold-induced metabolic changes in Arabidopsis. The lower acclimation capacity of Cvi in comparison to Wassilewskija-2 was associated with an attenuated accumulation of many cold-responsive metabolites (Cook et al., 2004 ). Hundreds of metabolite changes also occur in Col-0, but it has not been compared to other accessions (Kaplan et al., 2004). We compared metabolite changes during cold acclimation for 210 metabolites detected in all nine accessions. All accessions showed large increases (1.7- to 5.6-fold) in their total metabolite content during cold acclimation, with the exception of Can, whose content decreased. Te, similar to Can, behaved differently from all other accessions, with metabolite increases reduced in magnitude and a much higher proportion of metabolites decreasing during cold acclimation (Fig. 4
; Supplemental Table II). Our data confirm the reduced magnitude of changes in Cvi (Cook et al., 2004). However, the overall magnitude of changes in Ler and C24 in comparison to Te, Nd, and Col-0 indicated there was no simple correlation between global metabolic changes and acclimation capacity (Fig. 4). To prevent bias due to missing values, this analysis did not include all metabolites. Detected metabolites varied from 302 to 373 for the nine accessions. The proportion of these showing a significant (false discovery rate [FDR]; P < 0.05) change during cold acclimation was compared (Supplemental Table II). The proportion of metabolites that accumulated ranged from 0.76 to 0.87 for the accessions C24, Co, Col-0, Ler, Nd, and Rsch, while Cvi had 0.66, Te 0.45, and Can only 0.25. The proportion of metabolites decreasing was 0.11 for Cvi, 0.15 for Te, and 0.37 for Can, but less than 0.07 for all other accessions. There are probably more metabolites decreasing, as many were only reliably detected in nonacclimated samples (Supplemental Table II). Because acclimated samples were 1/5 split injected to prevent peak saturation, missing peaks could also represent metabolites that fell below their detection limit. These are included in Supplemental Table II but are not considered further.
To investigate functionally relevant changes (Hannah et al., 2005 ), we tested for overrepresentation of Mapman functional groups (Thimm et al., 2004; Usadel et al., 2005) among the 1,000 most significantly up- and down-regulated transcripts for each accession during cold acclimation. In most accessions, there was significant overrepresentation of secondary metabolism, transport, and much of primary metabolism among up-regulated transcripts and of the photosynthesis, lipid metabolism, nucleotide metabolism, hormone, and redox groups among down-regulated transcripts (Table II
; Supplemental Fig. 1). In Cvi, a smaller number of transcripts from the primary metabolism functional groups were up-regulated, but more were down-regulated as compared to the other accessions.
The CBF transcription factors CBF1 to 3 have a central role in cold acclimation (Gilmour et al., 2004 ), and the ZAT12 regulon contains cold-responsive genes and partially overlaps with that of CBF2 (Vogel et al., 2005). Our data revealed that the CBF and ZAT12 regulons were significantly overrepresented among genes changing during cold acclimation in all accessions (Table III
). Consistent with the report that the most highly up-regulated genes belong to the CBF and ZAT12 regulons (Vogel et al., 2005), overrepresentation was highest among the 1,000 most significant changes (data not shown). Col-0 was the only accession with significantly increased ZAT12 expression (Supplemental Table I) and also had the largest overrepresentation of genes up-regulated by ZAT12 overexpression (Table III). In contrast, CBF genes were significantly induced in all accessions except Nd and Te, with varying contributions of individual members of the CBF family. CBF1 to 3 were induced in Can and Cvi, CBF1 and CBF2 in Col-0, Ler, and Rsch, and CBF2 in C24 and Co (Supplemental Table I).
