CIRCADIAN CLOCK ASSOCIATED1 and LATE ELONGATED HYPOCOTYL function synergistically in the circadian clock of Arabidopsis.

The circadian clock is an endogenous mechanism that coordinates biological processes with daily and seasonal changes in the environment. Heterodimerization of central clock components is an important way of controlling clock function in several different circadian systems. CIRCADIAN CLOCK ASSOCIATED1 (CCA1) and LATE ELONGATED HYPOCOTYL (LHY) are Myb-related proteins that function in or close to the central oscillator in Arabidopsis (Arabidopsis thaliana). Single mutants of cca1 and lhy have a phenotype of short-period rhythms. cca1 lhy double mutants show an even shorter period phenotype than the cca1 single mutant, suggesting that CCA1 and LHY are only partially functionally redundant. To determine whether CCA1 and LHY act in parallel or synergistically in the circadian clock, we examined their expression in both light-grown and etiolated seedlings. We have shown that LHY and CCA1 bind to the same region of the promoter of a Light-harvesting chlorophyll a/b protein (Lhcb, also known as CAB). CCA1 and LHY can form homodimers, and they also colocalize in the nucleus and heterodimerize in vitro and in vivo. In Arabidopsis, CCA1 and LHY physically interact in a manner independent of photoperiod. Moreover, results from gel filtration chromatography indicate that CCA1 and LHY are present in the same large complex in plants. Taken together, these results imply that CCA1 and LHY function synergistically in regulating circadian rhythms of Arabidopsis.

Circadian clocks are autoregulatory, endogenous mechanisms that control numerous physiological and molecular processes in organisms ranging from cyanobacteria to humans. In plants, leaf movements, stomata opening, hypocotyl elongation, and the expression of a large number of genes show circadian rhythms (Harmer and Kay, 2000;Edwards et al., 2006;Yakir et al., 2007;Michael et al., 2008). Although circadian clock components are divergent across the kingdoms, most organisms share a conserved mechanism of generating rhythmicity: a central oscillator based on transcriptional negative feedback loops (Dunlap, 1999). In Arabidopsis (Arabidopsis thaliana), three genes have been suggested as core components of the central oscillator: CIRCADIAN CLOCK ASSO-CIATED1 (CCA1), LATE ELONGATED HYPOCOTYL (LHY), and TIMING OF CAB EXPRESSION1 (TOC1). CCA1 and LHY bind directly to the promoter of TOC1, negatively regulating TOC1 expression, and TOC1 participates in the positive regulation of CCA1 and LHY expression through an unknown mechanism (Alabadí et al., 2001). Other genes, such as LUX ARRHYTHMO, also known as PHYTOCLOCK1, GI-GANTEA (GI), EARLY FLOWERING3 (ELF3), ELF4, TIME FOR COFFEE, and PSEUDORESPONSE REGU-LATOR3/5/7/9 (PRR3/5/7/9), have also been suggested to function in or close to the central oscillator (Doyle et al., 2002;Hazen et al., 2005;Locke et al., 2005;Nakamichi et al., 2005;Edwards et al., 2006;Gould et al., 2006;Ding et al., 2007;McWatters et al., 2007). CCA1 and LHY are closely related transcription factors that contain a single Myb domain (Wang et al., 1997;Schaffer et al., 1998). They both have circadian rhythms of expression peaking soon after dawn. cca1 and lhy single mutants have short-period phenotypes, and overexpression of either gene causes arrhythmicity in expression of clock-regulated genes, leaf movement, and hypocotyl elongation and leads to dramatically reduced levels of endogenous CCA1 and LHY transcripts (Schaffer et al., 1998;Wang and Tobin, 1998;Green and Tobin, 1999;Alabadí et al., 2002;Mizoguchi et al., 2002). Therefore, CCA1 and LHY negatively regulate their own and each other's expression.
However, regulation of clock genes is not restricted to the level of transcription; a multitude of protein interactions are required for the generation and maintenance of rhythmicity within the central oscillator. Perhaps the best-characterized example is the Drosophila clock in which PERIOD (PER) protein interacts with TIMELESS (TIM) and inhibits their own tran-scription by repressing the activity of dCLOCK (dCLK) and CYCLE (CYC; Gekakis et al., 1995;Meyer et al., 2006). The dCLK/CYC heterodimer not only activates PER and TIM transcription but also the transcription of VRILLE and PAR DOMAIN PROTEIN1, which encode negative and positive regulators of CLK transcription, respectively (for review, see Hardin, 2004). Recent discoveries demonstrate that protein interactions are also important for proper clock function in Arabidopsis. The F-box protein ZEITLUPE (ZTL) binds TOC1 and causes it to be degraded through a CULLIN1containing SCF (for Skp1-Cullin-F-box protein) complex (Más et al., 2003;Harmon et al., 2008). ZTL also regulate PRR5 degradation through direct binding (Kiba et al., 2007). TOC1/ZTL interaction can be competed by the TOC1/PRR3 interaction and possibly prevents TOC1 degradation in the vascular tissues (Para et al., 2007;Fujiwara et al., 2008). In addition, ZTL protein is stabilized by direct interaction with GI in blue light (Kim et al., 2007).
In this work, we show that CCA1 and LHY have similar expression patterns and that they can form homodimers and heterodimers in vitro and in vivo. CCA1 and LHY also colocalize in the nucleus and physically interact in plants. Furthermore, we find that they are present in the same large complex. These observations suggest that CCA1 and LHY function synergistically in regulating the circadian rhythms of Arabidopsis.

