The Arabidopsis SPA1 gene is required for circadian clock function and photoperiodic flowering

Authors


(fax +1 212 327 8327; e-mail chua@rockefeller.edu).

Summary

Arabidopsis phytochrome A (phyA) regulates not only seed germination and seedling de-etiolation but also circadian rhythms and flowering time in adult plants. The SUPPRESSOR OF PHYA-105 (SPA1) acts as a negative regulator of phyA-mediated de-etiolation of young seedlings, but its roles in adult plants have not yet been described. Here, we show that SPA1 is involved in regulating circadian rhythms and flowering time in plants. Under constant light, the abundance of SPA1 protein exhibited circadian regulation, whereas under constant darkness, SPA1 protein levels remained unchanged. These results indicate that the SPA1 protein is controlled by the circadian clock and light signals. In addition, the spa1-3 mutation slightly shortened the circadian period of CCA1, TOC1/PRR1 and SPA1 transcript accumulation under constant light. Phenotypic analysis showed that the spa1-3 mutant flowers early under short-day (SD) but not long-day (LD) conditions. Consistent with this finding, transcripts encoding flowering locus T (FT), which promotes flowering, increased in spa1-3 under only SD conditions, although the CONSTANS (CO) transcript level was not affected under either SD nor LD conditions. Our results indicate that SPA1 not only negatively controls phyA-mediated signaling in seedlings, but also regulates circadian rhythms and flowering time in plants.

Introduction

In plants, the transition from the vegetative to the reproductive phase is influenced by environmental factors such as the photoperiod. Regulation of flowering initiation in response to day length is mediated by the interaction between external light signals and the circadian clock system (Hayama and Coupland, 2004). To date, several mutants affected in flowering time have been characterized in Arabidopsis thaliana that flower earlier when grown under long-day (LD) than under short-day (SD) conditions.

Mutations in CONSTANS (CO; Putterill et al., 1995), GIGANTEA (GI; Fowler et al., 1999; Park et al., 1999), FLOWERING LOCUS T (FT; Kardailsky et al., 1999; Kobayashi et al., 1999) and CRYPTOCHROME 2 (CRY2; Guo et al., 1998) delay flowering under LD conditions, but do not affect flowering time under SD conditions. Therefore, these mutations reduce sensitivity to photoperiod. FT is an early target of CO that links the control of flowering time to the circadian clock (Samach et al., 2000; Suarez-Lopez et al., 2001), whereas GI and CRY2 act upstream of CO (Guo et al., 1998; Suarez-Lopez et al., 2001). In contrast to these mutants, null mutations of EARLY FLOWERING 3 (ELF3; Hicks et al., 1996; Zagotta et al., 1996) cause the loss of photoperiodic regulation, resulting in flowering at the same time under LD and SD conditions. Many of these genes are associated with the photoperiodic pathway and encode either photoreceptors or proteins associated with the circadian clock.

The circadian clock system may be defined as the signaling system that is composed of three general parts: an input pathway, an output pathway and a central oscillator (Dunlap, 1999). The central oscillator of the circadian clock generates an oscillation with a period of 24 h, and regulates the expression of genes through an output pathway. Recently, it has been shown that the central oscillator is composed of an autoregulatory transcriptional and translational negative feedback loop. In Arabidopsis, CIRCADIAN CLOCK ASSOCIATED 1 (CCA1; Wang and Tobin, 1998), LATE ELONGATED HYPOCOTYL (LHY; Schaffer et al., 1998) and TIMING OF CAB EXPRESSION1 [TOC1; Strayer et al., 2000; also known as ARABIDOPSIS PSEUDO-RESPONSE REGULATOR1 (PRR1; Matsushika et al., 2000)] are putative components of the negative feedback loop (Alabadi et al., 2001).

Light signals perceived by specific photoreceptors, such as the red/far-red (FR) light receptors known as phytochromes (phyA to phyE) and the blue/UV-A light receptors known as cryptochromes (cry1 and cry2), reach the central oscillator of the circadian system through an input pathway and synchronize its phase to the actual periodic environmental changes. Indeed, mutations in phytochromes or cryptochromes cause the circadian rhythms of CAB2 promoter activity to oscillate at a pace slower than that of wild-type under various light conditions (Somers et al., 1998a). Furthermore, it was also shown that the expression of CCA1 and LHY is under the control of phytochrome signals (Monte et al., 2004).

