The Circadian Regulatory Proteins BMAL1 and Cryptochromes Are Substrates of Casein Kinase Iε*

The serine/threonine protein kinase casein kinase I ε (CKIε) is a key regulator of metazoan circadian rhythm. Genetic and biochemical data suggest that CKIε binds to and phosphorylates the PERIOD proteins. However, the PERIOD proteins interact with a variety of circadian regulators, suggesting the possibility that CKIε may interact with and phosphorylate additional clock components as well. We find that CRY1 and BMAL1 are phosphoproteins in cultured cells. Mammalian PERIOD proteins act as a scaffold with distinct domains that simultaneously bind CKIε and mCRY1 and mCRY2 (mCRY). mCRY is phosphorylated by CKIε only when both proteins are bound to mammalian PERIOD proteins. BMAL1 is also a substrate for CKIεin vitro, and CKIε kinase activity positively regulates BMAL1-dependent transcription from circadian promoters in reporter assays. We conclude that CKIε phosphorylates multiple circadian substrates and may exert its effects on circadian rhythm in part by a direct effect on BMAL1-dependent transcription.

Circadian rhythms allow organisms to optimize their metabolic and physiologic behavior in response to the 24 h day. The rhythm is generated by cell-autonomous transcription-translation feedback loops highly conserved in outline if not in detail in most eukaryotes studied to date (recently reviewed in Refs. [1][2][3][4]. Extensive genetic investigation coupled with molecular studies has uncovered many of the essential elements of the metazoan clock. The central feature of the clock is transcription, regulated by a heterodimeric transcription factor that drives expression of genes whose protein products are negative regulators of their own transcription. This negative feedback loop establishes stable molecular oscillations with a period of ϳ24 h. In mammals, the transcription factors are the PAS domain-containing proteins CLK 1 and BMAL1, whereas the negative factors include the PERIOD proteins PER1 and PER2 and the cryptochromes CRY1 and CRY2. The stability and period of the circadian oscillations are likely to be dependent on multiple factors, including the rate at which these regulators accumulate in the cell (i.e. the sum of their synthesis and degradation rates) and the rate at which they accumulate in the nucleus. Finally, the activities of circadian regulators are modified by post-translational modifications.
Analysis of animals and humans with altered circadian rhythms demonstrates the importance of phosphorylation in the regulation of the molecular clock. Mutations in the casein kinase I⑀ (CKI⑀) homolog double-time (dbt) in Drosophila markedly lengthen or shorten the circadian period, depending on the allele (5)(6)(7)(8). In hamsters, a point mutation in CKI⑀ that decreases kinase activity causes the semidominant short period tau phenotype (9). The functional role of CKI⑀ within the circadian clock has not been fully elucidated (1,10). A number of studies indicate the PER proteins are critical targets of CKI⑀. dbt-mutant flies have altered patterns of PER phosphorylation and accumulation (6,8,11). Mutations in a CKI⑀ phosphorylation site in human PER2 have been identified in a family with advanced sleep phase syndrome (FASPS), a dominantly inherited short circadian period disorder (12). Overexpression of CKI⑀ leads to decreased mPER1 protein half-life and alters the nucleocytoplasmic localization of the proteins (13)(14)(15). Whether CKI⑀ can phosphorylate other circadian regulatory proteins has not yet been explored.
One central feature of the circadian regulatory machinery is the multiple interactions between the clock proteins. CKI⑀ forms stable complexes with PER proteins. PER proteins heterodimerize with each other and interact with CRY1 and CRY2 (16 -18). CRY1 and CRY2 also interact with the heterodimeric transcription factor CLK/BMAL1 both functionally and in yeast two-hybrid assays (19,20). Hence, the potential for higher order complexes exists. Given the changing abundance of these factors over time, different complexes are likely to form at different times of day.
In the present study, we examined the ability of CKI⑀ to interact with and phosphorylate other clock proteins in vitro and in cell culture, both directly and when in multiprotein complexes. We find that mCRY1 and mCRY2 are substrates of CKI⑀ but only when both CRY and kinase are bound to PER1 or PER2. Furthermore, mCRY1 can overcome the CKI⑀-dependent cytoplasmic retention of mPER1 seen in human embryonic kidney 293 (HEK 293) cells and relocalize the entire multimeric complex to the nucleus. The ability of CKI⑀ to phosphorylate CLK and BMAL1 was examined as well. BMAL1 but not CLK is a substrate of CKI⑀ in vitro. BMAL1 phosphorylation by CKI⑀ is neither dependent on nor stimulated by the presence of PER or CRY proteins. BMAL1 is also a phosphoprotein when expressed in HEK 293 cells, and co-expression of a dominant-negative CKI⑀ inhibits BMAL1 phosphorylation. Inhibition of CKI⑀ activity using dominant-negative CKI⑀ or dsRNA-mediated interference also led to decreased CLK/BMAL1-dependent transcription. The data suggest CKI⑀ has multiple functional substrates in the circadian rhythm regulatory apparatus.

