A study of Gibberellin homeostasis and cryptochrome-mediated blue light inhibition of hypocotyl elongation

Cryptochromes mediate blue light-dependent photomorphogenic responses, such as inhibition of hypocotyl elongation. To investigate the underlying mechanism, we analyzed a genetic suppressor, scc7-D (s uppressors of cry1cry2 ), which suppressed the long-hypocotyl phenotype of the cry1cry2 mutant in a light-dependent but wavelength-independent manner. scc7-D is a gain-of-expression allele of the GA2ox8 gene encoding a gibberellin (GA)-inactivating enzyme, GA 2-oxidase. Although scc7-D is hypersensitive to light, transgenic seedlings expressing GA2ox at a level higher than scc7-D showed a constitutive photomorphogenic phenotype, confirming a general role of GA2ox and GA in the suppression of hypocotyl elongation. Prompted by this result, we investigated blue light regulation of mRNA expression of the GA metabolic and catabolic genes. We demonstrated that cryptochromes are required for the blue light regulation of GA2ox1 , GA20ox1, and GA3ox1 expression in transient induction, continuous illumination, and photoperiodic conditions. The kinetics of cryptochrome induction of GA2ox1 expression and cryptochrome suppression of GA20ox1 or GA3ox1 expression correlate with the cryptochrome-dependent transient reduction of GA 4 in etiolated wild-type seedlings exposed to blue light. Therefore we propose that in de-etiolating seedlings, chryptochromes mediate blue-light regulation of GA catabolic/metabolic genes, which affect GA levels and hypocotyl elongation. Surprisingly, no significant change in the GA 4 content was detected in the whole shoot samples of the wild type or cry1cry2 seedlings grown in the dark or continuous blue light, suggesting that cryptochromes may also regulate GA responseness and/or trigger cell-or tissue-specific changes of the level of bioactive GAs.

Phytochromes affect GA levels by regulating expression of the GA2ox, and GA3ox genes (Reid et al., 2002). Phytochromes may also regulate GA responsiveness (Reed et al., 1996;Cao et al., 2005;Foo et al., 2006). It has been recently shown that phyA and phyB mediate light stabilization of the DELLA proteins, which may, at least partially, result from the phytochrome-dependent regulation of GA homeostasis (Achard et al., 2007). In comparison to the phytochrome-regulated responses, the relationship between cryptochromes and GA in the blue light responses is less clear in Arabidopsis.
It has been found in pea that cry1 and phyA redundantly regulate GA2ox and GA3ox expression and GA signaling (Symons and Reid, 2003;Foo et al., 2006). Arabidopsis cry1 has been reported to regulate hypocotyl elongation via its effect on GA and auxin homeostasis or signaling (Folta et al., 2003).
Prompted by a study of the genetic suppressor of cry1cry2 that corresponds to the Arabidopsis GA2ox8 gene, we investigated the relationship between GA homeostasis and cryptochrome-mediated de-etiolation. We showed that increased expression of a GA2ox gene caused hypersensitive or constitutive photomorphogenesis, depending on the relative levels of GA2ox overexpression. We further demonstrated that cryptochromes are required for the blue light induction of GA2ox1 expression and blue light suppression of GA20ox1 and GA3ox1 expression. Although all those observations are consistent with a simple hypothesis that cryptochromes may inhibit accumulation of bioactive GAs to suppress hypocotyl growth, our analyses of the GA 4 content in whole shoot samples indicate the involvement of a more complex mode of regulation in the cryptochrome-mediated photomorphogenic responses.

