The CBF1-dependent low temperature signalling pathway, regulon and increase in freeze tolerance are conserved in Populus spp.


Vaughan Hurry. Fax: +46 90 7866676; e-mail:


The meristematic tissues of temperate woody perennials must acclimate to freezing temperatures to survive the winter and resume growth the following year. To determine whether the C-repeat binding factor (CBF) family of transcription factors contributing to this process in annual herbaceous species also functions in woody perennials, we investigated the changes in phenotype and transcript profile of transgenic Populus constitutively expressing CBF1 from Arabidopsis (AtCBF1). Ectopic expression of AtCBF1 was sufficient to significantly increase the freezing tolerance of non-acclimated leaves and stems relative to wild-type plants. cDNA microarray experiments identified genes up-regulated by ectopic AtCBF1 expression in Populus, demonstrated a strong conservation of the CBF regulon between Populus and Arabidopsis and identified differences between leaf and stem regulons. We studied the induction kinetics and tissue specificity of four CBF paralogues identified from the Populus balsamifera subsp. trichocarpa genome sequence (PtCBFs). All four PtCBFs are cold-inducible in leaves, but only PtCBF1 and PtCBF3 show significant induction in stems. Our results suggest that the central role played by the CBF family of transcriptional activators in cold acclimation of Arabidopsis has been maintained in Populus. However, the differential expression of the PtCBFs and differing clusters of CBF-responsive genes in annual (leaf) and perennial (stem) tissues suggest that the perennial-driven evolution of winter dormancy may have given rise to specific roles for these ‘master-switches’ in the different annual and perennial tissues of woody species.


Adaptation to low temperatures is one of the most important components in the evolutionary process in temperate and boreal tree species (Saxe et al. 2001). Even when non-lethal, the accumulated lifetime costs of recurring freeze damage to tree fitness are significant; spring frosts damage flushing buds and flowers, decreasing tree growth and seed production (Selas et al. 2002), and leaves prematurely lost to autumn frosts reduce the ability of the tree to cold harden (Howell & Stockhause 1973). Under field conditions, woody perennials develop deep winter hardiness in response to two environmental cues (Weiser 1970). Shortening day length first initiates the transition from active growth to winter dormancy, which may take many weeks to complete (Weiser 1970; Fuchigami, Weiser & Evert 1971; Junttila 1976). The onset of winter dormancy results in a moderate increase in freezing tolerance (FT), but subsequent exposure to low temperatures is required to promote the deep winter hardiness that is unique to woody perennials (Weiser 1970; Fuchigami et al. 1971). Woody perennials, like herbaceous species, can also acquire cold tolerance when exposed to low temperatures during long days, but full winter hardiness only develops under the combined stimuli of short photoperiods and low temperatures (Christersson 1978; Li et al. 2002, 2003a; Puhakainen et al. 2004).

The molecular mechanisms governing the acquisition of FT are largely unknown in woody plants. The accumulation of dehydrins has been associated with the development of FT in a number of woody species (Arora & Wisniewski 1994; Arora, Wisniewski & Rowland 1996; Campalans, Pages & Messeguer 2000; Richard et al. 2000) and have been shown to correlate with seasonal variations in FT (Arora et al. 1996; Wisniewski et al. 1996; Welling, Kaikuranta & Rinne 1997; Rinne, Welling & Kaikuranta 1998; Lim, Krebs & Arora 1999). However, although significant efforts have been made to assign roles to different dehydrins (Wisniewski et al. 1999; Bravo et al. 2003), their biochemical functions remain largely unknown (Svensson et al. 2002). Cold acclimation is better understood in herbaceous annuals such as Arabidopsis thaliana, where members of the C-repeat binding factor (CBF or DREB1) family of transcriptional activators, which bind the cis-element known as the C-repeat (CRT)/dehydration-responsive element (DRE) (Stockinger, Gilmour & Thomashow 1997), have been shown to control the transcription of a suite of genes that play important roles in cold acclimation and the development of FT. The Arabidopsis CBF/DREB1 family consists of six paralogues, but only three [CBF1 (DREB1b), CBF2 (DREB1c) and CBF3 (DREB1a)] are cold-inducible (Sakuma et al. 2002). These three low-temperature-responsive CBFs colocalize to an 8.7 kb region of chromosome 4 and share a conserved AP2 DNA binding region flanked by the characteristic amino acid motifs PKKPAGRxKFxETRHP and DSAWR (Jaglo et al. 2001). Based on a number of experiments using transgenic plants, at least 12% of the cold-induced transcriptional changes observed in Arabidopsis are accounted for by the members of the CBF transcription factor family (CBF1-3) (Fowler & Thomashow 2002; Vogel et al. 2005; van Buskirk & Thomashow 2006), and ectopic expression of Arabidopsis CBF1-4 has been shown to improve FT of non-acclimated plants (Jaglo-Ottosen et al. 1998; Gilmour et al. 2000, Haake et al. 2002; Gilmour, Fowler & Thomashow 2004).

The accumulation of CBF transcripts and the activity of the CRT regulatory motif in Arabidopsis is also modulated by the presence and quality of light during cold stress (Kim et al. 2002; Fowler, Cook & Thomashow 2005). A short period of exposure to red light has been shown to be sufficient to induce CRT-driven GUS reporter expression in the cold, and this induction is photo-reversible with far-red light. GUS reporter expression is also altered in phytochrome mutant backgrounds, suggesting that day length mediates expression of the CBF regulon through phytochrome B in Arabidopsis (Kim et al. 2002). Furthermore, the kinetics of endogenous CBF transcript accumulation is reduced in complete darkness (Kim et al. 2002), suggesting that light regulation occurs at a point upstream of CBF expression. These data raise the question of whether CBF or related transcriptional regulators may be involved in the day length-mediated growth cessation in woody species. It has been suggested that growth cessation in woody perennials, mediated by short days and phytochrome, operates independently of the low-temperature-driven acquisition of deep FT (Welling et al. 2002). However, the observation that pre-treatment with short days enhances the expression of low temperature response genes (Puhakainen et al. 2004) and the finding that a CBF-like transcription factor is induced by short days in the cambial meristem of poplar (Schrader et al. 2004) indicates that there may be cross-talk between these two acclimation mechanisms.

