Sequential and Compartment-specific Phosphorylation Controls the Life Cycle of the Circadian CLOCK Protein*

The circadian clock facilitates a temporal coordination of most homeostatic activities and their synchronization with the environmental cycles of day and night. The core oscillating activity of the circadian clock is formed by a heterodimer of the transcription factors CLOCK (CLK) and CYCLE (CYC). Post-translational regulation of CLK/CYC has previously been shown to be crucial for clock function and accurate timing of circadian transcription. Here we report that a sequential and compartment-specific phosphorylation of the Drosophila CLK protein assigns specific localization and activity patterns. Total and nuclear amounts of CLK protein were found to oscillate over the course of a day in circadian neurons. Detailed analysis of the cellular distribution and phosphorylation of CLK revealed that newly synthesized CLK is hypophosphorylated in the cytoplasm prior to nuclear import. In the nucleus, CLK is converted into an intermediate phosphorylation state that correlates with trans-activation of circadian transcription. Hyperphosphorylation and degradation are promoted by nuclear export of the CLK protein. Surprisingly, CLK localized to discrete nuclear foci in cell culture as well as in circadian neurons of the larval brain. These subnuclear sites likely contain a storage form of the transcription factor, while homogeneously distributed nuclear CLK appears to be the transcriptionally active form. These results show that sequential post-translational modifications and subcellular distribution regulate the activity of the CLK protein, indicating a core post-translational timing mechanism of the circadian clock.

The circadian clock provides a molecular mechanism that orchestrates behavior and physiology in a temporal fashion and synchronizes homeostatic functions with the environmental cycles of day and night (1)(2)(3). The central oscillating activity of the Drosophila and mammalian circadian clock is formed by the heterodimeric complex of the transcription factors CLOCK (CLK) 2 and CYCLE (CYC) that ultimately controls genome-wide transcription and activity states of key regulatory components (4 -7). Importantly, the circadian oscillator is synchronized by environmental cycles, primarily light/dark and temperature cycles (8,9), but keeps time on its own in the absence of environmental cues, therefore representing a true molecular clock.
Despite the crucial role of CLK/CYC for the circadian orchestration of physiology in Drosophila and mammals, little is known about their post-translational regulation (2,10,11). Transcript levels of Clk reveal robust circadian oscillations in Drosophila, due to rhythmic binding of the activator PAR-DOMAIN PROTEIN 1 and the repressor VRILLE to V/P-elements in the Clk-promoter (12)(13)(14). Rhythmic Clk transcription appears however not essential for self-sustained molecular oscillations, because expression of CLK from a perpromoter in anti-phase to its endogenous rhythm was found to support normal clock function (15). The post-translational regulation of CLK/CYC is however crucial for constitution of a circadian oscillator. Two CLK/CYC-activated genes period (per) and timeless (tim) feed back onto CLK/CYC activity, by binding and inhibition of CLK/CYC (16). Nucleocytoplasmic shuttling of TIM (17) allows a temporal control of nuclear import and inhibitor activity of PER (18 -21). PER-mediated inhibition of CLK/CYC (22,23) involves the recruitment of the casein kinase I⑀ homolog DOUBLE-TIME, which contributes to phosphorylation-induced inhibition and degradation of the CLK protein (24,25). Activation of CLK/CYC is affected by calcium/calmodulin-dependent kinase II and mitogen-activated protein kinase, which both phosphorylate CLK in vitro (26). Phosphorylation of CLK may also control recruitment of the transcription co-activator CREB-binding protein to the CLK/CYC complex, which is essential for trans-activation by CLK/CYC (27,28). These studies suggest a fundamental role of post-translational modifications for the temporal regulation of CLK/CYC-dependent transcription.
Here we show that sequential and compartment-specific phosphorylation controls the life cycle of the CLK protein.
Newly synthesized CLK accumulates in the cytoplasm in a hypophosphorylated state. Subsequent nuclear translocation converts the CLK protein into an intermediately phosphorylated form. Interestingly, we found that CLK is stored in subnuclear compartments, and homogeneously distributed CLK protein represents a transcriptionally active species. Hyperphosphorylation as well as degradation of CLK are promoted by nuclear export of the transcription factor. The data reveal specific localization and activity patterns for the phosphorylation states of the CLK protein that oscillate over the course of a day (29).

