SEARCH

SEARCH BY CITATION

Keywords:

  • cambium;
  • dormancy;
  • cold hardiness;
  • microarrays;
  • metabolic profiling

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

We have performed transcript and metabolite profiling of isolated cambial meristem cells of the model tree aspen during the course of their activity–dormancy cycle to better understand the environmental and hormonal regulation of this process in perennial plants. Considerable modulation of cambial transcriptome and metabolome occurs throughout the activity–dormancy cycle. However, in addition to transcription, post-transcriptional control is also an important regulatory mechanism as exemplified by the regulation of cell-cycle genes during the reactivation of cambial cell division in the spring. Genes related to cold hardiness display temporally distinct induction patterns in the autumn which could explain the step-wise development of cold hardiness. Factors other than low temperature regulate the induction of early cold hardiness-related genes whereas abscisic acid (ABA) could potentially regulate the induction of late cold hardiness-related genes in the autumn. Starch breakdown in the autumn appears to be regulated by the ‘short day’ signal and plays a key role in providing substrates for the production of energy, fatty acids and cryoprotectants. Catabolism of sucrose and fats provides energy during the early stages of reactivation in the spring, whereas the reducing equivalents are generated through activation of the pentose phosphate shunt. Modulation of gibberellin (GA) signaling and biosynthesis could play a key role in the regulation of cambial activity during the activity–dormancy cycle as suggested by the induction of PttRGA which encodes a negative regulator of growth in the autumn and that of a GA-20 oxidase, a key gibberellin biosynthesis gene during reactivation in spring. In summary, our data reveal the dynamics of transcriptional and metabolic networks and identify potential targets of environmental and hormonal signals in the regulation of the activity–dormancy cycle in cambial meristem.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Plant meristems frequently undergo transitions between active and dormant states. The ability of meristematic cells to terminate cell division and establish a dormant state plays a central role in processes as diverse as the protection of seeds (Bewley, 1997) and the control of the outgrowth of axillary buds (McSteen and Leyser, 2005). However, perennial plants of the boreal forest offer one of the most striking examples of the importance of activity–dormancy cycling for plant survival. Failure to establish the dormant state prior to winter, or precocious activation of growth in the spring, will severely compromise a boreal tree’s ability to survive (Aitken and Adams, 1997; Olsen et al., 1997). Thus, the capacity to synchronize the timing of active and dormant states with seasonal changes is of immense adaptive significance, since it underlies the ability of perennial trees to grow for hundreds of years under unpredictable and often hostile environmental conditions.

The transition between active and dormant states occurs gradually (Champagnat, 1983). The stages of the activity–dormancy cycle are defined by highly coordinated changes in cellular physiology, anatomy and metabolism that follow a strict temporal pattern in meristematic cells. The first stage in the transition to dormancy is characterized by cessation of cellular growth in the autumn. Simultaneously, the development of cold hardiness is initiated, as is a metabolic shift towards the accumulation of storage compounds (Farrar and Evert, 1997; Nelson and Dickson, 1981; Weiser, 1970). However, both these latter processes continue after cessation of growth has been completed. In the cambium, the vacuolar structure is altered as cessation of growth is completed and the cell walls thicken (Farrar and Evert, 1997). A developmental switch results in the shift from leaf development to bud scale formation in the shoot apex (Rohde et al., 2002) and subsequently, the plasmodesmata of meristem cells are blocked as dormancy is established (Rinne et al., 2001). Many of these changes are then reversed in the spring as cold hardiness is gradually lost, storage reserves are remobilized and cell division is reactivated.

Environmental signals play a central role in regulating the proper temporal execution of the various cellular responses during the activity–dormancy cycle. For example, reduction in day length induces cessation of growth in the autumn and initiates the induction of dormancy (Nitsch, 1957), whereas release from the dormant state requires prior exposure to chilling temperatures (Perry, 1971). Once the chilling requirement is met, warm temperatures in the spring can induce reactivation (Heide, 1993). Environmental signals are proposed to regulate activity–dormancy cycling via modulation of hormonal levels or through alteration of the sensitivity of the cells to hormones. For example, inhibition of elongation growth by the reduction in day length has been correlated with the lowering of gibberellin levels in the elongation zone below the apex (Juntilla and Jensen, 1988; Olsen et al., 1997). Similarly, the cambial meristem is rendered insensitive to the cell division-promoting effects of auxin as the establishment of dormancy is completed (Little and Bonga, 1974).

However, environmental and hormonal regulation of the activity–dormancy cycle in perennial plants remains poorly understood at the molecular level. Little attention has been paid to analysis of the modulation of transcriptional and metabolic networks during the transition to dormancy and even less is known about the reactivation process. The majority of the prior studies have either focused on analysis of a subset of genes (Espinosa-Ruiz et al., 2004; Rowland and Arora, 1997; Wisniewski et al., 2004) or compared gene expression between active and dormant stages (Schrader et al., 2004). As a result there is lack of suitable molecular markers that can be used to investigate how diverse signals regulate the cellular responses during the distinct stages of the activity–dormancy cycle. This has prompted us to perform transcript and metabolic profiling of isolated cambial meristem cells during the course of the activity–dormancy cycle in the model tree Populus tremula (aspen) growing under natural conditions. Our approach has allowed us to reveal the dynamic changes in the key transcriptional and metabolic networks during the distinct stages of the activity–dormancy cycle. These results provide insights regarding the molecular basis of the physiological and anatomical changes in meristem cells during the activity–dormancy cycle that has largely been lacking in previous studies (Espinosa-Ruiz et al., 2004; Schrader et al., 2004). Furthermore, using cambial meristem cells for analysis has allowed a much higher cellular resolution in defining the transcriptional and metabolic profiles in contrast to earlier studies. Much of the earlier work has used complex tissues consisting of several cell types, making it difficult to discern which of the studied changes occur in meristem cells (Renaut et al., 2004; Rowland and Arora, 1997; Sagisaka, 1974; Wisniewski et al., 2004). Thus our approach differs considerably from that of previous investigations and overcomes many of their inherent limitations and provides new information concerning the regulation of the activity–dormancy cycle at the molecular level.

Results and discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Temperature plays a role during cambial reactivation but not in growth cessation

We initially performed anatomical observations of the cambial zone to identify the timing of the two key events of the activity–dormancy cycle, induction and the cessation of cambial cell division. As seen in Figure 1, the first indications of cambial reactivation are evident in the 13 May sample where thin-walled cells display some cell expansion on the phloem side. In the 26 May sample, a dramatic increase in the width of the cambial zone reflects a fully activated cambium. By 20 August the cessation of cell division is apparent, as the cells again become thick-walled, much as they were in the 20 April sample. When this growth profile is mapped against the weather conditions over the same season, it appears that while the reactivation of the cambial cell division in the spring occurs after exposure to several days with minimal temperatures above zero degrees, cessation of cambial cell division in mid August occurs before the temperature becomes suboptimal for this process.

image

Figure 1.  Cross-section of aspen stem during the course of the activity–dormancy cycle. Aspen stems were sectioned and stained as described in Experimental procedures for all the time points for which microarray analysis was performed: (a) 20 April, (b) 13 May, (c) 26 May, (d) 9 June, (e) 7 July, (f) 16 August, (g) 11 September, (h) 5 October, (i) 13 December. The first indication of cambial reactivation can be observed by 13 May and active cambium is easily distinguishable by 26 May. Cambial activity appears to be terminated by 16 August with the cells becoming thick walled and being flanked by the late wood. Stars indicate the cambial cells.

