The depth of seed dormancy can be influenced by a number of different environmental signals, but whether a common mechanism underlies this apparently similar response has yet to be investigated. Full-genome microarrays were used for a global transcript analysis of Arabidopsis thaliana Cape Verde Island accession seeds exposed to dry after-ripening (AR), or low temperature, nitrate and light when imbibed. Germination studies showed that the sensitivity of imbibed seeds to low temperature, nitrate and light was dependant upon the length of time spent AR following harvest. Seeds had an absolute requirement for light to complete dormancy release in all conditions, but this effect required an exposure to a prior dormancy relieving environment. Principal component analyses of the expression patterns observed grouped physiological states in a way that related to the depth of seed dormancy, rather than the type of environmental exposure. Furthermore, opposite changes in transcript abundance of genes in sets associated with dormancy, or dormancy relief through AR, were also related to the depth of dormancy and common to different environments. Besides these common quantitative changes, environment-specific gene expression patterns during dormancy relief are also described. For example, higher transcript abundance for genes linked to the process of nitrate accumulation, and nitrate reduction was associated with dormancy relief. The quantity of GA3ox1 transcripts increased during dormancy relief in all conditions, in particular when dormancy relief was completed by exposure to light. This contrasts with transcripts linked to abscisic acid (ABA) synthesis, which declined. The results are consistent with a role for the ABA/gibberellic acid balance in integrating dormancy-relieving environmental signals.
Physiological dormancy is the most prevalent form of seed dormancy (Baskin and Baskin, 1998, 2004), and it is present throughout the plant kingdom with a wide biogeographical distribution. Seed dormancy has, therefore, probably evolved differently across species, through adaptation to various habitats and prevailing environments, so that the completion of germination occurs when conditions for establishing a new plant generation are likely to be suitable (Baskin and Baskin, 1998; Fenner and Thompson, 2005; Finch-Savage and Leubner-Metzger, 2006). As a result of this adaptation, the induction and loss of physiological dormancy can be triggered by divergent environmental cues activated through many, apparently different, physiological mechanisms. These cues can be seasonally characteristic (usually temperature) and integrated by the seed over time. For example, dormancy may be broken by higher or lower temperatures, depending on species, in order for the completion of germination to occur in the correct season (autumn or spring, respectively) for subsequent growth. However, additional environmental conditions may then be required to end dormancy and terminate germination (e.g. light and nitrate). In this way, flushes of radicle emergence are strongly promoted by disturbance of the soil, which terminates dormancy and induces the completion of germination of seeds by exposing them to light, but the timing of flushes is largely governed by temperature-driven changes in the depth of seed dormancy.
Even within the same species, several cues and responses may operate with differing importance in distinct ecotypes that may also have different depths of dormancy. For example, in Arabidopsis these differences in response can result in different seasonal patterns of seedling emergence, so that different populations can express either a winter or spring annual life history, and in some cases both (Donohue, 2005). These different environmental responses may not be necessarily species specific, and so may not be fundamentally physiologically different. However, they do have a genetic basis and can have an influence on plant demography and evolution (Donohue, 2005). In addition to the post-dispersal environment, conditions during seed maturation (photoperiod, temperature and light quality) can also alter the depth of dormancy to further modify the requirements for the completion of germination. Therefore, it is possible that the wide range of interactions between the environment and physiological dormancy may have common underlying genetic mechanisms, specifically adapted through natural allelic variation in key regulatory genes (Finch-Savage and Leubner-Metzger, 2006).
Although Arabidopsis thaliana is an ecologically appropriate species to study in this context, the commonly used accessions have only a shallow dormancy and a high rate of dry after-ripening so that dormancy can be transient. The Arabidopsis accession Cape Verde Island (Cvi) has deeper seed dormancy, which makes it particularly suitable for molecular studies of dormancy (Ali-Rachedi et al., 2004; Alonso-Blanco et al., 2003; Cadman et al., 2006; Koornneef et al., 2000). The Cvi accession has an absolute dependence on light to terminate dormancy and initiate radicle protrusion, and will cycle through different depths of dormancy unless exposed to light (Cadman et al., 2006). This feature is experimentally useful to separate the processes related to a change in the depth of dormancy from the processes that complete germination; in many systems that are studied, there appears to be a continuum between dormancy and the completion of germination, which has been a major constraint to the study of dormancy (Cohn, 1996).
Although dormancy can only be visually measured by the absence of radicle protrusion, the depth of dormancy in imbibed seeds is continually changing in response to the environment. Even in dry seeds, dormancy status changes with time (after-ripening). Therefore, dormancy can be considered a seed characteristic that determines the conditions required for the completion of germination (Fenner and Thompson, 2005; Thompson, 2000; Vleeshouwers et al., 1995). When viewed this way, any cue that widens the environmental conditions permitting the completion of germination should be regarded as a dormancy release factor, and when the seed no longer requires specific environmental cues, it is non-dormant. In the study reported here, we use full genome Arabidopsis microarrays to compare gene expression during release of primary dormancy in response to different environmental factors (dry after-ripening, low temperature, nitrate and light), and question whether a common underlying dormancy release mechanism exists.
Dry after-ripening, i.e. a period of dry storage for usually several months, is a common method used to relieve dormancy (Bewley, 1997; Kucera et al., 2005; Probert, 2000), and may represent a natural mechanism that can control dormancy in dry climates (Probert, 2000). However, the molecular mechanisms of dry after-ripening are not understood. It is also widely accepted that temperature is the major regulator of dormancy cycles in the soil (e.g. Probert, 2000; Baskin and Baskin, 1998; Fenner and Thompson, 2005). In addition, exogenous nitrate can influence primary dormancy and the requirement for light to promote the completion of Arabidopsis seed germination (Batak et al., 2002). The intensity of Arabidopsis dormancy at dehiscence is also influenced by the nitrate regime fed to the mother plant (Alboresi et al., 2005). Exposure to light often releases the final barrier to the completion of germination, thus terminating dormancy and germination. This phytochrome-mediated seed response to light can also be reversed in some cases by far-red light, until the seed is committed to the process of completing germination (Casal and Sánchez, 1998).
These treatments all influence dormancy status and germination in Arabidopsis Cvi (Ali-Rachedi et al., 2004; Cadman et al., 2006). However, studies in the field, with Arabidopsis and other species, highlight that there is an important distinction in seed response to these factors. (1) There are factors (e.g. dry after-ripening and temperature) that are related to slow seasonal changes, which are integrated over time to alter the depth of dormancy as well as the sensitivity to other factors (e.g. light, nitrate and temperature). (2) There are other factors that indicate in a more immediate way that conditions are suitable for the completion of germination (e.g. light), which could be considered to terminate dormancy and induce radicle emergence. Therefore, each factor could remove successive blocks to the progress of the germination process, and these factors usually need to be applied in a set order for them to work, i.e. in the process described, light must come last to be effective. However, it is not known how the responses to these factors interact at the molecular level.
In Cadman et al. (2006), a range of physiological states during cycling through secondary dormancy, following release of primary dormancy by after-ripening, were compared, and indicated that there was a common conserved mechanism underlying the dormant state. Collectively, gene expression patterns in these dormant states differed from those states where dormancy had been relieved by after-ripening, and gene sets characteristic of these two extremes were identified. As described above, the depth of dormancy can be influenced by a number of different environmental signals, but it is yet to be investigated whether a common dormancy breaking mechanism underlies these apparently similar responses. In the present work, the insights provided by Cadman et al. (2006) are built upon to compare the impact of dry after-ripening, temperature, nitrate, light, and combinations of these treatments, to address this question. To facilitate this, the extensive data sets presented by Cadman et al. (2006) have been re-analysed, to complement the original data sets collected from a range of treatments applied in the present study.
