Circadian rhythms are biochemical, physiological and behavioral processes that follow a 24-hr cycle, responding primarily to the periods of light and dark, and they have been observed in bacteria, fungi, plants and animals. The circadian clock that drives these rhythms—which dictate our sleep patterns and other processes—involves a set of genes and proteins that participate in a collection of positive and negative feedback loops.

Previous research has mainly focused on identifying core clock genes—that is, genes that make up the molecular clock—and studying the functions of these genes and the proteins they code for. However, it has become clear that other clock genes are also involved in circadian behavior, and it has been proposed that polymorphisms in these non-core clock genes could contribute to the variations in circadian behavior displayed by different mammals.

One important feedback loop in mammals involves two key transcription factors, CLOCK and BMAL1, that combine to form a complex that initiates the transcription of the negative feedback genes, Period and Cryptochrome. Shimomura et al. discovered that Usf1, a gene that codes for a transcription factor that is typically involved in lipid and carbohydrate metabolism, as well as other cellular processes, is also important. In particular, this transcription factor is capable of partially rescuing an abnormal circadian rhythm caused by a mutation in the Clock gene in mice.

Shimomura et al. showed that the proteins expressed by the mutant Clock gene can bind to the same regulatory sites in the genome as the normal CLOCK:BMAL1 complex, but that gene expression of these targets is reduced because transcriptional activation is lower and binding of the complex is not as strong. However, proteins expressed by the Usf1 gene are able to counter this by binding to the same sites in the genome and compensating for the mutant CLOCK protein. Further experiments are needed to explore how the interactions between the USF1 and CLOCK:BMAL1 transcriptional networks regulate circadian rhythms and, possibly, carbohydrate and lipid metabolism as well.

To adapt to daily environmental cycles, most organisms have evolved endogenous clocks composed of cell-autonomous, self-sustained oscillators that drive 24-hr rhythms in biochemistry, physiology and behavior (Bass and Takahashi, 2010; Lowrey and Takahashi, 2011). In mammals, the innate periodicity of the circadian clock is generated by transcriptional/translational feedback loops composed of a core set of clock genes expressed in cells throughout the body (Reppert and Weaver, 2002; Lowrey and Takahashi, 2011). Two members of the bHLH-PAS transcription factor family, CLOCK (and its paralog NPAS2) and BMAL1 (ARNTL), heterodimerize and initiate transcription of the Period (Per1, Per2) and Cryptochrome (Cry1, Cry2) genes through E-box regulatory sequences (King et al., 1997; Gekakis et al., 1998; Hogenesch et al., 1998; Kume et al., 1999; Bunger et al., 2000; Huang et al., 2012). The resulting PER and CRY proteins form multimeric complexes, translocate to the nucleus, and abrogate their own transcription by repressing CLOCK:BMAL1 (Lee et al., 2001; Koike et al., 2012). As the PER and CRY proteins are degraded, CLOCK:BMAL1 occupancy increases to initiate a new round of transcription (Lowrey and Takahashi, 2011; Koike et al., 2012). While this forms the major feedback loop of the mammalian molecular clock, the overall mechanism also depends on additional feedback loops driven by ROR and REV-ERBα/β (Preitner et al., 2002; Sato et al., 2004; Ueda et al., 2005; Cho et al., 2012).

Significant progress has been made in identifying the core clock genes and the functions of their protein products in the clock mechanism, yet it is clear from other studies, including mutagenesis screens, that additional genes are necessary for a fully functional circadian clock (Takahashi, 2004). We previously identified at least 13 loci in mice that affect circadian behavior through complex epistatic interactions (Shimomura et al., 2001), indicating that there are other clock-relevant genes in the mammalian genome. By extension, it is possible that the variance in circadian behavior observed in the human population results from polymorphisms in non-core circadian clock genes. Thus, identifying these genes is important for both mechanistic understanding and translational application.

The quantitative trait locus (QTL) approach has been successfully applied to detect loci that associate with phenotypes of interest. An important application of QTL analysis is the identification of loci that modify the function and/or expression of a particular gene (Nadeau and Topol, 2006). A major obstacle, however, has been the cloning of these genes—particularly those underlying behavioral QTLs (Nadeau and Frankel, 2000). To overcome this challenge, several approaches have been proposed, yet it remains difficult to obtain a mapping resolution suitable for gene cloning by the positional candidate method (Darvasi and Soller, 1995; Churchill et al., 2004; Valdar et al., 2006). To date there are few examples of behavioral QTLs having been cloned using high-resolution mapping strategies (Yalcin et al., 2004; Watanabe et al., 2007; Tomida et al., 2009).

The mouse Clock^Δ19 mutation was originally induced on the C57BL/6J (hereafter B6) isogenic background by ENU mutagenesis (Vitaterna et al., 1994). Clock^Δ19 lengthens circadian free-running period by about 1 hr in heterozygous mice, and by about 4 hr in homozygous mutant mice followed by arrhythmicity upon exposure to constant darkness (DD). We observed that the circadian period phenotype in Clock^Δ19/+ animals is suppressed by crossing to the BALB/cJ (hereafter BALB) inbred strain. On an isogenic B6 background, Clock^Δ19/+ lengthens period ∼0.6 hr, and this is suppressed to ∼0.3 hr on a (BALB x B6)F1 background. In ([BALB x B6]F1 x BALB)N4 mice, the Clock^Δ19/+ period phenotype is completely suppressed (Figure 1A,B). This suggests that at least one dominant suppressor allele exists in the BALB genome. To test whether this suppression is also cell or tissue autonomous, we monitored the circadian period of bioluminescence in Clock^Δ19/+ suprachiasmatic nucleus (SCN) and pituitary explants from Period2^Luciferase (Per2^Luc) reporter mice (Yoo et al., 2004). The circadian period in both Clock^Δ19/+ SCN and pituitary was shorter in (BALB x B6)F1 animals than in mice of the B6 background (Figure 1C). These results demonstrate that the BALB suppression phenotype is tissue autonomous and present in both central (SCN) and peripheral (pituitary) circadian clocks.

Figure 1

Download asset Open asset

To assess whether suppressor alleles segregate in the BALB x B6 hybrids, we analyzed the circadian locomotor activity rhythm of Clock^Δ19/+ mice from five different populations between the BALB and B6 strains (Figure 1D). Analysis of period in these crosses suggests that the BALB allele acts semidominantly to suppress the Clock^Δ19 mutation. To map the Clock suppressor loci, we used the (BALB x B6)F2 generation (Figure 1D). We performed QTL analysis on 222 (BALB x B6)F2 Clock^Δ19/+ mice with 87 simple sequence length polymorphism markers (SSLP) and detected a significant association between Clock^Δ19 phenotype suppression and a locus on mouse chromosome 1, which we named Suppressor of Clock (Soc) (Figure 1E).

