Light activates the degradation of PIL5 protein to promote seed germination through gibberellin in Arabidopsis

Authors


*(fax +82 42 869 2610; e-mail gchoi@kaist.ac.kr).

Summary

Angiosperm seeds integrate various environmental signals, such as water availability and light conditions, to make a proper decision to germinate. Once the optimal conditions are sensed, gibberellin (GA) is synthesized, triggering germination. Among environmental signals, light conditions are perceived by phytochromes. However, it is not well understood how phytochromes regulate GA biosynthesis. Here we investigated whether phytochromes regulate GA biosynthesis through PIL5, a phytochrome-interacting bHLH protein, in Arabidopsis. We found that pil5 seed germination was inhibited by paclobutrazol, the ga1 mutation was epistatic to the pil5 mutation, and the inhibitory effect of PIL5 overexpression on seed germination could be rescued by exogenous GA, collectively indicating that PIL5 regulates seed germination negatively through GA. Expression analysis revealed that PIL5 repressed the expression of GA biosynthetic genes (GA3ox1 and GA3ox2), and activated the expression of a GA catabolic gene (GA2ox) in both PHYA- and PHYB-dependent germination assays. Consistent with these gene-expression patterns, the amount of bioactive GA was higher in the pil5 mutant and lower in the PIL5 overexpression line. Lastly, we showed that red and far-red light signals trigger PIL5 protein degradation through the 26S proteasome, thus releasing the inhibition of bioactive GA biosynthesis by PIL5. Taken together, our data indicate that phytochromes promote seed germination by degrading PIL5, which leads to increased GA biosynthesis and decreased GA degradation.

Introduction

As photosynthetic organisms, plants must carefully monitor external light conditions and adjust their growth and development accordingly. In angiosperms, at least three photoreceptor systems – phototropins (PHOTs); cryptochromes (CRYs); and phytochromes (PHYs) – are responsible for monitoring light conditions and responding by making adjustments to various physiological and developmental processes (Fankhauser and Staiger, 2002). Among these, phytochromes perceive red and far-red lights and regulate various processes, including seed germination, seedling development, chloroplast development, shade avoidance and flowering (Chory et al., 1996; Neff et al., 2000; Sullivan and Deng, 2003).

In Arabidopsis, phytochromes consist of five members that can be grouped into the type I phytochrome (PHYA), which functions as a far-red light receptor; and the type II phytochromes (PHYB, PHYC, PHYD, and PHYE), which function as red light receptors (Quail, 1998). Regardless of their spectral specificities, both types of phytochrome regulate similar physiological processes by triggering transcriptional cascades that ultimately alter the expression of 10–30% of the entire transcriptome (Ma et al., 2001; Tepperman et al., 2001). Various transcription factors are known to be centrally involved in the transcriptional cascades of light signalling, including COG1 and OBP3 (encoding DOF family members); HY5 and HYH (encoding bZIP family members); LAF1 (encoding a MYB family member); and HFR1, PIL1, PIL5, PIL6, PIF3 and PIF4 (encoding bHLH family members) (Ballesteros et al., 2001; Chattopadhyay et al., 1998; Fairchild et al., 2000; Fujimori et al., 2004; Holm et al., 2002; Huq and Quail, 2002; Huq et al., 2004; Kim et al., 2003; Monte et al., 2004; Ni et al., 1998; Oh et al., 2004; Park et al., 2003; Salter et al., 2003; Ward et al., 2005). Phytochromes initiate transcriptional cascades by modulating the activities of these transcription factors at the transcriptional or post-translational levels (Bauer et al., 2004; Duek et al., 2004; Fairchild et al., 2000; Holm et al., 2002; Huq and Quail, 2002; Jang et al., 2005; Oyama et al., 1997; Park et al., 2003, 2004; Salter et al., 2003; Seo et al., 2003; Yamashino et al., 2003; Yang et al., 2005).

Different signalling components regulate different light responses. The best studied of these is the inhibition of hypocotyl elongation by phytochromes, in which a large proportion of the light signalling components interact in varying manners. Some signalling components regulate both PHYA- and PHYB-mediated inhibition of hypocotyl elongation (e.g. HY5); some regulate only PHYB-mediated inhibition of hypocotyl elongation (e.g. PIF3); while others regulate only PHYA-mediated inhibition of hypocotyl elongation (e.g. HFR1; Fairchild et al., 2000; Fankhauser and Chory, 2000; Kim et al., 2003; Koornneef et al., 1980; Soh et al., 2000). The exact molecular pathway that leads to the inhibition of hypocotyl elongation is not clear, but microarray analysis suggests that this process involves expressional changes of various genes related to the cell-wall metabolism (Ma et al., 2001). However, while the hypocotyl elongation process is relatively well understood, only a few signalling components have been identified in other important light responses, such as seed germination.

Seed germination is regulated by various factors, including abscisic acid (ABA), brassinosteroids (BR), ethylene and gibberellin (GA; Chiwocha et al., 2005; Debeaujon and Koornneef, 2000; Kepczynski and Kepczynska, 1997; Koornneef and van der Veen, 1980; Koornneef et al., 2002; Leubner-Metzger, 2001; Steber and McCourt, 2001). Among these, ABA and GA play antagonistic roles. During seed maturation, ABA levels increase and seed dormancy is established (Karssen et al., 1983). When the dormant seed is transferred to conditions favourable for germination, the level of ABA decreases and de novo GA biosynthesis commences, disrupting dormancy and triggering germination (Ogawa et al., 2003). ABA biosynthetic mutants such as aba1 and aba2 display reduced seed dormancy, while GA biosynthetic mutants such as ga1 are unable to germinate even under favourable conditions (Karssen et al., 1983; Koornneef and van der Veen, 1980; Marin et al., 1996). The importance of GA during seed germination was further shown in various GA-signalling mutants. A loss-of-function mutation in RGL2, a negative regulator of GA responses, allowed plants to germinate even in the absence of de novo GA biosynthesis, as did a mutation in SPINDLY, a Ser/Thr O-linked N-acetyl glucosamine (O-GlcNAc) transferase that also functions as a negative regulator of GA signalling (Jacobsen and Olszewski, 1993; Jacobsen et al., 1996; Lee et al., 2002; Tyler et al., 2004). Light-independent germination of a ga1 rgl2 rga gai quadruple mutant further suggests that GAI and RGA, homologues of RGL2, also play roles in seed germination (Cao et al., 2005). Collectively, these previous results indicate that ABA and GA biosynthesis are critical for seed dormancy and seed germination respectively.

