Correlated evolution between CK1δ Protein and the Serine-rich Motif Contributes to Regulating the Mammalian Circadian Clock*

Understanding the mechanism underlying the physiological divergence of species is a long-standing issue in evolutionary biology. The circadian clock is a highly conserved system existing in almost all organisms that regulates a wide range of physiological and behavioral events to adapt to the day-night cycle. Here, the interactions between hCK1ϵ/δ/DBT (Drosophila ortholog of CK1δ/ϵ) and serine-rich (SR) motifs from hPER2 (ortholog of Drosophila per) were reconstructed in a Drosophila circadian system. The results indicated that in Drosophila, the SR mutant form hPER2S662G does not recapitulate the mouse or human mutant phenotype. However, introducing hCK1δ (but not DBT) shortened the circadian period and restored the SR motif function. We found that hCK1δ is catalytically more efficient than DBT in phosphorylating the SR motif, which demonstrates that the evolution of CK1δ activity is required for SR motif modulation. Moreover, an abundance of phosphorylatable SR motifs and the striking emergence of putative SR motifs in vertebrate proteins were observed, which provides further evidence that the correlated evolution between kinase activity and its substrates set the stage for functional diversity in vertebrates. It is possible that such correlated evolution may serve as a biomarker associated with the adaptive benefits of diverse organisms. These results also provide a concrete example of how functional synthesis can be achieved through introducing evolutionary partners in vivo.

The assumption of conserved function between orthologs has been supported even between relatively distant species. However, orthologs are increasingly found to diverge in sequence, in expression, and even in knock-out phenotypes (1). As a kinase phosphorylates tens to hundreds of substrates, if we can catalog the genetic and functional differences of orthologous kinases and pick out relevant changes to adaptive functional regions in their substrates, then we can better understand the impact of orthologs on evolutionary novelty.
In Drosophila, DBT, the single Drosophila ortholog of CK1␦/⑀, binds to and phosphorylates PER, and DBT then regulates the subcellular localization of PER and signals the rapid degradation of PER by the proteasome (2)(3)(4). The mammalian orthologs of DBT are the CK1␦/⑀ proteins, which mediate the phosphorylation and degradation of mPER2 and facilitate its entry into the nucleus (5). Thus, the CK1␦/⑀/DBT/PER relationship is likely conserved between Drosophila and mammals. Interestingly, a mutation in CK1␦ or its substrate PER2 causes familial advanced sleep phase syndrome (FASPS) 3 (6,7). Moreover, the introduction of hPER2 S662G (the first serine in the SR motif is mutated) into mice faithfully recapitulates the human phenotype, whereas hPER2 S662D (which mimics the phosphorylated status) lengthened the circadian period (7). Correspondingly, a mutation in CK1␦ (hCK1␦ T44A ) also causes FASPS and hCK1␦ T44A expression in mice, thus shortening the circadian period (6). These studies suggest that PER2 SR phosphorylation is conserved by CK1␦ in the mammalian circadian clock. In Drosophila, however, overexpressing hCK1␦ T44A lengthened the circadian period without mimicking mouse or human phenotypes (6), suggesting that CK1␦ and/or its substrate PER2 may have diverged.
CK1␦ and CK1⑀ are close paralog of the CK1 family, which is found in most vertebrates (8). One consensus phosphorylation target for CK1 is (pS/pT)XX(S/T) (pS/pT, phospho-serine/ threonine; X, any amino acid) (9,10). Interestingly, many sequential phosphorylation examples in SR motifs by CK1␦/⑀ have been described to play important roles in various biological processes. Phosphorylation of the SR motif in the Yes-associated protein (YAP) by CK1␦/⑀ was reported to promote YAP degradation and therefore suppress its oncogenic function (11). A homologous mutation in the hPER1 SR motif (hPER1 S714G ) caused an advanced feeding rhythm in mice and disrupted their * This work was supported by grants from the National Science Foundation of  metabolic oscillation (12). These well recorded examples suggest critical functions for the SR motifs in the mammalian physiology. However, the apparent complexity of CK1␦/⑀ and SR motifs in mammals has hampered efforts to catalog the genetic differences and link them to specific phenotypic consequences.