Phytochrome A is also known to regulate early developmental processes such as seed germination and de-etiolation by perceiving continuous FR, very low fluences of white light and blue light of low fluence rates (Casal et al., 2003). To date, a large number of mutants involved in this step of phyA signaling have been isolated and characterized. The suppressor of phyA-105 (spa1) mutant was isolated as a negative regulator of phyA-mediated light responses in Arabidopsis seedlings (Hoecker et al., 1998). SPA1 encodes a protein that contains a protein kinase-like domain, a coiled-coil domain and a WD40 domain (Hoecker et al., 1999), and the SPA1 transcript markedly increases when dark-grown seedlings are exposed to FR or red light (Hoecker et al., 1999).

SPA1 is a member of a small gene family that encodes SPA1 and three SPA1-like proteins (SPA2, SPA3 and SPA4; Laubinger and Hoecker, 2003), which function redundantly in the regulation of photomorphogenesis (Laubinger and Hoecker, 2003; Laubinger et al., 2004). The SPA1 protein has been shown to modulate the in vitro ubiquitination of LAF1 and HY5 transcription factors by CONSTITUTIVE PHOTOMORPHOGENIC 1 (COP1) E3 ligase (Saijo et al., 2003; Seo et al., 2003). Furthermore, recent studies have suggested that SPA1 represses light signaling by modulating COP1-dependent degradation of HFR1 transcription factor at a post-translational level (Duek et al., 2004; Jang et al., 2005; Yang et al., 2005a,b). Thus, SPA1 and COP1 are likely to interact in the targeted degradation of certain phyA signaling intermediates.

Recently, a quadruple mutant of spa1/spa2/spa3/spa4 has been shown to exhibit an extreme dwarf phenotype in adult plants, although each single mutant does not exhibit any apparent morphological phenotype (Laubinger et al., 2004). Therefore, SPA genes regulate adult plant development and there is redundancy in the SPA gene family. Moreover, SPA1 transcript accumulation displays rhythmicity in a circadian manner (Harmer et al., 2000). However, our understanding about roles of SPA1 in plant development is still poor. Here, we describe the involvement of SPA1 in regulating the circadian clock and flowering time. SPA1 protein is regulated in a circadian manner and phenotypic analysis demonstrates that the spa1 mutation shortens the period of the circadian rhythm and reduces sensitivity to photoperiod. Therefore, clock-regulated SPA1 contributes to maintaining the circadian rhythm and the determination of flowering time in Arabidopsis plants.

Results

SPA1 transcripts are controlled by both the circadian clock and light signals

To confirm that SPA1 transcripts are under circadian clock control, we performed Northern blot hybridization with an SPA1-specific probe. Wild-type (WT) and spa1-3 mutant plants were grown in a chamber under day–night cycles (12 h light, 12 h dark), released into constant light (LL) or constant dark (DD) and harvested at 3 h intervals. Figure 1(a) shows that the SPA1 transcript level in WT plants under LL conditions peaked around the subjective end of night and decreased towards the subjective dusk, consistent with previous results using microarray analysis (Harmer et al., 2000). By contrast, the SPA1 transcript level decreased markedly in plants under DD conditions, making it difficult to determine whether it still displays rhythmicity (Figure 1b). These results indicate that the SPA1 transcript is regulated by both the circadian clock and light signals.

Figure 1.

 Diurnal and circadian cycling of SPA1 transcript.
(a) Circadian cycling of SPA1 transcripts under LL conditions. (b) Diurnal and circadian cycling of SPA1 transcripts under DD conditions. WT and spa1-3 mutant plants grown in day-night cycles (12 h light, 12 h dark) were transferred to (a) continuous light (LL: time 0) or (b) continuous darkness (DD: time 24) and harvested at 3 h intervals. Total RNAs were analyzed by Northern blot hybridizations using probes specific for SPA1 and UBQ10. UBQ10 levels were used as loading controls. SPA1 and UBQ10 transcript levels were quantified. The values of SPA1 transcript in WT at (a) the subjective end of night (23 h) and (b) dawn (2 h) were taken as 1.0. Each experiment was performed twice with similar results.

