Differential stage-specific regulation of cyclin-dependent kinases during cambial dormancy in hybrid aspen

Authors

  • Ana Espinosa-Ruiz,

    1. Department of Forest Genetics and Plant Physiology, Umea Plant Science Centre, Swedish University of Agricultural Sciences, 901 83 Umea, Sweden, and
    Search for more papers by this author
      Present address: Centro Nacional de Biotecnología-CSIC, Campus de la Universidad Autónoma de Madrid, Cantoblanco, 28049 Madrid, Spain. These authors contributed equally to this work.
  • Sangeeta Saxena,

    1. Department of Forest Genetics and Plant Physiology, Umea Plant Science Centre, Swedish University of Agricultural Sciences, 901 83 Umea, Sweden, and
    Search for more papers by this author
      These authors contributed equally to this work.
  • Julien Schmidt,

    1. Department of Forest Genetics and Plant Physiology, Umea Plant Science Centre, Swedish University of Agricultural Sciences, 901 83 Umea, Sweden, and
    Search for more papers by this author
      These authors contributed equally to this work.
  • Ewa Mellerowicz,

    1. Department of Forest Genetics and Plant Physiology, Umea Plant Science Centre, Swedish University of Agricultural Sciences, 901 83 Umea, Sweden, and
    Search for more papers by this author
  • Pál Miskolczi,

    1. Biological Research Center, Institute of Plant Biology, H-6701 Szeged, Hungary
    Search for more papers by this author
  • László Bakó,

    1. Department of Forest Genetics and Plant Physiology, Umea Plant Science Centre, Swedish University of Agricultural Sciences, 901 83 Umea, Sweden, and
    2. Biological Research Center, Institute of Plant Biology, H-6701 Szeged, Hungary
    Search for more papers by this author
  • Rishikesh P. Bhalerao

    Corresponding author
    1. Department of Forest Genetics and Plant Physiology, Umea Plant Science Centre, Swedish University of Agricultural Sciences, 901 83 Umea, Sweden, and
      For correspondence (fax +46 90 7865901; e-mail Rishikesh.Bhalerao@genfys.slu.se).
    Search for more papers by this author

For correspondence (fax +46 90 7865901; e-mail Rishikesh.Bhalerao@genfys.slu.se).

Present address: Centro Nacional de Biotecnología-CSIC, Campus de la Universidad Autónoma de Madrid, Cantoblanco, 28049 Madrid, Spain.

These authors contributed equally to this work.

Summary

The cambium of woody plants cycles between active and dormant states. Dormancy can be subdivided into eco- and endodormant stages. Ecodormant trees resume growth upon exposure to growth-promotive signals, while the establishment of endodormant state results in loss of the ability to respond to these signals. In this paper, we analysed the regulation of cyclin-dependent kinases (CDKs) to understand the differential response of cell division machinery to growth-promotive signals during the distinct stages of dormancy in hybrid aspen. We show that 4 weeks of short-day (SD) treatment causes termination of the cambial cell division and establishment of the ecodormant state. This coincides with a steady decline in the histone H1 kinase activity of the PSTAIRE-type poplar CDKA (PttCDKA) and the PPTTLRE-type PttCDKB kinase complexes. However, neither the transcript nor the polypeptide levels of PttCDKA and PttCDKB are reduced during ecodormancy. In contrast, 6 weeks of SD treatment establishes endodormancy, which is marked by the reduction and disappearance of the PttCDKA and PttCDKB protein levels and the PttCDKB transcript levels. The transition to endodormancy is preceded by an elevated E2F (adenosine E2 promoter binding factor) phosphorylation activity of the PttCDKA kinase that reduces the DNA-binding activity of E2F in vitro. The transition to endodormancy is followed by a reduction of retinoblastoma (Rb) phosphorylation activity of PttCDKA protein complexes. Both phosphorylation events could contribute to block the G1 to S phase transition upon the establishment of endodormancy. Our results indicate that eco- and endodormant stages of cambial dormancy involve a stage-specific regulation of the cell cycle effectors at multiple levels.

Introduction

During the course of development, the meristematic tissues in plants cycle between active and dormant states, the dormant state being defined by a temporary suspension of visible growth (Lang, 1987). This allows the plants to synchronise the timing of their growth period with favourable environmental conditions and protect the meristems from adverse environmental conditions. One of the most striking examples of such an adaptive mechanism is the temporary suspension of growth in the meristems of perennial trees in autumn.

Transition from active to dormant state occurs gradually (Champagnat, 1983; reviewed by Dennis, 1994), and termination of the meristematic cell division is one of the first visible indicators of the establishment of a dormant state. Physiologically, dormancy can be distinguished into two stages: ecodormancy and endodormancy (Champagnat, 1983; Lang, 1987; Lang et al., 1985). A major difference between them is the differential response of the arrested cell division machinery to growth-promotive signals. The cell division machinery can be described to be in online mode during active growth followed by a switch to standby mode in the ecodormant state (Figure 1). The transition from eco- to endodormant state is mirrored by a transition of the cell division machinery from standby to offline mode, becoming unable to respond to growth-promotive signals unless dormancy-breaking signals return the machinery to standby mode (Faust et al., 1997; Little and Bonga, 1974; Rinne et al., 2001).

Figure 1.

Schematic diagram showing the different physiological phases of dormancy (adapted from the study by Rinne et al., 2001).

Actively growing trees progress into ecodormancy as an adaptative response to shorter days. Low temperature may also play a role in dormancy entry. At ecodormant stage, moving the trees into a growth-favourable environment will still be able to reverse the process (standby). When transition to endodormant stage (offline) has taken place, the process cannot be reversed until a number of ‘chilling units’ has been accumulated – that is, exposure of the trees to a period of very low temperature. After the chilling period, the trees become ecodormant again (standby mode) and the transition to active growth will naturally occur as the trees are exposed to gradually increasing temperatures.

