Mutations lowering the kinase activity of Drosophila Doubletime (DBT) and vertebrate casein kinase Iε/δ (CKIε/δ) produce long-period, short-period, and arrhythmic circadian rhythms. Since most ckI short-period mutants have been isolated in mammals, while the long-period mutants have been found mostly in Drosophila, lowered kinase activity may have opposite consequences in flies and vertebrates, because of differences between the kinases or their circadian mechanisms. However, the results of this article establish that the Drosophila dbt mutations have similar effects on period (PER) protein phosphorylation by the fly and vertebrate enzymes in vitro and that Drosophila DBT has an inhibitory C-terminal domain and exhibits autophosphorylation, as does vertebrate CKIε/δ. Moreover, expression of either Drosophila DBT or the vertebrate CKIδ kinase carrying the Drosophila dbtS or vertebrate tau mutations in all circadian cells leads to short-period circadian rhythms. By contrast, vertebrate CKIδ carrying the dbtL mutation does not lengthen circadian rhythms, while Drosophila DBTL does. Different effects of the dbtS and tau mutations on the oscillations of PER phosphorylation suggest that the mutations shorten the circadian period differently. The results demonstrate a high degree of evolutionary conservation of fly and vertebrate CKIδ and of the functions affected by their period-shortening mutations.
ORGANISMS as phylogenetically diverse as cyanobacteria and man exhibit circadian rhythms, which produce cycles of behavior, physiology, and biochemistry that are synchronized (or entrained) by the environmental cycles. When organisms are maintained in the laboratory under constant conditions, these rhythms persist with a period length that is ∼24 hr, thus demonstrating the existence of an endogenous biological clock underlying the overt rhythms. The endogenous nature of this circadian clock is thought to allow environmental changes to be anticipated by physiological alterations that would not occur soon enough if they were produced only after the environmental changes (Pittendrigh 1974).
The molecular mechanism for the circadian clock has yielded to genetic analyses in flies, mammals, bread mold, plants, and cyanobacteria (Sehgal 2004). While different gene products are used in different phyla, in eukaryotes a general property has been transcriptional feedback loops, wherein transcriptional regulators increase and decrease over the course of the day to regulate target genes. In Drosophila, the period (PER) protein negatively regulates a bHLH-PAS clock/cycle transcription factor (CLK/CYC). Since the per gene is a target of CLK/CYC, PER represses its own mRNA. When PER protein levels fall, per mRNA is transcribed again, leading to antiphase cycling of per mRNA and nuclear PER protein (Price 2004).
To generate ∼24-hr oscillations, the timing of negative feedback must be regulated post-translationally by the activities of protein kinases and phosphatases (Harms et al. 2004). It was first shown that PER was rhythmically phosphorylated (Edery et al. 1994), and isolation and analysis of the dbt mutants in Drosophila then showed that casein kinase I (CKI) targets PER for phosphorylation and degradation (Kloss et al. 1998; Price et al. 1998; Rothenfluh et al. 2000a; Suri et al. 2000; Muskus et al. 2007). Doubletime (DBT) has also been proposed to regulate the nuclear localization (Bao et al. 2001; Cyran et al. 2005; Nawathean et al. 2007) and the repressor capacity of PER (Nawathean and Rosbash 2004; Kim et al. 2007; Nawathean et al. 2007), as well as the phosphorylation of CLK protein (Kim and Edery 2006; Yu et al. 2006). In addition, two other kinases [casein kinase II, CKII (Lin et al. 2002) and SGG (Martinek et al. 2001)] and two protein phosphatases [PP2A (Sathyanarayanan et al. 2004; Fang et al. 2007) and PP1(Fang et al. 2007)] contribute to rhythmic phosphorylation of clock proteins. Without the activities of these kinases and phosphatases, clock protein expression and clock-controlled processes are not rhythmic. DBT's mammalian orthologs CKIε (Lowrey et al. 2000; Vielhaber et al. 2000; Lee et al. 2001; Eide et al. 2005) and -δ (Lee et al. 2001; Xu et al. 2005), and PP1 (Gallego et al. 2006b), have also been shown to contribute to mammalian circadian rhythms, in part by regulating PER stability and nuclear localization.
The importance of dbt or vertebrate ckI protein kinases for the core time-keeping process is exemplified by mutants that can either shorten or lengthen the circadian period (Price et al. 1998; Lowrey et al. 2000; Rothenfluh et al. 2000a; Suri et al. 2000; Xu et al. 2005; Muskus et al. 2007). Intriguingly, all of the known mutant kinases exhibit lower kinase activity on complex substrates like casein and PER in vitro (Lowrey et al. 2000; Suri et al. 2000; Preuss et al. 2004; Xu et al. 2005), with all of the long-period mutants identified in fly DBT and all but one of the short-period mutants in vertebrate CKIε/δ, raising the possibility that there are differences between the mammalian and Drosophila circadian kinases and mechanisms (Xu et al. 2005; Sekine et al. 2008).
Several other findings are relevant to the evolutionary relationship between the roles of CKI in the Drosophila and vertebrate clocks. It is noteworthy that Drosophila DBT and vertebrate CKIε/δ are highly conserved in their N-terminal regions but diverge significantly in the C-terminal domains (Kloss et al. 1998). Moreover, their proteins differ biochemically, because the vertebrate CKIε/δ's are enzymatically active when purified from Escherichia coli (Graves and Roach 1995; Preuss et al. 2004) while Drosophila DBT is not (Suri et al. 2000; Kloss et al. 2001; Preuss et al. 2004). While a knock-out mutation of ckIε in mice lengthens period, the effect is relatively minor [∼20 min (Meng et al. 2008)]. Recently, it was shown that expression of mammalian CKIε in flies antagonized their circadian rhythms, suggesting that mammalian CKIε is not entirely evolutionarily conserved with fly DBT in its circadian function (Sekine et al. 2008). Finally, while both DBT and CKIε/δ target PER, additional circadian targets may differ, with mammalian CKIε/δ targeting BMALI and CRY (Eide et al. 2002) and DBT targeting CLK (Kim and Edery 2006; Yu et al. 2006).