EXPERIMENTAL PROCEDURES
Expression Constructs-CLK proteins were expressed from a pAc5.1/V5-HisA vector (Invitrogen) either without any tag (pAc-Clk) for Western blot analysis and transcription activation assays as described in a previous study (26,27), or as fusion proteins with a N-terminal DsRed1 red fluorescent protein tag (pDsRed-Clk) for fluorescence microscopy (30). Mutations and deletions were inserted into these constructs with the QuikChange II XL system (Stratagene, La Jolla, CA) according to the instructions by the manufacturer and using the following primers: CLK NLS (substitution of Pro-484 to Leu, Lys-485 to Glu, and Lys-487 to Glu; fwd, CCA CAG GAA TAT CGC TCG AGG CCG AAC GAA AGT GC; rev, GCA CTT TCG TTC GGC CTC GAG CGA TAT TCC TGT GG), CLK ⌬NES (deletion of 25 amino acids from Gln-836 to Leu-860; fwd, GGC TCG CAA TCC ACC ATT AAT C; rev, TTG TTG CAG CTG CAA CTG TTG CTG). A 25tag-CLK ⌬NES construct carries an Nterminal 25-amino acid tag (MDYKDDDDKDYKDDDDK-DYPRYFQS) that compensates for the change in protein size caused by the NES deletion.
Fluorescence Microscopy-For cellular localization studies, DsRed-CLK constructs were expressed in Drosophila S2 cells or S2Rϩ cells as specified in the figure legends after transient transfection with Lipofectin (Invitrogen) as instructed by the manufacturer. 5 ϫ 10 5 cells were transfected with 1 or 2 g of plasmid DNA and 10 l of Lipofectin, as described previously (16,22,26). Cells were cultured in poly-L-lysine (Sigma)-coated cover slide chambers (Nalgene, Rochester, NY). Fluorescence microscopy was performed using an Axiovert 200M (Zeiss, Jena, Germany) fluorescence microscope or a Nikon C1Si-CLEM confocal microscope (Fig. 6) after staining of nuclei with Hoechst 33342 (Invitrogen). At least 60 transfected cells from at least two independent experiments were analyzed per construct. Images were collected with a laser wavelength of 563 nm for DsRed-CLK and 352 nm for Hoechst 33342-stained nuclei.
Immunohistochemistry-Third instar larvae were raised and collected at time points as indicated in the figure in cycles of 12-h light and 12-h darkness. Larvae were dissected and brains were fixed in 4% formaldehyde in PEM buffer (100 mM Pipes, pH 6.9, 1 mM EGTA, 2 mM magnesium sulfate) for 2 h, blocked with 10% normal goat serum (NGS) in PT buffer (phosphatebuffered saline plus 0.3% Triton X-100) for 2 h at room temperature and subsequently incubated for 48 h at 4°C in 50 l of primary antiserum solution containing 1:500-diluted rabbit anti-CLK (27) and 1:200-diluted guinea pig anti-PAP (31) in PT buffer with 10% NGS. The rabbit anti-CLK antibody was incubated with Clk Jrk protein extract to remove unspecific interactions prior to the application in immunohistochemistry. Brains were rinsed with PT buffer and PT buffer plus 5% NGS for 20 min each and then incubated at room temperature for 2 h in secondary antiserum solution containing 1:200-diluted TRITC-conjugated donkey anti-guinea pig (Jackson Immuno Research, West Grove, PA) and 1:200-diluted Cy2-conjugated goat anti-rabbit (Jackson ImmunoResearch) in PT buffer with 10% NGS. After incubation, brains were rinsed with PT buffer and PT buffer plus 5% NGS for 20 min each and mounted in the mount-ing medium (50 mM Tris-Cl, pH 8, 90% glycerol, 2.5% DABCO (Sigma)).
Confocal Microscopy and Quantification of Staining Intensity-Optical sections of larval lateral neurons (LNs) were imaged on a Nikon C1Si-CLEM confocal microscope. For each Zeitgeber Time (ZT), LNs from at least 10 brain hemispheres were scanned. For each LN sample, PAP staining was used to identify and select an optical section, which was then scanned for CLK immunoreactivity. A 0.31-m Z-stack of images was taken per LN with single laser at 488 nm, 543 nm, and with both lasers together. Images were imported to NIH ImageJ 1.34S, and the localization of neuronal PAP and CLK staining was determined. Total pixel intensity of CLK staining was measured for the whole cell or nucleus only and the means Ϯ S.E. from all brains at a particular ZT are reported in the graphs.