Interestingly, the SPA1 transcript abundance in the spa1-3 mutant exhibited slightly shorter periods compared to WT, and the transcript level was slightly reduced (Figure 1a). The spa1-3 mutation results in the introduction of a premature stop codon into the SPA1 coding sequence (Hoecker et al., 1999). Therefore, it was not clear whether the reduced abundance of SPA1 transcript is due to the loss of SPA1 function or to the instability of the spa1-3 mRNA caused by the presence of the premature termination codon. Nevertheless, it seems likely that the SPA1 protein controls the pace of the circadian oscillator, which in turn regulates the circadian period of SPA1 transcription.

SPA1 protein displays rhythmicity in a circadian manner

To examine whether SPA1 is clock-regulated at the post-translational level, we prepared rabbit polyclonal antibodies against SPA1 as described in Experimental procedures. Figure 2(a) shows that in protein extracts prepared from WT seedlings grown under continuous white light, the antibodies recognized a protein band of about 120 kDa corresponding to the estimated molecular mass of SPA1. As this band was absent from spa1-3 mutant seedlings, we conclude that the antibodies recognized SPA1. The spa1-3 mutation is predicted to generate a truncated protein of 413 amino acids (Hoecker et al., 1999). Such a putative truncated form of SPA1 was not seen in the spa1-3 mutant (Figure 2a), suggesting that the mutant is null or expresses extremely low levels of the truncated mutant SPA1 protein.

Figure 2.

 Diurnal and circadian cycling of SPA1 protein.
(a) Specificity of anti-SPA1 antibodies. Protein extracts from 2-week-old WT and spa1-3 seedlings grown in continuous white light were analyzed by Western blots using anti-SPA1 antibodies. The asterisk indicates a major position of the cross-reacting band. (b) Circadian cycling of SPA1 protein under LL conditions. (c) Diurnal and circadian cycling of SPA1 protein under DD conditions. WT and spa1-3 mutant plants grown in day–night cycles (12 h light, 12 h dark) were transferred to (b) continuous light (LL: time 0) or (c) continuous darkness (DD: time 24) and harvested at 3 h intervals. Protein extracts from WT plants were analyzed by Western blots using anti-SPA1 antibodies. Tubulin levels were used as loading controls. SPA1 and tubulin protein levels were quantified and the values at (b) 23 h and (c) 2 h were taken as 1.0. Each experiment was performed twice with similar results.

To analyze the accumulation pattern of SPA1 in plants under LL or DD conditions, total protein extract from WT plants grown under LL or DD conditions was analyzed by Western blots. We detected a slight diurnal variation in the SPA1 protein level in WT plants grown under LL conditions: high in subjective daytime and low during the subjective dusk (Figure 2b). The peak of the SPA1 oscillation was shifted toward the subjective daytime compared to that of the SPA1 transcript (Figure 1a), suggesting that the oscillation of SPA1 protein levels follows transcriptional regulation. By contrast, the SPA1 protein remained more or less at the same level in plants under DD conditions and little oscillation of the SPA1 protein was detected (Figure 2c), despite the very low transcript level under these conditions (Figure 1b). Thus, SPA1 oscillation is likely to reflect the rhythmic transcription of the SPA1 gene.

The spa1 mutation affects the period of cycling of CCA1 and TOC1/PRR1

We next examined the effect of the spa1 mutation on cycling of the circadian-regulated genes, CCA1 and TOC1/PRR1, in plants grown under LL or DD conditions. We found that CCA1 and TOC1/PRR1 transcripts in spa1-3 mutant plants grown under LL conditions exhibited slightly shorter periods than those in WT plants, although the spa1-3 mutation did not affect the transcript levels (Figure 3a). The circadian phenotype of CCA1 and TOC1/PRR1 with shorter period was nearly identical to that observed for the SPA1 transcript (Figures 1a and 3a). By contrast, in plants under DD conditions, the rhythm of CCA1 and TOC1/PRR1 transcript accumulation damped after transfer to darkness (data not shown). We also examined the rhythm of CCR2 transcript accumulation in spa1-3 mutant plants under DD conditions (Kreps and Simon, 1997); however, the circadian rhythm of CCR2 became undetectable after one cycle under DD conditions (Figure 3b). For this reason, we could not determine whether the spa1 mutation also affects circadian periods under DD conditions.

Figure 3.

