The transcription factor CLOCK–BMAL1 is a core component of the molecular clock machinery that drives circadian gene expression and physiology in mammals. Recently, we reported that this heterodimeric transcription factor functions as a signaling molecule in response to the resetting stimuli via the Ca2+-dependent protein kinase C pathway. Here, we demonstrate that the CREB-binding protein (CBP) plays a key role in rapid activation of the CLOCK–BMAL1 heterodimer that leads to phase resetting of the circadian clock. Under physiological conditions, a bimolecular fluorescence complementation (BiFC) assay revealed that CLOCK and BMAL1 dimerize in the cytoplasm and subsequently translocate into the nucleus in response to serum stimuli (mean time duration was 29.2 minutes and mean velocity 0.7 μm/minute). Concomitantly, BMAL1 rapidly recruited CBP on Per1 promoter E-box, but not p300 (a functional analog of CBP), in the discrete nuclear foci. However, recruitment of CBP by cAMP/Ca2+ response element-binding (CREB) protein on CRE was not markedly increased upon delivery of the resetting stimuli. Furthermore, overexpression of CBP greatly potentiated the CLOCK–BMAL1-mediated Per1 transcription, and this effect was completely abolished by site-directed mutation of E-box elements, but not by the mutation of CRE in the Per1 promoter. Furthermore, molecular knockdown of CBP severely dampened circadian oscillation of clock gene expression triggered by the resetting stimuli. These findings suggest that CBP recruitment by BMAL1 mediates acute transactivation of CLOCK–BMAL1, thereby inducing immediate-early Per1 transcription and phase resetting of the circadian clock.
Within a variety of living species from bacteria to mammals, the circadian clock is an intrinsic timing system that allows organisms to anticipate and prepare for periodic environmental changes represented by day and night. To coordinate their daily physiology and behavior with environmental changes, the endogenous clock is entrained to the 24 hour cycle of the earth's rotation through daily phase resetting. In mammals, the circadian rhythm is maintained by the master clock (located in the suprachiasmatic nucleus, SCN, of the hypothalamus) that can be entrained by environmental cues called zeitgebers, the primary one of which is the day–night cycle (Reppert and Weaver, 2002). Recently, peripheral tissues and cells, even immortalized cell lines, have been found to possess their own self-sustaining oscillators that can be entrained by neural and humoral signals from the SCN (Schibler and Sassone-Corsi, 2002). Although the master and peripheral clocks are reset by distinct external cues, both use common molecular components and signaling pathways for time adjustment. One of the prominent molecular events that cause clock resetting is the immediate-early induction of the Per1 gene not only in SCN but also in peripheral tissues (Akashi and Nishida, 2000; Balsalobre et al., 2000a; Shigeyoshi et al., 1997; Yagita et al., 2001). Similar molecular events also occur in various cultured cells when they are reset by external stimuli such as serum shock and glucocorticoid (Balsalobre et al., 1998; Balsalobre et al., 2000a).
Thus far, intensive efforts have been made to elucidate the signal transduction pathways that lead to phase resetting of the circadian clock. These pathways include several kinase pathways such as mitogen-activated protein kinases (MAPKs), cAMP-dependent kinase (PKA), and Ca2+/calmodulin-dependent kinase II (CaMKII) (Akashi and Nishida, 2000; Balsalobre et al., 2000b; Ginty et al., 1993; Obrietan et al., 1998). These kinase-mediated signaling cascades are thought to ultimately converge onto activation of cAMP/Ca2+ response element-binding (CREB) protein (Obrietan et al., 1998; Obrietan et al., 1999; Yokota et al., 2001). However, the rapid activation of CREB alone cannot account for the specific ability of CREB to induce the immediate-early response of the Per genes because CREB activation by resetting signals can induce rapid activation of only a few genes, including fos, Per1 and Per2, despite a wide variety of genes containing functional cAMP-response elements (CREs) in their upstream regulatory regions. Even among these immediate-early genes, there are some time delays in reaching peak transcriptional levels in the SCN after the light pulse; induction of fos mRNA peaks within 30 minutes, whereas Per1 and Per2 respond more slowly and reach peak values ~1 and 2 hours after the stimulus, respectively (Albrecht et al., 1997; Crosio et al., 2000; Sherman et al., 1997; Wilsbacher et al., 2002). In addition, accumulating evidence emphasizes a pivotal role of clock-specific transcription factors in circadian clock resetting and Per1 transcription in response to external stimuli. In Drosophila, activation of MAPK and CaMKII increases CLOCK/CYCLE-dependent transcription via direct phosphorylation of CLOCK (Weber et al., 2006). In mammals, our previous studies demonstrated that acute Per1 induction by serum stimuli is mainly attributed to rapid activation of the CLOCK/BMAL1 complex via the Ca2+-dependent protein kinase C (PKC) pathway (Jung et al., 2003; Shim et al., 2007).
CLOCK and BMAL1 are members of the basic helix–loop–helix (bHLH)/PAS family of transcription factors that form a heterodimeric complex and then activate clock gene expression by binding to E-box elements located in the upstream regulatory region of target genes such as Per1 and Per2. Most transcription factors that belong to the bHLH/PAS family are implicated in the signal transduction pathways activated by diverse environmental changes (Kewley et al., 2004). This family of proteins can be categorized into two subgroups on the basis of their role in the signaling process. Class I proteins typically function as sensors of environmental signals, whereas Class II proteins dimerize with Class I members and translocate into the nucleus. Heterodimerization between these two groups is an essential step, not only for functional activation but also for specifying the transcriptional output (Gu et al., 2000). Phylogenetic analysis implies that CLOCK is an environmental sensor and its partner BMAL1 is a Class II member (Gu et al., 2000; Hogenesch et al., 1998). This phylogenetic interpretation is consistent with previous findings suggesting that CLOCK acts as a bona fide signaling molecule responding to the resetting stimuli via the Ca2+-dependent PKC pathway, while BMAL1 shuttles between the cytoplasm and nucleus, thereby facilitating nuclear accumulation of its partner CLOCK (Kwon et al., 2006; Shim et al., 2007).