RESULTS
A cca1 lhy Double Mutant Has More Severe Phenotypes Than a cca1 Single Mutant cca1 and lhy single loss-of-function mutants confer short periods (Green and Tobin, 1999;Alabadí et al., 2002;Mizoguchi et al., 2002) and an initial study showed that a cca1 lhy double loss-of-function mutant (cca1-1 lhy-11) is unable to maintain rhythms of clockcontrolled RNAs for more than a few cycles under constant light (LL; Alabadí et al., 2002;Mizoguchi et al., 2002). However, the lhy allele used in that study was not truly null, because a truncated LHY protein is still expressed (Supplemental Fig. S1). Therefore, it is possible that the truncated LHY retained partial clockassociated function in the cca1-1 lhy-11 double mutant. We therefore set out to test whether plants that are completely null for both the CCA1 and LHY proteins are able to sustain rhythmicity in LL. The CAB2::LUC (for LUCIFERASE) circadian reporter (Anderson et al., 1994;Knowles et al., 2008) was introduced into wildtype Wassilewskija (Ws), cca1-1 (a single-null mutant in the Ws background), and cca1-11 lhy-21 (a doublenull mutant in the Ws background) that is completely missing both proteins (Supplemental Fig. S1; Locke et al., 2005) seedlings by transformation (Clough and Bent, 1998). Figure 1 demonstrates that the CAB2::LUC rhythm in cca1-11 lhy-21 seedlings persists for at least 8 d in LL. Although some initial dampening (decreasing amplitude over time) was evident in the double mutant, the rhythm stabilized after three to four cycles. This is in contrast to reports that indicate rhythms dampen out completely in the double mutants (Mizoguchi et al., 2002;Locke et al., 2005). Compared to the cca1-1 single mutant, which displayed a short-period phenotype (24.77 6 0.07 h versus 26.41 6 0.16 h in wild type), the cca1-11 lhy-21 double mutant exhibits reduced amplitude as well as a very short period (19.73 6 0.12 h) of CAB2::LUC rhythm (Fig. 1). These results indicate that CCA1 and LHY are required for sustaining robust rhythms and proper period in LL; however, in their absence, other presumably closely related proteins can partially replace this function. The short period of the single cca1-1 mutant and the additive period phenotype of the cca1-11 lhy-21 double mutant indicate that CCA1 and LHY are only partially redun- Figure 1. The cca1-11 lhy-21 double mutant has a shorter period than the cca1-1 single mutant. A, Comparison of the CAB2::LUC activity in wild type (Ws) and cca1-1. B, Comparison of the CAB2::LUC activity in wild type (Ws) and cca1-11 lhy-21. CAB2::LUC seedlings in wild-type (Ws), cca1-1, and cca1-11 lhy-21 backgrounds were grown in 12L:12D for 8 d and then transferred to LL. Mean bioluminescence traces 6 SEM (n = 38-40) are shown. Each point was normalized to the average luminescence value for the entire run. Subjective day and subjective night are denoted by white and hatched bars, respectively. All of these experiments were done at least twice with similar results. WT, Wild type.
dant, in which case they might function in parallel and/or work together in regulating circadian rhythms.