Transcription Activity Assays-Transcriptional activation assays in Drosophila S2 and S2Rϩ cells were performed as described previously (16,22,26). In brief, cells from 500 l of culture (10 6 cells/ml) were transfected after 12-h incubation at 25°C with 200 l of serum-free Schneider's insect medium (Sigma) containing 1.5% Lipofectin (Invitrogen), 25 ng of pRLcopia, 10 ng of pGL3-(4-per-E-box)hs::lucϩ, 200 ng of pHT control vector, and 0.25 ng of pAc-Clk construct. Identical controls were performed for each condition without pAc-Clk. 200 l of 20% fetal bovine serum (Invitrogen) in Schneider's insect medium was added 4 -6 h after transfection. After 40 -48 h luciferase activities were determined using the dual-luciferase reporter assay system (Promega) according to the instructions by the manufacturer. Firefly luciferase activity was normalized toward Renilla luciferase activity to control for transfection efficiency and lysate concentration (16,26).
Subcellular Fractionation-w 1118 flies were entrained during eclosion in LD cycles of 12-h light and 12-h darkness. Flies, 1-7 days old, were harvested in LD 2 h after lights were on (ZT 2), and 2 h after lights were off (ZT 14). 400 -500 l of fly heads were subjected to subcellular fractionation according to a protocol that was adapted from a previous study (32). Fly heads were gently homogenized in liquid nitrogen and transferred to 800 l of buffer A (1 M sorbitol, 7% Ficoll, 20% glycerol, 50 mM Tris, pH 7.5, 5 mM magnesium acetate, 5 mM EGTA, 3 mM CaCl 2 , 3 mM dithiothreitol, 1ϫ complete protease inhibitor (Roche Applied Science)). Coarse debris was filtered by passing the homogenate through a nylon mesh. Subsequently, Nonidet P-40 was added to a final concentration of 0.1%, and the mixture was passed through a 20-gauge needle. 1200 l of ice-cold buffer B (10% glycerol, 25 mM Tris, pH 7.5, 5 mM magnesium acetate, 5 mM EGTA, 1ϫ complete protease inhibitor (Roche Applied Science)) were added during gentle mixing. The mixture was carefully added on top of 300 l of buffer AB (at a ratio of 1:1.7) and centrifuged at 700 ϫ g for 3 min. 120 l of the supernatant was collected as total protein fraction, and the rest was layered on top of a sucrose step-gradient (1 M sucrose, 10% glycerol, 25 mM Tris, pH 7.5, 5 mM magnesium acetate, 1 mM dithiothreitol, 1ϫ complete protease inhibitor (Roche Applied Science)). After centrifugation at 9000 ϫ g for 8 min, the nuclear pellet was resuspended in 50 l of SDS-loading buffer. The supernatant was collected as cytosolic fraction, and proteins were precipitated with trichloroacetic acid (Sigma) and re-dissolved in HU buffer (8 M urea, 200 mM Tris, pH 6.8, 5% SDS, 1 mM EDTA, 100 mM dithiothreitol).
Western Blot Analysis of Phosphorylation States and Degradation Kinetics-CLK proteins were analyzed from fly head extracts or after expression in S2 or S2Rϩ cells that were transiently transfected as described for fluorescence microscopy with 2 g of pAc-Clk constructs. After 40 -48 h cells were harvested and lysed in 4ϫ SDS loading buffer (240 mM Tris/HCl, pH 6.8, 8% SDS, 20% glycerol) and boiled at 95°C, and lysates were subjected to SDS-PAGE on 6% gels. CLK protein was analyzed by Western blot using an antibody raised in rabbits against the C terminus of CLK (27).