MATERIALS AND METHODS
Cell Line Maintenance and Transfection-HEK 293 cells were maintained in Dulbecco's modified Eagle's medium with 10% fetal calf serum in a humidified incubator at 37°C and 5% CO 2 . Cells were transfected using LipofectAMINE PLUS (Invitrogen) with between 0.3 and 0.7 g of expression constructs, as indicated in the figures, with the total amount of transfected DNA brought up to 1 g with the addition of empty vector.
Immunoprecipitations and Immunofluorescence-HEK 293 cells were transfected with the indicated expression constructs and harvested 18 h later in HTS buffer (10 mM HEPES, pH 7.5, 0.1% Triton X-100, 100 mM NaCl, 2 mM EDTA, 2 mM EGTA, and 2 mM DTT, supplemented with Complete TM (Roche Molecular Biochemicals) protease inhibitors). Myc epitope-tagged proteins were immunoprecipitated with 1 g of 9E10 anti-Myc monoclonal antibodies and protein Aagarose beads. The beads were then washed three times in HTS buffer, and proteins were eluted with SDS-PAGE loading buffer. After SDS-PAGE, the proteins were transferred to a nitrocellulose membrane (Hybond-C extra, Amersham Biosciences) and further analyzed by immunoblotting using the indicated antibodies. For immunofluorescence, HEK 293 cells were transiently transfected as indicated in Fig. 3. 18 h after transfection, the cells were fixed in 4% paraformaldehyde and permeabilized with 0.1% Triton X-100. Anti-Myc (9E10) and anti-influenza hemagglutinin epitope (12CA5) monoclonal antibodies were directly conjugated to Alexa 488 (green) or Alexa 594 (red) according to the manufacturer's directions (Molecular Probes). The antibodies and 4,6-diamidino-2-phenylindole nuclear stain were incubated with the cells for 1 h at room temperature. The epitope-tagged proteins were then visualized using an Olympus BX50 fluorescence microscope equipped with a cooled charge-coupled device camera (Photometrics Ltd.) at 60ϫ magnification. Images were acquired using the mFISH (Vysis, Inc) software package and further analyzed with PhotoShop 5.0 (Adobe).
Kinase Assays-HEK 293 cells were transiently transfected with expression constructs as indicated in the figures. Eighteen hours after transfection, Myc epitope-tagged proteins were immunoprecipitated from 300 g of soluble cell free extract with 1 g of 9E10 anti-Myc monoclonal antibodies as was done above. Beads were kept on ice and washed several times with HMB buffer (30 mM HEPES, pH 8, 7 mM MgCl 2 , 100 g/ml bovine serum albumin, and 25 M ATP). After washing, the beads were split three ways. To one bead fraction, HMB, [␥-32 P]ATP, and 50 ng of recombinant CKI⑀⌬C320 was added to a final volume of 20 l. This truncated form of CKI⑀ was used to avoid the auto-inhibition seen with full-length CKI⑀ in vitro (21). Another bead fraction was treated identically except for the omission of recombinant kinase as a negative control. The beads were then incubated at 37°for 15 min. The reactions were stopped by washing the beads three times with HM buffer (30 mM HEPES, pH 8, 7 mM MgCl 2 ). Proteins were eluted from the beads with Laemmli sample buffer and separated by SDS-PAGE, and dried gels were analyzed by PhosphorImager (Amersham Biosciences). To confirm the presence of proteins in the immunoprecipitate, the third bead fraction was also run out on SDS-PAGE and analyzed by immunoblotting as above.
For in vitro kinase assays (Fig. 5A), V5 epitope-tagged CLK and BMAL1 were synthesized in rabbit reticulocyte lysate using the TNT Coupled Reticulocyte System (Promega). The proteins were immunoprecipitated from 90 l of each synthesis reaction with 3 g of anti-V5 antibodies. Bead washes, in vitro kinase assays, and Western blotting were performed as described above.