Overexpression of GA2ox8 suppresses the long hypocotyl phenotype of the cry1cry2 mutant
To investigate the genetic mechanisms underlying cryptochrome-mediated light responses in plants, we prepared an activation-tagging population (see Materials and Methods) in the cry1cry2 double mutant, which exhibits a long hypocotyl phenotype when grown in blue light (Guo et al., 1998;Mockler et al., 1999). scc7-D was one of the dominant mutants identified in this screen. The scc7-D mutant suppressed the longhypocotyl phenotype of the cry1cry2 parent when grown in blue light, but it showed a normally elongated hypocotyl when grown in the dark ( Fig. 1 A-C). However, scc7-D is hypersensitive to not only blue light, but also red and FR lights ( Fig. 1D-F), suggesting that the corresponding gene is associated with light inhibition of cell elongation mediated by both cryptochromes and phytochromes.
The scc7-D locus contains a T-DNA inserted in an intergenic region, at 4293bp upstream from the start codon of GA2ox8 ( Fig. 1G-H), and GA2ox8 seems the only one overexpressed among the 7 T-DNA flanking genes tested (Fig. 1I). The lighthypersensitive phenotype of scc7-D can be rescued by exogenous GA 3 (Fig. 3J). In addition, scc7-D also showed a dwarf and late-flowering phenotype (not shown).
Similar dominant alleles of the GA2ox8 gene have been reported previously; and it was shown that GA2ox8 encodes a GA 2-oxidase that catalyzes 2β-hydroxylation of C 20 - GAs (Schomburg et al., 2003). Importantly, plants overexpressing GA2ox8 contain reduced amounts of bioactive GAs and precursors, while loss-of-function ga2ox7 and ga2ox8 mutants exhibited a long hypocotyl phenotype (Schomburg et al., 2003). We concluded that the exaggerated light response in the scc7-D allele was also caused by the increased GA2ox8 expression and reduced accumulation of bioactive GAs.
The level of GA2ox8 mRNA in scc7-D was probably high enough to cause exaggerated hypocotyl inhibition in blue light via phytochromes, some of which can act as blue light receptors, but not high enough to suppress hypocotyl elongation in the absence of light. Indeed, transgenic plants overexpressing the GFP-GA2ox8 fusion protein under the control of a strong 35S promoter exhibited a GA-rescuable constitutive photomorphogenic phenotype (Fig 2A-C). Similar to scc7-D, the 35S::GFP-GA2ox8 seedlings are also hypersensitive to light ( Fig. 2A-B). But contrary to scc7-D, the etiolated 35S::GFP-GA2ox8 transgenic seedlings grown in the dark also showed short hypocotyls, as well as un-hooked and partially opened cotyledons (Fig 2A-B). The phenotype of 35S::GFP-GA2ox8 was rescued by exogenous GA 3 (Fig. 2C, and data not shown). The levels of the 35S::GFP-GA2ox8 mRNA were markedly higher in transgenic 35S::GFP-GA2ox8 seedlings than that of the GA2ox8 mRNA in scc7-D or wild type seedlings (Fig. 2D), which explains why 35S::GFP-GA2ox8 lines are constitutively photomorphogenic regardless of light whereas scc7-D seedlings showed exaggerated photomorphogenic response only in light. Our observation that overexpression of a GA2ox enzyme, which catalyzes inactivation of GAs, caused constitutive photomorphogenesis is consistent with a previous report that the ga1 mutant impaired in the CPS (ent-copalyl diphosphate synthase) gene showed constitutive photomorphogenesis (Sun et al., 1992;Alabadi et al., 2004). Both observations suggest that photomorphogenic development is at least partially regulated by the levels of bioactive GA.

Association between GA homeostasis and blue light inhibition of hypocotyl elongation
The study of scc7-D prompted us to further explore a possible association between cryptochrome function and GA homeostasis. The wild type and the cry1cry2 mutants (grown at 22 o C) responded only weakly to bioactive GA 3 or GA 4 , regardless of light conditions (Fig. 3). Even under the highest fluence rate of blue light tested, whereby the wild-type seedlings normally exhibited minimum hypocotyl elongation, high concentrations of GA 3 or GA 4 (0.1-1 mM) failed to elicit a large increase in hypocotyl growth or to phenocopy the cry1cry2 mutant in the wild-type seedlings (Fig. 3A, 3E). It is noteworthy that although various hormones have been shown to antagonize light inhibition of hypocotyl growth when applied exogenously, such treatments never completely reverse the inhibitory effect of light on growth (Vandenbussche et al., 2005).