For phosphatase treatment, CLK proteins were similarly expressed in S2Rϩ cells. Cells were lysed in RBS buffer (10 mM HEPES, pH 7.5, 5 mM Tris, pH 7.5, 50 mM KCl, 10% glycerol, 2 mM EDTA, 1 mM dithiothreitol, 1% Triton X-100, 0.4% Nonidet P-40) (24) to which complete protease inhibitor (Roche Applied Science) was added. Proteins were subsequently de-phosphorylated by addition of 25 units of -phosphatase (New England Biolabs, Ipswich, MA) to a total of a 50-l reaction, which was incubated for 1 h at 37°C. After denaturation by addition of SDS loading buffer and boiling at 95°C, proteins were analyzed by 6% SDS-PAGE and Western blot.
To determine degradation kinetics, S2 cells (2 ml of 10 6 cells/ ml) were transfected as described for fluorescence microscopy with 2 g of pAc-Clk expression constructs. One day after transfection, cells were split in 24-well plates and incubated for a further 24 h. Subsequently, new protein synthesis was inhibited by the addition of cycloheximide at a final concentration of 260 M. Cells were harvested over a time course after the addition of cycloheximide, and CLK protein levels were determined by Western blot analysis on 6% SDS-PAGE. CLK protein levels were quantified for different time points by densitometry and normalized toward an unspecific control band, which was detected by the anti-CLK antibody. Amounts of CLK proteins are shown relative to the amount of CLK prior to the addition of cycloheximide (time point 0, set to 100).

RESULTS
Circadian Oscillations on the CLK Protein Level-We first investigated the cellular distribution of the CLK protein in circadian neurons of the larval brain (Fig. 1A). We found that the majority of the CLK protein localizes inside the nucleus (33). Interestingly, nuclear and total amounts of CLK revealed robust oscillations over the course of a day, with a maximum in the morning and trough levels in the evening (Fig. 1, A and B). These oscillations correlated with the circadian profile of Clk transcription that reaches a maximum toward the end of the night (16). In addition, phosphorylation states of the CLK protein oscillate in abundance over the course of a day (Fig. 1C) in accord with previous reports (25,29,33). Hypo-and hyperphosphorylated forms of CLK were mainly observed during late night and in the morning, whereas intermediate phosphorylation states were the predominant species in the evening, when CLK/CYC is transcriptionally active.
Compartment-specific Phosphorylation of CLK-To investigate the cellular localization of CLK in more detail, we expressed CLK with an N-terminal red fluorescent protein (DsRed) fusion tag in Drosophila Schneider 2 (S2) cells and analyzed the localization by fluorescence microscopy. The Drosophila CLK sequence reveals two consensus nuclear localization signals (NLSs) (Fig. 2A). The first consensus NLS is located in the N terminus at the beginning of the basic helix-loop-helix (bHLH) DNA-binding domain. Although this region is conserved in mammalian CLK orthologs, mutagenesis of this site did not affect the subcellular localization of the CLK protein, indicating that the consensus NLS in the bHLH domain is not involved in nuclear translocation (data not shown). A second consensus NLS located C-terminal of the PAS domains is unique to the Drosophila CLK protein. Rendering this site nonfunctional by mutagenesis ( Fig. 2A) caused a cytoplasmic localization of the CLK NLS mutant protein in S2 cells (Fig. 2B). These results show that the consensus NLS located C-terminal of the PAS domains is a functional nuclear translocation site and essential for nuclear import of the CLK protein. Interestingly, nuclear import-deficient CLK NLS localized largely in cytoplasmic speckles (Fig. 2B). These speckles were also observed for non-tagged CLK NLS mutants, indicating that the localization pattern is not due to the fluorescent protein fusion tag (data not shown). Similarly, PER/TIM complexes were found to localize in cytoplasmic foci shortly before they translocate into the nucleus (21). The nuclear import incompetent mutant of CLK may therefore be stalled in pre-import cytoplasmic foci.
Cellular fractionation experiments of mammalian CLK from mouse liver (34) revealed a hypophosphorylation of cytoplasmic mCLK, whereas nuclear mCLK was hyperphosphorylated. We therefore analyzed the phosphorylation state of nuclear and cytoplasmic CLK proteins (Fig. 2, B and C) in Drosophila S2 cells. Different hypo-, intermediate, and hyperphosphorylated forms of CLK could be resolved by one-dimensional gel electrophoresis. Phosphatase treatment converted these phosphorylated forms into two bands with increased electrophoretic mobility that likely represent a non-and a hypophosphorylated species (supplemental Fig. S1). Wild-type CLK showed pre-dominantly hyperphosphorylated, low electrophoretic mobility forms (Fig. 2C). In contrast, cytoplasmic CLK NLS accumulated mainly in a hypophosphorylated state, indicating that newly synthesized CLK protein is converted to a low phosphorylated species in the cytoplasm, whereas hyperphosphorylation requires nuclear import (Fig. 2C).