Metabolic Labeling of Cells-HEK 293 cells were transiently transfected as indicated in Fig. 4, A and B. Eighteen hours after transfection, cells were labeled for 3 h with [ 32 P]orthophosphoric acid (2 mCi/ml) in phosphate-free Dulbecco's modified Eagle's medium. After labeling, cells were washed and then lysed in radioimmune precipitation buffer (1% Nonidet P-40, 0.5% deoxycholic acid, 0.1% sodium dodecyl sulfate, 150 mM NaCl, 150 mM Tris, pH 8.0) supplemented with 2 mM dithiothreitol, 10 mM sodium fluoride, 10 mM ␤-glycerol phosphate, 200 nM okadaic acid, and Complete TM (Roche Molecular Biochemicals) protease inhibitors). Myc epitope-tagged CRY or BMAL1 was immunoprecipitated from 1 mg of soluble lysate with 8 g of 9E10 anti-Myc antibodies and protein A-agarose beads. After washing, bound proteins were eluted with Laemmli sample buffer and examined by SDS-PAGE and phosphorimaging analysis.
Transcription Assays and RNA Interference-HEK 293 cells seeded in a single well of a six-well dish were transiently transfected with between 25 and 250 ng of luciferase reporter plasmid using Lipo-fectAMINE Plus (Invitrogen) according to the manufacturer's direc-tions. Where indicated, 250 ng each of CLK and BMAL1 and 100 ng of CKI⑀ or CKI⑀(K38A) expression constructs were also co-transfected. In all cases 25 ng of a plasmid encoding ␤-galactosidase was added, and the total amount of DNA transfected was brought up to 1 g with the addition of empty vector. After 24 h, cell-free extracts were made by the addition of lysis buffer (100 mM potassium phosphate, pH 7.8, 0.2% Triton X-100, and 0.5 mM DTT). Luciferase activity was detected using the Tropix Dual-Light assay system (PE Biosystems) and measured using a microtiter plate luminometer and the Revelation MLX software package (Dynex Technologies). ␤-Galactosidase activity was measured as a transfection efficiency control and was used to normalize the luciferase data.
Endogenous CKI⑀ and CKI␦ mRNA was targeted in HEK 293 cells by the addition of a 21-nucleotide duplex small interfering RNA (siRNA). This duplex RNA targets nucleotides 520 -538 of the CKI⑀-coding sequence and was constructed using the ribooligonucleotide (r) pair with the sequences 5Ј-rCUGGGGAAGAAGGGCAACCdTT-3Ј and 5Ј-rG-GUUGCCCUUCUUCCCCAGdTT-3Ј. The identical nucleotide sequence is present in CKI␦ as well. As a control for the specificity of this duplex, a ribooligonucleotide pair with the inverse sequence was used with the sequences 5Ј-rCCAACGGGAAGAAGGGGUCdTT-3Ј and 5Ј-rGAC-CCCUUCUUCCCGUUGGdTT. The oligonucleotides were annealed essentially as described in Elbashir et al. (22). 4.5 g of each duplex was then introduced into 6-well tissue culture plate using Oligofectamine according to the manufacturer's directions (Invitrogen). After 48 h, the transfection mixture was removed, and the cells were transfected with mPer1-luc reporter, CLK, and BMAL1 expression constructs as described above. Twenty-four hours after transfection, luciferase activity was detected and normalized to ␤-galactosidase activity. CKI␦ and CKI⑀ abundance was assessed by immunoblot using UT31, a rabbit polyclonal antibody that recognizes a shared amino-terminal epitope on CKI⑀ and CKI␦ (23).

RESULTS
mCRY Binds a Conserved Domain in the Carboxyl Terminus of PER-Genetic, biochemical, and immunofluorescence data suggest there is a physical interaction between mCRY and mPER (17,18,20). To map the region of mPER1 and mPER2 that bind mCRY1 and mCRY2, amino-and carboxyl-terminal truncations of mPER1 and mPER2 were co-expressed with full-length mCRY1 or mCRY2, and the mPER protein was immunoprecipitated. As expected, mPER1 efficiently immunoprecipitated full-length mCRY1 (Fig. 1A, lane 1Ј). mCRY1 also binds mPER1⌬1219, a truncation of mPER1 lacking the carboxyl-terminal 72 residues (Fig. 1A, lane 5Ј). However, when mPER1 was further truncated by an additional 100 residues (mPER1⌬1118), all detectable mCRY binding was lost (Fig. 1A, lane 6Ј). These results indicate that mCRY1 binds to mPER1 in its carboxyl-terminal region bound by amino acids 1117 and 1219, outside the CKI⑀ binding domain. Supporting this, a fragment of mPER1 encompassing amino acids 1027 and 1291, mPER1(1027-1291), also binds mCRY1 (Fig. 1A, lane 4Ј). The efficiency of the mCRY1 interaction with mPER1(1027-1291) and mPER1⌬1219 is less than that seen with full-length mPER1, suggesting that additional contacts between CRY and PER proteins may occur in the full-length protein (Fig. 1A, lanes 4Ј and 5Ј). As reported by others, mCRY does not require the PAS domains of mPER for stable interaction (Fig. 1A, lane 2Ј) (20). The mCRY1 binding domain of mPER2 similarly maps to between residues 1127 and 1225 (Fig. 1B). mCRY2 interacts with mPER1 and mPER2 in the same domain as does mCRY1 (data not shown). This CRY binding region of mPER1 and mPER2 is conserved among vertebrate PER proteins but is absent in insect PER and other proteins in GenBank TM (data not shown).
