The CRTC1-SIK1 Pathway Regulates Entrainment of the Circadian Clock

1There has been remarkable progress in our understanding of the complex intracellular mechanisms that generate and regulate circadian rhythms across multiple species. In mammals, the molecular clock arises from a transcriptional translational-feedback loop, consisting of the transcription factors CLOCK and BMAL1 that drive the expression of their regulators, PERIOD and CRYPTOCHROME. These elements form part of an autoregulatory-feedback loop, the period of which is approximately 24 hr. Most cells of the body appear to possess such a molecular oscillator, and these autonomous clocks are synchronized to each other and the external environment primarily by a circadian pacemaker within the suprachiasmatic nuclei (SCN). The SCN is in-turn aligned (entrained) to the solar day by exposure to light at dawn and dusk, detected by photoreceptors located exclusively within the retina of the eye. Photosensitive retinal ganglion cells (pRGCs), utilizing the photopigment melanopsin (OPN4), provide the primary light input to the SCN, exemplified by the fact that mice or human subjects lacking all their rods and cones still possess a full cohort of circadian responses to light. Although rod and cone photoreceptors are not required for entrainment, they can partially compensate for the loss of melanopsin in Opn4−/− mice. In addition, a growing body of evidence demonstrates that rods, and the different classes of cone photoreceptor, can complement the role of the pRGCs in circadian entrainment by expanding the irradiance range of the response and in the detection of light transitions.

2There remains, however, only a very basic understanding of the molecular mechanisms whereby cellular oscillators within the SCN are entrained to environmental light. This can be summarized as follows: light activation of pRGCs causes the release of the neurotransmitters glutamate and PACAP at the terminals of the retinohypothalamic tract (RHT) that projects directly to the SCN. Neurotransmitter release results in a Ca2+ influx, which activates “phosphorelay” signaling pathways. A number of different kinases have been implicated in these pathways, including PKA, PKG, CaMK, and MAPK, although their relative roles remain unclear. All these pathways are thought to converge on the cAMP response element binding protein (CREB), which is activated by phosphorylation at Ser133 and Ser142. Phosphorylated CREB then binds to cAMP response elements (CRE) in the promoters of light-regulated genes. The promoters of many previously identified light-responsive genes in the SCN including Fos, Nr4a1, Egr1, Egr3, the MAP kinase phosphatase Dusp1, and miRNAs miR-132 and miR-217 all contain a highly conserved CRE. Significantly, the clock genes Per1 and Per2 also contain conserved CREs, which mediate their light induction. Elevated PER1 and PER2 adjust the molecular-feedback loop of the circadian oscillator to the light/dark (LD) cycle.

3This linear description of light-induced Per gene induction cannot explain a key feature of light-regulated circadian behavior. Namely, that circadian entrainment is slow, taking multiple days to adjust to an advanced or delayed LD cycle. In most mammals, including jet-lagged humans, behavioral shifts are limited to approximately 1 hr (one time zone) per day. Furthermore, at a molecular level, the effect of light on Per gene induction is limited. For example, under extended illumination, Per1 mRNA light-induction peaks after ∼1 hr and then returns to baseline, indicating CREB transcription is arrested one hour after it is induced. Mirroring this molecular response, high-irradiance light pulses greater than 1 hr cease to have a phase-shifting effect upon the murine clock. This saturation of the response suggests the presence of an unidentified negative-feedback mechanism that effectively limits the magnitude of phase-shifting responses to light within the SCN. Thus, entrainment is a gradual process requiring repeated shifting stimuli over multiple days. Our aims in this study were to understand the molecular mechanisms that limit the effects of entrainment stimuli on the clock.

4Here, we show both in vitro (via serum shock) and in vivo (via light) that an entraining stimulus causes CREB-regulated transcription coactivator 1 (CRTC1) to coactivate CREB, inducing the expression of both Per1 and Sik1. SIK1 then inhibits further expression of Per1 by phosphorylation of and deactivation of CRTC1. Critically, we show that knockdown of Sik1 within the SCN results in extended light-induced phase shifts and an enhanced rate of re-entrainment to an advanced LD cycle. As SIK1 acts to suppress the effects of light on the clock, this pathway provides a pharmacological target for the treatment of disturbed circadian rhythms, including jet lag, shift-work, and illnesses that disturb circadian timing.

5In order to identify signaling pathways involved in photic input, we examined the light-regulated transcriptome of the wild-type (Opn4+/+) SCN across a period of nocturnal light exposure. We collected SCN from mice that received a nocturnal light pulse delivered 4 hr after activity onset (circadian time/CT16) over a 2 hr time course and identified 536 genes as light-regulated in the wild-type SCN using Affymetrix whole-genome mouse exon arrays (Figure 1A and Figure S1 and Table S1 available online). Surprisingly, the majority (81%, 436 genes) of transcripts were downregulated in response to light, with a smaller number of genes showing upregulation (19%, 100 genes). These included the acutely upregulated immediate early genes (IEGs) previously shown to respond to light in the SCN, including Fos, Nr4a1, and Egr1, clock genes (Per1), and clock-controlled genes (Rorb). In addition, we confirmed Rrad, a GTPase modulating calcium-binding proteins such as CAMKII, which has an identified role in photic resetting and Dusp1, a MAP kinase phosphatase, characterized as light-induced and clock-controlled in the SCN.

6Among the genes we identified were Dusp4 and Sik1, which were found to cluster with Per1 (Figure 1A). Sik1 encodes salt inducible kinase 1, a Ser/Thr protein kinase of the AMP-dependent protein kinases (AMPK) family of which AMPK rhythmically phosphorylates Cry1 and Dusp4 encodes a dual Ser/Thr and Tyr phosphatase selectively controlling ERK. The downregulated transcripts include Tac1, which encodes substance P precursor, known to have a role in phase shifting; Sim1, a transcription factor of the bHLH-PAS family that also includes Clock; Plekho1, which shows homology with casein kinase and a number of receptors including the serotonin receptor Htr7; Lactrophilin (Lphn3); and the glycine receptors Glra1 and Glra2.

7To gain insight into the pathways represented by these transcripts, we performed a gene-enrichment analysis based on gene ontology (GO) terms (Figure 1B). One category of interest that is significantly responsive to light includes a number of transcripts relating to synaptic remodeling and plasticity, which were upregulated at a late stage, and included Raspgrp1, Grit, and Ntng1.

8A number of previous studies have highlighted the central role of CREB in phase shifting and the promoters of many previously identified light-responsive genes in the SCN including Fos, Per1, Nr4a1, and Dusp1 contain conserved CREs. We overlaid our data with those of ) to identify transcripts containing CRE sites in their promoters. We found that of the genes upregulated more than 1.5-fold, there was significant enrichment of those containing CRE sites near the TATA box (Figure 1C). This correlation implies that CREB drives the high-magnitude gene induction that underlies phase-shifting responses in the SCN. We provide a detailed characterization of CREB target genes that are induced by light in the SCN (Table S2). Analysis of individual clusters of the Opn4+/+ SCN light-induced transcriptome showed that in addition to CREB, binding sites for SP1 and KROX (in transcripts upregulated at 120 min) and TAL1, HNFalpha3, and WT1 (transcripts showing early repression followed by upregulation) were enriched.