Seed treatment and germination
A. thaliana accession Cvi seeds were harvested, cleaned and equilibrated with ambient relative humidity (c. 33% RH; 20°C), which resulted in a seed moisture content of 7% (dry-weight basis). Seeds were then placed in hermetic dry storage at 20°C to dry after-ripen following two harvest dates; 2 July 2003 and 6 April 2005. Seeds harvested at the second of those dates, after-ripened very slowly, so that 200 days after harvest only 30% of the seeds were able to complete germination in the light on water (Figure 1a). At harvest, no seeds would complete germination in the light on water, and seeds would not complete germination in the absence of light at any harvest. However, by 25 days sensitivity to nitrate had increased so that 15% of the seeds would complete germination in the light on a solution containing 10 mm KNO3, and higher percentages would complete germination on 20 and then 50 mm KNO3. By 75 days after harvest, a greater percentage of germination in the light was possible when the seeds were placed on a solution containing ≥ 10 mm KNO3.
During after-ripening the seeds also became increasingly sensitive to a period of exposure to low temperature (cold; 3°C) during imbibition. This exposure to cold resulted in the release of dormancy in a proportion of the population that would then complete germination in the light when placed back at 20°C. At each sample date the percentage radicle emergence was increased by 2 days of cold exposure, and this increased to a maximum after 4 days of cold exposure (Figure 1b). Exposure to cold for more than 4 days re-induced dormancy in a proportion of the seeds. This same pattern was followed at all sample dates. By 117 days of dry after-ripening, dormancy in 65% of the seed population was released by exposure to cold for 4 days, to allow the completion of germination in the light (Figure 1a,b). Continued after-ripening beyond this date did not increase the number of seeds in which dormancy was released by cold.
In contrast to seeds harvested on 6 April 2005, all seeds in the population harvested on 2 July 2003 were able to complete germination on 10 mm KNO3 by the time the seeds were cleaned and dried (20 days; data not shown). Indeed, the whole process of after-ripening was much more rapid in this seed lot (Figure 1c), even though it was produced from the same original accession, in the same glasshouse, and with the same minimum and venting temperatures. However, different light levels would have been experienced, and without cooling, different maximum temperatures would also have been experienced in the lots that were harvested in April 2005 and July 2003.
Principal component analysis (PCA) analysis reveals similarities between the effects of different dormancy-relieving treatments
To investigate similarities between the effects of different dormancy-relieving treatments on primary dormant seeds, microarray analyses were carried out to characterize gene expression in the nine different physiological states described in Table 1. PCA was applied to the expression of all the genes represented on the microarray over these different physiological states (Figure 2a). The analysis revealed clear differences between these states, but also identified similarities within three groups of treatments. The first dimension of the analysis (x-component) grouped seeds in the dry state (PDD and DDL) separately from a second group of imbibed seed treatments that were either primary dormant or had primary dormancy partially relieved (PD24 h, PDN, PDC and PDL), and those of a third group that were fully dry after-ripened or exposed to both nitrate and light (PDLN, LIG and DL). The dry states were likely to be grouped together and to be separate from the other states, despite the difference in dormancy status between them, because they would both retain transcripts stored during development on the plant that are subsequently lost in the early stages of imbibition (Nakabayashi et al., 2005). The second dimension (y-component) identified a difference between cold-treated seeds (PDC) and those of all other treatments. These first and second components explained 36 and 21% of the variance, respectively.
Table 1. Summary of the treatments carried out to produce physiological states for sampling and comparison by transcriptome analysis. Experimental treatments were conducted at 20°C on water in the dark unless otherwise stated. Comments in parenthesis describe the potential for seeds to germinate in the treatment applied. The characteristics of the PD24 h, DL and LIG samples were described in Cadman et al. (2006), characteristics of the remaining samples from imbibed treatments are shown in Figure 1a.
Primary dormant: seeds dry (will not complete germinate when imbibed in Light or dark)
Primary dormant seeds imbibed for 24 h (will not complete germination)
Primary dormant seeds after-ripened for 91 days and then imbibed for 24 h on a 10 mm KNO3 solution (will not complete germination unless exposed to light)
Primary dormant seeds after-ripened for 117 days and then imbibed for 4 days at 3°C (will not complete germination unless exposed to light)
Primary dormant seeds after-ripened for 91 days and then imbibed for 24 h in the light (will not complete germination)
Primary dormant seeds, dry after-ripened: seeds dry (will germinate when imbibed in the light)
Dry after-ripened seeds imbibed for 24 h (will germinate if placed in the light)
Primary dormant seeds after-ripened for 91 days and then imbibed in white light for 24 h on a 10 mm KNO3 solution (will complete germination)
Dry after-ripened seeds imbibed for 20 h in the dark and then 4 h in red light (will complete germination)
A second PCA was carried out to include microarray data from an additional four dormant samples, produced from the same accession of Cvi and described by Cadman et al. (2006). These samples were from primary dormant seeds imbibed for 48 h (PD48 h) and 30 days (PD30d), and from seeds with secondary dormancy induced when fully after-ripened by 25 days of imbibition at 20°C in the dark (SD1; one dormancy cycle), and then re-induced after dormancy loss by subsequent imbibition for 20 days at 3°C in the dark (SD2; two dormancy cycles). The impact of adding these extra treatments in the first two components of the analysis (Figure 2b) was to separate dormant samples (PD24 h, PD48 h, PD30d, SD1 and SD2) from those where dormancy had been only partially relieved (PDN, PDC and PDL), while retaining the other two groupings (PDD and DDL, and PDLN, DL and LIG). These two components accounted for 46% of the variance. These four groups of physiological states are explored further in the following sections.
It is interesting to note that, in the second PCA, these physiological states were grouped together in a way that related to the depth of dormancy of the seeds, rather than the type of environmental conditions to which they had been exposed. For example, primary dormant samples exposed to only nitrate, light or cold treatment (PDN, PDC and PDL) were grouped separately from the sample exposed to both nitrate and light, which was therefore capable of completing germination (PDLN), and consequently grouped with the after-ripened samples where primary dormancy had been fully relieved. These groupings provide an indication of similarity in the relief from dormancy by different environmental treatments, even though these different environments are also likely to have treatment-specific expression patterns characteristic of the environmental signal. This may have led to the separation of cold-treated samples from the others in the second component of the first PCA (Figure 2a). Below, we explore these similarities in gene expression that appear linked to the depth of dormancy, and then treatment-specific expression patterns that may induce these changes.
Relief of dormancy results in quantitative changes in gene expression that appear linked to depth of dormancy
Germination studies showed that primary dormancy was relieved by after-ripening (DDL and DL), although light was required to fully relieve dormancy and induce the completion of germination (LIG), and by the application of both light and nitrate (PDLN). Application of light (PDL), nitrate (PDN) or cold treatment (PDC), applied separately, did not fully relieve dormancy. To look for similarities in seed response to these different dormancy-relieving treatments, the transcripts in gene sets identified by Cadman et al. (2006), and associated with dormant and fully dry after-ripened seeds (Table S1), were investigated. The gene sets contained 442 genes associated with all dormant states (dormancy gene set, henceforth termed D-set), and a further set of 779 genes that was associated with dormancy relief in fully dry after-ripened states (after-ripened gene set, henceforth termed AR-set).