Nevertheless, nuclear abundance of CLOCK and BMAL1 does not merely indicate transcriptional activation of target genes. A growing body of evidence suggests that transactivation of the CLOCK–BMAL1 complex requires recruitment of transcriptional coactivators such as CREB binding protein (CBP) and its close relative, p300, which mediate histone acetylation and interaction with the basal transcription factor TFIIB (Kalkhoven, 2004; Yujnovsky et al., 2006). Primarily, CBP and p300 are characterized as coactivators of CREB and can bind specifically to the phosphorylated CREB at serine 133 (Chriva et al., 1993; Parker et al., 1996); the binding of CBP to CREB has been suggested to evoke immediate-early response of the Per1 gene during photic resetting of the circadian clock (Obrietan et al., 1998; Travinckova-Bendova et al., 2002). Recent studies, however, have demonstrated that rhythmic histone acetylation is an important regulatory mechanism underlying the clock gene activation mediated by the CLOCK–BMAL1 complex, although the coactivator responsible for the clock gene activation remains controversial (Curtis et al., 2004; Etchegaray et al., 2003; Takahata et al., 2000; Yujnovsky et al., 2006).
In the present study, we reveal the stimuli-induced protein–protein interactions that are responsible for CLOCK–BMAL1-mediated phase resetting of the circadian clock in vivo. By using a newly established bimolecular fluorescence complementation (BiFC) technique (Shyu et al., 2006), we first visualized dynamic changes in heterodimerization and nuclear translocation of CLOCK and BMAL1 in living cells, and demonstrated the kinetics of these molecular events. Then, combination of real-time measurements of complemented fluorescence and bioluminescence further showed that CBP is a key player in the rapid activation of the CLOCK–BMAL1 complex rather than of CREB in response to the resetting stimuli, and that direct interaction between CBP and BMAL1 is a crucial signaling event that leads to circadian clock resetting.
BiFC using Venus is a feasible method for assessing the CLOCK–BMAL1 interactions in living cells
To investigate dynamic changes in the molecular interactions between CLOCK and BMAL1 in living cells under physiological conditions, we employed a bimolecular fluorescence complementation (BiFC) assay using an improved yellow fluorescence protein, Venus, which enables visualization of direct protein interactions in living cells under physiologically relevant temperatures (Shyu et al., 2006). To optimize BiFC conditions between CLOCK and BMAL1, we constructed various plasmids encoding complementary fragments of Venus [Venus N-terminal (VN) and Venus C-terminal (VC)] fused with the N-terminal end of full-length BMAL1 and CLOCK (VC–BMAL1 and VN–CLOCK), respectively, and then coexpressed them in COS-7 cells (Fig. 1A). Microscopic observation revealed that coexpression of VC–BMAL1 and VN–CLOCK produced a strong BiFC signal and its predominant nuclear localization (Fig. 1C, upper) was consistent with our previous findings (Kwon et al., 2006). To test whether the binding activity of CLOCK and BMAL1 is crucial for the molecular complementation between VC–BMAL1 and VN–CLOCK, we generated several BMAL1 mutants (ΔbHLH, ΔPAS1, and ΔPAS2) lacking one of the domains responsible for dimerization with CLOCK and coexpressed them with wild-type CLOCK in COS-7 cells. As shown in Fig. 1B, immunoprecipitation assay demonstrated that the ΔPAS2 mutant had the most profound defects in the interaction with wild-type CLOCK (Fig. 1B). Consistent with this finding, VC–ΔPAS2 failed to generate BiFC signal with VN–CLOCK despite the fact that its expression level and subcellular localization were comparable to those of VC–BMAL1 (Fig. 1C, lower). Furthermore, the molecular complementation between the two fragmented fluorescent probes did not affect the transcription factor activity of the BMAL1/CLOCK complex or the inhibitory effects of Cryptochrome 1 (CRY1) on its transactivation (Fig. 1D). Thus, we concluded that the BiFC assay using Venus would be a feasible approach for investigating the dynamic changes of intracellular interaction between CLOCK and BMAL1 in vivo without any significant artifacts.