mPER Is a Scaffold That Brings mCRY1 and CKI⑀ into a Multimeric Complex-mPER1 and mPER2 bind to both endogenous and overexpressed CKI⑀ using a conserved domain in the approximate center of PER (e.g. mPER1 amino acids 496 -815) (12,13). We therefore tested whether CKI⑀ and CRY could simultaneously bind to PER, forming a ternary complex in intact cells. Full-length mCRY1 was expressed with mPER1 or mPER2 as indicated in Fig. 2. mCRY was immunoprecipitated, and the presence of mPER in the pellet was probed by Western blotting. mCRY1 efficiently interacted with both full-length mPER1 and mPER2 (Fig. 2, lanes 2Ј and 4Ј) but not mPER1⌬1118 or mPER2⌬905 (Fig. 2, lanes 3Ј and 5Ј), confirming that the carboxyl terminus of mPER is required for interaction. Immunoprecipitation of mCRY1 alone in the absence of mPER1 did not pull down endogenous CKI⑀ (Fig. 2, lane 1Ј). However, when either mPER1 or mPER2 was co-expressed, mCRY1 consistently co-immunoprecipitated both mPER and CKI⑀ (Fig. 2, lanes 2Ј and 4Ј and data not shown). Consistent with this result, when mPER1⌬1118 (deficient in mCRY binding) was co-expressed with mCRY1, neither truncated mPER1⌬1118 nor CKI⑀ co-immunoprecipitated with mCRY1 (Fig. 2, lane 3Ј). CKI⑀ also failed to immunoprecipitate with mCRY1 when mPER2⌬905 was expressed (Fig. 2, lane 5Ј). Taken together, these results indicate that CKI⑀ and mCRY1 are able to interact with each other only indirectly and that they require the presence of mPER to bring them into a trimeric complex. Thus, the mPER protein acts as a scaffold that allows formation of a mCRY1⅐CKI⑀⅐mPER multimeric complex.
mCRY1 Brings the Cytoplasmic mPER1⅐CKI⑀ Complex into the Nucleus-PER subcellular localization is a dynamic process regulated by at least two nuclear localization signals, a nuclear export signal, protein-protein interactions, and protein phosphorylation (11,15,24). CKI⑀ kinase activity can influence mPER1 subcellular localization by either sequestering it in the cytoplasm in HEK 293 cells (13) or driving it to the nuclear compartment in COS7 cells (15). mCRY1 protein is predominately localized to the nuclear compartment in transfected cells, where it inhibits CLK/BMAL1-dependent transcription (18). mCRY1 also binds and promotes nuclear translocation of mPER1 and mPER2 proteins (18). These observations, when taken together with the fact that mCRY1, mPER1, and CKI⑀ form a multiprotein complex suggest at least two possibilities with respect to subcellular localization. One, mCRY1 may translocate the CKI⑀⅐mPER1 complex to the nucleus, overcoming the CKI⑀-dependent mPER1 cytoplasmic localization seen in HEK 293 cells. Conversely, CKI⑀ may promote mCRY1 cytoplasmic localization that is dependent on its interaction with mPER1.
To distinguish between these two possibilities, epitopetagged CKI⑀ and mCRY1 were transiently expressed in HEK 293 cells with empty vector and either full-length or truncated mPER1 (Fig. 3), and their intracellular localization was examined by direct immunofluorescence microscopy. In the absence of mPER1, mCRY1 was nuclear in 97% of cells examined, whereas in the same cells, CKI⑀ was predominantly cytoplasmic (only nuclear in 3% of cells) (Fig. 3, panels a, d, and g). However, when mPER1 was co-expressed with CKI⑀ and mCRY1, CKI⑀ localization changed from predominantly cytoplasmic to nuclear (Fig. 3, panels b, e, and h). Thus, mPER1 allows mCRY1 to bring CKI⑀ to the nucleus despite the CKI⑀ kinase activity that otherwise promotes mPER1 cytoplasmic localization in HEK 293 cells (13). Consistent with this model, when the mCRY1 binding-deficient form of mPER1 (mPER1⌬1118) was co-expressed, CKI⑀ remained predominantly in the cytoplasm (Fig. 3, panels c, f, and i).
