Cambial meristem dormancy in trees involves extensive remodelling of the transcriptome


  • Jarmo Schrader,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, 90183 Umeå, Sweden
    Search for more papers by this author
    • Present address: ZMBP, Developmental Genetics, Auf der Morgenstelle 3, 72076 Tübingen, Germany.

  • Richard Moyle,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, 90183 Umeå, Sweden
    Search for more papers by this author
    • Present address: Department of Botany, University of Queensland, QLD 4072 Brisbane, Australia.

  • Rupali Bhalerao,

    1. Department of Plant Physiology, Umeå Plant Science Centre, Umeå University, 90187 Umeå, Sweden
    Search for more papers by this author
  • Magnus Hertzberg,

    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, 90183 Umeå, Sweden
    Search for more papers by this author
    • §Present address: SweTree Technologies, PO Box 7981, S-907 19, Umeå, Sweden.

  • Joakim Lundeberg,

    1. Department of Biotechnology, Royal Institute of Technology, 10044 Stockholm, Sweden
    Search for more papers by this author
  • Peter Nilsson,

    1. Department of Biotechnology, Royal Institute of Technology, 10044 Stockholm, Sweden
    Search for more papers by this author
  • Rishikesh P. Bhalerao

    Corresponding author
    1. Department of Forest Genetics and Plant Physiology, Umeå Plant Science Centre, Swedish University of Agricultural Sciences, 90183 Umeå, Sweden
    Search for more papers by this author

*For correspondence (fax +46 90 7868165; e-mail


The establishment of the dormant state in meristems involves considerable physiological and metabolic alterations necessary for surviving unfavourable growth conditions. However, a global molecular analysis of dormancy in meristems has been hampered by the difficulty in isolating meristem cells. We used cryosectioning to isolate purified cambial meristem cells from the woody plant Populus tremula during active growth and dormancy. These samples were used to generate meristem-specific cDNA libraries and for cDNA microarray experiments to define the global transcriptional changes underlying cambial dormancy. The results indicate a significant reduction in the complexity of the cambial transcriptome in the dormant state. Although cell division is terminated in the dormant cambium, the cell cycle machinery appears to be maintained in a skeletal state as suggested by the continued presence of transcripts for several cell cycle regulators. The downregulation of PttPIN1 and PttPIN2 transcripts explains the reduced basipetal polar auxin transport during dormancy. The induction of a member of the SINA family of ubiquitin ligases implicated in auxin signalling indicates a potential mechanism for modulation of auxin sensitivity during cambial dormancy. The metabolic alterations during dormancy are mirrored in the induction of genes involved in starch breakdown and the glyoxysomal cycle. Interestingly, the induction of RGA1 like gene suggests modification of gibberellin signalling in cambial dormancy. The induction of genes such as poplar orthologues of FIE and HAP2 indicates a potential role for these global regulators of transcription in orchestrating extensive changes in gene expression during dormancy.


Plant meristems have the ability to alternate between active growth and dormancy and the establishment of the dormant state plays a key role in the survival of adverse environmental conditions. For example, in the plants of the boreal forest, the transition to dormancy is essential to survive the harsh winter temperatures and failure to cease growth compromises the ability to survive (Olsen et al., 1997; Welling et al., 2002). While the transition to dormancy is apparent at the whole-plant level, some of the most dramatic changes occur in the meristems, which suspend their normal function of cell production and enter a state in which their ability to respond to growth-promoting signals is transiently inhibited. However, the difficulty in isolating meristems has precluded the molecular analysis of dormancy in this tissue. In this respect, the vascular cambium of aspen is an excellent experimental system as size and organization of the cambial meristem makes it possible to sample meristematic tissues with high purity and resolution using tangential cryosections (Uggla et al., 1996). Furthermore, techniques for global transcript profiling and hormone measurements in nearly single cell layers of cambial meristem have been developed (Hertzberg et al., 2001; Schrader, 2003; Uggla et al., 1996). Importantly, in contrast to seed or bud dormancy, there are fewer if any developmental alterations during cambial dormancy that can overlap with and therefore complicate the investigation of dormancy-related processes.

In addition to the cessation of cambial cell proliferation, a wide range of equally important alterations are associated with the establishment of dormancy in the cambium. These include induction of cold hardiness, desiccation tolerance and the accumulation of storage compounds (Clausen and Apel, 1991; Kozlowski and Pallardy, 2002; Rowland and Arora, 1997; Welling et al., 1997). The transition to dormancy also involves extensive changes in the cellular structure such as thickening of the cell walls and alterations in the vacuolar structure (Ermel et al., 2000; Farrar and Evert, 1997). The execution of these physiological and metabolic changes during dormancy is regulated through a complex interplay between environmental and hormonal signals. For example, day length acting via the phytochrome-signalling pathway plays a key role in growth cessation (Olsen et al., 1997). At the same time, the cambial meristem switches from an auxin sensitive to insensitive state (Little and Bonga, 1974). Another growth regulator, (abscisic acid) ABA, has also been implicated in various aspects of dormancy including growth cessation and induction of cold hardiness (Harrison and Saunders, 1975; Welling et al., 2002).

The hormonal and environmental signals most likely exert their effect on the induction and maintenance of dormancy via activation and repression of diverse gene expression programmes. Dormancy has been studied extensively in seeds and buds (for reviews see Anderson et al., 2001; Bradbeer, 1988; Lang, 1994) and the changes in gene expression during dormancy have been described to some extent. While these studies are highly informative for understanding dormancy (Bewley, 1997; Rohde et al., 2000), most of these experiments have been performed on complex tissue mixtures. As a result, it is unclear as to what extent the changes in gene expression describe the dormancy-associated molecular events in the meristem itself. Consequently, our knowledge of the molecular targets of diverse signals orchestrating the establishment of dormancy in the cambium and other experimental systems remains rudimentary.

Here we describe a genomics approach to investigate cambial dormancy utilizing a combination of large-scale expressed sequence tag (EST) sequencing and global transcript profiling. Our data identified dramatic changes in the expression of genes in pathways involved in the induction of cold and draught tolerance. We show that the transition to dormancy activates metabolic pathways related to starch degradation and glyoxysomal cycle as well as accumulation of storage proteins. While the genes related to the cell cycle machinery were coordinately downregulated it appears that even in the dormant state components of the cell division machinery are maintained at a skeletal level. Potential regulators of cell proliferation and differentiation showed gene-specific changes and suggest a close interaction between proliferation and meristem size maintenance machinery. While the results highlight the extensive remodelling of the cambial transcriptome during dormancy, the data also hint at a possible role of translational control in this process, which has not been studied in any detail to date.

