GIGANTEA acts in blue light signaling and has biochemically separable roles in circadian clock and flowering time regulation

: Circadian clocks are widespread in nature. In higher plants, they confer a selective advantage, providing information regarding not only time of day but also time of year. Forward genetic screens in Arabidopsis thaliana have led to the identification of many clock components, but the functions of most of these genes remain obscure. In order to identify both new constituents of the circadian clock and new alleles of known clock-associated genes, we performed a mutant screen. Using a clock-regulated luciferase reporter, we isolated new alleles of ZEITLUPE ( ZTL ), LATE ELONGATED HYPOCOTYL ( LHY ) and GIGANTEA ( GI ). GI has previously been reported to function in red light signaling, central clock function, and flowering time regulation. Characterization of this and other GI alleles has helped us to further define GI function in the circadian system. We found that GI acts in photomorphogenic and circadian blue light signaling pathways and is differentially required for clock function in constant red versus blue light. Gene expression and epistasis analyses show that TOC1 expression is not solely dependent upon GI and that GI expression is only indirectly affected by TOC1, suggesting that GI acts both in series with and in parallel to TOC1 within the central circadian oscillator. Finally, we found that the GI-dependent promotion of CONSTANS ( CO ) expression and flowering is intact in a gi mutant with altered circadian regulation. Thus GI function in the regulation of a clock output can be biochemically separated from its role within the circadian clock. dark (~100 µ mol m -2 s -1 ) or short days of 8 hours light and 16 hours dark (~200 µ mol m -2 s -1 ) provided by cool white fluorescent bulbs (Sylvania and Phillips). After germination flats were weeded allowing only one plant per pot and monitored daily for bolting. When a one centimeter bolt was present, the number of rosette leaves was noted. Total RNA with and 3 g each used for cDNA synthesis from oligo-dT (18) with SuperScript II Reverse Transcriptase (Invitrogen) following manufacturers Real time quantitative RT-PCR was performed using an iCycler (Bio-Rad) in 40 mM Tris HCL pH 8.4, 100 mM KCl, 6 mM MgCl2, 8% Glycerol, 20 nM 0.4X SYBR Green I (Molecular Probes), 1XBSA (New England Biolabs), 1.6 mM dNTPs, 2.5 µ M each primer, and 10% diluted cDNA using Taq polymerase. Samples were run in duplicate and starting quantity was estimated from critical thresholds compared to the standard curve of amplification. Data presented are normalized to PP2a expression level. All primer sets contain one primer which bridges an to melt curve analysis was performed following and control See supplemental for supplemental protocols for primer sequences). We found 30 cycles for FT and 20 cycles for UBQ-10 to be within the log-linear phase of amplification on a template dilution series. RT-PCR products were visualized on agarose gels with ethidium bromide staining and quantified using ImageQuant (GE Healthcare) software. CO expression was monitored by qRT-PCR as described above (see supplemental protocols for primer sequences).

CCR2::LUC seedlings with EMS, screened 10,000 M2 plants for alterations in period length ( Figure S2), and identified mutant lines for further study. Prior to further phenotypic analysis, confirmed mutants were backcrossed four to five times to the parental Col CCR2::LUC strain to remove extraneous EMS-induced mutations. Mutant M3 plants were also outcrossed to Landsberg erecta (Ler) to establish populations for SSLP mapping (Lukowitz et al., 2000, Jander et al., 2002.