Download figure to PowerPoint

Transcript and metabolic profiling reveals stage-specific dynamics of gene expression and metabolism during the activity–dormancy cycle

While the physiological and anatomical alterations associated with the distinct stages of the activity–dormancy cycle in poplar cambium follow a well-defined time-course (Farrar and Evert, 1997; Lachaud et al., 1999; Nelson and Dickson, 1981), the associated molecular changes remain largely unknown. We therefore performed transcript and metabolic profiling on cambial meristem cells at several time points during the course of the activity–dormancy cycle. We divided the transcript profiling data into two phases in order to capture the transcriptional changes specifically associated with reactivation and cessation of growth. In total, we identified 2266 and 1711 clones that are differentially regulated during the spring reactivation phase (lasting from 20 April to 7 July, phase 1), and autumn growth cessation and dormancy (9 June to 13 December, phase 2), respectively.The differentially expressed genes of each phase were clustered into 12 different patterns based on the kinetics of gene expression (Figure 2). During phase 1 (Figure 2a), two main groups of gene expression clusters can be coupled to (i) the early stage of cambial reactivation (clusters 1.2–1.4) and (ii) the peak of cambial activity (clusters 1.8–1.10). Similarly, during phase 2 (Figure 2b), gene expression clusters can be broadly divided into three main groups, two of which are associated with cessation of cambial activity (clusters 2.2, 2.3, 2.8–2.11 and clusters 2.5–2.7), while a third defines a set of genes whose induction is clearly separated from growth cessation (2.12). We then performed semi-quantitative RT-PCR on a subset of genes that were judged to be differentially expressed according to microarray data.The RT-PCR for these genes confirmed the results from microarray data, indicating a good robustness of microarrays in detecting differentially expressed genes in general. Our data on gene expression reported here have been derived from samples collected through a single year. However, we believe that the gene expression data (and metabolic profiling data) reported here are not simply representative for the year 2000 since we and others have previously reported that many of the genes display similar expression patterns to those reported here during the course of dormancy in poplar and other species (Arora et al., 1997; Schrader et al., 2003, 2004; Welling et al., 2002). In summary, considerable modulation of the cambial transcriptome occurs after growth cessation, perhaps associated with the later stages of development of cold hardiness and the establishment of dormancy.

image

Figure 2.  Cambial gene expression profiles during: (a) spring reactivation (phase 1) and (b) autumn growth cessation and dormancy (phase 2). Differentially expressed clones were clustered into 12 main classes each using TIGR MeV 4.0b (Saeed et al., 2003). The normalized expression profiles of individual clones (gray lines) and the mean expression profile of all transcripts belonging to the respective class (red lines) are shown. Values on the Y-axis are the expressed ratios in log2 scale. For each cluster the number of clones are shown in brackets and the description of the individual clones is summarized in Supplementary.

Download figure to PowerPoint

Metabolite profiling of cambial samples from April, June, August, September and December then allowed us to integrate these data with the transcript profiling results. Gas chromatography/MS analysis of cell extracts led to the detection of more than 1000 peaks, of which 227 were found to be significantly changed in one or pair-wise sample class comparisons using partial least-square discriminant analysis (PLS-DA) (Figure 3, Supplementary). The accumulation of these responsive metabolites follows distinct temporal patterns that can be associated with specific cellular processes occurring during the discrete stages of the activity–dormancy cycle. Some of the changes detected in the aspen cambial metabolome during autumn have also been observed during cold acclimation of Arabidopsis plants (Cook et al., 2004). However, it is worth noting that, in contrast to Arabidopsis where these changes in the metabolome are apparently triggered by exposure to low temperature, many of the changes in aspen cambial metabolome occur prior to the reduction in temperature in the autumn. This is consistent with the observation that, unlike Arabidopsis, short days can induce cold acclimation in trees independently of temperature reduction (Weiser, 1970). Nevertheless, the similarities in metabolic responses between cold-treated Arabidopsis and seasonal growth cessation in aspen suggest that it would be worthwhile comparing the underlying signaling pathways in these two species in the future.

image

Figure 3.  Seasonal changes in the cambial metabolite levels. Metabolites present at a higher level during a specific time point relative to others are highlighted. The metabolites are classified into five groups: carbohydrates, organic acids, amines (amines and amino acids) and sterols. Metabolites that do not belong to any of these groups are classified as ‘other’. Metabolites that could not be identified or classified according to their mass spectra are described as ‘unidentified’. All comparisons are based on the supervised multivariate statistical method PLS-DA (see Experimental procedures).

Download figure to PowerPoint

Regulation of cell-cycle genes during the activity–dormancy cycle involves transcriptional and post-transcriptional regulation

While activation and cessation of cell division are the two central events of the activity–dormancy cycle, the cell-cycle regulators targeted by the signals controlling the timing of these two events remain largely undefined. We therefore analyzed the expression patterns of the aspen homologs of the Arabidopsis core cell-cycle genes (Menges et al., 2005) during the course of the activity–dormancy cycle in aspen cambial cells. Of the 68 potential cell-cycle-related genes of aspen represented on the microarrays, 23 displayed significant changes in expression during reactivation and 21 did so during the cessation of cell division.The majority of these responsive cell-cycle-related genes displayed a highly coordinated expression pattern, falling essentially into only two major clusters for each phase (Figure 4).

image

Figure 4.  Expression of cell-cycle genes in cambial cells during the activity–dormancy cycle. Cell-cycle genes differentially expressed during (a) spring reactivation and (b) autumn growth cessation and dormancy were clustered using TIGR MeV 4.0b (Saeed et al., 2003). The normalized averaged expression profiles of individual clones (gray lines) and the mean expression profile of all transcripts belonging to the respective class (red lines) are shown. Values on the Y-axis are the expressed ratios in log2 scale. For each class the number of clones is indicated and the individual clones are described in Supplementary.

Download figure to PowerPoint

Interestingly, although anatomical analysis reveals a clear increase in the number of dividing cambial cells between the 20 April and 26 May tissue samples (compare Figures 1 and 4), this activity is not mirrored in increased levels of cell-cycle-related transcripts. This lack of correlation between the activation of cell division and the induction of cell-cycle gene expression would suggest a role for post-transcriptional control of cell-cycle genes during the early phase of cambial cell-cycle activation. This observation is in contrast to the situation in Arabidopsis, where the induction of the cell cycle is often correlated with the induction of cell-cycle gene expression (Menges et al., 2005). One possible hypothesis to explain our results is that the transcripts of key cell-cycle genes are being maintained at a low level after cessation of growth, with their translation being suppressed during the dormant phase. When signals for cambial reactivation are perceived in the spring, translation of these stored transcripts might be sufficient to initiate early cell divisions. This hypothesis is supported by our earlier results showing that while the polypeptide for the A-type cyclin-dependent protein kinase (CDK) is not detectable during dormancy, the corresponding transcripts are present at high levels (Espinosa-Ruiz et al., 2004). Similarly, experiments performed on dormant seeds indicate that translation of stored transcripts plays a key role during the release from dormancy, providing an interesting parallel with cambial reactivation (Nakabayashi et al., 2005; Rajjou et al., 2004). Furthermore, the expression of several other cell-cycle regulators, such as a subset of cyclins, E2F-like and DPB-like transcription factors, does not change appreciably through the course of the activity–dormancy cycle (Supplementary), again suggesting a pattern of post-transcriptional regulation of these gene products during the activity–dormancy cycle.