Like other meristematic tissues, the vascular cambium also undergoes a period of dormancy, and there is considerable amount of data describing the activity–dormancy transition at the anatomical and cytological levels in cambium (Lloyd et al., 1996; reviewed by Catesson, 1994). Also, several studies have investigated the differences between eco- and endodormant states at the physiological level in the cambium and other meristems (Dennis, 1994; Faust et al., 1997; Little and Bonga, 1974). However, the molecular basis of the transition from eco- to endodormancy and the differential response of the cell division machinery during the distinct stages of dormancy remain poorly characterised. Therefore, as a first step to understand this process, we have initiated the analysis of the cell division machinery so as to elucidate the molecular basis of its differential response to growth-promoting signals during eco- and endodormancy.

We have investigated the regulation of cyclin-dependent kinases (CDKs) during the stages leading to the establishment of ecodormancy and following the transition to and establishment of endodormancy. The CDKs are key regulators of the cell cycle in eukaryotes (Mironov et al., 1999; Nigg, 1995). They are activated by association with cyclin partners, and these CDK–cyclin protein complexes ensure the orderly transitions through cell cycle by phosphorylating protein substrates, only a few of which are currently known (Healy et al., 2001; Nakagami et al., 1999; Sorrel et al., 1999). One of the best characterised substrates of the CDKs is the retinoblastoma (Rb) protein. The phosphorylation of Rb by the CDKA/cyclinD results in the release of E2F (adenosine E2 promoter binding factor)–DP heterodimeric transcription factor from its binding with Rb, leading to the activation of a wide range of genes involved in G1 to S phase progression (Dyson, 1998; Gutiérrez et al., 2002; Ren et al., 2002). In contrast when E2F–DP activity is no longer required, phosphorylation by CDK2/cyclinA inactivates E2F (Krek et al., 1994; Marti et al., 1999). Thus, the regulated phosphorylation of E2F–DP and Rb by CDKs plays a crucial role in G1 to S phase progression (Harbour and Dean, 2000; Shen, 2002).

In order to analyse the changes taking place in cell cycle machinery at distinct stages of dormancy, we first established the conditions to differentiate between eco- and endodormant stages in hybrid aspen by using short days (SDs) to induce dormancy. Secondly, we cloned the cDNAs for two hybrid aspen CDKs, poplar CDKA (PttCDKA) and PttCDKB, and analysed their expression patterns in the dividing and non-dividing cells in woody tissues of hybrid aspen stem. Thirdly, we investigated the regulation of hybrid aspen CDKs during distinct stages of dormancy and the phosphorylation of Rb and E2F by A-type CDK during dormancy, given their role in G1 to S phase progression. Our results show that growth cessation and establishment of ecodormancy occur at the end of week 4 of SD treatment. Transition into ecodormancy is characterised by the decline of cambial mitotic activity and a major reduction in H1 kinase activity of PttCDKA and PttCDKB kinase complexes. Once the arrest of cambial cell division occurs, establishment of the endodormant state is preceded by a transient increase in the CDK-activity-phosphorylating E2F transcription factor, which decreases its DNA-binding affinity. Finally, loss of Rb phosphorylation activity, reduction of the CDK protein levels and a decrease of PttCDKB transcription occur once the endodormant state is established. These results indicate that eco- and endodormancy are characterised by distinctive changes in the regulation of key cell cycle components such as the CDKs reported here.

Results

Two distinct phases of SD-induced growth cessation in hybrid aspen

Hybrid aspen plants were exposed to SDs in a climate chamber under controlled conditions as described in Experimental procedures. Under these growth conditions, plants underwent growth cessation and set buds by the week 4 of SD treatment (Table 1). In the same set of plants, the termination of cambial cell division was analysed by measuring the alterations in the mitotic index of cambial cells at distinct time points (Figure 2). The number of dividing fusiform cambial cells declined sharply between weeks 2 and 4 and reached a constant low level by the week 4 of SD treatment. Thus, there is a good temporal correlation between the timing of growth cessation in the apical bud and the vascular cambium.

Table 1.  Summary of growth characteristics after SD treatment
Weeks in SDsBud setBud break after 3 weeks in long daysGrowth after 3 months in long days (cm)
  1. For each time point, three hybrid aspen plants were studied. At indicated time, three hybrid aspen plants were moved from growth chamber with SD conditions (8 h day, 16 h night) and placed in a long day (16 h day, 8 h night) growth chamber. Bud flush after 3 weeks and growth (increase in length) after a period of 3 months in long days is reported.

0Non.d.46.8 ± 5.2
2Non.d.43.1 ± 3.5
4YesYes20.1 ± 3.6
6YesNo0
8YesNo0
10YesNo0
12YesNo0
Figure 2.

Mitotic activity of fusiform cambial cells following SD treatment.

Cambial mitotic activity was measured by counting the number of fusiform cells in mitotic phase expressed as percentage of total fusiform cells counted (Y-axis) and plotted against time in weeks (X-axis), indicating the sample processed, 0 indicating tree sample immediately following shift to SDs.

To distinguish between eco- and endodormancy under our growth conditions, plants were removed at the indicated time points following SD treatment and grown under long days. As shown in Table 1, after 0 and 2 weeks of SD treatment, there was no bud set, and plants shifted to long days continued to grow as before. In contrast, after 4 weeks of SD treatment, plants had set buds, and when these plants were shifted back to long days, two plants out of three displayed bud burst after 3 weeks in long days, whereas a third plant displayed bud burst after 5 weeks. After 3 months in long days, these plants had undergone elongation growth of almost 150–200 mm following bud burst. However, plants that had set buds following 6, 8 or 10 weeks of SD treatment were unable to undergo bud burst and resume elongation growth even after 3 months of long-day treatment. In summary, at the end of the week 4 of SD treatment, buds are in ecodormant state, whereas after 6 weeks, plants are in endodormant state.