Isolation of genes responsible for short-and long-period phenotypes
The mapping population of a short period line demonstrated strong linkage to the top of chromosome I. Because of the proximity to the known clock locus LHY and the similar short period phenotype observed for lhy-20 (Michael et al., 2003b), we sequenced the LHY locus in the mutant and identified a nonsense mutation, C2136T, predicted to cause a stop codon 359 amino acids after the translation start site ( Figure S3). If translated, this would result in a truncated protein containing the myb-like DNA-binding domain but missing the C-terminal half of the wild-type gene product. The possibility that this mutation was responsible for the short-period phenotype was confirmed by lack of complementation by lhy-20 (Table SI) and we therefore designated the nonsense allele lhy-100. Like lhy-20, period length is shortened in lhy-100 in constant red or blue light as well as in the dark (Table SII), and lhy-100 exhibits a normal period response to increased fluence (data not shown).
Five long-period mutants demonstrated strong linkage to the bottom of chromosome V, near the ZTL locus. ZTL mutants have been previously reported to have a long-period 9 amino acids past the translation initiation site at the start of the kelch repeats. In addition, we have characterized ztl-105 (SALK-069091), which contains a T-DNA insertion located near the beginning of the kelch repeats ( Figure S3). ztl-105 failed to complement each of the EMS alleles (Table SI). ZTL protein can be detected at near normal levels in ztl-101 and ztl-102 but is undetectable in ztl-103 and ztl-105 (D. Somers, personal communication). As previously described for almost all ztl mutants (Somers et al., 2000;Somers et al., 2004;Kevei et al., 2006), the alleles we characterized exhibited a steeper fluence response curve in red light than wild type (data not shown), indicating that all our alleles have defects in red light signaling to the clock.
An additional short period mutant (Table I and Figure 1) mapped to a 63kb region on chromosome I encompassing the GI locus. Since GI has previously been implicated in clock function (Fowler et al., 1999;Park et al., 1999;Mizoguchi et al., 2005;Gould et al., 2006), the GI locus was sequenced as the most likely candidate gene within the region. A mutation (G3704A) was found that is predicted to induce a serine to alanine (S932A) amino acid change ( Figure S3). This serine residue is conserved across monocots and dicots (Edwards et al., 2005), suggesting it may play an important role in GI function. Since the short-period phenotype of this mutant was not complemented by the T-DNA allele gi-201 (SALK-092757) (Table SI) Figure 5D and data not shown), suggesting this is likely to be a null allele. Gene expression phenotypes similar to those seen for CCR2::LUC activity were also observed when we examined CCR2 mRNA levels by qRT-PCR in these gi mutants ( Figure 1D).
Luciferase activity was greatly reduced in the T-DNA mutant gi-201 ( Figure 1A); however, nearly normal levels of CCR2 mRNA could be detected ( Figure 1D), indicating that CCR2 expression is not damped low in these plants and that luciferase activity is reduced through an unknown mechanism. For more accurate measurements, we monitored the luciferase activity of clusters of gi-201 plants and found that CCR2::LUC rhythms were lower amplitude, damped more rapidly, and had a slightly altered period relative to wild-type ( Figure 1 and Table I). These phenotypes are similar to gene expression phenotypes in the likely null alleles gi-2 and gi-11 (Park et al., 1999;Gould et al., 2006) further supporting gi-201 as a null allele.
Since it had previously been reported that gi-1 mutants displayed a low amplitude phenotype in constant white light (Fowler et al., 1999;Park et al., 1999;Sothern et al., 2002;Tseng et al., 2004;Mizoguchi et al., 2005) but not in constant darkness (Park et al., 1999), whereas gi-3 rhythms are low amplitude in both conditions (Mizoguchi et al., 2005), we examined the light-dependence of the gi-200 phenotype. Similar to previous reports for gi -1 (Park et al., 1999), the severity of the period and amplitude phenotypes of gi-200 plants was reduced when they were maintained in DD (Table I and data not shown), however expression of CCR2::LUC still cycled with a short period.
To investigate the previous reports of variability in period phenotypes in gi-2 mutants (Park et al., 1999), we examined rhythmic CCR2::LUC expression in a variety of light conditions. In constant red light, the rhythmic amplitude of CCR2::LUC activity in gi-200 plants was significantly lower than that of wild type (p=0.0013) (Figure 2A), however, in constant blue light there was no difference in amplitude (p=0.72) ( Figure   2B). Interestingly, when held in red plus blue light, allowing for light perception through multiple pathways, gi-200 rhythms had an increased amplitude when compared to Col (p=0.0013) ( Figure 2C). Thus the gi-200 clock phenotype is wavelength-dependent, suggesting GI acts in both red and blue light input pathways to the clock and that these pathways may be differentially compromised by the gi-200 mutation. In contrast, the gi-constant red, constant blue, or the combination of constant red and blue light ( Figure 1).
Interestingly, gi-201 displayed a slightly long period phenotype in constant red light, but had a short period in constant blue light or the combination of red and blue light (Table   I), further supporting functionally distinct roles for GI in blue and red light signaling.
Although the circadian clock does not require light input to function, higher fluences of light cause the Arabidopsis clock to run with a shorter period, as is true for many other organisms (Aschoff, 1979). Because gi-1 was reported to demonstrate a lack of response to increasing red light (Park et al., 1999), we determined whether light input to the circadian clock was altered in the gi-200 mutant by examining the free-running period under various intensities of light. As previously reported (Somers et al., 1998a;Devlin and Kay, 2000), higher fluences of monochromatic red or blue light caused the clock to run with a shorter period in wild-type plants ( Figure 3). In contrast, there was no statistical difference (by one-way ANOVA or Student's t-test) in the period length of gi-200 plants when grown in constant red, blue, or red plus blue light across a wide fluence range ( Figure 3). Additionally, a significant interaction between genotype and fluence could be detected (p<0.05 in red and p<0.01 in both blue and red plus blue; ANOVA for general linear fixed-effects model), indicating the response to both red, blue, and the combination of red plus blue is altered in gi-200 relative to wild type. Our study extends the role of GI beyond red light signaling to indicate a function for GI in blue light input to the clock.
It has previously been reported that gi-1, gi-2 and gi-100 exhibit elongated hypocotyls in red light at all fluences tested and that gi-3 plants are tall when grown in blue light at elevated temperature (Huq et al., 2000;Paltiel et al., 2006). To further investigate the role of GI in photomorphogenesis, we examined hypocotyl elongation in many GI alleles under constant light of a variety of wavelengths and fluences. All alleles tested, in both the Col and Ler backgrounds, had significantly elongated hypocotyls (p<0.02 by constant darkness (data not shown). The degree of significance varied across a wide range of fluences, but the same trend of long hypocotyls was observed in all wavelengths of constant light examined ( Figure S4). This tall hypocotyl phenotype of these six GI alleles in blue light further supports the role for GI in blue light signaling.