9To identify any genes from our screen that could play a direct role in the core circadian oscillator, we reviewed data obtained from bioluminescent recordings from a genome-wide RNAi screen on Per2-luc transfected U2OS cells. Silencing of several of the genes caused significant changes in period length, which would indicate a role in regulating phase delays of the clock. Of particular interest were two kinases that caused period lengthening. Csnk2a1 (period = 31.2 hr versus 25.2 for control), which is downregulated by light, is a serine threonine kinase. It is closely related to CsnkE, which is mutated in the Tau mutant hamster, which also shows period lengthening. Silencing of Sik1, which is upregulated by light in the SCN, in a manner similar to Per1 (Figure 1A) also shows period lengthening (28.0 hr).

10Melanopsin-expressing retinal ganglion cells provide the major retinal projection to the SCN. The endogenous light sensitivity of these cells conferred by melanopsin (OPN4) is important for circadian entrainment. Mice lacking melanopsin (Opn4−/−) show attenuated phase-shifting responses to light, which correlate with attenuated induction of Fos and Per1 in the SCN. We compared SCN samples from Opn4−/− mice with the data from wild-type animals (Opn4+/+); 92% of the genes that were upregulated in response to light in the wild-type showed an attenuated light-regulated pattern of expression in the Opn4−/− SCN (Figures S2A and S2B). We verified the attenuated induction of Fos, Per1, and Egr1 (Figure S2C) in the Opn4−/− SCN using qPCR. Analysis of transcription factor binding site (TFBS) showed that those transcripts with the greatest attenuation in Opn4−/−mice, including Fos, Nr4a1 and Per1, were enriched for WT1, a repressor of Egr1 (Tables S3 and S4). Our results indicate that in Opn4−/− mice, transcriptional responses in the SCN are broadly attenuated, mirroring the attenuated Fos induction and circadian responses to light previously observed. These results suggest that in the absence of melanopsin, there is an overall reduction in signal amplitude reaching the SCN via the RHT.

11As CREB transcription plays a critical role in photic entrainment, and because SIK1 acts both as a repressor of CREB transcription and as a target of CREB regulation itself, Sik1 emerged as a strong candidate for an inducible repressor acting as a brake on photic input to the SCN. SIK1 functions by phosphorylating and thereby deactivating the CREB coactivator CRTC (CREB-regulated transcription coactivator), which exists in three isoforms, CRTC1–3. These isoforms exhibit a tissue-specific expression pattern and the dual requirement for CREB phosphorylation and the presence of CRTC provides the complexity and specificity required in the response to cAMP signaling. Indeed, the SIK/CRTC pathway can regulate CREB transcription in other cellular systems. In these cases, Ca2+ and cAMP signaling triggers calcineurin-mediated dephosphorylation of CRTC, and its nuclear translocation. CRTC in turn stimulates CREB transcription of genes containing CRE, including Sik1. The resulting induction of SIK1 represses CREB transcription through the phosphorylation of CRTC. In such a manner, SIK1 could provide negative feedback on photic input to the SCN by inhibiting the light-induced, CREB-mediated expression of Per1.

12To test this hypothesis, we investigated the function of the CRTC-SIK1 pathway in regulating the circadian clock. To this end, we studied the nuclear translocation of CRTC1; the resultant CREB-mediated Sik1 and Per1 induction; and the effect of SIK1 on the transcription of Per1 and on re-entrainment in both cellular clock models and the intact animal.

13In order to test the effects of SIK1 inhibition on the molecular oscillator, we used the immortalized mouse embryonic fibroblast cell line NIH 3T3 as an in vitro model of the circadian clock. These cells can be phase shifted using serum stimulation. We found that CRTC1 translocated to the nucleus following a serum shock (Figure 2A) and Sik1, which contains a conserved CRE/ATF (Figure 2B top) was induced (Figure 2B bottom). Increased levels of CRTC phosphorylation measured in cell lysates collected before and after serum shock, were observed as a consequence of Sik1 induction (Figure 2C). In order to specifically inhibit SIK1 activity, we used a gene-specific RNAi-based approach (Figures 2D and S3A–S3D and Table 1). Lysates of Sik1 knockdown cells showed no increase in CRTC phosphorylation after serum shock (Figure S3E). Cells transfected with siRNA against Sik1 (siSik1) showed severely attenuated induction of Sik1 resulting from serum treatment (Figure 2D and Table 1) with no significant change over time. To test the proposed role of Sik1 and Crtc in phase shifting of the clock, Per1 induction was assayed following silencing of Sik1 and Crtc isoforms expressed in NIH 3T3 cells (Crtc1 and Crtc3, Figure S3B). Significant changes of Per1 levels over the duration of the experiment were apparent after treatment with nontargeting control (siNT) and siSik1 (Figure 2E and Table 1). However, knockdown of Sik1 resulted in an increase in the duration and levels of Per1 induction as measured by a significant increase in the area under the curve for Sik1 (Figure 2G, Tables 1 and 2). In order to test that SIK1 acts via CRTC to attenuate CREB-mediated transcription of Per1, we assessed the effect of cosilencing Crtc1 and Crtc3 (siCrtc) on serum-shock-induced Per1 expression. Knockdown of Crtc attenuated the induction of Per1 (Figures 2F and 2G, Tables 1 and 2) with no significant difference over time after serum treatment. This result supports the proposed model whereby the effects of SIK1 are mediated through CRTC. Crucially, this result also demonstrates that CRTC is an important CREB coactivator required in the phase-shifting response. Circadian expression of Per1 is regulated by multiple promoters, including the E box; therefore, we confirmed this requirement for CRTC in the phase-shifting response by examining the induction of two other light/serum-responsive genes (Egr1 and Nr4a1) that contain CREs in their promoters (as identified previously) and show a robust response to phase-shifting stimuli. These transcripts showed increased induction in cells treated with siSik1 and attenuation with siCrtc (Figures 2H, 2I, S3F, and S3G and Tables 1 and 2). A summary of these results are presented in Tables 1 and 2. These data support our hypothesis that SIK1 acts to modulate the magnitude of phase shifts via the negative regulation of CREB-mediated transcription of Per1.