To isolate this region, we first selected mice carrying large fragments of BALB chromosome 1 by genotyping five SSLPs spanning this region (D1Mit22, 59.95 Mb–D1Mit155, 196.25 Mb). The congenic region was progressively narrowed to 25 Mb flanked by D1Mit452 (165.10 Mb) and D1Mit223 (190.45 Mb) (Figure 1E). We observed a significantly shorter free-running period in Clock^Δ19/+ animals from homozygous BALB Soc congenic lines than from Clock^Δ19/+ littermates lacking the BALB allele (Figure 1F). The effect of Soc was significant on Clock^Δ19/+ mutant, but not on wildtype or homozygous Clock^Δ19 backgrounds. The congenic analysis demonstrates that Soc is physically located within the 165.1–190.45-Mb portion of mouse chromosome 1 and thus defines a genomic interval containing a new clock-related gene.

Interval specific SNP haplotype analysis of the Soc locus

Because classical inbred laboratory mouse strains are derived from a limited number of progenitor species and the genomes of inbred strains are an admixture of different domesticated stocks (Wade et al., 2002; Frazer et al., 2007; Yang et al., 2011), polymorphisms at the Soc locus may be ancestral. If so, other inbred strains that share identity by descent with the Soc locus should also suppress Clock. To test this hypothesis, we characterized 14 additional laboratory inbred strains by crossing to B6 Clock^Δ19 mice to create F1 hybrids. Further, because strain background can affect wild-type circadian free-running period, a two-way ANOVA is required to distinguish the effects of the strain background (evident in wild-type) from the effects of the strain background on the expression of the Clock mutation (evident as a strain-by-genotype interaction effect). If we detected a significant strain-by-genotype interaction, we accepted this as evidence of suppression of the Clock mutation. Of the 14 inbred strains tested, we identified seven additional suppressor and seven non-suppressor inbred strains (Figure 2A; Table 1). These results suggest that the Soc phenotype occurs as a consequence of shared ancestral alleles.

Figure 2

Download asset Open asset

Figure 2—source data 1

Single nucleotide polymorphisms (SNP) in the Suppressor of Clock (Soc) candidate region among 16 mouse strains.

https://doi.org/10.7554/eLife.00426.005

Download elife-00426-fig2-data1-v1.xlsx

Table 1

Suppressor strain
Strain name	F-ratio (Clock × strain)	p value	df
BALB/cJ	14.698	<0.001	df(1,61)
A/J	13.13	<0.001	df(1,44)
AKR/J	9.41	0.004	df(1,44)
C3H/HeJ	56.82	<0.001	df(1,52)
DBA/2J	34.92	<0.001	df(1,46)
FVB/NJ	7.24	0.010	df(1,47)
SJL/J	12.03	0.001	df(1,55)
129S1/SvImJ	7.094	0.010	df(1,67)
Non-suppressor strain
C58/J	0.65	0.423	df(1,46)
BTBR_T+_tf/J	0.01	0.911	df(1,53)
I/LnJ	2.72	0.107	df(1,40)
MA/MyJ	0.02	0.891	df(1,52)
JF1/Ms	0.49	0.490	df(1,37)
MOLF/EiJ	0.90	0.348	df(1,37)
CZECHII/EiJ	0.08	0.780	df(1,46)

1. Only the F-ratio for interaction is shown. Two-way ANOVA analyses for circadian period were performed with four groups; C57BL/6J WT, F1 WT, C57BL/6J Clock/+ and F1 Clock/+. Suppressor strains show a significant interaction between Clock genotype and strain background (Clock × strain).

We performed pairwise SNP allele comparisons between B6 and the other 15 inbred strains in the 30-Mb interval from 160 to 190 Mb of mouse chromosome 1 using a total of 2714 SNPs (Figure 2B). Alternating regions of low and high SNP diversity are apparent in which low variation intervals represent strains sharing identity by descent, compared to high variation intervals that represent divergent ancestry. Within the 30-Mb interval, there is a region of high sequence variation between B6 and the suppressor strains, but identical by descent between B6 and the non-suppressor strains (Figure 2B). This region (green shading) spans 900 kb and contains 22 transcription units (Figure 2C; Figure 2—source data 1). Using the Mouse Phylogeny Viewer (Yang et al., 2011), the imputed subspecific origin of this 900 kb region reveals two sets of haplotypes that arise from Mus domesticus and M. musculus, respectively (Table 2). The suppressor strains all carry haplotypes originating from M. domesticus and the non-suppressor strains all carry haplotypes originating from M. musculus–thus confirming our hypothesis of an ancestral allele.

Table 2

Strain	Chromosome	Start position (bp)	End position (bp)	Subspecies	Sup or non-sup
BALB/cJ	1	173300000	174200000	Dom	S
A/J	1	173300000	174200000	Dom	S
AKR/J	1	173300000	173302142	Mus *	S
AKR/J	1	173302461	174200000	Dom	S
C3H/HeJ	1	173300000	174200000	Dom	S
DBA/2J	1	173300000	174200000	Dom	S
FVB/NJ	1	173300000	174200000	Dom	S
SJL/J	1	173300000	174200000	Dom	S
129S1SvlmJ	1	173300000	174200000	Dom	S
C57BL/6J	1	173300000	174200000	Mus	NS
C58/J	1	173300000	174028368	Mus	NS
C58/J	1	174187607	174200000	Dom*	NS
BTBR T<+>tf/J	1	173300000	174200000	Mus	NS
I/LnJ	1	173300000	174200000	Mus	NS
JF1/Ms	1	173300000	174200000	Mus	NS
MOLF/EiJ	1	173300000	174200000	Mus	NS
CZECHII/EiJ	1	173300000	174200000	Mus	NS

1. From the Mouse Phylogeny Viewer (Yang et al., 2011): http://msub.csbio.unc.edu. The AKR/J and C58/J strains contain the proximal and distal breakpoints, respectively, for the Soc locus, and thus there are two entries for these two strains. Dom = M. domesticus, Mus = M. musculus, S = suppressor, NS = non-suppressor.

2. *

   does not include Soc locus.

Identification of Soc

It is well established that circadian clocks exist in the SCN and throughout the body (Yoo et al., 2004). Because the Clock suppressor phenotype occurs in both SCN and pituitary tissue explants (Figure 1C), the gene encoded by the Soc locus should also be expressed in both central and peripheral tissues. To search for co-expression, we profiled the expression of the 22 Soc candidates as well as 9 circadian clock genes in 10 different tissues. Via cluster analysis, we found that all nine clock genes expressed a similar pattern among the tissues examined (Figure 3A), but that only seven Soc candidates (Vangl2, Usf1, Dcaf8, Copa, Pex19, Pea15, and Ncstn) expressed a similar pattern to the clock genes. We thus prioritized these seven Soc candidate genes for further analysis.