Complex interaction between GI and TOC1
Both GI and TOC1 have been suggested to function positively in regulation of CCA1 and LHY transcription (Fowler et al., 1999;Park et al., 1999;Alabadí et al., 2001;Mizoguchi et al., 2005). We therefore examined expression of CCA1 and LHY in gi mutants. Both We also examined GI expression in the strong loss-of-function allele toc1-2. Since it has previously been reported that overexpression of TOC1 causes GI message levels to damp to low levels (Makino et al., 2002), it was of interest to observe that GI message levels were very similar to wild type in toc1-2 mutants, although they revealed the expected short-period phenotype (Figures 5D). A recent model has suggested that TOC1 negatively regulates GI expression, leading to the prediction that GI levels would damp high in a toc1 loss-of-function mutant (Locke et al., 2005, Figures S5D). In contrast, our data suggest that TOC1 does not directly regulate GI expression levels.
To further examine the relationship between TOC1 and GI, we generated double gi-200 toc1-2 mutants and compared their phenotypes to those of the single mutants. We observed that in both monochromatic red and blue light the short period phenotypes of gi-200 and toc1-2 were nearly additive in the double mutant ( Figures 6A and 6B and Table   I). Similarly, the tall hypocotyl phenotype seen in both single mutants in constant red light was more exaggerated in the double mutant (data not shown). Furthermore, gi-201 toc1-2 mutants are largely sterile, a phenotype not present in either single mutant (data not shown). Thus a number of phenotypes are more extreme in the double mutant than in either single mutant, suggesting the two proteins have parallel functions rather than acting solely in series with each other. TOC1 is targeted for degradation by the F-box protein ZTL; in ztl mutants, there is a sustained accumulation of TOC1 protein and a consequent increase in free-running period length (Más et al., 2003b). Consistent with the regulated degradation of TOC1 being an important function of ZTL, the short-period phenotype of a toc1 loss-of-function allele is epistatic to the long-period ztl-1 phenotype in plants mutant for both genes (Más  Table I). Additionally, we found that the hypocotyl lengths of the double mutants in constant red light were slightly shorter than Col, but taller than ztl-105 (data not shown), again demonstrating an additive rather than epistatic effect of the two mutations. Thus unlike TOC1, a GI allele is not epistatic to ZTL, further suggesting that TOC1 and GI may act in parallel.