In this article, we have examined the evolutionary conservation of DBT and CKIδ function by further biochemical characterization of DBT and by expression in Drosophila of various mutations in both the fly DBT and the vertebrate CKIδ. Surprisingly, the results suggest a high degree of evolutionary conservation for CKIδ and DBT and for the circadian functions affected by the short-period mutations of DBT and CKIε/δ.
MATERIALS AND METHODS
Expression of DBT-MYC and PER-HA in cell culture lines:
Vectors expressing untagged DBT and MYC-epitope-tagged DBTWT, DBTS, DBTL (Preuss et al. 2004), or the catalytically inactive DBTK/R protein (Muskus et al. 2007) under the control of the metallothionein promoter have been previously described. DBTWT, DBTS, DBTL, the catalytically inactive DBTK/R protein, or truncated forms of DBT were used to produce stably transfected S2 cell lines as previously described (Muskus et al. 2007). To produce the truncated forms of DBT, PCR reactions were employed to amplify different extents of the DBT C terminus. Each reverse primer encoded an AgeI site and the MYC epitope tag and therefore contained the following sequence at its 5′ end: GCGCACCGGTGAGGTCTTCCTCGCTGATCAGCTTCTGCTC. This was followed by the noncoding strand for DBT starting at amino acid 296 (GCCAAACTTAAGCAGGTTCCAGTCAAACAC), 332 (GGCGGCTGCCGCTGCTGCCACCGCCGCTGC), or 387 (ATCGTCCATGTTGAGTCCGTTGCCGCCGAT). Each of these primers was used with a forward primer that annealed upstream of a unique NsiI site in the dbt cDNA. Digestion of the PCR product with NsiI and AgeI yielded a fragment that was ligated in place of the NsiI/AgeI fragment from pMT-DBT, yielding a plasmid that expressed an epitope-tagged truncation mutant. DBT-MYC-expressing constructs were transiently cotransfected into S2 cells with pAC-PER-HA, expression of DBT-MYC was induced with CuSO4, and PER and DBT levels were detected by immunoblot analysis with anti-HA and anti-MYC antibodies as previously described (Muskus et al. 2007). For the experiments examining autophosphorylation of DBT (Figure 3), stably transfected lines expressing DBTWT-MYC or DBTK/R-MYC from the pMT vector were used. These were induced with CuSO4 41 hr before harvest, and okadaic acid or vehicle (DMSO) was added 17 hr before harvest. The extracts were than analyzed by immunoblot analysis with an anti-MYC antibody.
In vitro kinase assays with DBTS, DBTL, or C-terminally truncated forms of DBT:
DBT-MYC expression was induced in stably transfected cell lines with CuSO4, the cell lysates were immunoprecipitated with anti-MYC antibody, the immunoprecipitates were incubated with PER expressed and purified from bacteria, and radiolabeled substrate was detected and quantified by phosphor-imager analysis of SDS–PAGE (Muskus et al. 2007). Aliquots of each reaction were also analyzed by immunoblot analysis with an antibody detecting the C terminus of DBT (Figure 1) or the MYC epitope (Figure 2), and the signal for 32P incorporation was normalized to the signal for DBT, as previously described (Muskus et al. 2007). The individual kinase activities measured in each of the experiments tabulated in Figure 1 were subjected to a one-way ANOVA, with each genotype as a categorical predictor (three altogether). Any comparisons between the kinase activities of two DBT genotypes were accomplished by applying a post hoc Tukey's honestly significant differences (HSD) test to this ANOVA. Significant differences were defined as those with P < 0.05. Group values are presented as means ± SEM. Statistics were performed using Statistica (StatSoft, Tulsa, OK).
Site-directed mutagenesis to produce mutant CKI and DBT:
Generation of the dbtS, dbtL, ckIS, ckIL (Preuss et al. 2004), and dbtK/R (Muskus et al. 2007) mutations has previously been described. The ckITau and ckID/N mutants were constructed by annealing two PCR fragments that each had the mutation at an end and then amplifying a larger fusion fragment with the mutation in the middle, as previously described (Preuss et al. 2004). The mutagenic primers were as follows: for ckIδTau, F5′-GGAACAGCTTGCTATGCATCTATTAACAAC and R5′-AGATGCATAGCAAGCTGTTCCGGTTAGGTT; and for ckIδD/N, R5′-CAGAAAATTGTTAGGCTTCACATCCCGATG and F5′-GTGAAGCCTAACAATTTTCTGATGGGACTT. In each reaction either the mutation-containing R primer was combined with a primer flanking an upstream PinAI site (5′-CACTTTTTTTGCTTCTTAGCGCCGGATGGA) or the mutation-containing F primer was combined with a primer flanking a downstream NsiI site (5′-TGGCTTATCATCAAAACGTAAGGACCGACA). The two fragments produced for each mutation were mixed, denatured, annealed through their complementary ends (including the mutation site), and extended by several rounds of PCR and then final PCR with both flanking primers, thereby amplifying a full-size fragment. Cloning of these fragments into a full-length Xenopus CKIδ cDNA has been previously described (Preuss et al. 2004).
The dbtTau mutation was generated by the Quik-Change procedure (Stratagene, La Jolla, CA), with a pMT-dbtWT vector as template. The primers were DBTtauF, 5′-CTCACGGGCACTGCCTGCTATGCCTCCATC-3′ and DBTtauR, 5′-GATGGAGGCATAGCAGGCAGTGCCCGTGAG-3′.