Figure 3

Download asset Open asset

The Clock^Δ19 mutation results from an A→T transversion in a splice donor site that causes exon skipping and disruption of the C-terminal transactivation domain of CLOCK (King et al., 1997). Consequently, the levels of Per and Cry mRNA are much lower in Clock^Δ19 mutants than in wild-type mice (Lowrey and Takahashi, 2011). Given the nature of the suppressor phenotype, we hypothesized that the product of the gene responsible for the Soc QTL should activate transcription from E-boxes in a manner similar to CLOCK:BMAL1. We tested E-boxes from Per1 and Per2 for activation by the seven Soc candidates in HEK293T cells. Of these, only Usf1 significantly activated transcription from both Per1 and Per2 E-boxes (Figure 3B). We observed a clear dose response of Per promoter activation by USF1 (Figure 3B). Unlike CLOCK:BMAL1, activation of E-boxes by USF1 was not inhibited by the CRYs (Figure 3B). Although it has been reported previously that USF1 binds E-box sequences (Ferré-D'Amaré et al., 1994), our finding that, of the seven Soc candidates, only USF1 effects significant transactivation from circadian-relevant E-boxes, provided initial molecular evidence that Usf1 is a candidate for the Soc locus.

We next tested the circadian expression of the seven candidate genes in liver tissue collected from Clock^Δ19/+ (BALB x B6)F1 and B6 animals every 4 hr on the third day of DD exposure. Among the genes tested, only the Usf1 transcript exhibited significant up-regulation in F1 mice (Figure 3C). By western blot analysis, USF1 protein levels in the liver were ∼40% higher in BALB animals (Figure 3D). To test whether the suppressor strain activates E-box-containing circadian clock genes in vivo, we examined the effect of strain background (B6 or [BALB x B6]F1) on expression levels of Per1, Per2, Cry1 and Cry2 in liver using quantitative PCR (qPCR). We detected significant up-regulation of all four circadian genes tested in the F1 background (Figure 3E), consistent with elevated expression of an E-box transcription factor such as USF1. Taken together these results strongly suggest that Usf1 is a prime candidate gene for the Soc locus. Because there are no coding mutations in Usf1 between the B6 and BALB mouse strains (Figure 2—source data 1), we hypothesized that the elevated expression of Usf1 in the BALB suppressor strain is likely the cause of Clock suppression.

Given that Soc is a dominant suppressor and the candidate gene, Usf1, exhibits up-regulated expression in the BALB Clock suppressor strain, loss-of-function tests for Usf1 are not appropriate. Instead, to test whether Usf1 is the gene responsible for the Soc QTL, we created Usf1 overexpression transgenic mice on an isogenic B6 genetic background. From previous work (Hong et al., 2007), we observed that a transgene fused to DNA fragments containing a CMV minimal promoter is induced at low levels, and thus created two independent Usf1 transgenic lines, both of which exhibited a significant behavioral period-shortening effect in Clock^Δ19/+ animals, thereby minimizing the possibility of transgene position effects (Figure 4A). Quantitation of Usf1 expression confirmed that the transgenic lines mimic the increase in Usf1 levels observed in the BALB suppressor strain (Figure 4B). Based on these results, we conclude that Usf1 is the gene encoded by the Soc locus. Consistent with the dominant action of Soc, Usf1 knockout animals do not show changes in period length, but do exhibit a reduction in circadian amplitude and locomotor activity levels (Figure 4C,D).

Figure 4

Download asset Open asset

For a gene to be a quantitative trait gene owing to an expression difference, the difference should be in a cis, rather than a trans regulatory element. We further examined the Usf1 allele-specific expression difference between B6 and BALB mice in (B6 x BALB)F1 animals, which controls for trans-effects and environmental influences (Cowles et al., 2002). Using three independent methods (Figure 5A,B,C), we found that the Usf1 expression difference between B6 and BALB is the result of polymorphisms in a cis-regulatory element causing allele-specific up-regulated expression of the BALB Usf1 transcript.

Figure 5

Download asset Open asset

To determine whether the allele-specific expression difference is the result of promoter polymorphisms between B6 and BALB, we tested basal promoter activity of ∼2.3 kb upstream of the Usf1 transcription start site using a luciferase reporter assay. Usf1 promoter sequences from both B6 and BALB were cloned into the PGL4 luciferase reporter vector and transfected into HEK293T cells. Although both the B6 and BALB sequences activated the luciferase reporter, the BALB sequence resulted in a significantly higher signal than that from B6 (Figure 5D). By exchanging EcoRI-XhoI fragments between the B6 and BALB reporter clones, we narrowed the critical polymorphisms to the EcoRI-XhoI fragment (∼1000 bp) and identified seven common polymorphisms (five SNPs and two deletions) that match the phenotypic distribution of 16 mouse strains (Figure 5E).

We next amplified ∼100-bp fragments containing each of the seven polymorphisms from BALB and B6 genomic DNA and assayed their reporter gene activity to test whether the candidate SNPs affect promoter activity. Among the six fragments tested (SNPs 5 and 6 were contained within the same fragment), only two (SNPs 3 and 7) showed strong promoter activity and statistically higher activity from the BALB allele compared to the B6 allele (Figure 5F). SNPs 3 and 7 correspond to rs31538551 and rs31538547, respectively, in the NCBI SNP database. Next, we used site-directed mutagenesis on the B6 promoter (EcoRI-XhoI fragment) to introduce the corresponding BALB polymorphism at each of the two SNP loci and examined promoter activity. We observed a significant increase in luciferase signal by replacing the B6 allele of SNP7 (rs31538547) with the BALB allele (Figure 5G). This strongly suggests that SNP7 from the BALB Usf1 promoter region is a quantitative trait nucleotide (QTN) responsible for the Soc phenotype.

CLOCK:BMAL1 and USF binding to the E-box motif

Because an elevation of Usf1 expression is most likely responsible for suppression of the Clock^Δ19/+ phenotype, we sought to determine the mechanism by which USF1 suppresses the Clock phenotype. Like CLOCK and BMAL1 (Ripperger and Schibler, 2006), USF1 binds to E-box motifs to regulate transcription (Ferré-D'Amaré et al., 1994). Using chromatin immunoprecipitation (ChIP) followed by qPCR analysis, we verified that CLOCK and USF1 can bind to common E-boxes in three circadian genes (Dbp, Per2 and Per1) (Figure 6A). Although not as robust as CLOCK, we observed USF1 binding to four E-boxes (Dbp E_P, Dbp E_I2, Per1 E_P1 and Per1 E_P2). In Clock^Δ19 mutants, however, USF1 occupancy increases and CLOCK^Δ19 occupancy decreases at these sites despite elevated CLOCK^Δ19 mutant protein levels (Figure 6B; Yoshitane et al., 2009). Further, we observed a time-dependent DNA binding pattern of USF1 with a peak at night that is antiphase to that of CLOCK which peaks at ZT8 (Figure 6C).