Results and discussion

The establishment of the dormant state is associated with a reduction in cambial transcriptome complexity

Tangential cryosections of actively growing and dormant vascular cambium of Populus tremula were used to isolate meristem-specific mRNA for the construction of cDNA libraries and microarray analysis. Following EST sequencing of the cDNA libraries, a total of 4406 active and 3649 dormant cambium sequences passed the quality control, representing a success rate of 80 and 86% for the active and dormant libraries respectively. The EST sequences described here were assembled into sequence contigs and clusters together with approximately 25 000 additional sequences isolated from various aspen tissues as part of a wider aspen EST sequencing programme (Bhalerao et al., 2003b). While the 4406 active cambium sequences assembled into 3076 clusters comprising 1598 contigs and 1478 singletons, the 3649 dormant cambium sequences exhibited a higher redundancy, assembling into only 1696 clusters with 950 contigs and 746 singletons. Only 378 contigs contained both an active and dormant cambium EST. The two largest clusters in the dormant cambial library contain more than 10% of the dormant clones sequenced and half of the dormant cambium clones can be assembled into only 120 clusters, that is 50% of all dormant clones are accounted for by only 7% of all dormant cambium clusters. In contrast, 33 clusters contain the most abundant 10% of the active cambium clones and 874 clusters are required to account for 50% of all isolated active cambium clones sequenced. A comparison of the cumulative percentage of total transcriptional activity further emphasizes the reduced complexity of the dormant cambium cDNA libraries (Figure 1). That the complexity of the dormant cambial library indeed is low is also supported by the observation that all other poplar cDNA libraries analysed so far (including active cambium, wood, tension wood and leaf) have nearly twice the number of assembled clusters from a comparable number of sequenced clones (Bhalerao, 2003). The reduction in dormant cambial transcriptome complexity is explained partly by reduction in the transcription of genes involved in processes typical for active meristems such as cell division, expansion and differentiation and simultaneous strong induction of a few genes such as those for storage protein and stress response (accounting for approximately 10% of all transcripts). In plants, such a radical change in transcriptome complexity has only been described for pollen tissue to date (Honys and Twell, 2003).

Figure 1.

Complexity of transcriptomes. The gene clusters from the active and dormant cambium libraries were ordered according to their clone abundance, from clusters containing the largest number of clones down to singletons. The figure indicates for a given percentage of clones in a library, how many clusters (i.e. how many different genes) are represented by these clones. For example, the most abundant 30% of dormant cambium clones fall into only 14 clusters while the same percentage of active cambium clones represents 296 clusters.

Functional classification of active and dormant cambial cDNA libraries

The active and dormant cambial ESTs were classified into functional classes using the MIPS classification system (Mewes et al., 2002; Schoof et al., 2002). A comparison of the distribution of active and dormant cambial ESTs into functional classes revealed surprisingly few differences between the transcriptomes represented by the two cDNA libraries (Figure 2). For this comparison, the ESTs for bark storage protein (Clausen and Apel, 1991; Coleman et al., 1993, 1994) were excluded from the analysis as its transcripts accounted for 20% of all ESTs in the dormant cDNA library and would have therefore strongly skewed the analysis. One of the notable differences between the two libraries was the comparatively larger ‘cell rescue and defence’ class in the dormant cambium library caused by the relatively large number of clones encoding the late embryogenesis abundant (LEA) family (Ingram and Bartels, 1996; Thomashow, 1999), osmotin (Larosa et al., 1992; Narasimhan et al., 2001) and ERD4 (Thomashow, 1999) that are normally induced in response to draught and cold stress. The ‘protein synthesis’ class was much larger in active cambium as evidenced by the large number of ESTs for EF1-alpha (Browning et al., 1990) and Polyubiquitin (Callis et al., 1995) reflecting the need to produce large amounts of proteins for newly forming cells as well as the turnover of proteins in the dividing cells of the active meristem. It is important to note however that in spite of the high level of activity, only 9% of active cambial transcriptome is devoted to metabolism and energy subclass compared with 43% in actively growing leaves (Bhalerao et al., 2003a).

Figure 2.

Functional classification. Clones from active and dormant libraries as well as clones changing more than fourfold in the microarray analysis were grouped into functional classes based on the MIPS classification system. The graph only shows the percentage of classified clones. Unclassified clones and clones with no significant hit against known proteins make up the difference to 100%. In case of the dormant library the 728 clones representing bark storage protein were removed. Note that the values for the libraries are based on individual expressed sequence tags (ESTs), so that the size of a functional class reflects the number of transcripts associated with that function. In particular, a functional class might contain many ESTs coming from a few distinct genes or it might contain ESTs representing a large number of different genes with only a few ESTs for any given gene. In case of the microarray results, each gene is only represented by one or very few clones on the array. The size of a functional class therefore more closely reflects the number of different genes in that class.

The remaining functional classes were comparable in size in active and dormant cambial cDNA libraries indicating that the relative number of transcripts associated with a certain type of cellular process remains unchanged between active and dormant cambium (Figure 2). However, it is likely that the type as well as the number of transcripts per gene that constitute a particular functional class might be significantly different in the active and dormant cambium transcriptome. In this respect it is interesting to note that while the dormant tissues are commonly assumed to have reduced metabolic activity, the fraction of ESTs that constitute metabolism and energy classes appear to be comparable between the two cDNA libraries. A closer look however indicates a qualitative difference between these two classes with a greater percentage of secondary metabolism in dormant compared with active cambium. Remarkably, a large fraction of transcripts in the active cambium, even some of the most abundant ones, lack functional information. While this may result from short sequence length for some of these genes, nevertheless the number of genes with undefined functional description in the active cambium is much greater than that observed for the dormant cambium. This illustrates our limited knowledge of factors involved in cambial meristem organization and the need to identify and characterize genes involved in this tissue.

Extensive differential gene expression during the establishment of cambial dormancy

The EST data have limited utility in defining the changes in the expression of individual genes especially if they are lowly expressed. Therefore we used the ‘POP1’ poplar cDNA microarray (Andersson et al., 2004) to identify genes differentially expressed between active and dormant cambium (Table S1). Initial analysis indicated far more differential expression than typically encountered in microarray experiments. Large numbers of changing genes provide a challenge for signal normalization as has been reported earlier in stationary phase yeast and human cell lines following heat shock treatment (van de Peppel et al., 2003). Therefore, two alternative normalization strategies were evaluated as described in Materials and methods. When the expression ratio of a given gene was calculated relative to the change in the majority of all clones present on the array, 1773 and 1620 clones were determined to be upregulated in the active and dormant cambium respectively. Alternatively, if the expression ratios reflected a change in the expression of a gene relative to all transcripts present in the cell, the corresponding numbers were 4674 and 706.