Here, we improved the available strategies for assessing circadian rhythms by reconstructing a kinase-substrate system in Drosophila and using a reliable circadian period as a readout together with a biochemistry assay and bioinformatics analysis to determine their correlation. We demonstrated that despite the highly conserved enzyme domains and sequences between Drosophila and vertebrates, CK1␦ has acquired novel regulatory features that may interact with a newly emerging SR motif in their vertebrate substrates. We suggest that divergences between orthologs effectively facilitate the evolution of new functions in vertebrates. Identifying the underlying intrinsic mechanisms that drive the differences among different species can provide valuable information for determining when and how new functions occur in the species and therefore realize functional synthesis in vivo.

Results
Incompatibility of the hPER2 SR Motif in the Drosophila Circadian System-The maximum likelihood indicates that the difference in CK1␦ in the circadian period of mammals and Drosophila may be related to different substrates and/or different kinase-substrate interactions. To clarify the functional differences, we compared the protein sequences of human and mouse PER1 and PER2 with those of Drosophila PER (dPER). A sequence alignment analysis shows that dPER does not contain the hPER2 SR motif conserved sequence ( Fig. 1A and supplemental Fig. S1). Because the phosphorylation status of PER2 SR motif is correlated with the period length in mice, we can use point mutation-mediated period changes to quantitatively analyze whether the SR motif is compatible with the Drosophila system (e.g. PER2 S662G abolishes sequential phosphorylation for a short period and PER2 S662D mimics phosphorylation for a long period). In mice, however, the expression of the period paralogous hPER1 S714G (12) or hPER3 S630G , 4 the first serinedirected mutation in the SR motif, failed to change the locomotor period as observed in the hPER2 S662G mice, suggesting that PER2 is an optimal candidate for quantitative analysis of locomotor period changes. Therefore, we introduced hPER2 S662 (wild type), hPER2 S662G , and hPER2 S662D into Drosophila. Driven by timeless-GAL4 (tim-GAL4), the overexpression of hPER2 S662 , hPER2 S662G , and hPER2 S662D all lengthened the Drosophila circadian period (Fig. 1B, upper row), indicating that hPER2 can regulate the circadian period in Drosophila. However, SR motif-dependent circadian period changes were not apparent in Drosophila. The period lengths from three genotypes of hPER2 transgenic flies were positively correlated with hPER2 expression but not with mutant types (Fig. 1B, bottom  row). These results indicated that the SR motif of hPER2 is incompatible with the Drosophila circadian system. To test whether the incompatibility of the SR motif with the Drosophila circadian system occurs because the SR motif cannot be modi-fied or because the Drosophila circadian system cannot respond to the phosphorylated SR motifs, we examined the hPER2 proteins in transgenic flies. As compared with the SR motif-dependent sequential phosphorylation of PER2 proteins in mice (Fig. 1C) (7), hPER2 S662 , hPER2 S662G , and hPER2 S662D showed a similar phosphorylation pattern in Drosophila (Fig.  1D), which may explain why the period lengths are not associated with their genotypes (Fig. 1B). These data suggest that the SR motif is not recognized by the kinases in the Drosophila background.