14Findings from the in vitro study using RNAi in fibroblasts were further supported by data from the direct pharmacological inhibition of SIK1 by indirubin-3′-monoxime (I3M) (Figure 3A), which produced both an increase in the level and the duration of serum-induced Per1 and Per2 expression in NIH 3T3 cells following a serum shock (Figures 3B and 3C). In order to quantify the phase shift induced, we used PER2::LUC mouse embryonic fibroblasts (MEFs) in a double-serum treatment experiment. The first serum shock was used to synchronize the cells and the second to allow precise quantification of the phase shift. We observed a 10.90 ± 1.02 hr phase delay of the PER2::LUC rhythm in I3M-treated cells, compared with 7.59 ± 1.00 hr delay in the control, a difference of 3.3 hr (Figures 3D and 3E). In order to assess the effect of I3M on the SCN, Per1:luc mice were exposed to light at ZT14 for 10 min, and the SCN was removed and maintained in culture. The rhythms of SCN treated with I3M peaked approximately 1 hr later than DMSO-treated controls (Figure 3F). It should be noted that I3M is a nonspecific inhibitor of SIK1, and although directly inhibiting SIK1 (Figure 3A), its targets also include the kinase GSK3B, which has a role in the circadian clock. Nevertheless, these results using I3M and those using Sik1-specific RNAi are consistent (Figure 2). Collectively, these data show that inhibition of SIK1 increases the magnitude of phase delays in the molecular clock.

15Of the different CRTC isoforms, CRTC1 is the most abundantly expressed in the SCN (Figure 4A). Therefore, we investigated whether phase-shifting stimuli trigger the nuclear translocation of CRTC1 in the SCN. Application of a serum shock to dissociated SCN cells in culture resulted in the translocation of CRTC1 from the nucleus to cytoplasm within 10 min (Figure 4B) and an upregulation of SIK1 in the SCN at 120 min after the light pulse (Figure 4C). We verified the expression patterns of Sik1 in the SCN seen on the exon arrays using qPCR (Figure 4D). Sik1 was light-induced with a peak mRNA expression at 60 min following both a CT16 phase-delaying and CT22 phase-advancing light pulse in wild-type mice, confirming a role for Sik1 in both phase advances as well as phase delays. Further, Sik1 induction is higher at CT22 when compared with CT16. In agreement with a role of SIK1 in negative feedback, Egr1 and Nr4a1 expression were found to be correspondingly lower at CT22.

16To address effect of SIK1 on circadian behavior, we used in vivo RNAi to provide an acute and specific inhibition of Sik1 within the SCN. Cy3-labeled nontargeting siRNA-lipid complex was administered via stereotaxic injection into the third ventricle adjacent to the SCN. Subsequent examination of the SCN showed the accumulation of siRNA in the cell bodies (Figure S4A), confirming transfection. We designed and tested siRNAs against Sik1 that achieved significant in vivo silencing in the SCN at both mRNA and protein levels (Figure S4B), causing little or no interferon responses (Figure S4C).

17We then assessed behavioral responses in Sik1-silenced mice to three light regimes: (1) a phase-delaying nocturnal light pulse; (2) re-entrainment to a 6 hr phase advanced LD cycle (jet-lag protocol) and (3) period length in constant darkness (DD). For the phase-shifting paradigm, mice injected with either a control siRNA or Sik1 siRNA were exposed 4 days later to a 30 min, type II phase-shifting light pulse (400 lux) at CT14.5 (Figure 5A). Compared with the controls, the mice with Sik1 silenced showed significantly enhanced phase shifts (98 min compared to 50 min, p = 0.036, Figure 5B). For the jet-lag protocol, 2 days after siRNA administration, the LD cycle was advanced 6 hr, and 10 days later, the cycle was advanced 6 hr again. Sik1 silenced mice showed significantly faster re-entrainment when compared to the controls in both cycles (Figure 5C). To illustrate the magnitude of this effect, on day 2 the difference in phase was over 3 hr in the siNT-injected animals but was less than 1 hr in the Sik1-silenced mice (Figure 5D).

18For period length, we measured period length (tau) in both cell lines (Per2::Luc U2OS) as well as in vivo after Sik1 knockdown. Period length was markedly increased in cell lines (24.8 hr with greater than 90% knockdown of Sik1 versus 22.1 hr with siNT-treated cells), and less so in vivo (23.27 hr with approximately 50% knockdown of Sik1 versus 23.04 siNT, Figure S5). These data suggest that in addition to acute effects on gene expression, CREB may play a role in constitutive clock gene regulation. Indeed, recent reports show that CREB silencing results in overall lower expression of clock genes and output hormones such as AVP and VIP. In this case, basal levels of SIK1 would be important in regulating the activity of CRTC, which in turn acts as a CREB coactivator. As a result, basal levels of SIK1 may affect period length via this CREB-dependent mechanism.

19Collectively, our in vitro studies combined with these behavioral data, provide critical support for a role of SIK1 acting as an inducible repressor inhibiting the effect of phase-shifting stimuli on the circadian clock by attenuating Per1 induction.

20The circadian system is entrained to environmental zeitgebers (time givers), which allow the appropriate alignment of internal and external time. However, the adjustment of the clock to such zeitgebers is rarely instantaneous, often occurring over repeated 24 hr cycles. What limits the effects of entraining stimuli on the circadian system represents a key, yet largely unanswered question. This issue provides the focus for the work presented here. We have identified >500 genes that are light regulated within the SCN. From these, we defined a key role for the CRTC-SIK1 pathway in regulating CREB-mediated clock gene expression. Our data show both in vitro and in vivo that CRTC1 translocates to the nucleus after phase-shifting stimuli and acts as a coactivator of CREB-driven transcription of Sik1 and Per1. Following in vitro Crtc silencing, induction of Per1 and other CREB-mediated genes is greatly attenuated. Elevated Per1 aligns the transcriptional/translational-feedback loop to the zeitgeber stimulus. We also show that SIK1 phosphorylates and deactivates CRTC with an appropriate time delay, which effectively blocks further CREB-driven Per1 transcription. Critically, we show that in vivo RNAi knockdown of Sik1 within the SCN greatly enhances light-induced behavioral phase shifts and re-entrainment to LD cycles. Collectively, these data provide empirical evidence for a molecular mechanism that limits the impact of an entraining stimulus on the circadian system.

21It is worth emphasizing that the SIK1-CRTC regulation of CREB-mediated gene expression we describe within the context of the circadian system has also been demonstrated in several nonclock pathways. For example, in skeletal myocytes, SIK1 functions as a Ca2+-responsive CREB-regulated HDAC kinase and a Ca2+-induced regulator of Na+ transport in kidney cells. In hippocampal neurons, CRTC1 functions as a sensor of stimulation, by traveling from the synapse to the nucleus in response to activation, triggering CREB mediated transcription and thereby regulating hippocampal plasticity. An appreciation of the role of CRTC1-SIK1 regulation in these noncircadian pathways very much supports our interpretation of how this signaling pathway functions within the molecular clock.