CCA1 and LHY Have Similar Expression Patterns
Both CCA1 and LHY are expressed rhythmically with peaks of expression occurring soon after dawn in plants grown under diurnal conditions or transferred to LL (Schaffer et al., 1998;Wang and Tobin, 1998). To determine whether CCA1 and LHY function in parallel or synergistically, their tissue-specific expression was examined. Different Arabidopsis tissues from light-grown plants were harvested at 1 h after dawn (zeitgeber time [ZT]-1) when CCA1 and LHY are most abundant, and quantitative real-time PCR (qRT-PCR) was used to measure their levels of expression. Figure  2A shows that both CCA1 and LHY are expressed at higher levels in leaves and stems compared to roots, inflorescences, and siliques. Although CCA1 and LHY have similar expression patterns in different tissues, LHY is generally more highly expressed than CCA1 ( Fig. 2A).
In etiolated seedlings, CCA1 and LHY are expressed at very low levels, but light can induce CCA1 and LHY expression Martinez-Garcia et al., 2000). To examine whether LHY could be regulated by light and the circadian clock in a similar manner to CCA1 in etiolated seedlings, we compared the expression of CCA1 and LHY in etiolated seedlings treated with a brief exposure to red light (R). This experimental approach allows the acute response to light and the circadian response to be separated (Millar et al., 1992;Anderson and Kay, 1997). Arabidopsis seedlings were grown for 5 d in the dark, exposed to 2 min of R, and tissue was harvested at various times up to 28 h following R exposure. Consistent with previous reports Martinez-Garcia et al., 2000), both CCA1 and LHY mRNA levels increased within 0.5 h and peaked at 1 h after R treatment (Fig. 2B), demonstrating that light induces both LHY and CCA1 expression in etiolated seedlings. Following the acute response to light, the expression of both CCA1 and LHY mRNAs reached trough levels in approximately 8 h and rose transiently to peak at approximately 16 h after R treatment (Fig.  2B). The second peak is most likely due to the action of an endogenous circadian oscillator (Millar and Kay, 1996). This result demonstrates that both light and the circadian clock regulate the expression of CCA1 and LHY in a similar way in etiolated seedlings. Taken together, the results indicate that CCA1 and LHY have similar expression patterns in both light-grown plants and etiolated seedlings.
LHY and CCA1 Bind to the Same Region of the Light-harvesting chlorophyll a/b protein Promoter CCA1 and LHY show similar regulation by light and the circadian clock. In addition, CCA1 mediates activation of Light-harvesting chlorophyll a/b protein (Lhcb) genes by binding directly to their promoters (Wang et al., 1997). It has previously been shown that the Myb domain of CCA1 interacts with two closely spaced binding sites (292 to 2105 and 2111 to 2122 relative to the transcription start site) in the Lhcb1*3 promoter (Wang et al., 1997). The extensive similarity (87% identity) between the Myb domains of the CCA1 and LHY proteins prompted us to examine whether LHY can bind to the same regions of the Lhcb1*3 promoter as CCA1. Figure 3 shows the results from a representative electrophoretic mobility shift assay (EMSA) using Escherichia coli-produced CCA1 and LHY proteins and wild-type and mutant promoter fragments of Lhcb1*3 as probes. DNA-binding activities were observed for both proteins using the wildtype A2 fragment of the Lhcb1*3 promoter (Fig. 3A). This A2 fragment was used previously to characterize CCA1 binding ability (Wang et al., 1997). The mutated m1 fragment carries several point mutations spanning both CCA1 binding sites in the Lhcb1*3 promoter (Fig.  3C). These mutations were previously shown to abolish binding of CCA1 to the wild-type A2 fragment (Wang et al., 1997). As shown in Figure 3A, the altered nucleotides of the m1 fragment also completely abolished the binding of LHY.
To characterize the LHY-binding site on the Lhcb1*3 promoter in more detail, we carried out 1,10-phenanthroline-copper footprinting analysis. We . Five-day-old (5d) seedlings were grown on plates, and the other tissues were harvested from 4-week-old soilgrown plants. All samples were collected at ZT-1. The RNA was subjected to qRT-PCR analysis to measure levels of CCA1 and LHY relative to RH8 (an internal control). Reactions were performed in triplicate. Error bars denote 6 SD. B, CCA1 and LHY are expressed with similar kinetics in etiolated seedlings exposed to R. Time course of CCA1 and LHY expression in 5-d-old etiolated seedlings after exposure to 2 min of R. Gel blots of total RNA were hybridized with RNA probes of CCA1, LHY, and UBQ10 (control). All of these experiments were done at least twice with similar results.
have previously used this method to characterize CCA1 binding to the same fragment of the Lhcb1*3 promoter (Wang et al., 1997). After performing an EMSA to resolve the DNA-protein complexes, the gel was treated with 1,10-phenanthroline-copper, and the DNA in the two complexes (B1 and B2; Fig. 3A) was recovered and resolved in a sequencing gel. As shown in Figure 3B, in complex B1, the 296 to 2107 region was protected and in complex B2, regions from 292 to 2107 and from 2111 to 2130 were protected. This result suggests that the two complexes of different mobilities are a result of the presence of two separate binding sites on this fragment and that the 296 to 2107 region probably is the higher affinity binding site for LHY. A similar observation in which one complex has a larger protected region than the other was also made in CCA1 footprinting analysis (Wang et al., 1997). Figure 3C summarizes the results from the phenanthroline-copper footprinting and compares the regions of the wild-type A2 fragment protected by CCA1 and LHY. These results demonstrate that the two proteins bind to the same region of the Lhcb1*3 promoter.