These results, when taken together with the immunoprecipitation data (Fig. 2), suggest that mPER1 must interact with both CKI⑀ and mCRY1 in order for mCRY1 to bring CKI⑀ into the nucleus. We note that there was some accumulation of CKI⑀ in the nucleus when mPER1⌬1118 was expressed (compare Fig. 3, panels d-f), again suggesting there is some additional interaction between mPER1⌬1118 and mCRY1 (Fig. 2).
mCRY1 Requires mPER1 or mPER2 to Be Phosphorylated by CKI⑀ in Vitro-The ability of CKI⑀ to phosphorylate PER1 requires that the kinase first bind to the substrate, presumably ensuring a high local concentration and/or appropriate positioning of the kinase (Ref. 13 and data not shown). Because mCRY1 can be found in a multimeric complex with PER and CKI⑀, mCRY1 present in this complex might similarly be phosphorylated by CKI⑀. We first assessed whether mCRY1 was in fact a phosphoprotein in intact cells. Epitope-tagged mCRY1 was transiently expressed in 32 P-metabolically labeled HEK 293 cells, immunoprecipitated, and analyzed by autoradiography (Fig. 4A). mCRY1 incorporated significant amounts of 32 P during a 3-h labeling, indicating it is in fact phosphorylated in cultured cells.
To test whether CKI⑀ could phosphorylate mCRY1, mCRY1 alone and mCRY1⅐mPER complexes were assembled by coexpression in HEK 293 cells, and the proteins were recovered by immunoprecipitation of mCRY1 (Fig. 4C). When mCRY1, expressed alone, was immunoprecipitated and incubated with recombinant CKI⑀, there was minimal phosphorylation of mCRY1 (Fig. 4B, lane 2). However, when full-length mPER1 or mPER2 was co-expressed with mCRY1, mCRY1 was efficiently phosphorylated by CKI⑀ (Fig. 4B, lanes 4 and 8). Phosphorylation of mCRY1 seen without added CKI⑀ may be due to endogenous CKI⑀ and/or CKI␦ co-precipitating with PER1 and PER2 (Fig. 4B, lanes 3 and 7). The interaction of CRY with mPER was required for CRY phosphorylation by CKI⑀, since mCRY1 phosphorylation fell back to background levels when truncated mPER proteins that do not interact with mCRY1 were co-expressed with mCRY1 (Fig. 4B, lanes 6 and 10). mCRY2 was similarly phosphorylated by CKI⑀ only in the presence of PER containing both CKI⑀ and CRY binding sites (data not shown). These results suggest that PER acts as a scaffold that brings mCRY and CKI⑀ into close proximity with each other. This effectively raises the local concentration of CKI⑀ around mCRY, allowing for efficient phosphorylation of mCRY by CKI⑀.