Figure 6

Download asset Open asset

To explore our observation that CLOCK^Δ19:BMAL1 occupancy decreases in Clock^Δ19/Clock^Δ19 mice we determined the relative affinity of CLOCK:BMAL1, CLOCK^Δ19:BMAL1 and USF1 complexes to E-boxes using EMSA. Given our results demonstrating Usf1 expression differences between the B6 (non-suppressor) and BALB (suppressor) mouse strains (Figure 3), we used isogenic B6 animals in these experiments to insure that USF1 levels were constant as we examined E-box occupancy of CLOCK, CLOCK^Δ19, and USF1. We examined the same E-box sites in the Dbp, Per1, and Per2 genes used for ChIP analysis (Figure 6A), and confirmed the identity of gel shift complexes by supershift using specific antibodies to CLOCK, BMAL1 and USF1 (Figure 6D). At all E-boxes, we detected three types of DNA:protein complexes: CLOCK:BMAL1 tandem heterodimers (CB2), CLOCK:BMAL1 heterodimers (CB1), and USF (consisting of USF1:USF1 and possibly USF1:USF2 or USF2:USF2 dimers). As shown previously, these E-box sites consist of tandem E-boxes spaced 6–10 nucleotides apart, and binding to these tandem sites appears to be cooperative (Figure 6E; Rey et al., 2011). Interestingly, wild-type CLOCK:BMAL1 complexes bind primarily as a tandem complex (CB2) at all four E-box sites with the Dbp E_P and Per2 E′ E-boxes showing the most bias towards tandem binding (Figure 6D). By contrast, the mutant CLOCK^Δ19:BMAL1 heterodimer binds primarily as a single complex (CB1) at all four sites. Antibody supershift experiments affected both the CB1 and CB2 complexes, consistent with the hypothesis that these complexes, whether containing mutant or wild-type CLOCK, represent single or tandem CLOCK:BMAL1 binding complexes, respectively (Figure 6D).

To measure the affinity of native CLOCK:BMAL1 protein-DNA interactions, we performed saturation binding experiments using EMSA to determine the apparent k_d and B_max values for the different complexes and E-box sites. To estimate the binding affinity of CLOCK:BMAL1, CLOCK^Δ19:BMAL1, and USF, liver nuclear extracts from ZT8 were incubated with increasing amounts of ³²P labeled Dbp E_P, Dbp E_I2, Per2 E′ or Per1 E_P1 E-box double-stranded oligonucleotide probes (Figure 7A). The binding of each of the complexes at the four E-boxes are presented as both absolute and relative binding plots to show both the relative amount of binding as well as the relative affinity (Figure 7B). At low probe concentrations, the wild-type CLOCK:BMAL1 complex (CB2) binds more strongly than CLOCK^Δ19:BMAL1 (CB1)—demonstrating the higher affinity of the wild-type complex for DNA. Although the affinity varies depending on the specific site, K_d values of wild-type CLOCK:BMAL1 range from 0.86 nM for Dbp E_I2 to 19 nM for Per1 E_P1, while the K_d of CLOCK^Δ19:BMAL1 ranges from 9 to 48 nM for Dbp E_I2 and Per1 E_P1, respectively (Table 2). K_d estimates indicate a binding affinity change from 2.5- to 11.8-fold between wild-type and mutant CLOCK:BMAL1 complexes depending on the E-box site (Figure 7B; Table 3). Parenthetically, these data also show that the affinity and maximal binding of CLOCK:BMAL1 vary widely among native E-box sites, thus it is important to base interpretations on more than one E-box site. There is no significant difference in USF1 binding affinity between the two genotypes and, in fact, the K_d of USF1 is similar to the mutant CLOCK^Δ19:BMAL1 complex (Figure 7B; Table 3).

Figure 7

Download asset Open asset

Table 3

	CLOCK:BMAL1	CLOCKΔ19:BMAL1	USF1 WT	USF1 ClockΔ19/ClockΔ19
Dbp EP-box
Bmax	0.97 ± 0.02	1.25 ± 0.05 *	0.83 ± 0.06	0.87 ± 0.04
Kd	2.01 ± 0.22*	32.94 ± 3.92	42.8 ± 9.30	49.56 ± 6.20
Dbp EI2-box
Bmax	0.79 ± 0.02*	2.38 ± 0.05	2.25 ± 0.05	2.10 ± 0.05
Kd	0.71 ± 0.10*	6.18 ± 0.54	6.10 ± 0.52	9.64 ± 0.96
Per2 E′-box
Bmax	0.26 ± 0.02*	0.43 ± 0.02	0.36 ± 0.03	0.41 ± 0.02
Kd	2.00 ± 0.64*	8.05 ± 1.57	17.81 ± 5.08	16.71 ± 2.46
Per1 EP1-box
Bmax	0.74 ± 0.12*	7.38 ± 1.18	3.06 ± 0.64	7.96 ± 2.01
Kd	16.55 ± 9.49	43.69 ± 19.40	33.36 ± 20.99	58.62 ± 37.82

1. Values are the mean + SEM derived from four-parameter nonlinear curve-fitting.

2. *

   indicates that the Bmax or Kd is significantly different from other groups, p<0.0001.

In addition to these overall trends, specific E-box sites also provide further insight into the affinity of CLOCK:BMAL1 binding. In the case of the Dbp E_I2 site (Figure 7A), where wild-type CLOCK:BMAL1 binds both as single and tandem complexes, the tandem CB2 site has much higher affinity than CB1, consistent with the prediction of previous work (Rey et al., 2011). In contrast, at the Dbp E_P site (Figure 7A), where mutant CLOCK^Δ19:BMAL1 complex can be seen binding as both single CB1 and tandem CB2 complexes, the affinities of the two mutant complexes are similar and lower than wild-type CLOCK:BMAL1 complex. This suggests that the affinity of the mutant CLOCK^Δ19:BMAL1 complex is lower for two reasons: first, it does not preferentially bind as a tandem complex, thus binding may not be cooperative; and second, its affinity, even as a tandem complex, remains lower suggesting that the Clock^Δ19 mutation interferes both with cooperativity and with affinity. Thus, our saturation binding experiments clearly demonstrate that the affinity of CLOCK^Δ19:BMAL1 is significantly lower relative to the wild-type complex, and that this lower affinity is similar to the affinity of USF1 at the same sites. This suggests that under normal conditions, the wild-type CLOCK:BMAL1 complex binds with much higher affinity than the USF1 complex, but when mutant, the affinities of the CLOCK^Δ19:BMAL1 and USF1 complexes are comparable which allows USF1 to bind more effectively to E-box sites.