At a higher temperature (26 o C), the wild-type Arabidopsis seedlings responded relatively more strongly to exogenous GAs as reported previously (Fig. 3F) (Collett et al., 2000). A thermoperiodic regulation of PsGA2ox expression has been reported to control thermoperiodic fluctuation of GA 1 content and stem elongation of pea stems (Stavang et al., 2005). Whether our observation represents a temperature sensitive GA responsiveness or that it may be related to the high temperature promotion of auxinmediated hypocotyl elongation (Gray et al., 1998) remains to be further investigated.
We next examined how the cry1cry2 mutant responded to GA biosynthesis inhibitors. The GA biosynthesis inhibitors, paclobutrazol (Fig. 3B) or ancymidol (Fig.   3D), can both rescue the long-hypocotyl phenotype of the cry1cry2 mutant, which is consistent with the observation of the cry1 mutant (Folta et al., 2003). These results suggest that GA is likely involved in the development of the long hypocotyl of cry1cry2, and that the level of bioactive GA is important for cryptochrome function. It is interesting that, when GA synthesis was inhibited, the cry1cry2 mutant seedlings grown in blue light with relatively high fluence rates exhibited greater hypocotyl elongation in response to GA 3 than the wild type seedlings (Fig. 3C, B3, B20, B100). These results indicate that cryptochromes may also be required to suppress GA response under blue light.

Cryptochromes mediate blue light-activated mRNA expression of GA2ox genes, especially GA2ox1, under dark-to-light transfer, continuous illumination, and photoperiodic conditions
The Arabidopsis genome encodes up to 8 GA2ox-related sequences, referred to as GA2ox1 to GA2ox8 (see Materials and Methods). However, the mRNA expression of only 6 members (GA2ox1, GA2ox2, GA2ox4, GA2ox6, GA2ox7, and GA2ox8) was readily detected in our study. GA2ox5 seems to be a pseudogene (Hedden et al., 2001;Schomburg et al., 2003), whereas GA2ox3 is normally expressed at a low level in the absence of auxin induction (Frigerio et al., 2006). Results of various DNA microarray studies concerning the blue light effect on GA2ox expression were not always consistent. For example, the expression of GA2ox1 shows an 8-fold increase in response to blue light in the wild-type seedlings in one study (https://www.genevestigator), but a 2.4-fold increase of GA2ox1 transcript was reported in the cry1 mutant in another study (Folta et al., 2003). We decided to systematically examine the blue light and cryptochrome effects on the expression of each member of the GA2ox gene family upon transfer from dark to blue light and under photoperiodic and continuous light conditions. We first analyzed blue light regulation of GA2ox expression in response to inductive blue light treatment. In this experiment, 6-day-old etiolated seedlings were exposed to blue light (100 µmolem -2 s -1 ) for 0.5 to 24 hours, and the level of mRNA expression of the GA2ox genes was examined at various time points. In wild-type seedlings, all six GA2ox genes tested showed various degrees of blue-light induction of mRNA expression (Fig. 4). For example, the expression of GA2ox1 and GA2ox8  (Folta et al., 2003). It is particularly interesting that GA2ox1 is much more responsive to cryptochrome-dependent blue light induction than other GA2ox genes, because GA2ox1 is one of the GA2ox genes that are not induced by auxin (Frigerio et al., 2006). The observation that GA2ox1 is more responsive to blue light than other auxin-inducible GA2ox genes indicate that GA2ox1 may be associated with the auxin-independent growth response reported previously (Collett et al., 2000).