Cloning and sequence analysis of the two hybrid aspen CDK homologues

We cloned the cDNAs for two CDK homologues from hybrid aspen. Their deduced amino acid sequences display high homology to other CDKs with sequence identity of up to 96% (results not shown). One of the hybrid aspen cDNA clones was found to encode a PSTAIRE-type CDK and was named PttCDKA whereas the other CDK was found to contain the PPTTLRE motif, indicating a B-type CDK, and thus was named PttCDKB. The deduced PttCDKA sequence was 294 amino acids (AA) in length, and exhibited most of the features conserved in A-type CDKs (Joubès et al., 1999), such as the threonine (Thr)-14 and tyrosine (Tyr)-15 phosphorylation sites, the PSTAIRE cyclin-binding domain (AA 45–51), the T-loop area (AA 148–173) and the SUC/CKS (suppressor of CDC2/cyclin dependent-kinase regulatory subunit)-binding motif (AA 207–244). The deduced PttCDKB protein was 306 AA long, again exhibiting the Thr-14 and Tyr-15 phosphorylation sites and harbouring the PPTTLRE motif characteristic of the CDKB2 protein family (Joubes et al., 2000; Figure 3). Likewise, alignment of both PttCDKA and PttCDKB protein sequences shows that the PttCDKA Phe (phenylalanine)-147 residue, characteristic of eukaryotic serine–Thr and Tyr protein kinases, is replaced by leucine (Leu)-159 in PttCDKB.

Figure 3.

Comparison of deduced PttCDKA (AAK16652) and PttCDKB (AAP73784) protein sequences.

Thick bars indicate regions of known function.

Differential expression patterns of PttCDKA and PttCDKB in the wood-forming tissues of hybrid aspen

We performed a high-resolution expression analysis of PttCDK genes in the cell layers representing the successive stages of wood formation in hybrid aspen. Distinct layers of cells undergoing division, expansion and secondary wall formation were isolated from this tissue by cryosectioning, and the expression of PttCDKA and PttCDKB was analysed using a technique described earlier by Moyle et al. (2002) and Uggla et al. (1996). This analysis allowed a direct assessment of the correlation between PttCDK gene expression and cell division in the wood-forming tissues. The results shown in Figure 4 indicate differential expression of the two PttCDK genes. While PttCDKA is expressed ubiquitously, PttCDKB expression is confined to dividing cells (young phloem, cambium and adjacent xylem cells), as previously seen with poplar cyclinH expression (Yamaguchi et al., 2000). These results are in agreement with previous studies in other plant species, showing a differential expression pattern for A- and B-type CDKs (Fobert et al., 1996; Hemerly et al., 1993; Joubès et al., 1999; Martínez et al., 1992), which indicate that A-type CDK genes are expressed in cells not only undergoing division but also that are competent to divide (Hemerly et al., 1993), whereas the expression of the B-type CDK genes is restricted to dividing cells.

Figure 4.

PttCDKA and PttCDKB genes are differentially expressed in cambial tissues.

Denatured cDNA samples prepared from mRNA extracted from the indicated tissue sections (see Experimental procedures) were spotted on nylon membrane, and PttCDKA and PttCDKB expression was detected by hybridising with radiolabelled probes prepared from ESTs corresponding to PttCDKA, PttCDKB and ubiquitin (as a control). Schematic representation of the cambial region is shown above. CO, cortex, between cork and the first fibre bundles; PF, phloem fibres; PH, functional phloem; CZ, cambial zone with differentiating phloem, DX, dividing xylem mother cells; EX, expanding xylem; and SW, zone of visible secondary wall formation.

The PttCDK genes are differentially regulated during dormancy

We investigated the transcriptional regulation of PttCDKA and PttCDKB genes in stem tissues every week following exposure to SDs during the induction of dormancy. PttCDKA and PttCDKB are differentially regulated during dormancy induction (Figure 5). While there is little alteration in the PttCDKA transcript levels throughout the entire experimental period, the PttCDKB transcript levels decline after 6–7 weeks of SD treatment and then disappear. This result indicates that, unlike PttCDKA, PttCDKB is regulated at transcriptional level, and this transcriptional downregulation of PttCDKB appears to correlate with the establishment of endodormancy. It is important to note that the reduction of neither the PttCDKA nor PttCDKB transcript levels occurs when the decline in cell division is observed (from week 2 onwards). In fact, the PttCDKA and PttCDKB transcript levels persist even after the cell division has ceased. Thus, the SD signals inducing a decline in cell division during transition to ecodormancy do not appear to act via downregulation of the PttCDKA or PttCDKB transcript levels, suggesting post-transcriptional regulation of PttCDKs during this process.

Figure 5.

Expression of PttCDKA and PttCDKB in stem tissues following SD treatment.

Total RNA extracted from internode segments 11–14 of hybrid aspen plants following SD treatment was used to investigate the expression of PttCDKA and PttCDKB. Total RNA was reverse transcribed and subjected to semiquantitative RT-PCR analysis using 18S rRNA as internal control. PCRs were performed in triplicates, signals were quantified and averaged results are shown on the graph. Bars represent standard deviation.

Regulation of protein levels and kinase activity of PttCDKA and PttCDKB during dormancy transitions

To further investigate this possible post-transcriptional regulation of PttCDKA and PttCDKB during transition to dormancy, we performed Western blot analysis using specific antibodies to follow the dynamics of the PttCDK proteins levels. The PttCDKA and PttCDKB protein levels decline around week 6 of the SD treatment and are nearly undetectable by week 9 (Figure 6a). This rules out the possibility that the decline in cell division following SD treatment between weeks 2 and 4 is because of a reduction in the PttCDKA or PttCDKB protein levels. As the PttCDKA protein levels decline whereas the corresponding transcript levels remain unchanged (Figure 5), either the translation of PttCDKA mRNA is inhibited and/or the PttCDKA protein is degraded during dormancy. Such a translational or post-translational control of an A-type CDK gene has not been previously reported. In contrast to PttCDKA, the decline in the PttCDKB protein levels nearly follows that of its mRNA levels. Importantly, for both the PttCDKs analysed here, the decline and loss of the protein levels coincide temporally with the establishment of endodormancy.