CCA1 Interacts with LHY in Vitro
In Drosophila, CYC heterodimerizes with dCLK and activates the transcription of PER and TIM through binding to the E-boxes of the PER and TIM promoters (Shirasu et al., 2003). Because CCA1 and LHY bind to the same region of the Lhcb1*3 promoter, we investigated whether CCA1 heterodimerizes with LHY using a reciprocal interaction assay. Glutathione S-transferase (GST) fusions of CCA1 and LHY were immobilized on glutathione-agarose beads and incubated with [ 35 S]Met-labeled CCA1 or LHY synthesized by in vitro transcription-translation. LUC was used as a control. Figure 4A shows that in the presence of either promoter. An EMSA gel with LHYand the wild-type A2 fragment was treated with phenanthroline-copper to cleave DNA, and the DNA from each band was recovered and separated on a sequencing gel. The lines between or next to the lanes highlight the regions protected by the proteins. F, Free probe. B1 and B2 refer to the DNA-protein complexes containing LHY used in the footprint analysis. C, Nucleotide sequence of the wild-type and m1 probes. WT and m1 denote the sequences of the wild-type A2 fragment of the Lhcb1*3 promoter and of the m1 mutated fragment, respectively. Dashes in the sequence of m1 represent nucleotides identical to those of the wild-type fragment. Solid and dashed lines above the sequence of the wild-type fragment represent the regions protected by LHY and CCA1, respectively. The region protected by CCA1 is from Wang et al. (1997). All of these experiments were done at least twice with similar results. GST-CCA1 or GST-LHY, but not with GST alone, both CCA1 and LHY proteins synthesized in vitro were efficiently bound, while LUC was not bound by either of the two GST fusion proteins. These results demonstrate that CCA1 and LHY not only homodimerize but can also form a heterodimer in vitro.
To identify the region of CCA1 that is responsible for interacting with LHY, we carried out interaction assays using [ 35 S]Met-labeled LHY synthesized in vitro and various deletion constructs of CCA1 fused to GST. Figure 4B shows that in the presence of immobilized GST-CCA1 proteins lacking either the 316 or 436 N-terminal amino acids of CCA1 or a 472-amino acid portion from the C terminus, LHY could not be efficiently bound. Taken together, these data suggest that a region of CCA1 between amino acids 136 and 316 is important for interaction with LHY. It is worth mentioning that the LHY-interactive domain of CCA1 lies outside the DNA-binding Myb domain.