 The effect of the spa1 mutation on circadian cycling of CCA1 and TOC1/PRR1 transcripts.
(a) Circadian cycling of CCA1 and TOC1/PRR1 transcripts under LL conditions. (b) Diurnal and circadian cycling of CCR2 transcripts under DD conditions. WT and spa1-3 mutant plants grown in day–night cycles (12 h light, 12 h dark) were transferred to (a) continuous light (LL: time 0) or (b) continuous darkness (DD: time 24), and then harvested at 3 h intervals. Total RNAs were analyzed by Northern blot hybridizations with probes specific for CCA1, TOC1/PRR1, CCR2 and UBQ10 sequences. UBQ10 transcript levels were used as loading controls. Transcript levels of CCA1, TOC1/PRR1, CCR2 and UBQ10 were quantified, and the levels of (a) CCA1 and TOC1/PRR1 in WT at 23 h and (b) CCR2 in WT at 2 h were taken as 1.0. Each experiment was performed twice with similar results.

The period-shortening effect of spa1-3 on cycling of CCA1 and TOC1/PRR1 is dependent on light quality

To see whether the circadian role of SPA1 is restricted to the input pathway or SPA1 functions more closely to the central oscillator of the circadian clock, we investigated the rhythms of CCA1 and TOC1/PRR1 under monochromatic light conditions. WT and spa1-3 mutant plants grown under day–night cycles (12 h light, 12 h dark) were released into continuous red (8 μmol m−2 sec−1) or continuous blue (8 μmol m−2 sec−1) light, and harvested at 3 h intervals. Figure 4(a) shows that, under continuous red light conditions, CCA1 and TOC1/PRR1 transcripts in spa1-3 mutant plants clearly exhibited shorter period than those in WT plants (Figure 4a). By contrast, under continuous blue light conditions, the rhythm of CCA1 and TOC1/PRR1 transcript accumulation in spa1-3 was almost similar to that in WT (Figure 4b). We also found that the CCR2 transcript accumulation in spa1-3 exhibited shorter period than that in WT under continuous red light conditions, but not under blue light conditions (data not shown). Thus, the circadian phenotype of spa1-3 was affected by altered light quality, suggesting that SPA1 functions in an input pathway rather than as a component of the central oscillator, although it is difficult to determine strictly whether SPA1’s function is restricted to the input pathway only.

Figure 4.

 The effect of the spa1 mutation on the circadian regulation of CCA1 and TOC1/PRR1 under monochromic light conditions.
(a) Circadian cycling of CCA1 and TOC1/PRR1 under continuous red light. (b) Circadian cycling of CCA1 and TOC1/PRR1 under continuous blue light. WT and spa1-3 mutant plants grown in day–night cycles (12 h light, 12 h dark) were transferred to (a) continuous red light (8 μmol m−2 sec−1: time 0) or (b) continuous blue light (8 μmol m−2 sec−1: time 0) and harvested at 3 h intervals. Total RNAs were analyzed by Northern blot hybridizations with probes specific for CCA1, TOC1/PRR1 and UBQ10. UBQ10 levels were used as loading controls. Transcript levels of CCA1 and TOC1/PRR1 were quantified and the values of CCA1 and TOC1/PRR1 in WT at 23 h under monochromic conditions were taken as 1.0. Each experiment was performed twice with similar results.

The spa1 mutation causes early flowering under SD conditions

Mutations in several Arabidopsis circadian clock-related genes result in a reduction or absence of sensitivity to day length, and the phenotypes observed for these mutants are often associated with flowering time (Doyle et al., 2002; Fowler et al., 1999; Park et al., 1999; Schaffer et al., 1998; Somers et al., 1998b; Wang and Tobin, 1998). We therefore analyzed the flowering time phenotype of spa1-3 mutant plants by measuring the number of days to bolting and the total leaf number at bolting. WT and spa1-3 mutant plants were grown under LD (16 h light, 8 h dark) or SD (10 h light, 14 h dark) conditions, and flowering time was scored. Figure 5(a) shows that, under LD conditions, the number of days to bolting of the spa1-3 mutant was similar to that of WT. Furthermore, there was no significant difference in the total number of leaves at bolting between WT and the spa1-3 mutant (Figure 5c). By contrast, under SD conditions, the spa1-3 mutant clearly flowered earlier than WT (Figure 5a,b), and the total number of leaves at bolting in the spa1-3 mutant distinctly decreased compared to that in WT (Figure 5c). This result indicates that the spa1-3 mutation reduces photoperiodic regulation of flowering time.

Figure 5.