Figure 6.

Detection of the PttCDKA and PttCDKB protein levels and histone H1 kinase activity of PttCDKA and PttCDKB complexes following SD treatment.

(a) Total proteins extracted from 11th−14th internode of hybrid aspen trees grown in SD conditions were fractionated by SDS–PAGE and immunoblotted with PttCDKA- and PttCDKB-specific antibodies. Sampling time (in weeks) is shown at the top of the figure, ‘0’ indicating protein sample prepared from trees immediately after shifting to SD conditions.

(b) Immunoprecipitated PttCDKA (broken line) and PttCDKB (continuous line) complexes were assayed for phosphorylation of histone H1. Kinase activity is expressed as arbitrary units corresponding to signal strength of phosphorylated histone H1 bands quantified using a phosphorimager and plotted against sampling time in weeks.

The results above suggest that termination of cell division during transition to dormancy does not involve the reduction of the PttCDKA and PttCDKB transcript or protein levels. Therefore, we measured the histone H1 kinase activity of immunoprecipitated PttCDKA and PttCDKB kinase complexes during transition to dormancy. For these experiments, the PttCDKA and PttCDKB kinase complexes were immunoprecipitated from protein extracts for those time points in which no reduction in the protein level was observed (Figure 6a). As shown in Figure 6(b), the H1 kinase activities of both the CDKs decline nearly simultaneously, and the timing of the reduction of H1 kinase activity is in good agreement with the decline in the amount of dividing cambial cells marking the transition to dormancy (Figure 2).

Arabidopsis E2Fa (AtE2Fa) DNA binding is inhibited by CDK-mediated phosphorylation

As plant E2Fs contain a number of potential CDK phosphorylation sites (Shen, 2002), we investigated if they can be phosphorylated by CDKs and examined the consequence of this phosphorylation on E2F's ability to bind DNA. As E2F proteins from different plant species are highly conserved (similarity above 80%), we used AtE2Fa for our protein studies in the absence of poplar E2F cDNA. AtE2Fa was chosen because it plays a crucial role in cell division and meristem activity in plants (De Veylder et al., 2002). The cDNAs for AtE2Fa and poplar DP1 (PttDP1) genes were cloned and the fusion proteins maltose-binding protein (MBP)–AtE2Fa and glutathione S-transferase (GST)–PttDP1 were produced. The purified proteins were then tested as substrates for kinase assays using immunoprecipitated CDK complexes. Both AtE2Fa and PttDP1 could be phosphorylated by CDKA kinases (Figure 7a), whereas B-type CDK complexes phosphorylated the AtE2Fa protein very weakly and the PttDP1 protein not at all (data not shown). As expected, AtE2Fa and PttDP1 were able to form heterodimers and bind the E2F DNA-binding consensus sequence (Figure 7b) in a specific manner as wild type (wt) consensus sequence, but not the mutated version of the oligonucleotide, effectively competed for the binding (Figure 7c).

Figure 7.

PttDP1 and AtE2Fa proteins are phosphorylated by CDKA kinases in vitro.

(a) GST–PttDP1 and MBP–AtE2Fa proteins were used as substrates for immunoprecipitated Medsa;CDKA;1 (MsCDKA) and PttCDKA kinases. In the control reactions pre-immune serums were used to immunoprecipitate the kinases. Arrows point at the expected size of the proteins, and additional bands in the case of PttDP1 may represent degradation products.

(b) EMSA of the DNA binding of AtE2Fa–PttDP1 heterodimer. Neither AtE2Fa nor PttDP1 alone produced a detectable signal on the retardation gel. On the lane marked with an asterisk, the free probe was loaded.

(c) DNA binding by AtE2Fa–PttDP1 heterodimer is specific for the E2F-binding site. Radioactive signal could be abolished with an excess of unlabelled wt oligonucleotide probe but not with an excess of mutated oligonucleotide in the binding reaction.

(d) Phosphorylation of AtE2Fa decreases its DNA-binding activity. Fifty nanograms of MBP–AtE2Fa was first phosphorylated with immunoprecipitated MsCDKA kinase in the presence of increasing amounts of ATP, and then incubated with 50 ng of GST–PttDP1 to allow heterodimerisation. 32P-labelled oligonucleotide probe harbouring E2F-binding site was added, and binding reactions were fractionated on native PAGE. The gel was dried and signals were detected by autoradiography.

In pull-down experiments, we did not detect any effect of the E2Fa and DP1 phosphorylation on the heterodimer formation (data not shown). However, phosphorylation of E2Fa by CDKA significantly reduced the AtE2Fa–PttDP1 complex ability to bind DNA (Figure 7d). Unlike AtE2Fa, phosphorylation of DP1 did not result in reduced DNA binding (data not shown).

Transition to endodormancy coincides with a transient peak of E2F phosphorylation

As CDK phosphorylation of E2F has a negative effect on its DNA-binding activity, we investigated whether phosphorylation of E2F is also part of the regulatory mechanism involved in termination of cell division during dormancy. PttCDKA kinase complex was immunoprecipitated from protein extracts prepared from hybrid aspen stem samples during induction of dormancy, and the phosphorylation of E2Fa was investigated. A peak of E2Fa phosphorylation is observed at weeks 4 and 5, preceding the transition from eco- to endodormancy (Figure 8). The peak of E2F phosphorylation is observable when H1 kinase activity of both PttCDKs has reached a low level (Figure 6b, week 4) and when termination of cell division has occurred (Figure 2, week 4).

Figure 8.

Alteration of E2F and retinoblastoma protein phosphorylation during transition to dormancy.