Entry into dormancy probably involves downregulation of many genes and simultaneous dilution of the transcriptome by a few highly expressed genes. Supposing the majority of genes were unchanged in their transcription, increasing the total transcript pool by the transcripts for massively induced genes such as that for storage protein would appear as an apparent downregulation of most genes when measured relative to all mRNA. Thus, while the detection of absolute transcripts per cell would clearly be the most desirable method for normalization, such measurements are not feasible using spotted cDNA microarrays and may not be amenable to samples consisting of one to three cell layers. As the relative abundance (provided by the first strategy) rather than the absolute level of a component was important for the analysis of metabolic pathways and regulatory networks, changes relative to all clones on the array were used for defining differentially expressed genes for further analysis. Furthermore, this strategy is simple, provided the best reproducibility between hybridizations and best reflected the changes in the relative contribution of different genes to a cellular process.

Changes in cambial meristem structure during dormancy influence the detection of differential gene expression

An important issue to be considered was the potential influence of alterations in cambial meristem structure accompanying the transition to dormancy on the identification of differentially expressed genes. The cambial meristem is reduced from five to 15 cell layers in the active state to less than four cell layers in the dormant state (Figure 3; Larson, 1994). In particular, proliferating xylem mother cells are absent from the dormant sample. Any gene expressed specifically in these cells will thus appear strongly downregulated in the dormant cambium. For such genes we can be confident that there is less gene product in the cambial zone during dormancy. However, we cannot distinguish whether this is caused by the lack of proliferating cells or gene-specific downregulation. However, on the phloem side, the situation is different. The smaller meristem size during dormancy implies that the dormant sample might contain small amounts of the flanking phloem tissues, despite careful sampling. As a result, an apparent upregulation of phloem-specific genes during dormancy could be caused by the presence of phloem tissue in the dormant cambial sample rather than actual transcriptional induction during dormancy. We therefore estimated the influence of differences in tissue composition on the detection of differentially expressed genes by looking at their expression profiles across the cambial zone as described in Materials and methods (Figure 4). A gene was only considered upregulated during dormancy if it exhibited at least threefold higher induction than genes with a similar expression profile across the cambial zone. This list contains over 650 genes and a subset of them with more than 15-fold upregulation in the dormant cambium is presented in Table 1.

Figure 3.

Schematic cross sections through active and dormant aspen cambial zones. The shaded area indicates the approximate location of the tangential cryosections used for library generation and microarray hybridization.

Figure 4.

Tissue effect on active–dormant expression ratio. As a measure of tissue effect on active–dormant ratios genes were grouped into 23 clusters based on their expression profile across the cambial region. The graphs show the average expression profile of each cluster, changes are in log2 scale and the black bar indicates the position of the cambial zone. The cluster number in brackets and the median fold change between active and dormant cambium for genes in that cluster are indicated.

Table 1.  Genes upregulated more than 15-fold in the dormant cambium. Fold change refers to expression ratios between active and dormant cambium with negative values indicating higher expression in dormant tissue. Cluster fold change is the median active/dormant expression ratio of similar genes based on their expression profiles across the cambial region. The corresponding clusters are shown in Figure 4. All genes in this list have a P-value for differential expression smaller than 10−10. For genes with an unknown function the accession number of the closest Arabidopsis homologue is given
Clone IDDescriptionFold changeCluster fold changeCluster
PU03653β-amylase (EC−20−1.6(4,2)
PU10165β-amylase (EC−17 No change
PU03786Carbonyl reductase, putative−35−1.3(4,1)
PU03391Cytochrome P-450−231.4(4,5)
PU03323Cytochrome P-450−46−1.2(5,4)
PU03161GDP-mannose pyrophosphorylase (EC−23−1.2(5,4)
PU03819Glutathione peroxidase, putative−20−1.3(4,4)
PU03118Inorganic pyrophosphatase−15−1.4(5,3)
PU03658Isocitrate lyase (EC−51 No change
PU04017Malate synthase, glyoxysomal (EC−35−1.3(4,1)
PU03275Cinnamyl-alcohol dehydrogenase, putative−22−1.4(5,3)
PU11418Polyphenol oxidase−25−1.6(4,2)
Nitrogen recycling
PU03385Amino acid carrier−281.1(3,3)
PU03073High affinity nitrate transporter-like protein−22−1.6(4,2)
PU03824Major storage protein−52−2.7(5,2)
PU03741Ubiquitin fusion-degradation protein−33−1.6(4,2)
Response to cold and osmotic stress
PU03458Dehydrin, LEA Class 2 (D-11 Family)−25−1.4(5,3)
PU03389LEA Class 3 (D-7 Family)−19−1.3(4,4)
PU03296LEA Class 3 (D-7 Family)−38−1.3(4,1)
PU03479LEA Class 4 (D-95 Family)−47−1.4(5,3)
PU03165Low temperature and salt-responsive protein−36−1.6(4,2)
PU03647SRC2-like cold-induced protein−361.3(3,4)
PU03827Thaumatin/osmotin-like protein−63 No change
PU03211Water stress-induced protein, putative−42−1.6(4,2)
Other stress response
PU03594Class I basic chitinase−40 No change
PU03656Class IV endochitinase−20 No data
PU10963Early light-induced, ELIP2-like−291.3(3,1)
PU03586GASA/LTCOR11 family−22−1.2(5,4)
PU03732GASA/LTCOR11 family−381.9(2,3)
PU03382Leucine-rich repeat disease resistance protein−27−1.2(4,3)
PU03488Senescence-associated protein−30−1.3(4,4)
PU03556Stress-related protein, putative−23−1.3(4,4)
Signal transduction
PU04119AP2 domain protein−243.3(2,5)
PU03208CONSTANS B-box zinc finger family protein−24−1.3(4,4)
PU03696Fertilization-independent endosperm protein (FIE)−55−1.4(5,3)
PU03397Peptidyl-prolyl cistrans isomerase (cyclophilin)−17−1.2(4,3)
PU03640RGA1-like, gibberellin response modulator−26−1.4(5,3)
Cell wall-related
PU03528Extensin-like protein−44 No change
PU03624Extensin-like protein−22−1.3(4,4)
PU03679Extensin-like protein−34 No data
PU03276Glycine-rich cell wall protein−30 No data
PU03340Pectin esterase inhibitor domain−26−1.2(4,3)
PU03171Xyloglucan endotransglycosylase, PttXET;E−32 No data
PU04020AAA-metalloprotease FtsH−28−1.2(5,4)
PU03530Copper transport protein−27−1.3(4,1)
PU03223Mannose-binding lectin family−34 No change
PU03277Plastid lipid-associated protein−19 No change
PU03486Pore protein−60−1.6(4,2)
PU12353Tumour-related protein−32 No change
Unknown function
PU03851Expressed protein; At1g19530−31−1.3(4,4)
PU03849Expressed protein; At4g39550−30−1.3(5,5)
PU03837Expressed protein; At2g15880−28−1.4(5,3)
PU03839Expressed protein; At2g38905−63−1.3(4,1)
PU03137Expressed protein; At2g47115−501.1(3,3)
PU03543Expressed protein; At4g39140−40−1.4(5,3)
PU03847Expressed protein; At4g27450−38 No change
PU03501Expressed protein; At1g29760−29−1.4(5,3)
PU03447Expressed protein; At2g15960−28−1.4(5,3)
PU03776Expressed protein; At5g66780−26−1.3(4,1)
PU01318Expressed protein; At1g29395−26−1.2(5,4)
PU03152Expressed protein; At1g53035−25 No change
PU03555Expressed protein; At2g42450−22−1.6(4,2)
PU03261Expressed protein; At5g51340−19 No change
PU03348Expressed protein; At3g55570−181.3(3,1)
PU03724Expressed protein; At5g03380−26−1.2(5,4)
PU03074Expressed protein; At1g64890−15−1.6(4,2)
PU03835Expressed protein; At1g71010−17−1.2(3,2)
PU03234Expressed protein; At5g07630−15−1.4(5,3)
PU05333Expressed protein; At5g63500−16−1.2(5,4)
PU03830No hit−92−1.2(5,4)
PU03588No hit−19−1.2(3,2)
PU03772No hit−26−1.4(5,3)
PU12425No hit−45−1.4(5,3)