Various external factors, such as light, also play critical roles in regulating seed germination. The promotion of seed germination by light was noted as early as the 19th century, and reversible regulation of lettuce seed germination by red and far-red light is reported by Borthwick et al. (1952). Later, studies in Arabidopsis mutants revealed that this reversible regulation of seed germination by red and far-red light was determined by phytochromes (Hennig et al., 2002; Shinomura et al., 1994). A direct connection between phytochrome signalling and de novo GA biosynthesis during seed germination was demonstrated by inhibition of light-induced seed germination in the presence of a GA biosynthesis inhibitor; by the insensitivity of the ga1 mutant to light induction; by the epistasis of the ga1 mutation to the phyB mutation (Peng and Harberd, 1997); and by the direct determination of increased GA levels after light induction (Hilhorst and Karssen, 1988; Koornneef and van der Veen, 1980; Ogawa et al., 2003). The increased de novo GA biosynthesis during light-induced germination of Arabidopsis seeds is due to increased expression of GA biosynthetic enzymes such as GA 3β-hydroxylase (Yamaguchi et al., 1998). It seems likely that some of the phytochrome-interacting proteins transduce light signals, leading to activation of GA biosynthetic gene transcription during seed germination.

We showed previously that PIF3-Like 5 (PIL5/PIF1/bHLH015), a phytochrome-interacting bHLH protein, is a key negative regulator of seed germination, and that light promotes seed germination partly by inhibiting the function of PIL5 (Oh et al., 2004). Here we demonstrate that PIL5 mediates seed germination by simultaneously regulating the expression of GA biosynthetic and catabolic genes, and also show that light inhibits PIL5 function by activating the phytochrome-mediated degradation of the PIL5 protein.

Results

PIL5 regulates seed germination through gibberellin

Our previous work showed that PIL5 is a key negative regulator in PHY-mediated promotion of seed germination (Oh et al., 2004). As phytochromes are known to promote seed germination by activating de novo GA biosynthesis, we examined whether PIL5 negatively regulates seed germination by repressing de novo GA biosynthesis (Yamaguchi et al., 1998). We first tested the effects of paclobutrazol, an inhibitor of GA biosynthesis, on the ability of the pil5 mutant to germinate irrespective of light conditions.

For the PHYB-dependent germination assay, we irradiated seeds with a far-red light pulse and transferred them directly to the dark with or without a second irradiation with a red light pulse. The far-red light pulse inactivated PHYB and other type II phytochromes by converting them to the Pr form, while the subsequent red light pulse-activated PHYB and other type II phytochromes by converting them to the Pfr form. As PHYB is the major photoreceptor responsible for promoting seed germination under these experimental conditions (Shinomura et al., 1994), the seeds fail to germinate when PHYB is either mutated or inactivated by a far-red pulse, but germinate when PHYB is activated by a red light pulse. As shown in Figure 1(a), wild-type Arabidopsis Columbia (Col-0) germinated only when PHYB was activated by the red light pulse, while the pil5 mutant germinated irrespective of red or far-red light irradiation. However, treatment with paclobutrazol blocked germination of both wild-type and pil5 mutant seeds, even under continuous white light irradiation. These results indicate that de novo GA biosynthesis is necessary for germination of both wild-type and pil5 mutant seeds under PHYB-dependent germination assay conditions.

Figure 1.

 Paclobutrazol inhibits germination of pil5 mutant seeds.
(a) Germination percentages of Col-0 and pil5-1 mutant seeds under PHYB-dependent germination conditions either in the presence (PAC) or absence (N) of paclobutrazol. Upper diagrams depict light treatment schemes for the experiments. Rp, a red-light pulse (20 μmol m−2 sec−1) for 5 min after a far-red light pulse (3.2 μmol m−2 sec−1) for 5 min. FRp, a far-red light pulse (3.2 μmol m−2 sec−1) for 5 min. WLc, continuous white light (80–100 μmol m−2 sec−1).
(b) Germination percentages of the phyB-9 and pil5-1/phyB-9 mutants under PHYB-dependent germination conditions in the presence (PAC) or absence (N) of paclobutrazol. Upper diagrams depict light treatment schemes for the experiments. No, imbibition without far-red light irradiation. FRi, imbibition followed by far-red light irradiation (3.2 μmol m−2 sec−1) for 4 h. Error bars indicate standard deviations.

Arabidopsis seed germination is also promoted by PHYA if the phytochrome is allowed to accumulate during a prolonged imbibition (Oh et al., 2004; Shinomura et al., 1996). To test whether de novo GA biosynthesis is also required for the role of PIL5 in the PHYA-dependent germination, we irradiated seeds with a far-red pulse to inactivate the type II phytochromes, then imbibed the seeds for 48 h to allow accumulation of PHYA. After imbibition, seeds were transferred to darkness either directly or after a second irradiation with far-red light. To exclude the effect of PHYB, we used a phyB mutant and a pil5 phyB double mutant for the experiment. As shown in Figure 1(b), the phyB mutant did not germinate if transferred directly to darkness, but germinated if the imbibed seeds were irradiated with far-red light. In contrast, the pil5 phyB double mutants germinated partially even if transferred directly to darkness, and the rate increased further by the irradiation of far-red light. None of the tested seeds germinated in the presence of paclobutrazol. These results collectively suggest that de novo GA biosynthesis is also required for PHYA-dependent germination.

To further prove that PIL5 regulates seed germination through GA, we generated a pil5 ga1 double mutant and examined the rate of seed germination. GA1 encodes ent-copalyl diphosphate synthase, which converts geranylgeranyl diphosphate to ent-copalyl diphosphate during GA biosynthesis. The ga1 mutant is devoid of de novo GA biosynthesis, and seed germination does not occur in this mutant under all light conditions unless exogenous GA is supplied (Koornneef and van der Veen, 1980). Our germination assay showed that both the ga1 mutant and the pil5 ga1 double mutant failed to germinate under all light conditions, while the pil5 mutant germinated irrespective of light conditions (Figure 2). These results indicate that the ga1 mutation is epistatic to the pil5 mutation for seed germination in the absence of exogenous GA, suggesting that GA biosynthesis is a target for the action of PIL5.

Figure 2.

 The ga1 mutation is epistatic to the pil5 mutation in seed germination.
Germination percentages were examined in the ga1, pil5-1 and pil5-1/ga1 mutants under PHYB-dependent germination conditions. Upper diagrams, abbreviations as in Figure 1.

Our observation that the ga1 mutation is epistatic to the pil5 mutation, coupled with our finding that paclobutrazol inhibits germination of the pil5 mutant, suggested that PIL5 inhibits seed germination by inhibiting GA biosynthesis. To test this hypothesis, we examined whether application of exogenous GA could overcome the inhibitory effect of PIL5 overexpression on seed germination by treating wild-type and various mutant seeds with GA and determining the germination percentages.

We first determined the amount of GA required to promote seed germination in the ga1 mutant. We used GA4, a bioactive form of GA, for a gibberellin, and determined the germination percentages in the PHYB-dependent germination assay when seeds were treated with GA concentrations ranging from 0 to 10 μm. As shown in Figure 3(a), over 10 μm GA was sufficient to induce near 100% seed germination in the ga1 mutant. To ensure a sufficient GA supply, 10 μm GA was used for the following experiments unless otherwise indicated.