PttCDKA was immunoprecipitated from protein extracts prepared from internodes 11–14 at the indicated time points and kinase activity of the immunocomplexes were assayed with either (a) MBP–AtE2Fa or (b) GST–hRb proteins as substrate. Reaction mixtures were resolved on SDS–PA gel, and the intensity of radiolabelled bands was quantified by phosphorimager scanning. Results from two independent measurements are shown for AtE2Fa phosphorylation.

Decline of Rb phosphorylation occurs upon transition to endodormancy

As the Rb protein has a crucial role in the regulation of E2F/DP1 transcription factor activity, we also investigated Rb phosphorylation by PttCDKA during the transition to dormancy. We used the human Rb (hRb) protein fused to GST, which was shown to be an in vitro substrate for plant CDKA complexes (Nakagami et al., 2002). Immunoprecipitation–protein kinase assays detected a decline of Rb kinase activity upon transition to endodormancy (Figure 8).

Discussion

Temporary suspension of growth and induction of dormancy in the meristem occurs during the normal course of plant development (Shimizu-Sato and Mori, 2001). Earlier work on dormancy focused on the differences between active and dormant tissues or described changes taking place in the dormant tissues following exposure to dormancy-breaking signals (Devitt and Stafstrom, 1995; Shimizu and Mori, 1998; and reviewed by Shimizu-Sato and Mori, 2001). However, far less is known about the establishment of dormancy, which is a stepwise process. In the transition from eco- to endodormancy, the cell division machinery loses its ability to respond to growth-promoting signals, and this change is poorly understood at the molecular level (Champagnat, 1983; Little and Bonga, 1974; and reviewed by Dennis, 1994; Figure 1). We established growth conditions to temporally separate eco- and endodormancy stages, which occurred after 4 and 6 weeks of SD treatment, respectively, in hybrid aspen, and analysed the regulation of CDKs to understand the differential stage-specific response of the cell division machinery during these stages.

For our analysis described here, we investigated two distinct CDKs from hybrid aspen, PttCDKA and PttCDKB. During the first 4 weeks of SD treatment, the decline in cambial cell division coincided with a reduction in the H1 kinase activity. This decline could be a result of reduced levels of CDK transcript and/or proteins, given that a good correlation between CDK transcription and activity has been described in other experimental systems (Hemerly et al., 1993; John et al., 1990; Setter and Flannigan, 2001). However, the transcript levels of PttCDKA and PttCDKB remain unaffected during that time. Thus, post-transcriptional mechanisms, e.g. inhibition of the translation of PttCDK mRNA and/or degradation of these proteins, may explain the observed decline of the H1 kinase activity of the two PttCDKs and the termination of cell division during this period. Nevertheless, the PttCDK protein levels remain also unchanged. Therefore, an alternative regulatory mechanism is probably involved, such as increased levels of CDK inhibitors or enhanced inhibitory phosphorylation of PttCDKs (Liu et al., 2000; Sun et al., 1999; Wang et al., 1998). It has been shown in Arabidopsis that ABA induces the expression of ICK1 (cyclin dependent-kinase inhibitor), a CDK inhibitor (Wang et al., 1998). Given the role of ABA in dormancy-related processes (Powell, 1987), hybrid aspen CDK inhibitor levels may be enhanced during the establishment of cambial dormancy. Alternatively, a reduction of the cyclins either at the transcriptional or post-transcriptional level would lower the amount of active cyclin–CDK complexes. This hypothesis is supported by the observation that SDs can reduce the transcription from an Arabidopsis B-type cyclin promoter in poplar buds (Rohde et al., 1997).

Cell division terminates after 4 weeks of SD treatment, but neither the PttCDKA nor PttCDKB transcript levels decline during this period. The maintenance of the PttCDKA transcript levels during dormancy is not surprising, as the expression of A-type CDKs is often coupled to competence to divide, which is not altered during dormancy in case of cambial cells (Hemerly et al., 1993). However, the continued expression of PttCDKB even after the termination of cell division is intriguing (Figure 5). Normally, the expression of B-type CDKs is under strict cell cycle control being linked to G2 to M phase and is downregulated, once cell division terminates (Fobert et al., 1996; Magyar et al., 1997; Umeda et al., 1999) or when cells exit the division mode as observed for PttCDKB in the cambial sections (Figure 4). Whether the presence of PttCDKB mRNA after termination of cambial cell division indicates differential regulation of PttCDKB and other cell cycle genes in the meristematic cells remains to be seen. If this is the case, switching off the transcription of certain cell cycle genes in meristematic cells may not be coupled to simply cessation of cell division. Instead, a more profound change, e.g. transition to endodormant state, may be necessary for the repression of the transcription of cell cycle genes such as PttCDKB. At this stage, the existence of such an alternative cell cycle regulatory mechanism in meristematic cells during dormancy remains a speculation and needs further investigation. Nevertheless, the PttCDKB transcript levels can thus be used as a molecular marker suitable for discriminating eco- and endodormant stages of cambial dormancy.