Cessation of cell division does not cause a global repression of cell cycle genes during dormancy

During dormancy, cell division is suppressed and the cambial cells are supposed to be arrested in the G1 phase (Mellerowicz et al., 1989; Zhong et al., 1995). In agreement with this, several aspen homologues of core cell cycle genes (Vandepoele et al., 2002) present on the microarray were downregulated in the dormant cambium (Table 2). Interestingly however, the expression of a number of core cell cycle genes did not change significantly upon entry into dormancy (Table 2 and data not shown). These include aspen homologues of D-type and G2-M phase mitotic cyclins as well as homologues of critical regulators of the G1 to S phase transition like E2F, DP1 and Retinoblastoma. The cyclin-dependent kinase CDKC was even slightly upregulated. While the continued presence of a potential regulator blocking G1 to S phase transition such as Retinoblastoma could be expected (Dyson, 1998; Gutierrez, 1998), unchanged expression of D-type and mitotic cyclins is intriguing given their association with actively dividing cells (Sorrell et al., 1999). It is possible that these genes could be regulated post-transcriptionally during dormancy as observed for PttCDKA, an A-type CDK in poplar (Espinosa-Ruiz et al., 2004). Importantly, these data indicate that the entry into dormancy does not lead to a global downregulation of cell cycle gene transcription. The continued presence of certain cell cycle components at the transcriptional level might reflect a mechanism that maintains the cell cycle machinery in a skeletal state ready for reactivation during spring.

Table 2.  Genes involved in cell proliferation, starch degradation and signal transduction. Fold change refers to expression ratios between active and dormant cambium with positive and negative values indicating higher expression in active and dormant tissue respectively. Cluster fold change is the median active/dormant expression ratio of similar genes based on their expression profiles across the cambial region. The corresponding clusters are shown in Figure 4
Clone IDDescriptionFold changet-test P-valueCluster fold changeCluster
Cell cycle
PU06527Cyclin A13.81.2 × 10−23.3(2,5)
PU13262Cyclin A22.37.1 × 10−32.1(1,3)
PU01217Cyclin B22.76.9 × 10−32.5(2,4)
PU13230Cyclin D38.83.9 × 10−34.3(1,2)
PU10071Cyclin-dependent kinase CDKA12.52.4 × 10−4 No change
PU00348Cyclin-dependent kinase CDKB214.22.4 × 10−47.7(1,5)
PU03544Cyclin-dependent kinase CDKC-1.53.7 × 10−4 No change
PU10385DPB-like transcriptional regulator1.07.0 × 10−1 No change
PU04263E2F-like transcriptional regulator2.85.6 × 10−37.7(1,5)
PU01146Retinoblastoma protein1.63.7 × 10−3 No change
PU02867Histone H45.83.9 × 10−52.5(2,4)
Meristem regulation
PU04170AINTEGUMENTA-like, PttANT6.86.9 × 10−57.2(3,5)
PU04960CLAVATA1-like, PttCLV11.14.3 × 10−1−1.2(5,4)
PU00931CLAVATA1-like, PttRLK38.23.0 × 10−48.3(1,4)
PU07962Homeobox protein, PttHB8−1.07.4 × 10−18.3(1,4)
PU07175Homeobox protein, PttHB311.37.4 × 10−48.3(1,4)
Cell wall modification
PU00431Basic cellulase22.94.0 × 10−37.7(1,5)
PU02594Expansin, PttEXP132.31.1 × 10−57.7(1,5)
PU07472Pectin methylesterase11.39.9 × 10−48.3(1,4)
PU07373PttXET;C9.91.2 × 10−37.7(1,5)
PU01546PttXET16A3.66.7 × 10−32.1(1,3)
PU02607Callose synthase (1,3-beta-glucan synthase)−4.21.1 × 10−14−1.2(5,4)
PU02018Callose synthase (1,3-beta-glucan synthase)−1.51.6 × 10−5 No change
Starch degradation
PU11145α-amylase (EC−1.98.5 × 10−2−2.7(5,2)
PU07898α-amylase (EC−2.52.7 × 10−4 No data
PU12255Debranching enzyme (EC−1.21.2 × 10−1−1.3(4,1)
PU03913Starch phosphorylase (EC−4.57.0 × 10−13−1.2(3,2)
PU02095SEX1-like protein−11.90.0 × 10−99−1.6(4,2)
PU05226SEX1-like protein−7.98.9 × 10−16−1.6(4,2)
Regulation of transcription
PU03802HAP2-like CCAAT box-binding factor−10.23.1 × 10−13−1.6(4,2)
PU03412CBF1-like DRE-binding protein−6.86.0 × 10−10−1.6(4,2)
PU11255Ribosomal protein S6 kinase−3.42.1 × 10−08 No change
PU03825Leucine zipper transcription factor PTBF1−5.05.5 × 10−10 No change