 Photoperiodic responses of WT and spa1-3 mutant plants.
(a) The bolting time of WT and spa1-3 mutant plants. WT and spa1-3 mutant plants were grown on soil under SD (10 h light, 14 h dark) or LD (16 h light, 8 h dark) conditions. The bolting time was measured as the time of emergence of the first flower bud in WT and spa1-3 mutant plants. Two independent experiments were carried out, and 20–30 plants were measured for each experiment. Mean values are given along with standard deviations.
(b) Forty-day-old plants of WT RLD ecotype (left) and spa1-3 mutant (right) grown on soil under SD conditions.
(c) The number of leaves at the time of bolting. Leaf numbers at bolting were counted in WT and spa1-3 mutant plants grown under LD or SD conditions. Thirty to forty plants were measured and mean values with standard deviations are presented.

The spa1 mutation does not affect diurnal oscillations of photoreceptors

We next examined the effect of the spa1-3 mutation on phyA protein levels in plants under SD or LD conditions. Two-week old WT and spa1-3 mutant seedlings grown under SD or LD conditions were harvested at 4 h intervals after dawn (0 h). Total protein extract from the seedlings was analyzed by Western blots using anti-phyA monoclonal antibodies (Shinomura et al., 1996). Figure 6 shows that the spa1-3 mutation did not affect the diurnal oscillation in phyA levels under either SD or LD conditions.

Figure 6.

 Expression level of photoreceptors under SD or LD conditions.
(a) Expression levels of photoreceptors under SD conditions. (b) Expression levels of photoreceptors under LD conditions. Two-week-old WT and spa1-3 mutant plants grown under SD or LD conditions were harvested at 4 h intervals, starting at 0 h after dawn. Total protein extracts were analyzed using anti-phyA, anti-phyB, anti-cry1 and anti-cry2 antibodies. Tubulin levels were used as a loading control. The relative protein levels were quantified and the value of each photoreceptor in WT at 0 h was taken as 1.0. Each experiment was performed twice with similar results.

Recently, it has been shown that a single amino acid substitution in cry2 results in a reduction in light-induced down-regulation of cry2 in plants grown under SD conditions, leading to early flowering (El-Din El-Assal et al., 2001). Therefore, we also examined diurnal oscillations of cry2, as well as the photoreceptors cry1 and phyB using specific antibodies (Shinomura et al., 1996; see Experimental procedures). Figure 6 shows that the protein level and the oscillation patterns of these photoreceptors in spa1-3 mutant plants were nearly identical to those in WT plants. These results indicate that the early flowering phenotype of the spa1 mutant is not caused by an altered protein level or an altered daily oscillation of the photoreceptors.

The early flowering phenotype of spa1-3 correlates with increased FT transcript levels

Our phenotypic analysis indicates that spa1-3 mutant plants have reduced sensitivity to photoperiod (Figure 5). Therefore, we examined whether the spa1-3 mutation affects the diurnal oscillation of CO under SD or LD conditions. Figure 7(a) shows that, under SD conditions, the spa1-3 mutation did not affect the diurnal oscillation in the CO transcript level, which peaked around the beginning of night and decreased towards dusk. In addition, no significant difference in the CO oscillation was observed in spa1-3 mutant plants under LD conditions (Figure 7b).

Figure 7.

 Diurnal oscillation of CO and FT transcripts.
(a) Transcript accumulation of CO and FT under SD conditions. (b) Transcript accumulation of CO and FT under LD conditions. Two-week-old WT and spa1-3 mutant plants grown under SD or LD conditions were harvested at 4 h interval, starting at 0 h after dawn. Total RNAs were analyzed by RT-PCR using primers specific for CO and FT transcripts. Products of the control PCR were amplified with primers specific for α-tubulin transcripts. Transcript levels were quantified and the levels of CO and FT transcript in WT at 0 h were taken as 1.0. Mean values of two independent experiments are given along with standard deviations.

We next examined FT transcript levels under SD or LD conditions, as studies of CO reveal that FT is an early target gene of CO and the FT transcript level correlates with CO protein level (Samach et al., 2000; Valverde et al., 2004). Under LD conditions, the oscillation in spa1-3 mutant plants was similar to that in WT plants (Figure 7b). By contrast, under SD conditions, the FT transcript level in spa1-3 mutant plants markedly increased throughout the day compared to that in WT plants (Figure 7a). This result is consistent with the early flowering phenotype of spa1-3 under SD conditions (Figure 5).