Sequential and Compartment-specific Phosphorylation of the CLK Protein-The nuclear localization of wild-type CLK in S2 cells (Fig. 2) mimics the predominant nuclear localization of CLK in circadian neurons of the larval brain (Fig. 1). When we investigated the cellular localization of CLK in S2Rϩ cells, we made the surprising observation that CLK showed a strong cytoplasmic localization in most cells, in addition to nuclear staining (Fig. 3A). Co-expression of CYC facilitated, however, an efficient nuclear localization of CLK in S2Rϩ cells, suggesting that CYC promotes the nuclear localization of CLK. Such a role of CYC is consistent with the finding of reduced endogenous cyc levels in S2Rϩ cells that reached only 50% of the transcript levels found in S2 cells, as determined by quantitative real-time PCR. In addition, pharmacological inhibition of CRM1-dependent nuclear export by leptomycin B significantly increased the amount of nuclear CLK, suggesting that rapid nuclear export causes an inefficient nuclear localization of CLK in S2Rϩ cells (Fig. 3A). Interestingly, deletion of a C-terminal consensus nuclear export signal (NES) resulted in a complete nuclear localization of the mutant CLK ⌬NES protein in most cells, indicating that this sequence is important for nuclear export of CLK (Fig. 3A). The CLK ⌬NES mutant activated transcription of a luciferase reporter gene in S2Rϩ cells similar as wild type (Fig. 3B) demonstrating that deletion of the NES did not compromise CLK function and allowed a native fold of the mutant protein. The results are consistent with a nucleocytoplasmic shuttling of the CLK protein, which, in the absence of CYC, results in a cytoplasmic localization of CLK, due to inefficient nuclear retention.
When the phosphorylation states of cytoplasmic and nuclear CLK proteins were investigated after expression in S2Rϩ cells, we observed that cytoplasmic CLK NLS/⌬NES was hypophosphorylated ( Fig. 3C and data not shown), whereas wild-type CLK showed predominantly hyperphosphorylated forms (Fig. 3C). Interestingly, the nuclear export-deficient CLK ⌬NES protein accumulated in an intermediate and a hyperphosphorylated state. The phosphorylation level of CLK ⌬NES was increased compared with nuclear import-deficient CLK NLS/⌬NES , indicating that nuclear translocation is important for hyperphosphorylation of CLK. On the other hand, CLK ⌬NES showed a stronger accumulation in the intermediate phosphorylation state compared with shuttling wild-type CLK (Fig. 3C), which was mainly hyperphosphorylated. These findings suggest that nuclear export promotes phosphorylation of the CLK protein and further phosphorylation events may take place in the cytoplasm after nuclear export. Because the CLK ⌬NES construct carries a 25-amino acid deletion of the NES, we validated these findings using a CLK ⌬NES construct that carries a 25-amino acid tag (25tag-CLK ⌬NES ) (supplemental Fig. S2). No significant differences were however observed in the phosphorylation patterns of CLK ⌬NES and 25tag-CLK ⌬NES (supplemental Fig. S2A). Similarly, 25tag-CLK ⌬NES accumulated in a phosphorylation state that was intermediate compared with mainly hyperphosphorylated wild-type CLK and hypophosphorylated cytoplasmic CLK NLS (supplemental Fig. S2B).
The above observations reveal a sequential and compartment-specific phosphorylation of the CLK protein, which starts with hypophosphorylation in the cytoplasm prior to nuclear import, followed by further phosphorylation in the nucleus toward an intermediate phosphorylation state, and finally proceeds to hyperphosphorylation, which is promoted by nuclear export. The observations are consistent with a transcriptionally active low or moderately phosphorylated form of nuclear CLK that is the major phosphorylation species in the evening of a circadian cycle, when CLK/CYC activity is at its peak ( Fig. 1) (24,29). During late night and early morning hypo-and strongly hyperphosphorylated CLK can be observed (Fig. 1) (29). The hypophosphorylated protein likely represents newly synthesized CLK that enters the nucleus, whereas the hyperphosphorylated form represents CLK protein from the previous circadian cycle that is exported and targeted to degradation.