GI functions in the central clock and in regulation of a clock output are biochemically separable
Previously-characterized lesions in GI cause a near complete loss of photoperiodism, resulting in late flowering in inductive photoperiods (Rédei, 1962;Koornneef et al., 1991;Araki and Komeda, 1993;Park et al., 1999;Huq et al., 2000). Consistent with these results, the T-DNA allele gi-201 flowered late in long days (LD) but with the normal number of leaves in short days (SD) ( Figure 7A). But to our surprise, we found that under inductive photoperiods gi-200 flowered normally, and actually flowered earlier than wild type under SD ( Figure 7A). Therefore, although gi-200 mutants have a similar clock phenotype to the previously characterized gi-1 and gi-3 alleles, gi-200 plants completely lack the late-flowering phenotype that gave GI its name.  Figures 7C and 7D). Thus a coincidence between light and CO expression in both toc1-2 and gi-200 in SD is correlated with high FT expression and early flowering. The flowering phenotype in gi-200 mutants is therefore due to a change in the phase in CO expression and thus results from the short-period clock phenotype of these plants, rather than being caused directly by a change in CO expression levels. This indicates gi-200 retains its normal biochemical function in CO regulation but lacks the ability to properly regulate circadian timing, demonstrating that the roles of GI in clock function and CO regulation are separable.

GI functions in red and blue light signaling
GI has previously been reported to be involved in red light signaling, both in photomorphogenesis and input to the circadian clock (Park et al., 1999;Huq et al., 2000).  , 1993;Lin et al., 1998;Devlin and Kay, 2000), it is likely that GI is involved in cryptochrome signaling. However, the flattening of the fluence response curve to blue light seen in gi-200 ( Figure 3) was not observed in cry1 cry2 double mutants (Devlin and Kay, 2000), indicating that the gi light input defects are not solely due to a decrease in blue light signaling through cryptochrome.
GI has been proposed to act within the central clock, either as a positive regulator of TOC1 expression or in parallel with TOC1 (Locke et al., 2005;Mizoguchi et al., 2005, Gould et al., 2006. Like gi mutants, toc1 loss-of-function alleles also have long hypocotyls in red light (Más et al., 2003a). However, the semi-dominant short period toc1-1 mutant has no hypocotyl phenotype in red light despite its circadian phenotype (Somers et al., 1998b). These data suggest TOC1 plays biochemically distinct roles in phytochrome signaling and central clock function. Since gi mutants have clock (but not hypocotyl) phenotypes even in DD (Park et al., 1999;Mizoguchi et al., 2005; Table I, and data not shown), GI's functions in red and blue light signaling are likely distinct from its role in the central circadian oscillator. Notably, all gi mutants examined had similar hypomorphic hypocotyl phenotypes (Figures 4 and S4) but disparate and often antimorphic circadian phenotypes (Table I; Park et al., 1999;Mizoguchi et al., 2005). In particular, we found that the long-period clock phenotype of the likely null gi-201 was recessive while the short-period phenotype of the EMS allele gi-200 was semi-dominant (Table SI), consistent with the short-period phenotype of plants overexpressing GI (Mizoguchi et al., 2005). Taken together, these data support the hypothesis that the biochemical roles of GI in light signaling and the circadian clock are separable (Mizoguchi et al., 2005). The observation that gi but not toc1 mutants have hypocotyl phenotypes in blue light (Más et al., 2003a;Figures 4 and S4) also suggests that in at least blue light signaling GI and TOC1 act independently of each other.