Function of NEMO/NLK in Setting the Clock Speed Is Not Conserved-The recent discovery of phosphorylation by the NEMO/NLK kinase at the "per-short" domain ( Fig. 1A) on dPER stimulates a hierarchical phosphorylation by DBT to set the circadian period in the Drosophila (13,14), strikingly similar to the PER2 SR motif in mammals (a potential SR motif on dPER). To estimate the conservation of this kinase and its potential role as PER1 and PER2 kinase within a more functional context in mice, we generated Nlk knock-out mice with targeted embryonic stem cells obtained from the International Mouse Mutagenesis Program ( Fig. 2A). Because Nlk-deficient mice died within 2 days after birth, we tried to examine the function of NLK in the central clock by examining the circadian phenotype of the Nestin-cre;Nlk f/f mice, in which Nlk is knocked out in central and peripheral neuron cells. However, as with Nlk conventional knock-out mice, most Nestin-cre;Nlk f/f mice died after birth. We only generated one Nestin-cre;Nlk f/f mouse that survived. Tested by wheel running, the circadian period of this mouse is 23.29 h, as compared with the 23.44 h of the control mice (Fig. 2B). Given the limited number of Nestincre;Nlk f/f mice, we cannot come to a confident conclusion of the function of NLK in the central clock. Because Nlk expressed at a high level in the liver (data not shown), Alb-Cre (Cre recombinase transcription under the control of the albumin promoter) mice were crossed with Nlk f/f mice to generate liver-specific Nlk knock-out mice (Alb;Nlk f/f ) ( Fig. 2C) (15,16). To determine the role of NLK in clock function, we crossed the Alb;Nlk f/Ϫ mice to the Nlk f/f ;mPER2 Luc knock-in reporter mice (17) and monitored Per2-luciferase oscillation in liver tissues. The circadian period of PER2::LUC expression from the liver tissues was comparable in the WT and Alb;Nlk f/f mice (Fig. 2D). Furthermore, we found that the discernible expression changes of the core clock genes such as Per2 and Cry1, and time-dependent changes in phosphorylated state of PER1 and PER2 relative to the WT and Alb;Nlk f/f mice, did not occur (Fig. 2

, E and F).
These data indicate that the phosphorylation function of NEMO/NLK on the SR-like or SR motif was not conserved in Drosophila and mammals. hCK1␦, hPER2, and NEMO/NLK cannot recapitulate the function in Drosophila or mammals, which suggests that divergences may have occurred in both substrates and kinases.
CK1␦ Restores the Function of the SR Motif in the Drosophila Circadian System-The mechanisms by which new functions evolved can be identified by introducing new components into ancestral backgrounds and characterizing the effects of their interactions (18,19). We attempted to restore the SR motif function by reconstructing hPER2 S662 , hPER2 S662G , and hPER2 S662D expression in hCK1⑀, hCK1␦, or DBT backgrounds 4 Z. Liu and Y. Xu, unpublished data.

CK1␦ Expands Function of Substrates with Serine-rich Motifs
in Drosophila. First, we introduced DBT, hCK1⑀, or hCK1␦ into Drosophila and fused a V5 tag to their C terminus to compare the expression levels among different lines and different proteins. Consistent with previous studies (6), the transgenic lines expressing hCK1␦, hCK1⑀, or DBT driven by tim-GAL4 all showed a significantly prolonged circadian period ( Table 1), indicating that CK1⑀ and CK1␦ are able to regulate components for circadian function in a Drosophila genetic background. Although the expression levels of DBT, hCK1⑀, and hCK1␦ in transgenic flies are comparable among different lines (Fig. 3A), the overexpression of hCK1⑀ and hCK1␦ produced a longer circadian period than DBT ( Table 1), suggesting that hCK1⑀ and hCK1␦ may display different activity and/or interact with different substrates than DBT in Drosophila. Interestingly, when hCK1␦ was coexpressed with hPER2 S662 , hPER2 S662G , or hPER2 S662D in Drosophila under tim-GAL4 driver, hPER2 S662G ::CK1␦ double transgenic flies showed a significantly shorter period relative to hPER2 S662 ::CK1␦ and hPER2 S662D ::CK1␦ flies (Fig. 3B), thus recapitulating the mouse function (7). However, the expression of hPER2 S662 , hPER2 S662D , or hPER2 S662G from CK1⑀-or DBT-overexpressed background produced weak or no effects on the circadian period (Fig. 3B). These phenotypes suggest that hCK1␦ preferably (or at least partially) interacts with the SR motif to generate Significance was calculated using Student's t test. **, p Ͻ 0.01. Lower panel, Western blotting analysis of hPER2 protein from three lines of tim-GAL4:hPER2 S662 ; tim-GAL4:hPER2 S662G ; and tim-GAL4:hPER2 S662D transgenic flies. The proteins were extracted from 100 fly heads at ZT2 and ZT14. The values are representative of three repeats. Because the size of the PER2 protein is more than 130 kDa and the fly head samples are small, we used the same stained membrane for the loading control instead of actin staining. The relative intensities are labeled in the top lane and thereafter. C, Western blotting analysis of the hPER2 protein from hPER2 S662 , hPER2 S662G , and hPER2 S662D transgenic mouse liver extracts at different circadian time points as described in a previous study (7). CT, Circadian time. D, Western blotting analysis of the hPER2 protein in tim-GAL4:hPER2 S662 ; tim-GAL4:hPER2 S662G ; tim-GAL4:hPER2 S662D transgenic fly heads at ZT0, ZT6, ZT12, and ZT18. The values are representative of the WB of three independent repeats.