In the dry dormant state (PDD), and the dry after-ripened state (DDL), average transcript abundance of genes in the AR-set tends to be lower than those from the D-set (Figure 3a). This overall pattern remains in dormant seeds whether they were imbibed for 24 h or not (Figure 3b). However, in dry after-ripened seed, transcript abundance of the D-set decreases, and the AR-set increases, following 24 h of imbibition (DL), relative to unimbibed, dry after-ripened seed (Figure 3c). This is consistent with a developmental switch occurring during dry after-ripening (Bove et al., 2005).
In a further comparison of quantitative changes in gene expression during dormancy relief, the treatments were grouped according to the first component of the PCA analysis in Figure 2b (i.e. PD24 h/PDC, PDL, PDN/PDLN, DL, LIG), and the mean transcript abundance for each of the groups was plotted against that of dry, primary dormant seeds (PDD; Figure 3d,e). Comparison of the regression lines, fitted to these data, show that transcript abundance in the D-set (Figure 3d) was highest in dormant imbibed seeds (PD24 h), lower in the partial dormancy-releasing conditions (PDC, PDL and PDN), and lowest in PDLN, DL and LIG. The opposite pattern was observed for genes in the AR-set (Figure 3e). In both sets, the regression coefficient (slope of the fitted line) for the third group (PDLN, DL and LIG) tended to be lower than for the other two groups. This indicated that genes in the AR-set (Figure 3e) with lower expression increased more than those with high expression, whereas genes in the D-set (Figure 3d) with higher expression decreased more than those with low expression. Analysis of variance confirmed this biased change in transcript abundance of D and AR genes in the third group (PDLN, DL and LIG), relative to the dry dormant state (PDD; P < 0.01).
These quantitative changes in gene expression in response to dormancy-relieving treatments appear characteristic of all genes within the sets (Figure 3). These differences in response are reflected in the average transcript abundance of both gene sets when expressed relative to PDD seeds (Figure 4a). Analysis of variance showed that the average abundance of the D-set decreased, and the AR-set increased, as dormancy was relieved (P < 0.01). For example, the genes in PDLN seeds were expressed at similar levels to those in imbibed fully after-ripened seeds (DL and LIG). However, when light, nitrate or cold were applied singly, there was an intermediate shift in transcript abundance compared with combined light plus nitrate, although completion of germination was not induced. In comparison, the genomic average expression was fairly constant, with a maximal shift of 0.066 in relative expression levels between any treatment and PDD. As nitrate and cold-treated seeds retained an absolute requirement for light (no radicle protrusion without light), the intermediate expression level does not result from a population artifact, such as that which would occur when only a subset of the seeds in the sample population had dormancy relieved, and were capable of completing germination.
Cadman et al. (2006) reported that primary dormancy deepens with increased imbibition time (PD24 h < PD48 h < PD30d), and that secondary dormancy induced by cold (SD2) is deeper than when induced by warm temperatures (SD1). To illustrate this, re-normalized data from these treatments are also included in Figure 4a, and they show the reverse trend in average transcript abundance to that occurring when dormancy is relieved. Figure 3 shows that these trends in transcript abundance are not just evident in the average value, but universal in the genes that are represented in both gene sets. For example, the same trend is shown in genes of the AR- and D-sets that encode transcriptional regulators and proteins involved in ribosomal biogenesis (Tables S2 and S3, respectively). Collectively, these changes in transcript abundance of the two gene sets appear characteristic, and provide a quantitative impression of changes in dormancy status in response to environmental conditions.
During relief of dormancy by cold, nitrate and light, there is an increase in the transcription of genes associated with the preparation for germination before all the necessary environmental signals for dormancy release have been encountered. This pattern is evident in genes concerned with the construction of translation machinery (Table S2), and those encoding proteins involved in cell-wall modification and reserve mobilization (Figure 4b), which are required for the completion of germination. In this way, the seeds appear to have been prepared for rapid completion of germination when the environmental signal that removes the final block to radicle protrusion (e.g. light) is encountered, indicating that conditions are suitable for completion of germination and plant establishment. As imbibed seeds that do not encounter this signal can enter into the secondary dormant state (SD1 and SD2) after DL, these changes in transcript abundance are reversible, consistent with the model of dormancy cycling presented by Cadman et al. (2006).
These quantitative changes, therefore, appear common to all the environments that influence dormancy status (relief and induction). The extent of these changes is influenced by the timing of the seeds exposure to an environmental signal, and the order in which the environments are experienced.
Changes in transcript abundance during AR
After-ripening enhances seed sensitivity to factors that relieve dormancy and stimulate radicle protrusion (e.g. light, nitrate and gibberellins; Hilhorst and Karssen, 1992). This is evident in the results presented here in Figure 1a. The rate of AR is a function of both temperature and seed moisture content. Hydrothermal AR time models for Bromus tectorum (Bair et al., 2006) and Lolium rigidum (Steadman et al., 2003) indicate that AR occurs at a moisture content range between approximately 5% and 20% (dry-weight basis). This implies that for AR to occur some water is required, presumably the bound water of region 1 of the sorption isotherm (Leopold et al., 1988). The AR conditions of Arabidopsis Cvi seeds in the present work were 20°C and 33% RH, with a measured moisture content of 7% (dry-weight basis), presumably representing the bound-water fraction of region 1 (Hay et al., 2003). The oil content of Cvi seeds is approximately 28% of the dry weight (O’Neill et al., 2003). Hence, the overall moisture content of the non-oily seed parts is maximally 9.7%, which is assumed to be too low for any ordered metabolic activity (Vertucci and Farrant, 1995), yet this value is in the range where the rate of AR progress can be linearly related to moisture content (Bair et al., 2006).
In the present work it is not possible to compare directly the transcript profiles of the two dry samples (PDD and DDL) taken before and following after-ripening, as they come from different seed-production occasions. Differences in transcript abundance that may result from dry after-ripening are therefore likely to be confounded with differences in the abundance of stored transcripts left from seed development, as the seeds have not undergone imbibition and consequent transcriptional changes (Nakabayashi et al., 2005). These two samples are therefore useful to show general trends in expression (e.g. Figure 3), but direct comparison of the transcript levels of individual genes between only these two samples is liable to misinterpretation. Therefore, the following strategy was used to identify genes that are likely to be specifically related to dormancy relief, and the transcript abundance of which may have declined during AR. Those genes that had consistently higher transcript abundance (P < 0.05) in PDD and the dormant imbibed states (PD24 h, PD48 h, PD30d, SD1 and SD2), when compared with the dry (DDL) and imbibed (DL) AR samples, were sought. These were thought likely to be specifically related to dormancy relief, and that their transcript abundance may have declined during AR. This resulted in a set of 30 genes with similar expression patterns (Table S4). Of these 30 genes 14 were present in the D-set.
One of these genes was the recently cloned DOG1 (delay of germination; At5 g45830) encoding an unknown protein (Bentsink et al., 2006). DOG1 is functionally related to primary dormancy in Arabidopsis Cvi. Its expression decreased with after-ripening and during imbibition, which leads to the conclusion that the gene is related to dormancy induction during seed development (Bentsink et al., 2006). However, here we show that DOG1 expression increased during the induction of dormancy in imbibed seeds and most prominently so in the deep-dormant states (Figure 5), a result confirmed by QRT-PCR analysis (Table S5). Expression of DOG1 was lower in 24-h imbibed after-ripened seeds (DL) than in imbibed dormant seeds (PD24 h). Transcript abundance on the microarray in cold-treated seeds (PDC) was at a similar level to PDD and the deep-dormant states, but this was not confirmed by QRT-PCR analysis, which indicated that expression in PDC was at a similar level to that in the shallow-dormant and germinating states. The reason for this result is not known. Other dormancy-relieving treatments reduced expression levels from those of the deep-dormant states. As the other genes in the set of 30 showed similar expression patterns to DOG1, they may also be important candidate genes for the control of dormancy.