Acclimated Freezing Tolerance
There were 731 transcripts that correlated significantly (FDR P < 0.05) with acclimated LT50 for the eight core accessions. Of these, 275 were positively correlated and 456 were negatively correlated with the freezing tolerance level (Supplemental Table I). Most of these transcripts also changed in abundance during cold acclimation. Photosynthesis, tetrapyrrole synthesis, cell wall, lipid, and nucleotide metabolism were significantly overrepresented among transcripts that negatively correlated with acclimated freezing tolerance (Table II; Supplemental Fig. 1). Carbohydrate, amino acid, and secondary metabolism as well as stress functional groups were overrepresented among positively correlating transcripts. CBF1 and CBF2 had significant positive correlations (FDR P = 0.045 and 0.053, respectively) with acclimated freezing tolerance (Fig. 5
) but CBF3 and ZAT12 did not. Among transcripts positively correlating with acclimated freezing tolerance, those up-regulated by ZAT12 or CBF2 (Vogel et al., 2005
At the metabolic level, there were large differences between nonacclimated and acclimated samples, but the differences between acclimated samples were smaller (Fig. 6 ). In general, there was not a strong cluster separation among acclimated samples, indicating no obvious global metabolic determination of differences in acclimated freezing tolerance. There was moderate support for the clustering of C24, Nd, Co, and Cvi versus Col-0, Rsch, Ler, and Te. Although this would support association between freezing tolerance and the global metabolite profile, the position of Ler and Nd are not in agreement with a clear separation of low and high acclimated tolerance (Fig. 6). To identify individual metabolites that may contribute to freezing tolerance, we correlated metabolite levels with acclimated LT50 for the core accessions. Fifteen metabolites, including Glc, Fru, Gal, raffinose, and Xyl, correlated significantly (FDR P < 0.1) with acclimated freezing tolerance (Supplemental Table II). Suc had the next most significant correlation (FDR P = 0.107). Glc, Fru, and raffinose are among the top four metabolites increased in response to CBF3 overexpression (Cook et al., 2004 ). CBF3 overexpression increased many other metabolites, and 35 increased more than 7.5-fold (Cook et al., 2004). The majority of these were part of a small cluster of metabolites in our dataset (Fig. 6).
Nonacclimated Freezing Tolerance
In contrast to acclimated freezing tolerance, nonacclimated freezing tolerance has received little attention. There have been some genetic analyses (Stone et al., 1993
Among transcripts significantly different in Te were CBF1 and 2, which were elevated about 14.5- and 4.7-fold, respectively (Supplemental Table I). CBF3 was increased 6.6-fold but this was only significant in four of seven comparisons. This was paralleled by the significant enrichment of the CBF regulon (Fisher exact test, P = 6 x 10–18) among transcripts elevated in Te, but there was no enrichment of CBF2 down-regulated transcripts among those reduced in Te (Table III). There was also a large overlap with metabolites increased by CBF3 overexpression, of which galactinol, raffinose, Fru, Glc, Pro, and Man were the most prominent (Cook et al., 2004
The second late flowering accession, Can, was very different from the other accessions before and after cold acclimation and was the only accession in which the total metabolite content decreased during acclimation. Despite these differences, nonacclimated Can showed similarities to Te and to acclimated rather than nonacclimated samples of other accessions at the metabolite level (Fig. 6). However, because its freezing tolerance was low, these similarities may represent transcripts and metabolites that correlate with life history traits or plant age but are not sufficient to increase freezing tolerance. We identified 1,767 transcripts that were higher in nonacclimated Can than in all accessions except Te (Supplemental Table I). In nonacclimated Can, total metabolite content was similar to acclimated samples, and our previously reported 114 consensus cold-regulated genes (Hannah et al., 2005 ) were weakly overrepresented (Fisher exact test, P = 0.057) among transcripts increased in nonacclimated Can. Although the expression of the CBF transcription factors was not different from other accessions, the CBF regulon was also significantly overrepresented among these transcripts (Fisher exact test, P = 0.005; Table III). With the exception of Pro, all metabolites discussed in relation to the CBF pathway were similarly or more elevated in nonacclimated Can in comparison to Te (Fig. 7). In addition, the functional groups of transcripts differentially expressed in Can compared to the other accessions, particularly those that were reduced, were similar to those identified in Te and in the analyses of acclimation capacity and acclimated freezing tolerance of the core accessions (Table II).