The PKC pathway mediates the serum-evoked heterodimerization of CLOCK and BMAL1
Previously, we demonstrated that CLOCK phosphorylation by Ca2+-dependent PKC is essential for transactivation of the CLOCK–BMAL1 complex during the serum-evoked phase resetting of the clock (Shim et al., 2007). To examine the effects of the resetting stimuli on the dimerization of CLOCK and BMAL1 in vivo, we performed the BiFC assay between VN–CLOCK and VC–BMAL1 in living cells after serum shock. Within 1 hour of serum shock, fluorescence complementation between VN–CLOCK and VC–BMAL1 was drastically increased, predominantly in the nucleus, although tiny speckles were distributed in the cytoplasm in COS7 and NIH3T3 cells (Fig. 2A; supplementary material Fig. S1A). In addition, quantitative analysis demonstrated that the number of cells with nuclear fluorescent signals was augmented more than fourfold after the brief serum shock (Fig. 2B). These effects returned to control level by pretreatment with the PKC inhibitor GF109203X (GF), but not by pretreatment with cyclohexamide and other kinase inhibitors, including the PKA inhibitor KT5720 and the MAPK inhibitor PD98509 (Fig. 2A,B; supplementary material Fig. S1A and Fig. S2). Consistently, treatment with phorbol 12-myristate 13-acetate (PMA), a PKC kinase activator, markedly induced CLOCK–BMAL1 BiFC complexes in the cells, as did serum shock, which was also inhibited by pretreatment with GF. However, addition of other activators, including forskolin (FSK), PKA activator, epidermal growth factor (EGF) and MAPK activator, did not elicit the stimuli-induced effect. (supplementary material Fig. S2). To further verify the role of PKC in heterodimerization of CLOCK and BMAL1, red fluorescence protein (RFP)-tagged PKCγ (PKCγ–RFP) was coexpressed with VC–BMAL1 and VN–CLOCK in COS-7 cells. As expected, coexpression of PKCγ–RFP markedly increased the BiFC signals of CLOCK and BMAL1 in COS-7 cells compared to those in cells expressing the BiFC-competent proteins alone (Fig. 2C). In addition, more BiFC-positive nuclei were found in PKCγ–RFP-transfected cells than in control RFP-expressing ones (Fig. 2D). These observations indicate that Ca2+-dependent PKC stimulates the protein–protein interaction between CLOCK and BMAL1 and enhances the nuclear localization of these proteins.
Time-lapse BiFC imaging demonstrates rapid nuclear translocation of the CLOCK–BMAL1 complex in living cells
As shown in Fig. 2C, Ca2+-dependent PKC was predominantly localized in the cytoplasm, but the BiFC signal of CLOCK–BMAL1 heterodimer was concentrated in the nucleus. These observations raise the possibility that the two clock proteins dimerize in the cytoplasm and then translocate into the nucleus. To explore the nuclear translocation of the CLOCK–BMAL1 complex in response to the resetting stimuli, we conducted time-lapse imaging experiments in COS-7 and NIH3T3 cells after serum shock. In these experiments, cells were expressed with VC–BMAL1 and VN–CLOCK, and the BiFC signal of the cells was then imaged under normal culture conditions. As shown in the upper panels of Fig. 3A, the complemented fluorescence signal rapidly emerged in both the cytoplasm and nucleus after serum treatment. More specifically, the cytosolic fluorescence was abruptly speckled and appeared to move into the nucleus, whereas the nuclear signal was gradually enhanced throughout the nucleoplasm and was saturated within 1 hour of serum shock (Fig. 3B; supplementary material Fig. S1B and Movie 1). However, when the cells were transfected with the plasmid encoding a BMAL1 mutant that lacked a functional nuclear localization signal (VC–BMAL1 ΔNLS) (Kwon et al., 2006) instead of wild-type BMAL1, the fluorescence signal rapidly accumulated in the cytoplasm but failed to increase in the nucleus in response to the resetting stimuli (Fig. 3A lower panels; Fig. 3B). All these effects were abolished by pretreatment of the cells with a PKC inhibitor (Fig. 3B; supplementary material Fig. S1B,C). These data suggest that the rapid nuclear accumulation of the BiFC signal would be mainly attributed to its cytosolic dimerization in a PKC-dependent manner upon serum stimuli.
To further clarify the nuclear translocation of the cytoplasmic CLOCK–BMAL1 complex, we randomly selected 13 fluorescent speckles in the cytoplasm and characterized their migration by using image analysis software (supplementary material Fig. S3 and Movie 2). All the cytosolic speckles rapidly moved into the nucleus (mean time duration: 29.2 minutes; mean velocity: 0.7 μm/minute) with concomitant increase in nuclear fluorescence, although the particles did not move straight toward the nucleus and the parameters of their migration varied according to each speckle (Table 1; supplementary material Movie 2). Therefore, these findings clearly demonstrate that the resetting stimuli rapidly induce dimerization and nuclear translocation of the cytosolic CLOCK and BMAL1. Moreover, these signaling processes provide a good explanation for the kinetics of immediate-early Per1 expression, which peaked within 90 minutes of delivery of the resetting stimuli (Shigeyoshi et al., 1997).
Rapid nuclear translocation of CLOCK–BMAL1 is essential for resetting of the circadian clock
Although individual cells intrinsically express circadian genes with ~24-hour period, the cells gradually lose coherence at the population level and, hence, the circadian amplitude is damped (Nagoshi et al., 2004; Welsh et al., 2004). Therefore, without the synchronization of individual cellular clocks by resetting stimuli, clock genes in cultured cells exhibit constant expression profiles. To assess the physiological relevance of the rapid nuclear translocation of CLOCK–BMAL1 to circadian clock synchronization, we conducted real-time bioluminescence assays in living cells by using a destabilized luciferase reporter gene (dLuc) controlled by the Per2 promoter (Per2 Pro-dLuc), which drives robust circadian gene expression (Kwon et al., 2006). Cells were co-transfected with vectors carrying Per2 Pro-dLuc and the indicated genes, following which their bioluminescence was measured continuously for 72 hours after administration of the resetting stimuli, such as dexamethasone (Dex), a synthetic glucocorticoid (Fig. 4). Time-course changes in the bioluminescent signal exhibited a circadian rhythm in the cells transfected with Per2 Pro-dLuc alone (Fig. 4A,B, black). Intriguingly, the circadian oscillation of cellular bioluminescence was more robust, displaying significantly higher amplitude when cells were coexpressed with wild-type BMAL1 (Fig. 4, green). The oscillation and amplitude of the reporter activity were severely dampened by pretreatment with a PKC inhibitor (Fig. 4, red). More importantly, the cells coexpressed with mutant BMAL1 lacking NLS (BMAL1 ΔNLS) also markedly decreased the amplitude of the rhythmic expression of the reporter gene (Fig. 4, violet). These results suggest that both the nuclear localizing activity of BMAL1 and its dimerization with CLOCK in a PKC-dependent manner are essential for the rhythmic gene expression in the given cell population.