Generation of transgenic flies expressing CKIδ and DBT-MYC:
The cloning of the ckIδ genes into the GAL4-inducible fly transformation vector pUAST was performed by using the restriction enzymes SalI and XbaI and isolation of a 1.6-kb fragment from pVAX-ckIδ (Preuss et al. 2004), which was inserted into the opened vector plasmid after digestion of the vector with XhoI and XbaI. This procedure placed ckIδ in a 5′–3′ orientation downstream of the UAS promoter sequence in pUAST. The dbt-myc cDNAs were cloned into a pUAST vector modified with a PmeI site, as previously described (Muskus et al. 2007). Embryos from a w1118 line (Bloomington Drosophila Stock Center, stock no. 3605) were injected by Duke Model Systems Genomics, and lines containing different transgene insertion sites were generated by crossing the injected adults to a w; TM3 or a w; CyO balancer stock, selecting transgenic progeny by their w+ phenotype, and establishing lines by at least two backcrosses to the w; TM3 or w; CyO balancer stock. The insertions were mapped to chromosome II or III on the basis of the segregation of the w+ P element relative to the dominant markers on the balancer chromosomes, and the line was crossed to the other balancer stock if its insertion site was not on the homologous chromosome of the balancer stock used initially. Independent insertions were produced by ensuring that all lines tested herein were derived from different injected w1118 embryos.
Rearing of flies and behavioral analysis:
Males from UAS-responder transgenic flies were mated to virgin females carrying a GAL4 driver transgene. Male progeny were collected and carried one copy of the driver and responder transgene. The tim-GAL4 driver transgene is expressed in all circadian cells [P(w[+mC]=GAL4-tim.E)62; Bloomington stock no. 7126 (Kaneko and Hall 2000)]. This driver produced expression of MYC-tagged DBTs in both photoreceptors and brain neurons, including the lateral neurons that secrete PDF and control locomotor behavior (data not shown). Progeny were reared for at least 3 days in a 12-hr:12-hr light:dark cycle. Locomotor activity was then monitored in constant darkness with Trikinetics activity monitors, and period length and rhythmicity were determined by ClockLab analysis software, as previously described (Muskus et al. 2007). The average periods of all the individual lines in Table 1 were subjected to a one-way ANOVA, with each genotype (nine altogether) as a categorical predictor. In addition, the average periods of the responder without driver controls (seven averages of this type in Table 1) composed a 10th categorical predictor in this ANOVA. Any comparisons between two genotypes were accomplished by applying a post hoc Tukey's HSD test. Significant differences were defined as those with P < 0.05. Group values are presented as means ± SEM. Statistics were performed using Statistica (StatSoft).
Immunoblot analysis of DBT, CKI, and PER levels in the heads of transgenic flies:
tim-GAL4>UAS-DBT-MYC or -CKI flies were reared, entrained to light:dark (LD), and collected at various times in an LD cycle as described above. Heads were cut off by razor blade and analyzed by immunoblot analysis as previously described (Muskus et al. 2007). Depending on the experimental setup, immunoblot analysis was performed using a 1:5000 dilution of anti-MYC antibody (Invitrogen, Carlsbad, CA), a 1:2000 dilution of rabbit anti-DBT antibody, a 1:25,000 dilution of rabbit anti-PER antibody, a 1:2000 dilution of mouse monoclonal anti-CKIδ 128A (kindly provided by Anthony DiMaggio of ICOS), or a 1:5000 dilution of mouse monoclonal anti-tubulin (Developmental Studies Hybridoma Bank, University of Iowa), followed by a 1:5000 dilution of the appropriate horseradish peroxidase-labeled secondary antibody and visualization of the secondary antibody using Amersham (Piscataway, NJ) ECL Plus Western blotting detection reagents (cat. no. RPN2132).
DBTS and DBTL phosphorylate PER less actively in vitro than does DBTWT:
We have previously shown that Xenopus CKIδ carrying the Drosophila dbtS or dbtL mutations (CKIS or CKIL) phosphorylates casein and PER less actively than does wild-type CKIδ [CKIWT (Preuss et al. 2004; Constance et al. 2005)], and both the short-period and long-period mutant forms of DBT (DBTS and DBTL) phosphorylate casein less readily than does wild-type DBT [DBTWT (Preuss et al. 2004; Kivimae et al. 2008)]. Here, it is shown that DBTS and DBTL immunoprecipitated from Drosophila S2 cells also phosphorylated recombinant Drosophila PER less actively than wild-type DBT, although a comparison with a catalytically inactive form of DBT [DBTK/R (Muskus et al. 2007)] demonstrates that both mutants retained some kinase activity (Figure 1; two reaction aliquots are shown from two separate reactions for each, and the average relative activity of DBTWT was set to 1 in all assays). An ANOVA indicated a highly significant effect of genotype on kinase activity [F(2, 6) = 14.3, P < 0.01]. Both the DBTS and the DBTL proteins produced significantly less phosphorylation of PER than did DBTWT (P < 0.015 by ANOVA and post hoc Tukey's test), while the differences between DBTS and DBTL were not significant (P > 0.800). Therefore, in the context of either vertebrate CKIδ or Drosophila DBT, the dbtS and dbtL mutations lower general kinase activity toward both PER and casein in vitro.