Figure 3.

 Germination percentages retain light dependency in the presence of exogenous GA.
(a) Determination of GA concentration needed for the rescue of germination defects in the ga1 mutant under PHYB-dependent germination assay conditions.
(b) Germination percentages of Col-0, pil5-1, phyB-9 and pil5-1/phyB-9 seeds under PHYB-dependent germination conditions in the presence of 10 μm GA (GA4) or the absence of GA (N). Upper diagrams, abbreviations as in Figure 1. Error bars, standard deviations.
(c) Light-dependent germination percentages of ga1 and pil5/ga1 mutant seeds in the presence of exogenous GA. Light treatment schemes and notations are as described in (b).

As shown in Figure 3(b), wild-type seeds germinated following a pulse with red light, but failed to germinate with far-red light alone. When exogenous GA was supplied, all wild-type seeds germinated even if only a far-red light pulse was applied. Consistent with the notion that PHYB is the major photoreceptor under these germination conditions, the untreated phyB mutant failed to germinate under all light conditions. In the presence of exogenous GA, the phyB mutant seeds germinated irrespective of light conditions, but the germination percentage was around 40% that of the GA-treated wild-type seeds. The lower germination percentage of the phyB mutant was not due to an intrinsic defect in germination, as 100% of phyB mutant seeds germinated under continuous light (data not shown). GA treatment also substituted for red light in the pil5 phyB double mutant. The pil5 mutant germinated irrespective of light conditions, and the pil5 phyB double mutant germinated partially in the absence of exogenous GA. When supplied with exogenous GA, the pil5 phyB double mutant seeds germinated near 100%, irrespective of light conditions (Figure 3b). The lower germination percentage of far-red treated pil5 phyB double mutant than that of far-red treated pil5 mutant is probably due to the presence of other components and the functional role of PHYB, either in dry seeds or in maturing seeds, as discussed previously (Oh et al., 2004). Consistent with this, the lower expression levels of GA biosynthetic genes were observed in the pil5 phyB double mutant compared with the pil5 single mutant (Figure S2c).

The lower germination percentage of the phyB mutant compared with wild type in the presence of the same amount of exogenous GA suggests that light may regulate seed germination through additional processes other than de novo GA biosynthesis. To investigate this possibility, we examined whether the germination percentage of the ga1 mutant showed light dependency in the presence of exogenous GA. If de novo GA biosynthesis was the sole light-regulated process, the germination percentage of GA-treated ga1 mutants should not show any light dependency, because GA cannot be synthesized endogenously in the ga1 mutant. However, as shown in Figure 3(c), the germination percentage of the ga1 mutant was light-dependent. Mutant seeds supplemented with 1 μm GA4 showed nearly 95% germination when irradiated with red light, but only about 40% germination when irradiated with far-red light. Treatment with higher concentrations of GA reduced this light dependency, but the trend seemed to indicate that light regulates seed germination not only through de novo GA biosynthesis, but also through other processes. Unlike the ga1 single mutant, the pil5 ga1 double mutant showed no light dependency of germination, suggesting that PIL5 plays an important role in the additional processes other than GA biosynthesis (Figure 3c).

We then examined whether exogenous GA could overcome the inhibitory effect of PIL5 overexpression on seed germination. As shown in Figure 4(a), the PIL5 overexpression lines [PIL5OX1 and PIL5OX3 (Myc-tagged PIL5)] failed to germinate even after red light irradiation, but GA treatment increased their germination percentages in a concentration-dependent manner. In addition, the rates of GA-induced germination in the PIL5OX lines were inversely related to the expression levels of PIL5, and red light was more effective than far-red light in promoting seed germination in the presence of exogenous GA (Figure 4a). In PIL5OX3, the germination percentage was nearly 100% when red light was irradiated in the presence of 10 μm GA4, but 70% when far-red light was irradiated with the same GA treatment. Collectively, these results indicate that PIL5 inhibits seed germination not only through de novo GA biosynthesis, but also through other processes.

Figure 4.

 Exogenous GA overcomes the inhibitory effect of PIL5 overexpression on PHYB-dependent seed germination.
(a) Germination percentages of PIL5OX1 and PIL5OX3 seeds were examined under PHYB-dependent germination conditions by varying exogenously supplied GA. Upper diagrams, abbreviations as in Figure 1. Error bars, standard deviations. RT-PCR data show expression levels of PIL5 in the PIL5Oxs seeds. Expression of ubiquitin (UBQ) was used as a loading control.
(b) Additive regulation of seed germination by light and exogenous GA in PIL5OX1 and PIL5OX3 seeds in the presence of 10 μm GA (GA4) or the absence of GA (N). After the irradiation of a far-red pulse, seeds were irradiated by red light (20 μmol m−2 sec−1) for various durations (0–12 h).

As we previously reported that the PIL5OX lines required a longer irradiation time for germination, we next examined whether GA treatment altered the required irradiation time for germination (Oh et al., 2004). We tested the germination percentages of PIL5OX seeds in the presence or absence of GA coupled with red light irradiations of various durations. As shown in Figure 4(b), 5 min red light irradiation was sufficient to fully induce germination in wild-type seeds, while much longer irradiation times were required for the PIL5OXs. Wild-type seeds germinated even in the absence of red light irradiation if exogenous GA was supplied. In the case of PIL5OX3, GA treatment reduced the required irradiation time to achieve 100% germination from 720 to 5 min, indicating that the negative role of PIL5 on seed germination is overcome additively by light and GA.

As PIL5 is also a negative regulator of PHYA-dependent seed germination, we investigated whether GA treatment could overcome the inhibitory role of PIL5 on PHYA-dependent seed germination by determining the germination percentages of wild-type and various mutant seeds in the presence or absence of exogenous GA. In the absence of GA, the germination percentages of wild-type seeds increased in response to longer far-red light irradiation times, while the phyA mutant failed to germinate regardless of irradiation time (Figure 5). In the case of the PIL5OX lines, increased far-red light irradiation times increased the germination percentages especially in the weaker line (PIL5OX3). In the presence of GA, both wild-type and phyA mutant seeds germinated well, even in the absence of far-red light irradiation (Figure 5). In the case of PIL5OX seeds, the germination percentage increased further if GA was supplied. Taken together with the inhibition of PHYA-dependent seed germination by paclobutrazol, these results indicate that PIL5 regulates seed germination negatively through GA also under PHYA-dependent seed germination conditions.

Figure 5.

 Exogenous GA overcomes the inhibitory effect of PIL5 overexpression on PHYA-dependent seed germination.
Germination percentages of Col-0, pil5-1, phyA-211, pil5-1/phyA-211, PIL5OX1 and PIL5OX3 seeds under PHYA-dependent germination conditions in the presence of 10 μm GA (GA4) or in the absence of GA (N). After irradiation of a far-red light pulse, seeds were imbibed for 48 h, then irradiated by far-red light (3.2 μmol m−2 sec−1) for 0, 4 or 12 h. Error bars, standard deviations.