In contrast to the reactivation phase, the decline in the transcript levels of the core cell-cycle genes in the autumn correlates well with the cessation of cell division between July and August (see Figure 1). In a prior study we had hypothesized that the growth cessation signal could downregulate cyclin and/or induce the expression of cyclin-dependent kinase inhibitor (CKI) expression, although no cyclins or CKIs were specifically investigated (Espinosa-Ruiz et al., 2004). While the expression of CKIs did not change significantly in the present experiment, the expression of a subset of cyclins, e.g. D-type cyclins that are key regulators of the G1–S phase transition, is downregulated as cell division terminates (Figure 4). We cannot exclude the possibility that the growth cessation signal(s) also act via induction of CKIs since only five out of ten or more CKIs in the poplar genome (http://genome.jgi-psf.org/Poptr1/Poptr1.home.html) were analyzed here, but cyclins clearly do appear to be targets of that signal.

Acquisition of cold hardiness is a complex process involving the activation of several distinct mechanisms

In addition to growth cessation, appropriate acquisition of cold hardiness is critical for the survival of perennial plants in boreal forest. In spite of the importance of this process, our knowledge of the regulation of cold hardiness in trees at the molecular level is rudimentary. Firstly, the temporal overlap between the development of cold hardiness and dormancy makes it problematic to discriminate between these two processes (Arora et al., 1997; Weiser, 1970). Secondly, to date the molecular analysis of cold hardiness in trees is based on the investigation of a very limited set of genes, such as dehydrins, whose role in this process is far from clear (Arora et al., 1997; Welling et al., 2002). In particular, it is also unclear which of these putative cold hardiness-related genes are actually expressed in the meristem cells. Therefore, to obtain a better understanding of the regulation of cold hardiness in aspen cambial meristem, we retrieved from our microarray data set all those genes whose expression was significantly induced when cold hardiness develops in the autumn.

In Umeå, cold hardiness in aspen and related species starts developing from August onwards, peaking by late October (Sennerby-Forsse, 1987). Our data indicate that the transcript levels of dehydrins, cold-regulated proteins (CORs) and biosynthetic enzymes for cryoprotectants such as raffinose that have been implicated in cold tolerance in other species (Welling and Palva, 2006a) increase in aspen cambial tissue as cold hardiness develops in the autumn.Simultaneously, the transcript levels for enzymes involved in phospholipid biosynthesis and lipid desaturation are also induced. The induction of the expression of a number of genes involved in oxidative stress responses and enzymes of radical scavenging pathways indicates that plants activate mechanisms to protect cell membranes from the photo-oxidative damage associated with exposure to low temperature (Foyer et al., 1997; Kocsy et al., 2001). Although these patterns reflect considerable similarity between the development of cold hardiness and cold shock responses in terms of the type of genes induced, there are significant differences between the two processes. For instance, the proline biosynthesis pathway is not induced in aspen cambial tissue as cold hardiness is established whereas it is induced by cold shock in Arabidopsis (Benedict et al., 2006; Vogel et al., 2005; Wanner and Junttila, 1999).

Temporally distinct activation of gene expression programs underlies the stepwise development of cold hardiness in the autumn

The development of cold hardiness in aspen during the autumn occurs in a stepwise manner (Weiser, 1970; Welling and Palva, 2006a) but the underlying molecular mechanisms are not well understood. To address this we investigated the timing of induction of putative cold hardiness-related genes in the cambial cells during the autumn. We first compiled a list of genes potentially involved in cold hardiness from three data sets that included: (i) poplar genes induced by low temperature and AtCBF overexpression (Benedict et al., 2006; Renaut et al., 2004), (ii) poplar homologs of Arabidopsis genes induced by low temperatures (Vogel et al., 2005), and (iii) poplar homologs of genes involved in the development of cold hardiness in other tree species (Rowland and Arora, 1997; Wisniewski et al., 2004). We then clustered the expression patterns of these genes, based on their timing of induction. The clustering yielded six main classes (Figure 5), three associated with the autumn transition (classes I–III) and three with the spring (classes IV–VI). Interestingly, the expression of several genes, e.g. WCOR, ERD10 and dehydrins of classes I and II that have been shown to be induced in Arabidopsis and poplar by low temperature (Benedict et al., 2006; Fowler and Thomashow, 2002; Gilmour et al., 2004; Vogel et al., 2005) are induced in the aspen cambial cells before the plants experience any reduction in temperature in the autumn (Figure 5). This implies that a signal other than low temperature must trigger the induction of these genes in the autumn under natural conditions. The temporal overlap between this induction pattern and cessation of growth suggests that they may share a common regulatory signal such as short days. This suggestion is supported, for example, by the induction of dehydrins (Welling et al., 2002) and other genes by short days alone (L. Resman, R.P. Bhalerao, in preparation). Importantly, this observation provides a potential explanation for the molecular basis of low-temperature-independent induction of cold hardiness in perennial plants in the autumn (Weiser, 1970). The expression of class III genes is induced significantly later compared than that of classes I and II (Figure 5). Thus, if the short day signal is involved in the regulation of class III genes it could play an indirect role, serving perhaps to modulate their responsiveness to low temperature. This hypothesis is also supported by the observation in birch where prior exposure to short days enhances the inducibility of Lti6a, a dehydrin-like gene, by low temperature (Puhakainen et al., 2004). Thus, differential temporal induction of cold hardiness-related genes shown here provides a possible explanation for the stepwise development of autumnal cold hardiness that is independently regulated on the one hand by short days alone and on the other by short days acting in concert with low temperature during autumn (Weiser, 1970).

image

Figure 5.  Analysis of the expression of low-temperature-induced genes in cambial cells during the activity–dormancy cycle. The low-temperature-induced and cold hardiness-related genes (see Supplementary for a detailed description) were clustered using TIGR MeV 4.0b (Saeed et al., 2003) into three main patterns of expression during autumn growth cessation (I, II, III) and spring reactivation (IV, V, VI), respectively, in addition to several minor patterns (data not shown). The normalized averaged expression profiles of individual clones (gray lines) and the mean expression profile of all transcripts belonging to the respective class (red lines) are shown. Values on the Y-axis are the expressed ratios in log2 scale. For each class the number of clones is indicated. Additionally, the expression of PttDRTY and PttHB12 genes is highlighted.

Download figure to PowerPoint

Cold hardiness-related genes are superinduced during the early stage of reactivation

In contrast to the autumn transition, less is known about the molecular regulation of cold hardiness in the cambium and other meristems in the spring (Aitken and Adams, 1997; Sennerby-Forsse, 1987). Interestingly, we detected a superinduction of several cold-responsive genes (classes V and VI) during the early stage of reactivation in the cambial cells (Figure 5). This superinduction of cold-regulated genes in spring has not been described before and is surprising in view of earlier reports that a rapid loss of cold hardiness parallels the reactivation of the cambial meristem and bud burst in Salix which is closely related to aspen (Sennerby-Forsse, 1987). This raises questions concerning the role of these superinduced genes during early stages of reactivation in the cambium. In contrast to autumn, it appears unlikely that short days and/or low temperatures could be involved in the superinduction of these genes since both temperature and day length are increasing during this period.Thus, these genes appear to be responding to some other signal during cambial reactivation in the spring. While the exact role of this superinduction is also not clear, it might reflect a need to protect the very sensitive dividing cambial cells from sudden drops in temperature during early spring.