Following their original discovery in Arabidopsis, the list of plants with cold-inducible CBF orthologues has expanded to include cold-sensitive and cold-tolerant annual species (Jaglo et al. 2001) and woody perennials (Kitashiba et al. 2002; Owens et al. 2002). Puhakainen et al. (2004) have also demonstrated that AtCBF3 can activate gene transcription via the birch Bplti36 promoter sequence in transgenic Arabidopsis, indicating that the structure and function of the active domain in the promoters of CBF-mediated cold-responsive genes have been conserved in herbaceous and woody perennials. However, detailed studies of the role of CBF-related proteins in cold acclimation in tree species have been lacking. Based on the demonstrated effects of both low temperature and light on CBF transcript accumulation in Arabidopsis, we have examined how CBF expression affected the FT of non-dormant poplar leaf and meristematic stem tissue under long-day growth conditions. This experimental design enabled us to focus on the role the CBF-mediated signalling pathway plays in the temperature response of different tissues in a woody perennial, without complicating interactions with day length and dormancy. Our data demonstrate: (1) that CBFs are involved in the FT mechanism of temperate trees species such as poplar; (2) that distinct regulons control stem (meristematic) versus leaf (non-meristematic) FT; and (3) that the Populus genome contains multiple CBFs that are likely to be differentially employed in perennial (stem) and annual (leaf) tissues.


Construction of transformation vector

A CaMV 35SPromoter::Arabidopsis CBF1 cDNA::35STerminator cassette (Jaglo-Ottosen et al. 1998) was ligated as a HindIII fragment into the binary vector pGAH (Onouchi et al. 1991) that had been linearised with HindIII.

Poplar transformation

Plants of poplar clone 717-1B4 (Populus tremula × alba) were grown in vitro in a culture room at 24 °C under cool-white fluorescent lights (40–60 µmol m−2 s−1, 16 h photoperiod). Stem internodes (7–10 mm in length) and leaf discs (5 mm in diameter) from fully expanded young leaves were used as the explants for the transformation (DeBlock 1990).

Agrobacterium tumefaciens strain EHA105 carrying the plasmid pGAH35SAtCBF1 was grown overnight at 28 °C in liquid YEP medium supplemented with 50 mg L−1 hygromycin and 50 mg L−1 kanamycin. The cells were collected by centrifugation at 1500 g for 20 min and resuspended to an OD600 = 0.4–0.6 in Murashige and Skoog (MS) medium with 20 µg L−1 acetosyringone (AS). Explants (stem internodes or leaf discs) were soaked for 10–20 min in the bacterial suspension, incubated on a shaker (300 g) for 1 h at room temperature, followed by cultivation on callus induction medium (CIM, pH 5.2) supplemented with 20 µg L−1 AS at 25 °C in the dark for 2–3 d. After 2–3 d of cocultivation, explants were washed four times with double-distilled water and once with wash solution (WS, pH 5.8) supplemented with 500 mg L−1 cefotaxime. The washed explants were blotted dry with paper towels, allowed to stand in a laminar flow hood for drying for about 30 s and then transferred to CIM supplemented with 250 mg L−1 cefotaxime and 50 mg L−1 kanamycin for 2 weeks at 25 °C in darkness. Explants were transferred to shoot induction medium (SIM, pH 5.8) supplemented with 50 mg L−1 kanamycin and 250 mg L−1 cefotaxime to induce shoot formation for 2–3 weeks. Explants were then transferred to shoot development and elongation medium (SEM, pH 5.8) supplemented with 50 mg L−1 kanamycin and 250 mg L−1 cefotaxime until shoots were well developed (> 1 cm in length). During this period, the explants were transferred to fresh SEM supplemented with 50 mg L−1 kanamycin and 250 mg L−1 cefotaxime every two weeks. Regenerated shoots (> 1.0 cm) were excised from the explants and further screened for kanamycin resistance on rooting medium (RM, pH 5.8) supplemented with 50 mg L−1 kanamycin and 250 mg L−1 cefotaxime. Leaves from the rooted plants were collected for DNA extraction for PCR to verify the presence of CaMV 35S promoter::AtCBF1 transgene in the poplar genome, and for RNA extraction for Northern blot analysis to verify lines expressing detectable levels of the transgene.

Leaf and stem FT assays

Plants of wild-type (WT) and transgenic lines 1 and 2 were propagated by cuttings in vitro, and rooted plants were transplanted into 1 gallon pots containing Sunshine SB40 Professional Growing Mix (Sun Gro Horticulture Inc., Bellevue, WA) supplemented with 7% pumice (weight/volume) and 0.14% perlite, and fertilized with 0.04% osmocote (The Scotts Company, Marysville, OH). Plants were grown in a greenhouse maintained at a 16 h day (25 ± 3 °C) and 8 h night (20 ± 3 °C) with supplemental lighting (mercury halide lamps providing 400–500 µmol/m2 s−1). After 6 weeks of growth, plants of both transgenic lines and WT were divided into two groups. One group of plants was transferred to a cold room (2 °C, 16 h photoperiod) for 1 week, while the other group was maintained in the greenhouse under the greenhouse conditions mentioned earlier. For freezing tests tissue samples, 1-cm-diameter leaf discs were excised from each side of the two to three uppermost fully expanded leaves, and 2 cm stem segments were harvested. Three leaf discs or stem segments were used per assay, and FT was determined by ion leakage (Sukumaran & Weiser 1972).

RNA extraction for microarray analysis

Leaf and stem samples were collected from three WT plants and three individuals from each transgenic line (1 and 2) grown under either the warm greenhouse conditions discussed earlier or after 1 week of cold acclimation (2 °C with 16 h photoperiod). Total RNA was extracted from frozen stem and leaf samples (Hughes & Galau 1988), the RNA samples were reconstituted in TE buffer and the total RNA was further purified using QIAGEN RNeasy columns (Qiagen, Valencia, CA, USA). Aliquots of these total RNA preparations from each clonally propagated genetic background were then pooled before being used for fluorescent probe synthesis. This experimental protocol was repeated on clonally propagated material grown at a later date, resulting in two independent biologically replicated experiments.