22The same regulatory processes that induce Per1 also induce Sik1. Following the induction of these genes, phosphorylation then plays a key, yet distinct, role in the molecular clockwork. The phosphorylation of core clock proteins is a major posttranslational event whereby delays are introduced into the transcriptional-translational-feedback loop to generate a ∼24 hr oscillation. Several proteins have been described in this process, including FBXL3, an F box protein that targets phosphorylated CRY for degradation; CSNK1e, in the Tau mutant hamster, which phosphorylates PER1, CRY1, and BMAL1 and NEMO/NLK that phosphorylates PER to initiate a time-delay phosphorylation circuit. In contrast to these kinases, SIK1 does not target the core clock proteins to control their degradation; rather, it attenuates the induction of clock genes in response to entraining stimuli, functioning as a negative-feedback mechanism to deactivate the molecular pathways that are activated following a phase-shifting stimulus.

23But why limit the impact of an entraining stimulus on the clock? Almost all physiological systems benefit from buffering mechanisms that prevent sudden and large changes, and in this regard the circadian system is no exception. This would be particularly important for the clock because it has to maintain internal synchrony between the multiple cell-autonomous circadian oscillators throughout the body. If sudden and large shifts are imposed upon one part of the circadian network, such as the SCN, internal desynchrony among the cellular clocks of the body will result. Under such circumstances normal physiology will be severely compromised. By limiting the phase-shifting effects of light and other clock-resetting stimuli, such as food or activity the circadian system would be shielded from such disruption. Such a buffering mechanism in the natural world would protect the clock from abnormal Zeitgeber stimuli. In addition, limiting the impact of light on the SCN would be useful in nonequatorial latitudes during the equinoxes when the time of dawn and dusk, along with activity and food availability, moves very rapidly. With a rapidly changing twilight transition, the circadian system has to maintain a balance between entrainment and the prevention of internal de-synchrony. Under these conditions, CRTC1-SIK1-driven inhibition of Per transcription may help achieve this compromise.

24Sleep and circadian rhythms disruption (SCRD) is endemic within the 24/7 economies of the developed and developing industrialized societies. In addition, SCRD is a common feature of both neuropsychiatric and neurodegenerative disease. In shift-work, jet lag, and brain abnormalities in general, SCRD is associated with chronic diseases such as cancer, metabolic disorders, cardiovascular disease, immune dysfunction, and impaired cognition. Currently, there are very few options available to stabilize or adjust the circadian axis and sleep timing in individuals with SCRD. However, the discovery of the molecular mechanisms that generate and regulate the circadian system provides targets for the development of therapeutic agents that regulate internal time. Indeed, there have been recent efforts in this direction, including the use of small molecule activators of cryptochrome and an inhibitor of CSNK1E. Such modulators act on the central clock to lengthen or shorten period. An alternative approach would be to shift the clock to an appropriate phase. As SIK1 acts to suppress the effects on light on the clock, this pathway presents a tractable target.

25Opn4−/− mice were maintained on a C57Bl/6_129Sv background as heterozygous breeders. For SCN sample collection and RNA extraction, see Extended Experimental Procedures.

26Sense probes were prepared for hybridization from 300 ng of total RNA for Opn4+/+ shams SCN (n = 6), LP30, LP60, and LP120 (n = 4) and Opn4−/− shams (n = 4), LP30, LP60, and LP120 (n = 3) using the GeneChip WT Sense Target Labeling kit (Affymetrix, Santa Clara, CA, USA) and hybridized overnight to GeneChip Mouse Exon 1.0 ST Arrays (Affymetrix Santa Clara, CA, USA). Arrays were washed, stained, and scanned according to manufacturer’s guidelines.

27CEL file data were extracted, log transformed and quantile normalized with Expression Console (v1.1) software from Affymetrix using the RMA-sketch algorithm at the exon-core probe set level including the detection above background (DABG) p value for each probe set. Exon array quality control was assessed using Expression Console (V1.1) and any outliers removed. Subsequent analysis steps were carried out in Microsoft Excel 2007. Probe sets were filtered by DABG p value for each array sequentially, discarding data for probe sets for which p ≥ 0.05. Only probe sets called as “present” on all arrays (n = 18) were further analyzed (110,026 probe sets) by one-way ANOVA and filtered by p value (p ≤0.05) (corresponding to 4,384 probe sets that respond to light). Genes that were light-responsive were identified by significant probe set enrichment. Probe set enrichment was evaluated by comparing the number of significant probe sets for an individual gene to (1) number of probe sets on the array for that gene, (2) number of probe sets significantly changing, and (3) the total number of probe sets on the array based upon a binomial distribution. Finally, probe sets were filtered by probe set enrichment p value of p < 0.05 (1,417 probe sets corresponding to 536 genes with the false-positive rate determined to be 0.02). The same probe sets were subsequently analyzed in the Opn4−/− data to assess attenuation. Refer to extended experimental procedures for hierarchical clustering, functional pathway analysis and TFBS prediction.

28For qPCR, primer sequences, western blotting, immunostaining, microscopy, CRTC phosphorylation assay, and I3M-based studies, see Supplemental Information.

29siRNAs were synthesized as siGENOME (Dharmacon) for in vitro experiments, siSTABLE (Dharmacon) or custom synthesized by Roche for in vivo work. The siRNAs supplied by Roche were moderately stabilized for in vivo use and tagged with a cholesterol conjugate to aid transfection. The sequences are as described in Supplemental Information.

30NIH 3T3 cells (mouse embryonic fibroblast cell line) were cultured in DMEM supplemented with 10% FBS at 37°C, 5% CO2. Where required, cells were transfected with 10 nM siRNA using Lipofectamine RNAimax (Invitrogen, Life Technologies) according to manufacturer’s instructions. The cells were maintained for at least 24 hr in 1% FBS in DMEM at confluence. At 0 min, the cells were synchronized by application of 50% horse serum returned to 1% FBS in DMEM after 30 min (serum shock treatment). RNA was isolated from cells at 30 min intervals over 3 hr. RNA was extracted using the RNeasy mini kit (QIAGEN) and qPCR was performed as discussed in supplementary material to measure knockdown/gene expression.

31siRNA complexed with Invivofectamine (Invitrogen) prepared according to manufacturer’s instructions were delivered using microinjection into the 3V with sterotaxic equipment with coordinates as described previously. Mice were allowed to recover and returned to the light-tight chambers for further experiments. For visualization of siRNA, 5′ Cy3-labeled nontargeting siRNA was injected into the 3V, and 24 hr later, the animals were sacrificed. Tissue processing and imaging were conducted as described in Supplemental Information. To measure silencing of Sik1, mice were injected with either siNT or siSik1 and 96 hr later were sacrificed, SCN collected on which RNA extraction and qPCR were conducted for the required mRNAs. To check for activation of an interferon response, a common off-target effect of siRNA, Oas1 mRNA levels from siRNA (siNT and siSik1)-injected mice were compared with noninjected controls (Figure S4). For phase-shifting studies, C57Bl/6 mice were maintained on running wheels in light tight chambers on a 12:12 hr LD cycle (400 lux from white LED lamps) and injected with siRNA as above. Four days after the injection, the mice were placed in DD for 24 hr and received a 30 min light pulse at approximately CT14.5. This time point was chosen to allow all experimental subjects to receive a light pulse within the maximal phase delay zone of CT14–16. The mice were then allowed to free run for 14 days in DD, running wheel activity data were collected and analyzed on Clocklab (Actimetrics, Wilmette, IL, USA). For jet-lag studies, C57Bl/6 mice were maintained on running wheels in light tight chambers on a 12:12 hr LD cycle (400 lux from white LED lamps) and injected with siRNA as above. Two days after injection, the LD cycle was shifted 6 hr in advance. Ten days later, the mice were shifted 6 hr in advance again. Onset of activity on each day was used to measure phase relative to the LD cycle.