Transient increase in cambial ABA levels is correlated with the late stage of development of cold hardiness

As outlined above, short days and short days acting in conjunction with low temperature appear to regulate the stepwise development of cold hardiness in the autumn. While factors acting downstream of these two signals in regulating gene expression are not known, a role for ABA in this process has been proposed (Gusta et al., 2005; Welling et al., 2002). However, it is unclear whether ABA is involved in early and/or the late stage of acquisition of cold hardiness in the autumn. Our measurements (Table 1) indicate that the major increase in cambial ABA levels occurs after the induction of the class I and II genes. In contrast, a good correlation is observed between increases in ABA levels and the induction of class III genes in the cambial cells. This suggests that ABA might act downstream of the short day plus low temperature signal in inducing the late stage of cold hardiness. It is important to note, that we cannot rule out the possibility that ABA might also be involved in the early stage of the development of cold hardiness in the autumn. For instance, a short day signal could potentially act via modulation of the sensitivity of cells to ABA without affecting the cellular levels of the hormone. Interestingly, while ABA levels increased in the cambial cells late in the autumn, we did not notice a corresponding increase in the transcript levels of poplar genes similar to ABA1 (Qin and Zeevaart, 1999) and ABA2 (Cheng et al., 2002) that are represented on the arrays (data not shown).

Table 1.   Cambial abscissic acid (ABA) levels during the course of activity-dormancy cycle. Results show the ABA content in cambial sections from two independent trees.
Sampling dateABA level (pg)
May 13363
May 13225
May 26149
May 26127
June 09113
June 09111
July 07 94
July 07 93
August 16158
August 16100
September 11295
September 11312
December 13118
December 13198

In contrast to the autumn transition, little is known about the role of ABA during spring reactivation. Interestingly, high cambial ABA levels were observed in early spring at the 13 May time point which overlap with the superinduction of cold-responsive genes of classes IV and V and could potentially implicate ABA in their regulation. While ABA has received considerable attention as a potential regulator of cold-responsive genes in plants (Gusta et al., 2005; Welling and Palva, 2006b), it may not be the only signal involved in this process. Interestingly, the level of gamma-amino butyric acid (GABA), a signaling molecule previously implicated in plant stress responses (Bouche et al., 2003), increases in the cambial cells in the same time-frame as the induction of cold hardiness-related genes.Since GABA levels have also been shown to increase in Arabidopsis upon exposure to low temperature (Cook et al., 2004; Kaplan et al., 2004), it would be worthwhile to investigate more closely the role of GABA in the induction of cold hardiness-related genes in trees.

Transcriptional control of cold hardiness-related gene expression

Although the acquisition of cold hardiness involves considerable modulation of gene expression, the transcriptional regulators involved in this process remain largely unknown in trees. One exception to this is the recent finding that overexpression of Arabidopsis CBF1 in poplar improves cold hardiness and induces a specific subset of cold-responsive genes (Benedict et al., 2006). However, the poplar CBF homolog (PU12514) represented on the array used in the present study is not induced when cold hardiness develops in the autumn, and similar results have been obtained in birch (Welling and Palva, 2006a). Nevertheless, a role for CBFs in establishing cold hardiness cannot be ruled out since we did not examine the expression of other poplar CBFs in this experiment. It is also possible that we might have failed to detect changes in poplar CBF expression if its expression is upregulated only transiently (Benedict et al., 2006; Maruyama et al., 2004; Vogel et al., 2005).

Given the complexity of the development of cold hardiness, it is likely that multiple transcription factors could also be involved in regulating this process. One such gene is PttDRTY (PU03412) a trans-acting factor similar to the CBFs whose expression increases when cold hardiness is acquired in the autumn (Figure 5). It remains to be seen whether this gene is indeed involved in cold hardiness. Another potential candidate transcription factor that could be involved in cold hardiness is PttHB12 (PU09211) a poplar gene encoding a HD-ZIP type transcription factor that is also induced in the autumn (Figure 5). Given the role of ABA in cold hardiness, it is worth noting that both PttHB12 and its close relative in Arabidopsis, AtHB12, are induced by ABA (Olsson et al., 2004). Thus, one possibility is that PttHB12 could act downstream of ABA in regulating the expression of cold hardiness-related genes. Interestingly, PttHB12 and PttDRTY display differential timing of expression with PttDRTY being induced early whereas PttHB12 is induced late in the autumn (Figure 5). Furthermore, in contrast to PttDRTY, PttHB12 is superinduced during spring reactivation (Figure 5) suggesting that these two factors could regulate different sets of target genes. Finally, in addition to the genes mentioned above a host of other transcription factors are induced when cold hardiness develops and at least some of these could also be involved in regulating this process.

Breakdown of starch provides substrates for the generation of energy and cryoprotectants

Acquisition of cold hardiness requires considerable investment in terms of energy and carbon as new proteins and cryoprotectants need to be synthesized (Hurry et al., 2000; Stitt and Hurry, 2002). However, a major phase of development of cold hardiness in the cambium occurs when both energy and carbon becomes limiting as photosynthesis declines (Keskitalo et al., 2005). At this point plants must derive the energy and carbon required for the development of cold hardiness from other sources. Interestingly, the breakdown of starch coincides with the development of cold hardiness and the resulting carbohydrate pools could therefore potentially provide the necessary substrates for the generation of energy and metabolites. The observation that reducing starch breakdown in Arabidopsis via reduction of beta-amylase expression leads to reduced freezing tolerance is consistent with such a model (Kaplan and Guy, 2005). Starch breakdown has previously been proposed to be induced by low temperature in poplar (Sauter, 1988), but the transcriptional induction of key enzymes of this pathway, such as beta-amylase, glucan–water dikinase (SEX1) and starch phosphorylase, occurs prior to a reduction in the temperature (Figure 6a). Thus factors other than low temperature, e.g. short day length, are likely to act as a trigger in the transcriptional induction of the starch breakdown pathway. Starch breakdown provides substrates that appear to be utilized in a range of pathways. Our metabolite profiling data indicate an increase in the levels of sucrose, raffinose, the raffinose precursor, galactinol, and other potential cryoprotectants from August onwards (Figure 2) during the period in which starch breakdown occurs. The expression of the genes of the raffinose biosynthesis pathway is also induced during this period (Figure 6b), which supports the metabolic profiling data. Simultaneously with starch breakdown, transcriptional induction of the genes encoding glycolytic enzymes occurs as well (Figure 6c) indicating the activation of this pathway to meet increased demand for energy and/or carbon skeletons during this phase of the acquisition of cold hardiness. Taken together, these data suggest that starch breakdown plays a key role in the autumn transition by supporting the production of cryoprotectants and energy generation as photosynthesis declines.

image

Figure 6.  Expression of the carbon metabolism-related genes during the autumn. (a) Starch breakdown pathway: beta-amylase (PU03585, PU03653, PU10165), SEX1 (PU05226), starch phosphorylase (PU03913). (b) Raffinose biosynthesis pathway-related genes: raffinose synthase (PU03091), inositol monophosphatase (PU03461, PU03821). (c) Glycolysis-related genes: enolase (PU07901, PU06779), glyceraldehyde 3-phosphate dehydrogenase (PU06261, PU01445, PU02442), phosphofructokinase (PU04051), phosphoglycerate kinase (PU00556, PU00536), phosphoglycerate mutase (PU03188, PU03310), pyruvate kinase (PU03187, PU06775, PU05377, PU03877, PU13112, PU04889), triosephosphate isomerase (PU02106, PU07749, PU02496). (d) Expression of fatty acid and acetyl-CoA biosynthesis-related genes: acyl carrier protein (PU00675, PU01108), acyl-CoA thioesterase (PU07433), ATP citrate lyase (PU06429), long chain fatty acid-CoA ligase (PU03668), phosphoenolpyruvate carboxylase (PU02953), pyruvate dehydrogenase E1a component (PU09583), stearol-ACP desaturase (PU01531, PU04034). All expression data are presented in log2 scale.