The transient nature of the blue-light induction of GA2ox expression observed in the 24 hr inductive blue light treatment (Fig. 4) may be due to the diurnal or circadian rhythmic expression of the GA2ox genes. To test this possibility, seedlings were entrained in long-day (16hL/8hD) and short-day (8hL/16hD) photoperiods for 10 days and then transferred to continuous white light for two days. Samples were collected every 3hr for one day before transfer and two days after transfer to the free-running condition, and the expression of each GA2ox gene was analyzed. The mRNA expression of all six GA2ox genes exhibited either a diurnal rhythm or circadian rhythm (Fig. 5, and data not shown). Among them, GA2ox1 and GA2ox2 showed the most robust circadian rhythms that were sustained in seedlings transferred from the long-day or short-day photoperiod to continuous white light for at least two days ( Fig. 5A-B, and data not shown). The phase of the circadian rhythm of GA2ox2 was dramatically different in plants grown in long-day and short-day photoperiods (Fig. 5B). It has been shown recently that GA 4 is the active gibberellin regulating floral initiation (Eriksson et al., 2006). Whether the photoperiod-dependent phase change of GA2ox2 expression affects photoperiodic flowering in Arabidopsis remains to be further investigated (Blazquez and Weigel, 1999;Yu et al., 2006;Zhao et al., 2007). The high-amplitude rhythmic expression of GA2ox1 was all but abolished in the cry1cry2 mutant grown in long-day photoperiod ( We next compared the steady-state levels of GA2ox expression in seedlings grown in continuous blue light or dark (Fig. 6). Four GA2ox genes (GA2ox1, GA2ox2, GA2ox4, GA2ox7) tested showed higher steady-state levels of mRNA in seedlings grown in continuous blue light than in etiolated seedlings (Fig. 6A). Under continuous blue light, the expression of GA2ox1 showed a more pronounced blue light response than other GA2ox genes; the mRNA expression of GA2ox1 was at least 7-fold higher in light-grown wild-type seedlings than that in etiolated wild-type seedlings (Fig. 6A). A more pronounced blue light effect on the GA2ox1 expression is consistent with that observed under other conditions  and that found in continuous white light (Achard et al., 2007). Although our study was initially prompted by the observation that overexpression of GA2ox8 suppressed cry1cry2 phenotype in the scc7-D mutant (Fig.   1), the level of GA2ox8 mRNA expression was not dramatically affected by continuous blue light or cryptochromes (Fig. 2D, and Fig. 6B). The blue light-induced GA2ox1 expression in seedlings grown under continuous blue light was moderately impaired in the cry1 mutant (Fig. 6B), but it was markedly reduced in the cry1cry2 (Fig. 6A), cry1phyA (Fig. 6B), and cry1cry2phyA mutants (Fig. 6B). These results demonstrate that CRY1 acts redundantly with CRY2 or phyA to mediate blue-light induction of GA2ox1 expression. We concluded that cryptochromes are the major blue light receptors required for the transient blue light induction of GA2ox1 expression in etiolated seedlings exposed to blue light, for the circadian rhythmic GA2ox1 expression in seedlings grown in long-day photoperiods under white light, and for the high level steady-state mRNA expression of GA2ox1 in seedlings grown in continuous blue light.

Cryptochromes mediate blue light suppression of GA20ox1 and GA3ox1 mRNA expression
In addition to the GA2ox enzymes that catalyze catabolism or inactivation of bioactive GAs such as GA 4 , two other dioxygenases, GA 20-oxidase (GA20ox) and GA 3β-hydroxygenase (GA3ox) that catalyze synthesis of bioactive GAs are also critical to GA homeostasis (Hedden and Phillips, 2000;Reid et al., 2004). Therefore, we next examined blue light and cryptochrome effects on the expression of GA20ox and GA3ox genes. Fig. 7 shows that expression of the three GA20ox genes and one of the two GA3ox genes tested were lower (to differing extents) in seedlings grown in continuous blue light than in etiolated seedlings (Fig. 7A). The blue light suppression of GA20ox1 and GA3ox1 expression are dependent on cryptochromes, because the sustained blue light suppression of both genes was largely abolished in the cry1cry2 mutant (Fig. 7).