GI acts within the central clock
Rhythmic luciferase activity quickly damped in gi-201 plants transferred to constant light, a more extreme phenotype than observed in the gi-200 mutant ( Figure 1) and consistent with the suggestion that GI may act in the central clock along with TOC1, CCA1 and LHY (Locke et al., 2005;Mizoguchi et al., 2005). TOC1 and GI are both evening-phased genes whose expression is negatively regulated by CCA1 and LHY (Fowler et al., 1999;Alabadí et al., 2001). Further analogies with TOC1 are suggested by the wavelength-dependent phenotype of gi-200 plants: like toc1 loss-of-function mutants (Más et al., 2003a), gi-200 demonstrates a more severe clock phenotype in constant red than in constant blue light (Figures 1, 2, 6A, and 6B). Finally, both TOC1 and GI are required for normal peak levels of CCA1 and LHY expression (Figures 5A and 5B, Fowler et al., 1999;Park et al., 1999;Alabadí et al., 2001;Mizoguchi et al., 2002).
This and other data was incorporated into a recent model of the plant circadian oscillator.
In this model, a component termed 'Y' positively regulates TOC1 expression, and the expression of 'Y' is in turn negatively regulated by TOC1, LHY, and CCA1. It has been proposed that GI represents all or part of component 'Y' (Locke et al., 2005) cannot constitute all of the 'Y' activity, since TOC1 mRNA levels do not show a marked reduction in the strong gi-201 mutant ( Figure 5C). However, the apparent damping of TOC1 levels towards the median in this mutant is consistent with GI providing a portion of 'Y' activity ( Figures S5A and S5B). On the other hand, a different prediction of the two-loop model is not supported by our data. The model predicts that a strong loss-offunction toc1 allele would cause expression of 'Y' to be high and arrhythmic (Locke et al., 2005, Figure S5D), but we found GI transcript levels were very similar in toc1-2 and wild-type plants (Figures 5D and S5C). This suggests that either the decreased GI expression seen in TOC1 overexpressing plants (Makino et al., 2002) is a non-specific effect or that another factor acts redundantly with TOC1 to inhibit GI expression.
The circadian phenotype of gi-200 toc1-2 double mutants is more severe than that observed in either single mutant and is in fact nearly additive ( Figures 6A and 6B and Table I). Since neither allele is a null, this result must be interpreted with caution.
However, combined with the observation that overexpression of TOC1 causes a long period while overexpression of GI causes a short period phenotype (Más et al., 2003a;Mizoguchi et al., 2005), the extreme short-period phenotype in the double mutant suggests that GI has functions in the circadian clock independent of regulation of TOC1 expression. This possibility is reinforced by the observation that CCA1 and LHY message levels are reduced in both gi-200 and gi-201 ( Figures 5A and 5B) despite the different period phenotypes of these mutants (Table I and Figure 1). Furthermore, 35S::GI plants demonstrate near wild type levels of CCA1 and LHY despite their short period phenotype (Mizoguchi et al., 2005). Since loss of LHY or CCA1 function causes a shortened period (Green and Tobin, 1999;Michael et al., 2003b), these data suggest that GI plays a role within the central clock that is independent of the TOC1/CCA1/LHY feedback loop. We therefore propose that GI acts both in series and in parallel to TOC1 within the central oscillator.
Further distinctions between TOC1 and GI were revealed by epistasis analysis with ZTL.
It therefore seems unlikely that accumulation of GI protein contributes significantly to the ztl long-period phenotype. This conclusion is underscored by the observation that plants that constitutively overexpress GI have a short period, in contrast to the long period seen in ztl mutants (Mizoguchi et al., 2005;Somers et al., 2000). These data suggest that GI is not a substrate for ZTL-mediated degradation and that the dark-induced proteolysis observed for GI (David et al., 2006) is likely directed by another mechanism.
GI has separable functions within the circadian clock and in the regulation of a clock output GI has been implicated in flowering time regulation, red light signaling, and central clock function (Rédei, 1962;Koornneef et al., 1991;Araki and Komeda, 1993;Huq et al., 2000;Locke et al., 2005;Mizoguchi et al., 2005;Gould et al., 2006).   Table I). However, two mutants with premature stop codons near the C-terminus (gi-1 and gi-3) both have short-period phenotypes (Park et al., 1999;Mizoguchi et al., 2005), similar to plants overexpressing GI (Mizoguchi et al., 2005). This suggests that the N-terminus of GI acts to stimulate the pace of the clock and that the C-terminus may function to block N-terminal activity. Alteration of a C-terminal residue in gi-200 leads to a semi-dominant short-period phenotype, suggesting that the Ser 932 A mutation impairs the putative negative regulatory role of the C-terminus.