Differential regulation of key regulators of meristem activity during dormancy

The control of meristem identity and size is intricately linked with the regulation of cell division. This has been well studied in the shoot apical meristem and involves meristem regulators such as CLAVATA, SHOOTMERISTEMLESS and WUSCHEL that have also been implicated in the regulation of cell cycle genes (Brand et al., 2000; Lenhard et al., 2002; Schoof et al., 2000; Weigel and Jürgens, 2002). We find that poplar homologues of some of the key meristem regulators mentioned above are expressed in the cambial region (this study and Schrader, 2003). We analysed whether the expression of these meristem regulators is altered during cambial dormancy as cell proliferation ceases (Table 2). The expression of PttANT, the aspen homologue of ArabidopsisAINTEGUMENTA (Mizukami and Fischer, 2000) a known regulator of cell proliferation, was strongly reduced in the dormant cambium, supporting a role for this gene in proliferation-related processes in the cambial meristem (Table 2). Furthermore, two aspen CLAVATA1 (Clark et al., 1997) homologues, PttCLV1 and PttRLK3 displayed differential regulation upon entry into dormancy with PttRLK3 being downregulated whereas PttCLV1 remained unchanged. Similarly, PttHB3, a gene closely related to WUSCHEL (Mayer et al., 1998) was strongly downregulated during dormancy (Table 2). Downregulation of the CLAVATA1-like PttRLK3 and the WUSCHEL-like PttHB3 during dormancy suggests that these genes are mainly required to balance cell proliferation and maintenance and specification of undifferentiated cells during active growth but might not be needed during dormancy. The other CLAVATA1 orthologue, PttCLV1, might play a role in maintaining cell identity even during dormancy or alternatively, this gene may be required to ensure the repression of cell division during dormancy. Interestingly, PttHB8, a close homologue of the Arabidopsis transcription factor Athb-8 with expression in the procambium (Baima et al., 1995), is also not downregulated during dormancy. Additionally, several members of the KNOTTED and SCARECROW families of transcription factors were also found to be unaffected by the transition to dormancy (data not shown).

Differential regulation of auxin transport components during dormancy

Auxin has been implicated as a key signal regulating cambial cell proliferation and cambial meristem identity. During the transition to dormancy, polar auxin transport is reduced and the cambium is rendered insensitive to applied auxin (Lachaud and Bonnemain, 1984; Little and Bonga, 1974; Odani, 1975; Schrader, 2003). However, the molecular basis of these changes in auxin transport and response during dormancy are not well characterized. Therefore, we investigated the changes in the components of the auxin signalling and transport machinery during cambial dormancy (Table 3). The results indicate differential regulation of the auxin transporters during dormancy. Of all the potential transporters represented on the array, only the expression of PttPIN1 and PttPIN2 efflux carriers is clearly downregulated in dormancy. Surprisingly, no change in the expression of two other potential efflux and three influx carriers was observed during dormancy. In contrast to other carriers the downregulation of PttPIN1 and PttPIN2 might indicate that the major decrease in polar auxin transport during dormancy could be attributed to a reduction in the expression of these two genes. However, the lack of change in expression of other carriers during dormancy is intriguing and whether these genes are regulated at the post-transcriptional level needs to be analysed.

Table 3.  Genes involved in auxin transport and response. Fold change refers to expression ratios between active and dormant cambium with positive and negative values indicating higher expression in active and dormant tissue respectively. Cluster fold change is the median active/dormant expression ratio of similar genes based on their expression profiles across the cambial region. The corresponding clusters are shown in Figure 4
Clone IDDescriptionFold changet-test P-valueCluster fold changeCluster
PU02176Auxin influx carrier PttLAX11.05.2 × 10−13.3(2,5)
PU13526Auxin influx carrier PttLAX31.17.0 × 10−1 No change
PU04390Auxin influx carrier PttLAX71.07.4 × 10−11.2(2,2)
PU02299Auxin efflux carrier component PttPIN111.52.9 × 10−22.5(2,4)
PU02471Auxin efflux carrier component PttPIN24.53.5 × 10−52.5(2,4)
PU13283Auxin efflux carrier component PttPIN31.07.9 × 10−1 No change
PU12696Auxin efflux carrier component PttPIN42.07.6 × 10−3 No data
PU00681Aux/IAA family protein PttIAA12.42.8 × 10−3−1.2(3,2)
PU01384Aux/IAA family protein PttIAA21.83.7 × 10−23.3(2,5)
PU02187Aux/IAA family protein PttIAA33.51.0 × 10−2 No change
PU00544Aux/IAA family protein PttIAA41.68.7 × 10−23.3(2,5)
PU13252Aux/IAA family protein PttIAA5−1.22.2 × 10−1 No change
PU09140Aux/IAA family protein PttIAA7−1.11.7 × 10−1 No change
PU01813Aux/IAA family protein PttIAA84.77.5 × 10−44.3(1,2)
PU12361Aux/IAA family protein PttIAA92.21.6 × 10−31.9(2,3)
PU13492Aux/IAA family protein PttIAA12−1.11.5 × 10−11.1(3,3)
PU12917Auxin response factor (ARF family)−1.14.4 × 10−11.9(2,3)
PU13459Auxin response factor (ARF family)1.12.1 × 10−1 No data
PU12041Auxin response factor (ARF family)1.71.9 × 10−21.5(2,1)
PU11863Auxin response factor (ARF family)2.35.3 × 10−31.5(2,1)
PU06171Auxin response factor (ARF family)1.06.2 × 10−1 No change
PU08615Auxin response factor (ARF family)−3.62.1 × 10−3−1.6(4,2)
PU04864Auxin response factor (ARF family)−1.41.0 × 10−2 No change
PU01346Auxin response factor (ARF family)−1.11.8 × 10−1 No change
PU12989Auxin response factor (ARF family)−1.68.3 × 10−3 No change
PU07193Auxin response factor (ARF family)1.43.0 × 10−2 No change
PU13175RUB-activating enzyme AXR1−1.13.4 × 10−2 No data
PU11951RUB-activating enzyme AXR11.56.0 × 10−3 No change
PU00526Cullin1-like protein1.61.5 × 10−4 No change
PU07634Cullin1-like protein−1.76.0 × 10−5−1.3(4,1)
PU07079Cullin1-like protein−1.64.1 × 10−5 No change
PU05163TIR1-like F-Box protein−1.11.8 × 10−11.1(3,3)
PU13503TIR1-like F-Box protein1.32.4 × 10−21.5(2,1)
PU05197TIR1-like F-Box protein−1.52.2 × 10−4 No change
PU01339SCF complex subunit SKP11.66.8 × 10−4 No change
PU07143SCF complex subunit SKP1−1.91.6 × 10−7 No change
PU09690SCF complex subunit SKP1−2.13.1 × 10−9−1.2(4,3)
PU01815SCF complex subunit SKP1−1.63.6 × 10−7 No change
PU00867Seven in absentia (SINA) family−7.54.0 × 10−15 No change