Therefore, GA20ox1 and GA3ox1 are also major targets of cryptochromes. Similar to what was observed for the GA2ox genes (Fig. 4), but in the opposite direction, the expression of most GA20ox and GA3ox genes tested (except GA3ox2) showed a transient decrease in etiolated seedlings exposed to blue light (Supplemental Fig. 1).
The blue light suppression of GA20ox1 and GA3ox1 expression was moderately affected in the cry1cry2 mutant upon transfer from dark to blue light, although their expression was more significantly affected in continuous blue light (Fig. 7, Supplemental   Fig. 1). In contrast to all the other GA biosynthesis genes tested, GA3ox2 expression exhibited cryptochrome-dependent blue-light induction upon transfer from dark to blue light and under continuous light conditions (Fig. 7, and Supplemental Fig. 1). In contrast to all the other GA biosynthesis genes tested, GA3ox2 expression exhibited blue-light induction upon transfer from dark to blue light and under continuous light conditions, and it was impaired in the cry1cry2 mutant only under the inductive condition (Fig. 7, and Supplemental Fig. 1). It is known that phyB mediates red light activation of GA3ox2 expression in germinating seeds to facilitate germination (Yamaguchi et al., 1998). The blue light activation of GA3ox2 expression may also be associated with light stimulation of cotyledon expansion rather than light inhibition of hypocotyl elongation, but this response appears redundantly regulated by phytochromes and cryptochromes. We might predict that overall, the cryptochrome-dependent blue-light suppression of the expression of GA biosynthesis genes such as GA20ox1 and GA3ox1, and cryptochrome-dependent blue-light activation of the expression of GA catabolic genes such as GA2ox1, may cause a reduction in bioactive GA levels. However, the discrepancy regarding GA3ox2 expression (discussed above) and the relatively small effect of blue light on the expression of the GA20ox2, GA20ox3, GA2ox2, GA2ox6 and GA2ox8 genes make such predictions very uncertain. Direct measurement of endogenous bioactive GAs is clearly required to accurately assess the effect of blue light on GA level.

Cryptochromes mediate transient blue light suppression of GA 4 accumulation
To test whether cryptochrome-mediated blue light regulation of gene expression changes correlate with changes in the level of bioactive GA 4 , we examined GA 4 levels in etiolated wild type and cry1cry2 mutant seedlings exposed to blue light. As expected, the level of GA 4 showed a statistically significant decline in the wild type seedlings exposed to blue light for 4hrs (Fig. 8A). The blue light-induced decrease of GA 4 was dependent on cryptochromes, because no statistically significant reduction of GA 4 was detected in the cry1cry2 mutant (Fig. 8A). After a blue light treatment for 24hrs, the level of GA 4 in the etiolated wild-type seedlings exposed to blue light became indistinguishable from that in the etiolated seedlings (Fig. 8A), suggesting the cryptochrome-dependent blue light inhibition of GA 4 accumulation was transient or rhythmic. The transient decrease of GA 4 upon transfer from dark to blue light appears to correlate with the transient (or rhythmic) increase of GA2ox1 expression (Fig. 4-5) and transient (or rhythmic) decrease of GA20ox1 and GA3ox1 expression (Supplemental Fig. 1-2). A transient decline of bioactive GA accumulation in etiolated seedlings exposed to blue light has also been reported previously in pea (Foo et al., 2006). In this case the major bioactive GA in pea, GA 1 , showed a significant but transient decrease in etiolated seedlings exposed to blue light. In contrast to Arabidopsis, wherein cryptochromes mediate the transient blue light suppression of GA 4 , cry1 and phyA redundantly control the transient blue light suppression of GA 1 in pea (Foo et al., 2006).
Finally, we compared GA 4 levels in etiolated wild type and cry1cry2 seedlings with those grown under continuous blue light. Interestingly, we detected no significant difference in GA 4 level in those samples (Fig. 8B). However, because whole-shoot samples were used in our GA analyses, we cannot exclude the possibility that a localized or cell-specific change of GA 4 content may occur in seedlings grown in continuous blue light.