Genome-wide location analysis of CLOCK:BMAL1 and USF1 binding sites

Because USF1 and CLOCK:BMAL1 compete for binding at the same E-boxes, and because CLOCK^Δ19:BMAL1 exhibits a lower affinity for these sites, we predicted that there should be a global increase in USF1 occupancy at CLOCK:BMAL1 binding sites in Clock^Δ19 mutant mice. To test this hypothesis we performed genome-wide location analysis of USF1, CLOCK and BMAL1 using ChIP-Seq on liver nuclei collected at ZT8 from wild-type and Clock^Δ19/Clock^Δ19 mice. Again, we used isogenic B6 mice in order to avoid the complication of different levels of USF1 in suppressor strains. Analysis of USF1 and CLOCK:BMAL1 binding sites reveals extensive overlap in binding in wild-type mice, with 497 of 1885 USF1 peaks binding either CLOCK or BMAL1 (Figure 8A). As predicted for Clock^Δ19/Clock^Δ19 animals, the overlap between USF1 and CLOCK:BMAL1 binding sites increases significantly to 1916 sites, an almost fourfold increase compared to wild-type controls (Figure 8A). In Clock^Δ19/Clock^Δ19 mutant animals, USF1 co-occupies 38% of CLOCK:BMAL1 sites (1916/5072)—a dramatic increase from 14% (497/3412) observed in wild-type animals. Surprisingly, USF1 binding also increases globally in homozygous Clock^Δ19 animals, even at sites that do not bind CLOCK or BMAL1, from 1885 peaks in wild-type animals to 6091 peaks in Clock^Δ19/Clock^Δ19 mutant animals.

Figure 8

Download asset Open asset

Figure 8—source data 1

ChIP-seq peak list for USF1.

https://doi.org/10.7554/eLife.00426.015

Download elife-00426-fig8-data1-v1.xlsx

Figure 8—source data 2

ChIP-seq peak list for CLOCK.

https://doi.org/10.7554/eLife.00426.016

Download elife-00426-fig8-data2-v1.xlsx

Figure 8—source data 3

ChIP-seq peak list for BMAL1.

https://doi.org/10.7554/eLife.00426.017

Download elife-00426-fig8-data3-v1.xlsx

Heat map analysis of the 1916 sites that are co-occupied by both USF1 and CLOCK:BMAL1 in Clock^Δ19/Clock^Δ19 mutant animals reveals a dramatic increase in binding intensity of USF1 in mutant mice (Figure 8B,C). Not only do the number of USF1 peaks increase, but also the binding intensity at existing sites increases in mutant animals. Concomitantly, in the mutant animals there is a decline in CLOCK binding intensity at these 1916 common binding sites (Figure 8B,C). We quantitated the binding intensity of the top 10% of the binding sites for all three ChIP-Seq experiments and performed comparisons between wild-type and Clock^Δ19/Clock^Δ19 animals (Figure 8C). For USF1 and CLOCK there was a significant increase and decrease, respectively, in binding between wild-type and mutant animals, but no change in BMAL1 binding (Figure 8C). Motif analysis of each group of binding sites identified the canonical E-box binding site, CACGTG, as the most significant motif, indicating both USF1 and CLOCK:BMAL1 bind to the same consensus sequence in both wild-type and Clock^Δ19/Clock^Δ19 mice (Figure 8D).

UCSC genome browser profiles are shown for examples of four genes (Figure 8E). Per1 has at least five E-boxes, however ChIP-Seq analysis reveals two major peaks approximately 5 kb upstream of the transcription start site as well as two minor peaks further downstream (Figure 8E). The peak height of USF1 binding increases dramatically in Clock^Δ19/Clock^Δ19 animals, and there is a slight reduction in CLOCK binding in mutant animals. Within the Dbp gene there are also multiple E-boxes, and USF1 binding increases in Clock^Δ19/Clock^Δ19 animals on the promoter, and intron two E-boxes. On the Dbp promoter CLOCK and BMAL1 binding decreases in the Clock^Δ19/Clock^Δ19 animals. A similar pattern is observed for other clock genes such as Rev-erbα which contains multiple E-boxes (Figure 8E) and other target genes such as Ilf3 (Figure 8E).

We extended this analysis of ChIP-Seq data to 41 E-boxes in 15 ‘reference clock genes' that are thought to be core regulators of the circadian clock (Rey et al., 2011). We specifically analyzed these E-box sites in the two genotypes for USF1 binding changes. Of these 41 sites, 23 bound USF1 with a significant increase in USF1 binding in the Clock^Δ19/Clock^Δ19 animals (Figure 8F,G). Thus, USF1 binds to a majority of reference clock genes and its occupancy on these sites increases in the Clock^Δ19/Clock^Δ19 mice.

These data suggest an intriguing mechanism of rescue of the Clock^Δ19 heterozygous phenotype by USF1. In this model, USF1 expression is higher in BALB than in B6 strain owing to a promoter polymorphism. Our data show that USF1 can competitively bind to the same E-boxes in vivo as CLOCK:BMAL1. Furthermore quantitative analysis of binding kinetics indicates that the CLOCK^Δ19:BMAL1 complex has a much lower binding affinity for E-boxes than its wild-type counterpart. This lowered affinity combined with increased expression of USF1 in BALB mice accounts for the rescue of the Clock^Δ19/+ phenotype in this strain. We propose that USF1 acts as a partial agonist whose transcriptional regulation of circadian genes can compensate for the reduction in transcriptional activation by the CLOCK^Δ19:BMAL1 complex.