PIL5 regulates both GA synthesis and GA degradation

PHY-mediated light signalling induces de novo GA biosynthesis by increasing the expression of two GA 3β-hydroxylases that convert biologically inactive GA9 to biologically active GA4 (Yamaguchi et al., 1998). As PIL5 is a phytochrome-interacting protein that negatively regulates seed germination through GA, we investigated the expression of GA-biosynthesis genes in both the pil5 mutant and PIL5OX seeds (Oh et al., 2004). For the experiment, various mutant seeds were irradiated with a far-red pulse with or without a subsequent red light pulse and incubated in the dark for 12 h before sampling. This experimental condition is identical to the PHYB-dependent germination assay condition, except that seeds were sampled for expression analysis 12 h after irradiation.

We first examined the expression of two representative marker genes (EXP1 and CP1, encoding expansin 1 and cystein proteinase 1 respectively) that are induced by GA during seed germination (Ogawa et al., 2003; Yamauchi et al., 2004). The wild-type seeds showed induced marker gene expression when PHYB was activated by a red light pulse (inductive condition), but not when PHYB was inactivated by a far-red light pulse (non-inductive condition) (Figure 6). The pil5 mutant seeds, which germinated irrespective of light conditions, showed increased expression of two marker genes irrespective of light conditions. The PIL5OX seeds, which failed to germinate under all light conditions, showed lower levels of EXP1 and CP1 irrespective of light conditions. These results indicated that the expression patterns of the two GA-inducible marker genes are consistent with the germination patterns of the various mutants.

Figure 6.

 PIL5 regulates GA biosynthesis by repressing the expression of GA 3β-hydroxylase genes (GA3ox1 and GA3ox2) and by activating the expression of a GA 2-oxidase gene (GA2ox2) under PHYB-dependent germination conditions (for quantitive RT-PCR data see Figure S2).
(a) Expression patterns of GA3ox1, GA3ox2, GA2ox2, EXP1 and CP1 in Col-0, pil5-1 and PIL5OX1 seeds under PHYB-dependent germination conditions. Upper diagrams, abbreviations as in Figure 1. Expression of S18 was used as a loading control. Germination percentages indicated below figure.
(b) Light-dependent expression of GA2ox2 in ga1 and pil5/ga1 mutant seeds. Light treatment scheme and notations are as described in (a).
(c) Expression patterns of GA3ox1, GA3ox2 and GA2ox2 in Col-0, pil5-1, PIL5OX1, phyB-9 and pil5-1/phyB-9 seeds.

We next determined the expression of two GA 3β-hydroxylase genes (GA3ox1 and GA3ox2). As shown in Figure 6, both GA3ox1 and GA3ox2 were expressed in wild-type seeds following irradiation with red light but not far-red light. In the pil5 mutant, GA3ox1 and GA3ox2 expression was constitutively high, irrespective of light conditions. In contrast, no GA3ox1 and GA3ox2 expression was detected in the PIL5OX mutants, irrespective of light conditions. Overall, the expression patterns of GA3ox1 and GA3ox2 were very similar to those of the two GA-inducible marker genes, collectively suggesting that PIL5 negatively regulates seed germination, at least partly by repressing the expression of GA biosynthetic genes.

Our germination analyses indicated that both light and PIL5 regulate seed germination through additional processes other than GA biosynthesis. One candidate process is the degradation of biologically active GA. During seed germination, the level of a bioactive GA shows an initial rapid increase and reaches a subsequent plateau concomitant with the increase of an inactive GA, suggesting that both synthesis and degradation of biologically active GA occur during seed germination (Ogawa et al., 2003). Thus it is possible that PIL5 inhibits seed germination not only by repressing the expression of GA biosynthetic genes, but also by activating the expression of GA catabolic genes.

To investigate this possibility, we examined the GA2ox genes, which encode the GA 2-oxidases responsible for converting biologically active GA4 and/or its precursors to biologically inactive forms. Of the eight GA2ox genes present in the Arabidopsis genome, we focused on the expression of GA2ox2, because it was expressed at high levels and its expression was light-dependent in germinating seeds (data not shown). As shown in Figure 6(a), GA2ox2 was highly expressed in seeds that failed to germinate: far-red light-irradiated wild-type seeds, and both red and far-red light-irradiated PIL5OX seeds. In contrast, GA2ox2 was expressed at low levels in seeds that germinated well: red light-irradiated wild-type seeds, and both red and far-red light-irradiated pil5 mutant seeds. Taken together, these results indicate that PIL5 negatively regulates the biosynthesis of bioactive GA in germinating seeds, not only by repressing GA3ox1 and GA3ox2, but also by activating the expression of GA2ox2.

Although GA2ox2 was expressed at high levels in PIL5OX mutant seeds, irrespective of light, the expression level of GA2ox2 was higher in far-red light-irradiated PIL5OX seeds than in red light-irradiated PIL5OX seeds (Figure 6a). This light-dependent expression of GA2ox2 in the PIL5OX seeds might partly explain the light-dependent germination percentage of PIL5OX in the presence of exogenous GA (Figure 4a). Similarly, light-dependent germination was also observed in the ga1 mutant in the presence of exogenous GA (Figure 3c), although the pil5 ga1 double mutant did not show any light dependency. To investigate whether light-dependent expression of GA2ox2 is also partly responsible for the light-dependent germination percentages of these mutants, we determined the expression of GA2ox2 in the ga1 mutant and the pil5 ga1 double mutant. As shown in Figure 6(b), GA2ox2 expression was higher in the far-red light-irradiated ga1 mutant seeds than in the red light-irradiated ga1 mutant seeds. In the pil5 ga1 double mutant, GA2ox2 expression was slightly higher in the far-red light-irradiated seeds, but the overall expression level was lower than that of far-red light-irradiated ga1 seeds. As GA 2-oxidase degrades biologically active GAs, these results suggest that the light-dependent expression of GA2ox2 partly explains the light-dependent germination percentages of PIL5OXs and ga1 mutant seeds in the presence of exogenous GA (Thomas et al., 1999).

To investigate the relationship between PIL5 and PHYB in seed germination, we determined the expression levels of GA3ox1, GA3ox2 and GA2ox2 in the phyB mutant and in the pil5 phyB double mutant. Under PHYB-dependent germination assay conditions, the phyB and PIL5OX seeds failed to germinate, while the pil5 phyB double mutant germinated only partially (Figure 3b). Consistent with these findings, seeds that germinated under the assay conditions (wild type and pil5 mutant) showed increased GA3ox1 and GA3ox2 expression levels, but decreased expression levels of GA2ox2 (Figure 6c), while seeds that failed to germinate (PIL5OX and phyB mutant) showed decreased GA3ox1 and GA3ox2 expression and increased GA2ox2 expression. The pil5 phyB double mutant showed intermediate expression levels of both GA biosynthetic and GA catabolic genes, which is consistent with its intermediate germination percentage. These data indicate that both PHYB and PIL5 regulate bioactive GA levels during seed germination via mediating both synthesis and degradation of GA. As PIL5 is a PHYB-interacting, downstream-signalling component, these data further suggest that PHYB promotes seed germination partly by inhibiting the negative role of PIL5 in the biosynthesis of bioactive GA.