DISCUSSION
In this study, we showed that increased expression of a GA2ox gene genetically suppressed the cry1cry2 mutant ( Fig. 1-2). Although this observation by itself may have alternative interpretations, our follow-up photophysiological (Fig. 3) and gene expression studies (Fig. 3-5) demonstrate that cryptochromes are indeed positive regulators of GA2ox genes, especially GA2ox1. We showed that cryptochromes are required for the transient induction of GA2ox1 expression in etiolated seedlings exposed to blue light, for the sustained elevation of GA2ox1 expression in seedlings grown in continuous blue light, and for maintaining a high amplitude of the circadian rhythm of GA2ox1 expression in seedlings grown in long-day photoperiods. Consistent with the cryptochrome-mediated blue-light stimulation of the expression of the GA catabolic gene GA2ox1, we also demonstrated that cryptochromes mediate blue light suppression of the expression of GA biosynthesis genes GA20ox1 and GA3ox1 (Fig. 7).
We concluded that cryptochromes are positive regulators of GA2ox1 but negative regulators of GA20ox1 and GA3ox1. These cryptochrome-regulated gene expression changes may result in a blue light-dependent reduction of bioactive GAs. Given the well-established role of GA as a growth promoter, we propose that cryptochromeregulated change in GA homeostasis is an important mechanism underlying blue-light inhibition of hypocotyl elongation.
Consistent with this hypothesis, our analyses of GA 4 content showed a cryptochrome-dependent transient reduction of GA 4 in etiolated wild-type seedlings exposed to blue light (Fig. 8A). This result correlates with the transitory/rhythmic expression patterns of many GA metabolism/catabolism genes ( Fig. 4-5, Supplemental Fig. 1-2). However, in contrast to the transient reduction in GA levels in de-etiolating seedlings exposed to blue light, we did not detect a significant reduction of GA 4 in the wild type seedlings grown in continuous blue light (Fig. 8B) or white light (Symons and Reid, unpublished). Neither did we detect a significant effect of the cry1cry2 mutation on the GA 4 level in seedlings grown in continuous blue light (Fig. 8B). These results impose a significant challenge to the hypothesis that cryptochromes inhibits hypocotyl elongation solely by reducing GA 4 levels. Similar observations have been previously reported in pea (O'Neill et al., 2000;Symons and Reid, 2003;Foo et al., 2006). In comparison to etiolated seedlings, the level of bioactive GA also showed a transient reduction in pea seedlings transferred from dark to blue light but not in seedlings grown under prolonged illumination (Reid et al., 2002;Foo et al., 2006). It was hypothesized that phyA and cry1 regulate not only GA homeostasis but also GA responsiveness to affect shoot elongation (Reid et al., 2002;Foo et al., 2006). Therefore, it is possible that light regulation of GA responsiveness or GA signal transduction may account for the inhibition of hypocotyl elongation in Arabidopsis plants grown under continuous blue light.
However, we cannot escape the question why continuous blue light caused markedly changed mRNA expression of the GA2ox1, GA20ox1, and GA3ox1 genes ( Fig. 4-7) without a significant change in the level of GA 4 (Fig. 8B). It would be interesting to examine whether those mRNA changes actually resulted in corresponding changes in the protein levels of the respective key enzymes in GA homeostasis.