Preparation of fluorescent targets

Targets were labelled indirectly by incorporating aminoallyl-dUTPs (Sigma, St Louis, MO, USA; Pharmacia, Fairfield, CT, USA) and subsequently coupling with either Cy3 or Cy5 dye. The reverse transcription reaction mixture (30 µL) contained 20 µg total RNA with 5 µg of oligo(dT) 20 mer, 10 mm DTT, 500 µm each of dATP, dCTP, and dGTP, 300 µm dTTP, 200 µm aa-dUTP, 30–40 units RNase inhibitor, and 200 units of SuperScript II reverse transcriptase (Life Technologies, Grand Island, NY) in 1X Superscript first-strand buffer (50 mm Tris-HCl, pH 8.3, 75 mm KCl, 3 mm MgCl2, and 20 mm DTT) (Life Technologies). After incubation at 42 °C for 2 h, the reaction products were treated with 10 µL 1 m NaOH, 10 µL 0.5 m EDTA, incubated at 65 °C for 15 min, and neutralized with 50 µL 1 m HEPES (pH 7.0). Samples were purified on QIAGEN MinElute Clean-up columns (Qiagen) as directed, except that the final elution was performed with 15 µL of NaHCO3, pH 9.0 into an amber eppendorf containing the appropriate vacuum-dried Cy dye pellet [1/7 of the commercial aliquot (Amersham Pharmacia Biotech; 1 mg protein-labelling aliquots)]. The eppendorf contents were mixed until the Cy dye pellet was fully resuspended, spun down and incubated in darkness at room temperature for 2 h. At the end of the incubation period, the labelling mixture was diluted to 100 µL using sterile water and QIAGEN PCR purification columns (Qiagen) were used as directed, except that the final elution was performed using 30 µL sterile water. Respective Cy3 and Cy5 samples for comparison were combined and reduced under vacuum to a volume of 13 µL at 40 °C. This probe mixture was combined with 0.5 µL of 25 µµL−1 yeast tRNA, 2.0 µL of 10 µµL−1 oligo(dA), 6.0 µL of 20X SSC, 7.5 µL of deonized formamide and 1.0 µL 10% sodium dodecyl sulphate (SDS) and denatured at 95 °C for 1 min. After 2 min cooling on ice, the prepared probe mixture was used for hybridization.

Microarray hybridization and scanning

Populus 13K duplicate array slides (Andersson et al. 2004) were prehybridized and hybridized using the solutions and protocols reported previously (Hertzberg et al. 2001). The slides were scanned with a ScanArray Lite scanner (Perkin Elmer, Wellesley, MA, USA) and ScanArray express software. Separate images were acquired for each fluor at a resolution of 10 µm per pixel. Scanning photomultiplier and laser power settings were optimized by the software to minimize saturated spots and maximize the signal-to-noise ratio. Microarray spots were flagged and quantified using Genepix 5.0 (Axon Instruments, Foster City, CA, USA) software. The resulting mean signal and median background data for each channel and spot were imported and normalized using the Bioconductor package (, applying Edwards background correction and loess normalizations within individual arrays and Aquartile between-array normalization before import into Genespring 6.0 (Silicon Genetics, Redwood City, CA, USA) for visualization. Gene lists for each comparison (Supplemental Tables) were constructed from the data collected from four hybridizations of each line (including two independent biological replicates and dye-swapping). Duplicate printing of probes on the hybridized UPSC–KTH 13K Populus slides also increased the total replication twofold for each probe. Therefore, the final gene lists (incorporating expression data from both transgenic lines) represent the mean expression change and statistical data from 16 replicates. We set statistical significance at P < 0.05 plus an arbitrary fold-change (FC) cut-off of FC > 1.75X to construct the gene lists. The raw and normalized microarray data are publicly available at

Multiplex and real-time RT-PCR transcript quantification

Total RNA was isolated from 50 to 100 mg of powder from leaf or stem tissue from 6-week-old hybrid aspen (Populus tremula × tremuloides clone T89) trees using Trizol reagent (Gibco BRL, Paisley, UK) according to the manufacturer’s instructions except for an extra 10 min centrifugation step at 4 °C after the addition of Trizol, the use of 0.5 volumes of bromochloropropane in place of 1 volume of chloroform during phase separation, and use of 1:1 isopropanol:high salt solution (0.8 m sodium citrate, 1.2 m sodium chloride) in place of isopropanol during RNA precipitation. QIAGEN RNeasy columns (Qiagen) were used as directed to purify samples. Three micrograms total RNA was used in reverse transcription with the First Strand cDNA kit (Amersham) according to the manufacturers instructions. Semi-quantitative multiplex PCR was used to determine relative amounts of AtCBF1 transcript in the transgenic Populus lines using the Cbf-f (ACGAATCCCGGAGT CAACATGC) and Cbf-r (ccttcgctctgttccggtgtataaat) gene-specific primers and Universal 18S (Quantum RNA 18S standards, Ambion, Austin, TX, USA) primers with competimer (2:8 ratio) as the endogenous standard. Real-time PCR quantification was performed using iQ SYBR green supermix (Bio-Rad, Hercules, CA, USA) as instructed with gene-specific primers [PtCBF1F, ACACAGGATGCCT TGTTTCC; PtCBF1R, GGAGTTCAACCAGGTG CAAT; PtCBF2F, GGGAGGTGAGTTGATGAGGA; PtCBF2R, TATTAGCCAACAACCCTGGC; PtCBF3F, TTTCAAATGAGGCCAAGGAC; PtCBF3R, CCTC CCTCCTGAAATCTTCC; PtCBF4F, GGCAGCAAAT GAGGCAGCAG; PtCBF4R, CTTGAGCAATCCTCTA GCACTGCAT; Universal 18S rRNA primers (Ambion), and ∼2 ng cDNA template. Relative PtCBF abundance was quantified and then normalized using the ΔΔct method of reference gene: 18S; calibrator sample: 2 ng, 3 h cold-treated leaf cDNA].