CK1␦ Expands Function of Substrates with Serine-rich Motifs
a behavioral response in the Drosophila circadian clock. In other words, hCK1␦ (rather than CK1⑀ or DBT) is the preferred kinase for the PER2 SR motif. Different Kinase Activity of DBT and hCK1␦ on the SR Motif-To provide more direct evidence to support this conclusion, we examined hPER2 protein levels from different kinase backgrounds and found that hCK1␦ has differential effects on PER2 S662 , PER2 S662G , and PER2 S662D protein levels, leading to varied and distinct protein oscillation profiles (Fig.  4A). In contrast, DBT has a uniform destabilizing effect on these very different PER2 mutants (Fig. 4A). The comparable interaction of DBT and hCK1␦ with hPER2 ( Fig. 4B) suggests that DBT phosphorylates other sites in PER2. Although the mechanisms by which the hPER2 S662G /hPER2 S662 /hPER2 S662D proteins were affected and how PER2 protein change was linked to a period event are not well understood, we speculate that the kinase activity of hCK1␦ increased during evolution. To test this hypothesis, we generated transgenic flies expressing hCK1␦ T44A , which reduced the kinase activity on PER2 (6). Similar to DBT and CK1⑀, hCK1␦ T44A also lengthened the Drosophila circadian period like DBT and CK1⑀ (Table 1), and the expression levels among these kinases are comparable (Fig. 3A). Interestingly, hCK1␦ T44A has an impaired ability to shorten period length in hPER2 S662G transgenic flies (Fig. 3B), suggesting that the kinase activity of hCK1␦ is critical for the SR motif.
To further compare the kinase activity among DBT, CK1⑀, and CK1␦ on the SR motif, we immunoprecipitated DBT, CK1⑀, and CK1␦ from transgenic Drosophila heads with a V5 antibody because DBT purified from Escherichia coli is enzymatically inactive (21)(22)(23)(24). Replacing the first serine with an aspartate residue in the PER2 SR motif could facilitate the sequential phosphorylation (7,25). We thus substituted the first serine with an aspartate residue in the SR motif (DVASLTSQCSYSS) (termed SD) to test the phosphorylation reactions in vitro. GST was linked to SD (GST-SD: GST-DVASLTSQCSYSS), and this peptide was expressed in E. coli by induction with isopropyl ␤-D-1-thiogalactopyranoside and purified with a glutathione-Sepharose affinity resin. The in vitro kinase assay showed that CK1␦ had the greatest activity on SD (Fig. 4C). Because the level of DBT auto-phosphorylation was much lower than that of CK1␦, we reasoned that the difference in kinase activity over the SR motif was not caused by auto-phosphorylation because auto-phosphorylation likely inhibits the activity of kinases (26 -28). Finally, we compared    the effect of these kinases on the generalized CK1 peptide (GST-RRKDLHDDEEDEAMSITA) (29), where serine phosphorylation is primed by a series of aspartic or glutamic acids. Consistent with the effect on the SR motif, hCK1␦ phosphorylated the CK1 peptide much more efficiently than DBT (Fig.  4D). Altogether, these data suggest that hCK1␦ strongly enhanced phosphorylation at the SR motif and did so more efficiently than DBT and CK1⑀. Correlated Evolution of CK1␦ and SR Motifs-If CK1␦ had evolved to obtain the ability to catalyze the SR motif in the PER2 protein, it should be able to perform similar reactions on the SR motif in other substrates or utilize SR motifs in other protein to expand the repertoire of its enzymatic activities and result in functional adaptation (30). We found that an increased number of studies have suggested that CK1␦/⑀ phosphorylate SR motifs in many proteins including PER2, PER1, APC, NFATC, and PGC1a and YAP1 (7,11,12,(31)(32)(33)(34). These SR motifs are well conserved in vertebrates (Fig. 5A), which is consistent with the highly conserved vertebrate CK1␦/⑀ protein sequence (supplemental Fig. S2 and Fig. 5B). In addition, we found that SR motifs are divergent among 48% (13/27) of PER3 proteins lacking one or two serine residues at SR motifs (Fig. 5C). However, the SR motifs in PER1 and PER2 are well conserved in different vertebrates (Fig. 1A). Interestingly, mouse PER3 does not show interaction with CK1⑀/␦, which is inconsistent with PER1 and PER2 (35). Both observations suggest that the functional evolution of CK1␦ kinase activity is likely correlated with the evolution of the SR motif and/or that this activity is likely to preserve SR motifs in CK1␦ substrates.