32Animals were housed under a reversed 12:12 hr LD cycle for 2 weeks with food and water ad libitum. A light pulse (LP) of 400-500 lux was administered at CT16 for 30 min, using a SpectraNova halogen white light (Schott AG, Mainz, Germany). Animals were sacrificed under dim red light by cervical dislocation at 30, 60, and 120 min after the onset of the light pulse. The eyes were immediately removed to prevent any photic stimulation to the SCN. Sham-treated mice not given a light pulse were dissected at each time point. Brains were removed and placed into a brain matrix (Kent Scientific, Torrington CT, USA). A skin graft blade (Swann-Morton, Sheffield, UK) was positioned at Bregma −0.10 mm (Figure S1). A second blade was placed 1 mm (Bregma −1.10) caudal from the first, and a 1 mm thick brain slice was dissected. SCN punches were taken using a sample corer (1 mm internal diameter, Fine Science Tools GmbH, Heidelberg, Germany) from the brain slice (n = 4), flash frozen on dry ice and stored at −80°C prior to RNA extraction. All procedures were performed in accordance with the UK Home Office Animals (Scientific Procedures) Act 1986 and the University of Oxford’s Policy on the Use of Animals in Scientific Research.Total RNA was extracted using the microRNeasy column method (QIAGEN, Hilden, Germany). Quality and quantity of RNA were measured using an Agilent Bioanalyzer and a Nanodrop1000 (Thermo Fisher Scientific, Waltham, MA USA), respectively. To ensure that punches of the defined areas were accurately dissected, qPCR of the SCN-specific mRNA Six6 and Per1 were conducted, revealing selective enrichment (Figure S1).The transcripts found to be significantly altered following the light pulse in the Opn4+/+ animals were and those then downregulated in the Opn4−/− were examined using the DAVID ontology tools[REMOVED HYPERLINK FIELD]. 536 transcript cluster IDs from the Opn4+/+ corresponded to 527 DAVID identifiers and these were used to generate functional annotation charts using DAVID default ontologies for GO, pathways and protein domains (as of 28 Nov 2011). These annotation charts were visualized using the enrichment map plugin for Cytoscape.We used a comparative genomics approach based on multiple sequence alignments of genomic sequences from several mammalian species: human, dog, horse, cow, pig, rabbit, mouse and rat. TFBSs are represented by position-specific scoring matrices (PSSMs) as obtained from the TransFac database (version 11.4) and the JASPAR collection. We used the ProfileStats package to regularize the count matrices. The resulting scoring matrices were then used to scan genomic sequences (excluding known protein-coding stretches) from all species studied individually, and on both forward and reverse strands. We considered those matches that were conserved in at least two species.TFBSs are characterized by two parameters: an “expectation ratio” (ER), estimating the relative probability of finding a PSSM match of the given quality in a random sequence of 500 bp length and genome background base composition, and the number of allowed mismatching species in the alignment. Thus, predictions with a low ER and few or none mismatching species have the strongest support.RNA samples were prepared as described for microarray hybridization. cDNA was synthesized with a qScript cDNA synthesis kit (Quanta Biosciences, Gaithersburg, MD), and quantitative PCR (qPCR) was conducted with Sybr green I and an SDS7700 thermal cycler (Applied Biosystems, Foster City, CA). Relative quantification of transcript levels was done as described previously. The geometric mean of three housekeeping genes was used for normalization (Gapdh, Actb, and Arbp). Primer sequences are provided in Table S5.P0 neonatal mice(from C57 BL/6 and C3H backgrounds) were killed by decapitation and the brains dissected in to ice-cold PBS. Micro-dissection was carried out under a stereomicroscope to isolate the optic chiasm and the surrounding hypothalamus up to ∼1 mm caudal and ventral to the chiasm. This region should contain the SCN, but was not considered to yield purely SCN-specific cells. The tissue was then dissociated using a Papain dissociation system (Worthington Biochemical Corp, Lakewood, NJ, USA), using incubation times from 50–70 min. single-cell suspensions were generated using a series of sterile plastic Pasteur pipettes with decreasing bores (Appleton Woods, UK). Final cell densities of 105 cells/ml were prepared in DMEM as before (with the addition of B27 supplement, Invitrogen, Oregon, USA) and cells were plated on sterile, poly-L-lysine-treated glass slides, prepared using silicone chamber gaskets (CoverWell, Invitrogen, Oregon, USA) and allowed to recover for at least 48hr.Images are NIH 3T3 cells/dissociated SCN cell culture grown on glass slides using CoverWell perfusion chamber gaskets. Individual chambers on the same slide were changed to media containing 50% horse serum at 10 and 0 min prior to washing with PBS and fixation with PBS +4% PFA. Slides were processed using a monoclonal rabbit antibody to CRTC1 (C71D11, Cell Signaling Technology, Danvers, MA, USA) at a concentration of 1:100. Images were collected using a Zeiss LSM710 confocal microscope (Carl Zeiss Ltd., Cambridge, UK).NIH 3T3 cells (mouse embryonic fibroblast cell line) were cultured in DMEM supplemented with 10% FBS at 37°C, 5% CO2. Cells were cultured to confluence in 6 well plates, incubated in 1% FBS for at least 24 hr followed by application of 50% horse serum for 30 min (serum shock) to synchronize cellular rhythms and induce expression of Sik1 and Per1 as shown in Figure 3. Cells then were then washed with PBS lysed at the required time points in 200 μl kinase lysis buffer (25 mM Tris (pH 7.5), 150 mM NaCl, 50 mM NaF, 5% glycerol, 5 mM β-glycerol phosphate, 1 mM EDTA, 1 mM NaVO4, 1 mM DTT, complete mini protease inhibitor cocktail (Roche Diagnostics, Penzberg, Germany) and Phosphatase Inhibitor cocktail (Sigma – Aldrich, Dorset, England). 