Conclusions:
The isolation and characterization of new alleles of known clock-associated genes has provided insights into the functioning of GI in the circadian system. GI has a wavelength-dependent role in circadian clock function, acting in both red and blue light signaling to the clock. Characterization of a missense allele revealed that its action in the clock is biochemically distinct from its regulation of the flowering time pathway. In fact, it may be that most plant clock proteins also act in non-clock dependent processes (Más et al., 2003b;Kevei et al., 2006). Further biochemical characterization of GI function will shed light on the mechanisms underlying its diverse roles in plant signaling.

Mutagenesis and screening
Plants of the Columbia-0 ecotype were transformed with the CCR2::LUC reporter construct (Strayer et al., 2000) which confers gentamycin resistance. Seeds homozygous for CCR2::LUC were mutagenized with EMS by soaking two grams of seeds in 0.25% EMS at 21°C for 19 hours. Approximately 10,000 M1 plants were grown in pots of about 100 plants, each pot producing one pool of M2 seeds. M2 seeds were sterilized and plated on MS containing 3% sucrose and stratified for three to four days at 4°C before release to 12:12 light:dark cycles. Five days post-germination, seedlings were transferred to 96-well Packard plates and entrained for 3 more days. Seedlings were then assayed in constant darkness in a Packard multi-well scintillation counter for 6 days. See Harmer and Kreps (2001) for more details.
Data collected from the assay was analyzed for rhythmicity following the luciferase activity analysis methods found in Plautz et al. (1997) Mutations were confirmed with CAPS markers according to supplemental protocols.
Seeds were plated on MS (MP Biomedicals) 0.7% agar (Sigma A1296) plates containing 3% sucrose (EMD chemicals) and 75mg/L gentamycin (EMD chemicals) and stratified at 4°C for 4 days before 12:12 light/dark entrainment at 22°C. After 6 days in light/dark cycles plants were sprayed with 3mM D-luciferin (Biosynth AG) and monitored with a cooled CCD camera (either an ORCA II ER (Hamamatsu) or a DU434-BV (Andor Technology)). Plants were grown at a constant 22°C. Light was provided by red and/or blue LED SnapLites (Quantum Devices) or cool white fluorescent bulbs. Neutral density filters (RoscoLux #98 and #398) were used to obtain the different fluence levels for the period length fluence response curves. Images were analyzed with MetaMorph (Molecular Devices) software and the pattern of luciferase activity was fit to a cosine wave through Fourier Fast Transform-Non-Linear Least Squares (FFT-NLLS) (Plautz et al., 1997) allowing for estimates of period length, amplitude, and phase.

Hypocotyl analysis
Hypocotyl length was assayed by sowing seeds on MS-agar plates containing 3% sucrose before stratification at 4°C for 4 days. After a four hour white light pulse (100 µmol m -2 s -1 ), seeds were released to constant white light (cool white fluorescent bulbs, Sylvania and Phillips), or constant red or blue light (LED SnapLites, Quantum Devices) at 22°C under neutral density filters (RoscoLux #98 and #398) for 6 days before collection. On day 6 seedlings were transferred to transparencies and scanned. Individual measurements were obtained from the scans using the Seed Vigor Index System (Hoffmaster et al., 2003) with program modifications implemented by Kikuo Fujimura and Lijie Xu (manuscript in preparation).