There are both common and unique patterns of gene expression in response to exposure with different dormancy-relieving environments during imbibition
PCA analysis grouped together all samples that were exposed to single, but different dormancy-relieving environmental signals (PDC, PDL and PDN; Figure 2b). This provided evidence of similarity in gene expression patterns during relief from dormancy in imbibed seeds exposed to these different treatments. There were 991, 250 and 474 genes that were expressed more highly (P < 0.05) in treatments that individually applied cold, nitrate and light, respectively, than in all dormant states (PDD, PD24 h, PD48 h, PD30d, SD1 and SD2). Of these, 59 were common to all conditions, and a high proportion (47) was present in the AR-set, including many ribosomal protein and cell-wall modifying genes (summarizing the Venn diagram in Table S6). In the reverse case, there were 893, 84 and 72 genes expressed more highly, respectively, but only six that were common to all conditions. Up to 58% of the genes in any one condition were in the D- or AR-sets, and this is reflected in the data presented in Figures 3 and 4.
However, in the second dimension of the PCA analysis (Figure 2a), PDC was separated from PDL and PDN, indicating that there were significant treatment-specific expression patterns characteristic of the cold signal. Indeed, there were 231 genes that were more highly expressed in PDC than the other 12 physiological states, compared with only 31 in PDL and three in PDN. In the reverse case (lower expression than the other 12 states) there were 224 genes in PDC, but only two in PDL and six in PDN. There was a maximum of only one gene in any of these selections that also appeared in the D- or AR-sets. Therefore, in addition to the common quantitative changes in gene expression described above, there are unique gene expression patterns in response to specific dormancy-relieving environmental signals, which will be considered in the following sections.
The effect of cold treatment on transcript abundance – a role for gibberellic acid (GA) metabolism in dormancy release
Ogawa et al. (2003) reported 357 GA-responsive genes, whereas Yamauchi et al. (2004) reported 313 cold-regulated genes, both in a Landsberg erecta (Ler) background. Of these two gene sets 83 genes overlapped, indicating that the cold response and GA response share some common pathways. In the present work, cold-regulated genes during dormancy relief were identified by comparing PDC with all dormant conditions (i.e. PDC expression greater than or less than PDD, PD24 h, PD48 h, PD30d, SD1 and SD2 expression). This gene set will therefore contain many genes similarly expressed in other dormancy-relieving treatments, and form part of the common quantitative changes in gene expression illustrated above with the D- and AR-sets. The cold-regulated genes identified were then compared with the data sets of Ogawa et al. (2003) and Yamauchi et al. (2004). Of the 991 genes that had higher expression upon cold treatment of Cvi seeds, 23 genes were identified as cold induced by Yamauchi et al. (2004) and 24 as GA induced by Ogawa et al. (2003). Of these, only six genes were shared by all three data sets. The expression of a set of 893 genes was lowered by cold treatment of Cvi seeds, and 29 of these genes were identified as cold reduced by Yamauchi et al. (2004), and 23 were identified as GA reduced by Ogawa et al. (2003). Of these, only 15 genes were shared by all three data sets. This implies that release from dormancy by cold treatment shares little overlap with other cold-induced and GA-regulated pathways, as investigated by Ogawa et al. (2003) and Yamauchi et al. (2004), and forms a rather unique pathway. Alternatively, the use of a different genetic background (Cvi versus Ler) and the different sampling time may have resulted in the small number of overlapping genes with our dataset.
GA biosynthesis and catabolism are clearly linked to germination (Ogawa et al., 2003), as well as inducing the completion of germination during low-temperature treatment (Yamauchi et al., 2004), and the key enzymes in GA biosynthesis are now well described (Olszewski et al., 2002). GA 20-oxidase catalyses the penultimate steps in the production of bioactive GAs and GA 3-β-dioxygenase catalyses the production of bioactive GAs from these precursors, whereas GA 2-oxidase not only deactivates the bioactive GAs but also the immediate precursors. Cadman et al. (2006) hypothesized that in Cvi the relative expression of genes encoding these enzymes indicated that a dynamic state of synthesis and catabolism maintained lower endogenous concentrations of GA in dormant states, compared with dormancy-relieved after-ripened states. They further hypothesized the reverse case for abscisic acid (ABA) synthesis and catabolism, and these genes are discussed further below. The data presented in Figure 6 shows that, in general, the expression of these genes in other dormancy-relieving environments (nitrate, cold and light) is consistent with this hypothesis. This result is supported by QRT-PCR analysis (Table S5). In the case of GA synthesis, GA20ox2 (At5 g51810) expression is greatly increased in the PDC sample compared with that in other dormancy-relieving treatments (Tables S5 and S6; Figure 6a). Two further GA 20-oxidases (At5 g07200, GA20ox3; At4 g25420, GA20ox1; Figure 6a) are also more highly expressed in PDC than samples exposed to other dormancy-relieving treatments; other members of the GA 20-oxidase gene family differ little between samples (data not shown). These higher levels of expression of GA 20-oxidase genes appear to be balanced by higher expression of a GA 2-oxidase gene (At1 g30040, GA2ox2; Figure 6a). Conversely, the pronounced decrease in transcript abundance of GA20ox2 and GA20ox3, as compared with PD24 h expression level, in PDL and PDLN is associated with a diminishing GA2ox2 expression. Yamauchi et al. (2004) have suggested that repression of the GA 2-oxidase gene family may contribute to elevating bioactive GAs in response to cold. This is consistent with the light and light plus nitrate treatment, but not for the cold treatment, in the present work. They also suggested that this repression of the GA 2-oxidase gene family is in contrast to the effect of light, where expression levels of this gene were unaffected (Ogawa et al., 2003), which is consistent with the data presented. Transcript numbers of this gene family were not significantly affected by light, nitrate or after-ripening.
The expression of one member of the GA 3-β-dioxygenase (At1 g15550; GA3ox1; Figure 6a) gene family increased in a similar fashion when exposed to all environmental signals that partially relieve dormancy (PDC, PDN and PDL), but increased further in PDLN, where dormancy had been fully relieved to initiate the completion of germination. This result is reflected in the expression of some genes encoding GA-responsive proteins such as putative GASA4 (At1 g74670; Figure 6a). The transcript levels from other members of this gene family were not altered by exposure to cold or other dormancy-relieving environments (data not shown). Yamauchi et al. (2004) showed that both red light and GA deficiency act in addition to cold to elevate the level of GA3ox1, indicating that multiple signals can be integrated by this gene, as we see here with the effect of light, nitrate, and combinations of these treatments and after-ripening (Cadman et al., 2006). In addition, mutant analysis showed that transcripts of the gene were required to increase bioactive GA levels, and increased transcript levels were shown to result in the increased production of active GAs (Yamauchi et al., 2004). Earlier work also showed a positive relationship between exposure to cold and endogenous GA concentration in Arabidopsis (Derkx et al., 1994).