CCA1 and LHY Can Form Homodimers in Vivo
To examine the homodimerization of CCA1 in vivo, we carried out in vivo transient expression in Nicotiana benthamiana. Expression constructs of 35S::YFP (yellow fluorescent protein)-CCA1 and 35S::MYC-CCA1 were coinfiltrated into N. benthamiana and total protein was extracted from infiltrated leaves. Immunoprecipitation with anti-GFP antibody and subsequent detection of MYC-CCA1 using anti-MYC antibody (Fig. 5A) showed that anti-GFP antibodies not only immunoprecipitate YFP-CCA1, but also coimmunoprecipitate MYC-CCA1. To examine the homodimerization of LHY in vivo, similar immunoprecipitation was performed using anti-GFP antibodies with total protein extracts from N. benthamiana leaves coexpressing YFP-LHY and haemagglutinin (HA)-LHY. Figure 5B shows that HA-LHY is coprecipitated with YFP-LHY. These results indicate that both CCA1 and LHY can homodimerize in vivo.

CCA1 and LHY Colocalize in the Nucleus and Heterodimerize in Vivo
Because both CCA1 and LHY have been reported in the nucleus in plants (Wang et al., 1997;Carré and Kim, 2002), we first examined whether they colocalize in vivo. Expression constructs of 35S::YFP-CCA1 and 35S::LHY-CFP (cyan fluorescent protein)-HA were coinfiltrated into N. benthamiana. Confocal microscopy results (Fig. 5C) show that both YFP-CCA1 and LHY-CFP are localized in the nucleus and their expression overlaps. To verify the heterodimerization between CCA1 and LHY, MYC-CCA1 and HA-LHY were coexpressed in N. benthamiana. Immunoprecipitation with anti-LHY antibody and subsequent detection of MYC-CCA1 (Fig. 5D) showed that CCA1 interacts strongly with LHY in vivo. We extended these trials to test different regions of CCA1 and found that a region of CCA1 between amino acids 136 and 316 is important for interaction with LHY ( Fig. 5D) in vivo. Protein product from one of the CCA1 deletion constructs (35S::MYC-CCA1D316N) can only be detected by anti-CCA1 antibody, but not anti-MYC antibody (Fig. 5D), indicating that the MYC tag may be cleaved off from CCA1D316N protein. As expected, protein products from CCA1 C-terminal deletion constructs (35S::MYC-CCA1D270C and 35S::MYC-CCA1D472C) are not recognized by anti-CCA1 antibody, which is generated against the C-terminal portion of the CCA1 protein (Fig. 5D). Taken together, the results support the in vitro findings, demonstrating that CCA1 and LHY colocalize in the nucleus, and the region of CCA1 downstream of the Myb domain contains the determinants for LHY interactions in planta.

CCA1 and LHY Physically Interact in Arabidopsis
The physical interaction between CCA1 and LHY was also observed in wild-type (ecotype Columbia [Col]) Arabidopsis (Fig. 6A). Immunoprecipitation was performed using anti-LHY antibodies with total extracts from 2-week-old seedlings grown in 12 h light:12 h darkness (12L:12D) and harvested at ZT-1 and ZT-23 (1 h before dawn). Figure 6A shows that Glutathione-agarose beads containing approximately 1 mg of GST, GST-CCA1, or GST-LHY were mixed with 20 mL of 35 S-labeled CCA1, LHY, or LUC. Proteins interacting with GST, GST-CCA1, or GST-LHY were eluted and resolved in SDS-PAGE, followed by autoradiography. Input represents 10% of the 35 S-labeled protein used. B, The region between amino acids 136 and 316 of CCA1 is essential for interaction with LHY in vitro. The Myb domain of CCA1 is shown. The region of CCA1 filled with black and denoted with an asterisk indicates the 136-to 316-amino acid region that interacts with LHY. All of these experiments were done at least twice with similar results.