Ectopic expression of AtCBF1 in Populus

Arabidopsis CBF paralogues have been used to manipulate FT in a number of herbaceous plant species (Jaglo et al. 2001; Hsieh et al. 2002; Lee et al. 2003, 2004). To test whether a CBF regulon exists and plays a role in the FT of overwintering woody perennials, we generated transgenic Populus tremula × alba (clone 717-1B4) ectopically expressing the well-characterized AtCBF1 gene from Arabidopsis (Fig. 1a). We generated 19 independent transgenic lines expressing AtCBF1 and characterized two of these lines expressing AtCBF1 at rates equivalent to endogenous AtCBF1 transcript levels in Arabidopsis after 3 h cold treatment, as assessed by RT-PCR (Fig. 1b & c). The selected lines, with high AtCBF1 transcript abundance (AtCBF1-Poplar), had fewer roots and slower growth rates when cultivated in vitro(Fig. 2a), but fully recovered after 3–6 weeks growth in soil (Fig. 2b–d).

Figure 1.

Expression of the AtCBF1 transgene. (a) pGAH35SAtCBF1 expression construct. (b) Relative expression of the AtCBF1 gene in the stem (S) and leaf (L) tissue of wild-type Populus (PtWT), AtCBF1-Poplar line 1, AtCBF1-Poplar line 2 grown for 6 weeks in the greenhouse at 23 °C, and in warm (23 °C) grown wild-type Arabidopsis, Col 0 (AtWT) after 3 h exposure to 5 °C, respectively. (c) Semi-quantitative multiplex RT-PCR analysis was performed on first-strand cDNA generated from total RNA extracted from wild-type and transgenic plants 6 weeks after transfer to soil in the greenhouse. Differences in template starting quantity in the wild-type and transgenic lines were corrected by normalizing to the expression of the 18S rRNA (18S) in the respective samples. Error bars indicate standard errors of six independent experiments.

Figure 2.

Morphology of the AtCBF1-Poplar trees. (a) AtCBF1-Poplar were propagated on agar under long-day conditions. (b) AtCBF1-Poplar 3 weeks after transfer from propagation on agar to soil. (c) AtCBF1-Poplar 10 weeks after transfer from propagation on agar to soil. (d) Average plant height during 6 weeks growth on soil in wild-type and AtCBF1-Poplar lines. Error bars represent standard error in three independent experiments. Open circles, wild type; closed squares, AtCBF1-Poplar line 1; closed triangles, AtCBF1-Poplar line 2.

AtCBF1-Poplar lines have improved leaf and stem FT

The effect of ectopic expression of AtCBF1 in poplar on FT of leaves and stems was tested using an electrolyte leakage test. This assay provides an estimate of cell damage (Dexter, Tottingham & Garber 1932; Sukumaran & Weiser 1972) and is quantified by determining the temperature leading to leakage of 50% of cellular electrolytes (TEL50). Non-acclimated leaves from both AtCBF1-Poplar lines showed a significant (< 0.001) gain in FT, from −3.9 °C for non-acclimated WT to an average of −6.9 °C for AtCBF1-Poplar (Table 1). Ectopic expression of AtCBF1 also significantly improved stem FT in non-acclimated plants (TEL50−5.4 °C for the AtCBF1 lines versus −4.1 °C for WT, P < 0.001) (Table 1). These FT data demonstrate that the cryoprotective benefit of ectopic AtCBF1 expression shown in Arabidopsis leaves (Jaglo-Ottosen et al. 1998; Gilmour et al. 2000) extends to leaves and stems of poplar, and suggests that Populus spp. utilize a CBF-mediated signalling pathway to control at least part of the cold acclimation response.

Table 1. TEL50 measurements for leaf and stem tissues of non-acclimated (NA) and 1 week cold-acclimated (CA) hybrid aspen trees. Values represent the mean of six independent experiments ± standard error. Statistical significance was tested first by one-way anova with Dunnett’s post-test using GraphPad Prism version 4.00 for Windows (GraphPad Software, San Diego, CA, USA)
  • WT, wild type.

  • ***

    P < 0.001

WT−3.9 ± 0.2−17.3 ± 1.2−4.1 ± 0.3 −9.9 ± 0.2
Line 1−6.7 ± 1.0***−18.3 ± 1.1−5.2 ± 0.4***−10.7 ± 1.0
Line 2−7.1 ± 0.7***−18.7 ± 2.5−5.7 ± 0.6***−11.3 ± 1.7

The AtCBF1 regulon(s) in Populus

To determine whether AtCBF1-driven transcriptional changes mirrored the endogenous transcriptional response to low temperature, the transcriptomes of WT warm-grown leaf and stem tissue were compared with the transcriptomes of leaf and stem tissues of warm-grown poplar ectopically expressing AtCBF1, and with the transcriptomes of leaf and stem tissues from WT trees after 7 d of cold acclimation at 2 °C. The cDNA microarray experiments utilized mRNA from WT plants, and the two independent AtCBF1-Poplar lines and gene lists were constructed using the average result of two independent growth experiments. These experiments enabled us to determine: (1) which genes were susceptible to AtCBF1 regulation; (2) whether the genes that respond to AtCBF1 control showed differential expression in the meristematic stem tissues versus leaf tissue; and (3) how much overlap there was between the CBF regulon and the endogenous (untransformed) response of poplar leaves and stems to cold.

Using our cut-offs (FC > 1.8, P < 0.05), we identified 63 genes up-regulated in leaves from warm (23 °C) grown AtCBF1-Poplar relative to leaves from warm-grown WT plants (Table S1a). Almost half of the total list (29/63) of AtCBF1-Poplar leaf regulon members was composed of novel ‘expressed proteins’ and other proteins with no known function. Of the 34 AtCBF1-Poplar leaf regulon genes annotated to have a known function, more than one-third (35%) encoded metabolic enzymes. Together, the classes ‘Cell rescue’ (15%) and ‘Transcription’ (21%) made up another third, with the remaining classes represented being: ‘Development’ (9%), ‘Cellular communication/signal transduction’ (6%), ‘Storage’ (6%), ‘Cellular biogenesis’ (6%) and ‘Protein fate’ (3%) (Fig. 3). This functional distribution of genes in the AtCBF1-Poplar leaf regulon mirrored that of the genes endogenously up-regulated in WT leaves by 7 d at low temperature (Fig. 3, Table S2). In contrast, the functional class ‘Transcription’ made up the dominant fraction (41%) of the AtCBF1-Poplar regulon in warm-grown perennial stem tissue, followed by ‘Cellular communication/signal transduction’ (18%), ‘Cell rescue’ (12%) and ‘Metabolism’ (12%) (Fig. 3).