To test the potential biological function of mCRY phosphorylation by CKI⑀, we tested the effect of CKI⑀ co-expression on the ability of CRY1 and CRY2 to repress transcription from the mPer1 promoter (18). No consistent effect of active or dominant-negative CKI⑀ on CRY activity was found (data not shown). Similarly, no effect of dominant-negative CKI⑀ (K38A) on the level of CRY1 phosphorylation in transfected cells was detectable. Thus, although CKI⑀ can clearly phosphorylate mCRY in a multimeric complex in vitro, the physiological significance of this phosphorylation remains to be determined. BMAL1 Phosphorylation Is Regulated by CKI⑀-CKI⑀ is found both in the nucleus and cytoplasm depending on cell and tissue type (13,15,25). Depending on its subcellular localization CKI⑀ may phosphorylate distinct sets of substrates. Because CKI⑀ forms intermolecular complexes with and phosphorylates clock proteins such as mPER1, mPER2, and mCRY1, we next asked if CKI⑀ could phosphorylate other circadian regulators. CKI⑀ was able to phosphorylate BMAL1 immunoprecipitated from programmed reticulocyte lysates and from transfected cells (Fig. 5, A and B). Immunoprecipitated CLK was not phosphorylated by CKI⑀ (Fig. 5A). We did not observe a reproducible, robust interaction between CKI⑀ and BMAL1 by co-immunoprecipitation assays. Unlike the stimulation of mCRY phosphorylation seen in the presence of mPER, BMAL1 phosphorylation by CKI⑀ was not appreciably altered by the A, mCRY1 is a phosphoprotein in intact cells. HEK 293 cells were transiently transfected with either empty vector (Ϫ) or with a plasmid encoding Myc-tagged mCRY1 (ϩ). 18 h after transfection, cells were metabolically labeled with [ 32 P]orthophosphoric acid for 3 h, at which time cell free lysates were prepared and subjected to immunoprecipitation with anti-Myc antibody. 32 P incorporation into mCRY1 was assessed by SDS-PAGE and phosphorimaging analysis. B, mCRY1 is phosphorylated by CKI⑀ in a CKI⑀⅐PER⅐mCRY1 complex in vitro. HEK 293 cells were transiently transfected with mCRY1 and PER expression plasmids as indicated. mCRY1 protein was then immunoprecipitated (IP) with anti-Myc antibodies, and the immunoprecipitates were split three ways. Two of the three fractions were subjected to an in vitro kinase assay in the presence of [␥-32 P]ATP and without (Ϫ) or with (ϩ) added recombinant CKI⑀(⌬C320). mCRY1 phosphorylation was then visualized by SDS-PAGE and phosphorimaging. C, the third bead fraction was subjected to immunoblotting after SDS-PAGE to confirm the presence of CRY and PER proteins in the immunoprecipitates as indicated in the figure. formation of a BMAL1⅐mPER2 complex (Fig. 5B and data not shown). BMAL1 was found by metabolic labeling to be a phosphoprotein in cells (Fig. 5C). Because CKI⑀ was able to directly phosphorylate BMAL1 in vitro, the effect of overexpression of a dominant-negative form of CKI⑀ on BMAL1 phosphorylation in intact cells was also examined. CKI⑀(K38A) co-expression reduced BMAL1 phosphorylation by about 40% (Fig. 5, C and D). The decrease in BMAL1 phosphorylation was not due to a decrease in BMAL1 protein levels, which were unchanged regardless of the presence or absence of CKI⑀(K38A) (Fig. 5C,  lower panel). The partial reduction in BMAL1 phosphorylation could be due to complete inhibition of phosphorylation of a subset of sites or a partial reduction in CKI⑀ activity. Phosphopeptide mapping of BMAL1 expressed without and with CKI⑀(K38A) demonstrated that there was a global decrease in BMAL1 phosphorylation (data not shown), consistent with CKI⑀(K38A) partially blocking the endogenous kinases phosphorylating BMAL1.
CKI⑀ Regulates BMAL1-dependent Transcription-Phosphorylation regulates the activity of multiple transcription factors (26, 27). Because BMAL1 is phosphorylated by CKI⑀ in vitro, and expression of a dominant-negative form of CKI⑀ decreases BMAL1 phosphorylation in intact cells, we examined whether CKI⑀ regulated the activity of a CLK/BMAL1-dependent promoter.TranscriptionalactivitywasassessedbytheCLK/BMAL1dependent expression of luciferase from the mPer1 promoter (Fig. 6A). Co-expression of active CKI⑀ had no effect on luciferase expression, whereas expression of dominant-negative CKI⑀, CKI⑀(K38A), caused a 40% decrease in luciferase activity. This change in luciferase expression is similar in magnitude to the decrease in BMAL1 phosphorylation seen with CKI⑀(K38A) co-expression. CKI⑀ may regulate the expression of multiple genes, as it has been implicated in processes as diverse as Wnt/␤-catenin signaling and NFAT (nuclear factor of activated T cells) nuclear localization (28,29). However, CKI⑀(K38A) is not a global transcriptional repressor, as it had no effect on luciferase expression from actin and cdc2 promoters (Fig. 6, B and C).