DBT is inhibited by a C-terminal domain and has autokinase activity:
Kinase activity of vertebrate CKIε/δ is inhibited by its C-terminal domain, as removal of this domain by proteolysis or expression of truncated proteins leads to elevated activity in vitro (Graves and Roach 1995; Cegielska et al. 1998). While the DBT C terminus does not exhibit any significant homology with vertebrate CKIε/δ, this domain differs significantly between vertebrate CKIε and CKIδ, which most likely were produced by a gene duplication event and have evolved from a common progenitor gene (Constance et al. 2005). Despite the relatively low sequence conservation, the C termini of CKIδ and CKIε seem to have similar properties, since both mediate autophosphorylation-dependent inhibition of CKI's kinase activity (Graves and Roach 1995; Cegielska et al. 1998) Unlike the catalytic activity of the kinase, this function may not require much sequence conservation (perhaps just some serines or threonine targets). DBT's C-terminal domain has several serines and threonines that could mediate autophosphorylation-dependent inhibition, and it might serve the same function as the C terminus of CKIε/δ, despite the lack of apparent sequence conservation. Therefore, DBT truncation mutants were assayed to determine whether DBT's C terminus might inhibit DBT's kinase activity, as does the C-terminal domain in vertebrate CKIε/δ. The shortest of these terminated at amino acid 296 and was missing all of the evolutionarily divergent C-terminal domain, which extends from amino acid 296 to the C terminus at amino acid 440 (Constance et al. 2005). All of these were expressed as MYC-tagged proteins in Drosophila S2 cells, immunoprecipitated with an antibody to the MYC epitope, and assayed with PER in vitro. All of the truncated DBTs phosphorylated PER, and the DBT-296 produced increased phosphorylation of PER in comparison with full-length DBTWT (Figure 2A). This qualitative result was reproduced in five experiments; although the low levels of DBT-296 recovery often precluded quantification of its chemifluorescent signal, the amount of PER phosphorylation was greater than or equal to that produced by higher levels of DBT-440.
When DBT is coexpressed with PER in S2 cells, PER is progressively phosphorylated by DBT, and this phosphorylation causes PER to exhibit reduced electrophoretic mobility and to be targeted for degradation by the proteasome (Ko et al. 2002; Muskus et al. 2007). In these experiments, the amount of DBT can be altered by transfection of different amounts of dbt plasmid, and by the addition of different amounts of CuSO4, which induces the metallothionein promoter in the pMT-dbt plasmid. This S2 cell phosphorylation and degradation program recapitulates the circadian phosphorylation and degradation program produced by DBT in vivo, and it has recently been employed to map DBT target sites in PER (Chiu et al. 2008). When expressed together with PER in S2 cells, all of the DBT truncation mutants targeted PER for phosphorylation and degradation (Figure 2B). The activity of the DBT-296 protein was particularly noteworthy, as it accumulated only to low levels but was more effective than DBTWT in targeting PER. Hence, deletion of the C-terminal domain leads to hyperactivity for Drosophila DBT, as it does for vertebrate CKIε/δ.
In vertebrate CKIε/δ, the inhibitory effects of the C terminus are mediated by autophosphorylation of the C terminus (Graves and Roach 1995; Cegielska et al. 1998; Rivers et al. 1998). To address whether DBT might be autophosphorylated, we expressed MYC-tagged DBTWT or DBTK/R in Drosophila S2 cells in the presence of increasing amounts of the phosphatase inhibitor okadaic acid (nanomolar concentrations given) or vehicle alone (DMSO; 0 nm) and analyzed the tagged proteins by immunoblot analysis (Figure 3). Lower mobility forms were detected for DBTWT but not for DBTK/R. Because the novel forms in DBTWT require the kinase activity of DBT and are enhanced by phosphatase inhibition, they are likely to be produced by autophosphorylation of DBT. Taken together with the data in Figure 2, they argue that the DBT C-terminal domain inhibits the kinase activity of DBT and that DBT is autophosphorylated, like vertebrate CKIε/δ.
The dbtS and tau mutations shorten the circadian period in both CKIδ and DBT:
To compare the circadian phenotypes of various mutant vertebrate CKIδ proteins and Drosophila DBT, both were placed under the control of a yeast UAS promoter element, which allows these kinase “responder” genes to be expressed when a GAL4 transcription factor “driver” gene is also present. Expression of the kinases in progeny flies is produced by crossing parents that carry a driver with those with a responder. We did not attempt to rescue a dbt null mutant with this approach because prior work had shown incomplete rescue of the null phenotype even with widespread expression of Drosophila UAS-DBT, and vertebrate UAS-CKIε did not produce any viable adults whose circadian rhythms could be assayed (Sekine et al. 2008). Instead, we overexpressed UAS-DBT and -CKIδ transgenes in a genotype that included dbtWT at the endogenous locus. We had previously established that it is possible to produce fully penetrant dbt mutant phenotypes by overexpression of mutant DBT proteins that outcompetes endogenous DBT for interactions with clock components (Muskus et al. 2007). By restricting the expression with clock-cell-specific drivers, it is possible to avoid lethality and reduce the possibility of general developmental effects or activation of diverse signaling pathways (e.g., wingless and hedgehog pathways) that could indirectly affect rhythms.
Locomotor activity was assayed for flies expressing the wild-type and various mutant forms of CKIδ and DBT with a tim-GAL4 driver, which produces expression in all circadian cells (Kaneko and Hall 2000). As with overexpression of DBTWT, overexpression of CKIδWT examines whether the introduction of a vertebrate casein kinase into fly circadian cells causes lethality or disrupts the circadian system of the fly, and it serves as a baseline against which the effects of period-altering mutations are measured. There was no lethality or disruption of external morphology when CKIδWT was expressed with tim-GAL4 (data not shown). While ANOVA indicated a highly significant effect of genotype on circadian period [F(9, 37) = 73.4, P < 0.0001], the UAS-DBTWT or -CKIWT responders did not produce a statistically significant change in period with the tim-GAL4 driver relative to the average period of the responders without the driver (Table 1, P > 0.05 by ANOVA with post hoc Tukey's test; however, the average period of the tim-GAL4>UAS-DBTWT or -CKIWT lines was slightly longer than the average period of the UAS-DBTWT or -CKIWT responders without the driver). They did produce substantial arrhythmicity. Marginal period lengthening like that shown here has been previously observed for overexpression of both mammalian CKIδ and Drosophila DBT in flies (Xu et al. 2005; Muskus et al. 2007). The period lengthening caused by CKIδ in the previous analysis was somewhat more than shown here, while the arrhythmicity was significantly less, probably because of subtle differences in the transgenes or rearing conditions. Our molecular analyses of these lines show that the oscillations of PER are robustly rhythmic in all of our lines except the tau mutants (Figure 6), so the apparent arrhythmicity of locomotor activity is not likely to arise from a damping of the underlying circadian mechanism but rather from an effect of transgene overexpression on the locomotor output pathway.