Compartment-specific CLK Phosphorylation in Vivo-To analyze a compartment-specific phosphorylation of CLK in vivo, we performed a nucleocytoplasmic fractionation of fly heads that were collected in the morning 2 h after lights were on (ZT 2), and in the evening 2 h after lights were off (ZT 14) (Fig. 4A). In the morning, when CLK/CYC-dependent transcription is at its minimum, we found hyperphosphorylated CLK in the nucleus and in the cytoplasm, which likely represents CLK protein from the previous circadian cycle that is targeted to degradation (Fig. 4A). A weak band of hypophosphorylated CLK could be detected in the cytoplasm, consistent with newly synthesized CLK protein.
During early night, a strong band of intermediately phosphorylated CLK could be observed inside the nucleus, which correlates with maximal transcriptional activity of CLK/CYC (Fig.  4A). In addition, hyperphosphorylated CLK is present in the nucleus and in the cytoplasm. Interestingly, in the evening nuclear CLK protein showed a similar phosphorylation pattern as the CLK ⌬NES construct in cell culture. These findings support the observations in cell culture, indicating that hypophosphorylated cytoplasmic CLK protein is converted to an intermediately phosphorylated species after translocation into the nucleus. The accumulation of intermediately phosphorylated nuclear CLK correlates with CLK/CYC-activated transcription in the evening, which is consistent with a transcriptional activity of intermediately phosphorylated CLK. Significant amounts of intermediately phosphorylated CLK were only observed inside the nucleus (Fig. 4A). In contrast, hyperphosphorylated CLK is present in the nucleus as well as in the cytoplasm, which is consistent with a nuclear export of the CLK protein after transcriptional inhibition by PER-mediated hyperphosphorylation (24,25,35). Because in cell culture, nuclear CLK ⌬NES revealed a stronger accumulation in the intermediate phosphorylation state, whereas shuttling wild-type CLK accumulated mainly in a hyperphosphorylated form, nuclear export likely enhances the hyperphosphorylation of the CLK protein.
Nuclear Export Promotes Turnover of the CLK Protein-The above results suggested that nuclear export plays a role in the regulation of the CLK life cycle. We therefore analyzed compartment-specific degradation kinetics of the CLK protein in cell culture, to address a role of nuclear export for turnover of CLK (Fig. 4B). Degradation kinetics were analyzed by cycloheximide chase experiments for cytoplasmic CLK NLS , nuclear CLK ⌬NES , and shuttling wild-type CLK. We found that cytoplasmic and shuttling CLK proteins were degraded with similar kinetics, suggesting that phosphorylation events, which trigger CLK degradation, occur in the cytoplasm. CLK ⌬NES was significantly more stable, demonstrating that nuclear localization stabilizes the CLK protein and nuclear export is important for turnover of CLK.
CLK Is Stored in Nuclear Foci-An interesting observation from cell culture was the finding that CLK localized to nuclear foci (Fig. 5A). To investigate whether this localization pattern is relevant in vivo, we performed a detailed analysis of endogenous CLK localization in wild-type larval brains by immunohistochemistry (Fig. 5). Similar to the observations in cell culture we found that endogenous CLK localizes in nuclear foci of lateral neurons (Fig. 5B), demonstrating that the nuclear localization pattern is not an artifact of CLK expression in cell culture but occurs under normal physiological conditions in vivo. The size of the nuclear foci is similar to the size of secretory vesicles to which the circadian neuropeptide PDF localizes.
A detailed investigation of CLK localization in S2 cells using confocal microscopy confirmed the localization of wild-type CLK in nuclear foci (Fig. 6), whereas the CLK NLS mutant localized mainly to the cytoplasm (see also Fig. 2). Surprisingly, a    AUGUST 28, 2009 • VOLUME 284 • NUMBER 35 weak nuclear staining could however be observed for the NLS mutant (Fig. 6A). In contrast to wild-type CLK, the small amount of nuclear CLK NLS revealed a homogeneous distribution. When cells that expressed different levels of wild-type CLK were compared, those cells that expressed high levels of CLK showed an accumulation of CLK in nuclear foci, whereas cells that expressed lower levels of CLK revealed a more homogeneous nuclear staining, similar to the small amount of nuclear CLK NLS (Fig. 6A). These results suggested that the amount of nuclear CLK affects the subnuclear distribution.