CBP activates the nuclear CLOCK–BMAL1 complex through direct binding to BMAL1
As described above, nuclear translocation of the CLOCK–BMAL1 complex is essential for cellular clock resetting by the extracellular stimuli. Nonetheless, it is important to note that nuclear abundance of CLOCK and BMAL1 during the circadian cycle does not correlate with their transcriptional activity (Kondratov et al., 2003; Kwon et al., 2006; Lee et al., 2001). This paradoxical feature is mediated by potent negative regulators of the CLOCK–BMAL1 heterodimer, including PERs and CRYs. In particular, the binding of CRY to the heterodimeric transcription factor facilitates the nuclear accumulation of this complex by suppressing its degradation and transactivation (Kwon et al., 2006; Lee et al., 2001). Indeed, coexpression of CRY1 with the fusion proteins VC–BMAL1 and VN–CLOCK noticeably enhanced the BiFC signal in the nucleus without the resetting stimuli (supplementary material Fig. S4). These observations raise the possibility that there is another regulatory mechanism underlying the transcriptional activation of the CLOCK–BMAL1 heterodimer. In addition, growing evidence emphasizes a pivotal role of histone acetylation in circadian gene activation, although the key player responsible for this enzyme modification is still elusive (Doi et al., 2006; Etchegaray et al., 2003). This prompted us to investigate the potential role of the transcriptional coactivators CBP and p300, which are known histone acetyltransferases (HATs) that bind to the CLOCK–BMAL1 complex, in circadian clock resetting. As shown in Fig. 5A, coexpression of either CBP or p300 with CLOCK–BMAL1 greatly increased CLOCK–BMAL1-dependent Per1 transcription. Subsequent immunoprecipitation assay demonstrated that both CBP and p300 can interact with the CLOCK–BMAL1 complex (Fig. 5B). However, the results from the immunoprecipitation experiment do not provide details about how one interacts specifically with the other within the protein complex. Hence, to further dissect the protein–protein interactions between the heterodimeric complex and the coactivators, we developed BiFC constructs for CBP and p300 and analyzed their interaction with either CLOCK or BMAL1 in living cells by using all possible combinations with the BiFC constructs of CLOCK and BMAL1 (Fig. 5C; supplementary material Fig. S5). Interestingly, BMAL1 was capable of interacting with both CBP and p300, but CLOCK failed to interact with CBP. All these fluorescence signals were strongly detected as discrete foci distributed throughout the nucleus.
Next, to examine the effects of the resetting stimuli on the interaction between the coactivators and the CLOCK–BMAL1 complex, we monitored consecutive changes in Venus fluorescence in living cells expressing possible combinations of the BiFC fusion proteins after serum stimulation. The time-lapse imaging data revealed that the BiFC signal between CBP and BMAL1 was dramatically enhanced in the nucleus shortly after delivery of the resetting stimuli and was saturated within 2 hours (Fig. 5D upper panels; see supplementary material Movie 3). The mean value of the fluorescence intensity of 50 nuclei further confirmed the kinetics of the interaction between CBP and BMAL1 after treatment with the extracellular stimuli (Fig. 5E). In sharp contrast, there were no significant changes in the BiFC signal produced by either VC–p300/VN–BMAL1 or VC–p300/VN–CLOCK (Fig. 5D middle and lower panels, Fig. 5E; see supplementary material Movies 4 and 5). It is therefore likely that immediate recruitment of CBP to BMAL1 after nuclear translocation of the CLOCK–BMAL1 complex is a key molecular event in the activation of transcription of target genes.
Recruitment of CBP, not by CREB but by BMAL1, is crucial for clock gene induction
CBP has been identified as a coactivator for the transcription factor CREB and binds when CREB is phosphorylated (Kwok et al., 1994; Mayr and Montminy, 2001). In addition, several previous reports have suggested that CREB acts as a major effector molecule of multiple signaling pathways that lead to immediate-early response of Per1 gene transcription during resetting of the circadian clock (Gau et al., 2002; Ginty et al., 1993; Travinckova-Bendova et al., 2002). Therefore, we assessed the possibility that CBP plays a substantial role in CREB activation, thereby inducing the immediate-early Per1 gene expression. First, we analyzed whether the resetting stimuli promote CBP binding to CREB in COS-7 cells by using a BiFC assay. The CREB protein fused with the C-terminal fragment of Venus (CREB-VC) successfully produced moderate fluorescence exclusively in the nucleus when coexpressed with CBP–VN (data not shown). However, the BiFC signal of CREB and CBP did not increase to a great extent after serum stimuli when compared to that of BMAL1 and CBP (Fig. 6A,B; see supplementary material Movie 6). Next, to investigate the functional relevance of CBP recruitment by BMAL1 and CREB to Per1 gene induction, cells were co-transfected with a Per1 promoter reporter and various combinations of plasmids expressing CBP, CLOCK, BMAL1 and CREB. As shown in Fig. 6C, both CLOCK–BMAL1 and CREB induced Per1 promoter activation at comparable levels, but the synergistic effect of CBP with CLOCK–BMAL1 was much greater than that combined with CREB on Per1 promoter activation. These results are consistent with the BiFC data demonstrating that CBP binding to BMAL1 is more efficient than binding to CREB (Fig. 6A,B).