Taken together, we demonstrate that a dominant suppressor of the Clock^Δ19 mutation in the BALB genetic background is an ancestral allele of Usf1 which carries a cis-regulatory SNP that enhances Usf1 expression (Figure 9A). USF1 is ubiquitously expressed in both mouse and human and is a member of the evolutionarily conserved bHLH-Zip transcription factor family. USF1 participates in the regulation of several processes including lipid and carbohydrate metabolism, immune responses, and the cell cycle (Corre and Galibert, 2005). USF1 interacts with the circadian clock gene pathway by binding to E-box regulatory sites in common with those bound by CLOCK:BMAL1 to regulate circadian gene expression (Figure 9B). We find that the mutant CLOCK^Δ19:BMAL1 complex binds with much lower affinity than wild-type CLOCK:BMAL1 complexes, arising in part from the absence of cooperativity of binding to tandem E-box sites seen with the wild-type CLOCK:BMAL1 complex (Figure 9C). Because USF1 binds as a dimer, the 40% increase in USF1 levels observed in the Clock suppressor strain could lead to a substantial increase in DNA binding if those levels are below the K_d for USF1 binding. In addition, because the PER:CRY negative feedback complex cannot repress USF1-mediated transactivation (Figure 3B), a modest increase in USF1 could activate target genes more effectively than CLOCK^Δ19:BMAL1. Thus, USF1 can suppress the Clock^Δ19 mutant by increased occupancy of CLOCK:BMAL1 E-box sites, acting as a partial agonist for CLOCK:BMAL1-mediated transcription. What might the role of USF1 be in a Clock wild-type background? We have found that as CLOCK:BMAL1 DNA binding decreases at night (Rey et al., 2011; Koike et al., 2012), USF1 occupancy increases at these sites, exhibiting an antiphase temporal DNA binding pattern (Figure 6C). We speculate that the role of USF1 at CLOCK:BMAL1 sites could be to maintain an open chromatin state to facilitate CLOCK:BMAL1 binding on the following circadian cycle although additional work would be required to test this hypothesis. In addition, in Usf1 knockout mice, we observe a reduction in circadian amplitude and activity levels showing that USF1 plays a role under normal conditions to enhance circadian rhythmicity and robustness. While it is tempting to speculate that these effects of Usf1 might work at the level of the cell autonomous circadian oscillator, additional experiments would be required to determine at what level of organization (cell, circuit or higher) these changes in amplitude at the behavioral level originate.

Figure 9

Download asset Open asset

In summary, the global interactions between CLOCK:BMAL1 and USF1 reveal an extensive and previously unknown interface linking these two transcriptional networks. It will be interesting in future work to determine whether the shared targets of these two pathways affect phenotypes beyond the circadian system. Taken together, these results show that USF1 is a significant modulator of molecular and behavioral circadian rhythms in mammals.

1. 1. Shimomura K
   2. Koike N
   3. Takahashi JS

   (2012) Usf1, a suppressor of the circadian clock mutant, reveals the nature of the DNA-binding of the CLOCK:BMAL1 complex

   ID GSE44609. Publicly available at the NCBI Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/).

   http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE44609

1. 1. Bass J
   2. Takahashi JS

   (2010) Circadian integration of metabolism and energetics

   Science 330:1349–1354.

   https://doi.org/10.1126/science.1195027

   • Google Scholar
2. 1. Bunger MK
   2. Wilsbacher LD
   3. Moran SM
   4. Clendenin C
   5. Radcliffe LA
   6. Hogenesch JB

   et al. (2000) Mop3 is an essential component of the master circadian pacemaker in mammals

   Cell 103:1009–1017.

   https://doi.org/10.1016/S0092-8674(00)00205-1

   • Google Scholar
3. 1. Cho H
   2. Zhao X
   3. Hatori M
   4. Yu RT
   5. Barish GD
   6. Lam MT

   et al. (2012) Regulation of circadian behaviour and metabolism by REV-ERB-α and REV-ERB-β

   Nature 485:123–127.

   https://doi.org/10.1038/nature11048

   • Google Scholar
4. 1. Churchill GA
   2. Airey DC
   3. Allayee H
   4. Angel JM
   5. Attie AD
   6. Beatty J

   et al. (2004) The collaborative cross, a community resource for the genetic analysis of complex traits

   Nat Genet 36:1133–1137.

   https://doi.org/10.1038/ng1104-1133

   • Google Scholar
5. 1. Corre S
   2. Galibert MD

   (2005) Upstream stimulating factors: highly versatile stress-responsive transcription factors

   Pigment Cell Res 18:337–348.

   https://doi.org/10.1111/j.1600-0749.2005.00262.x

   • Google Scholar
6. 1. Cowles CR
   2. Hirschhorn JN
   3. Altshuler D
   4. Lander ES

   (2002) Detection of regulatory variation in mouse genes

   Nat Genet 32:432–437.

   https://doi.org/10.1038/ng992

   • Google Scholar
7. 1. Darvasi A
   2. Soller M

   (1995)

   Advanced intercross lines, an experimental population for fine genetic mapping

   Genetics 141:1199–1207.

   • Google Scholar
8. 1. Ferré-D'Amaré AR
   2. Pognonec P
   3. Roeder RG
   4. Burley SK

   (1994)

   Structure and function of the b/HLH/Z domain of USF

   EMBO J 13:180–189.

   • Google Scholar
9. 1. Frazer KA
   2. Eskin E
   3. Kang HM
   4. Bogue MA
   5. Hinds DA
   6. Beilharz EJ

   et al. (2007) A sequence-based variation map of 8.27 million SNPs in inbred mouse strains

   Nature 448:1050–1053.

   https://doi.org/10.1038/nature06067

   • Google Scholar
10. 1. Gekakis N
    2. Staknis D
    3. Nguyen HB
    4. Davis FC
    5. Wilsbacher LD
    6. King DP

    et al. (1998) Role of the CLOCK protein in the mammalian circadian mechanism

    Science 280:1564–1569.

    https://doi.org/10.1126/science.280.5369.1564

    • Google Scholar
11. 1. Heinz S
    2. Benner C
    3. Spann N
    4. Bertolino E
    5. Lin YC
    6. Laslo P

    et al. (2010) Simple combinations of lineage-determining transcription factors prime cis-regulatory elements required for macrophage and B cell identities

    Mol Cell 38:576–589.

    https://doi.org/10.1016/j.molcel.2010.05.004

    • Google Scholar
12. 1. Hogenesch JB
    2. Gu YZ
    3. Jain S
    4. Bradfield CA

    (1998)

    The basic-helix-loop-helix-PAS orphan MOP3 forms transcriptionally active complexes with circadian and hypoxia factors

    Proc Natl Acad Sci USA 95:5474–5479.