10 μg CRTC peptide (rlpsalnrts sdsalhtsvm, containing the SIK1 phosphorylation site of all three CRTC isoforms) was blotted onto circles of Whatmann Nitrocellulose 0.25um membrane (GE Healthcare, Waukesha, WI, USA) punched with a paper punch and allowed to dry. The membrane was then blocked with 10% BSA for 30 min. 35 μl lysate or 0.1 μg SIK1 (ab89693, Abcam, Cambridge, UK) or 0.1 μg SIK1 + drug (20 μM Indirubin-3-Monoxime or 100 nM Staurosporine) in 50 μl kinase buffer and 2 μCi (γ32P)ATP were added to the membrane in a 50 μl reaction and allowed to incubate at RT for 20 min. The membranes were then washed 5x with TBS – 0.2% Tween-20, rinsed with water and detected using a Typhoon Phosphor Imager System (GE Healthcare, Waukesha, WI, USA).Mouse embryonic fibroblasts (MEFs) derived from PER2::LUCIFERASE C57Bl/6 mice (PER2:LUC); were cultured in DMEM + 10% FBS + antibiotic (HyClone, Logan, UT). Confluent cells were maintained in DMEM + 1% FBS + antibiotics for 24–48 hr. Confluent cells were treated with I3M (20 μM) (Sigma-Aldrich, St. Louis, MO) with DMSO (0.06%) or DMSO alone (vehicle control) 30 min before time 0 min. I3M has been shown to inhibit SIK1 activity on TORC by preventing the phosphorylation and consequent activation of SIK1 and also directly inhibits SIK1 activity as shown by the CRTC phosphorylation assay as described above. The autonomous cellular rhythms were synchronized with a serum treatment. At 0 min, cells were treated with 50% horse serum (HS) plus I3M + DMSO or DMSO in DMEM, and RNA was isolated from cells at 30 min intervals over 4 hr. Treatment medium was maintained on the cells throughout the experiment. qRT-PCR using cDNA prepared from each time point was used to measure relative per1 and per2 mRNA levels and conducted as previously described using the standard curve method and normalized to GAPDH. per2 Taqman reagents (forward 5′-CCCATCCCACACTTGCCTC-3′, reverse 5′-CACTGTGCCAGCCGGG-3′; probe 6FAM-CCAGGCTGAGTTCCCTAGTCGGACCT-TAMRA).PER2::LUC MEFs were cultured in 35 mm Petri dishes in DMEM + 10% FBS + antibiotic (HyClone). Once confluent, the cells were maintained in DMEM + 1% FBS + antibiotics for 24 – 48 hr. The cellular rhythms were synchronized with a 30 min 50% HS treatment. Ten hours later, the cells were treated with a second 30 min 50% HS pulse with either I3M (20 μM) and DMSO (0.06%) or DMSO alone (vehicle control), which results in a delay shift of the rhythm of a defined magnitude. Experimental cells were pretreated 30 min prior with either I3M or DMSO alone. The cells were then placed in Leibovitz’s-15 media (GIBCO, Grand Island, NY) with 1% FBS and 0.1μM Luciferin (Gold Bio Technology, St. Louis, MO), dishes sealed and placed into a LumiCycle instrument (Actimetrics, Wilmette, IL). For the first 3 hr, media also contained I3M + DMSO or DMSO. Cultures were maintained at 37°C during media exchanges. Rhythmic expression of PER2::LUC activity was measured by luminometry. Peaks were determined using the Actimetrics LumiCycle Analysis software and determining the time after the start of the first serum pulse at which each peak reached its maximum. Phase differences between treatment groups were determined using one-way ANOVA with post hoc Tukey’s HSD tests (p < 0.05; n = 6-13 cultures per treatment group).Adult Per1:luciferase mice (Per1:luc, CD1 background; 3–4 months; n = 8) maintained on a 12:12 hr LD cycle were exposed to a 10 min light pulse (400 lux) at ZT14-14.5. Mice were immediately sacrificed and SCN tissue isolated and cultured as previously described. SCN were maintained in media containing 0.1 μM Luciferin and either 20 μM I3M + DMSO or DMSO alone. After 3 hr, the treatment media was replaced with fresh recording media, including 2x 15 min washes. Cultures were maintained at 37°C during media exchanges. Explant cultures were monitored for several days in the LumiCycle instrument, per1:luc expression monitored by luminescence, and peak phase within the first circadian cycle determined. Phase differences between treatment groups were determined using Student’s t test (p < 0.05). All procedures were approved by the University of Notre Dame Animal care and Use Committee and performed in accordance with NIH Guidelines for the Care and Use of Laboratory Animals.siRNA sequences are as follows; Sik1 (in vivo): 5′ GCUCAACCCUCCUUGCAUA 3′; Nontargeting siRNA: 5′ CUUACGCUGAGUACUUCGA 3′; Sik1 (2, in vitro): 5′ UGAACAAGAUCAAAGGGUU 3′; Sik1 (3, in vitro): 5′ AAACGCAGGUUGCAAUAAA 3′; Crtc1: 5′ UGGACAGAGUAUAUCGUGA 3′; Crtc3: 5′ GUUUCGAGCUGACCGGCUA 3′; Sik1 (human): 5′ ACGAUUAGAUUCAAGCAAU 3′; Cry1 (human): 5′ GGAGUUAAUUAUCCUAAAC 3′.As shown in Figure 4C, 2 μg SCN lysate in RIPA buffer were run on 4%–20% SDS-PAGE gels (Bio-Rad, Hercules, CA, USA), transferred using a semi-dry protocol on the Turbo Transfer system (Bio-Rad) onto PVDF membranes, blocked with 3% BSA, incubated with rabbit polyclonal antibodies to SIK1 (SIK1 (Y-20) sc-83754, Santa Cruz Biotechnology, Santa Cruz, CA, USA, 1:1000) and Beta-ACTIN (ab8227, Abcam, 1:5000), subsequently with a donkey anti-rabbit IgG HRP conjugated antibody (ab98503, Abcam, 1:5000) and developed with a standard ECL protocol (Luminata Classico, Millipore, Billerica, MA, USA).As shown in Figure 4E: 5 μg of SCN (from NT or Sik1 siRNA-injected animals) in RIPA buffer were run on 4%–20% SDS-PAGE gels (NuPAGE, Life Technologies), transferred using standard protocols (Bio-Rad) onto Immobilon FL PVDF membranes (Millipore), blocked with Odyssey blocking buffer (Li-Cor Biosciences, Lincoln, NE, USA) incubated with rabbit polyclonal antibodies to SIK1 (ab64428, Abcam, 1:500) and Beta-Actin (ab8227, Abcam, 1:1000), subsequently with donkey anti-rabbit IgG 488 Dylite conjugated secondary antibody (Pierce Biotechnology, Rockford, IL, USA, 1:200) and scanned with a Typhoon Scanner (GE Healthcare).U2OS cells stably transfected with a Per2::Luc reporter were cultured in DMEM supplemented with 10% FBS. For siRNA-based experiments, cells were seeded at 5,000 per well into white 384-well plates and transfected with 20 nM siRNA the next day. Two days later, the cells were synchronized with 100 nM dexamethasone and the medium replaced with phenol-red free DMEM supplemented with B-27 and 100 μM luciferin potassium salt and sealed. Per2::Luc rhythms were recorded from a BMG Labtech Fluostar Omega plate reader maintained at 36°C and readings taken from each well every hour. Data were then analyzed using BRASS (http://millar.bio.ed.ac.uk/PEBrown/BRASS/BrassPage.htm).