CCA1 coprecipitated with LHY and their interaction is not light dependent in Arabidopsis.
To further determine whether CCA1/LHY interaction is affected by photoperiod, wild-type Col Arabidopsis plants grown in 12L:12D were harvested at different times and a similar immunoprecipitation was conducted. The amount of LHY immunoprecipitated and the amount of CCA1 coimmunoprecipitated were correlated with the LHY and CCA1 levels in the total extract, suggesting that CCA1 can interact with LHY whenever they both are present. Similar results were obtained when Arabidopsis plants were grown in LL and harvested at different circadian times (data not shown). These data demonstrate that CCA1 and LHY physically interact in a manner independent of photoperiod in plants (Fig. 6).

CCA1 and LHY Are Found in a Large Complex
To examine whether the interaction of CCA1 and LHY is stable in plants, gel filtration chromatography was performed using total extract from wild-type (Col) Arabidopsis, which was prepared from 2-weekold seedlings grown in 12L:12D and harvested at ZT-1. We expected to observe heterodimers of about 150 kD. Surprisingly, we were unable to detect any monomers or heterodimers of CCA1 and LHY. CCA1 and LHY monomers are predicted to be 67 and 70 kD, respectively, but migrate at 75 and 80 kD, respectively, when subject to electrophoresis in denaturing conditions Daniel et al., 2004). We found that both CCA1 and LHY are present in a large complex (.440 kD; Fig. 7A).To determine whether CCA1 and LHY are present in the same complex, fractions 5, 6, and 7 were subjected to the immunoprecipitation with anti-LHY antibodies. CCA1 coprecipitated with LHY in those fractions, indicating that CCA1 and LHY are present in the same large complex (Fig. 7B). This observation suggests that CCA1 and LHY could function as part of a complex and work cooperatively with other proteins in regulating circadian rhythms.  12, and 19), 13,and 20), and MYC-CCA1D472C (lanes 7, 14, and 21). Input and supernatant after immunoprecipitation were run next to the immunoprecipitation product as a control. One hundred micrograms of total protein was analyzed as input. All of these experiments were done at least twice with similar results.