Regulation of hormonal responses during dormancy

While plant hormones play an important role in dormancy, our knowledge of the hormone-signalling components and their regulation during bud and cambial dormancy is limited. In the cambium, auxin responsiveness is repressed during endodormancy and this transient repression of auxin responsiveness is thought to play a key role in maintaining the dormant state (Little and Bonga, 1974). Therefore, we examined the expression of key auxin-signalling components, the auxin-inducible Aux/IAA genes encoding short-lived transcriptional repressors (Gray et al., 2001; Tiwari et al., 2001) and their interacting partners – the auxin response factors (ARFs, Ulmasov et al., 1997). Several Aux/IAA genes were downregulated, but the expression of the majority of ARF and Aux/IAA genes was unaltered during dormancy (Table 3). These results suggest that dormancy-inducing signals could differentially modulate auxin-signalling components. Similarly, the transcripts for components of the SCF complex known to mediate auxin-dependent turnover of AUX/IAA proteins (Gray et al., 2001) such as SKP, Cullins, RUB1 conjugating enzyme as well as AXR1 and TIR1 homologues were also not affected by the transition to dormancy (Table 3). Based on the present study we can conclude that downregulation of the AXR1/TIR1 pathway for auxin-mediated degradation of Aux/IAA genes is probably not the cause for decreased auxin sensitivity as none of the components of this pathway changes notably upon entry into dormancy. It is worth mentioning, however, that a gene that is closely related to SINAT5 is upregulated during dormancy (Table 3). ArabidopsisSINAT5 modulates auxin responses by targeting the NAC1 transcription factor for degradation (Xie et al., 2002), and its overexpression results in repression of lateral root formation, a process stimulated by auxin. We find several NAC1-like genes expressed in cambial tissues (data not shown) and it will be interesting to find out whether these could be potential targets of poplar SINA during dormancy. Thus, the upregulation of a poplar gene closely related to SINAT5 suggests a potential mechanism contributing to the repression of auxin responses during dormancy.

Gibberellins are another class of hormones that have been implicated in the transition to dormancy. The levels of gibberellins decrease upon exposure to short days preceding the cessation of apical growth (Olsen and Junttila, 1997). However, there has been little experimental evidence for the role of gibberellins in cambial dormancy. In this respect it is worth mentioning that a poplar gene PttRGA1 highly similar to RGA1, a repressor of gibberellin responses (Dill et al., 2001), is highly upregulated in the dormant cambium. This may be an indication of a modulation of gibberellin signalling in cambial dormancy and if confirmed in functional studies, would provide yet another example of conservation of regulatory mechanism underlying dormancy in the apical meristem and cambium.

Stress and adaptive response genes are strongly upregulated in the dormant cambium

Freezing temperatures in winter can cause damage to cellular structures and consequently a significant fraction of the dormant cambial transcriptome is devoted to combating cold stress. The survey of the genes induced in the dormant cambium suggests the activation of multiple mechanisms that enable the cambium to survive low temperatures. The induced genes belong to several categories, for example osmotin, dehydrins, chitinases, poplar homologues of the GASA1-like genes LtCOR11 and LtCOR12 and other proteins induced by water, salt and cold stress of unknown function (Table 1). Many of these, for example the LEA proteins, accumulate in a wide variety of plants in response to water deficit (Ingram and Bartels, 1996; Thomashow, 1999) and cold-induced members of chitinases and osmotin gene families function as cryoprotectants (Griffith et al., 1997). There is an intimate connection between osmotic stress and cold tolerance (Close, 1997; Siminovitch and Cloutier, 1983; Thomashow, 1999). Cold tolerance involves reduction in cellular water content (Rinne et al., 1998) and both processes involve members of the same gene families (Rowland and Arora, 1997). It is therefore likely that most, if not all the osmotic stress-related genes upregulated in the dormant cambium are part of the acquisition of cold hardiness process. Additionally, genes involved in the generation of sugars such as sucrose and raffinose that have been proposed to act as cryoprotectants are also upregulated in the dormant cambium (data not shown). It is important to note that low temperature induces the production of hydrogen peroxide that can damage membrane lipids via lipid peroxidation (Foyer et al., 1997; Kocsy et al., 2001). This might explain the upregulation of genes encoding glutathione-S-transferase and glutathione peroxidase involved in combating oxidative stress caused by low temperatures during cambial dormancy.

Starch breakdown during dormancy involves induction of amylolytic genes

The establishment of dormancy involves considerable alterations in cellular metabolism. Conversion of starch to sugar is a key metabolic process associated with the entry into dormancy as starch-derived sugars serve several purposes, for example as cryoprotectants as well as a source of energy. Starch breakdown is a multi-step process involving several enzymes whose regulation with respect to dormancy is not fully understood. A key finding is a temporal correlation between upregulation of total- and alpha-amylase activity with the period of starch breakdown during dormancy (Witt and Sauter, 1994). However, in the absence of purified enzymes and in complex mixtures, it is difficult to judge the relative contribution of different amylolytic enzymes to the observed increase in degradation. For example, alpha-glucosidases are capable of degrading starch granules at rates comparable to alpha-amylases (Sun and Henson, 1990). It is further difficult to judge the relative importance of transcriptional and post-transcriptional regulation. For example, an isoform of alpha-amylase is present at all times in stems of poplar but only binds to starch granules upon lowering of temperatures (Sauter et al., 1998). Our data show that in addition to the two genes encoding beta-amylase that are induced more than 15-fold, transcripts for starch phosphorylase as well as for a recently described R1 protein involved in starch breakdown (Ritte et al., 2002; Yu et al., 2001) are also significantly induced during dormancy (Table 2). In contrast, only a moderate increase in the expression of two aspen genes for alpha-amylase is observed during dormancy. These results identify potential components of the starch degradation pathway during dormancy and indicate the role of transcriptional regulation in this process.