Sequential Post-translational Regulation of CLOCK
We next asked whether the CLK protein present in nuclear foci or the homogeneously distributed CLK represents the transcriptionally active species. Because CLK NLS showed a weak but homogeneous distribution in the nucleus, we analyzed the transcriptional activity of this mutant in cell culture. Surprisingly, CLK NLS activated the expression of a luciferase reporter gene almost as efficient as wild-type CLK (Fig. 6B). Because only small amounts of CLK NLS enter the nucleus and show a homogeneous nuclear distribution (Figs. 2 and 5A), the results suggest that homogeneously distributed nuclear CLK is the transcriptionally active species.

DISCUSSION
The heterodimeric complex of the transcription factors CLK and CYC forms the core oscillating activity of the circadian clock in Drosophila and mammals, ultimately controlling genome wide transcription and activity states of key regulatory factors in a temporal fashion. The circadian oscillation of CLK/ CYC activity thereby allows a temporal orchestration of homeostatic functions and their synchronization with the environmental cycles of day and night. Although precise timing of CLK/CYC activation and inhibition appears to be crucial for clock function and accurate timing of circadian transcription (36), little is known about the molecular mechanisms that facilitate these regulatory events. CYC is constitutively expressed in Drosophila, whereas CLK transcription and protein phosphorylation reveal robust oscillations over the course of a day (16,29). A number of studies indicated an important role of posttranslational regulation of the CLK protein for clock function and accurate timing of circadian transcription (15,24,25,29,37). Because many transcription factors are regulated at the level of their cellular distribution, we investigated the contribution of nucleocytoplasmic transport and compartment-specific phosphorylation to the control of CLK/CYC activity.
We found that CLK shuttles between the nucleus and the cytoplasm and interaction with the heterodimerization partner CYC promotes an efficient nuclear localization. Similar nucleocytoplasmic shuttling has been observed for the mammalian CYC homolog BMAL1, which promotes the nuclear localization of mCLK (38 -40). Because CLK is the limiting factor for circadian transcription (41), CYC levels are likely sufficient to facilitate the nuclear localization of CLK in neurons throughout a circadian cycle (Fig. 1). Nucleocytoplasmic shuttling of CLK suggests a regulatory role of cellular transport for trans-activation by CLK/CYC. Such regulation is supported by oscillating total and nuclear amounts of CLK protein in circadian neurons of the larval brain (Fig. 1) and by compartment-specific phosphorylation states of the CLK protein (Figs. 2-4).
Cytoplasmic CLK is hypophosphorylated, while phosphorylation increases after nuclear import (Figs. 3 and 4), similar to the situation in mammals (30,34). The initial phosphorylation of newly synthesized CLK is likely crucial for nuclear import, because we identified consensus phosphorylation sites that prevent nuclear import after mutagenesis to the constitutively unphosphorylated form (data not shown). Upon nuclear import, the CLK protein was found to be further phosphorylated (Fig. 3). Similarly, a weak band of hypophosphorylated CLK was observed in cytoplasmic fractions of fly head extracts during morning, while in the evening intermediate and hyperphosphorylated forms of CLK accumulated in the nucleus in vivo (Fig. 4). These findings reveal a temporal and compartmentspecific sequence of phosphorylation events prior to, and after nuclear translocation of the CLK protein.
