The CRTC1-SIK1 pathway regulates entrainment of the circadian clock

Abstract

Retinal photoreceptors entrain the circadian system to the solar day. This photic resetting involves cAMP response element binding protein (CREB)-mediated upregulation of Per genes within individual cells of the suprachiasmatic nuclei (SCN). Our detailed understanding of this pathway is poor, and it remains unclear why entrainment to a new time zone takes several days. By analyzing the light-regulated transcriptome of the SCN, we have identified a key role for salt inducible kinase 1 (SIK1) and CREB-regulated transcription coactivator 1 (CRTC1) in clock re-setting. An entrainment stimulus causes CRTC1 to coactivate CREB, inducing the expression of Per1 and Sik1. SIK1 then inhibits further shifts of the clock by phosphorylation and deactivation of CRTC1. Knockdown of Sik1 within the SCN results in increased behavioral phase shifts and rapid re-entrainment following experimental jet lag. Thus SIK1 provides negative feedback, acting to suppress the effects of light on the clock. This pathway provides a potential target for the regulation of circadian rhythms.

Graphical abstract

Figure 1

Analysis of Light-Regulated Gene Expression in the SCN (A) Hierarchical clustering of the wild-type (Opn4^+/+) SCN transcriptome. Sham = SCN from sham-pulsed mice; LP30, LP60, and LP120 = SCN from 30, 60, and 120 min after light pulse, respectively (n = 4). Two gene clusters are enlarged: The upper cluster shows acute upregulation and contains immediate early genes (Fos, Egr1, Nr4a1) and clock genes (Per1); the lower cluster shows downregulated transcripts including transcription factors (Sim1 and Nr2f2), signaling peptides (Tac1) and receptors (Glra1 and Glra2). Scale bar shows fold change; yellow = upregulation, blue = downregulation. Numbers beneath scale bar show levels of expression relative to Sham. (B) Functional annotation of the 536 genes changing in response to light using DAVID and visualized using the enrichment map plugin for Cytoscape. (C) Enrichment of the CREB-activated promoter CRE in the promoters of 108 genes showing >1.5-fold change in response to light in the Opn4^+/+ SCN. Data are shown as observed/expected frequency of appearance of promoter elements (CRE_TATA: CRE within 300 bp of the TATA box, CRE_NoTATA: CRE site further upstream, others: absence of CRE sites) within the 108 genes. CRE_TATA shows significant enrichment (^∗p = 0.02, Chi-square test), highlighting the role of CREB in high amplitude gene induction responses to light. See also Figures S1 and S2, and Tables S1, S2, S3, and S4.

Figure S1

SCN Sample Collection, Gene Expression Arrays, and Analysis, Related to Figure 1 and Experimental Procedures (A) Schematic of SCN sample collection: Mice housed under 12:12 hr LD were released into one day of darkness and subjected to a 30 min light pulse (400 lux) at CT16. Animals were sacrificed at the time points as indicated under dim red light; LP30, LP60, and LP120 = SCN from 30, 60, and 120 min, respectively following the start of a 30 min light pulse at CT16. Time matched sham-pulsed controls were collected at each time point and pooled. Brains were collected, sections were taken by inserting blades as indicated and SCN punches were cored and stored frozen until use. (B) Real time PCR conducted on SCN or control (SCN CT) for (B) Six6 and Per1 normalized to three housekeeping genes. (C) RNA quality for SCN samples used for microarrays, Affymetrix sample data provided for reference. (D) Schematic of data analysis from Affymetrix Exon arrays; 48% of core probe sets were detected as reliably present above background in the SCN (DABG p < 0.05, 110,026 probe sets out of a total of 228,090). Of these, 4.1% were found to significantly change in expression in response to light (4,489 probe sets corresponding to 3,187 genes p < 0.05, one-way ANOVA). Of these probe sets, 1,470 were found to be significantly enriched, that is, significantly more probe sets per gene than would be expected by chance alone. These probe sets correspond to 536 genes. Error bars = SEM.