As PIL5 is also a negative regulator of seed germination under PHYA-dependent germination conditions, we investigated whether PIL5 regulates the expression levels of GA3ox1, GA3ox2 and GA2ox2 under these conditions. For the experiment, far-red-irradiated seeds were imbibed for 48 h to allow the accumulation of PHYA. After imbibition, far-red light was applied again to activate PHYA, and the seeds were incubated for 8 h in the dark before sampling. Under the PHYA-dependent germination assay conditions, wild-type, pil5 mutant and pil5 phyA double mutant seeds germinated following far-red irradiation, but PIL5OX and phyA mutant seeds did not. Consistent with our findings under PHYB-dependent germination conditions, the GA-inducible marker genes (EXP1 and CP1) were highly expressed in seeds that germinated under PHYA-dependent germination conditions, but were expressed at low levels in seeds that did not germinate (Figure 7a). The seeds that germinated showed higher expression levels of GA3ox1 and GA3ox2, but lower expression levels of GA2ox2. In contrast, the seeds that failed to germinate showed lower expression levels of GA3ox1 and GA3ox2, but higher expression levels of GA2ox2 (Figure 7a,b). These data indicate that PIL5 regulates both GA synthesis and degradation under PHYA-dependent germination conditions. Our data suggest that PHYA promotes seed germination partly by inhibiting the negative role of PIL5 on the biosynthesis of bioactive GA, being similar to the relationship observed between PIL5 and PHYB. In addition, the higher expression of GA biosynthetic genes in the pil5 single mutant compared with the pil5 phyA double mutant after inductive FR irradiation further suggests the presence of other factors that mediate PHYA signalling to GA biosynthesis.

Figure 7.

 PIL5 regulates GA biosynthesis by repressing the expression of GA 3β-hydroxylase genes (GA3ox1 and GA3ox2) and activating the expression of a GA 2-oxidase gene (GA2ox2) under PHYA-dependent germination conditions (for quantitive RT–PCR data see Figure S3).
(a) Expression patterns of GA3ox1, GA3ox2, GA2ox2, EXP1 and CP1 in Col-0, pil5-1 and PIL5OX1 seeds under PHYA-dependent germination conditions. After irradiation of a far-red light pulse (3.2 μmol m−2 sec−1), seeds were imbibed for 48 h, then irradiated with far-red light (3.2 μmol m−2 sec−1) for 4 h (FR) and transferred to the dark, or left unirradiated (N) and transferred to the dark. Expression of S18 was used as a loading control. Germination percentages indicated below figure.
(b) Expression patterns of GA3ox1, GA3ox2 and GA2ox2 in Col-0, pil5-1, PIL5OX1, phyA-211 and pil5-1/phyA-211 seeds.

Our expression analysis of GA biosynthetic genes suggested that the amount of bioactive GA was likely to be increased in the pil5 mutant and decreased in PIL5OX plants. To confirm this, we irradiated various mutant seeds with far-red pulses with or without subsequent red light pulses, and quantified the GA4 contents. As shown in Figure 8, the GA4 level was twofold higher in red light-treated wild-type seeds versus far-red light-treated wild-type seeds. In the pil5 mutant, GA4 levels were similar to those found in red-light treated wild-type seeds, regardless of light conditions. In PIL5OX plants, the GA4 level was slightly lower than that found in far-red-treated wild-type seeds, regardless of light conditions. The higher contents of bioactive GA in pil5 mutants and the lower contents of bioactive GA in the PIL5OX plants are consistent with the expression patterns of GA biosynthetic and catabolic genes (Figure 6). Taken together, our data indicate that PIL5 negatively regulates seed germination by lowering bioactive GA levels.

Figure 8.

 Endogenous GA levels
The pil5 mutant contains increased levels of bioactive GA (GA4), whereas PIL5OX plants have lower levels of GA4, regardless of light conditions. Upper diagrams, abbreviations as in Figure 1. Error bar, standard deviation. Experiments were repeated with similar results using independently incubated seed sets.

Our expression analysis of GA2ox2 gene suggested that GA catabolic activity should be lower in the pil5 mutant and higher in the PIL5OX. As GA9 and GA4 are catabolized to GA51 and GA34, respectively, by GA2ox, we determined levels of two GA catabolic products (GA34 and GA51) (Table S1). Consistent with the lower expression of GA2ox2 gene in red light-treated wild-type seeds, levels of two GA catabolic products were higher in far-red light-treated seeds than those in red light-treated wild-type seeds. In the pil5 mutant, the GA catabolic products were similar to those of red light-treated wild-type seeds regardless of light conditions. In contrast, levels of GA catabolic products were very low in the PIL5OX. As the level of catabolic products depends on substrate availability, we determined the ratio of GA catabolic products and their substrates (GA9 and GA4) and found that the ratio is lower in the pil5 mutant and higher in PIL5OX (Table S1). Taken together, these results indicated that GA catabolic activity is lower in the pil5 mutant and higher in PIL5OX.

PIL5 protein is degraded by both red and far-red light irradiation

Previous reports indicated that PIF3, another phytochrome-interacting bHLH protein, is degraded through the 26S proteasome by red and far-red lights (Bauer et al., 2004; Park et al., 2004). The degradation of PIF3 by light partly explains how phytochromes inhibit the function of PIF3 in certain light responses. PIL5 is very similar to PIF3 in terms of its amino acid sequence, its ability to interact with phytochromes, and its functional relationship with phytochromes (Huq et al., 2004; Oh et al., 2004). Thus it is tempting to speculate that phytochromes inhibit PIL5 by activating the degradation of PIL5 during seed germination. To determine whether PIL5 protein is degraded by red and far-red light, we generated transgenic Arabidopsis plants expressing a Myc-tag fused PIL5 under the control of the CaMV 35S promoter. The Myc-tagged PIL5 was found to be functional, and one Myc-tagged PIL5 overexpression line (PIL5OX3) was chosen for use in the following Western blot experiments.

As the primary role of PIL5 is regulating seed germination, we examined whether red or far-red light regulates the stability of PIL5 protein in germinating seeds. Under PHYB-dependent germination conditions, PIL5 protein was degraded following irradiation of seeds with a pulse of red light, but not far-red light (Figure 9a). However, after imbibition both red and far-red light effectively triggered degradation of PIL5 protein (Figure 9b). Similar degradation of PIF1/PIL5 protein by light has recently been reported in seedlings (Shen et al., 2005). These results suggest that the degradation of PIL5 protein, like PIF3 protein, is also activated by phytochromes in germinating seeds.