Phosphorylation of CLK in the nucleus is likely linked to activation and inhibition of CLK/CYC-dependent transcription. The enhanced intermediate phosphorylation state of nuclear CLK ⌬NES in cell culture (Fig. 3), together with the accumulation of an intermediately phosphorylated form of nuclear CLK during the evening of a circadian cycle in vivo (Fig. 4), correlates with CLK/CYC-dependent transcription, indicating that an intermediately phosphorylated form of CLK is transcriptionally active. Recently, an inhibitory PER-dependent hyperphosphorylation of chromatin-bound CLK by the casein kinase I⑀ homolog DOUBLE-TIME has been shown (24,35). PER-mediated hyperphosphorylation is thought to trigger inhibition and degradation of the CLK protein (24,25,35). Interestingly, hyperphosphorylated forms of CLK are present in nuclear and cytoplasmic fractions of fly head extracts, while the intermediate phosphorylation form is mainly nuclear (Fig. 4). This finding suggests that hyperphosphorylated CLK is exported from the nucleus after transcriptional inhibition. Similarly, export-deficient nuclear CLK ⌬NES revealed a reduced hyperphosphorylation compared with shuttling wild-type CLK, suggesting that nuclear export promotes the hyperphosphorylation of CLK. These observations in cell culture and in vivo indicate that nuclear export is an integral step of the CLK life cycle. This conclusion is supported by the finding that nuclear CLK protein is more stable than cytoplasmic and shuttling wild-type CLK. Nuclear export therefore promotes degradation of the CLK protein. Because wild-type CLK and a cytoplasmic CLK mutant reveal similar degradation kinetics, the modifications that trigger CLK degradation, such as phosphorylation and/or ubiquitination, likely occur in the cytoplasm.
In conclusion, these results uncover a sequential and compartment-specific phosphorylation of the CLK protein, which assigns specific localization and activity patterns to the phosphorylation states of CLK that are found to oscillate over the course of a day (Fig. 7). Importantly, the subcellular and, as discussed later, subnuclear distribution together with the sequential and compartment-specific phosphorylation of CLK provides time of day specific modification states of the CLK protein throughout the circadian cycle (Fig. 7). In the morning, when mRNA levels and new protein synthesis of CLK are high, newly synthesized, hypophosphorylated forms of CLK that enter the nucleus can be observed, together with hyperphosphorylated CLK from the previous circadian cycle that is targeted to degradation after nuclear export (Figs. 1 and 4) (29). In the evening, when CLK/CYC transcriptional activity reaches its peak, intermediate phosphorylated forms of CLK are most abundant (Figs. 1 and 4) (24,25,29). These findings identify a temporal and compartment-specific sequence of phosphorylation events that allow a precise post-translational regulation of the CLK life cycle and likely provide a core post-translational interval-timer of the circadian clock.
Interestingly, localization of CLK in nuclear foci was not only observed in cell culture, but also under endogenous conditions in neurons of wild-type larval brains (Fig. 5). These foci likely contain a storage form of the CLK protein, because the heterodimerization partner CYC revealed a homogeneous nuclear distribution (30,42). In addition, cells that express low levels of CLK revealed a homogeneous distribution of nuclear CLK protein, indicating that excessive amounts of CLK are targeted to the foci (Fig. 6). The CLK NLS mutant localized mainly to the cytoplasm, but a small amount of the mutant protein was found inside the nucleus with a homogeneous distribution (Fig. 6). Interestingly, the CLK NLS mutant activated transcription almost as efficient as wild-type CLK, revealing that the small amount of homogeneously distributed nuclear CLK NLS is transcriptionally active. To this end the post-translational modifications that target CLK into nuclear foci are not known. These modifications may involve phosphorylation or alternatively sumoylation. BMAL1 was shown to be sumoylated (43,44), and this modification is known to control transcription factor activities by targeting to specific subnuclear compartments (45,46). Storage of CLK in nuclear foci provides an explanation for the puzzling observation that overexpression, as well as ectopic expression of Clk from a per-promoter in anti-phase to its endogenous rhythm is able to sustain normal clock function (15). The oscillation of Clk transcription may contribute to the robustness of molecular oscillations, but apparently is not essential for clock function. Storage of CLK in nuclear foci sug-gests that as long as these storage compartments are supplied with new CLK protein, molecular oscillations can be sustained. Activation and re-distribution of CLK from these subnuclear compartments may instead be crucial for clock function and timing of circadian transcription.
In conclusion, our results show a sequential and compartment specific phosphorylation of the CLK protein that is accompanied by nucleocytoplasmic transport and specific subnuclear distribution (Fig. 7). The specific phosphorylation patterns observed prior to nuclear import, inside the nucleus and after transit through the nucleus assign specific cellular localization and activity patterns to the different phosphorylated species of the CLK protein that oscillate over the course of a circadian cycle. These findings suggest a post-translational timing mechanism of the circadian clock that is based on a temporally controlled sequential phosphorylation and specific subcellular localization of the circadian transcription factor CLK.