Our survey of nine geographically diverse accessions of Arabidopsis identified considerable natural variation for acclimated and nonacclimated freezing tolerance as well as for acclimation capacity. This is the widest survey of Arabidopsis freezing tolerance reported to date, and the observed variation was greater than reported in previous pairwise comparisons between Cvi and Wassilewskija-2 (Cook et al., 2004 ), Cvi and Col-0 (Zuther et al., 2004) or Ler (Alonso-Blanco et al., 2005), or between Col-0 and C24 (Klotke et al., 2004; Rohde et al., 2004). In Arabidopsis, natural variation has been reported for various traits (Koornneef et al., 2004), and latitudinal clines have been demonstrated for hypocotyl length (Maloof et al., 2001), circadian period length (Michael et al., 2003), and flowering time (Stinchcombe et al., 2004). However, the selective environmental factors, their ecological importance, i.e. the influence on the genetic structure of populations, and their relationship to underlying molecular mechanisms are not fully understood. Considering the general lack of evidence for environmental selection on genomic polymorphisms in Arabidopsis (Nordborg et al., 2005), our North-South transect revealing the significant correlation of freezing tolerance with habitat temperature (Fig. 2) assigns temperature a high probability of being an important selective force, with freezing tolerance levels being closely adjusted to habitat requirements. Such environmental selection is long known from studies of tree ecotypes being adapted to local climatic conditions. Genotype x environment interactions are well documented for low temperature as well as photoperiod, which are important signaling cues that trigger the onset of winter dormancy and bud break in the spring. More recently, a relationship between latitude of origin and the development of freezing tolerance has been demonstrated for both evergreen and deciduous trees (Repo et al., 2000; Li et al., 2002). Our data demonstrate a significant association of the complexity of the transcriptional response with acclimation capacity. Accessions with strong cold acclimation had about 1.5- to 2-fold more significant changes in gene expression than those that acclimated poorly (Fig. 3A). The amplitude of gene expression changes common to all accessions was also greater in accessions with higher acclimation capacity (Fig. 3B). As larger magnitudes of change may favor statistical detection, the number and magnitude of expression changes may be related. However, in contrast to the changes that were significant for all accessions, there was no relationship between magnitude and acclimation capacity when all significant changes in each accession were considered. Although the association of number and/or magnitude of expression changes could be a cause or consequence of increased cold acclimation, metabolic changes during acclimation (Supplemental Table II), particularly starch accumulation, argue against generally reduced metabolic activity as the cause of smaller transcriptome changes in poorly acclimating accessions. This suggests that a large number of changes may be functionally relevant for acclimation capacity and is good genomic evidence for the complexity of the trait observed in physiological investigations. For genetic engineering of freezing tolerance, such complexity likely needs to be addressed; overexpressing transcription factors to up-regulate entire regulons may be more promising than trying to identify single, nonregulatory genes with large positive effects on freezing tolerance.
Because secondary treatment effects like reduced light intensity or altered water availability are inevitably coupled to temperature shifts, pairwise comparisons of acclimated and nonacclimated plants cannot unambiguously identify genes directly involved in determining acclimated freezing tolerance. The correlation analysis shown here circumvents that problem by quantitatively relating transcriptome changes with acclimated freezing tolerance of identically treated accessions. This allows for improved functional assignment of transcript changes based on phenotypic variation among accessions. The phenotypic variation of acclimated freezing tolerance between accessions was high (5.9°C) and was 3-fold greater than the minimum acclimation capacity of an individual accession (1.7°C; Fig. 1). Obviously, genes that do not vary between accessions but are still necessary for freezing tolerance cannot be identified by this approach. Likewise, as all accessions were able to cold acclimate to some degree, genes identified may not be essential for the minimal acclimation response. However, the accordance of functional groups of genes changing during cold acclimation and those correlating with acclimated freezing tolerance indicates that most changes in response to low