Figure 3.

Functional distribution of the genes annotated with known functions significantly up-regulated in warm-grown AtCBF1-Poplar (51 genes in total) and 7 d cold-treated wild-type Populus (387 genes in total) relative to warm-grown wild-type Populus.

The functional analysis of the genes induced by 7 d of cold (2 °C) treatment in WT Populus leaves and stems showed that these tissue-specific regulons were not as innately different as the AtCBF1-Poplar leaf and stem regulon data (Fig. 3). This comparison held true even at the level of individual genes: more than half (57%) of the up-regulated genes with known function from WT cold-treated stem tissue were also significantly up-regulated in WT cold-treated leaf tissue. In fact, in terms of functional composition, the AtCBF1-Poplar leaf, WT cold leaf and WT cold stem regulons showed high similarity, while the AtCBF1-Poplar stem regulon differed largely on the basis of a substantial depletion of metabolic genes (12% versus 35% in AtCBF1-Poplar leaf tissue and 38% in WT cold stem) and an enrichment of genes regulating transcription (41% versus 21% in AtCBF1-Poplar leaf tissue and 22% for WT cold stem) (Fig. 3).

As indicated by the functional analysis, the normalized expression data for all genes significantly up-regulated by AtCBF1 in at least one tissue showed that the leaf (Table S1a) and stem (Table S1b) AtCBF1-Poplar regulons contained clusters specific for that tissue (Fig. 4 gene subsets B and C, respectively), suggesting different functional roles for the AtCBF1-Poplar regulons in the annual (leaf) and perennial (stem) tissues. However, the leaf and stem AtCBF1-Poplar regulons also shared clusters of up-regulated genes with each other and with the WT leaf and stem cold regulons (Fig. 4 gene subset A and Table 2). Twenty-two of the 63 AtCBF1-Poplar leaf regulon genes identified were also up-regulated in the WT 7 d cold-acclimated leaves (Table 2). Five of the 26 AtCBF1-Poplar stem regulon genes were also up-regulated in WT stems by 7 d cold treatment (Table 2). Although the total AtCBF1-Poplar gene number in stems was lower than that in leaves, approximately the same fraction of the total 7 d cold induced list could be linked to AtCBF1 expression in the stem (5/152 or 3%, see Table 2 and Table S3) as in the leaf (22/462 or 5%, Table 2 and Table S2). Twenty-nine genes were also found to be down-regulated by AtCBF1 expression in leaves, and 24 genes were down-regulated in stems (Table S1c).

Figure 4.

Hierarchical tree of genes belonging to the CBF regulon based on similarities in expression profiles of the leaf and stem tissues of AtCBF1-Poplar and 7 d cold-treated wild-type Populus. Gene subsets up-regulated in the leaf and stem tissue of AtCBF1-Poplar are indicated to the right with brackets, where (A) indicates the gene subset similarly up-regulated in both AtCBF1-Poplar and wild-type cold-treated Populus tissues, (B) indicates the AtCBF1-Poplar leaf-specific regulon, and (C) indicates the AtCBF1-Poplar stem-specific regulon.

Table 2.  A selection of genes up-regulated in AtCBF1-Poplar (Poplar tremula × alba) and wild-type (WT) Populus tremula × alba cold acclimated for 1 week at 2 °C (versus WT Populus tremula × alba grown at 23 °C). Normalized mean fold-change (FC) values are shown (bold face indicates P < 0.05). The poplar gene probe (PU) number is matched to its closest Arabidopsis gene (AGI) based on homology to the Arabidopsis genome sequence. This table corresponds to gene subset ‘A’ in Fig. 4.
CBF1 Leaf FCCBF1 Stem FCWT Leaf FCWT Stem FCClone IDAGIDescriptionCategory rescue proteinUnknown role protein. Oryza sativaMetabolism proteinUnknown role proteinUnknown role proteinUnknown role proteinCell rescue proteinUnknown role
1.91.910.05.8PU03418At3g05880LTI6ACell rescue
2.62.414.54.1PU03755At5g38760Pollen coat proteinDevelopment protein AGP21Development proteinUnknown role lipase/hydrolaseMetabolism pyrophosphataseMetabolism
2.62.514.01.3PU03390At2g28680Legumin-like proteinStorage proteinUnknown role rescue B-box zinc finger proteinTranscription−1.3PU10341At5g55430Expressed proteinUnknown role
1.8− proteinUnknown role abundant proteinUnknown role binding proteinSignal transduction transcription factor FBP22Transcription proteinUnknown role

Comparison of Populus and Arabidopsis CBF regulons

Comprehensive transcriptome changes due to ectopic AtCBF1 expression have previously only been published for Arabidopsis as a combined AtCBF1/CBF2/CBF3 regulon (Fowler & Thomashow 2002). However, paralogue-specific studies have identified gene members of the AtCBF2 (Vogel et al. 2005) and AtCBF3 regulons in Arabidopsis (Maruyama et al. 2004). We next examined the similarity between our identified annual (leaf) and perennial (stem) tissue AtCBF1 regulons in Populus and the reported Arabidopsis CBF regulons. We surveyed our Populus cDNA microarray DB (Sterky et al. 2004) to find all the closest protein homologues to previously reported CBF-responsive genes from the combined AtCBF1-3 regulon and the AtCBF2- and AtCBF3-paralogue regulons. The Arabidopsis CBF3 regulon shared the most similarity with the CBF regulon in Populus. Twelve of the 38 reported AtCBF3-up-regulated genes that had representative orthologues on our microarray, and a gene comparison showed that more than half (7/12 having FC > 1.3) of the AtCBF3-up-regulated gene orthologues present on the POP1 13K cDNA array were similarly up-regulated in one or both AtCBF1-Poplar tissues and possessed DREs in their 1500 bp promoters (Table 3). Examination of the induction of AtCBF2- and AtCBF1/2/3 regulon orthologues in Populus (Table S4) showed greater disagreement. Whether this reduced response of the CBF2 and CBF1/2/3 regulon orthologues in Populus to ectopic AtCBF1 expression was indicative of differences in species and/or paralogue-specific regulon composition remains a question that will require additional experiments to answer.