Genes of the glyoxalate cycle are induced during dormancy

Similar to other experimental systems, the cambium accumulates storage compounds such as carbohydrates, proteins or lipids (Nelson and Dickson, 1981). In contrast to carbohydrates and storage proteins, the role of storage lipids in the context of cambial dormancy is unclear. Therefore, the induction of malate synthase and isocitrate lyase genes of the glyoxysomal cycle during dormancy is intriguing. The glyoxysomal cycle can either be used to generate energy from lipids by channelling them into the gluconeogenetic pathway (Beevers, 1961) or to replenish the intermediates of the citric acid cycle when these are removed at high rates to furnish carbon skeletons (Eastmond et al., 2000). Although the induction of these glyoxysome cycle genes may suggest that lipids are metabolized through the glyoxysomal cycle during dormancy to provide energy, there are several observations that do not support such a function. For example, it is difficult to imagine why lipids would be utilized via conversion to carbohydrates when massive amounts of carbohydrates are derived from starch. In addition, there is no evidence for a reduction in lipid bodies during dormancy. Finally, in Tilia, feeding of substrates for the above-mentioned enzymes to dormant tissues does not lead to their utilization (Höll, 2000). In view of this combined evidence, another possibility may be that while these enzymes are transcribed during dormancy, the transcripts are not translated before the reactivation phase in spring when demand for energy is very high.

A role for stored transcripts in reactivation during seed germination has been reported recently (Rajjou et al., 2004). If the two enzymes of the glyoxysomal cycle are indeed stored until spring, then their regulation together with that of other such transcripts would point to a hitherto not well characterized role of translational regulation during dormancy. In this respect, the induction of a poplar gene with high similarity to S6 kinase (PU11255) is interesting. The S6 kinase phosphorylates the ribosomal S6 protein thereby regulating the translation of a selected group of mRNAs (Kawasome et al., 1998; Pearson and Thomas, 1995). Interestingly, this gene is also induced during seed dormancy as well as by ABA and cold (Mizoguchi et al., 1995) and recent experiments suggest a role for this gene in the entry of stationary phase cell cultures into the active cell division phase (Turck et al., 2004).

Modulation of cell walls during dormancy

During cambial dormancy, increases in the xyloglucan and esterified galacturonan content of cambial cell walls and cell wall thickening has been reported earlier (Ermel et al., 2000; Guglielmino et al., 1997) but there is little data on the nature of enzymes involved in this process. Although a large number of genes encoding cell wall-related enzymes are downregulated in the dormant cambium when proliferation ceases, several other genes of this category are induced during dormancy (Table 1), thus providing candidates for genes involved in dormancy-related cell wall modification. The induced genes include glucan synthase, extensin as well as those encoding proteins containing a domain common to glycosyl hydrolases. Interestingly, a single member of the xyloglucan endotransglycosylase family, PttXET;E is induced during dormancy in contrast to other family members suggesting a distinct dormancy-related function for this gene. Of particular interest is the induction of a glucan synthase gene during dormancy. It has been shown that the apical meristem becomes symplastically isolated during dormancy by plugging of plasmodesmata with glucan polymers (Rinne et al., 2001). The plugged plasmodesmata prevent the plant growth regulator gibberrellic acid from entering the apical meristem cells (Rinne et al., 2001) and are supposed to contribute to the maintenance of dormancy. The induction of glucan synthase in the dormant cambium may suggest that symplastic isolation could be a regulatory mechanism for maintenance of dormancy that is conserved between the apical meristem and the cambium.

Transcriptional changes in the dormant cambium are mirrored in the induction of several transcriptional regulators

The establishment of dormancy involves the induction of a large number of genes and therefore identification of transcription factors that play a role in dormancy-regulated gene expression would be an important step in understanding this process. As shown in Table 2 we observe the induction of several transcription factors. Similarities of dormancy-induced transcription factors to well characterized proteins from other systems provide hints regarding their function in cambial dormancy. For example, an aspen gene highly similar to CBF1 is induced 6.7-fold in the dormant cambium. The Arabidopsis CBF gene family regulates gene expression in response to osmotic stress and cold temperature (Thomashow et al., 2001) and thus the induced poplar gene of this family might be involved in the regulation of some of the above-mentioned genes. Similarly, prior work has identified the poplar PFBF1 protein that binds to the promoter of the bark storage protein gene. Induction of this gene in our experiments lends further support to its role in regulating storage protein gene expression. An intriguing finding was the strong upregulation of an aspen homologue of FERTILISATION INDEPENDENT ENDOSPERM (FIE), a polycomb-family protein during cambial dormancy (Table 1). It is believed that FIE acts as part of a complex that silences the transcription of genes necessary for proliferation through modification of the chromatin structure (Grossniklaus et al., 2001; Katz et al., 2004). In this regard, induction of an aspen gene with similarities to the CCAAT-binding factor HAP2, which is part of a complex that includes the seed dormancy regulator LEC1, is potentially interesting (Table 2; Kwong et al., 2003; Lee et al., 2003). The upregulation of these two genes provides hints at how global changes in transcription could be orchestrated during the induction of dormancy.


The transition of a meristem from active growth to dormancy involves a considerable modulation of various cellular processes. This is reflected in our data in the form of extensive changes in the cambial transcriptome upon entry into dormancy. The changes include repression of numerous cellular processes as well as the induction of dormancy-specific genes leading to an overall reduction in transcriptome complexity in the dormant cambium. The central role of the activation of survival mechanisms is evident in the large proportion of adaptation and stress tolerance genes in the dormant cambium. The massive change in the physiological status of the cambium during dormancy involves global alteration of gene expression. This appears to involve not only specific transcription factors regulating a specific transcriptional programme, for example control of cold/stress-responsive genes by CBFs, but also global regulators of transcription such as FIE and LEC1-like genes. The data also underscore the similarities in dormancy-related programmes between seeds, buds and cambium as many of the regulators such as FIE and LEC1-like genes have known functions either in dormancy or related processes such as seed development. Although our data identified transcriptional control as an important mechanism in cambial dormancy, unexpected changes in several genes point towards an as yet uncharacterized role for translational regulation in this process. Importantly, the high-resolution expression patterns presented here provide a hint in understanding the function of many genes whose sequence does not provide any information with regard to their function.