Recently, Penfield et al. (2005) have shown how the bHLH transcription factor SPATULA (SPT; At4 g36930) can act as a light-stable repressor of GA3ox expression controlling seed response to cold. The related PIL5 (At2 g20180), although not required for seed dormancy, was involved in the repression of germination completion in the dark after cold treatment. The two may therefore form part of a regulatory network coupling seed germination and the expression of GA3ox genes to light and temperature signalling in the seed. Based on this work, we might have expected a pattern of expression from SPT to be the reverse of GA3ox, at least in the cold-treated seeds, but in the present work we found no effect of cold on SPT expression and sound similarity in the expression patterns of both SPT and GA3ox(Figure 6a). Although it is relevant to show this result because of the potential importance of this putative regulation mechanism for the control of dormancy, no clear conclusion can be drawn about the merits of this system in Cvi. In addition, this ecotype is known to have a limited response to GA during dormancy (Ali-Rachedi et al., 2004), and so other factors are likely to be dominant if the action of PIL5 and SPT are only mediated through their effect on GA3ox.
The effect of nitrate on transcript abundance
Nitrate transport. Nitrate proved capable of releasing different forms of dormancy in Arabidopsis Cvi Seeds, including secondary dormancy induced by warm temperatures in the dark (Figure 1, and Cadman et al., 2006). However, nitrate was not effective in releasing secondary dormancy induced by prolonged exposure to low temperature in the dark (SD2; Cadman et al., 2006). Transcript abundance of the nitrate transporter NRT1.1 (At1 g12110) was at a similar level in both PDD and DDL seeds, increased to high levels in the imbibed PD and SD1 seeds and other states susceptible to nitrate, but was lowest in the SD2 seeds that were not sensitive to nitrate (Figure 7a; Table S7). Lower expression of NRT1.1 in SD2 seeds compared with other deeply dormant states was confirmed by QRT-PCR analysis (Table S5). There was no similar pattern of expression with six putative NRT2.1 genes (At1 g08090, At1 g12940, At3 g45060, At5 g14570, At5 g60770 and At5 g60780) and dormancy status (Table S7). The expression of other nitrate transporters including NTP3 (At3 g21670) and NTL1 (At1 g69850) also showed a pattern that was not consistent with a role for nitrate uptake in dormancy release (Figure 7a; Table S7).
Nitrate reduction. We investigated the expression of genes encoding nitrate-reducing enzymes to see if any relationship with dormancy release could be found. Upon uptake, nitrate is reduced to nitrite in the cytoplasm by nitrate reductase (NaR). The Arabidopsis genome contains two genes encoding nitrate reductase, NR1 (At1 g77760) and NR2 (At1 g37130). Of the two genes, NR1 was expressed more highly in imbibed seeds, and in particular when dormancy was being relieved by cold, light or nitrate, but was lowest in SD2, the state that was not sensitive to nitrate (Cadman et al., 2006). In general, NR2 increased comparatively little upon imbibition (Figure 7b; Table S7). Nitrite may be further reduced to ammonia in the chloroplast by nitrite reductase (NiR). One putative NiR encoding gene (At2 g15620) was found in the Arabidopsis genome (Wang et al., 2000). An increase in transcript abundance was most distinct after nitrate application, and in seeds induced to complete germination (Figure 7b; Table S7). Several (putative) glutamine synthetase genes (At1 g48470, At1 g66200, At3 g17820, At3 g53180, At5 g16570, At5 g35630 and At5 g37600) did not display major differences in expression over the different physiological states (Table S7).
Oxidative pentose phosphate pathway. The reducing power for the reduction of nitrate and nitrite may be provided by NADPH, which is generated by the oxidative pentose phosphate pathway (oxPPP). In the dark, NADPH from the oxPPP reduces ferredoxin that acts as a source for electrons required for nitrate and nitrite reduction. Our expression data shows that two of the multiple genes encoding for 6-phosphogluconate dehydrogenase (EC 188.8.131.52) and ribose 5-phosphate isomerase (EC 184.108.40.206) are significantly more highly expressed in the several dormancy-relieving conditions, compared with all the imbibed dormant states (Figure 7c), whereas genes encoding all the other oxPPP enzymes were expressed at constitutive levels with very little variation over the different physiological states (Table S7). Increased gene expression and, possibly, activity of both oxPPP enzymes suggest a demand for nucleotide synthesis through ribose-5-phosphate upon the release of dormancy and the completion of germination. Apparently, this may occur with constitutive expression of glucose-6-phosphate dehydrogenase gene activity (Table S7). Transcript abundance of two genes encoding for Fd-NADP+ oxidoreductase (At1 g30510 and At4 g05390) is higher after nitrate or light treatment, but not after cold (Figure 7d, Table S7). Concomitantly, transcript abundance of genes encoding for two pentose phosphate translocators, At1 g61800 (AtGPT2) and At5 g17630, increased upon the breaking of dormancy, except for At5 g17630, in PDC seeds (Figure. 7c).
Nitric oxide metabolism. Nitric oxide (NO) is a potent dormancy-releasing agent in many species, including Arabidopsis, and has been suggested to be an endogenous regulator of dormancy (Bethke et al., 2004). In plants NO is synthesized by reduction of nitrite by NR, particularly in the absence of nitrate or under anoxia (Crawford, 2006). NO may also be synthesized from arginine by an NO synthase. As this is a possible explanation for nitrate-induced dormancy release, we compared published NO-upregulated genes with our data set. Genes from the set of 37 genes upregulated in Arabidopsis leaves treated with the NO donor sodium nitroprusside did not show any relevant variation over the different physiological conditions, apart from those that were previously identified as stress-related genes (data not shown; Cadman et al., 2006; Parani et al., 2004). In addition, expression of the gene encoding NO synthase (NOS1) was constant over all conditions. This was also the case for the guanylyl cyclase gene (GC1) that may play a role in the induction of cGMP synthesis by NO (Delledonne, 2005).
The effect of light on transcript abundance
It has been shown above that the expression of GA3ox1 (At1 g15550; Figure 6a) increased when exposed to all environmental signals that partially relieve dormancy (PDC, PDN and PDL), but was increased further in PDLN where dormancy had been fully relieved to initiate the completion of germination by exposure to light. This was also true in fully after-ripened seeds exposed to light (LIG; Cadman et al., 2006). A search was carried out for other genes that are differentially expressed in seeds where dormancy was relieved and seeds were induced to complete germination by light (PDLN and LIG, but not PDL, greater than other 24-h dormancy-breaking incubations, PDN, PDC and DL), and this revealed a set of only two genes with significantly (P < 0.05) higher expression (At3 g57040, At3 g47980; Figure 8; Table S8). There was only one gene that had significantly (P < 0.05) lower gene expression (At5 g07200; Supplemental Table S8). At5 g07200 encoding GA20ox3 shows a contrasting expression pattern to that of At1 g15550 encoding GA3ox1, as discussed above (Figure 6a).
At3 g57040, encoding response reactor 4 (ARR9), is a negative regulator of cytokinin signalling involved in cytokinin-mediated inhibition of lateral root formation (To et al., 2004). Lateral roots are not usually observed until after the completion of germination, when the primary root is sufficiently long. Expression of this gene under conditions that induce radicle protrusion is presumably required to suppress lateral root formation until root elongation has advanced sufficiently to penetrate the soil. The function of At3 g47980 is not known; sequence analysis suggests that it encodes an HPP family protein, an integral membrane protein with possible transporter function (Marchler-Bauer et al., 2005).