During dormancy, the cambial meristem cells are arrested in the G1 phase once cell division is terminated (Mellerowicz et al., 1989, 1992; Zhong et al., 1995). Thus, when the cell division machinery is rendered insensitive to growth-promotive signals during endodormancy stage, G1 to S phase transition is probably a major point of regulation. As the E2F/Rb pathway is crucial in G1 to S phase transition (Shen, 2002; Trimarchi and Lees, 2002), we examined the alteration in phosphorylation of E2F and Rb by PttCDKA upon SD treatment. Importantly, the transition to endodormancy correlates with a decline in Rb phosphorylation by PttCDKA. The potential consequence of this decline would be a reduction in the release of the E2F–DP complex by Rb for activating the S phase genes at this stage. Notably, the E2F kinase activity of PttCDKA increases dramatically during weeks 4 and 5 following SD treatment. While it is not possible here to conclude whether PttCDKA-dependent phosphorylation targets E2F for degradation, other data certainly support this possibility in poplar. In mammalian systems, E2F phosphorylation by CDKs is a prerequisite for E2F inactivation and subsequent degradation by the ubiquitin–proteasome pathway (Ohta and Xiong, 2001). Similarly, in plants, AtE2Fc can be phosphorylated by AtCDKA, and only the phosphorylated form of AtE2Fc binds AtSKP2, a component of the SCF-type E3 ubiquitin ligases (del Pozo et al., 2002). Our data show that AtE2Fa phosphorylation by CDKA results in a decreased ability of this transcription factor to bind its target DNA sequence (Figure 7). This loss of DNA-binding activity points to yet another mechanism negatively regulating E2F activity. Given that E2F transcription factors are major integrators of the signals for proceeding into a new round of DNA replication (De Veylder et al., 2002; Lukas et al., 1996; Wu et al., 1996), the timing of elevated E2F phosphorylation preceding the establishment of endodormancy is intriguing. Increased E2F phosphorylation would stimulate its inactivation and/or degradation and play a role in terminating cell division, inhibiting the transition from G1 to S phase and repressing the resumption of cell division as the transition to the endodormant stage occurs. E2F is phosphorylated before cell division terminates, showing a peak around weeks 4–6 of SD treatment. However, E2F might be differentially phosphorylated during the course of dormancy induction, possibly at multiple sites. It is unknown whether the two phosphorylation events (during the first 4 weeks of SD treatment and later during weeks 4–7) differentially affect E2F stability and activity. Increased E2F phosphorylation by PttCDKA kinase complex occurring subsequently to the reduction of its H1 kinase activity suggests that the CDK complexes phosphorylating E2F must have distinct cyclin partners that are either induced or stabilised during the period when a peak of E2F phosphorylation is observed.

The differential response of cell division machinery to growth-promotive signals during eco- and endodormancy has never been described in molecular terms. Our analysis presented here distinguishes both states in terms of the changes in the regulation of key cell cycle components. Our results define three major sets of molecular events following SD treatment that can be roughly correlated with the different stages of dormancy. These are: (i) reduction of H1 kinase activity of PttCDKA and PttCDKB that coincides with the decline and termination of cell division; (ii) transient increase in E2F phosphorylation activity preceding the transition to endodormancy; and (iii) reduction of Rb phosphorylation, and a decline and loss of the PttCDKA and PttCDKB proteins upon endodormancy. Based on the molecular and physiological data presented here, a model can be proposed to distinguish growth cessation and ecodormant state from the subsequent endodormant state. During the ecodormant stage, the termination of cell division occurs without the reduction in either the transcription or translation of the PttCDKs. Thus, the key cell cycle components CDKs are retained during ecodormant state, suggesting a potential mechanism for maintenance of the cell division machinery in a standby mode that is able to respond to growth-promotive signals. Cell division arrest during ecodormancy thus appears to resemble the cell division arrest that occurs upon exposure of the plants to stress, for which a decline of cell division takes place without the reduction of the CDK levels, and cell division can resume merely upon transfer to favourable conditions (Engelen-Eigles et al., 2000; Schuppler et al., 1998; Swiatek et al., 2002). In contrast, the transition to endodormancy involves a hitherto undescribed molecular switch acting via degradation of PttCDK proteins, leading to a block in G1 to S phase transition by its action on E2F and Rb. Thus, the transition to endodormancy involves loss of key cell division components and consequently, the cell division machinery enters offline mode, becoming insensitive to growth-promotive signals (Little and Bonga, 1974). This provides a tighter regulation of cell division, rendering its resumption unlikely, unless the process is reversed by dormancy-breaking signals. Hence, the changes in cell division machinery leading to ecodormancy are distinguishable from those that occur upon transition to endodormancy, and these findings are summarised as a model in Figure 9.

Figure 9.

Schematic representation of the changes in cell cycle components during distinct stages of dormancy.

Here the process of cambial dormancy was investigated, but the ability of plant meristems to terminate and resume growth has been studied in other experimental systems as well, e.g. in case of seed dormancy or axillary bud growth (Raz et al., 2001; Rohde et al., 1997; Shimizu-Sato and Mori, 2001). However, until recently, there was paucity of knowledge about the signals that induce cell division arrest and how this arrest can trigger other downstream developmental events. Earlier work on the control of embryonic cell divisions has indicated the role of FUS3 (Fusea 3) and LEC1 (Leafy colyledon 1) in termination of cell division during seed dormancy in Arabidopsis (Brocard-Gifford et al., 2003; Keith et al., 1994; Nambara et al., 2000; West et al., 1994). Similarly, genes such as ABA1 (abscisic acid deficient 1) and ABI3 (abscisic acid insensitive 3) are required to prevent premature resumption of growth in the same system (Raz et al., 2001). These examples demonstrate that separate genetic pathways may be involved in ensuring termination of cell division and maintaining the state of growth arrest. However, the targets of tree homologues of such genes and their involvement in cambial dormancy remain presently unknown. In conclusion establishing dormancy is a complex process involving cross-talk between multiple signalling pathways responding to diverse environmental cues. Having identified the potential targets of signals regulating eco- and endodormancy, the next step will be to identify the components of the signal transduction pathway, mediating the environmental and hormonal control of eco- and endodormancy.

Experimental procedures

Cloning and sequencing of PttCDK2A, PttCDK2B, PttDP1

Expressed sequence tags (ESTs) corresponding to PttCDKA (AF194820), PttCDKB (AY307372) and PttDP1 (AY307373) were identified during the course of a large-scale EST-sequencing project of a cambial cDNA library of poplar (Populus tremula L.×Populus tremuloides; Sterky et al., 1998). Full-length clones were obtained for these ESTs and sequenced by the dideoxynucleotide chain termination method (Sanger et al., 1977) using the ABI PRISM system (Perkin Elmer, Warrington, USA).