To further assess the role of CBP in the Per1 gene activation induced by CLOCK–BMAL1 and/or CREB, we generated site-directed mutants of the Per1 promoter lacking functional CRE or E-box elements (the binding site of CLOCK–BMAL1) and conducted promoter reporter assays using these mutant promoters under the same experimental conditions as given in Fig. 6C (Fig. 6D,E). Unexpectedly, the mutant promoter lacking CRE was activated with a pattern similar to that of the intact Per1 promoter, although the overall promoter activity was slightly decreased (Fig. 6D). However, the transcriptional enhancing effect of CBP on CLOCK–BMAL1 was completely abolished when the consensus E-boxes were destroyed (Fig. 6E). More importantly, chromatin immunoprecipitation (ChIP) assay demonstrated that CBP recruitment to the first proximal E-box of Per1 promoter was more significantly increased than its recruitment to CRE shortly after serum shock; however, p300 binding to both the E-box element and the CRE were not significantly altered after serum shock, although its binding activity was comparable to that of CBP (Fig. 7). Thus, CBP appears to play a crucial role in the rapid induction of Per1 gene transcription, not by binding to CREB but by binding to BMAL1.
CBP is a key player in the CLOCK–BMAL1-mediated circadian clock resetting
Despite the essential role of CBP in rapid Per1 induction, its relevance to circadian physiology remains obscure. To evaluate the physiological relevance of CBP to the CLOCK–BMAL1-mediated clock resetting, we employed an siRNA technique to knockdown CBP and other related molecules in living cells. The synthesized siRNAs specific to CBP, p300 and CREB effectively blocked the target gene expression without any significant effect on cell morphology and expressions of other genes (Fig. 8A and data not shown). We then analyzed the responsiveness of these knockdown cells to the resetting stimuli by a real-time bioluminescence assay. To this end, cells were co-transfected with Per2 promoter-driven luciferase reporter plasmid and specific siRNA against the indicated genes (Fig. 8B), and were then exposed to the resetting stimuli. A long-time measurement of bioluminescence in the living cells demonstrated a pivotal role of CBP in producing circadian oscillation in gene expression. Depletion of CBP severely dampened circadian oscillation of reporter gene expression (Fig. 8B). By contrast, siRNA against p300, CREB or both did not significantly affect the periodic Per2 gene expression in response to the resetting stimuli. These findings strongly suggest that CBP is essential not only for the CLOCK–BMAL1-mediated clock gene transcription but also for circadian clock resetting.
Rapid signaling events via protein–protein interactions in response to extracellular stimuli are widely required for triggering timely physiological responses to environmental changes. By performing real-time analysis of molecular interactions in living cells with BiFC assay, we demonstrated that rapid recruitment of CBP to BMAL1 concomitant with nuclear translocation of the CLOCK–BMAL1 heterodimer is a key molecular event that leads to the synchronization of individual cellular clocks in the presence of entrainment stimuli (Fig. 9).
Accumulating evidence suggests that a number of transcription factors can interact with their binding partners to functionally mediate various signaling events in a phosphorylation-dependent manner (Lassot et al., 2001; Nateri et al., 2005). Considering our previous report demonstrating the rapid activation of CLOCK by PKC-mediated phosphorylation (Shim et al., 2007), the present data using BiFC analysis suggests that phosphorylated CLOCK acquires the ability to interact with BMAL1 to form a transactivation complex. Our BiFC data account for previous results that PKC activation by treatment with PMA promoted the binding of both CLOCK and BMAL1 to E-box elements of the Per1 promoter, Per1 mRNA expression, and eventually the phase shifts of the circadian rhythm in cultured fibroblasts (Shim et al., 2007). Moreover, the real-time bioluminescence data revealed that inhibition of PKC virtually abolished immediate-early induction and rhythmic oscillation of the Per2 gene upon treatment with Dex (a potent clock resetting agent), further emphasizing the physiological significance of the role of PKC in molecular clock resetting (Fig. 4). Consistent with these results, PKC activation in the SCN caused advanced circadian rhythm in neurons, whereas inhibition of the PKC pathway eliminated the phase shift of SCN activity induced by melatonin and neuropeptide Y (Biello et al., 1997; McArther et al., 1997; Schak and Harrington, 1999). Taken together, our BiFC data unveiled the functional relevance of PKC activation for CLOCK–BMAL1-mediated phase resetting of central and peripheral clocks.
Given that CLOCK acts as a signaling molecule via PKC-mediated phosphorylation, initially it was predicted that CLOCK might be a transcription factor responsible for signal-induced CBP recruitment, as in the case of phosphorylation-dependent coactivator recruitment of several transcription factors in response to various stimuli (Chriva et al., 1993; Foulds et al., 2004). However, BiFC analyses after treatment with the resetting stimuli revealed that BMAL1 rapidly interacted with CBP after nuclear entry, but CLOCK did not (Fig. 5C,D). In line with these observations, a previous study suggested that CBP can interact with a transcriptional activation domain (TAD) at the C-terminal region of BMAL1 when it is heterodimerized with CLOCK (Takahata et al., 2000). Indeed, deletion of the PAS2 domain of BMAL1, which eliminated the interaction with CLOCK (Fig. 1D), substantially weakened the BiFC between BMAL1 and CBP, whereas overexpression of CLOCK strengthened the signal (Y.L., unpublished data). Intriguingly, the BMAL1 C-terminal domain was found to also associate with CRY1, a potent repressor of the CLOCK–BMAL1 heterodimer, thereby suppressing its target gene transcription (Kiyohara et al., 2006; Sato et al., 2006). These findings present the possibility that the BMAL1 C-terminal has antagonistic dual functions and serves as a molecular switch for the gene transcription controlled by the CLOCK–BMAL1 heterodimer.