When the search for light-induced genes was extended to all three light treatments (LIG, PDLN and PDL), more genes were identified, some of which showed a relationship with dormancy (Table S8). A number of genes (At1 g68240, At2 g40080, At2 g46340, At3 g02910, At3 g51240, At3 g61220, At5 g08640, At5 g13930 and At5 g57785) are expressed more highly after a light treatment (PDL, PDLN and LIG) compared with other 24-h incubations (PDN, PDC and DL; Table S8). Of this set, two genes encode proteins that participate in the flavonoid pathway, namely CHALCONE SYNTHASE, alias TRANSPARENT TESTA 4 (CHS/TT4; At5 g13930) and FLAVANONE 3-HYDROXYLASE (F3H; At3 g51240). There is also a putative flavonol synthesis gene (At5 g07480) and other flavonol-related genes (13), which have either enhanced or reduced expression in response to cold. The CHS/TT4 protein has been reported to demonstrate interaction with F3H, as well as with the proteins encoded by three more genes: bZIP protein HY5 (HY5, At5 g11260); CHALCONE ISOMERASE, alias TRANSPARENT TESTA 5 (TT5, At3 g55120); and DIHYDROFLAVONOL-4-REDUCTASE (DFR, At5 g42800; Table S8) (Ang et al., 1998; Burbulis and Winkel-Shirley, 1999). These last two genes, unlike CHS/TT4 and F3H, did not show differences in gene expression; HY5 appeared to have a similar, albeit non-significant, light-regulated expression pattern as CHS/TT4 and F3H, and was described as essential for light activation of CHS/TT4 (Ang et al., 1998). Apart from acting as a flavonoid enzyme, CHS/TT4 has also been described to inhibit polar transport of auxin, a plant hormone involved in elongation, phototropism and gravitropism (Brown et al., 2001). The role of auxins in germination has not been established unequivocally (Kucera et al., 2005). Debeaujon and Koornneef (2000) described a possible role for CHS/TT4 in reducing the germination restriction via a thinner testa. As the testa in mature seeds is dead tissue, and de novo transcription in the testa is not likely during germination, any effect of testa-located CHS/TT4 is probably established during seed development. It is likely that the enhanced CHS/TT4 transcription observed here in living seed tissues is solely a light response.
Only one gene showed significantly (P < 0.05) lower expression in seeds exposed to all three light treatments, At4 g16780 encoding HAT4, a homeobox-leucine zipper protein that strongly responds to changes in R/FR (Carabelli et al., 1993). Gene expression was higher in more deeply-dormant states, as well as in PDN and PDC. Light decreased expression in both dormant seeds imbibed for short periods and after-ripened seeds. Nitrate seemed to play an ambiguous role: although nitrate alone increased expression of HAT4 in dormant seeds (PD24 h versus PDN), it induced a further decrease in the presence of light (PDL versus PDLN). Expression seemed to decrease in seeds that are further advanced towards completion of germination, and mainly regulated by light.
The efficiency of seed dormancy release in Cvi by cold, nitrate and light is determined by the length of dry after-ripening following harvest
Dormancy was released by dry after-ripening, or by exposure to cold or nitrate followed by light exposure. However, the sensitivity of seeds to cold, nitrate and light was dependant upon the length of time spent dry after-ripening following harvest. The seeds first became sensitive to nitrate, then cold and finally light. In addition, seeds had an absolute requirement for light whatever conditions they were exposed to, but dormancy in seeds was not released by light without them first being exposed to either prolonged after-ripening or a shorter period of after-ripening followed by imbibition on a nitrate solution or in the cold. However, the rate of increase in sensitivity to environmental signals that break dormancy during dry after-ripening following harvest was not fixed, and seeds produced in July 2003 changed more rapidly than those harvested in April 2005. This is consistent with the depth of dormancy not only being determined genetically, but also through the interaction of genes with the ambient environment during production (Donohue, 2005). Therefore, seeds from plants with the same genetic background are likely to follow the same pattern of dormancy loss, but this change may occur at different rates with the potential to alter ecological behaviour.
The oxidative pentose phosphate pathway is not primarily involved in the breaking of dormancy
In non-photosynthetic cells the oxPPP is a primary source of reducing power, in the form of NADPH for the assimilation of inorganic nitrogen, as well as for the biosynthesis of fatty acids and nucleotides. Moreover, the oxPPP maintains the redox potential required for protection against oxidative stress (reviewed by Kruger and von Schaewen, 2003).
Activation of the oxPPP has long been implicated in the breaking of seed dormancy (Bewley and Black, 1994; Roberts and Smith, 1977). This was mainly based on the observation that inhibitors of the cytochrome pathway of respiration were able to break dormancy, suggesting that diversion to the oxPPP was a prerequisite for dormancy relief. NADPH, generated by the oxPPP, was thought to supply the reducing power for the reduction of nitrate and nitrite, both strong dormancy-releasing agents in many species, via ferredoxin:NADP+ oxidoreductase. However, experimental verification of this hypothesis has so far yielded mixed results.
The enzyme 6-phosphogluconate dehydrogenase (At1 g64190) catalyzes one of the two steps in the oxPPP that generate NADPH. One of the multiple genes encoding for this enzyme was upregulated in dormancy-releasing conditions, and is predicted to be located in the plastid (Kruger and von Schaewen, 2003). The occurrence of multiple genes encoding for oxPPP enzymes may result from the demand for different isozymes in the cytosol and plastids. The concomitant increase in transcript abundance of genes encoding for pentose phosphate translocators suggests that the plastidic oxPPP is provided with cytosolic carbon skeletons, most likely in the form of xylulose-5-phosphate (Eicks et al., 2002). This particularly occurs under conditions of a high demand for PPP intermediates. The increase in abundance of two ferredoxin:NADP+ oxidoreductase transcripts suggests a mechanism of electron transfer to the dependent enzymes in the plastids, induced by dormancy-releasing conditions.
Our results suggest that, at the transcriptome level, the oxPPP is operational. As nitrate reduction is not essential to the breaking of dormancy, the oxPPP may be primarily directed towards the synthesis of such essential compounds as nucleotides, rather than to supply reducing power in the form of NADPH.
Nitrate is a signalling molecule in the breaking of dormancy, and acts in concert with light and cold signalling
Nitrate breaks seed dormancy in a large number of species, but little is known about the possible mechanisms involved. Nitrate transporters, nitrate reductase, the oxPPP and NO signalling have all been implied in the signal transduction path leading to dormancy breaking (Alboresi et al., 2005; Bethke et al., 2004; Hilhorst, 1990b; Roberts and Smith, 1977). Nitrate uptake by NRT1.1, but not by NRT2.1, has been hypothesized to play a role in the regulation of dormancy (Alboresi et al., 2005). Our results are consistent with this hypothesis, suggesting that nitrate uptake through NRT1.1 might form one regulatory step through which this dormancy-releasing agent acts. Our results show that one of the genes encoding for nitrate reductase, NR1, was more highly expressed in conditions that break dormancy, including cold, light and nitrate. Thus, expression of this gene was not exclusively induced by nitrate. Contrary to NR1, an increase in expression of the NiR encoding gene was observed after a treatment of nitrate with or without light, or shortly after initiation of the completion of germination (LIG), but not after light or cold treatment. This suggests that expression of this gene may depend on the presence of nitrate. If nitrate and nitrite reduction does occur in the seeds, further metabolization of nitrogen through glutamine synthetase to glutamine would be expected. However, none of the seven (putative) glutamine synthetase (GS1) genes displayed increased transcript abundance during the breaking of dormancy. Thus at the level of the transcriptome these results appear to preclude a nutritional effect of nitrate on the breaking of dormancy. Obviously, control of nitrate assimilation may also occur at a (post) translational level. However, inhibition of nitrate reductase activity by chlorate and tungstate in Sisymbrium officinale, did not affect germination on nitrate (Hilhorst and Karssen, 1989), and the nitrate reductase mutant G’4-3 of Arabidopis, possessing only 0.5% of the nitrate reductase activity of the corresponding wild type, was even less dormant than the wild type (Alboresi et al., 2005). Thus, it appears that release of dormancy by nitrate was facilitated by the NRT1.1 transporter, and was accompanied by increased expression of NR1 and of an NiR encoding gene, but not of NR2 or glutamine synthetase genes. A tentative conclusion is that increased nitrate accumulation and reduction seem to convey a signal for dormancy breaking rather than function as a nitrogen source for nutrition. However, it has been suggested that during dormancy breaking in Arabidopsis the transcriptome is anticipating the next likely step in development, e.g. initiation of growth (Cadman et al., 2006). Thus, the occurrence of increased transcript abundance of NR1 may be a preparation for nitrate assimilation during seedling growth.