temperature are likely related to cold acclimation (Table II). All analyses indicate the down-regulation of photosynthesis and the up-regulation of photoprotective flavonoids are associated with improved acclimated freezing tolerance. These pathways were not only cold regulated but were also more overrepresented in accessions with higher acclimation capacity (Table II). However, as the largest photosynthetic changes occurred in Can, which acclimated poorly, down-regulation of photosynthesis appears insufficient for increasing freezing tolerance on its own. We identified the induction of genes encoding enzymes of flavonoid biosynthesis as quantitatively related to freezing tolerance in Arabidopsis. Genes encoding flavonoid biosynthetic enzymes were prominent among transcripts up-regulated by cold (4°C) exposure (this study) as well as chilling (13°C), despite an overall low correlation of changes between these treatments (Provart et al., 2003
Flavonoid metabolism is regulated by the closely related MYB transcription factors production of anthocyanin pigment 1 and 2 (PAP1 and 2; Borevitz et al., 2000
Increased freezing tolerance during cold acclimation is associated with many metabolic changes (Cook et al., 2004 ; Kaplan et al., 2004). Using multiple accessions, we reach the conclusion that cold acclimation is always associated with numerous and complex metabolic changes, but there is no clear relationship between global metabolite changes and differences in acclimation capacity or differences between the accessions in acclimated freezing tolerance (Figs. 4 and 6). Our data show that the reduced scale of metabolic changes in Cvi compared to other accessions (Cook et al., 2004; Fig. 4) is not characteristic of low acclimation capacity but offers a possible explanation why it occurs. It could be traced to the transcript level, where carbohydrate metabolism was less up-regulated and glycolysis more down-regulated than in the other accessions (Table II). Differences in the TCA cycle may also contribute to differences in the magnitude of metabolic changes, as Can, Cvi, and Te, which showed reduced metabolic changes during cold acclimation (Fig. 4), were the only accessions without significant up-regulation of this group (Table II). With respect to the involvement of metabolites in freezing tolerance, the probable importance of central carbohydrate metabolism is indicated by the identification of Glc, Fru, and Suc among positively correlating metabolites (Supplemental Table II). This is further supported by its overrepresentation also among transcripts positively correlating with acclimated freezing tolerance. The function in plant stress tolerance of soluble sugars, particularly of raffinose, which was identified by our analyses at the metabolite as well as the transcript level, has been highlighted by many investigators (for review, see Smallwood and Bowles, 2002). However, raffinose accumulation alone is neither necessary nor sufficient for cold acclimation in Arabidopsis (Zuther et al., 2004).
The two winter annuals, Can and Te, differed markedly from all other accessions, because in the nonacclimated state, they were metabolically similar to cold acclimated plants of the summer annuals due to their elevated metabolite levels (Fig. 6). During cold acclimation, metabolite increases were reduced in magnitude and more metabolites decreased in Can and Te (Fig. 4; Supplemental Table II). Can was the only accession that showed a net decrease in metabolite content during cold acclimation. As plants were harvested at the bolting stage, the late flowering of Te and Can meant they were about 13 and 32 d older, respectively, than the other accessions that likely contributed to the transcriptional and metabolic differences we observed. However, our analyses provided strong evidence that a substantial proportion of the metabolic and transcriptional differences between Te and the summer annuals were likely related to its increased freezing tolerance. Te was collected in Finland at the northernmost edge of the Arabidopsis range. It showed exceptionally high nonacclimated freezing tolerance, being greater than 2°C more tolerant than all other nonacclimated accessions and more tolerant than C24, Can, Cvi, or Co even when they were cold acclimated (Fig. 1). In nonacclimated plants, 167 metabolites were more abundant in Te in comparison to the summer annual accessions (Supplemental Table II). Of these, four of the top five of known identity (raffinose, Man, Gal, Glc, and Fru) correlated with acclimated freezing tolerance. Likewise, among the 298 transcripts significantly more abundant in Te than in the summer annuals, there was massive enrichment of both cold induced genes (Fisher exact test; P = 2.1 x 10–32) and those positively correlating with acclimated freezing tolerance (Fisher exact test; P = 5.7 x 