Alternatively, a localized change of GA 4 levels in response to continuous blue light may provide another possible explanation of our puzzling observations. Specifically, a bluelight-dependent reduction of GA 4 may be limited to specific organs or cells. For example, light inhibits hypocotyl elongation but stimulates cotyledon expansion, suggesting that light may trigger a decrease or increase of GA 4 in hypocotyls or cotyledons, respectively. Moreover, hypocotyl elongation is accomplished mostly by a limited number of cells in the elongation zone (Vandenbussche et al., 2005). Therefore, it is conceivable that light may suppress GA 4 accumulation only in cells located in the elongation zone, but not in cells located in other regions of the same hypocotyl. Such localized changes of GA 4 homeostasis may not be readily discernable when the whole shoot samples were examined as in the present study. The localized change of hormone homeostasis is not uncommon in plants. For example, differential distribution of GA 1 was previously reported in pea seedlings (O'Neill et al., 2000). Developmental regulation by cell-specific auxin and cytokinin biosynthesis have also been reported recently in Arabidopsis and rice, respectively (Cheng et al., 2006;Kurakawa et al., 2007). A cell-specific change of GA levels in response to blue light may be regulated by a post-translational mechanism specific to the respective cells, because the blue lightdependent changes of mRNA expression of GA metabolism/catabolism genes were readily detectable in the whole shoot samples (Fig. 4-7). Whether cryptochromes differentially regulate changes of GA 4 levels in specific cells and what molecular mechanism(s) may be responsible for such changes remain to be further investigated.

Plant Materials
Arabidopsis mutants cry1, cry2, cry1cry2, phyA, and cry1cry2phyA used in this study are in the Col background as described previously (Mockler et al., 2003).
To prepare an activation-tagging population, the cry1cry2 plants were transformed using the binary vector pSKI015 and Agrobacterium strain GV3101 as described (Weigel et al., 2000). The T 1 seeds were harvested in pools, and each pool contains seeds harvested from approximately 50 T 0 plants. T 1 seeds were germinated in white light and grown on compound soil, submerged in the herbicide Basta (ca 0.006% ammonium glufosinate), and approximate 250,000 Basta-resistant T1 individuals were obtained. The herbicide-resistant individuals that showed hypocotyl length shorter than that of cry1cry2 parent were selected as putative scc-D mutants, and they were separated from the rest of the mutant populations. Those scc-D lines that showed short hypocotyls in the T 2 generation grown in continuous blue light were subject to further genetic analysis. Genomic sequence flanking the T-DNA insert was identified using plasmid rescue (Weigel et al., 2000) or the Tail-PCR method (Liu et al., 1995). Hypocotyl lengths were measured manually for 6-day-old or 7-day-old seedlings (Lin et al., 1998). The sample size is larger than 20 seedlings, and the standard deviations are calculated.
To investigate the response of hypocotyl elongation to exogenous GA 3 , paclobutrazol, or ancymidol, seeds were surface sterilized for 30 sec in 70 % ethanol, placed in 0.1% Hgcl 2 for 8 min, and rinsed five times with sterile, distilled water. About 100 seeds were placed in MS agar growth medium. All hormone and inhibitor stocks For studies of light-or clock-regulated gene expression, about 300 sterile seeds were sown on MS agar medium, cold treated at 4 o C for 4 days, exposed to white light for 12 hours, and grown in the dark for 6 days before transfer to various light treatments. Alternatively, seedlings were grown under white light (20±3 µmol m -2 s -1 ) with LD (16hL/8hD) or SD (8hL/16hD) photoperiod for 10 days, and some Petri dishes were transferred to continuous white light for two days. At the end of treatment, Petri dishes were dipped in liquid nitrogen, and tissues (mostly shoot devoid of roots) were harvested by gentle scraping, and stored at -80 o C for RNA extraction.

GA analysis
Whole shoot (all tissues except root) samples were used to analyze the GA content. For inductive light experiments, seeds were sown thickly on pots filled with potting mix, covered with a fine mesh, and placed in weak fluorescent light at 4 o C for 4 days. Plants were transferred to dark at 22 o C for 6 days and then transferred to 60 µmol m -2 s -1 blue light at 22 o C using light source described (Platten et al., 2005).