Download figure to PowerPoint

The fatty acid biosynthesis pathway is stimulated during the induction of dormancy and is interlinked with starch breakdown

During the induction of dormancy, the central vacuole in the cambial cells disintegrates into several smaller vacuoles, necessitating the synthesis of new vacuolar membranes and membrane lipids (Farrar and Evert, 1997). This is reflected in the increased levels of transcripts encoding enzymes involved in fatty acid biosynthesis and elongation [acyl-CoA synthetase, acyl-CoA thioesterase and acyl carrier protein (ACP) during this period (Figure 6d)]. The induction in the autumn of transcripts encoding pyruvate dehydrogenase, phosphoenol pyruvate carboxylase and ATP-citrate lyase (Figure 6d) suggest that pyruvate and/or phosphoenol pyruvate derived from glycolysis could be used to generate the acetyl-CoA necessary for fatty acid synthesis. The increased levels of glyceric acid and free fatty acids detected during this period may also be related to this shift in metabolic focus as cessation of growth occurs in the autumn. Thus it appears that starch breakdown during autumn also plays an important role in supporting fatty acid biosynthesis, thereby linking carbohydrate metabolism to fatty acid metabolism.

The early phase of cambial reactivation is associated with sucrose breakdown and induction of the glyoxylate cycle

The reactivation of cambium in the spring occurs before any significant photosynthesis activity is under way and therefore the dividing cells require an alternative source of energy and carbon skeletons during this period. Induction of sucrose synthase and various invertases during the early phase of reactivation in spring (Figure 7a) indicates that sucrose catabolism generates hexoses that can be metabolized via glycolysis during this phase. The reducing power on the other hand appears to be generated through the induction of the oxidative pentose phosphate shunt, as indicated by the induction of glucose-6-dehydrogenase, transketolase and 6-phosphogluconolactonase (Figure 7b). This pattern is fully consistent with earlier enzyme activity data (Sagisaka, 1974). The expression of several genes encoding enzymes of beta-oxidation and the glyoxylate cycle, such as AIM1/MFP, KAT5 (beta-keto-acyl-CoA thiolase), ICL, citrate synthase, malate synthase and aconitate hydratase, is induced either in late autumn or early spring (Figure 7c,d). The exact role of the induction of the expression of these genes in autumn is currently unclear, since fats are stored rather than utilized during this period. However, some of these genes are superinduced in spring and high transcript levels for other genes of this pathway are maintained in the spring. These data suggest that fats are metabolized during cambial reactivation in spring via beta-oxidation and the glyoxylate cycle. This suggestion is supported by the metabolic profiling data which record an increase in cis-aconitic acid and itaconitic acid with a simultaneous reduction in trisaccharides, sterols and fatty acids during this period (Figure 3, Supplementary). Citrate, malate and succinate derived from activity of the glyoxylate cycle are likely to be used to provide substrates for the tricarboxylic acid (TCA) cycle, suggesting an anaplerotic role for glyoxylate cycle during reactivation.

image

Figure 7.  Expression patterns of the carbon metabolism-related genes during the spring. (a) Sucrose breakdown: inline image beta fructosidase (PU04013, PU03467), inline image sucrose synthase (PU01379, PU03976, PU09676, PU12930, PU12928). (b) Oxidative pentose phosphate shunt: inline image glucose 6-phosphate 1-dehydrogenase (PU08630, PU08312), inline image transketolase (PU02587), inline image 6-phosphogluconolactonase (PU02575). (c), (d) Beta-oxidation and glyoxylate cycle: inline image malate synthase (PU04017, PU03896), inline image isocitrate lyase (PU03380, PU03658), inline image phosphoenolpyruvate carboxylase (PU02953), inline image phosphoenolpyruvate carboxykinase (PU00909), inline image aconitate hydratase (PU00663, PU02155, PU01352, PU12725), inline image acyl-CoA thioesterase (PU07433), inline image fatty acid multifunctional protein (PU12326, PU02057), inline image 3-ketoacyl-CoA thiolase (PU01071), inline image citrate synthase (PU02335), inline image pyruvate dehydrogenase E1 (PU08187). All expression data are presented in log2 scale.

Download figure to PowerPoint

Expression of the bark storage protein gene is regulated at both transcriptional and post-transcriptional levels during the activity–dormancy cycle

A perennial growth habit requires the ability to store and remobilize various types of resources at the appropriate time. A major nitrogen resource that undergoes cycles of storage and remobilization in poplar is bark storage protein (Clausen and Apel, 1992). Bark storage protein transcripts are highly induced in the cambium as the transition to dormancy takes place from August onwards (Figure 8a) (Clausen and Apel, 1992).

image

Figure 8.  Expression of bark storage protein and amino acid metabolism-related genes during the activity–dormancy cycle. Bark storage protein gene expression profiles during (a) autumn, (b) spring and (c) expression analysis of amino acid metabolism-related genes alanine-glyoxylate aminotransferase (AGT), aspartate aminotransferase (AST), glutamate decarboxylase (GAD) and glutamate-1-semialdehyde aminotransferase (GSA-AT). Values on the Y-axis are the expressed ratios in log2 scale.

Download figure to PowerPoint

Metabolic profiling experiments also show an accumulation of amino acids that are essential for biosynthesis of storage protein in the cambial cells during this period (Figure 3). Reactivation of cambial activity in the spring overlaps with the degradation of storage proteins and the metabolic profiling data also indicate an increase in the levels of amino acids during this period (Figure 3). Interestingly, high levels of storage protein mRNA still persist until late May, long after storage protein synthesis has presumably been switched off (Figure 8b). This raises questions concerning the mechanism underlying the ability of the translational machinery to specifically terminate the translation of storage protein transcripts when corresponding protein synthesis is no longer necessary. In Arabidopsis seeds, there have been indications that specific factors prevent the translation of stored maturation phase RNAs such as those for late embryogenesis abundant (LEA) proteins during germination (Rajjou et al., 2004). Whether a similar mechanism might prevent the translation from mRNA encoding storage proteins while allowing other proteins to be synthesized during the reactivation stage is not clear at this stage.