To further assess the role of CKI⑀ in regulation of circadian E box-containing promoters, we attempted to eliminate CKI⑀ and CKI␦ expression using siRNA (22). A 21-base pair RNA duplex (CKI⑀ nucleotides 520 -538, a sequence conserved in FIG. 5. BMAL1 phosphorylation is regulated by CKI⑀ in vitro and in intact cells. A, BMAL1 but not CLK is a substrate of CKI. V5 epitope-tagged CLK (lanes 1, 3, and 4) and BMAL1 (lanes 2, 5, and 6) were expressed in rabbit reticulocyte lysates and immunoprecipitated, and the beads were split three ways. One fraction was probed with anti-V5 antibodies to verify the presence of protein in the pellet ( lanes  1 and 2), and the remaining fractions were subjected to an in vitro kinase reaction in the absence (lanes 3 and 5) or presence (lanes 4 and 6) of added recombinant CKI⑀(⌬C320) as indicated. B, BMAL1 was phosphorylated by CKI⑀ in vitro independent of mPER2. HEK 293 cells were transiently transfected with plasmids encoding Myc-tagged BMAL1 along with FLAG-tagged mPER2 or empty vector. The Myctagged BMAL1 was immunoprecipitated and incubated with [␥-32 P]ATP and without (Ϫ) or with (ϩ) recombinant CKI⑀(⌬C320). Co-immunoprecipitation of mPER2 with BMAL1 was confirmed by immunoblotting a fraction of the immunoprecipitates with anti-FLAG antibodies (lanes 5 and 6). The presence of co-immunoprecipitating mPER2 did not stimulate phosphorylation of BMAL1. C, BMAL phosphorylation in intact cells is reduced by co-expression of dominantnegative CKI⑀. HEK 293 cells were transiently transfected with plasmids encoding Myc epitope-tagged BMAL1 and either vector (Ϫ) or a kinase-inactive form of CKI⑀(K38A) (ϩ). At 18 h post-transfection, cells were labeled with [ 32 P]orthophosphoric acid for 3 h, and then the BMAL1 protein was immunoprecipitated (IP) from lysates with anti-Myc monoclonal antibodies. 32 P incorporation into BMAL1 was assessed by SDS-PAGE and phosphorimaging analysis. BMAL1 protein expression was determined by immunoblot of cell free lysate. The data shown is a representative result of three separate experiments. D, Quantitation of BMAL1 phosphorylation without (-) or with (ϩ) coexpression of CKI⑀(K38A). The results shown are the average of three independent experiments.
FIG. 6. CKI⑀ regulates BMAL1 driven transcription. A, overexpression of dominant-negative CKI⑀ leads to decreased BMAL1-driven transcription. HEK 293 cells were transiently transfected with a CLK⅐BMAL1-responsive promoter (mPer1:luc) and CLK, BMAL1, and CKI⑀ expression constructs or empty vector as indicated. After 18 h, cell free extracts were prepared, and the luciferase activity was analyzed. B and C, CKI⑀(K38A) has minimal effect on other promoters. HEK 293 cells were transiently transfected with plasmids encoding either CKI⑀ or dominant-negative CKI⑀(K38A) and a reporter with the actin promoter (B) or the cdc2 promoter (C) driving luciferase expression. 18 h after transfection, luciferase activity was analyzed as in A. D, dsRNAmediated interference leads to depletion of CKI⑀/␦ and inhibition of CLK/BMAL1-driven gene expression. Endogenous CKI⑀/␦ was partially eliminated in HEK 293 cells by transfection of a 21-nucleotide RNA duplex (siRNA) directed against the CKI⑀/␦ sequence. As a control, the sequence was introduced in the inverted orientation (Inv). After 48 h, the transfection mixture was removed, and plasmids encoding both CLK and BMAL1 (where indicated) along with a reporter containing luciferase behind the mPer1 promoter (18) were introduced. After an additional 24 h, whole cell lysates were prepared, and luciferase activity analyzed as described in A. The lower panel shows a representative immunoblot of CKI⑀. The relative amount of endogenous CKI⑀ was analyzed using NIH Image software and the quantitation shown beneath. CKI␦) was transiently transfected into HEK 293 cells. A dsRNA with the same sequence but in reverse order (denoted Inv) was similarly transfected in control experiments. The effect of siRNA on protein expression was assessed by immunoblot of cell extracts (Fig. 6D). Under these conditions, the dsRNA targeting CKI⑀ and CKI␦ reduced their abundance by almost 50%. As Fig. 6D illustrates, treatment of cells with the CKI⑀/␦ dsRNA led to a 40% decrease in luciferase expression driven by the mPer1 promoter compared with mock-treated and control-treated cells. Hence, both dominant-negative CKI⑀ and siRNA-mediated inhibition of CKI activity led to similar decreases in BMAL1-dependent transcriptional activity. DISCUSSION CKI⑀ is firmly established as a critical component of the circadian clock. Studies in flies and mammals suggest that the stability and localization of the PERIOD proteins are regulated by CKI⑀-dependent phosphorylation. In this study, we present evidence that CKI⑀ acts on additional circadian substrates, including a PER-dependent phosphorylation of CRY and direct phosphorylation of BMAL1. Inhibition of CKI activity by either expression of dominant-negative CKI⑀ or use of small interfering RNAs reduced expression from the mPer1 but not the cdc2 nor actin promoters. Although the data cannot exclude the possibility that additional CKI substrates may influence transcription from the mPer1 promoter, they are consistent with a model in which direct phosphorylation of BMAL1 by CKI⑀ positively regulates the activity of BMAL1 in transcription.