Plant growth conditions

Tissue culture-grown hybrid aspen plants were transferred to soil and grown for 2 months at 25°C with 16 h of day and 8 h of night in greenhouse. At this time, healthy plants with similar height (approximately 70 cm) were transferred to climate chamber for SD treatment. SD treatment consisted of 8 h of day and 16 h of night with day temperature at 20°C and 15°C during the night for 7 weeks. After 7 weeks, the temperature was reduced to 15°C during the day and 10°C in the night until the end of the experimental period (week 11). Samples were collected from leaves, stem and apical dome every week, frozen in liquid nitrogen and stored at −80°C until the time of use. In order to assay for the ability of the plants to resume growth following the SD treatment, three plants for each time point (2, 4, 6 and 8 weeks of SD treatment) were shifted to long days and 25°C, and the ability to resume growth was measured by investigating bud flush at the end of 3 weeks after the shift to long days. The plants were kept in long days for a period of 3 months in total, and growth (length) was measured at the end of 3 months.

Analysis of cambial growth cessation

Stem segments containing the sandwiched cambium were fixed in FAE (5% formaldehyde, 5% acetic acid, 50% ethanol) for 48 h and stored in ethanol until needed. For measurement of dividing cells the cambial tissue was scrapped from the fixed stem for each of the time points indicated. The bark was peeled and scrapping was performed either on the bark side (in active cambium) or on the xylem side (in dormant cambium) under the dissecting microscope with a fine needle. Following the scrapping, the cells were placed in a drop of water on a gelatin-coated slide and dispersed by vibrating the needle in a drop. The slides were air dried on a slide warmer at about 40–50°C and stained in toluidine blue O, rinsed in water and either observed directly or air dried and mounted in Euparal. Number of fusiform cells in mitotic phase was counted and expressed as percentage of total number of fusiform cells.

RNA extraction and reverse transcription (RT)-PCR

Total RNA was extracted from plant tissue using RNeasy kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. For expression analysis of candidate genes in wood-forming tissue during dormancy transition, RNA was isolated from stem segments from internodes 11 to 14 for each indicated time point. Two micrograms of total RNA was reverse transcribed into cDNA using the first strand cDNA synthesis kit (Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, England). First strand cDNA was used as template for RT-PCR amplification, ensuring that the amount of amplified product remained in linear proportion to the initial template present in the reaction. For RT-PCR, QUANTUM 18S RNA internal standard kit (Ambion Inc., Austin, USA) was used according to the manufacturer's instructions to provide internal control. Gene-specific primers used for RT-PCR were 5′-ggattcttctcctgaatttgctaaggatccacgcc and 5′-gaaatggcatttatggtacaaaaccaatatcc (for PttCDKA), 5′-tgtcccacccaccactcttcgcgaagtctcc and 5 ‘gcccatccttgtccaaattagtaacagctga (for PttCDKB).

High-resolution analysis of PttCDKA and PttCDKB expression in the wood-forming tissues

In order to analyse the expression of the PttCDKA and PttCDKB in the wood-forming region, 30-µm-thick tangential sections were prepared from a 17.0 mm × 2.5 mm block from the cambial region of a hybrid aspen stem using a cryomicrotome. The sections represented the following developmental zones: cortex, phloem fibres, functional phloem, cambial zone, dividing xylem mother cells and expanding xylem and secondary walls. From each section, poly(A)+ mRNA was extracted using Dynabeads (Dynal, Oslo, Norway), reverse transcribed and PCR amplified as described before by Moyle et al. (2002), and approximately 500 ng of PCR product were dotted onto a Hybond-N nylon membrane (Amersham Pharmacia Biotech) and hybridised with 32P-labelled probes derived from PttCDKA and PttCDKB EST clones.

Production of GST–PttDP1 and MBP–AtE2Fa fusion proteins

Full-length cDNA for PttDP1 was PCR amplified using oligonucleotide primers DP6 (5′-ccggggatccgtatggtcgctggtggggcccacctggaag) and DP4 (5′-ggacgtcgacgccgatctgcttgtcatctcag), which introduced BamHI and SalI restriction sites (underlined) at the 5′ and 3′ ends of the amplified product, respectively. The PCR amplified fragments were digested with BamHI and SalI restriction enzymes and cloned into pGEX5X1 plasmid (Amersham Pharmacia Biotech) digested with the same set of enzymes to obtain the plasmid PttDP1–pGEX5X1. Several independent clones were sequenced to identify a plasmid clone with no sequence errors in the coding region. To amplify AtE2Fa cDNA, poly(A)+ mRNA was isolated from 8-day-old Arabidopsis seedlings using Dynabeads (Dynal) according to the manufacturer's instructions. The RNA was reverse transcribed using a first strand cDNA synthesis kit (Amersham Pharmacia Biotech). The resulting cDNA samples were used as templates for PCRs using oligonucleotide primers 5′-agccgaattcatgtccggtgtcgtacgatcttct and 5′-ctgggtcgactcatctcggggttgagtcaacagct, which add EcoRI and SalI restriction sites (underlined) to the AtE2Fa open-reading frame. The PCR product was digested with EcoRI/SalI and ligated into pMALC2SN vector (a kind gift from Dr Jonas Lidholm, Pharmacia AB, Sweden) previously digested with the same enzymes, creating AtE2Fa–pMALC2SN. Again several independent clones were sequenced to identify a clone without sequence errors in the AtE2Fa open-reading frame. To produce GST–PttDP1 and MBP–AtE2Fa fusion proteins, PttDP1–pGEX5X1, and AtE2Fa–pMALC2SN plasmids were introduced into Escherichia coli strain BL21 (Stratagene, La Jolla, USA). Hundred millilitres of E. coli cultures were grown at 28°C, and fusion protein production was induced by adding 2 mm IPTG (isopropyl-β-d-thiogalactoside) for 2 h. At the end of this period, the cells were harvested, re-suspended in lysis buffer (25 mm Tris–HCl, 100 mm NaCl, 2 mm EDTA, 1 mm DTT, 1 mm PMSF and 1 mm benzamidine, pH 7.8), sonicated and centrifuged for 10 min at 13 000 g at 4°C. GST–PttDP1 and MBP–AtE2Fa fusion proteins were then purified from the supernatants by affinity chromatography on glutathione-Sepharose (Amersham Pharmacia Biotech) and amylose-Sepharose (New England Biolabs, Beverly, USA) columns, respectively, according to the instructions provided by the manufacturers. Yield and purity of the eluted fusion proteins was estimated by SDS–polyacrylamide gel electrophoresis (PAGE) analysis. The pGEX-Rb plasmid for the expression of GST–hRb (AA 379–928 of the hRb protein fused to GST) was obtained from W. Kaelin and purified as described by Kaelin et al. (1992).