Figure S2

Light-Regulated Gene Expression in the Opn4^−/− SCN, Related to Figure 1 (A) Scatter plot comparing fold change in the light-regulated SCN transcriptome in Opn4^+/+ and Opn4^−/− mice showing attenuated regulation in the melanopsin deficient animals. Upregulation in response to light = yellow; downregulation = blue. Fos, Egr1, Nr4a1, Per1, Sim1, and Tac1 highlighted. (B) Heat map comparison of a section of the light-regulated SCN transcriptome of Opn4^+/+ (+/+) and Opn4^−/− (−/−) mice. Scale shows downregulation in blue (up to 0-fold) and upregulation in yellow (2-fold). (C) Temporal expression profiles of Fos, Per1 and Egr1 from SCN of Opn4^+/+ (black) and Opn4^−/− (red) mice from array (upper plots) and qPCR (lower plots). Error bars = SEM.

Figure 2

The Role of Sik1 in Phase-Shifting Responses in NIH 3T3 Embryonic Fibroblasts (A) CRTC1 shows nuclear translocation after serum shock (before serum, top; after 10 min serum, bottom). (B and C) Bioinformatic analysis of the Sik1 promoter region showing conservation of the CRE across several mammalian species (B, top). Sik1 was induced after serum shock (B, bottom, n = 4), leading to (C) increased phosphorylation of CRTC. A CRTC peptide is phosphorylated by purified SIK1, measured by the incorporation of ³²P from γ³²P ATP (+SIK1). When incubated with cell lysates (lysate, 0 min), serum shock (lysate, 120 min) increased significantly phosphorylation of CRTC (p = 0.006, Student’s t test, n = 8). Experiments containing no lysate are shown as a negative control (− lysate). Representative blots are shown (left). (D) Sik1 induction following serum shock was attenuated significantly following silencing of Sik1 or Crtc compared with nontargeting control siRNA (siNT: p = 1.5 × 10⁻⁶, siSik1: p = 4.1 × 10⁻¹, siCrtc: p = 0.01, n = 4, area under curve [AUC] analysis shown in Figure S3). (E) Sik1 silencing enhanced Per1 induction following serum shock (siNT: p = 2.2 × 10⁻⁵, siSik1: p = 9.3 × 10⁻⁵). (F) Following Crtc silencing, no significant induction of Per1 expression was observed (siCrtc: p = 0.06). AUC analysis (normalized to NT control) for Per1 expression following silencing of Sik1 showed a significant increase (p = 0.00047 siSik1 versus siNT, Student’s t test), not seen with Crtc silencing (p = 0.799 siCRTC versus siNT, Students’ t test). (G) Normalised data for Per1 induction following Sik1 and Crtc knockdown as measured by Area Under Curve (AUC) from (E) and (F). (H and I) Egr1 (H) and (I) Nr4a1, both CRE-regulated transcripts, show enhanced expression with Sik1 silencing and attenuated or no induction with Crtc silencing. AUC analysis shown in Figure S3. All mRNA levels normalized to housekeeping controls and t = 0, change over time analyzed by one-way ANOVA, ^∗∗∗ = p < 0.001, ^∗ = p < 0.05, n.s. = p > 0.05. Error bars = SEM; n = 4. See also Figure S3 and Tables 1 and 2.