Amino acids derived from storage breakdown could be utilized for replenishing TCA cycle substrates as well as for generating GABA

To date, the amino acids derived from the breakdown of storage proteins have been thought to be mainly required for the synthesis of new proteins as cambial reactivation occurs in the spring (Cooke and Weih, 2005). However, the transient induction of aspartate aminotransferase during cambial reactivation (Figure 8c) would suggest that glutamate could be converted to alpha-ketoglutarate and aspartate. This would not only allow the regeneration of carbon skeletons necessary for nitrogen assimilation but also permit primary assimilation to take place even when alpha-ketoglutarate is limiting during this phase (Cooper and Meister, 1985; Given, 1980). Similarly, glutamate could potentially be converted to succinate via the GABA shunt, thereby providing an alternative pathway for entry into the TCA cycle, as suggested by the induction of transcripts for glutamate decarboxylase and glutamate semialdehyde aminotransferase in the spring (Figure 8c; Vandewalle and Olsson, 1983). Thus, this pathway could have relevance for replenishing the substrates for the TCA cycle during the early phase of reactivation before leaf photosynthesis becomes sufficiently active. Finally, the increased levels of GABA detected by metabolite profiling indicate that glutamate-derived GABA could act as a signaling molecule in the activation of stress response genes in the autumn and early spring as cold hardiness is acquired (Bouche et al., 2003). These data would suggest that amino acids derived from the breakdown of storage protein may be used not only for the synthesis of new proteins but also for other purposes such as replenishing TCA cycle substrates during the early phase of cambial reactivation.

A role for gibberellin in regulating cambial activity during the activity–dormancy cycle

Several reports have suggested a role for hormones, particularly ABA and gibberellins (GAs), acting downstream of environmental signals in regulating cellular responses during the activity–dormancy cycle (Eriksson and Moritz, 2002; Gusta et al., 2005; Juntilla and Jensen, 1988; Olsen et al., 1997; Powell, 1987). We therefore analyzed biosynthesis and signaling pathways connected to these hormones during the activity–dormancy cycle. The role of ABA in the induction of cessation and reactivation of growth in trees has been controversial (Powell, 1987), but in this study we did observe an increase in cambial ABA levels after the induction of growth cessation had occurred (compare Table 1 and Figure 1) Similarly, our data reveal overlap between high levels of ABA in cambial cells and the early stages of cambial reactivation (compare Table 1 and Figure 1). We interpret this pattern to mean that, contrary to previous suggestions, increases in the cambial ABA levels in aspen do not influence the cessation of growth or negatively influence its reactivation.

A reduction of GA levels induced by exposure of plants to short days has also been thought to play an important role in triggering cessation of growth in the apex (Juntilla and Jensen, 1988; Olsen et al., 1997). Indeed, plants overexpressing GA-20 oxidase, a gene encoding a key gibberellin biosynthetic enzyme, have been shown to exhibit a delay in growth cessation (Eriksson and Moritz, 2002). However, none of the GA-20 oxidase genes represented on the array were transcriptionally downregulated during the cambial transition to dormancy. While this might reflect differences in the regulation of these genes in the cambium as opposed to the apex during growth cessation, we note that the expression of the poplar homolog of RGA (PttRGA; Figure 9a) is induced during growth cessation. Arabidopsis RGA and related genes are key components of GA signaling, and interestingly, these have recently been implicated in negatively regulating plant growth (Achard et al., 2006; Busov et al., 2006). In contrast to autumn, there is a transient induction of the poplar GA-20 oxidase gene PttGA20ox (Eriksson and Moritz, 2002) during the early phase of cambial reactivation (Figure 9b). This is an interesting parallel to activation of seed germination in Arabidopsis, where a transient cold-induced increase in GA levels is observed (Yamauchi et al., 2004). The induction of PttGA-20ox transcripts during the early stages of reactivation is particularly relevant in view of the finding that GA stimulates the breakdown of the RGA protein (Silverstone et al., 2001). Thus, one possible hypothesis is that the signals inducing reactivation could act by inducing GA-20 oxidase transcripts and a corresponding increase in GA levels. These in turn could stimulate the breakdown of RGA, a growth inhibitor. Based on these results, we propose that the induction of a negative regulator of GA signaling would contribute to restricting growth in the autumn during the induction of dormancy, whereas the induction of GA biosynthesis in early spring would stimulate growth as reactivation occurs.

image

Figure 9.  Expression profiles of selected chromatin remodeling proteins and GA biosynthesis and response pathway-related genes during the activity–dormancy cycle. (a) Expression during the autumn and (b) the reactivation phase. Values on the Y-axis are the expressed ratios in log2 scale. Data are presented for the following clones: PttFIE (PU03696); PttRGA (PU03640); PttGA20ox (PU09349, PU07797; PttEZA, PU03388).

Download figure to PowerPoint

A role for chromatin remodeling in regulating global transcriptional changes during the activity–dormancy cycle

Our data reveal massive reprogramming of the cambial transcriptome and metabolome during the distinct stages of the cambial activity–dormancy cycle. This transcriptional reprogramming can be orchestrated through stage-specific induction of transcriptional activators and repressors. However, the scale of alteration in the transcriptional programming also suggests a key role for mechanisms such as chromatin remodeling that act on a more global scale. In this respect it is interesting to note the transcriptional induction of poplar homologs of the Fertilisation Independent Endosoperm (PttFIE) gene (Ohad et al., 1999; Figure 9a) during the dormancy transition and of Enhancer of Zeste (PttEZA;Riechmann et al., 2000; Figure 9b) during the early phase of cambial reactivation. Drosophila homologs of both these proteins have been shown to be a part of the polycomb repression complexes that are involved in transcriptional repression. The polycomb repression complex represses gene expression by targeting histone deacetylases to specific chromosomal regions and creating transcriptional repression domains bearing deacetylated histones (Breiling et al., 2001). Importantly, suppression of Arabidopsis FIE expression leads to the aberrant activation of several key regulators of meristem activity (Katz et al., 2004). This suggests an attractive hypothesis in which the poplar FIE homolog would also be involved in the regulation and/or maintenance of dormancy through its regulation of meristem activity genes by modification of changes in the histone deacetylation patterns in cambial meristem cells.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Plant material and RNA preparation

Wild-grown Populus tremula (referred to as aspen) with a height of approximately 6 m and a stem diameter of approximately 40 mm, located at latitude 63°48′ N, longitude 20°17′ E about 500 m from Umeå airport, were used for sampling in all the experiments. Sampling was performed from 20 April (before bud break) until 13 December of the year 2000 at nine different time points to cover all the major stages of the activity–dormancy cycle. Three 10-cm stem sections with a diameter of 3–4 cm were sampled at approximately 1.5 m above ground from three independent trees at each time point. Samples were flash frozen in liquid nitrogen and stored at –80°C. Frozen stem segments from the nine sampling dates were trimmed to blocks of approximately 15 mm length (axial) and 2.5 mm width (tangential). Tangential sections through the cambial region of the stem were obtained using a cryomicrotome and frozen in liquid nitrogen as described earlier (Uggla et al., 1996). Briefly, 30-μm sections were isolated by tangential cryosectioning at –20°C with a HM 505 E microtome equipped with a steel knife. In order to ensure that tangential sections were parallel to the cambium, transverse sections were cut from each end of the specimen and these were mounted in 20% glycerol and observed under the microscope using Nomarski optics to determine deviations from parallel. The block was reoriented in relation to the knife and the procedure was repeated until the sample was properly oriented. The position of the zone of cambial cells was identified by using transverse sections that marked the secondary xylem and phloem cells that had a distinct appearance from the radially narrow fusiform cambial cells and flanked and marked the position of the cambial cells. This position of cambial cells was then utilized to isolate sections for further analysis. From each stem segment, two pairs of cryosections representing the cambial cells (30 μm × 2 mm × 20 mm, about 0.5 mg) were pooled (a total of four sections) and sections from two independent trees per time point were used as biological replicates.

Anatomical observations of the cambium

For anatomical studies, transverse sections were prepared from the stems (corresponding to those used for microarray analysis) using cryomicrotome as described earlier (Schrader et al., 2003). The sections were stained with toluidine blue in 20% calcium chloride followed by ruthenium red and mounted in 50% glycerol.