Protein extraction and kinase assay

Proteins were extracted from 3-day-old culture of Medicago sativa cells or poplar stem from the dormancy samples by grinding the samples in liquid nitrogen and adding extraction buffer (20 mm Tris–HCl, pH 7.8, 15 mm MgCl2, 15 mm EGTA, 20 mm NaF, 0.1 mm Na3VO4, 2 mm DTT, 0.2% Nonidet P40, 10% glycerol and 1X protease inhibitor cocktail (Boehringer Mannheim, Germany). After incubation for 10 min at 4°C, samples were cleared by centrifugation (10 min at 13 000 g), and protein concentrations was determined by Bradford (1976). Equivalent amounts of proteins (150 µg) were incubated with antibodies raised against the indicated CDKs for 2 h at 4°C, then 40 µl of 50% (v/v) slurry of protein A-Sepharose was added and incubated for an additional 1 h at 4°C. Beads were washed three times with wash buffer (150 mm NaCl, 25 mm Tris–HCl, pH 7.8, 0.1% Triton X-100 and 1 mm DTT) and two times with kinase buffer (25 mm Tris–HCl, pH 7.8, 10 mm MgCl2 and 1 mm DTT). Kinase reactions were performed by adding 15 µg of histone H1 or 3 µg of MBP–E2Fa or GST–hRb proteins to the beads in a final volume of 30 µl of kinase buffer in the presence of 5 µCi [γ-32P]ATP and 10 µm unlabelled ATP (hot reactions) or in the presence of 0, 25 or 50 µm unlabelled ATP (cold reactions). Kinase reactions were allowed to proceed for 30 min at room temperature, and then reaction mixtures were fractionated by SDS–PAGE. Phosphorylated substrate proteins were either detected in a phosphorimager and quantified with imagequant software (Amersham Biosciences, Uppsala, Sweden), or the gel was dried and the phosphorylated proteins were detected by autoradiography (hot reactions). Alternatively, phosphorylated proteins were used for DNA-binding experiments (cold reactions).

Electrophoretic mobility shift assays (EMSAs)

Oligonucleotide probes, 200 pmol containing either a consensus binding site for E2F (wt, 5′-atttaagtttcgcgccctttctcaa) or a mutated version of the E2F-binding site (mutant (mut) 5′-atttaagtttcgtaccctttctcaa) were annealed in TE (Tris-EDTA) + 50 mm NaCl. The double-stranded wt probe was 5′ end labelled with [γ-32P]ATP and T4PNK (T4 polynucleotide-kinase). In each binding reaction, 50 000 c.p.m. of labelled probe was used. The DNA-binding reactions were conducted by incubating 20–100 ng of recombinant purified PttDP1 and AtE2Fa proteins for 20 min at 4°C, adding the oligonucleotide and incubating for additional 20 min at room temperature. Reactions contained 25 mm HEPES–KOH with pH 7.5, 100 mm KCl, 1 mm MgCl2, 1 mm EDTA, 10% glycerol, 1 mm DTT and 0.5 µg of poly dI-dC (Amersham Pharmacia Biotech). Samples were fractionated by electrophoresis on 5% polyacrylamide (PA) gels in 0.5× TBE (Tris-borate-EDTA) at 4°C. For the competition experiments, 30–100-fold cold oligonucleotide (wt or mut) was added to the binding reaction mixture.

Antibodies and Western blotting

Polyclonal antibody against the C-terminal peptide of alfalfa Medsa;CDKA;1 (EYFKDIKFVP) was generated and affinity purified as described by Magyar et al. (1997). This polypeptide sequence is identical to that of PttCDKA C-terminal sequence; thus, this antibody was used for the immunodetection and immunoprecipitation of PttCDKA. For immunodetection and immunoprecipitation of PttCDKB, a polyclonal antibody raised against the C-terminal CLQYDPSKRISA peptide of the PttCDKB protein was used after affinity purification. MBP–AtE2Fa protein was detected using a rabbit polyclonal anti-MBP antibody (New England Biolabs). To determine the changes in the PttCDKA and PttCDKB protein levels, 50 µg of total protein was fractionated on SDS–PA gel and transferred to nitrocellulose membrane as described by Magyar et al. (1997). PttCDKA and PttCDKB polypeptides were detected by incubating the membrane first with affinity purified anti-Medsa;CDKA;1 and anti-PttCDKB antibodies at a dilution of 1 : 2000, and then with a peroxidase-conjugated anti rabbit IgG (Vector Laboratory, USA) diluted to 1 : 15 000. Signals were detected with enhanced chemiluminescence reagents (Amersham Pharmacia Biotech).

Acknowledgements

This work was supported by a joint grant of the Swedish and Hungarian Academy of Sciences, Vetenskapsrådet and the Kempe Foundation to Dr Rishikesh P. Bhalerao. Financial support from Energimyndigheten to Dr Rishikesh P. Bhalerao is acknowledged. A.E.-R. was recipient of a postdoctoral fellowship from Vetenskapsrådet and the spanish Ministerio de Educación, Cultura y Deporte.

Accession numbers: PttCDKA, AF194820; PttCDKB, AY307372; and PttDP1, AY307373.

Ancillary