Discussion

Light signal is important for the SPA1 rhythmicity

Immunoblot analyses have demonstrated that the SPA1 protein level in plants grown under LL conditions is regulated in a circadian manner with peaks in the subjective daytime (Figure 2b). This circadian regulation of SPA1 protein was similar to that observed for the SPA1 transcript (Figure 1a), suggesting that the transcriptional regulation of SPA1 is a key mechanism controlling SPA1 function under LL conditions. However, when plants were shifted from day–night cycles to constant dark, the SPA1 transcript levels decreased markedly and remained low (Figure 1b). Moreover, the SPA1 protein constitutively accumulated with little oscillation under DD conditions (Figure 2c). Although we did not determine whether the circadian regulation of SPA1 transcript continued in the subsequent DD conditions, under constant darkness the SPA1 protein might not be subject to circadian regulation, at least at the post-translational level (Figures 1b and 2c). Therefore, under these conditions, SPA1 protein is likely to be relatively stable. Taken together, our results indicate that light is required not only for the expression of the SPA1 transcript, which cycles in a circadian fashion, but also for the post-translational degradation of SPA1 protein to generate oscillations at the post-translational level under LL conditions.

Involvement of SPA1 in the input pathway to the clock

We found that spa1-3 exhibits slightly shorter circadian periods of accumulation of CCA1, TOC1/PRR1 and SPA1 transcripts compared to WT grown under LL conditions (Figures 1a and 3a), but the amplitude of CCA1 and TOC1/PRR1 oscillations remains essentially the same (Figure 3a). These results suggest that SPA1 is required to maintain the proper period length of circadian rhythms.

It is currently accepted that the circadian system is composed of three components: an input pathway, a central oscillator and an output pathway. Both the clock-regulated oscillation of SPA1 and the circadian phenotype of spa1-3 suggest that SPA1 functions in the input pathway and/or the central oscillator. If SPA1 is a component of the input pathway, the circadian phenotype of spa1-3 should be restricted to LL conditions only. When plants were shifted from the day–night cycle (12 h light, 12 h dark) to DD conditions, the rhythms of CCR2, CCA1 and TOC1/PRR1 damped rapidly (Figure 3b; data not shown), making it difficult to determine whether SPA1 is an input component. Nevertheless, our experiments with monochromatic lights demonstrated that the shorter circadian period of spa1-3 was dependent on light quality (Figure 4). Therefore, it seems likely that SPA1 functions more closely to the input pathway rather than the central oscillator, although the mechanism by which the circadian phenotype is generated in spa1-3 mutants has yet to be examined.

In plants, the circadian period length is inversely related to light intensity (McClung, 2001). In Arabidopsis, phytochromes and cryptochromes contribute to the establishment of the period length. Indeed, mutation of PHYA causes the circadian rhythms of CAB2 promoter activity to oscillate at a pace slower than that of WT in dim red light and at low fluence rates of blue light below 3 to 5 μmol m−2 sec−1 (Somers et al., 1998a). The period-shortening effect of spa1-3 on circadian cycling was detected under red light (8 μmol m−2 sec−1) but not blue light conditions (8 μmol m−2 sec−1) (Figure 4). In addition to the circadian phenotype of spa1-3 under red light, the mutant seedlings also display increased responsiveness to red light of broad fluence rates in a phyA-dependent manner (Hoecker et al., 1998). Taken together, these results suggest that SPA1 may negatively regulate the phyA-mediated input pathway to the clock. The absence of SPA1 might enhance phyA-mediated light signaling, resulting in a shorter circadian period of the clock under red light, but not under higher fluence rates of blue light. Hence, light signals perceived by phyA might reach the circadian clock, allowing expression of CCA1, TOC1/PRR1 and SPA1 transcripts, which in turn might lead to the control of phyA-mediated circadian regulation by SPA1.

SPA1 might be involved in the photoperiod pathway to regulate flowering

Our phenotypic analysis has demonstrated that spa1-3 mutant plants flower earlier, especially under SD conditions (Figure 5), indicating that the spa1 mutation results in a reduction of photoperiodic regulation of flowering time.