10–57). Nonacclimated Can showed similarities, both at the metabolite and the transcript level, to nonacclimated Te and all acclimated accessions (Fig. 6; Table II) but had only weak freezing tolerance (Fig. 1). Many of the similarities of nonacclimated Can and Te may be related to the common late flowering habit, causing increased age of these plants at bolting. Importantly, these data indicate that high metabolite content is not sufficient for improved freezing tolerance. It is therefore tempting to speculate that only a few mutations could be responsible for the loss of tolerance in Can that never experiences temperatures below 10°C in its natural habitat. The absence of significant amounts of the stress-related metabolite Pro in Can (Fig. 7) may be a candidate for such an effect. Also, sensitive to freezing mutants have been isolated in Arabidopsis (Warren et al., 1996
Prime candidates for large effects on freezing tolerance are the major regulators, i.e. transcription factors that control large sets of temperature regulated genes. The best characterized of these are the CBF transcription factors. These have been shown to be strongly induced in response to short-term cold exposure but continue to be up-regulated after 7 d (Fowler and Thomashow, 2002
In agreement with a significant but limited role of the CBF pathway, only about 5% to 15% of the transcript changes we identified are established to be under the control of CBF. In addition, this overlap was only seen for transcripts up-regulated by CBF overexpression, but there was no significant overlap among down-regulated transcripts. Genes affected by CBF overexpression but not regulated during acclimation may indicate that some CBF regulated genes do not remain changed in the long term or may represent pleiotropic effects of ectopic overexpression. As our analyses support important roles for the down-regulation of many pathways, these changes are likely to be mediated by pathways independent of CBF. We identified 27 transcription factors, in addition to the CBFs and PAP1/2 that correlated positively with acclimated freezing tolerance, and 24 that showed negative correlations (Supplemental Table I). Negative regulators of freezing tolerance have received little attention. Here we identify two auxin-induced AUX/indole-3-acetic acid (Dharmasiri and Estelle, 2004
Plant Material
Nine Arabidopsis (Arabidopsis thaliana) accessions were greenhouse grown as described (Rohde et al., 2004
Whole rosettes pooled from 10 plants were harvested. Extraction, labeling, hybridization, and scanning were as described (Redman et al., 2004
In the first experiment, leaf discs were harvested from 30 individual plants. In experiments 2 and 3, six pooled samples from three plants were taken. Sugars and Pro were quantified independently as described (Rohde et al., 2004 Twenty spectra per second were recorded between mass-to-charge ratio 85 and 500. Peak identification and quantification were performed using the Pegasus software (Leco). Reference chromatograms at a minimum signal/noise threshold of 10:1 were used for automated chromatogram comparisons. Artifact peaks were removed from the result files. Areas were computed using reference unique ions as quantifiers. GC-mass spectrometry raw peak areas were normalized to the independent measure of Suc concentration and log2 transformed. The independent measures of Glc, Fru, raffinose, and Pro indicated good agreement (R2 = 0.83–0.93). Only data from experiment 1 is presented while, due to the reduced replication, experiments 2 and 3 were used for confirmation. Comparisons of known metabolites indicated that 0.83 and 0.85 were significantly (P < 0.05) correlated between experiments 1 and 2 and between experiments 1 and 3, respectively. As some data showed nonnormal distributions and/or unequal variances we used Wilcoxan tests to identify metabolite differences. Correlations were performed by linear regression using the median. Only metabolites detected in at least 10 samples were considered.
Statistical comparisons of the overlap between different gene lists, such as cold-responsive genes among functional groups or transcriptional regulons, were made using a Fisher exact test in the R software (Hannah et al., 2005
We thank Aenne Eckhardt for running the GC-mass spectrometry samples, Thomas Altmann for providing seeds, the German Resource Center for Genome Research for ATH1 hybridizations, the R and Bioconductor community for software, Ellen Zuther for help with sugar measurements, Anat Reiner for advice on the FDR corrections, and the Mapman team at the Max-Planck-Institut für Molekulare Pflanzenphysiologie for providing annotations. Received March 29, 2006; accepted July 6, 2006.
1 Present address: Universität Mainz Institut für Klinische Chemie und Laboratoriumsmedizin, Langenbeckstr. 1, D–55131 Mainz, Germany. 