Although the mechanism of nitrate action is still largely unclear, the present study is consistent with the suggestion by Ali-Rachedi et al. (2004) that nitrate can enhance ABA catabolism and inhibit ABA synthesis. These authors have shown in Arabidopsis Cvi seeds that nitrate applied to dormant seeds decreased the ABA content and prevented de novo synthesis, which occurred in the absence of nitrate in the presence of light. Intimate cross-talk between light and nitrate signalling has also been implied in the breaking of seed dormancy (Hilhorst, 1990a,b). The results presented here provide more evidence for complementary mechanisms in which each of the dormancy-relieving treatments (viz. cold, light and nitrate) influence different components of the ABA/GA hormone balance in Arabidopsis Cvi seeds. Cold appears to induce genes involved in the synthesis of precursors of the active GAs, whereas nitrate action is aimed at ABA metabolism, resulting in a decrease in ABA content; when applied in addition to other treatments to fully relieve dormancy, the action of light is associated with the production of active GAs. This view is expanded upon below.
Dormancy relief appears linked to quantitative patterns of gene expression directed by changes in the dynamic balance of ABA/GA, which has the potential to integrate environment-specific signals
The PCA analysis grouped physiological states together in a way that related to the depth of dormancy of the seeds, rather than the type of environmental conditions to which they had been exposed (Figure 2). This strongly suggests similarity in the relief from dormancy by different environmental signals through a cascade of gene expression that is universal in dormancy relief. The latter was evident also in the transcript abundance of genes in D- and AR-sets (Cadman et al., 2006), which showed opposite patterns related to the depth of dormancy. These quantitative changes took place following seed exposure to single dormancy-relieving factors (cold, light and nitrate), even though the seeds were still dormant in the absence of a second factor. When the appropriate factors were present in the correct order (e.g. nitrate plus light, and light following after-ripening) to fully relieve dormancy and complete germination, transcript abundance of the D-set was further reduced and abundance in the AR-set set was further increased (Figures 3e and 9). These data are consistent with physiological observations that environmental factors remove successive blocks to germination when experienced in the appropriate order. In addition, these data support the view that there is preparation for germination during the process of dormancy release, and before all the necessary environmental signals for dormancy release have been encountered. Thus seeds can be prepared for rapid completion of germination when the environmental signal that removes the final block to radicle protrusion (e.g. light) is encountered, indicating that conditions are suitable for the completion of germination and plant establishment.
Light has a more immediate response than other treatments, for example, transcript levels of GA20ox1, GA20ox2, GA20ox3 and GA2ox2 in Cvi indicate that there is turnover of the precursors of the active GAs when seeds are exposed to dormancy-relieving conditions (cold and nitrate), so that when seeds are subsequently exposed to light, a rapid response is possible through greatly enhanced AtGA3ox1 transcript levels leading to active GAs. However, Cvi has a limited response to gibberellins (Ali-Rachedi et al., 2004), and the control of germination completion through dormancy cycling, in a variable environment, is not regulated by GAs alone, but is more likely to be regulated through a dynamic balance of the synthesis and catabolism of both GA and ABA to drive changes in dormancy status (Cadman et al., 2006; Finch-Savage and Leubner-Metzger, 2006). Toyomasu et al. (1998) have shown that endogenous levels of ABA in lettuce seeds were substantially lowered by both light and exogenous GA3 via catabolism or conjugation. They concluded that inactivation of ABA was mediated by GAs during dormancy breaking by light.
Cadman et al. (2006) reported that the expression of genes encoding enzymes involved in the synthesis and catabolism of ABA and GA followed reverse patterns during dormancy cycling, and this is apparent in the results presented here during dormancy loss (Figure 6); for example, compare the expression patterns of genes encoding GA 3-β-dioxygenase (Figure 6a) and 9-cis-epoxycarotenoid dioxygenase (NCED; Figure 6b) genes, which are thought to be the key regulators of the synthesis of active GAs and ABA, respectively. Their expression changes, in response to a range of environments that induce and relieve dormancy, support the view that multiple (additive) signals can be integrated by these genes regulated at the level of transcription (Yamauchi et al., 2004, Cadman et al., 2006; and summarized in Figure 9). Additive effects of other environmental signals on the ABA/GA balance and sensitivity to these hormones are also known, such as smoke and light on Nicotiana attenuata (Schwachtje and Baldwin, 2004). Cadman et al. (2006) show that the changes reported here are consistent with differential expression of large numbers of transcription factors present in the D- and AR-sets (see also figures incorporated into Table S2), along with genes encoding histones involved in transcriptional regulation, which are suggestive of a complete switch in gene expression resulting from a change in chromatin structure. Many genes of the AR-set are concerned with the establishment of translation machinery, the potential for cell wall modification and reserve mobilization (Figure 4b; see also Table S3 and incorporated figure) in advance of germination completion. The present work indicates that transcription of these gene sets responds in a quantitative way to specific environmental signals, when they are presented to the seeds in the order appropriate to relieve dormancy and facilitate the completion of germination in seasonal conditions that are suitable to sustain subsequent growth. Our work also suggests that changes in the transcriptome, at least partly, anticipate the next likely stage of development, i.e. radicle extension and subsequent seedling growth. Results from Rajjou et al. (2004) have shown that chemical inhibition of transcription in non-dormant Arabidopsis seeds did not affect the eventual completion of germination, but inhibited further growth of the seedling after radicle protrusion. However, translation inhibitors effectively blocked the completion of germination. This suggests that the transcripts for the completion of germination of non-dormant seeds are pre-formed during their development, and then translated to enable progress of germination all the way to completion. However, the present work shows that the relief of dormancy is accompanied by transcription of genes associated with the completion of germination (AR-set) including those encoding for proteins involved in translation machinery. Thus, we hypothesize that an important molecular event of dormancy relief is establishing the capacity for the completion of germination. Initiation of the processes resulting in germination completion and radicle protrusion may then be regulated at a post-transcriptional level.