Figure S3

RNAi to Validate the Role of SIK1 in Regulating Phase Shifting of the Clock in NIH 3T3 Fibroblasts, Related to Figure 2 (A) Relative expression in NIH 3T3 fibroblasts of Sik1 after transfection siSik1-1 (red) or a nontargeting control siRNA (siNT; white). siSik1-1 is modified (siStable formulation) for in vivo use, resulting in marginally lower efficiency compared with unmodified siRNA. (B) Relative expression of Crtc1 (light green), Crtc2 (green) and Crtc3 (dark green) mRNA in NIH 3T3 fibroblasts. (C) Relative expression levels to measure silencing of Crtc1 (light green) and Crtc3 (dark green) after transfection of siRNA against Crtc1 (siCrtc1) and Crtc3 (siCrtc3) respectively. (D) Relative expression levels of Sik1 (red) Crtc1 (green) and Crtc3 (blue) after transfection with nontargeting control (siNT); Sik1 siRNA (siSik1) or Crtc1 and Crtc3 siRNAs (siCrtc). (E) Level of CRTC1 phosphorylation, indicated by incorporation of ³²P from ³²P ATP by NIH 3T3 cell lysates. Cells treated with nontargeting siRNA (siNT) show significant increase in CRTC1 phosphorylation after serum shock (siNT 120 min) when compared with before (siNT 0 min). Cells treated with Sik1 siRNA show no increase in CRTC1 phosphorylation after serum shock (siSik1 120 min) when compared with before (siSik1 0 min). (F–I) Areas under the curve from traces in Figure 2H (Egr1 - F), 2I (Nr4a1- G) and 2D (Sik1- H) showing increased expression of Egr1 (siSik1 versus siNT: p = 0.0006) and Nr4a1 (siSik1 versus siNT: p = 0.046) and decreased expression of Sik1 (siSik1 versus siNT: 0.008) with Sik1 siRNA (siSik1). Crtc siRNA (siCrtc) induced reduced or unchanged expression of all three transcripts as indicated by the area under the curve (siCrtc versus siNT: p = 0.38 for Egr1, p = 0.07 for Nr4a1, p = 7.7 × 10⁻⁷ for Sik1) All p values derived from Students’ t test, error bars = SEM and n = 4. See Table 1 and Table 2 for results of individual comparisons and statistical tests in Figures 2D–2I and S3F–S3H. In order to ensure the increases of Per1, Egr1 and Nr4a1 induction seen after Sik1 silencing are not due to off-target effects, we confirmed the expression patterns with two separate siRNA sequences against Sik1 (siSik1-2 and siSik1-3) (I). (J–L) Per1 (J) Nr4a1 (K) and Egr1 (L) in NIH 3T3 fibroblasts treated with either nontargeting control siRNA (siNT; blue); siRNA sequence 2 against Sik1 (siSik1-2; red) or siRNA sequence 3 against Sik1 (siSik1-3; orange) all show increased induction following a serum shock after silencing of Sik1. Error bars = SEM, n = 4.

Figure 3

Pharmacological Inhibition of SIK1 by I3M Causes Enhanced Phase Shifting (A) Incorporation of ³²P from ³²P ATP into CRTC peptide by purified SIK1, SIK1 + 20 μM I3M (SIK1+I3M) and SIK1 + 100 nM Staurosporine (SIK1+Stau, staurosporine being a broad-spectrum kinase inhibitor); I3M directly inhibits SIK1. (B and C) Period1 and 2 induction following 30 min serum treatment in PER2::LUC MEFs treated with indirubin-3′-monoxime (I3M) in DMSO or DMSO alone for the duration of the experiment. Relative gene expression of (B) per1 and (C) per2 normalized to GAPDH. (D) Representative baseline detrended bioluminescence recordings from PER2::LUC MEFS treated with a single (orange, first red arrow) or second serum shock 10 hr later (second red arrow) in the presence of I3M (red) or DMSO alone (green), with I3M creating a larger phase shift than the DMSO-treated controls. Timing of I3M or DMSO treatment is shown by the horizontal gray bar. The first serum pulse is given to synchronize the cells and the second allows for phase shift from synchronized conditions, allowing precise quantification of the phase shift. (E) Time of peak bioluminescence determined from data as shown in (C). Data are mean ± SEM first and second peaks after the second serum pulse. Cells treated with I3M peak significantly later than DMSO-treated controls (one-way ANOVA; p < 0.01, with post hoc t tests, ^∗p < 0.05). (F) Zeitgeber time of second peak of bioluminescence from SCN collected from light-pulsed per1::luc mice treated with I3M versus DMSO-treated (DMSO) controls (^∗t test, p < 0.05). Error bars = SEM.