Table 3.  Normalized microarray expression values for Populus genes matched to reported Arabidopsis CBF3 regulon members in AtCBF1-Poplar and trees acclimated for 1 week at 2 °C. The poplar gene probe (PU) number is matched to its closest Arabidopsis gene (AGI) based on homology to the Arabidopsis genome sequence. Gene names presented in bold have a dehydration-responsive element (DRE) cis-element in within 1500 bp of the start codon.
   AtCBF1 treesWT cold trees
Leaf FCStem FCLeaf FCStem FC
  1. WT, wild type; FC, fold-change.

CBF3 regulon
expressed proteinPU07867At4g353001.
expressed proteinPU04708At5g625301.
expressed proteinPU10962At4g140001.
expressed proteinPU11420At1g277301.
expressed proteinPU07579At1g510901.
expressed proteinPU05694At4g159101.

AtCBF1 regulon promoter analysis in Populus reveals conserved linkage of the DRE and abscisic acid-responsive element (ABRE)

Given the broad similarity between the Populus and Arabidopsis CBF regulons, we determined whether there was also a conserved enrichment of the DRE in the promoters of the AtCBF1-Poplar regulon genes. Sixteen of the 63 (25%) AtCBF1-up-regulated leaf regulon gene promoters (defined as the 1500 bp of genomic sequence found upstream of the ATG site) contained the basic DRE nucleic acid sequence ‘RCCGAC’. Eight of 26 (31%) AtCBF1-up-regulated stem regulon promoters contained the DRE. Positional analysis of the RCCGAC revealed no significant enrichment along the 1500 bp promoter. However, it is worth noting that 7 of 33 DREs appeared between 200 and 400 bp upstream of ATG (Table S5), leaving the possibility that a larger sample set, using currently unavailable transcriptional start sites instead of the ATG, would verify the −450/−51 positional enrichment of the DRE reported for Arabidopsis (Maruyama et al. 2004) in Populus. Recent studies have also noted an enrichment of the ABRE (Marcotte, Russell & Quatrano 1989) ‘ACGTGTC’ in the CBF3 regulon of ArabidopsisMaruyama et al. 2004). In Populus, an ABRE was found in 12 of 87 AtCBF1 regulon 1500 bp promoters, representing a significant enrichment (χ2, P < 0.05). There was also a significant overlap between ABRE- and DRE-containing gene promoters in Populus (4/12 ABRE-containing gene promoters also contained the RCCGAC consensus: χ2, P < 0.001) and positional enrichment of the ABRE at positions 100–199 bp and 300–399 bp upstream of the start codon (χ2, P < 0.001).

The Populus balsamifera ssp. trichocarpa genome encodes six CBF-like transcription factors

The AtCBF1-Poplar leaf and stem regulons contained many genes previously linked to the process of cold acclimation in herbaceous annuals such as Arabidopsis and, when combined with the observed improvement in FT in AtCBF1-Poplar, implied that cold-responsive CBF transcription factors were encoded by the Populus genome and involved in the cold acclimation process. Iterative tBLASTn searches of the Populus balsamifera ssp. trichocarpa (Torr. & A. Gray) Brayshaw genome ( using the conserved AP2 and flanking CBF ID sequences from Arabidopsis (Jaglo et al. 2001) identified six potential CBF-encoding genes. An amino acid sequence alignment of the highly conserved AP2 domain and flanking CBF ID regions of the candidate poplar CBFs with other dicotyledonous CBFs (Fig. 5a) shows that two of the six candidate genes (PtCBF1 and PtCBF2) have 100% conservation of the previously reported Arabidopsis CBF consensus sequences (PKKR/PRAGRxKFxETRHP and DSAWR), two other candidate genes (PtCBF3 and PtCBF4) possess a single amino acid substitution in the N-terminal consensus (I instead of K in position 10) with the remaining two sequences (PtCBFL1 and PtCBFL2) deviating from the consensus sequences by four and five amino acids, respectively. Interestingly, none of the poplar CBFs utilizes the commonly occurring proline at the fourth position, but PtCBF1-4 use the common alternate arginine residue while PtCBFL1-2 have lysine at this position, in contrast to all the other dicot CBFs. A phylogenetic analysis of the six poplar CBF candidates with known full-length sequences from dicotyledonous plants, using the TINY AP2 transcription factor as the out-group, confirms that PtCBFL1 and PtCBFL2 share the least amino acid similarity to the characterized CBFs from Arabidopsis (Fig. 5b). Also of some note, a Populus gene previously reported to be ‘CBF1-like’ and up-regulated in dormant cambium (PtDRTY) (Schrader et al. 2004) groups with TINY, not the CBFs. However, unlike TINY, PtDRTY shares some extended blocks of homology in the C-terminal domain with the CBFs. Investigation of the 1500 bp promoters of the PtCBFs also demonstrated that they possess numerous potential ICE-binding sites (CANNTG) and a tBLASTn search of the Populus balsamifera ssp. trichocarpa genome identified Populus-encoded ICE proteins with high similarity to Arabidopsis ICE1 (data not shown), indicating that ICE-mediation of the CBF signalling pathway may also be conserved in this woody perennial.

Figure 5.