Figure 9.

 Phytochromes activate the degradation of PIL5 protein.
(a) Degradation of PIL5 protein by red or far-red light under PHYB-dependent germination conditions. PIL5 indicates Myc-tagged PIL5 protein detected by an anti-Myc antibody. Tubulin (TUB) was detected by anti-tubulin antibody as a loading control.
(b) Degradation of PIL5 protein by red or far-red light under PHYA-dependent germination conditions.
(c) Degradation of PIL5 protein in PIL5OX3 and PIL5OX3/phyA mutant seeds by far-red light under PHYA-dependent germination conditions.
(d) Degradation of PIL5 protein in PIL5OX3 and PIL5OX3/phyB mutant seeds by red light under PHYB-dependent germination conditions.
(e) Inhibition of PIL5 degradation by MG132 in far-red light-irradiated seedlings.
(f) Inhibition of PIL5 degradation by MG132 in red light-irradiated seedlings. Five DAG dark-grown seedlings were treated by either DMSO or MG132 (200 μm), irradiated by red light pulse for 5 min, and incubated in the dark for 30 min until harvesting.

To provide additional evidence that phytochromes are photoreceptors for the activation of PIL5 degradation, we determined the degradation of PIL5 protein in the phyA and phyB mutants. As shown in Figure 9(c), PIL5 was degraded by far-red light under PHYA-dependent germination conditions in the wild-type background but not in the phyA mutant background (Figure 9c), indicating that PHYA is a main photoreceptor for the activation of PIL5 degradation in response to far-red light. In the phyB mutant, the degradation pattern of PIL5 was slightly more complicated. As shown in Figure 9(d), the PIL5 protein was initially degraded more or less equally in both PIL5OX3 and PIL5OX phyB mutants following irradiation. At 24 h after irradiation, PIL5 protein was still degraded well in PIL5OX, but virtually no degradation occurred in the PIL5OX phyB mutant. These results suggested that the degradation of PIL5 is activated by both PHYB and other type II phytochromes in response to red light, and the functional PHYB is required for the prolonged degradation of PIL5 protein after a red-light pulse.

As PIF3 is degraded through the 26S proteasome, we examined the degradation of PIL5 in the presence of a 26S proteasome inhibitor (MG132) in seedlings. As shown in Figures 9(e) and 8(f), the degradation of PIL5 by red and far-red lights was greatly reduced in the presence of MG132, suggesting that PIL5 is also degraded through the 26S proteasome both under red and far-red lights. MG132 had a much weaker effect on PIL5 degradation in seeds, when compared with seedlings, possibly due to lower permeation by this chemical (Figure S1). Taken together, our results indicate that phytochromes act as photoreceptors to activate the degradation of PIL5 protein through the 26S proteasome.

Discussion

We report here that PIL5, a phytochrome-interacting bHLH protein, negatively regulates seed germination by repressing two GA 3β-hydroxylases genes (GA3ox1 and GA3ox2) and activating a GA 2-oxidase gene (GA2ox2) at the transcriptional level. Repression of GA biosynthetic genes and activation of GA catabolic genes by PIL5 leads to increased levels of bioactive GA in the pil5 mutant and decreased levels of bioactive GA in PIL5OX seeds, regardless of light conditions. Our functional relationship studies further suggest that phytochromes promote seed germination by inhibiting PIL5, at least partly via activation of PIL5 protein degradation through the 26S proteasome. Based on these results, we propose a photogermination model in which PIL5 plays a negative role in the biosynthesis of bioactive GA, and PIL5 degradation by phytochromes is activated in response to inductive light signals (Figure 10).

Figure 10.

 Proposed model for the phytochrome-mediated promotion of seed germination in Arabidopsis.
PIL5, a phytochrome-interacting bHLH protein, inhibits seed germination by repressing the transcription of GA 3β-hydroxylase genes and activating the transcription of a GA 2-oxidase gene. Under inductive light conditions, phytochromes activate the degradation of PIL5 through the 26S proteasome, thus increasing the biosynthesis of bioactive GAs. Bioactive GAs promote seed germination by activating the degradation of GAI, RGA and RGL2 through SCFSLY1,SNE. For simplicity, other factors implicated in mediating light signalling to GA biosynthesis and the degradation of GA9 to GA51 are not indicated in the model.

GA synthesis and degradation are regulated during phytochrome-mediated promotion of seed germination

De novo GA biosynthesis plays a critical role in the phytochrome-mediated promotion of seed germination, as shown by the ability of a GA biosynthetic inhibitor (paclobutrazol) to inhibit seed germination, and the fact that treatment with exogenous GA could initiate germination in the absence of light stimulation (Hilhorst and Karssen, 1988; Yang et al., 1995). The observation that GA biosynthetic mutants such as ga1 failed to germinate even after inductive light treatment further supports the role of GA in the phytochrome-mediated promotion of seed germination (Koornneef and van der Veen, 1980). Previous studies have shown that GA biosynthesis increases during seed germination due to the transcriptional activation of GA 3β-hydroxylase genes, which encode enzymes required for the rate-limiting step of GA biosynthesis (Yamaguchi et al., 1998).

The present work reveals that light regulates the biosynthesis of bioactive GA not only by activating the transcription of GA biosynthetic genes, but also by suppressing the transcription of at least one GA catabolic gene. Expression of two GA 3β-hydroxylase genes (GA3ox1 and GA3ox2) increased dramatically on the perception of inductive light signals, whereas expression of a GA 2-oxidase gene (GA2ox2) decreased under inductive conditions and increased under non-inductive conditions. The net effect of this reciprocal regulation of biosynthetic and catabolic genes is probably a sharper increase or decrease of bioactive GA levels, compared with regulation of either alone. This reciprocal regulation was found to be under the control of phytochromes. Under PHYB-dependent germination assay conditions, induction of the GA 3β-hydroxylase genes and repression of the GA 2-oxidase gene were abolished in the phyB mutant. Similarly, under PHYA-dependent germination assay conditions, this reciprocal regulation was abolished in the phyA mutant. These data indicate that both PHYA and PHYB promote the biosynthesis of bioactive GA during seed germination by activating expression of GA 3β-hydroxylase genes and repressing the expression of at least one GA 2-oxidase gene.