    • Google Scholar
13. 1. Hong HK
    2. Chong JL
    3. Song W
    4. Song EJ
    5. Jyawook AA
    6. Schook AC

    et al. (2007) Inducible and reversible Clock gene expression in brain using the tTA system for the study of circadian behavior

    PLoS Genet 3:e33.

    https://doi.org/10.1371/journal.pgen.0030033

    • Google Scholar
14. 1. Huang N
    2. Chelliah Y
    3. Shan Y
    4. Taylor CA
    5. Yoo SH
    6. Partch C

    et al. (2012) Crystal structure of the heterodimeric CLOCK: BMAL1 transcriptional activator complex

    Science 337:189–194.

    https://doi.org/10.1126/science.1222804

    • Google Scholar
15. 1. Kim TK
    2. Hemberg M
    3. Gray JM
    4. Costa AM
    5. Bear DM
    6. Wu J

    et al. (2010) Widespread transcription at neuronal activity-regulated enhancers

    Nature 465:182–187.

    https://doi.org/10.1038/nature09033

    • Google Scholar
16. 1. King DP
    2. Zhao Y
    3. Sangoram AM
    4. Wilsbacher LD
    5. Tanaka M
    6. Antoch MP

    et al. (1997) Positional cloning of the mouse circadian Clock gene

    Cell 89:641–653.

    https://doi.org/10.1016/S0092-8674(00)80245-7

    • Google Scholar
17. 1. Koike N
    2. Yoo SH
    3. Huang HC
    4. Kumar V
    5. Lee C
    6. Kim TK

    et al. (2012) Transcriptional architecture and chromatin landscape of the core circadian clock in mammals

    Science 338:349–354.

    https://doi.org/10.1126/science.1226339

    • Google Scholar
18. 1. Kume K
    2. Zylka MJ
    3. Sriram S
    4. Shearman LP
    5. Weaver DR
    6. Jin X

    et al. (1999) mCRY1 and mCRY2 are essential components of the negative limb of the circadian clock feedback loop

    Cell 98:193–205.

    https://doi.org/10.1016/S0092-8674(00)81014-4

    • Google Scholar
19. 1. Lee C
    2. Etchegaray JP
    3. Cagampang FR
    4. Loudon AS
    5. Reppert SM

    (2001) Posttranslational mechanisms regulate the mammalian circadian clock

    Cell 107:855–867.

    https://doi.org/10.1016/S0092-8674(01)00610-9

    • Google Scholar
20. 1. Lowrey PL
    2. Takahashi JS

    (2011) Genetics of circadian rhythms in mammalian model organisms

    Adv Genet 74:175–230.

    https://doi.org/10.1016/B978-0-12-387690-4.00006-4

    • Google Scholar
21. 1. Nadeau JH
    2. Frankel WN

    (2000) The roads from phenotypic variation to gene discovery: mutagenesis versus QTLs

    Nat Genet 25:381–384.

    https://doi.org/10.1038/78051

    • Google Scholar
22. 1. Nadeau JH
    2. Topol EJ

    (2006) The genetics of health

    Nat Genet 38:1095–1098.

    https://doi.org/10.1038/ng1006-1095

    • Google Scholar
23. 1. Preitner N
    2. Damiola F
    3. Lopez-Molina L
    4. Zakany J
    5. Duboule D
    6. Albrecht U

    et al. (2002) The orphan nuclear receptor REV-ERBalpha controls circadian transcription within the positive limb of the mammalian circadian oscillator

    Cell 110:251–260.

    https://doi.org/10.1016/S0092-8674(02)00825-5

    • Google Scholar
24. 1. Reppert SM
    2. Weaver DR

    (2002) Coordination of circadian timing in mammals

    Nature 418:935–941.

    https://doi.org/10.1038/nature00965

    • Google Scholar
25. 1. Rey G
    2. Cesbron F
    3. Rougemont J
    4. Reinke H
    5. Brunner M
    6. Naef F

    (2011) Genome-wide and phase-specific DNA-binding rhythms of BMAL1 control circadian output functions in mouse liver

    PLoS Biol 9:e1000595.

    https://doi.org/10.1371/journal.pbio.1000595

    • Google Scholar
26. 1. Ripperger JA
    2. Schibler U

    (2006) Rhythmic CLOCK-BMAL1 binding to multiple E-box motifs drives circadian Dbp transcription and chromatin transitions

    Nat Genet 38:369–374.

    https://doi.org/10.1038/ng1738

    • Google Scholar
27. 1. Salmon-Divon M
    2. Dvinge H
    3. Tammoja K
    4. Bertone P

    (2010) PeakAnalyzer: genome-wide annotation of chromatin binding and modification loci

    BMC Bioinformatics 11:415.

    https://doi.org/10.1186/1471-2105-11-415

    • Google Scholar
28. 1. Sato TK
    2. Panda S
    3. Miraglia LJ
    4. Reyes TM
    5. Rudic RD
    6. McNamara P

    et al. (2004) A functional genomics strategy reveals Rora as a component of the mammalian circadian clock

    Neuron 43:527–537.

    https://doi.org/10.1016/j.neuron.2004.07.018

    • Google Scholar
29. 1. Shimomura K
    2. Low-Zeddies SS
    3. King DP
    4. Steeves TD
    5. Whiteley A
    6. Kushla J

    et al. (2001) Genome-wide epistatic interaction analysis reveals complex genetic determinants of circadian behavior in mice

    Genome Res 11:959–980.

    https://doi.org/10.1101/gr.171601

    • Google Scholar
30. 1. Takahashi JS

    (2004) Finding new clock components: past and future

    J Biol Rhythms 19:339–347.

    https://doi.org/10.1177/0748730404269151

    • Google Scholar
31. 1. Tomida S
    2. Mamiya T
    3. Sakamaki H
    4. Miura M
    5. Aosaki T
    6. Masuda M

    et al. (2009) Usp46 is a quantitative trait gene regulating mouse immobile behavior in the tail suspension and forced swimming tests

    Nat Genet 41:688–695.

    https://doi.org/10.1038/ng.344

    • Google Scholar
32. 1. Ueda HR
    2. Hayashi S
    3. Chen W
    4. Sano M
    5. Machida M
    6. Shigeyoshi Y

    et al. (2005) System-level identification of transcriptional circuits underlying mammalian circadian clocks

    Nat Genet 37:187–192.

    https://doi.org/10.1038/ng1504

    • Google Scholar
33. 1. Uno K
    2. Ueda HR

    (2007)

    Microarrays: quality control and hybridization protocol

    Methods Mol Biol 362:225–243.

    • Google Scholar
34. 1. Valdar W
    2. Solberg LC
    3. Gauguier D
    4. Burnett S
    5. Klenerman P
    6. Cookson WO

    et al. (2006) Genome-wide genetic association of complex traits in heterogeneous stock mice

    Nat Genet 38:879–887.

    https://doi.org/10.1038/ng1840

    • Google Scholar
35. 1. Vitaterna MH
    2. King DP
    3. Chang AM
    4. Kornhauser JM
    5. Lowrey PL
    6. McDonald JD

    et al. (1994) Mutagenesis and mapping of a mouse gene, Clock, essential for circadian behavior

    Science 264:719–725.

    https://doi.org/10.1126/science.8171325

    • Google Scholar
36. 1. Wade CM
    2. Kulbokas EJ III
    3. Kirby AW
    4. Zody MC
    5. Mullikin JC
    6. Lander ES

    et al. (2002) The mosaic structure of variation in the laboratory mouse genome

    Nature 420:574–578.