Finally, a statistical analysis was performed of the conserved SR motifs in various species using protein sequence data retrieved from the OrthoMcl database. We searched for proteins in each species that contain a conserved SXXSXXSXX-SXXS (5S) motif with their human orthologs and compared them with proteins containing other conserved similar pattern motifs such as AXXAXXAXXAXXA (Fig. 5D, A is an any amino acid). Interestingly, the number of conserved 5S motif-containing proteins was higher relative to other conserved 5A motifcontaining proteins in vertebrates (Fig. 5D, right side, and supplemental Table S). Due to the evolutionary distance with humans, the number of conserved 5S and 5A motif-containing proteins in non-vertebrate species was much lower than that in vertebrates, and the number of conserved 5S motif-containing proteins was average (Fig. 5D, left side). An almost identical trend was also observed in proteins containing the 4S motif (Fig. 5E), although additional proteins contained the 4S motif rather than the 5S motif (compare Fig. 5E with Fig. 5D). As described above, the emergence of a conserved SR motif in vertebrates suggests that the critical function of the SR motif might be evolutionarily correlated with the kinase activity of CK1␦/⑀.

Discussion
The different phenotypes induced by CK1␦ mutations in mouse and Drosophila circadian system challenged the notion that the CK1␦-PER interaction is functionally conserved (6). This conclusion is supported by Sekine et al. (36), in which they introduced hCK1⑀ into Drosophila and compared the function of DBT and hCK1⑀ on fly circadian rhythm and PER phosphor-ylation. They found that hCK1⑀ cannot replace the function of DBT in the fly, suggesting the evolution of kinase substrate interaction (36).
We have tried to directly examine the activity of DBT and CK1␦/⑀ on hPER2 S662 /hPER2 S662G /hPER2 S662D in vitro. However, we failed to find a proper in vitro system to perform the test. When expressed in mammalian cell culture, Drosophila DBT seems enzymatically inactive. It can barely phosphorylate its own substrate PER in human embryonic kidney 293 cells. When overexpressed in the Drosophila S2 cell, hPER2 proteins were easily degraded, especially when overexpressed together with kinases (data not shown), which makes the detection of phospho-hPER2s very difficult.