We further found that PIL5 plays a critical role for this reciprocal regulation of GA synthesis and catabolism during phytochrome-mediated promotion of seed germination. We previously showed that PIL5 is a phytochrome-interacting bHLH protein that preferentially interacts with the Pfr forms of both PHYA and PHYB. Our characterizations of mutant and overexpression lines revealed that PIL5 is a key negative regulator for both PHYA- and PHYB-dependent germination processes (Oh et al., 2004). In the present work we found that expression of GA3ox1 and GA3ox2 is increased in the pil5 mutant and decreased in PIL5OX seeds, indicating that PIL5 represses expression of these GA biosynthetic genes. Similar higher expression of GA3ox1 and GA3ox2 in the pil5 mutant have been reported recently by Penfield et al. (2005). In contrast, GA2ox2 is expressed at lower levels in the pil5 mutant and at higher levels in PIL5OX seeds, indicating that PIL5 activates expression of this GA catabolic gene. The expression patterns of GA3ox1, GA3ox2 and GA2ox2 in the phyB, pil5 phyB, phyA and pil5 phyA mutants further indicated that PIL5 reciprocally regulates these genes during phytochrome-mediated promotion of seed germination. The consequences of GA biosynthetic and catabolic gene-expression patterns are reflected in the levels of bioactive GA in germinating seeds. Consistent with the higher expression of GA biosynthetic genes and the lower expression of GA catabolic genes, the levels of bioactive GA were increased and decreased in the pil5 and PIL5OX plants, respectively, regardless of light conditions. Taken together, these data indicate that PIL5 is a key negative component that links phytochrome signals to the expression of GA biosynthetic and catabolic genes during seed germination.

Although our finding that light signals trigger induction of GA 3β-hydroxylase genes is consistent with the previous report, the relationship between phytochromes and these genes observed in the present study differs slightly from the earlier findings (Yamaguchi et al., 1998). The previous study showed that PHYB controlled GA3ox2 expression, while GA3ox1 expression was controlled by other phytochromes. However, our data clearly indicate that these genes were not expressed in the phyB mutant under the PHYB-dependent germination assay conditions, showing that the expression of both GA3ox1 and GA3ox2 are both controlled by PHYB (Figure S2c). We further showed that PHYA also regulates the expression of both GA3ox1 and GA3ox2 under PHYA-dependent germination assay conditions. The basis of these discrepancies is not immediately clear, but may be related to the use of different Arabidopsis ecotypes. The previous reports were based on analyses of the Landsberg erecta (Ler) ecotype, while we utilized the Col-0 ecotype in the present work. Differences have been reported in light signalling and seed germination between the Ler and Col ecotypes, suggesting the possibility that light-signalling processes such as expression of GA3ox1 might differ between these ecotypes (Borevitz et al., 2002; van der Schaar et al., 1997; Wolyn et al., 2004; Yanovsky et al., 1997). It will be interesting to determine if any of the quantitative trait loci between Ler and Col are related to the expression of GA biosynthetic genes, such as GA3ox1. Alternatively, the use of slightly different protocols in the PHYB-dependent germination assays might account for the differences between our results and those of other groups.

It is not yet known how PIL5 reciprocally regulates the transcription of GA biosynthetic and catabolic genes. Many transcription factors have been shown to activate one set of genes while repressing another, although the molecular mechanisms for these reciprocal regulations vary (Holm et al., 2002; Monte et al., 2004). For example, PIL5 may activate transcription of two different downstream transcription factors, one that activates GA2ox2 transcription and one that represses expression of GA3ox1 and GA3ox2. Alternatively, PIL5 may regulate these genes directly, with its activity as an activator or a repressor depending on the context created by promoters or binding cofactors. Further analysis will be required to elucidate the reciprocal regulation of these genes by PIL5, leading to greater understanding of light signalling in general.

Phytochromes inhibit PIL5 by promoting its protein degradation

Protein degradation plays a key role in PHY-mediated light responses. Mutations in the COP/DET/FUS genes cause constitutive photomorphogenic phenotypes, even in the dark, and analysis of these genes indicated that the majority of them encode proteins involved in protein degradation (Wei and Deng, 1996). Among these, COP1 encodes a RING-finger type ubiquitin E3 ligase; COP10 encodes a ubiquitin E2 conjugation enzyme; and most of the others encode subunits of a regulatory complex similar to the 19S lid complex of the 26S proteasome (von Arnim and Deng, 1993; Suzuki et al., 2002; Wei and Deng, 1996). In the light-signalling process, HFR1, HY5 and LAF1, three positive light-signalling components, are ubiquitinated by COP1 in the dark, and the ubiquitinated components are subsequently degraded by 26S proteasome (Duek et al., 2004; Jang et al., 2005; Osterlund et al., 2000; Seo et al., 2003; Yang et al., 2005). Photoreceptors such as phytochromes or cryptochromes inhibit the E3 ligase activity of COP1 partly by excluding COP1 from the nucleus (von Arnim and Deng, 1994). Thus photoreceptors potentiate light responses that are mediated by HFR1, HY5 and LAF1 by inhibiting the degradation of these factors by COP1.

Phytochromes also modulate the stability of directly interacting components. PIF3, a bHLH protein, was the first identified phytochrome-interacting protein (Ni et al., 1998). The physiological function of PIF3 is complex, depending on the light responses examined. For hypocotyl elongation, it acts as a negative component in PHYB- but not PHYA-mediated inhibition of hypocotyl elongation. For anthocyanin synthesis, PIF3 acts as a positive component in PHY-mediated accumulation of anthocyanin; in terms of chloroplast development, PIF3 plays a positive role during the transition from dark to light (Kim et al., 2003; Monte et al., 2004). Expression analysis indicated that the level of PIF3 transcript is not significantly affected by light, whereas PIF3 protein is rapidly degraded by red or far-red light (Bauer et al., 2004; Park et al., 2004). This degradation is activated by PHYA in response to far-red light, and by PHYB and PHYD in response to red light. In both cases, PIF3 is degraded by ubiquitination followed by degradation via the 26S proteasome.

We show here that phytochromes also activate the degradation of PIL5, another phytochrome-interacting bHLH protein, through the 26S proteasome. Similarly to PIF3, PIL5 regulates various light responses, including seed germination, hypocotyl elongation, shoot gravitropism and protochlorophyllide biosynthesis (Huq et al., 2004; Oh et al., 2004). Genetically, phytochromes inhibit the function of PIL5 in those PHY-mediated light responses. Thus our data, showing the degradation of PIL5 by red or far-red light, demonstrate a mechanism by which phytochromes inhibit the function of PIL5 to initiate those light responses.

Our work, and the previous analyses, show that phytochromes are able to regulate the degradation of various light-signalling components through at least two different pathways. One pathway acts through COP1, which leads to degradation of largely positive components such as HFR1, HY5 and LAF1 in the dark (Duek et al., 2004; Jang et al., 2005; Osterlund et al., 2000; Seo et al., 2003; Yang et al., 2005). The regulation of COP1 activity by phytochromes is partly due to the exclusion of COP1 from the nucleus under light conditions (von Arnim and Deng, 1994). However, HFR1 accumulates more quickly under far-red light irradiation than COP1 exclusion proceeds, indicating that this exclusion may not be the sole regulatory mechanism (Yang et al., 2005). The other pathway, acting through a yet-unknown mechanism, degrades directly interacting, largely negative components such as PIF3 and PIL5 through the 26S proteasome under light conditions (Bauer et al., 2004; Park et al., 2004). Further investigations will be required to determine how phytochromes activate the degradation of their directly interacting proteins in the light.