Ribonucleic acid isolation and probe preparation for microarray hybridization

For each time point, total RNA was prepared (Chang et al., 1993), and messenger RNA was obtained from 1 μg total RNA using paramagnetic oligo(dT) beads (Dynabeads, Dynal Biotech, http://www.invitrogen.com/dynal) following the manufacturer’s instructions. Messenger RNA from the cambial sections at each time point was PCR amplified as described (Moreau et al., 2005). For each sample an internal standard consisting of equal amounts of Lucidea Universal Score Card control cDNA (http://www.amershambiosciences.com) was added to 100 ng of amplified cDNA. A common reference experimental design was used to allow a comparison of each sample against any other. The RNA for the common reference was derived by pooling of total RNA extracted from all the sections. Labeling of the amplified cDNA samples was performed by direct incorporation of 3 μl Cy3-dUTP or Cy5-dUTP (Amersham Biosciences, http://www5.amershambiosciences.com/) in an asymmetric PCR reaction with 100 ng cDNA, 1 μm MaraAP1 primer, 0.6 μl AmpliTaq, 67 mm 2-amino-2-(hydroxymethyl)-1,3-propanediol (TRIS)-HCl (pH 8.8), 4 mm MgCl2, 16 mm (NH4)2SO4, 80 μm of each of dATP, dCTP, dGTP and 20 μm dTTP in a total reaction volume of 50 μl. The PCR conditions were 95°C for 1 min followed by nine cycles of 95°C for 30 sec, 50°C for 30 sec and 72°C for 10 min. The PCR product was purified with a QIAquick PCR purification kit (www1.qiagen.com) and eluted twice in 30 μl of 4 mm KPO4 buffer (pH 8.5). The final volume was reduced to 41 μl.

Microarray hybridizations and data analysis

Microarray hybridizations were performed as described (Schrader et al., 2004). Three to four replicate hybridizations, including a dye-swap, were utilized for the analysis for all samples. Taking into account the duplicates on the slides, six to eight data points for each clone were obtained. Arrays were scanned with a ScanArray 4000 (Perkin-Elmer Sverige AB, http://www.perkinelmer.com) at wavelengths of 543 and 633 nm for the two fluorescence dyes Cy3 and Cy5, respectively, at a high resolution (5 μm). Image analysis was performed with GenePix Pro4.1 software (Axon Instruments, http://www.axon.com/).

The entire data set was divided into two phases to simplify the analysis of data and investigate cambial activation in the spring and dormancy in the autumn. Phase 1 comprises April to July and phase 2 June to December. The two data sets were subsequently treated independently of each other. Variation in gene expression was assessed by using a limma (Linear Models for Microarray data) package (Smyth, 2004) in Bioconductor R 1.9.0 (http://ww.bioconductor.org). Data were treated by print-tip loess normalization after background subtraction (Edwards, 2003). Linear models with B statistics implemented in the limma package (Smyth, 2004) for the statistical software package R 1.9.0 (http://ww.r-project.org) were used to obtain relative expression values for individual genes as well as to identify differentially expressed genes (Diaz et al., 2003). The biological replicates were treated as separate replicates during the statistical analysis and averaged for the clustering and visualization. A gene was considered differentially expressed if its associated B statistic was ≥5 for at least one time point of the series. The B statistic indicates the probability of a gene being differentially expressed, with a B value of 0 representing a 50:50 chance of differential expression. Differentially expressed genes (2266 in phase 1 and 1711 in phase 2) were clustered using K-means clustering as implemented in the TIGR MultiExperiment Viewer 3.0.1 software (Saeed et al., 2003). Clusters were manually optimized to remove redundancy in expression patterns. The microarray raw data, including tiff and gpr files from scanning and image analysis, can be publicly accessed at the UPSC-BASE microarray database (Sjodin et al., 2006) under experiment ID UMA-0014.

Reverse transcriptase-PCR analysis

Total RNA was extracted from stem tissues using RNeasy kit (Qiagen,http://www.qiagen.com/) used according to the manufacturer’s instructions. Total RNA (2 μg) was reverse transcribed using a First Strand cDNA Kit (Amersham Pharmacia Biotech). First-strand cDNA was used as template for RT-PCR quantifications performed using a Quantum 18S RNA internal standard kit (Ambion Inc., http://www.ambion.com/) according to the manufacturer’s instructions. Relative expression was calculated by comparison with 15S RNA. The primer sequences used for amplification were as follows: PU03474, bark storage protein (5′-TTTATGAGCCTGACAGTGAAAATC and 3′-CAGAAGCACTTGCTGTATCGGCAG); PU04013, beta fructosidase (5′-GAGAACGTGAATCAAGTCCATAG and 3′-GCTGCCTATCAAAATCCTCCAATG); PU05693, histone H4 (5′-CCAAGAGGCATCGTAAGGTTCTTCG and 3′-ATCATGAAGCCAAACAGATCGGC); PU07797, GA20oxidase (5′-CTACTGTGAGGCCATGAGCACTTTG and 3′-GTGAACACCTCAAGTGTCTTCATG); PU03658, isocitrate lyase (5′-ATTGGAGAGTACGAGAAGAGGAG and 3′-GCATCTGCATGGAAACCTGCC); PU03091, raffinose synthase (5′-TTTGCCCTGCCCACTAGGGATTG and 3′-GGAGATTCAAGCTCAGTTTGCC); PU03585, beta-amylase (5′-TGCATGGAGATAATATCCTCAATG and 3′-GGGCAGTAATCTTGATCAGCATG).

Metabolomic analysis

For metabolite profiling analysis samples were taken at seven time points over the year, (20 April, 9 June, 7 July, 16 August, 11 September, 5 October and 13 December). Six sections from each time point were analyzed, except for the April time point where only four samples were analyzed. The metabolites were extracted with 1 ml methanol, chloroform and water mixture as described (Gullberg et al., 2004). Nine hundred microliters of the extract was evaporated to dryness, and thereafter further fractionated on a mixed mode solid-phase column (Oasis®MCX, 60 mg, Waters, http://www.waters.com/). The dried extracts were dissolved in 600 μl Milli-Q water and the pH adjusted to about three with acetic acid prior to the fractionation. The column was conditioned with methanol, equilibrated with 1% acetic acid, the extract applied and the run-through extract collected. The column was washed with 2 ml 1% acetic acid and the run-through was pooled with the previous (acidic fraction). To elute neutral and non-polar compounds 2 ml methanol was added and the run-through collected (organic fraction). As a final step 2 ml 0.35 m NH4OH:methanol (40:60) was added to elute basic metabolites (basic fraction). As solid-phase extraction (SPE) fractionation can result in sub-fractionated samples, we verified that the sub-fractionation did not affect the statistical analysis by manually verifying how many of the statistically significant metabolites were in found in two or more fractions. More than 90% of the putative derivatized metabolites were occurring in one fraction and in those cases where sub-fractionation did occur; it did not affect the statistical analysis.