Materials and methods

Plant material and RNA preparation

Samples for active and dormant cambium were obtained from stems of plantation-grown P. tremula, located near Umeå in northern Sweden. The active cambial samples were taken in July 2000 and the dormant cambial samples were taken in October 2000, approximately 8 weeks after bud set when all leaves were shed and cambial activity was absent. In the autumn of 2000, prior to taking the dormant sample, there were three nights with temperatures below 0°C while average temperatures were 6–11°C for most of September.

Serial tangential 30 μm-thick cryosections were obtained from frozen stem tissues as described earlier (Uggla et al., 1996). Transverse hand sections taken after every third tangential section were used to identify the section to be selected for mRNA extraction covering the cambial zone. For each time point, one section was used to prepare mRNA for the cDNA library while three sections from individual trees were pooled to provide material for microarray analysis.

Library construction, plasmid preparation and sequencing

Messenger RNA was obtained from the cambial sections using Dynabeads (Dynal AS, Oslo, Norway) according to the manufacturer's instructions. cDNA was prepared from mRNA using the SMART cDNA synthesis kit (Clontech; This cDNA was PCR-amplified, digested with SfiI restriction enzyme and size-fractionated by agarose gel electrophoresis. Four independent cDNA libraries were prepared from size-fractionated cDNAs of 0.5–1, 1–2, 1.5–2 and >2 kb size by ligating the cDNAs into SfiI-digested pTriplEX2 plasmid (Clontech) and transforming into Escherichia coli strain XL1-Blue and selecting for transformants on LB plates supplemented with carbenicillin. The colonies with plasmid-bearing inserts were hand picked using blue-white screening protocol.

Plasmid DNA was isolated from 1 ml overnight E. coli cultures using miniprep kits (Qiagen; on the Biomek robotic workstation (Beckman Instruments; The cDNA insert sizes of each plasmid were estimated by restriction mapping and the size distribution of cDNA inserts was compared with distribution profile of the original cDNA based on the ethidium bromide staining intensity after gel electrophoresis. A representative sequence library was maintained by preferentially sequencing clones constructed from cDNA size fractions that were found to be under-represented in the distribution of sequenced clone inserts. The bioinformatic data analysis including quality trimming of sequences, assembly of contigs and functional annotations were carried out as described previously (Bhalerao, 2003; Bhalerao et al., 2003a).

Microarray analysis

The POP1 arrays containing 13824 spots, representing 33 000 aspen ESTs was used for transcript profiling (Andersson et al., 2004). In addition to 13530 aspen clones the POP1 array contains six copies of the full set of Lucidea Universal Score Card synthetic controls (Amersham Biosciences; for normalization purposes.

Messenger RNA from active and dormant cambial sections was PCR-amplified as described (Hertzberg et al., 2001). For each sample an internal standard consisting of equal amounts of Lucidea Universal Score Card control cDNA was added to 100 ng of amplified cDNA. Subsequent labelling with fluorescent dyes (Cy3, Cy5) and hybridization to POP1 microarrays was performed as described (Hertzberg et al., 2001). Six replicate hybridizations, three of them with swapped dyes, were performed. Images were scanned on a ScanArray4000 scanner (Perkin-Elmer Life Sciences, Boston, MA, USA) at 10 μm resolution. The scanned images were quantified using QuantArray (Perkin-Elmer; with the AdaptiveCircle method calculating median signal and background intensities.

The raw signal was background-subtracted and a quality filter was applied. Genes were flagged present if signal signal ≥2 * median background in the other channel). After filtering, the data were log2 transformed. Background corrected data were imported into GeneSpring (SiliconGenetics; Two strategies were used for normalization: (i) relative to all clones on the array by applying ‘intensity dependent normalization (Lowess)’ in GeneSpring; (ii) relative to all transcripts in the tissue by additionally dividing all ratios by the median ratio (1.7) of the Lucidea Score Card control spots. Option (i) was used in the final analysis.

Genes showing significant change were selected using the CyberT web server ( This program implements a Bayesian approach where the significance cut-off for each gene depends on its signal strength and the observed variation (Baldi and Long, 2001). The following parameters were used: Window size 101 and Trust 18. Genes were declared differentially expressed if their Bonferroni-corrected P-value was less than 0.25.

Tissue effects on expression ratios

In order to estimate the influence of tissue composition on the active/dormant cambium expression ratios and correct for this potential effect on selection of differentially expressed genes, a recently completed analysis of gene expression across the cells of the cambial meristem of aspen during active growth (Schrader, 2003) was used. In their analysis Schrader and coworkers used cDNA amplified from tangential cryosections covering the cambial zone to generate expression profiles for the same set of genes as in the present study. Using this set of expression profiles, genes showing more than twofold change across the cambial meristem were divided into 23 clusters (Figure 4; Schrader, 2003). For each cluster, the median of the active versus dormant cambium ratio was calculated as an estimate of the amount of differential expression that is associated with tissue-specific expression (Figure 4a). The data reveal a correlation between expression pattern and active/dormant ratio whereby phloem- and xylem-specific genes have higher and lower expression in the dormant cambium respectively. For genes with a specific expression on the xylem side, this behaviour could be expected as many genes in this group are associated with processes such as cell proliferation and expansion which do not take place during dormancy. However, on the phloem side, an apparent upregulation of phloem-specific genes during dormancy could be explained by a larger amount of contaminating phloem tissue in the dormant cambium sample. It is important to note, however, that not all genes followed the trend and that a number of phloem-specific genes appeared downregulated in the dormant cambium. In the tables the results of the active/dormant fold change as well as the average fold change of genes in the same cluster are reported, allowing judgement of a possible tissue effect on a gene expression ratio.


We thank I. Sandström and A. Degueret for technical assistance. This work was supported by grants from Energimyndigheten and the Swedish Research Council to R.P.B.

Supplementary material

The following material is available from

Table S1 Normalized microarray expression data for 13525 EST clones found on the POP1 array in the form of a Microsoft Excel document. Data are subdivided into clones that are significantly upregulated in active cambium, in the dormant cambium and clones that do not show any significant change.