Comparison of representative C-repeat binding factor (CBF) protein family members in dicots. (a) Alignment of AP2 and flanking region of CBF factors. (b) Phylogenetic analysis, based on minimum evolution, was performed with the full-length protein sequences (Fig. S1). Polypeptides used for alignments were the six CBFs of Arabidopsis: AtCBF1 (U77378), AtCBF2 (AF074601), AtCBF3 (AF074602), AtCBF4 (NM_124578), AtDDF1/At1g12610 (NM_101131), AtDDF2/At1g63030 and (NM_104981); three Brassica napus CBFs: BnCBF-A (AAL38242), BnCBF-B (AAL38243), BnCBF-C (AAD45623); two tobacco EREBPs: Nt111A (AAG43548), Nt111B (AAG43549), Prunus avium PaDREB1 (AB121674); and the six poplar CBFs identified in this study: PtCBF1 (JGI666968), PtCBF2 (JGI346104), PtCBF3 (JGI548519), PtCBF4 (JGI63608), PtCBFL1 (JGI594822), and PtCBFL2 (JGI649103). PtDRTY (JGI253714) was also used and is a related AP2 domain protein with extended homology to CBFs in the C-terminal domain. Arabidopsis TINY (AtTINY) (AAC29139) was utilised as an outlier and represents the closest AP2 domain-containing protein group that falls outside the CBF/DREB1 family clade. A likely poplar TINY orthologue, PtTINY-Like (JGI653993), was also utilised as an outlier and assembled from the draft poplar genome sequence data. CBF proteins were initially aligned using CLUSTAL-W, the output alignments hand refined using GeneDoc Version 2.6 software (, and used to conduct phylogenetic analysis using MEGA Version 2.1 software ( Phylogenetic trees were generated using MEGA Version 2.1 Neighbor Joining and Minimum Evolution methodologies on 1000 bootstrap replications. Bootstrap numbers are shown on the dendrogram branch points.

PtCBFs are cold-inducible in leaf and stem tissues

Based on the phylogenetic analysis, we selected PtCBF1-4 for further study, and real-time PCR demonstrated that all were cold-inducible, although the kinetics of induction and tissue specificity differed between the orthologues (Fig. 6). PtCBF1 was cold-inducible in leaf and stem tissue. PtCBF1 transcript levels in both tissue types peaked 6 h after transfer to cold and decreased again overnight (Fig. 6a). Similarly, PtCBF3 was inducible in both leaf and stem tissue (Fig. 6c). PtCBF3 transcript accumulated rapidly after cold-shifting, peaking at 3 h, and returned to near starting levels within 6 h. In contrast, both PtCBF2 and PtCBF4 were cold-inducible in leaf tissue but were only weakly cold-induced in the stem (Fig. 6b & d). Leaf PtCBF2 expression peaked 9 h after transfer, with stem transcripts peaking between 3 and 6 h. PtCBF4 expression peaked at 3 h (Fig. 6d). These data strongly support the conclusion that poplar utilizes a cold-inducible CBF-based response to cold.

Figure 6.

Expression response of Populus CBF paralogues to low temperature in wild-type trees: (a) PtCBF1; (b) PtCBF2; (c) PtCBF3; (d) PtCBF4. Expression was determined by RT-PCR. Each reaction used 30 ng first-strand cDNA and normalized to 18S rRNA quantity. Open circle, leaf; closed circle, stem. Error bars represent the standard error on three independent experiments.


In this study, we have characterised the role of the CBF transcription factors in leaves and stems of a woody perennial, Populus. As shown previously for herbaceous plants (Jaglo et al. 2001; Hsieh et al. 2002; Lee et al. 2003, 2004), ectopic expression of AtCBF1 from Arabidopsis increased the constitutive FT of non-acclimated meristematic stem and leaf tissues in both independent AtCBF1-Poplar lines, demonstrating that the CBF-mediated signalling pathway and regulon are conserved in annual and perennial tissues of woody perennials. This conclusion is supported by earlier reports that the promoters of cold-responsive genes such as Bplti36 from birch are activated by Arabidopsis CBFs (Puhakainen et al. 2004) and that CBF orthologues from woody species are functional when transferred to Arabidopsis (Kitashiba et al. 2004). Our subsequent microarray analysis, showing broad agreement between the transcriptomes of AtCBF1-Poplar and that reported for the AtCBF1-3 regulon orthologues in Arabidopsis (Fowler & Thomashow 2002; Vogel et al. 2005), demonstrates that AtCBF1 acts in a similar manner in Populus as AtCBF1-3 act in Arabidopsis.

Interestingly, the functional distribution of the Populus AtCBF1 regulon genes differed between the annual (leaf) and perennial (stem) tissues of this woody species. In stem tissues, 59% of the AtCBF1-Poplar regulon was composed of genes involved in the processes of ‘Transcription’ and ‘Cellular Communication/Signal Transduction’, while these two processes composed only 26% of the AtCBF1-Poplar leaf regulon. It has been shown previously that the CBF regulon contains a number of transcription factors, which themselves target additional gene sets (Fowler & Thomashow 2002; Gilmour et al. 2004; Novillo et al. 2004; Vogel et al. 2005; Nakashima & Yamaguchi-Shinozaki 2006). However, the difference in composition of the poplar leaf and stem regulons was unexpected and could not be attributed to a general difference in the low temperature transcriptomes of the annual and perennial tissues because the same comparison made between 7 d cold-acclimated WT leaf and stem tissues demonstrated general agreement in the distribution of up-regulated functional categories (Fig. 3). Examination of the identities of the leaf and stem cold regulon genes showed that their functional similarities were not superficial: the WT cold-responsive regulons of the leaf and stem were compositionally similar. It is possible that the AtCBF1-Poplar leaf/stem regulon functional difference is due to the small size of the stem regulon; however, examination of the largest functional group (‘Metabolism’) in the leaf and stem regulons contained significantly (χ2, P < 0.05) fewer genes (2) than expected (6) in the stem regulon. The composition of the AtCBF1-Poplar stem regulon was also not representative of the functional distribution of the genes printed on our array, indicating that this compositional difference has a biological basis.