Flowering time analysis
Seeds were stratified in water for 4 days before sowing directly to soil. Flats were then placed in either long days of 16 hours light and 8 hours dark (~100 µmol m -2 s -1 ) or short days of 8 hours light and 16 hours dark (~200 µmol m -2 s -1 ) provided by cool white fluorescent bulbs (Sylvania and Phillips). After germination flats were weeded allowing only one plant per pot and monitored daily for bolting. When a one centimeter bolt was present, the number of rosette leaves was noted.

qRT-PCR assay for expression of CCA1, LHY, TOC1, GI, CCR2, and PP2a
Approximately 40 plants of each genotype for each time point were germinated on Whatman filter paper atop MS-agar, 3% sucrose plates and entrained in 12:12 light/dark cycles under cool white fluorescent bulbs (55 µmol m -2 s -1 ) for 8 days before release to constant light (55 µmol m -2 s -1 ) and sample collection at three hour intervals. Total RNA was prepared with TRIzol reagent (Invitrogen) and 3µg each were used for cDNA synthesis from oligo-dT (18) with SuperScript II Reverse Transcriptase (Invitrogen) following manufacturers protocol. Real time quantitative RT-PCR was performed using an iCycler (Bio-Rad) in 40 mM Tris HCL pH 8.4, 100 mM KCl, 6 mM MgCl2, 8% Glycerol, 20 nM fluorescein, 0.4X SYBR Green I (Molecular Probes), 1XBSA (New England Biolabs), 1.6 mM dNTPs, 2.5 µM each primer, and 10% diluted cDNA using Taq polymerase. Samples were run in duplicate and starting quantity was estimated from critical thresholds compared to the standard curve of amplification. Data presented are normalized to PP2a expression level. All primer sets contain one primer which bridges an intron to reduce genomic amplification, melt curve analysis was performed following amplification to confirm specificity of products over primer dimers, and a no RT control was used to ensure products detected were from cDNA rather than genomic. See supplemental protocols for primer sequences.
FT and CO expression analysis cDNA samples were generated as for qRT-PCR analysis except that plants were entrained in short days (8:16) under cool white fluorescent bulbs (200 µmol m -2 s -1 ) for 8 days before collection. cDNA samples were diluted 1:5 for FT expression and 1:20 for CO expression. FT expression was monitored by PCR using FT and UBQ10 specific primers (see supplemental protocols for primer sequences). We found 30 cycles for FT and 20 cycles for UBQ-10 to be within the log-linear phase of amplification on a template dilution series. RT-PCR products were visualized on agarose gels with ethidium bromide staining and quantified using ImageQuant (GE Healthcare) software. CO expression was monitored by qRT-PCR as described above (see supplemental protocols for primer sequences).

Supplemental Material:
Supplemental Protocols: See this section for a description of the CAPS markers used to genotype the EMS alleles described in this study, primer sequences for used for genotyping lines obtained from ABRC, the sequences of primers used in qRT-PCR experiments, and the sequences of primers used for semi-quantitative RT-PCR, and supplemental references.

EMS allele CAPS markers
EMS alleles generated through this screen can be detected with CAPS or dCAPS markers. ztl-100, ztl-101, and ztl-104 were amplified with CAPS primers F-5'-AGC AAG GTT TGC TGA ACG AT-3' and R-5'-CAG GCA GCT GGA ATC TCT CT-3' followed by digest with AccI (New England Biolabs); wild type produces a 238bp band and the mutant produces two bands of 186bp and 52bp. ztl-102 can be detected with Table I -Phenotypes in different light conditions at 22 o C. Variance weighted mean period ± variance weighted standard error (n) was calculated for the period of the freerunning rhythms of CCR2::LUC expression. The indicated genotypes were entrained as described in Figure 1 and then transferred to the indicated light conditions. Period estimates were generated using the method described by Plautz et al. (1997) and statistical comparisons were made using a two-tailed Student's t-test.

Dark
Red    conditions. Plants were entrained as described in Figure 1 and then released to constant red, blue, or red plus blue light. Luciferase activity was monitored in (A) constant red light of 55 µmol m -2 s -1 (n=11-12), (B) constant blue light of 12.5 µmol m -2 s -1 (n=11-12), and (C) the combination of constant red and blue light of the above fluence rates, delivering a total fluence rate of 67.5 µmol m -2 s -1 (n=15). cRL = constant red light, cBL = constant blue light, and cR+BL = constant red plus blue light. Amplitude was determined using FFT-NLLS as described in Plautz et al., (1997). Data are indicative of at least two independent assays. Statistical comparisons were made using a two-tailed Student's t-test.