DISCUSSION
It has been shown that heterodimerization of central clock components is an important way of controlling clock function in other circadian systems (Dunlap, 1999). Examples include the interaction of PER and TIM in Drosophila as a means of controlling the nuclear entry and degradation of both proteins (Edery, 1999;Meyer et al., 2006), the interactions among Kai proteins in cyanobacteria (Iwasaki et al., 1999) and heterodimers of the transcription factors mCLOCK and BMAL1, and interactions among three mPER (PER1-PER3) and two CRYPTOCHROME (mCRY1 and mCRY2) proteins in mammals (Reppert and Weaver, 2002;Lowrey and Takahashi, 2004). In this study, we demonstrate that CCA1 and LHY have similar expression patterns, bind to a similar region of the Lhcb1*3 promoter, colocalize in the nucleus, and that they can physically interact both in vitro and in plants. These findings are consistent with the evidence showing the involvement of dimerization between core oscillator components in the mammalian system (Kwon et al., 2006) and in Drosophila (Meyer et al., 2006), suggesting that homodimerization and heterodimerization might be conserved aspects in the regulation of eukaryotic circadian clocks.
The observation that cca1-1 plants show persistent rhythms in LL, albeit with a shorter period length than wild type, has been attributed to the presence of functionally redundant components in the circadian system, particularly to the CCA1-homolog LHY (Green and Tobin, 1999). We have shown that in mutants lacking both CCA1 and LHY, rhythms are still generated and persist in LL, but with a reduced amplitude and shortened period length (Fig. 1). Similar results have been reported elsewhere (Locke et al., 2005). Hence, a compromised, but functioning oscillator continues to operate in the absence of the two proteins. It is known that CCA1 and LHY belong to a 10-member family of Myb-related proteins that have significant homology in their Myb domains (Yanhui et al., 2006). Moreover, some of the single Myb domain proteins in this family have been reported to be under circadian control (Somers, 1999;Strayer and Kay, 1999;Kuno et al., 2003;Zhang et al., 2007). One can postulate that in the absence of CCA1 and LHY, some or all of the eight remaining members can substitute for them to some extent so that some rhythmic expression is maintained.
The phenotype of the single mutants and the enhanced phenotype of cca1 lhy double mutants (Fig. 1) indicate that CCA1 and LHY are not entirely functionally redundant. One possibility is that CCA1 and LHY have distinct roles and work in parallel in the control of circadian rhythms. Alternatively, CCA1 and LHY may have similar functions, but either the dosage of their combined gene products or the interactions between them are crucial for their function. cca1 and lhy single mutants have similar phenotypes (Green and Tobin, 1999;Alabadí et al., 2002;Mizoguchi et al., 2002), and overexpression of each gene has similar effects on circadian rhythms (Schaffer et al., 1998;Wang and Tobin, 1998). Together with the evidence presented here that CCA1 and LHY have similar expression patterns (Fig. 2) and that LHY and CCA1 bind to the same region of the Lhcb1*3 promoter (Fig. 3), this suggests that CCA1 and LHY have closely related functions Figure 6. CCA1 and LHY can interact in plants and their interaction is independent of photoperiod. Western-blot analysis of immunoprecipitation (IP) with a-LHY antibody and detection with antibody to LHY and CCA1. A, CCA1 coimmunoprecipitates with LHY in plant extract. Two-week-old seedlings grown in 12L:12D were harvested at ZT-1 and ZT-23. B, CCA1 coimmunoprecipitates with LHY throughout the diurnal cycle. Two-week-old seedlings grown in 12L:12D were harvested at different times as indicated. One hundred micrograms of total protein was analyzed as input. Light and dark periods are denoted by white and black bars, respectively. All of these experiments were done at least twice with similar results. Shown is western-blot analysis of fractions collected from gel filtration and detection with antibody to LHYand CCA1. The arrow indicates the band of CCA1 and the asterisks indicate cross-reactive bands. Total (T) represents 5% of the protein extract used for gel filtration. B, CCA1 coimmunoprecipitates with LHY in fractions 5, 6, and 7. Protein extracts were prepared from 2-week-old seedlings grown in 12L:12D and harvested at ZT-1. All of these experiments were done at least twice with similar results. within the circadian system. It has been shown that proteins containing a single Myb domain could bind DNA as dimers (Jin and Martin, 1999). In Drosophila, the transcription factors CYC and dCLK are localized in the nucleus and activated by dimerization. The complex of CYC and dCLK binds to the E-boxes of the PER and TIM promoters and activates their expression (Shirasu et al., 2003). Our results showed that CCA1 and LHY not only homodimerize but also colocalize in the nucleus and form heterodimers in vitro and in vivo (Figs. 4 and 5). The interaction between CCA1 and LHY was confirmed in plants using coimmunoprecipitation with total plant extract of wild-type Arabidopsis (Fig. 6). In addition, we determined that the region of CCA1 between amino acids 136 and 316 is responsible for interacting with LHY (Figs. 4B and 5D). Although the functional significance of CCA1 and LHY dimerization has yet to be elucidated, one hypothesis is that homodimers and heterodimers may have different effects on nuclear entry, DNA binding, and/or protein degradation. In wild-type plants, heterodimers of CCA1 and LHY could be important for proper feedback inhibition and in single mutants of cca1 and lhy, CCA1 or LHY may form homodimers, which could maintain rhythmic oscillations but cause the feedback inhibition to occur faster and result in a short-period phenotype. This is consistent with the notion that CCA1 and LHY have different biochemical activities (Gould et al., 2006) and they are only partially redundant in the circadian system.
In an effort to detect CCA1/LHY heterodimers in plants, wild-type plant extracts were fractionated by gel filtration chromatography. Interestingly, both CCA1 and LHY are found in a large complex (Fig. 7A). This novel finding will open up new opportunities for future work. Because CCA1 and LHY physically interact (Fig.  7B), it is likely that they are present in the same complex. The calculated molecular mass of heterodimers of CCA1/LHY is about 150 kD, yet the CCA1/LHYcontaining complex is .400 kD (Fig. 7). Discovery of other components in the complex will significantly increase our knowledge of the functions of CCA1 and LHY and of the central oscillator of the circadian clock in Arabidopsis. To discover other components of the protein complex, transgenic plants expressing tandem affinity purification-tagged CCA1 have been generated and will be used to purify the complex.
In summary, our study shows that CCA1 and LHY not only homodimerize, but also colocalize in the nucleus and heterodimerize. They are also present in a protein complex, suggesting that CCA1 and LHY function synergistically in the control of circadian rhythms of Arabidopsis.