33Animals were housed under a reversed 12:12 hr LD cycle for 2 weeks with food and water ad libitum. A light pulse (LP) of 400-500 lux was administered at CT16 for 30 min, using a SpectraNova halogen white light (Schott AG, Mainz, Germany). Animals were sacrificed under dim red light by cervical dislocation at 30, 60, and 120 min after the onset of the light pulse. The eyes were immediately removed to prevent any photic stimulation to the SCN. Sham-treated mice not given a light pulse were dissected at each time point. Brains were removed and placed into a brain matrix (Kent Scientific, Torrington CT, USA). A skin graft blade (Swann-Morton, Sheffield, UK) was positioned at Bregma −0.10 mm (Figure S1). A second blade was placed 1 mm (Bregma −1.10) caudal from the first, and a 1 mm thick brain slice was dissected. SCN punches were taken using a sample corer (1 mm internal diameter, Fine Science Tools GmbH, Heidelberg, Germany) from the brain slice (n = 4), flash frozen on dry ice and stored at −80°C prior to RNA extraction. All procedures were performed in accordance with the UK Home Office Animals (Scientific Procedures) Act 1986 and the University of Oxford’s Policy on the Use of Animals in Scientific Research.

34Total RNA was extracted using the microRNeasy column method (QIAGEN, Hilden, Germany). Quality and quantity of RNA were measured using an Agilent Bioanalyzer and a Nanodrop1000 (Thermo Fisher Scientific, Waltham, MA USA), respectively. To ensure that punches of the defined areas were accurately dissected, qPCR of the SCN-specific mRNA Six6 and Per1 were conducted, revealing selective enrichment (Figure S1).

35The transcripts found to be significantly altered following the light pulse in the Opn4+/+ animals were and those then downregulated in the Opn4−/− were examined using the DAVID ontology tools[REMOVED HYPERLINK FIELD]. 536 transcript cluster IDs from the Opn4+/+ corresponded to 527 DAVID identifiers and these were used to generate functional annotation charts using DAVID default ontologies for GO, pathways and protein domains (as of 28 Nov 2011). These annotation charts were visualized using the enrichment map plugin for Cytoscape.

36We used a comparative genomics approach based on multiple sequence alignments of genomic sequences from several mammalian species: human, dog, horse, cow, pig, rabbit, mouse and rat. TFBSs are represented by position-specific scoring matrices (PSSMs) as obtained from the TransFac database (version 11.4) and the JASPAR collection. We used the ProfileStats package to regularize the count matrices. The resulting scoring matrices were then used to scan genomic sequences (excluding known protein-coding stretches) from all species studied individually, and on both forward and reverse strands. We considered those matches that were conserved in at least two species.

37TFBSs are characterized by two parameters: an “expectation ratio” (ER), estimating the relative probability of finding a PSSM match of the given quality in a random sequence of 500 bp length and genome background base composition, and the number of allowed mismatching species in the alignment. Thus, predictions with a low ER and few or none mismatching species have the strongest support.

38RNA samples were prepared as described for microarray hybridization. cDNA was synthesized with a qScript cDNA synthesis kit (Quanta Biosciences, Gaithersburg, MD), and quantitative PCR (qPCR) was conducted with Sybr green I and an SDS7700 thermal cycler (Applied Biosystems, Foster City, CA). Relative quantification of transcript levels was done as described previously. The geometric mean of three housekeeping genes was used for normalization (Gapdh, Actb, and Arbp). Primer sequences are provided in Table S5.

39P0 neonatal mice(from C57 BL/6 and C3H backgrounds) were killed by decapitation and the brains dissected in to ice-cold PBS. Micro-dissection was carried out under a stereomicroscope to isolate the optic chiasm and the surrounding hypothalamus up to ∼1 mm caudal and ventral to the chiasm. This region should contain the SCN, but was not considered to yield purely SCN-specific cells. The tissue was then dissociated using a Papain dissociation system (Worthington Biochemical Corp, Lakewood, NJ, USA), using incubation times from 50–70 min. single-cell suspensions were generated using a series of sterile plastic Pasteur pipettes with decreasing bores (Appleton Woods, UK). Final cell densities of 105 cells/ml were prepared in DMEM as before (with the addition of B27 supplement, Invitrogen, Oregon, USA) and cells were plated on sterile, poly-L-lysine-treated glass slides, prepared using silicone chamber gaskets (CoverWell, Invitrogen, Oregon, USA) and allowed to recover for at least 48hr.

40Images are NIH 3T3 cells/dissociated SCN cell culture grown on glass slides using CoverWell perfusion chamber gaskets. Individual chambers on the same slide were changed to media containing 50% horse serum at 10 and 0 min prior to washing with PBS and fixation with PBS +4% PFA. Slides were processed using a monoclonal rabbit antibody to CRTC1 (C71D11, Cell Signaling Technology, Danvers, MA, USA) at a concentration of 1:100. Images were collected using a Zeiss LSM710 confocal microscope (Carl Zeiss Ltd., Cambridge, UK).

41NIH 3T3 cells (mouse embryonic fibroblast cell line) were cultured in DMEM supplemented with 10% FBS at 37°C, 5% CO2. Cells were cultured to confluence in 6 well plates, incubated in 1% FBS for at least 24 hr followed by application of 50% horse serum for 30 min (serum shock) to synchronize cellular rhythms and induce expression of Sik1 and Per1 as shown in Figure 3. Cells then were then washed with PBS lysed at the required time points in 200 μl kinase lysis buffer (25 mM Tris (pH 7.5), 150 mM NaCl, 50 mM NaF, 5% glycerol, 5 mM β-glycerol phosphate, 1 mM EDTA, 1 mM NaVO4, 1 mM DTT, complete mini protease inhibitor cocktail (Roche Diagnostics, Penzberg, Germany) and Phosphatase Inhibitor cocktail (Sigma – Aldrich, Dorset, England). 10 μg CRTC peptide (rlpsalnrts sdsalhtsvm, containing the SIK1 phosphorylation site of all three CRTC isoforms) was blotted onto circles of Whatmann Nitrocellulose 0.25um membrane (GE Healthcare, Waukesha, WI, USA) punched with a paper punch and allowed to dry. The membrane was then blocked with 10% BSA for 30 min. 35 μl lysate or 0.1 μg SIK1 (ab89693, Abcam, Cambridge, UK) or 0.1 μg SIK1 + drug (20 μM Indirubin-3-Monoxime or 100 nM Staurosporine) in 50 μl kinase buffer and 2 μCi (γ32P)ATP were added to the membrane in a 50 μl reaction and allowed to incubate at RT for 20 min. The membranes were then washed 5x with TBS – 0.2% Tween-20, rinsed with water and detected using a Typhoon Phosphor Imager System (GE Healthcare, Waukesha, WI, USA).