Six different insertion lines of UAS-CKIS responder transgenic lines were analyzed and showed a shortened average circadian period ranging from 20 to 22 hr when expressed with the tim-GAL4 driver (Table 1 and Figure 4; average period of lines = 20.8 ± 0.3 hr, with P < 0.0002 for the difference with tim-GAL4>UAS-CKIWT by ANOVA and post hoc Tukey's test). This period shortening was similar to the one observed in flies expressing a copy of the UAS-DBTS transgene with the tim-GAL4 driver (see Table 1, Figure 4, and Muskus et al. 2007; average period of lines = 18.3 ± 0.1 hr, with P < 0.0002 for the difference with tim-GAL4>UAS-DBTWT by ANOVA with post hoc Tukey's test). Transgenic DBTWT-MYC and DBTS-MYC were more highly expressed than endogenous DBT in clock cells (Figure 5A), and they should therefore titrate endogenous DBT for stable interactions with PER to dictate an allele-specific set point for circadian period. Both DBT-MYC and CKIδ under the control of the tim-GAL4 driver are expressed at all times of day in the head, and the levels of both wild-type and short-period transgenic protein are comparable for both CKIδ and DBT (Figure 5, A and B). The periods for CKIS are significantly longer (difference between tim-GAL4>UAS-DBTS and tim-GAL4>UAS-CKIS, P < 0.0002 by ANOVA with post hoc Tukey's test) and show a higher variance than those produced by DBTS (Table 1), suggesting the CKIδ may not titrate endogenous DBT as effectively as transgenic DBT. However, it is clear that the period-shortening effect of the amino acid change in CKIS is manifest and confers on the vertebrate protein the ability to act dominantly in the same manner as DBTS. This finding strongly supports the proposed evolutionary conservation of function between CKIδ and DBT and further establishes the validity of assessing the biochemical effects of the dbtS and dbtL mutations in the context of the vertebrate gene.
Besides the dbtS mutation from Drosophila (Price et al. 1998) and the CKIδ familial advanced sleep phase syndrome (FASPS) mutation in humans (Xu et al. 2005), a mutation of a single amino acid in CKIε of Syrian hamsters has been identified that gives rise to a semidominant period shortening in hamsters (Lowrey et al. 2000). Similar to DBTS, this CKIεTau exhibits a reduced kinase activity in vitro (Lowrey et al. 2000). We therefore sought to determine whether effects of the tau mutation, which maps to the highly identical N-terminal region of ckI [amino acid R178C (Lowrey et al. 2000)], would be functionally conserved if introduced into the vertebrate CKIδ (a paralog of CKIε) or Drosophila DBT and expressed in circadian clock cells in Drosophila. Expression of either UAS-CKIδTau or UAS-DBTTau with the tim-GAL4 driver produced comparable period shortening of up to 3 hr in all responder lines tested (21- to 23-hr periods in Table 1 and Figure 4), consistent with the short period length described for a homozygous ckIεTau hamster [20 hr (Lowrey et al. 2000)]. The difference between the average periods of the tim-GAL4>UAS-DBTTau (21.6 ± 0.5 hr) or -UAS-CKITau (21.1 ± 0.1 hr) lines and the average periods of the corresponding tim-GAL4>UAS-DBTWT (24.7 ± 0.1 hr) or >UAS-CKIWT (24.2 ± 0.1 hr) lines was significant (P < 0.0002 by ANOVA with post hoc Tukey's test), while the difference between the average periods of the tim-GAL4>UAS-DBTTau and >UAS-CKITau lines was not (P > 0.9000). The level of expression of the tau mutant proteins was comparable to those of the wild-type and dbtS forms of CKIδ and DBT, and there was again no evidence of oscillation (Figure 5, A and B).
The reproduction of the period alteration caused by a mutation in CKIε when introduced into CKIδ or Drosophila DBT and tested in a transgenic fly is very intriguing, since it is strong evidence for the high amount of functional conservation between ckIε and ckIδ, both of which are proposed regulators of the vertebrate clocks (Lowrey et al. 2000; Vielhaber et al. 2000; Lee et al. 2001; Constance et al. 2005; Eide et al. 2005; Xu et al. 2005). In addition, these results show the mechanistic conservation of the clock between flies and vertebrates, as the combined analyses of the dbtS and tau mutations show that mutations that appear hypomorphic when assessed in vitro but produce short periods in flies or vertebrates do so in both types of organisms and in the context of either the fly or the vertebrate enzyme, rather than producing opposite effects on period.