Associated with the difference in regulon composition between annual leaf tissue and the perennial stem meristem tissue, another finding was that a FRY2 orthologue was down-regulated in AtCBF1-Poplar leaves. FRY2 is a double stranded RNA-binding transcriptional repressor that is known to repress expression of CBF regulon members by working at a point upstream of their transcription (Xiong et al. 2002). Like the majority of genes down-regulated in the AtCBF1-Poplar, the FRY2 1500 bp promoter lacks the DRE consensus, indicating that a CBF-responsive transcription factor or post-transcriptional process, and not CBF itself, leads to its differential expression. Nevertheless, because the fry2 mutation is known to alter ABA sensitivity in Arabidopsis seeds and seedlings (Xiong et al. 2002) and the CBF regulon in Populus was significantly enriched for the ABRE, differential expression of FRY2 in leaves (versus stems) of AtCBF1-Poplar may also have contributed to the CBF leaf/stem regulon differences. The up-regulation of PtCBF3 in AtCBF1-Poplar stem tissues also suggests that individual Populus CBF factors are, as in Arabidopsis (Novillo et al. 2004), regulated by their own paralogues. PtCBF promoters contain numerous ICE1 binding sites, and the genome encodes ICE1-like proteins (data not shown).

Despite the broad similarity between the Populus and Arabidopsis CBF leaf regulons and the substantial overlap of the AtCBF1-Poplar leaf regulon with the endogenous transcriptional response to low temperature, we found no significant enrichment of the DRE (RCCGAC) in the 1500 bp promoters of our identified Populus CBF regulon. The lack of DRE enrichment may have been due to the small number of genes identified in the regulon, but the current lack of annotated transcriptional start sites in the Populus genome assembly, the larger genome and intergenic size of Populus or an altered AtCBF1-binding consensus sequence may all have contributed to this. In contrast, the ABRE, previously shown to be enriched in CBF3 regulon promoters and to impart ABA-responsiveness (Hattori et al. 2002), showed significant positional enrichment (between 100 and 400 bp upstream of the ATG) and co-occurrence with the DRE in our Populus CBF regulon, indicating that this cis-element relationship is important.

As noted earlier, the accumulation of CBF transcripts and the activity of the CRT regulatory motif in Arabidopsis are also modulated by the presence and quality of light during cold stress (Kim et al. 2002; Fowler et al. 2005) and that CRT-driven GUS reporter activity is altered in phytochrome mutant backgrounds, suggesting that day length mediates expression of the CBF regulon through phytochrome B in Arabidopsis (Kim et al. 2002). Pre-treatment with short days has also been shown to enhance the expression of low temperature-responsive, CRT-containing genes in birch (Puhakainen et al. 2004) and a CBF-like transcription factor is induced by short days in the cambial meristem of poplar (Schrader et al. 2004). These data suggest that CBF or related transcriptional regulators may be involved in the day-length-mediated growth cessation in woody species. ABA is known to mediate the response of many abiotic stress-response genes (Chen, Li & Brenner 1983; Mäntyla, Lång & Palva 1995) via activation of the ABRE (Marcotte et al. 1989; Lam & Chua 1991), and initiation and maintenance of bud dormancy in woody species has been associated with high endogenous ABA levels (Wright 1975; Rinne, Saarelainen & Junttila 1994; Rinne, Tuominen & Junttila 1994; Rohde et al. 2002; Li et al. 2003b). The strong concurrence of the DRE and ABRE in the promoters of our Populus CBF regulon supports the conclusion that there is direct interaction between the day-length-dependent ABA and low-temperature-dependent CBF signalling pathways. Although unravelling the role(s) that the different PtCBFs play in this interaction will require additional studies, our data nevertheless indicate that in poplar the initial perception of low temperature (biologically indicated by the accumulation of CBF transcripts in WT trees and artificially mimicked by ectopic expression of AtCBF1 transcript in AtCBF1-Poplar) results in up-regulation of a functionally different CBF regulon in annual leaf and perennial stem tissues. This conclusion is supported by the differential inducibility of the endogenous PtCBFs in Populus. Although all four PtCBFs are cold-inducible in leaf tissue, only PtCBF1 and PtCBF3 are significantly induced in the stem. PtCBF1/2 and PtCBF3/4 form distinct phylogenetic clusters, and one gene from each cluster is tissue-specific, suggesting that the different CBF gene clusters may be controlling slightly different CBF regulons in the different tissues. In accordance with their predicted role in cold acclimation in trees (protection of temperature-sensitive tissues such as buds and flowers during spring flushing), the poplar expressed sequence tag (EST) sequencing project (Sterky et al. 2004) has also detected PtCBF3 expression in flower buds (as well as roots), and PtCBF1 in dormant buds.

Overall, our studies of the CBF regulon in Populus have demonstrated that the central role played by the CBF family of transcriptional activators in cold acclimation of herbaceous annuals such as Arabidopsis has been maintained in temperate woody perennials, despite the evolutionary pressure imposed by the development of a winter dormant phenotype. However, these results also point to several key differences when we compare results from Populus with Arabidopsis: (1) unlike data from all herbaceous species, two of the four poplar CBF paralogues show differential expression in leaf (annual) and stem (perennial) tissues; (2) the functional composition of the stem and leaf AtCBF1-Poplar regulons are different, perhaps reflecting in part the different transcriptomes of these divergent tissues and also possibly indicating that different functional requirements in these tissues, have driven the evolution of specific leaf and stem cold-responsive CBF paralogues; and (3) this divergence in functional composition is reinforced by the finding of leaf- and stem-specific AtCBF1-Poplar regulon members. Taken together, these data suggest that the perennial-driven evolution of winter dormancy has led to the development of specific roles for abiotic stress response regulators, such as the CBFs, in annual and perennial tissues. Future studies exploring the FT and transcriptomic changes in poplar constitutively/inducibly expressing native PtCBFs should help elucidate these specific roles in greater detail.


We thank Joel Davis for excellent technical assistance. We also thank Dr Mike Thomashow for providing the original AtCBF1 expression cassette, Zoran Jeknic for providing the plasmid construct pGAH35SAtCBF1 used in this study, and Andreas Sjödin for assistance with programming the normalization protocol in ‘R’. This study was supported in part by grants from the Swedish Forestry and Agricultural Research Council to VH, from Statens Energimyndigheten (STEM) to RB, from Natural Sciences and Engineering Research Council (NSERC) to NPAH and from the National Science Foundation (NSF) Plant Genome Project (DBI 0110124) to THHC.