By reconstructing the interactions of DBT/hCK1␦/⑀ and SR motif mutated hPER2s in the Drosophila circadian system, we can speculate on the function of DBT and CK1␦/⑀ on hPER2 SR motif by comparing the phenotype of different double transgenic flies. We discovered that hCK1␦ is required for the function of the SR motif, which is not observed in Drosophila PER. One limitation of our in vivo experimental design is that we examined the activity of CK1␦/⑀ on hPER2 in the presence of endogenous DBT. Although the overexpressed CK1␦/⑀ is much higher than the endogenous DBT, it is still possible that the endogenous DBT complicated the activity of CK1␦/⑀ on hPER2s. Given the fact that CK1⑀ cannot rescue the circadian arrhythmicity of DBT mutant fly (36), we did not repeat the experiment in DBT-deficient background. By in vitro kinase assay using kinases immunoprecipitated from transgenic flies, we directly demonstrated varied kinase activity of DBT and CK1␦ on the SR motif. Although more work is needed to address the mechanistic details of their coevolution, we have shown that DBT is less effective than hCK1␦ in phosphorylating the SR motif, which suggests that the co-variation between kinase and substrates may be responsible for the functional diversification of the interaction and therefore may play a role in the phenotypic evolution of organisms. This study also provides an example of the correlated evolution of orthologous kinase and their substrates, and shows their contribution to physiological evolution. These results are consistent those of other studies in which protein kinases and phosphorylation have been reported to provide important sources of phenotypic diversity (37,38).
Despite a pivotal role in several biological processes, the specific function of SR motifs is poorly understood. The phosphorylation of SR motifs may play different regulatory roles in various proteins. Vanselow et al. (39) reported that the hPER2 S662G mutation destabilizes the PER2 protein. Phosphorylation of YAP and PGC1␣ SR motifs was reported to facilitate the degradation of proteins (11,31). For NFATC, phosphorylation of the SR motif increases the localization of NFATC to the nucleus (32). APC is phosphorylated by the mutual priming of CK1⑀ and GSK-3␤ and then competes with axin for binding to ␤-catenin, which leads to the degradation of the complex (33). These studies indicate that the conserved SR identified in each species' proteome would be versatile. Generally, the multiple phosphorylation targets in SR motifs enable the hierarchical regulation of the protein by different kinases and the formation

CK1␦ Expands Function of Substrates with Serine-rich Motifs
of different complexes. Moreover, various conformations would be elicited by hierarchical phosphorylation that may allow the protein to serve multiple functions (40). Here, we searched a collection of proteins containing SR motifs, most of which remain unidentified to date. Thus, future studies are needed to obtain additional data regarding the specific substrates and functions of the motifs in each protein and further understand how new functions are developed in vertebrates. Our observations also suggest that with genetic synthesis, phenotypes caused by single or additive co-evolution changes can likely be recapitulated in other species.
Besides the SR motif in PER2 and per-short domain in dPER (Fig. 1A), other CK1 phosphorylation sites have been identified in PER proteins. Ser-47 of dPER (supplemental Fig. S1, marked in blue) was reported to be phosphorylated by DBT, providing a binding site for the F-box protein, SLIMB (ortholog of ␤TrCP) and therefore mediating the degradation of dPER (41). Sequence alignment showed that Ser-47 is well conserved among PER proteins from different species (supplemental Fig.  S1). In mammals, mutations in Ser-477 and Gly-479 in mPER2 abolished the CK1⑀-dependent ␤TrCP binding, indicating that Ser-477 (supplemental Fig. S1, marked in yellow) might be the phosphorylation target of CK1⑀, which recruits ␤TrCP and mediates the degradation of mPER2 (42). Using mass spectrometric analysis, multiple phosphorylation sites have been identified in dPER and mPER2 (24,39,41,43). Among these phospho-peptides, another SR motif of mPER2 (Ser-525, Ser-528, Ser-531) (supplemental Fig. S1, marked in brown) was identified by two separate groups (39,43). This SR motif is conserved between mouse and human but not present in other species, indicating that it is a newly emerged CK1␦ target (supplemental Fig. S1). One SR motif (perSD domain) was found to be phosphorylated by DBT in dPER (Ser-604, Ser-607, Thr-610, Ser-613, Ser-629) (supplemental Fig. S1, marked in red) (24), which was shown to be important for the transcriptional repression activity of dPER (24). Despite the similar pattern with the PER2 SR motif, they are not evolutionarily conserved domains (supplemental Fig. S1). Kivimae et al. (24) suggest that the function of perSD domain is tightly regulated by the phosphorylation of Ser-589, which is 15 residues N-terminal to perSD domain, suggesting different regulation of perSD domain with hPER2SR motif. Thus, it is possible that the weak activity of DBT on the SR motif might be sufficient to regulate the activity of perSD domain.