The fractions were concentrated in a speed-vac transferred to GC vials, evaporated to dryness and derivatized according to Gullberg et al. (2004). The fractions were analyzed on GC-TOFMS (Leco Corp., http://www.leco.com/) as described (Gullberg et al., 2004; Jonsson et al., 2005). All non-processed MS files from the metabolic analysis were exported from the ChromaTOF software in NetCDF format to matlab™ software 7.0 (Mathworks, http://www.mathworks.com/), where necessary data pre-treatment procedures such as base-line correction, peak alignment and hierarchical multivariate curve resolution (H-MCR) were performed using custom scripts according to Jonsson et al. (2005). The result is a data table where each row represents one sample, and the columns correspond to its resolved peak area intensities. Normalization of each row was performed by dividing its values by their total summed intensity. In addition, column centering and scaling to unit variance was also done prior to modeling. Partial least square discriminant analysis was used to reveal metabolite differences between the samples. Partial least square discriminant analysis is a method where a quantitative relationship between two data tables X and Y is sought. Here, X is the data table of resolved GC/MS peak areas and the corresponding rows in Y contain information about its class belonging. All multivariate statistical investigations (principal components analysis, PLS-DA) were performed using simca-p software 10.5.0.0 (Umetrics, http://www.umetrics.com/).

For the different fractions, PLS-DA models between April and June, July and August, August and September, September and December were calculated. The PLS-DA model interpretation was performed using the loading profile of the first component (w1) in combination with the explained variation (R2VX) of each variable. Variables with a R2VX > 0.5 and significant according to the jack-knifing results using 99% confidence interval were considered putative metabolites. These putative metabolites were identified by comparisons of their retention index and mass spectra with retention time index and mass spectra libraries (Schauer et al., 2005).

Abscisic acid measurements

Frozen samples were individually placed, with 500 μl of extraction medium (MeOH:H2O:CH3COOH, 800:190:10; MeOH = methanol) including stable isotope internal standard [1 ng 2H6-ABA (purchased from Professor Suzanne R. Abrams, Simon Fraser University, Saskatoon, Canada)] into 1.5-ml Eppendorf tubes. After extraction using an MM 301 Vibration Mill (Retsch GmbH & Co. KG, http://www.retsch.com/) the supernatant was evaporated to dryness. The residue was dissolved in H2O and applied to a pre-equilibrated 100-mg C8-EC ISOLUTE cartridge (Sorbent AB, http://www.sorbentab.se), after adjusting the pH to 3.0. The column was washed with 1% aqueous acetic acid, and then ABA was eluted with 80% MeOH. After evaporation to dryness, the samples were methylated with ethereal diazomethane, and thereafter analyzed by combined GC/MS high-resolution selected ion monitoring (GC/MS-HR-SIM; Moritz and Olsen, 1995) using a JMS–700 MS station mass spectrometer (JEOL, http://www.jeol.com). The resolution was set to 10 000.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

The work in the laboratory of RPB was funded by grants from Vetenskapsrådet and STEM. We thank Erik Walfridsson and Kjell Olofsson for anatomical analyses and Dr Vaughan Hurry, Professor Brian Ellis and Dr Antje Rohde for careful reading of the manuscript and helpful comments.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Results and discussion
  5. Experimental procedures
  6. Acknowledgements
  7. References
  8. Supporting Information

Figure S1. Ambient temperature during the experimental period. The red line and the blue line represent the maximum and minimum temperatures, respectively in degrees Celsius. Figure S2. Semi-quantitative RT-PCR analysis of the expression of selected clones during activity-dormancy cycle. (a) bark storage protein (PU03474), (b) beta-fructosidase (PU04013), (c) histone H4 (PU05693), (d) PttGA20oxidase (PU07797), (e) isocitrate lyase (PU03658 (f) bark storage protein (PU03474), (g) histone H4 (PU05693), (h) isocitrate lyase (PU03658), (i) raffinose synthase (PU03091), (j) beta-amylase (PU03585). Values on the Y-axis are relative expression levels. Table S1. Differentially expressed clones during spring reactivation (Phase 1) and autumn growth cessation and dormancy (Phase 2). Table S1 describes the clones whose expression is significantly altered (See Figure 2 for expression profiles) during the period April to June (phase 1) and June to December (phase 2). The table describes the cluster, PU number, poplar EST corresponding to the PU number (see http://www.populusdb.umu.se for details), closest Arabidopsis gene (AGI number), poplar gene model corresponding to the EST, annotation and relative gene expression level. Poplar gene models sequences were utilized to obtain annotation as described earlier (Sterky et al., 2004). Table S2. Analysis of cambial metabolome during the course of activity-dormancy cycle. GC/TOFMS peaks according to PLS-DA identified as important for explaining the differences between cambial samples at different stages of activity-dormancy cycle. aSpA=peak more abundant in April than June; SpJ=peak more abundant in June than April; SuJ=+June/-August; SuA=-June/+Aug; AuA=+Aug/-Sep; AuS=-Aug/+Sep; WiS=+Sep/-Dec; WiD=-Sep/+Dec; bPeaks are named according to UPSC- in-house mass spectra library. cAnnotation of peaks were performed by comparing mass spectrum and retention index (RI) with the dUPSC in-house mass spectra library or the mass spectra library maintained by the Max Planck Institute (MPI) in Golm (http://csbdb.mpimp-golm.mpg.de/csbdb/gmd/gmd.html). ePeaks annotated or classified according “M000000…” are identical or similar to non-annotated mass spectra in the MPI-library. Naming refers to MPI-spectra numbering. UPSC mass spectra will shortly be available for download on UPSC homepage (http://www.upsc.se). Table S3. Analysis of the core cell cycle gene expression during spring reactivation and autumn growth cessation and dormancy. Table S3 describes the poplar cell cycle genes represented on the array. Poplar homologs of Arabidopsis cell cycle genes were identified by comparing the sequences of poplar clones represented on the array with that of Arabidopsis core cell cycle genes described by Menges et al. (2005) using tblastX (Altschul et al., 1990). The cell cycle genes whose expression was significantly altered in phase 1 or Phase 2 is indicated with a cross in the table and their expression profiles are shown in figure 4. Table S4. Analysis of the expression of cold hardiness related genes during the course of activity-dormancy cycle. Table S4 describes the low temperature induced and cold hardiness related genes in poplar. Genes compiled in this table were obtained by: (i) identifying poplar homologs of Arabidopsis genes shown to be low temperature induced by (Vogel et al., 2005), (ii) poplar genes induced by low temperature or by overexpression of AtCBF1 (Benedict et al., 2006) and (iii) poplar genes homologous to those that have been implicated earlier in cold hardiness related processes in tree species (Rowland and Arora, 1997; Renaut et al., 2004). Where known, specific classes of genes are color coded. Expression profiles of clones in bold letters are shown in figure 5 and the cluster to which they belong are indicated in the table. ReferencesAltschul, S.F., Gish, W., Miller, W., Myers, E.W. and Lipman, D.J. (1990) Basic local alignment search tool. J. Mol. Biol., 215, 403-410. Sterky, F., Bhalerao, R.R., Unneberg, P., Segerman, B., Nilsson, P., Brunner, A.M., Charbonnel-Campaa, L., Jonsson-Lindvall, J., Tandre, K., Strauss, S.H., Sundberg, B., Gustafsson, P., Uhl�n, M., Bhalerao, R.P., Nilsson, O., Sandberg, G., Karlsson, J., Lundeberg, J. and Jansson, S. (2004) A populus EST resource for plant functional genomics. Proc Natl Acad Sci USA, 101, 13951-13956.

FilenameFormatSizeDescription
TPJ_3077_sm_FigS1.pdf156KSupporting info item
TPJ_3077_sm_FigS2.pdf278KSupporting info item
TPJ_3077_sm_Tables.zip516KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.