CKIδ proteins with the dbtL mutation do not produce period lengthening when expressed in Drosophila:
Six independent insertion lines of ckIδL were tested for alterations in their locomotor behavior when the transgene was expressed by the tim-GAL4 driver. None of the ckIδL transgenes gave rise to a clear change of period length of the transgenic flies, which had periods in the wild-type range (Table 1). The average period of the tim-GAL4>UAS-CKIL lines (24.1 ± 0.1 hr) was not significantly different from the average period of the tim-GAL4>UAS-CKIWT lines (24.2 ± 0.1 hr; P > 0.9 by ANOVA with post hoc Tukey's test). By contrast, all of the tim-GAL4>UAS-DBTL flies exhibited ∼27-hr periods, similar to the original dbtL allele (Price et al. 1998) and significantly different from the tim-GAL4>UAS-DBTWT lines (averages of 27.3 ± 0.1 and 24.7 ± 0.1 hr, respectively; P < 0.0002 by ANOVA with post hoc Tukey's test), as previously shown (Muskus et al. 2007). The expression levels of these UAS responders varied somewhat from line to line, and this was particularly true for the CKIL lines (supplemental Figure 1). However, the lines that produced wild-type periods in combination with the tim-GAL4 driver included ones with high levels of CKIL expression. For instance, line 4F2B did not lengthen period (Table 1), despite the fact that it expressed CKIδ at as high a level as did any of the lines expressing CKIWT (supplemental Figure 1). Hence, the lack of period lengthening in tim-GAL4>UAS-CKIL lines is not a consequence of low expression levels.
To address the possibility that long periods cannot be produced by lowered kinase activity of vertebrate CKIδ but can be in the context of the Drosophila DBT enzyme, another mutation that generated lower kinase activity (data not shown) was generated in vertebrate CKIδ by altering an amino acid (D132N) in the catalytic loop and introduced into transgenic flies. The effect on the circadian behavior is summarized in Table 1. The results show that the introduction of the hypomorphic form of the CKIδ protein leads to variable period lengthening in the fly in most tested lines. The average period of the tim-GAL4>UAS-CKID/N lines (25.6 ± 0.4 hr) was significantly longer than the average for the tim-GAL4>UAS-CKIWT lines (Table 1, 24.2 ± 0.1 hr; P < 0.05 by ANOVA with post hoc Tukey's test). However, while it is possible to lengthen period by expression of a CKIδ with reduced kinase activity, the effect is not as penetrant as with period shortening caused by the dbtS and tau mutations (only two of six lines produced >1.5 hr period lengthening).
The dbtS and tau mutations lead to distinct phosphorylation profiles for PER:
Since DBT is known to affect circadian rhythms by phosphorylating PER and targeting it for degradation, the molecular cycles of PER were examined in CKIδ and DBT-overexpressing flies. Flies were collected throughout the day, and the level and electrophoretic mobility (altered by phosphorylation state) of PER were assessed by immunoblot analysis of head extracts. In wild-type flies, PER exhibits a high mobility as it accumulates during the day and early night [Zeitgeber time (ZT) 7–13; lights on from ZT 0–12 and off from ZT 12–24] but becomes progressively less mobile on SDS–PAGE as it becomes phosphorylated during the late night and early day [e.g., ZT 19–1 (Edery et al. 1994)]. In flies overexpressing CKIWT or DBTWT, this pattern of phosphorylation was observed (Figure 6A; compare with nontransgenic + flies in Figure 6A), as we have previously observed in tim-GAL4>UAS-DBTWT flies (Muskus et al. 2007). Oscillations of PER were affected in tim-GAL4>UAS-DBTS and -UAS-DBTL flies as they are in the original EMS-induced mutations (Price et al. 1998) and are consistent with the more rapid and less rapid (respectively) phosphorylation kinetics observed in Drosophila S2 cells (Ko et al. 2002). In tim-GAL4>UAS-DBTS flies, PER's electrophoretic mobility retardation (caused by its progressive phosphorylation) was more striking at ZT 19 than in tim-GAL4>UAS-DBTWT flies, and PER disappeared more rapidly in the short-period genotype (compare amounts at ZT 1). In tim-GAL4>UAS-DBTL flies, this expedited phosphorylation of PER was not observed, and PER levels persisted at higher levels at ZT 1 than in tim-GAL4>UAS-DBTS flies (Figure 6A). PER accumulation and mobility profiles in tim-GAL4>UAS-CKIL flies were essentially wild type, as expected because behavioral rhythms were not altered in these flies (Table 1 and Figure 6A). The phase advance of PER cycling was not as dramatic in tim-GAL4>UAS-CKIS flies as in tim-GAL4>UAS-DBTS flies (note, however, the more rapid decline of PER levels at ZT 1 than in tim-GAL4>UAS-CKIWT), but this is consistent with the somewhat reduced period shortening caused by the UAS-CKIS transgene relative to the UAS-DBTS transgene.
The pattern of PER phosphorylation was distinctly different in tim-GAL4>UAS-CKITau and -UAS-DBTTau flies from other genotypes. PER exhibits more dispersed mobilities at ZT 13 and ZT 19 in both tim-GAL4>UAS-DBTTau and -UAS-CKITau flies, as well as at ZT 1 and ZT 7 in tim-GAL4>UAS-DBTTau (Figure 6, A and B), suggesting a more heterogeneous phosphorylation state for PER and a damped oscillation of phosphorylation. In addition, the oscillation in level was less robust than in other genotypes. Note in particular the slow mobility form of PER at ZT 7 in tim-GAL4>UAS-DBTTau flies and the lack of this form in any of the other genotypes (Figure 6B). Thus, while both the dbtS and tau mutations produce short circadian periods, their effects on the phosphorylation program of PER are different. Presumably, the sequence in which sites are phosphorylated, or their extent of phosphorylation, is different in the dbtS and tau mutants.
DBT is essential for circadian molecular oscillations because it introduces time delays into the negative feedback exerted by PER (Price et al. 1998). These phosphorylation-dependent delays are thought to be mediated via several mechanisms. For instance, DBT is thought to destabilize PER in the cytoplasm, while TIM (PER's dimerization partner) is thought to prevent this destabilization and trigger the movement of both proteins to the nucleus, perhaps indirectly (Vosshall et al. 1994; Price et al. 1995; Shafer et al. 2002; Cyran et al. 2005; Meyer et al. 2006). The destabilization of PER in the cytoplasm delays its nuclear accumulation, while the stabilization of PER in the nucleus by TIM delays the DBT-dependent decrease of nuclear PER (Price et al. 1995; Rothenfluh et al. 2000b; Kloss et al. 2001). Because DBT controls the timing of PER's nuclear accumulation so that it does not occur until after the per/tim mRNA levels have peaked, molecular rhythms of per and tim mRNA are possible (Price et al. 1998). Additional regulation has been proposed to occur at the level of PER's capacity to negatively regulate its transcription factor target (CLK/CYC) (Nawathean and Rosbash 2004; Kim and Edery 2006; Yu et al. 2006; Kim et al. 2007; Nawathean et al. 2007).