More importantly, we found that CBP was rapidly recruited to BMAL1 on Per1 promoter E-box shortly after stimulation, whereas p300, a closely related protein, failed to do so. Despite their functional overlap, accumulating evidence emphasizes that CBP and p300 have distinct roles in various physiological settings such as cell proliferation, differentiation and apoptosis (Vo and Goodman, 2001). In particular, it has been reported that the transcriptional activity of CBP and p300 can be differentially regulated by post-translational modifications such as phosphorylation. For instance, in vivo phosphorylation at specific sites of CBP (sites that p300 lacks) by PKC and CaMKIV controls CBP recruitment to the related transcription complexes and subsequent transcriptional activation of these elements (Impey et al., 2002; Zanger et al., 2001). Furthermore, the present real-time bioluminescence data demonstrate that specific knockdown of CBP, but not p300, disrupted circadian gene transcription (Fig. 8). This notion is further supported by a recent study suggesting that CBP is phosphorylated and recruited to the CLOCK–BMAL1 complex via D2R receptor signaling, thereby potentiating Per1 gene expression in cultured retinal neurons (Yujnovsky et al., 2006). By contrast, co-immunoprecipitation experiments using mouse liver and HeLa cells showed that p300, but not CBP, interacts with CLOCK in a time-dependent manner to activate CLOCK–BMAL1-mediated circadian transcription through rhythmic histone acetylation (Curtis et al., 2004; Etchegaray et al., 2003). Accordingly, unlike p300, CBP seems to have unique regulatory properties for signal-regulated transcriptional activation mostly via phosphorylation-dependent interaction with related transcription factor complexes.
Recently, we demonstrated that BMAL1 can be modified by SUMO2/3 and that this modification localizes BMAL1 exclusively to the promyelocytic leukemia (PML) nuclear body and potentiates its transcription factor activity (Lee et al., 2008). In fact, the PML nuclear body has been reported to serve as a discrete nuclear compartment in which a number of transcription factors and their co-regulators such as p53, CBP and RNA polymerase II (Pol II) reside (Zhong et al., 2000). Moreover, CBP is associated with RNA Pol II and PML in a subset of nuclear bodies (von Mikecz et al., 2000). It was reported that the prompt formation of nuclear bodies upon serum stimulation coincides with the immediate-early induction of the fos gene in cultured fibroblasts (Matsuzaki et al., 2003). Consistent with previous findings, our results from BiFC analysis revealed that binding of CBP to BMAL1 predominantly occurs at discrete nuclear foci (Fig. 5C,D), and that the BiFC signal colocalizes with immunostaining of the PML protein (Y.L., K.H.L. and K.K., unpublished data). These observations raise the possibility that formation of nuclear bodies containing the BMAL1/CBP complex after stimulation provides transcriptionally active sites of downstream target genes, including Per1 and Per2.
We have shown that CREB exhibits much less efficiency than BMAL1 for CBP recruitment in response to serum stimuli (Fig. 6A,B). CREB is generally known as a central mediator in light-induced clock resetting, on the basis of the observation that CREB was phosphorylated at serine 133 in vivo in response to photic stimulation (Gau et al., 2002; Ginty et al., 1993). However, several lines of evidence have shown that formation of the CREB–CBP complex was readily induced by a cAMP agonist, but not well promoted or even inhibited in cells exposed to a variety of mitogenic or stress signals, including epidermal growth factor, PMA and other growth factors, despite the comparable levels of CREB phosphorylation (Mayr et al., 2001; Nakajima et al., 1996; Sun et al., 1994; Parker et al., 1998; Wagner et al., 2000; Xing et al., 1996). This could be a possible explanation for the minor increase in the CREB/CBP complex-induced BiFC signal after serum shock, which includes diverse stimuli (Fig. 6A,B). Consistent with the BiFC data, our transcriptional assay showed that CBP-potentiated upregulation of Per1 promoter activity by CREB was relatively far less than by CLOCK–BMAL1 (Fig. 6C). Even in ChIP analysis, CBP was recruited on the proximal E-box of Per1 promoter far more effectively than on the CRE region in response to serum shock stimuli. (Fig. 7). Moreover, specific knockdown of CREB by siRNA did not efficiently abrogate rhythmic Per2 expression (Fig. 8). In fact, diverse extracellular stimuli, including fibroblast growth factor, platelet-derived growth factor, serum and PMA, have been shown to induce the immediate-early expression of Per1 or Per2, resulting in a robust circadian oscillation of gene expression (Akashi et al., 2000). Taken together, it appears that without significant interference by the other signaling cascades, CBP recruitment to BMAL1, rather than to CREB, can serve as a major signaling route to elicit rapid transcriptional response upon entrainment stimuli. This notion is further supported by the data showing that the upregulation of Per1 expression by CBP coactivation of CLOCK–BMAL1 or CREB was substantially inhibited by the disrupted E-boxes, but was conserved even in the absence of the CREB responsive element (Fig. 6C,E).
Materials and Methods
Cell culture, transfection, and luciferase assay
COS-7 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin (Invitrogen) at 37°C under 9% CO2. For luciferase reporter assays, COS-7 cells were transfected using Lipofectamine PLUS reagents (Invitrogen). At 24 hours after transfection, the cells were lysed, and luciferase activities in the cell extracts were analyzed by a Dual-Luciferase reporter assay system (Promega).