Experimental procedures

Plant materials and growth conditions

Arabidopsis thaliana plants were grown in a growth room with a 16-h light/8-h dark cycle at 22–24°C for general growth and seed harvesting. The ga1 mutant was obtained from the Arabidopsis Stock Center (Salk_109115; Alonso et al., 2003). For generation of the PIL5OX3 transgenic line, the full-length PIL5 cDNA was amplified with specific primers (5′-AGA GTG ATC AAA AAT GCA TCA TTT TGT CCC TGA C-3′ and 5′-AGA GTG ATC ACC ACC TGT TGT GTG GTT TCC-3′), cloned into a pBI-HTM vector for expression of a fusion protein bearing both His and Myc tags, and transformed into wild-type plants (Col-0) by Agrobacterium-mediated transformation. Three independent homozygous lines (PIL5OX3, PIL5OX4 and PIL5OX5) were established. As they all showed phenotypes consistent with the previously established PIL5OX1 and PIL5OX2 lines expressing the native PIL5 gene, we used PIL5OX3 seeds for our experiments where indicated. All plants used in the experiments (pil5-1, phyA-211, phyB-9, ga1, PIL5OX1 and PIL5OX3) were Col-0 ecotype background. Different mutants used in each figure were grown at the same time, in the same growth room, in the same tray. Seeds were stored at 22°C in white paper bags.

Germination assay

For the PHYB-dependent germination assay, triplicate sets of 60 seeds for each mutant were surface sterilized and plated on aqueous agar medium (0.6% phytoagar, pH 5.7). At 1 h after the start of seed sterilization, the plated seeds were irradiated with red (20 μmol m−2 sec−1) or far-red (3.2 μmol m−2 sec−1) light for 5 min. After 5 days’ incubation in the dark, germinated seeds were determined by the emergence of radicles. For the PHYA-dependent germination assay, the plated seeds were first irradiated with far-red light (3.2 μmol m−2 sec−1) for 5 min, incubated in the dark for 48 h, then further irradiated with far-red light (3.2 μmol m−2 sec−1) for various times. Four days later, the germinated seeds were counted. To determine the effect of paclobutrazol on seed germination, triplicate sets of 60 seeds for each mutant were spotted on agar plates containing 100 μm paclobutrazol, incubated under PHYA- and PHYB-dependent germination assay conditions, and counted for germinated seeds. For the germination assay in the presence of exogenous GA, triplicate sets of 60 seeds for each mutant were spotted on agar plates containing various concentrations of GA4 (0–10 μm).

Gene-expression analysis

A total of 50 μl seeds were plated on wet filter paper and incubated under PHYA- or PHYB-dependent germination assay conditions, with the exception that the plates were incubated for 12 h before seeds were harvested for expression analysis. Total RNA was extracted from the seeds using an Ambion RNA extraction kit according to the manufacturer's guidelines (Ambion, Austin, TX, USA).

The following primers were used for RT–PCR analysis of gene expression: GA3ox1 (5′-CCG AAG GTT TCA CCA TCA CT-3′ and 5′-CCC CAA AGG AAT GCT ACA GA-3′); GA3ox2 (5′-TAG ATC GCA TCC CAT TCA CA-3′ and 5′- TGG ATA ACT GCT TGG GTT CC-3′); GA2ox2 (5′-TGA CTC GGT TAG AGC AGG AG-3′ and 5′-CTT GAA CCT CCC GTT AGT CA-3′); EXP1 (5′-TCA CAT GTC AAT GGT TAC GC-3′ and 5′-TGT CCC CAG TTT CTT GAC AT-3′); CP1 (5′-TGA TGA GTC CAT CAT CAA CG-3′ and 5′-TGT AGG ATT TGT CGC AGT CA-3′); S18 (5′-CCA GCG ATC GTT TAT TGC TT-3′ and 5′-AGT CTT TCC TCT GCG ACC AG-3′); PIL5 (5′-GGG GAT TTT AAT AAC GGT-3′ and 5′-GAG ATT ATG AAC TTC AGC AGC ACG-3′); UBQ (5′-GAT CTT TGC CGG AAA ACA ATT GGA GGA TGG T-3′ and 5′-CGA CTT GTC ATT AGA AAG AAA GAG ATA ACA GG-3′).

Analysis of endogenous GAs

Various mutant seeds were surface sterilized and plated on aqueous agar medium (0.6% phytoagar, pH 5.7). At 1 h after the start of seed sterilization, the plated seeds were irradiated with red (20 μmol m−2 sec−1) or far-red (3.2 μmol m−2 sec−1) light for 5 min. After 26 h incubation in the dark, seeds were collected for GA quantification. Quantitative analysis of GAs was carried out by GC-selected ion monitoring (SIM) using 2H-labelled GAs as internal standards, as described previously (Gawronska et al., 1995). Briefly, a pre-purified ethyl acetate-soluble fraction containing GAs was subjected to HPLC purification using a reverse-phase column (Capcell Pak C18 SG120; Shiseido Fine Chemicals, Tokyo, Japan). When necessary, GA-containing fractions after reverse-phase HPLC were further purified through another round of HPLC using an ion-exchange column (Senshu Pak N[CH3]2, 1151-N; Senshu Scientific, Tokyo, Japan) to ensure the removal of impurities for reliable quantification. The purified fractions were subjected to GC–SIM analysis using a mass spectrometer (Automass Sun; JEOL, Tokyo, Japan) equipped with a GC (6890N; Agilent Technologies, Palo Alto, CA, USA) and a capillary column (DB-1; Agilent Technologies) after derivatization. Authentic GA samples and 2H-labelled internal standards were purchased from Professor Lewis Mander (Australian National University, Canberra).

Protein extraction and protein gel blotting

A total of 50 μl seeds were plated on wet filter paper, incubated under PHYA- or PHYB-dependent germination assay conditions, and harvested at the times indicated. The harvested seeds were ground in liquid nitrogen and solubilized in buffer (100 mm NaH2PO4, 10 mm Tris–Cl, 8 m urea pH 8.0) by vigorous vortexing. The debris was cleared by centrifugation at 20 000 g for 10 min at 4°C. Western blot analyses were performed as described by Park et al. (2004) to determine the PIL5 protein levels. The Myc tag was used to assess PIL5 protein levels with an anti-Myc antibody. To determine the effects of MG132, five DAG dark-grown seedlings were pre-treated by either DMSO or MG132 (200 μm) and incubated under various light conditions until harvesting.

Acknowledgements

We thank Ms Masayo Sekimoto and Mr Atushi Hanada (RIKEN Plant Science Center) for GA analysis. This work was supported in part by grants from the KRF (C00044), KOSEF (R21-2003-000-10002-0), the Plant Diversity Research Center of the 21st Frontier Research Program (PF0330508-00), and the Plant Metabolism Research Center funded by KOSEF.

Ancillary