Plant Material and Growth Conditions
Arabidopsis (Arabidopsis thaliana) ecotype Col was used for all experiments described unless stated otherwise. Seeds were stratified for 3 d in the dark at 4°C and then sown onto either solid Murashige and Skoog medium (Murashige and Skoog, 1962) containing 1.5% agarose or soil. Seedlings were grown under a 12-h-fluorescent light (40 mmol m 22 s 21 ):12-h-dark photoperiod at a constant temperature of 22°C. Growth of etiolated seedlings and brief R treatments of seedlings were carried out as previously described (Wang et al., 1997).
PCR amplifications were performed using Pfu DNA polymerase (Stratagene) following standard PCR protocol. All vectors prepared were direct sequenced.

Recombinant Proteins
GST-CCA1 and GST-LHY fusion proteins were expressed in Escherichia coli and were purified as described previously (Sugano et al., 1998). The GST-CCA1 deletions were generated by deleting the appropriate fragment from pGEX-CCA1 (Sugano et al., 1998) and re-ligating the remaining portion of the construct (e.g. D86N represents 86-amino acid deletion from N terminus and D86C represents 86-amino acid deletion from C terminus, etc.).

EMSA and Footprinting
The wild-type A2 and mutant m1 probes have been described previously (Sun et al., 1993). CCA1 and LHY proteins were produced in E. coli as GST fusions and released from the GST portion by cleavage with factor Xa as described by Sugano et al. (1998). Each DNA-binding reaction included 0.3 ng Interaction of CCA1 and LHY Plant Physiol. Vol. 150, 2009 of 32 P-labeled DNA probe, 1 mg of bovine serum albumin, and 0.5 mg of poly (dI-dC). Labeled probes were incubated in the presence or absence of approximately 5 ng of recombinant CCA1 or LHY proteins for 15 min prior to electrophoresis. The EMSA buffer and electrophoresis conditions have been described previously (Sun et al., 1993). DNA-protein complexes were detected by autoradiography.
Phenanthroline-copper footprinting was carried out as described previously (Wang et al., 1997). The A2 fragment was labeled with 32 P at the 3#-end of the antisense strand (Sun et al., 1993). The EMSA reactions contained 1.5 3 10 6 cpm of end-labeled probe, 120 ng of Factor Xa-digested LHY protein, and 3 mg poly(dI-dC).
In Vitro Interaction Assay CCA1, LHY, and LUC proteins labeled with [ 35 S]Met were synthesized in vitro in 50 mL of rabbit reticulocyte lysate using a coupled transcription translation system (Promega). In vitro interaction assays were performed as described previously (Sugano et al., 1998).

Nicotiana benthamiana Infiltration
Nicotiana benthamiana leaves were agroinfiltrated as described previously (Voinnet et al., 2003;Kim et al., 2007). Agrobacterium tumefaciens strain C58C1 cells were transformed with the appropriate binary vectors using electroporation. A. tumefaciens positive clones were grown overnight and resuspended in infiltration buffer (10 mM MgCl 2 , 10 mM MES, and 100 mM acetosyringone) to reach a concentration of 0.8 OD 600 . This suspension was pressure-infiltrated into tobacco leaves using a syringe. Leaves were analyzed or harvested 2 d after infiltration.

Confocal Microscopy
All imaging was done using a Carl Zeiss 510 Meta laser scanning confocal microscope with a Plan-Apochromat 633/1.4 oil DIC objective. To image YFP and CFP together, we used multitracking in stack mode and a 488/543 main dichroic. YFP was excited with an argon laser at 25% to 50% of its output that was attenuated to 7% to 9% at 514 nm. The emission was sent through a 530-to 600-nm band-pass filter for detection of YFP. CFP was stimulated by a 458-nm laser line that was attenuated to 7% to 9%. The emission was passed through a 475-to 525-nm band-pass filter for detection of CFP.

Supplemental Data
The following materials are available in the online version of this article.