In Arabidopsis, there are at least four pathways to regulate the flowering time: the gibberellin (GA), autonomous, vernalization and photoperiod pathways. The autonomous and vernalization pathways promote flowering by reducing the accumulation of the FLC transcript and FLC protein (Michaels and Amasino, 1999; Sheldon et al., 2000). The major role of FLC is to repress flowering by inhibiting expression of the genes FT and SOC1/AGL20 that promote flowering, which is opposite to the role of CO (Mouradov et al., 2002). The FLC transcript level in spa1-3 grown under SD conditions is similar to that in WT (data not shown). In addition, the spa1 mutation does not affect the GAI transcript level, which is involved in the GA pathway (data not shown; Mouradov et al., 2002; Olszewski et al., 2002). Therefore, the early flowering phenotype observed for spa1-3 might not be caused by up- and/or down-regulation of the autonomous, vernalization and GA pathways.

It is known that phyA not only regulates de-etiolation by perceiving continuous FR light, but also promotes flowering. Indeed, phyA mutants flower later than WT plants (Johnson et al., 1994), and transgenic Arabidopsis plants over-expressing phyA flower earlier than WT plants (Bagnall et al., 1995). Our Western blot analyses have demonstrated that diurnal oscillations of photoreceptors are unaffected by the spa1-3 mutation (Figure 6). These observations suggest that the spa1-3 mutation enhances phyA signaling without elevating phyA protein levels, thereby promoting early flowering under SD conditions. Thus, SPA1 might negatively regulate phyA-mediated light signals for flowering.

The role of SPA1 in flowering

Mutations in the TOC1 gene lead to an early flowering phenotype specifically under SD conditions (Somers et al., 1998b) and affect the CO transcript level that is regulated by the circadian clock, with an expression peak in the second half of the day (Yanovsky and Kay, 2002). Recently, it has been proposed that the phase of diurnal oscillations of the CO transcript in the toc1 mutant is shifted towards daytime compared to that in WT, resulting in relatively high levels of the CO transcript at dusk (Yanovsky and Kay, 2002) and early flowering under SD conditions. The flowering phenotype observed for toc1 is apparently similar to that of spa1-3. However, it is notable that, in contrast to toc1, the phase of diurnal oscillations of the CO transcript under either SD or LD conditions was not affected by spa1-3, despite an increase in FT transcript levels (Figure 7a). This observation suggests that the CO protein level and/or the CO protein activity are up-regulated.

Little is known about the molecular mechanism by which the CO protein is activated by light to induce the expression of FT. However, a recent analysis of photoreceptor mutants showed that phyA and cry2 antagonize the degradation of CO protein in the evening, allowing the activation of FT (Valverde et al., 2004). Apparently, these photoreceptors balance the abundance of CO protein in plants. Therefore, it is possible that the activity of the photoreceptors may oscillate during the light phase of a long day, although the levels of these proteins remain constant during a long day (Figure 6). This hypothesis might be supported by the observation that the abundance of SPA1 protein displays circadian oscillation, increasing in the daytime and decreasing in the evening (Figure 2). Hence, clock-regulated SPA1 may indirectly control the activity of these photoreceptors, especially phyA activity, thereby regulating the stability and/or the activity of the CO protein through a photoperiod pathway.

Apart from the known photoperiod pathway that regulate(s) the expression of FT and SOC1/AGL20, SPA1 may regulate phyA-dependent flowering via a different pathway. Recently, a light-quality pathway has been proposed to explain the role of phyB, phyD and phyE in flowering (Simpson and Dean, 2002). phyB is known to play a role in shade avoidance responses and regulate flowering in response to an altered ratio of red to FR light. More recently, it has been suggested that phyB regulates FT expression level by a PHYTOCHROME AND FLOWERING TIME 1 (PFT1)-dependent mechanism that does not involve the transcriptional regulation of CO or SOC1/AGL20 (Cerdan and Chory, 2003). At present, it is not known how the spa1-3 mutant flowering phenotype is affected by altered light quality. Nevertheless, SPA1 may be involved in the light-quality pathway and thereby negatively regulate FT without the transcriptional regulation of CO. In either case, further experiments should be directed to the analysis of CO protein levels in the spa1-3 mutant.