42Mouse embryonic fibroblasts (MEFs) derived from PER2::LUCIFERASE C57Bl/6 mice (PER2:LUC); were cultured in DMEM + 10% FBS + antibiotic (HyClone, Logan, UT). Confluent cells were maintained in DMEM + 1% FBS + antibiotics for 24–48 hr. Confluent cells were treated with I3M (20 μM) (Sigma-Aldrich, St. Louis, MO) with DMSO (0.06%) or DMSO alone (vehicle control) 30 min before time 0 min. I3M has been shown to inhibit SIK1 activity on TORC by preventing the phosphorylation and consequent activation of SIK1 and also directly inhibits SIK1 activity as shown by the CRTC phosphorylation assay as described above. The autonomous cellular rhythms were synchronized with a serum treatment. At 0 min, cells were treated with 50% horse serum (HS) plus I3M + DMSO or DMSO in DMEM, and RNA was isolated from cells at 30 min intervals over 4 hr. Treatment medium was maintained on the cells throughout the experiment. qRT-PCR using cDNA prepared from each time point was used to measure relative per1 and per2 mRNA levels and conducted as previously described using the standard curve method and normalized to GAPDH. per2 Taqman reagents (forward 5′-CCCATCCCACACTTGCCTC-3′, reverse 5′-CACTGTGCCAGCCGGG-3′; probe 6FAM-CCAGGCTGAGTTCCCTAGTCGGACCT-TAMRA).

43PER2::LUC MEFs were cultured in 35 mm Petri dishes in DMEM + 10% FBS + antibiotic (HyClone). Once confluent, the cells were maintained in DMEM + 1% FBS + antibiotics for 24 – 48 hr. The cellular rhythms were synchronized with a 30 min 50% HS treatment. Ten hours later, the cells were treated with a second 30 min 50% HS pulse with either I3M (20 μM) and DMSO (0.06%) or DMSO alone (vehicle control), which results in a delay shift of the rhythm of a defined magnitude. Experimental cells were pretreated 30 min prior with either I3M or DMSO alone. The cells were then placed in Leibovitz’s-15 media (GIBCO, Grand Island, NY) with 1% FBS and 0.1μM Luciferin (Gold Bio Technology, St. Louis, MO), dishes sealed and placed into a LumiCycle instrument (Actimetrics, Wilmette, IL). For the first 3 hr, media also contained I3M + DMSO or DMSO. Cultures were maintained at 37°C during media exchanges. Rhythmic expression of PER2::LUC activity was measured by luminometry. Peaks were determined using the Actimetrics LumiCycle Analysis software and determining the time after the start of the first serum pulse at which each peak reached its maximum. Phase differences between treatment groups were determined using one-way ANOVA with post hoc Tukey’s HSD tests (p < 0.05; n = 6-13 cultures per treatment group).

44Adult Per1:luciferase mice (Per1:luc, CD1 background; 3–4 months; n = 8) maintained on a 12:12 hr LD cycle were exposed to a 10 min light pulse (400 lux) at ZT14-14.5. Mice were immediately sacrificed and SCN tissue isolated and cultured as previously described. SCN were maintained in media containing 0.1 μM Luciferin and either 20 μM I3M + DMSO or DMSO alone. After 3 hr, the treatment media was replaced with fresh recording media, including 2x 15 min washes. Cultures were maintained at 37°C during media exchanges. Explant cultures were monitored for several days in the LumiCycle instrument, per1:luc expression monitored by luminescence, and peak phase within the first circadian cycle determined. Phase differences between treatment groups were determined using Student’s t test (p < 0.05). All procedures were approved by the University of Notre Dame Animal care and Use Committee and performed in accordance with NIH Guidelines for the Care and Use of Laboratory Animals.

45siRNA sequences are as follows; Sik1 (in vivo): 5′ GCUCAACCCUCCUUGCAUA 3′; Nontargeting siRNA: 5′ CUUACGCUGAGUACUUCGA 3′; Sik1 (2, in vitro): 5′ UGAACAAGAUCAAAGGGUU 3′; Sik1 (3, in vitro): 5′ AAACGCAGGUUGCAAUAAA 3′; Crtc1: 5′ UGGACAGAGUAUAUCGUGA 3′; Crtc3: 5′ GUUUCGAGCUGACCGGCUA 3′; Sik1 (human): 5′ ACGAUUAGAUUCAAGCAAU 3′; Cry1 (human): 5′ GGAGUUAAUUAUCCUAAAC 3′.

46As shown in Figure 4C, 2 μg SCN lysate in RIPA buffer were run on 4%–20% SDS-PAGE gels (Bio-Rad, Hercules, CA, USA), transferred using a semi-dry protocol on the Turbo Transfer system (Bio-Rad) onto PVDF membranes, blocked with 3% BSA, incubated with rabbit polyclonal antibodies to SIK1 (SIK1 (Y-20) sc-83754, Santa Cruz Biotechnology, Santa Cruz, CA, USA, 1:1000) and Beta-ACTIN (ab8227, Abcam, 1:5000), subsequently with a donkey anti-rabbit IgG HRP conjugated antibody (ab98503, Abcam, 1:5000) and developed with a standard ECL protocol (Luminata Classico, Millipore, Billerica, MA, USA).

47As shown in Figure 4E: 5 μg of SCN (from NT or Sik1 siRNA-injected animals) in RIPA buffer were run on 4%–20% SDS-PAGE gels (NuPAGE, Life Technologies), transferred using standard protocols (Bio-Rad) onto Immobilon FL PVDF membranes (Millipore), blocked with Odyssey blocking buffer (Li-Cor Biosciences, Lincoln, NE, USA) incubated with rabbit polyclonal antibodies to SIK1 (ab64428, Abcam, 1:500) and Beta-Actin (ab8227, Abcam, 1:1000), subsequently with donkey anti-rabbit IgG 488 Dylite conjugated secondary antibody (Pierce Biotechnology, Rockford, IL, USA, 1:200) and scanned with a Typhoon Scanner (GE Healthcare).

48U2OS cells stably transfected with a Per2::Luc reporter were cultured in DMEM supplemented with 10% FBS. For siRNA-based experiments, cells were seeded at 5,000 per well into white 384-well plates and transfected with 20 nM siRNA the next day. Two days later, the cells were synchronized with 100 nM dexamethasone and the medium replaced with phenol-red free DMEM supplemented with B-27 and 100 μM luciferin potassium salt and sealed. Per2::Luc rhythms were recorded from a BMG Labtech Fluostar Omega plate reader maintained at 36°C and readings taken from each well every hour. Data were then analyzed using BRASS (http://millar.bio.ed.ac.uk/PEBrown/BRASS/BrassPage.htm).