Although DBT's orthologs CKIδ and CKIε are involved in the mammalian circadian clock, the extent of evolutionary conservation of their circadian mechanisms has not been clear, as discussed in the Introduction. Our cellular and biochemical analysis argues for a significant degree of conservation. We have previously shown that both the original long-period mutation (dbtL) and the short-period dbt mutation (dbtS) reduce the enzymatic activity of Drosophila DBT and a Xenopus CKI ortholog of DBT on casein (Preuss et al. 2004), and here we showed that Drosophila DBTS and DBTL also exhibit reduced activity on PER. In our current work, deletion of the DBT C terminus was shown to increase the kinase activity of DBT and its capacity to target PER for degradation. The latter result is consistent with interaction studies, as it is the N-terminal catalytic domain rather than the C-terminal domain that interacts with PER (Preuss et al. 2004). Finally, it was shown here that in the presence of a general phosphatase inhibitor DBT produces forms that migrate more slowly on SDS–PAGE, in a manner that requires its kinase activity. These results suggest that DBT is autophosphorylated. Inhibitory autophosphorylation of the C-terminal domain is a common feature of both vertebrate CKIε and -δ (Graves and Roach 1995; Cegielska et al. 1998; Rivers et al. 1998). All of these biochemical results demonstrate a high degree of evolutionary conservation between Drosophila DBT and vertebrate CKIε/δ.
Our in vivo analysis herein of vertebrate CKIδ and Drosophila DBT further establishes the evolutionary conservation of these kinases and the clock mechanisms in which they participate. Mutations that shorten the circadian period in the context of DBT produced corresponding shortening in the context of CKI in flies, and the tau mutation produced almost identical period shortening in both the mammalian and the Drosophila clocks, in the context of either the fly or the vertebrate enzyme. The data for the tau mutation are particularly complete for analysis of evolutionary conservation, as this mutation has been tested in all possible combinations of organism (fly or mammal) and kinase (DBT or CKI) except one (DBTTau in mammals). These results argue against the possibility that reduced kinase activity produces only long periods in flies and short periods in mammals. On the other hand, the CKIδ FASPS mutation, which has a small effect on period, produces opposite effects on period in flies and mammals (Xu et al. 2005). The mutations analyzed herein have much stronger effects on period and clearly show similar effects in both vertebrates and flies. These findings strongly support the proposed evolutionary conservation of CKIδ and DBT protein kinases and of at least some of the circadian processes in which they are involved.
The reproduction of the period alteration caused by a mutation in CKIε when introduced into CKIδ or Drosophila DBT and tested in a transgenic fly is strong evidence for the high amount of functional conservation between ckIε and ckIδ, both of which are proposed regulators of the vertebrate clocks (Lowrey et al. 2000; Vielhaber et al. 2000; Lee et al. 2001; Constance et al. 2005; Eide et al. 2005; Xu et al. 2005). However, CKIδ and -ε are not completely interchangeable, as similar overexpression experiments with CKIε in flies have produced a very different set of results from the ones presented here for CKIδ. A catalytically active CKIε produces a dominant negative effect on circadian rhythms in Drosophila, with relatively constant expression of hypophosphorylated PER at all times of day, while overexpression of a catalytically inactive form of CKIε produces only a mild lengthening of circadian period (Sekine et al. 2008). Taken together with our results, this result suggests that vertebrate CKIδ is better able to interact with fly clock proteins than is CKIε, and in a manner more comparable to Drosophila DBT in the same expression protocol. The difference between the finding of Sekine et al. (2008) and our results, as well as those of Xu et al. (2005) (who also employed overexpression of CKIδ in flies and did not observe dominant negative phenotypes), may be due to the divergent C-terminal domains of CKIδ and -ε.
We have shown that flies expressing the dominant negative form of DBT have very long periods or are arrhythmic, and this result argues that general reductions in DBT's kinase activity produce long-period rhythms that grade to arrhythmicity (Muskus et al. 2007). While the lower activity of DBTL is predicted to (and in fact does) lengthen circadian period, the short-period PER oscillation produced by DBTS and tau mutations is not readily explained by their lower kinase activity in vitro. One possible explanation is that the short-period mutants affect something besides kinase activity (Muskus et al. 2007)—an interaction with a regulator, for example.
Another possible explanation is that reduced activity of short-period CKI enzymes in vitro may not translate into lower phosphorylation of PER in vivo because other kinases provide compensatory phosphorylation. In fact, progressive phosphorylation of PER, at least as assessed by a reduction in PER's electrophoretic mobility, occurs more rapidly in dbtS flies or cells than in wild-type or dbtL flies or cells (Price et al. 1998; Ko et al. 2002), as is also shown herein for tim-GAL4>UAS-DBTS flies, so the in vivo phosphorylation profile indicates more rapid phosphorylation in the dbtS mutant. Likewise, in the tau mutant, the phosphorylation of PER is only slightly delayed and ultimately appears to be as complete as in wild-type hamsters (Lee et al. 2001), despite the strong reduction in kinase activity caused by the tau mutation in vitro. In mammals, at least two kinases (CKIε and CKIδ) (Lowrey et al. 2000; Vielhaber et al. 2000; Lee et al. 2001; Constance et al. 2005; Eide et al. 2005; Xu et al. 2005) associate with PER and phosphorylate it, while in flies casein kinase II phosphorylates PER together with DBT (Lin et al. 2002; Nawathean and Rosbash 2004).