PKCγ–RFP was kindly provided by Dineke S. Verbeek (University of Amsterdam Academic Medical Center, Amsterdam, The Netherlands). pcDNA 3.1-mouse CBP and pcDNA 3.1-mouse p300 were gifted by Ronald Cohen (University of Chicago Medical School, Chicago, IL). RSV–rat CREB was kindly provided by Jea Bum Kim (Seoul National University, School of Biological Sciences, Seoul, Korea). CFP–CRY1 was constructed by subcloning the cDNA of cyan fluorescence protein (CFP) into the upstream region of mouse CRY1 cDNA, which was cloned in pFLAG–CMV2 (Sigma). The pGL3–mPer1-Luc promoter was previously described (Travinckova-Bendova et al., 2002). The CRE-defective mouse Per1 (mPer1) promoter was generated by mutation of 4 nt (5′-TGACGTCA→AAAAAAAA-3′). E-boxes within the mPer1 promoter were abrogated by mutation of 2 nt (5′-CACGTG→AGCGTG-3′). A subset of Myc–BMAL1 expression clones was previously described (Kwon et al., 2006). Internal deletions of BMAL1 produced Myc–BMAL1 Δ71–140 (ΔbHLH), Myc–BMAL1 Δ210–320 (ΔPAS1) and Myc–BMAL1 Δ350–480 (ΔPAS2), respectively.
Proteins were resolved on 6 or 8% sodium dodecyl sulfate–polyacrylamide gels and transferred to polyvinylidene difluoride membranes (Immobilon P; Millipore). Target proteins were detected with anti-BMAL1 (Kwon et al., 2006), anti-CLOCK (S-19; Santa Cruz Biotechnology), anti-CBP (A22; Santa Cruz Biotechnology), anti-p300 (N-15; Santa Cruz Biotechnology) and anti-CREB (48HE; Cell Signaling Technology). The immune complexes were visualized with an ECL detection kit (Pierce).
Preparation of nuclear extracts and immunoprecipitation
NIH3T3 cells were transfected with the indicated plasmids. At 36 hours post-transfection, the cells were washed twice in phosphate-buffered saline (PBS) and scraped in hypotonic lysis buffer A (50 mM HEPES, 10 mM KCl, 1.5 mM MgCl2, 1% NP-40). After centrifugation for 10 minutes at 4000 g (4°C), the supernatant was discarded and the crude nuclear pellet was resuspended in hypotonic buffer C [50 mM HEPES pH 7.6, 1.5 mM MgCl2, 450 mM NaCl, 10 mM KCl, 1 mM EDTA, 1 mM EGTA, 1 mM Na3VO4, 1 mM NaF, 1 mM PMSF, protease inhibitor cocktail (Sigma)], vortex-mixed for 5 seconds, incubated on ice for 30 minutes, and centrifuged for 30 minutes at 12,000 g (4°C). The supernatant was transferred to a fresh tube and diluted in two volumes of hypotonic buffer C without NaCl before immunoprecipitation. For immunoprecipitation, whole cell lysates or nuclear extracts were incubated with 2 μg of precipitating antibodies, as indicated, for 1.5 hours at 4°C and then added to protein A/G-Sepharose bead slurry (Amersham). The final immune complexes were analyzed by immunoblotting as indicated above.
Chromatin immunoprecipitation assay
ChIP assays were performed using a commercial ChIP assay kit (Upstate Biotech) according to the manufacturer's instructions. The pre-cleared chromatin was immunoprecipitated for 2 hours at 4°C by agitating with 2 μg of anti-CBP antibody and anti-p300 antibody. The cell lysates without incubation of antibody were used for input control. Immune complexes were collected by incubation with protein-A Sepharose beads and sheared salmon sperm DNA. The primer sets used for ChIP assays were as follows: 1st (proximal) mPer1 E-Box (E-box1) forward primer, 5′-CCCTCACTTCCCTTTCATTATTGACG-3′ and reverse E-box1 primer, 5′-TGCATAATGCCAGGCCCTGCCCCTCATTGG-3′ (amplified 250 bp) (Lee et al., 2001); the mPer1 CRE forward primer was 5′-CAGCTGCCTCGCCCCGCCTC-3′, and reverse CRE primer, 5′-CCCAAGCAGCCATTGCTCGC-3′ (amplified 215 bp).
Transfection of siRNA
siRNAs targeting mouse CBP or mouse p300 and a non-targeting control siRNA were commercially available (Santa Cruz Biotechnology). These siRNAs are composed of a pool of three or four individual siRNAs designed to different regions of their respective target mRNA. Double-stranded RNAi oligonucleotides for mouse CREB were designed and synthesized by Shanghai GenePharma (Shanghai, China). The sequence (forward: 5′-CAGCCACAGATTGCCACATTA-3′, reverse: 5′-TAATGTGGCAATCTGTGGCTG-3′) targets mouse CREB (NM_009952, 313 to 333). Transfection of 100 pmole siRNAs per well in NIH3T3 cells was conducted with Lipofectamine 2000 (Invitrogen) in 12-well plates according to the manufacturer's instructions. Immunoblotting was performed 48 hours after transfection.
Real-time measurement of bioluminescence and data analysis
NIH3T3 cells were plated in a 35-mm dish and transfected with the Per2 promoter fused with destabilized firefly luciferase (Per2 Pro-dLuc) alone or co-transfected with pcDNA BMAL1 or pcDNA BMAL1 NLSΔ (Fig. 4) and siRNAs targeting mouse CBP, mouse p300 or mouse CREB as indicated (Fig. 7). The cells were pre-incubated in serum-free DMEM and synchronized with 100 nM Dex (Sigma) for 2 hours, and the medium was replaced with a recording medium [DMEM supplemented with 10% FBS and 0.1 mM D-luciferin (Promega)], as previously described (Kwon et al., 2006). GF109203X (5 μM) was added 1 hour before Dex treatment. Bioluminescence was measured with photomultiplier tube detector assemblies (AB-2550 Kronos-Dio; ATTO, Tokyo, Japan). For the detrended data, rhythms were plotted following a standard detrending procedure whereby the 24-hour running mean was subtracted from the raw data to remove the baseline drift that can mask the circadian rhythm, as previously described (Tsuchiya et al., 2009). Quantitative and statistical analysis of amplitudes from the bioluminescence rhythms were performed using cosinor periodogram provided by Roberto Refinetti (University of South Carolina, Salkehatchie, SC).