Experimental procedures

Plant materials and growth conditions

The Arabidopsis thaliana ecotype RLD was used in this work. The spa1-3 mutant was kindly provided by Dr Ute Hoecker, University of Dusseldorf. To measure the number of days to bolting and total leaf number, seeds were sown on soil, kept in the dark at 4°C for 3 days, and then transferred to a growth chamber at 22°C under LD (16 h light, 8 h dark) and SD (10 h light, 14 h dark) as described previously (Putterill et al., 1995). Seeds were surface-sterilized, sown on Murashige and Skoog media plates with 1% sucrose and then cold-treated for 3 days at 4°C. Plates were illuminated with white fluorescent light in a growth chamber at 22°C and kept under various light conditions.

RNA expression analysis

For Northern blotting analysis, WT and spa1-3 mutant plants were grown in a chamber under day–night cycles (12 h light, 12 h dark) for 20 days, transferred to continuous white, red, blue lights or darkness, and harvested at 3 h intervals. Total RNA purified using an RNeasy Plant Mini Kit (QIAGEN, Valencia, CA, USA) was subjected to Northern blotting analysis as described previously (Matsushika et al., 2000). 32P-labeled RNA probes specific for SPA1, CCA1, TOC1/PRR1, CCR2 and UBQ10 sequences were used to detect each transcript.

For RT-PCR analysis, total RNA was purified from 2-week-old WT and spa1-3 mutant seedlings grown under SD or LD conditions, and the first strand of cDNA was synthesized using Ready-To-Go You Prime First-Strand Beads (Amersham Biosciences, Piscataway, NJ, USA). Sample volumes were normalized for equal amplification of DNA fragments with primers specific for α–tubulin cDNA. PCR was performed as described by Semiarti et al. (2001), and primer sets used in RT-PCR were as follow: 5′-CATTAACCATAACGCATACATTTC-3′ and 5′-CTCCTCGGCTTCGATTTCTC-3′ for CO; 5′-TAAGCAGAGTTGTTGGAGACG-3′ and 5′-TCTAAAGTCTTCTTCCTCCGCAG-3′ for FT; 5′-GGACAAGCTGGGATCCAGG-3′ and 5′-CGTCTCCACCTTCAGCACC-3′ for α-tubulin. Transcript levels were analyzed using NIH ImageJ software (http://rsb.info.nih.gov/ij/).

Antibody production and purification

The cDNAs encoding peptides corresponding to amino acid residues 510-681 of CRY1 (CRY1C), 478-612 of CRY2 (CRY2C) and 1-114 of SPA1 (SPA1N) were cloned into the vector pET-28a as translation fusions to a His tag. The recombinant proteins, CRY1C, CRY2C and SPA1N, expressed in Escherichia coli were purified using Ni-affinity columns and purified proteins were used to prepare rabbit polyclonal antibodies. To purify antibodies specific to the recombinant proteins from the serum, 300 μg of each recombinant protein were fractionated by SDS–PAGE and transferred to Immobilon-P membranes (Millipore, Billerica, MA, USA). Membranes were briefly stained with Ponceau S, and bands containing the protein were excised and incubated with antisera in TBS containing 0.1% Tween-20 and 5% skim milk for 2 h at room temperature. Membrane strips were eluted with 1 ml of 0.2 m glycine (pH 2.5) and the eluate immediately mixed with 0.1 ml of 1.5 mTris-Cl (pH 8.8). Eluted antibodies were dialyzed in TBS at 4°C for 16 h before addition of 500 ng μl−1 BSA.

Immunoblotting analysis

Two-week-old plants of WT and the spa1-3 mutant under LD or SD conditions and 3-week-old plants of WT and the spa1-3 mutant under LL or DD conditions were harvested and frozen immediately with liquid nitrogen. The frozen plants were ground in liquid nitrogen, and the powder was added to the same volume of sample buffer for SDS–PAGE and boiled for 10 min. The protein extract was analyzed with SDS–PAGE (6% acryl-amide gel, 0.67% bis-acrylamide). Proteins in the gels were transferred to Immobilon-P membrane and detected with antibodies specific to PHYA, PHYB, CRY1, CRY2 and SPA1.

Acknowledgements

We thank Peter Hare for additional valuable comments on the manuscript, Ute Hoecker for providing spa1-3 mutant seeds, and Akira Nagatani for providing antibodies to phyA and phyB. M.I. was supported by a fellowship from the Human Frontier Science Program Organization (LT00725/2003-C). T.K. was in part supported by a fellowship from the Japan Society for the Promotion of Science. This work was supported by NIH grant GM 44640 to N-H.C.

Ancillary