For GA measurement, 3 to 4 g whole shoot samples were harvested, placed in ice cold 80% methanol, homogenized, and the extract was filtered as described (Symons and Reid, 2003). Samples were concentrated to approximately 1ml, and loaded onto a pre-conditioned Sep-pak C 18 cartridge in 0.4 % acetic acid. GAs were eluted in 80% methanol in 0.4% acetic acid, dried, and fractionated using the reversephase C18 HPLC system (Jager et al., 2005). HPLC fractions corresponding to retention times of the relevant GAs were pooled, dried, methylated, and further purified by solvent partitioning as described (Jones et al., 2005). 10µl dry pyridine and 40µl N,O-bis(trimethylsilyl)trifluoroacetamide (BSTFA) were added to each vial and samples were heated at 80ºC for 20 mins. The samples were then dried before the addition of 15µl BSTFA, incubated at 80ºC for 15 mins, dried under nitrogen gas, resuspended in 20µl of chloroform, and analysed by GC/MS/MS on a Varian 1200 triple quadrupole mass spectrometer (Jones et al., 2005). Injections of 1.5µl were made in splitless mode at an injection temperature of 260°C onto a Varian VF5-ms column (30m x 0.25mm x 0.25micron). The oven temperature was held at 50°C for 2 min, ramped to 230°C at 30°C per min, increased to 270°C at 5°C per min, and held at 270°C for 3 min. The transfer line temperature was 290°C, and the ion source was held at 220°C. The electron multiplier gain was 1950 V. Carrier gas was helium at 1.2ml per minute. The Although the transient reduction in GA 4 level may result from reduced conversion of the precursor GA 9 to GA 4 and/or increased inactivation of GA 4 to the 2β-hydroxylated GA 34 , we detected no significant changes in the GA 9 or GA 34 levels. These discrepancies may be due to the fact that GA 9 level (less than 0.0025 ng.g -1 ) was below our detection limit, whereas GA 34 levels were approximately 10-fold higher than that of GA 4 , resulting in technical difficulties to detect relatively small or transient changes in the GA 34 level under our experimental conditions. To compare GA content in etiolated seedlings with that grown in continuous blue light, 5-day-old whole shoot samples were used.

RNA analyses
The accession numbers of genes discussed in this report are: GA2ox1 (At1g78440), GA2ox2 (At1g30040), GA2ox3 (At2g34555, mRNA not detectable), GA20ox3 (At5g07200), GA3ox1 (At1g15550), and GA3ox2 (At1g80340) Total RNA was isolated using Puprep RNAeasy Mini Kit (Ambiogen Life Tech Ltd. Changsha). DNA-free RNA was obtained by RQ1 DNase I treatment according to the manufacturer's instructions (Promega). The amount of mRNA was analyzed using semi-quantitative RT-PCR as decribed (Mockler et al., 2003). cDNA was prepared from 2 µg of total RNA by using Moloney murine leukemia virus reverse transcriptase according to the manufacturer's instructions (Promega). The cDNA was generally diluted 10 fold, and 1 µl of diluted cDNA was used in a 20 µl PCR reaction. DNA sequences of the PCR primers used in this study are the following.
The GA 4 content of 5-day-old wild type (WT) and cry1cry2 mutant seedlings grown in continuous dark or continuous blue light are shown in (B). Asterisk indicates a significant difference (P < 0.05, two-way ANOVA); D: dark; B4 and B24, etiolated seedlings exposed to blue light for 4hrs and 24hrs, respectively; Bc: seedlings grown in continuous blue light. GA 4 in whole seedling samples except roots was separated by HPLC and analyzed by GC-MS.

SUPPLEMENTAL MATERIALS
Supplemental Fig. 1. Blue light-reduced mRNA expression of members of the GA20ox and GA3ox genes 6-day-old etiolated wild-type or cry1cry2 mutant seedlings were exposed to 3 µmolem -2 s -1 (A) or 100 µmolem -2 s -1 blue light (B), and samples were collected at the time indicated for RNA analyses. Levels of mRNA expression of GA20ox genes (A) and GA3ox genes (B) are shown as the RT-PCR gel images (top) and the relative signal intensities (bottom).