Plant material, seed production and germination conditions
Plants of the A. thaliana accession Cvi (accession number N8580) were grown in media containing peat/sand/vermiculite from seeds donated by Maarten Koornneef (Wageningen University, the Netherlands). For the first 4 weeks, four replicate blocks of 24 seedlings were grown in 2.5-cm2 square plastic modules at 20°C under daylight fluorescent tubes (16-h photoperiod). Seedlings were then transplanted to 5-cm2 square plastic modules and placed in a glasshouse that was vented and heated to control temperature, and had supplementary lighting to maintain light levels and photoperiod. When the seeds were mature, the plants were no longer watered. Seeds were harvested when siliques were dry on the plant. Seeds were bulked from each replicate block of plants. The four replicate seed lots were then dried to an eRH of 33% at 20°C (moisture content 7%, dry-weight basis), placed in separate sealed moisture-proof containers and stored at 20°C to dry after-ripen. Initial primary dormant samples (3 × 25 mg per replicate) and further samples were taken following a range of treatments described in Table 1. Physiological status was determined in sub-samples of seeds from each replicate in germination tests described below. The very low numbers of seeds that completed germination in treatments during preparation of the samples were removed, and the remainder of each seed sample was immediately frozen and stored at –80°C until used for RNA extraction.
Completion of germination (radicle protrusion) was recorded on a minimum of 50 seeds from each of at least four replicates maintained at 20 ± 0.5°C on two layers of filter paper (3MM Chr.; Whatman International Ltd., http://www.whatman.com) in clear polystyrene boxes. The filter paper was kept moist throughout treatment with either water or a 10 mm KNO3 solution. The boxes were either placed in the dark or light (daylight fluorescent tubes; 60 μmol m−2 sec−1; PPFD). Seeds kept in the dark were set up under green light, and where observations of percentage germination were necessary, these were also made under green light. Germination was monitored for a minimum of 28 days to record all seeds capable of its completion. In some experiments, seeds on moist filter paper were initially exposed to 3 ± 0.5°C (cold stratification) for different periods, or were exposed to 4 h of red light after 20 h of imbibition, before radicle protrusion was observed at 20 ± 0.5°C. Seeds were not sterilized to avoid influencing their dormancy status; however, fungal infection was minimal in the seedlot used.
Total RNA isolation from seeds
Following treatment, seeds were stored in 2-ml tubes at –80°C. Total RNA was extracted using a hot borate method (modified from Wan and Wilkins, 1994). Seeds were homogenized with single steel cone balls (Retsch GmbH, http://www.retsch.com) in 2-ml tubes, using a Retsch Mixer Mill MM200 for three times 2 min at 30 Hz, and submersing tubes and tube adapter blocks in liquid nitrogen between each 2-min session. Ground seeds were kept submersed in liquid nitrogen prior to RNA extraction. Ground seeds were re-suspended in 700 μl XT buffer [(0.2 m sodium tetraborate decahydrate, 30 mm EGTA, 1% w/v SDS, 1% w/v sodium deoxycholate, pH 9.0) containing 14 mg polyvinylpyrilidone and 1.1 mg dithiothreitol], which had been heated to 80°C. The seed suspension was added to 0.35 mg proteinase K in a 2-ml round-bottomed tube and incubated at 42°C for 90 min. KCl (2 m, 56 μl) was then added, and samples incubated on ice for a further 2–3 h. Samples were centrifuged at 16 000 g at 4°C, after which the supernatant was removed to a new tube and its volume determined. A one third volume of 8 m LiCl (to give a final 2 m LiCl concentration) was added and samples incubated at 4°C overnight. Samples were centrifuged, and pellets washed by vortexing in 2 m LiCl until the supernatant remained colorless. Pellets were re-suspended in 300 μl diethylpyrocarbonate (DEPC)-treated water, and RNA precipitated by the addition of 30 μl 3 m sodium acetate and 990 μl ice-cold 100% ethanol, and incubated at –80°C for a minimum of 3 h. RNA was pelleted by centrifugation at 16 000g for 20 min at 4°C, followed by a wash with 70% ethanol. Partially dried pellets were re-suspended in 100 μl of DEPC-treated water and purified with RNeasy columns (Qiagen, http://www.qiagen.com). All kits were used according to the manufacturer’s instructions. RNA concentration was determined using a NanoDrop ND-1000 spectrophotometer (Nanodrop® Technologies Inc., http://www.nanodrop.com), and RNA quality was checked using formaldehyde gel electrophoresis.
Total RNA (20 μg per biological replicate) was submitted to Nottingham Arabidopsis Stock Centre (NASC, http://arabidopsis.info) for synthesis of labelled cRNA, hybridization to the Affymetrix ATH1 Arabidopsis GeneChip (Affymetrix, http://www.affymetrix.com), chip scanning and data accumulation (using affymetrix microarray suite version 5.0 software) (Craigon et al., 2004). Four independent biological replicates were submitted for each of six physiological states (PDD, DDL, PDN, PDC, PDL and PDLN, described in Table 1), and the resulting expression data sets are available at http://affy.arabidopsis.info/narrays/experimentbrowse.pl. Previously unpublished microarray data and re-analysed data displayed on this website from a further seven physiological states (PD24 h, PD48 h, PD30d, SD1, SD2, DL and LIG) described in Cadman et al. (2006) have also been used in the present work.
Microarray data handling and analysis
Data were normalized in bioconductor (Gentleman et al., 2004) using the VSN algorithm (Huber et al., 2003). The algorithm transforms raw data from each array with an affine transformation, and then transforms the data with a generalized logarithm to stabilize the variance. PCA was performed in genemaths (version 2.01; Applied Maths, http://www.applied-maths.com). PCA was based on the expression patterns of all genes over either nine or 13 different physiological states described in the text.
The normalized data were subjected to analysis of variance using genstat for windows (version 8.1; VSN International, http://www.vsni.co.uk). Variance estimates were pooled across the different genes. Comparisons of gene transcript levels between pairs of physiological states were made on the transformed scale using the Benjamini–Hochberg method (Benjamini and Hochberg, 1995) to control the false discovery rate to 5%. For the pairwise comparisons made, the difference in transformed transcript levels required for significance (P < 0.05) varied from 0.64 to 0.85. To study differences in gene expression related to different physiological states, genes were identified that had a significant (P < 0.05) difference between individual states or between all states in a group, and were compared with all those in another group of physiological states to identify characteristic gene sets.
The expression of selected genes was quantified by QRT-PCR. Seed total RNA (2 μg) from four biological replicates per dormancy stage were treated with 2 U RNase-free DNase (Promega, http://www.promega.com ). cDNA was synthesized using TaqMan® Reverse Transcription Reagents (Applied Biosystems, http://www.appliedbiosystems.com) according to the manufacturer’s instructions. Primers for real-time PCR were generated using the Primer3 web interface (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi, Rozen and Skaletsky, 2000) with sequence data from TAIR. Gene-specific primer sequences are given in Table S9. Genes chosen for validation were normalized to 18S rRNA. The absence of genomic DNA was confirmed by analysis of dissociation curves and agarose gel electrophoresis of the PCR products, where applicable. PCR reactions were performed in 384-well plates using a SYBR® Green PCR Master Mix (Applied Biosystems) and an ABI Prism® 7900HT Sequence Detection System (Applied Biosystems) to monitor cDNA amplification. Reactions were performed in triplicate for each biological replicate (four biological replicates for each of the seven dormancy stages) in a mixture containing 1 μl of first-strand cDNA in a volume of 15 μl (corresponding to approximately 20 ng cDNA). The standard thermal profile used was: 2 min at 50°C, 10 min at 95°C, followed by 40 cycles of 15 s at 95°C and 60 sec at 60°C. Average threshold cycle (Ct) and relative quantities were calculated using sds software v2.1 (Applied Biosystems) and used for subsequent analysis.
We thank K. C. Dent for technical assistance with germination experiments, Z. Paniwynk for QRT-PCR, Jan Kodde for PCA analysis, and the UK Department for Environment Food and Rural Affairs (Defra) for funding.