    https://doi.org/10.1038/nature01252

    • Google Scholar
37. 1. Watanabe A
    2. Toyota T
    3. Owada Y
    4. Hayashi T
    5. Iwayama Y
    6. Matsumata M

    et al. (2007) Fabp7 maps to a quantitative trait locus for a schizophrenia endophenotype

    PLoS Biol 5:e297.

    https://doi.org/10.1371/journal.pbio.0050297

    • Google Scholar
38. 1. Yalcin B
    2. Willis-Owen SA
    3. Fullerton J
    4. Meesaq A
    5. Deacon RM
    6. Rawlins JN

    et al. (2004) Genetic dissection of a behavioral quantitative trait locus shows that Rgs2 modulates anxiety in mice

    Nat Genet 36:1197–1202.

    https://doi.org/10.1038/ng1450

    • Google Scholar
39. 1. Yamazaki S
    2. Takahashi JS

    (2005) Real-time luminescence reporting of circadian gene expression in mammals

    Meth Enzymol 393:288–301.

    https://doi.org/10.1016/S0076-6879(05)93012-7

    • Google Scholar
40. 1. Yang H
    2. Wang JR
    3. Didion JP
    4. Buus RJ
    5. Welsh CE
    6. Bonhomme F

    et al. (2011) Subspecific origin and haplotype diversity in the laboratory mouse

    Nat Genet 43:648–655.

    https://doi.org/10.1038/ng.847

    • Google Scholar
41. 1. Yoo SH
    2. Yamazaki S
    3. Lowrey PL
    4. Shimomura K
    5. Ko CH
    6. Buhr ED

    et al. (2004) PERIOD2::LUCIFERASE real-time reporting of circadian dynamics reveals persistent circadian oscillations in mouse peripheral tissues

    Proc Natl Acad Sci USA 101:5339–5346.

    https://doi.org/10.1073/pnas.0308709101

    • Google Scholar
42. 1. Yoshitane H
    2. Takao T
    3. Satomi Y
    4. Du NH
    5. Okano T
    6. Fukada Y

    (2009) Roles of CLOCK phosphorylation in suppression of E-box-dependent transcription

    Mol Cell Biol 29:3675–3686.

    https://doi.org/10.1128/MCB.01864-08

    • Google Scholar
43. 1. Zhang Y
    2. Liu T
    3. Meyer CA
    4. Eeckhoute J
    5. Johnson DS
    6. Bernstein BE

    et al. (2008) Model-based analysis of ChIP-Seq (MACS)

    Genome Biol 9:R137.

    https://doi.org/10.1186/gb-2008-9-9-r137

    • Google Scholar

Author details

1. Kazuhiro Shimomura

   Center for Functional Genomics, Department of Neurobiology, Center for Sleep and Circadian Biology, Northwestern University, Evanston, United States

   Contribution

   KS, Conception and design, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

2. Vivek Kumar

   Department of Neuroscience, Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, United States

   Contribution

   VK, Acquisition of data, Analysis and interpretation of data, Drafting or revising the article

   Competing interests

   The authors declare that no competing interests exist.

3. Nobuya Koike

   Department of Neuroscience, University of Texas Southwestern Medical Center, Dallas, United States

   Contribution

   NK, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

4. Tae-Kyung Kim

   Department of Neuroscience, University of Texas Southwestern Medical Center, Dallas, United States

   Contribution

   T-KK, Conception and design, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

5. Jason Chong

   Department of Neurobiology, Northwestern University, Evanston, United States

   Contribution

   JC, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

6. Ethan D Buhr

   Department of Neurobiology, Northwestern University, Evanston, United States

   Contribution

   EDB, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

7. Andrew R Whiteley

   Department of Neurobiology, Northwestern University, Evanston, United States

   Contribution

   ARW, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

8. Sharon S Low

   Department of Neurobiology, Northwestern University, Evanston, United States

   Present address

   Center for Scientific Review, National Institutes of Health, Bethesda, United States

   Contribution

   SSL, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

9. Chiaki Omura

   Department of Neurobiology, Center for Functional Genomics, Northwestern University, Evanston, United States

   Contribution

   CO, Acquisition of data, Analysis and interpretation of data

   Competing interests

   The authors declare that no competing interests exist.

10. Deborah Fenner

    Department of Neurobiology, Center for Functional Genomics, Northwestern University, Evanston, United States

    Contribution

    DF, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

11. Joseph R Owens

    Department of Neurobiology, Center for Sleep and Circadian Biology, Northwestern University, Evanston, United States

    Contribution

    JRO, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

12. Marc Richards

    Department of Neurobiology, Northwestern University, Evanston, United States

    Contribution

    MR, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

13. Seung-Hee Yoo

    Department of Neuroscience, Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, United States

    Contribution

    S-HY, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

14. Hee-Kyung Hong

    Department of Neurobiology, Center for Functional Genomics, Northwestern University, Evanston, United States

    Contribution

    H-KH, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

15. Martha H Vitaterna

    Center for Functional Genomics, Department of Neurobiology, Center for Sleep and Circadian Biology, Northwestern University, Evanston, United States

    Contribution

    MHV, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

16. Joseph Bass

    Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, United States

    Contribution

    JB, Conception and design, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

17. Mathew T Pletcher

    Department of Genomics, Genomics Institute of the Novartis Research Foundation, San Diego, United States

    Present address

    Rare Diseases Research Unit, Pfizer Worldwide Research and Development, Cambridge, United States

    Contribution

    MTP, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

18. Tim Wiltshire

    Department of Genomics, Genomics Institute of the Novartis Research Foundation, San Diego, United States

    Present address

    Division of of Pharmacotherapy and Experimental Therapeutics, UNC Eshelman School of Pharmacy, Chapel Hill, United States

    Contribution

    TW, Acquisition of data, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

19. John Hogenesch

    Department of Pharmacology, Perelman School of Medicine, University of Pennsylvania, Philadelphia, United States

    Contribution

    JH, Conception and design, Analysis and interpretation of data

    Competing interests

    The authors declare that no competing interests exist.

20. Phillip L Lowrey

    Department of Neurobiology, Northwestern University, Evanston, United States

    Present address

    Department of Biology, Rider University, Lawrenceville, United States

    Contribution

    PLL, Analysis and interpretation of data, Drafting or revising the article

    Competing interests

    The authors declare that no competing interests exist.

21. Joseph S Takahashi

    Department of Neuroscience, Howard Hughes Medical Institute, University of Texas Southwestern Medical Center, Dallas, United States

    Contribution

    JST, Conception and design, Analysis and interpretation of data, Drafting or revising the article

    For correspondence

    joseph.takahashi@utsouthwestern.edu

    Competing interests

    The authors declare that no competing interests exist.

• 2,440

  views

• 405

  downloads

• 64

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.