While the phosphorylation profiles of PER in tim-GAL4>UAS-DBTS, -CKIS, and -DBTL resemble the phosphorylation profiles of the original dbt mutants, the PER phosphorylation profiles of tim-GAL4>UAS-DBTTau and -CKITau do not resemble the profile that has been previously reported for the endogenous mutant (Lee et al. 2001); in particular, the oscillation of PER phosphorylation is notably blunted by overexpression of the DBTTau mutant kinases. It is possible the overexpression of the DBTTau has a stronger effect on the phosphorylation profile of PER in flies than in mammals because the tau mutant effects are partially masked in mammals by compensatory activity of CKIδ, with which CKIε may be partially redundant (Meng et al. 2008). Nevertheless, the period of the circadian clock is dramatically shortened by overexpression of both DBTS and DBTTau mutant kinases, and the circadian period of the tau mutation in flies is similar to that for the hamster tau mutant, suggesting that the alterations in the DBT/CKI kinase are still altering period as they do in the original mutants.
Another possible reason for the phenotypes of the short-period mutants is that phosphorylation of specific sites in PER affects multiple, specific aspects of its regulation with opposite effects on period length. Phosphorylation of Drosophila PER at multiple sites, only some of which affect stability, has recently been demonstrated (Chiu et al. 2008; Kivimae et al. 2008). Along this line of thinking, it has been proposed that the ckIεTau mutation is a gain-of-function mutation that enhances phosphorylation of PER at specific sites—an enhancement that is missed in global analysis of multisite substrates like PER (Gallego et al. 2006a; Meng et al. 2008). Others have offered explanations for the tau mutant that include lowered phosphorylation at all sites, but with cytoplasmic destabilization produced by phosphorylation at some sites and increased nuclear retention (and stabilization) produced by phosphorylation at other sites (Vanselow et al. 2006). If these hypotheses for site-specific effects are correct, the dispersed phosphorylation profile detected for Drosophila PER in our CKITau- and DBTTau-expressing flies is indicative of general changes in phosphorylation, most of which are not relevant to the period shortening, gain-of-function phenotype produced at a subset of sites. While both the dbtS and tau mutations produce short circadian periods, their effects on the phosphorylation program of PER are different, as reflected in their different effects on circadian changes in PER electrophoretic mobility.
Why does CKIL overexpression not lengthen the period of the Drosophila rhythm, while DBTL overexpression does? It is likely that the dbtL mutation compromises CKIδ function and reduces its ability to compete with endogenous wild-type DBT for interactions with PER. In fact, CKIδ may have generally less ability to compete with endogenous DBT than transgenic DBT, as overexpression of CKIS also has weaker effects than overexpression of the corresponding DBTS protein. Not all mutations that reduce the kinase activity of vertebrate CKI produce short periods, or no effect like CKIL. We show that the D/N mutation, which like the K/R mutation is predicted to have a very specific effect on the catalytic properties of the enzyme and not other aspects of its function, produced variable period lengthening. The variability is most likely a consequence of chromosomal position effects at the P-element insertion site on the transgene expression levels. The stronger effects of period-lengthening mutations in the context of fly DBT than in the vertebrate CKI suggest that there may be differences in the way they affect the fly and vertebrate CKI orthologs, with a consequence that fly DBT can be mutated more readily to produce a long period. These differences argue against the idea that lack of long periods is produced by a difference in circadian targets, as the lack of long periods correlates with the vertebrate enzyme rather than the species in which the enzyme is expressed (i.e., DBTL can produce long periods in flies, while CKIL cannot).
The involvement of protein kinases with circadian clocks spans a phylogeny that is larger than the one separating vertebrates and fruit flies. Casein kinases I and II are also involved in the bread mold (Neurospora) clock and target both the FRQ transcriptional repressor and the WCC transcriptional activator in a mechanism reminiscent of the one involving DBT (Garceau et al. 1997; Schafmeier et al. 2005; He et al. 2006; Huang et al. 2007). Recently, a kinase involved in DNA replication control was also shown to target Neurospora FRQ, with implications for the interplay between circadian rhythms and cell cycle control in higher eukaryotes (Pregueiro et al. 2006). The core circadian oscillator mechanism in cyanobacteria involves rhythmic phosphorylation and dephosphorylation of the KaiC protein (Nakajima et al. 2005). The evolutionary conservation of kinase function has led to a synergy between research in different organisms—for instance, with work in Drosophila identifying a circadian role for CKI (Kloss et al. 1998; Price et al. 1998) that has now been shown in diverse phyla and work in Neurospora showing a kinase-targeting role for FRQ (Schafmeier et al. 2005; He et al. 2006) that was subsequently shown for PER (Kim and Edery 2006; Yu et al. 2006) as well. Further elucidation of the evolutionarily conserved processes regulated by phosphorylation will reveal general mechanisms at the core of the circadian mechanism.
We gratefully acknowledge Ravi Allada for the construct used to produce recombinant PER, Anthony DiMaggio of the ICOS corporation for the anti-CKIδ antibody, the Developmental Studies Hybridoma Bank of the University of Iowa for the anti-tubulin antibody, the Bloomington Stock Center for the GAL4 driver line, and the Duke University transgenic fly facility for injections of our responder constructs. This work was supported by a Public Health Service grant from the National Institutes of Health (R01 MH056895).
↵1 Present address: Department of Neurobiology and Physiology, Northwestern University, Evanston, IL 60208.
Communicating editor: R. Anholt
- Received August 3, 2008.
- Accepted October 23, 2008.
- Copyright © 2009 by the Genetics Society of America