2 Present address: University of California Davis Genome Center, 1315 GBSF Building, Health Sciences Drive, Davis, CA 95616. The author responsible for distribution of materials integral to the findings presented in this article in accordance with the policy described in the Instructions for Authors (www.plantphysiol.org) is: Matthew A. Hannah (hannah{at}mpimp-golm.mpg.de).
[W] The online version of this article contains Web-only data.
[OA] Open Access articles can be viewed online without a subscription. www.plantphysiol.org/cgi/doi/10.1104/pp.106.081141 * Corresponding author; e-mail hannah{at}mpimp-golm.mpg.de; fax 49–331–567–8250.
Alonso-Blanco C, Gomez-Mena C, Llorente F, Koornneef M, Salinas J, Martinez-Zapater JM (2005) Genetic and molecular analyses of natural variation indicate CBF2 as a candidate gene for underlying a freezing tolerance quantitative trait locus in Arabidopsis. Plant Physiol 139: 1304–1312 Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing. J Roy Stat Soc B Met 57: 289–300 Borevitz JO, Xia Y, Blount J, Dixon RA, Lamb C (2000) Activation tagging identifies a conserved MYB regulator of phenylpropanoid biosynthesis. Plant Cell 12: 2383–2394 Chinnusamy V, Ohta M, Kanrar S, Lee BH, Hong XH, Agarwal M, Zhu JK (2003) ICE1: a regulator of cold-induced transcriptome and freezing tolerance in Arabidopsis. Genes Dev 17: 1043–1054 Cook D, Fowler S, Fiehn O, Thomashow MF (2004) A prominent role for the CBF cold response pathway in configuring the low-temperature metabolome of Arabidopsis. Proc Natl Acad Sci USA 101: 15243–15248 Dharmasiri N, Estelle M (2004) Auxin signaling and regulated protein degradation. Trends Plant Sci 9: 302–308[CrossRef][ISI][Medline] Fowler S, Thomashow MF (2002) Arabidopsis transcriptome profiling indicates that multiple regulatory pathways are activated during cold acclimation in addition to the CBF cold response pathway. Plant Cell 14: 1675–1690 Gautier L, Cope L, Bolstad BM, Irizarry RA (2004) Affy-analysis of Affymetrix GeneChip data at the probe level. Bioinformatics 20: 307–315 Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J, et al (2004) Bioconductor: open software development for computational biology and bioinformatics. Genome Biol 5: R80[CrossRef][Medline] Gilmour SJ, Fowler SG, Thomashow MF (2004) Arabidopsis transcriptional activators CBF1, CBF2, and CBF3 have matching functional activities. Plant Mol Biol 54: 767–781[CrossRef][ISI][Medline] Gilmour SJ, Zarka DG, Stockinger EJ, Salazar MP, Houghton JM, Thomashow MF (1998) Low temperature regulation of the Arabidopsis CBF family of AP2 transcriptional activators as an early step in cold-induced COR gene expression. Plant J 16: 433–442[CrossRef][ISI][Medline] Gong Z, Dong CH, Lee H, Zhu J, Xiong L, Gong D, Stevenson B, Zhu JK (2005) A DEAD box RNA helicase is essential for mRNA export and important for development and stress responses in Arabidopsis. Plant Cell 17: 256–267 Guo Y, Xiong L, Ishitani M, Zhu JK (2002) An Arabidopsis mutation in translation elongation factor 2 causes superinduction of CBF/DREB1 transcription factor genes but blocks the induction of their downstream targets under low temperatures. Proc Natl Acad Sci USA 99: 7786–7791 Hannah MA, Heyer AG, Hincha DK (2005) A global survey of gene regulation during cold acclimation in Arabidopsis thaliana. PLoS Genet 1: e26[Medline] Hoffmann MH (2002) Biogeography of Arabidopsis thaliana (L.) Heynh. (Brassicaceae). J Biogeogr 29: 125–134[CrossRef] Hwang I, Sheen J (2001) Two-component circuitry in Arabidopsis cytokinin signal transduction. Nature 413: 383–389[CrossRef][Medline] Kaplan F, Kopka J, Haskell DW, Zhao W, Schiller KC, Gatzke N, Sung DY, Guy CL | ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
RESULTS