The PER protein acts as a scaffold, bringing CKI⑀ and CRY proteins into close proximity. Both CRY proteins interact with the carboxyl termini of mPER1 and mPER2 in a region of high sequence homology between the two proteins and distant from the previously defined PAS domain and CKI⑀ binding site. Identification of a similar CRY binding site on PER has been reported recently by others (30). Notably, the CRY binding domain has no homologous region in insect PER, consistent with a distinct role for cryptochromes as light-sensing proteins in non-vertebrates. The presence of distinct CRY and CKI⑀ binding sites on PER allows formation of a ternary complex between CKI⑀, CRY, and PER proteins. Formation of this complex appears to have at least two consequences. First, CRY can overcome the CKI⑀-mediated PER1 cytoplasmic localization. This is consistent with immunolocalization studies showing nuclear CKI⑀ in neurons within the suprachiasmatic nucleus (25,31) and positions CKI⑀ to phosphorylate nuclear circadian regulators. This conclusion is supported by the recent report of Lee et al. (32), who found CKI⑀ and CKI␦ in a complex with both PER and CRY in rat and hamster liver and additionally found circadian regulation of CKI nuclear localization. Second, CRY present in this complex can be efficiently phosphorylated by CKI⑀ in vitro. The functional consequence of CKI⑀-dependent CRY phosphorylation is the subject of ongoing investigation, but it does not appear to modify the inhibitory effect of CRY upon circadian promoters (data not shown).
We also examined whether or not BMAL1 was a substrate for CKI⑀. Given that CRY requires the formation of a CRY⅐PER⅐CKI⑀ multimeric complex for its phosphorylation by CKI⑀, it was unexpected that CKI⑀ can in fact directly phosphorylate BMAL1. This phosphorylation appears to be physiologically relevant, since overexpression of dominant-negative CKI⑀ led to both a decrease in BMAL1 phosphorylation and a decrease in BMAL1-dependent transcription from E-box-containing promoters. Similarly, a dsRNA-mediated decrease in CKI⑀ protein in intact cells led to a decrease in BMAL1 transcriptional activation function. That the CKI⑀ with the R178C mutation found in the tau hamster is likely to exhibit a dominant-negative phenotype (i.e. bind to substrates and block ac-cess of active kinases) (1,9) suggests that decreased phosphorylation of BMAL1 may contribute to the tau short period phenotype. An alternative hypothesis not excluded here is that in the cell CKI⑀ may regulate a pathway that then regulates BMAL1 phosphorylation and activity. Although in these studies we used CKI⑀, the closely related CKI␦ that can similarly bind to and phosphorylate PER proteins is likely to similarly phosphorylate BMAL1 (13,25). Recent work from two groups suggests both BMAL1 and CLK are phosphorylated in intact animals. Both proteins have phosphorylation-dependent electrophoretic mobility shifts that change through the circadian day (32). Finally, we note that Sanada and co-workers (33) recently show that mitogen-activated protein kinase phosphorylates BMAL1 on several sites and modestly inhibits the activity of BMAL1 in transcription assays. Mitogen-activated protein kinase and CKI⑀ are likely to phosphorylate distinct sites on BMAL1. BMAL1 may therefore be the target of several distinct signaling pathways, with divergent effects on its activity.
Immunolocalization studies of CKI⑀ have produced conflicted results, dependent on cell type and co-expression of additional circadian proteins. The finding of a multimeric CKI⑀⅐PER⅐CRY complex that can relocalize CKI⑀ from cytoplasm to nucleus implies that some of the variability in CKI⑀ localization seen in cultured cells is due to co-expression of circadian regulators. For example, PER and CRY levels may be higher in COS7 than in HEK 293 cells and, hence, target CKI⑀ to the nucleus more readily in those cells. These results also suggest that one function of the PER⅐CRY complex in vivo may be to relocalize CKI⑀ to the nucleus during periods of CRY abundance, facilitating phosphorylation of BMAL1 and perhaps additional nuclear circadian regulators. What remains to be resolved is how the stimulatory effect of CKI⑀ phosphorylation and the inhibitory effect of CRY on BMAL1 activity are coordinated to produce stable oscillations in expression from circadian promoters.