BiFC and immunofluorescence analyses
For BiFC assay, cDNAs encoding N-terminal residues 1–172 (VN173) and C-terminal residues 155–238 (VC155) of Venus were fused upstream of sequences encoding BMAL1 using linker sequences for 5′-GGGGSGGGGS-3′, and the chimeric coding regions were cloned into pcDNA 3.1 (Invitrogen) to produce VN–BMAL1 and VC–BMAL1 respectively. For BMAL1–VN and BMAL1–VC, cDNA encoding BMAL1 was subcloned into pFLAG–VN173 and pHA–VC155, respectively, which were kindly provided by Chang-Deng Hu (Purdue University School of Pharmacy, West Lafayette, IN) (Shyu et al., 2006). For VC–ΔPAS 2, cDNA encoding BMAL1 Δ350–480 (ΔPAS2) was subcloned in place of wild-type BMAL1 in VC–BMAL1. VC–BMAL1 ΔNLS was generated by fusing VC fragment upstream of BMAL1 ΔNLS, which lacks a functional nuclear localization signal (Kwon et al., 2006). For CLOCK BiFC plasmids, VN173 and VC155 were fused upstream of sequences encoding CLOCK using a linker for 5′-ANSSIDLISVPVEYPYDVPDYASR-3′, and the chimeric coding regions were cloned into pFLAG–CMV2 (Sigma) to produce VN–CLOCK and VC–CLOCK, respectively. For CLOCK–VN and CLOCK–VC, CLOCK cDNA was subcloned to pFLAG–VN173 and pHA–VC155. For BiFC analysis of CBP recruitment, cDNA encoding full-length mouse CBP was cloned into pFLAG–VN173. For VC–p300, VC155 was fused upstream of mouse p300 cDNA in pcDNA 3.1 expression vector. rCREB–VC was generated by subcloning cDNA encoding full-length rCREB into pHA–VC155.
For BiFC analyses, COS-7 cells were cultivated in 24-well plates overnight and then co-transfected with expression vectors as indicated in each experiment (200 ng each) using Lipofectamine Plus reagents (Invitrogen). At 12 hours post-transfection, the cells were placed in 1% serum media for 12–16 hours of serum starvation followed by 1–2 hours exposure to media containing 50% horse serum or various kinase activators, including 1 μM PMA (Sigma), 1 μM forskolin (Calbiochem) and 10 ng/ml EGF (Calbiochem). Various inhibitors, including 5 μM cycloheximide (Sigma), 0.5 μM KT5720 (Calbiochem), 10 μM PD98509 (Calbiochem) and 5 μM GF109203X (Calbiochem), were added 2 hours before the addition of 50% horse serum or the indicated kinase activators. The samples were examined under a fluorescence microscope, and fluorescent images were captured using a YFP filter set (EX=500 nm, DM=515 nm, EM=535 nm). For quantitative analyses, the cells showing nuclear signals were counted among BiFC signal-positive cells after serum shock in the presence or absence of the indicated inhibitors.
For immunofluorescence analyses, cells were fixed with 3.75% paraformaldehyde in PBS and incubated with anti-BMAL1, anti-CLOCK, anti-c-Fos (PC38; Calbiochem), anti-phospho-RORαSer35 (Lee et al., 2010), anti-phospho-CREB (#9191; Cell Signaling Technology) and anti-phospho-pERK (#9101: Cell Signaling Technology) antibodies. The cells were visualized with secondary antibodies conjugated with TRITC (Jackson ImmunoResearch Laboratories).
Time lapse imaging analyses/deconvolution microscopy
Cells were plated in a live cell chamber (Live Cell Instrument, Seoul, South Korea) in 10% FBS-containing growth medium. For serum shock experiments, the transfected cells, as indicated above, were incubated in a humidified chamber at 37°C and at ambient CO2 levels for 12 hours and starved in medium containing 1% FBS for 12–16 hours before 50% horse serum treatment. Images were collected using a Delta Vision deconvolution microscope (Applied Precision, Isaquah, WA) at 37°C. At every time point, a fluorescence image was taken with specific filter sets for YFP. For quantification of changes in BiFC signals, fluorescence intensity in the nuclei of more than 50 individual fluorescent cells was measured using an automated intensity recognition feature of the Delta Vision imaging analysis system. Four-dimensional particle tracking analyses of BiFC signal in cells were conducted using Imeris imaging analysis software provided by Applied Precision.
We thank Chang-Deng Hu, (Purdue University School of Pharmacy, West Lafayette, IN) for generously providing the BiFC plasmids. Anti-phospho-RORαSer35 antibody was kindly provided by Sung Hee Baek (Department of Biological Sciences, Seoul National University, Seoul, Korea). This work was supported by the Brain Research Center of the 21st Century Frontier Program in Neuroscience from the Korea Ministry of Education, Science and Technology (MEST), National Research Foundation of Korea (2009-0070022), and the BioImaging Research Center at GIST. Y.L., J.L. and I.K. are supported by the Brain Korea 21 Research Fellowships from MEST.