Generation of Transgenic Flies and Behavioral Analyses-
The coding sequences for hPER2 S662 , hPER2 S662G , and hPER2 S662D were inserted into the pUAST vector. The V5 tag coding sequence was added to the C terminus of the DBT, hCK1␦, and CK1␦T44A coding sequences. DBT-V5/hCK1␦-V5/CK1␦T44A-V5-pUAST vectors were constructed, and transgenic flies were generated by injecting each vector into w 1118 Drosophila embryos using a standard protocol. Three independent lines for each genotype were selected and then crossed to tim-GAL4 (45) and w 1118 (control) flies.
Drosophila were reared on standard cornmeal/agar medium and maintained under 12:12 light-dark (LD) cycles at 25°C. The circadian periods of the flies were analyzed using the ClockLab (Actimetrics) software as described previously (Pfeiffenberger et al. (46)). For each Drosophila line, the locomotor activity tests were repeated in at least two trials. The results were consistent between trials.
Western Blotting Analysis-One hundred fly heads of the specified genotype were collected at the indicated time points after 3 days of rearing under LD conditions. Western blots were conducted as described previously (20,44). The hPER2 proteins were probed with homemade hPER2 antibody (7) with 1:1000 dilution.
In Vitro Kinase Assay-One hundred heads from w 1118 , DBT, and hCK1␦ transgenic flies were collected at ZT20. The tissue lysates were incubated with an anti-V5 monoclonal antibody (Sigma-Aldrich) followed by incubation with 60 l of a 50% slurry of protein G-Sepharose (BioVision). Sepharose was washed three times with lysis buffer and then subjected to a kinase assay in a 20-l reaction mix (30 mM HEPES, pH 7.5, 10 mM MgCl 2 , 0.5 M DTT, 200 mM ATP, 50 mg/ml BSA, and 100 M [␥-32 P]ATP (0.5 Ci) (PerkinElmer)) with 1 g of substrate (GST-SD or GST-CK1 peptide) for 30 min at 30°C. The reaction was terminated by adding 20 l of 4ϫ sample buffer. The products were resolved on 12% SDS-PAGE gels, and ␥-32 P incorporation was detected in the dried gels by exposure of X-ray films (Kodak) at Ϫ80°C for 3 days.
Bioinformatic Analysis of SR Motifs-The protein sequences for different species were retrieved from the OrthoMCL DB database and then screened to identify the 5A or 4A motifs. The proteins containing 5A or 4A motifs were aligned with their human orthologs. The number of proteins containing conserved 5A or 4A motifs with human orthologs was counted for each species.  Drosophila melanogaster). B, phylogenetic tree for CK1⑀ and CK1␦ retrieved from ensemble database. CK1⑀ and CK1␦ exist in most vertebrates. C, phylogenetic tree of PER3 derived from the Ensembl database. The PER3 proteins that do not contain the integrated SR motif are shown in red. Variations of SR motifs in different species are illustrated. D and E, number of proteins containing conserved 5A (D) and 4A (E) motifs with Homo sapiens orthologs. The species included in this study from the distal to the proximal x axis are as follows: Pan troglodytes, Equus caballus, Canis lupus familiaris, Mus musculus, Rattus norvegicus, Gallus gallus, Takifugu rubripes, Tetraodon nigroviridis, Danio rerio, Acyrthosiphon pisum, Ciona intestinalis, Anopheles gambiae str. PEST, Aedes aegypti, Nematostella vectensis, Drosophila melanogaster, Ixodes scapularis, Culex pipiens, Caenorhabditis elegans, Caenorhabditis briggsae AF16, Chlamydomonas reinhardtii, and Neospora caninum. The species on the right side of the blue line are vertebrates. The numbers on the x axis represent the conservation of each species as compared with humans, which is calculated by multiplying the percentage of proteins that have human orthologs by the orthologous sequence identity. From the 19 control motifs, we selected 7 motifs for the actograms. For the other 12 motifs, the number of 5A motif-containing proteins was less than 100 in most species and the conservation values were extremely low.