Figure 4

Induction of the CRTC1-SIK1 Pathway in the SCN in Response to Phase-Shifting Stimuli (A) Expression data from exon arrays indicating relative levels of Crtc isoforms in the SCN, showing Crtc1 is the most abundant isoform. (B) CRTC1 (green) translocates from the cytoplasm (upper left, 0 min) to the nucleus (upper right, 10 min) 10 min after the application of 50% horse serum to dissociated SCN cell cultures. Merged figures with DAPI staining (blue, nuclei) and phalloidin (white, cytoskeleton) are indicated in the respective lower panels. (C) SIK1 protein is increased in the SCN 120 min after a CT16 light pulse, representative blots shown. β-actin levels shown for comparison. (D) Sik1 was upregulated by light in the SCN of animals after a CT16 phase-delaying and CT22 phase-advancing light pulse, as measured by qPCR (CT16: p = 0.0002 ^∗∗∗, n = 5, CT22: p = 0.009 ^∗∗, n = 3). Egr1 and Nr4a1 levels are shown for comparison, where in contrast, their induction is lower at CT22 than at CT16. Sham = SCN from sham-pulsed mice; LP30, LP60, and LP120 = SCN from 30, 60, and 120 min after light pulse, respectively. Error bars = SEM.

Figure S4

In Vivo Knockdown of Sik1, Related to Figure 5 (A) Cy3-labeled siRNA (red) was delivered via intracerebellar ventricular injection (ICV) into the third ventricle (3V) above the SCN (left panel). siRNA was subsequently localized in SCN neurons (right panel). DAPI staining for nuclei is shown in blue. (B) Sik1 mRNA (left panel) and protein levels (right panel) were attenuated in the SCN after ICV injection of Sik1 siRNA (siSik1) when compared to nontargeting siRNA (siNT) (54% knockdown of mRNA, p = 0.03 and 49% protein, p = 0.02, Student’s t test, n = 4). Representative blots are shown with beta-ACTIN levels for comparison. (C) Oas1 (induction indicates interferon response) levels in WT SCN from either mice injected in the 3V with Sik1 siRNA (siSik1) or NT siRNA (siNT), (siRNA-injected SCN) versus noninjected controls (noninjected SCN), n = 5. Error bars = SEM.

Figure 5

In Vivo Knockdown of Sik1 Results in Enhanced Phase Shifting and Rapid Re-Entrainment in a Jet-Lag Protocol (A) C57Bl/6 mice were housed under a 12:12 hr LD cycle before siRNA ICV injection into the 3V (indicated by red asterisk on actograms). Ninety-six hours postinjection, the mice were given a 30 min light pulse (red arrow) at CT14.5, then placed in DD to enable phase-shift magnitude to be determined. Representative actograms from light-pulsed animals receiving siNT (siNT Light, top) or siSik1 (siSik1 Light, bottom) shown. Actograms are enlarged around the day of the light pulse for clarity (right). (B) Phase-shifting responses to light are significantly larger following knockdown of Sik1 in the SCN (siSik1) when compared to nontargeting siRNA (siNT) controls (98 min versus 59 min, p = 0.036, Student’s t test, n = 5). Error bars = SEM. See also Figure S4. (C) C57Bl/6 mice were housed under a 12:12 hr LD cycle before siRNA ICV injection into the 3V (indicated by red asterisk). Fourty-eight hours postinjection, the LD cycle was advanced by 6 hr and 10 days after the first shift, the LD cycle was advanced 6 hr again. Faster re-entrainment was observed with the Sik1 knockdown (siSik1) mice. Three actograms are displayed for each treatment (siNT or siSik1), showing that regardless of activity levels, Sik1 knockdown accelerates re-entrainment. (D) Phase relative to new LD cycle (second shift) plotted against days after the shift in cycle. Day before shift indicated as 0 (before the dotted line). ^∗∗ = p < 0.01, ^∗ = p < 0.05 Student’s t test, phase of siSik1 versus siNT-treated animals on each day, n = 6 for siNT, n = 11 for siSik1. Representative actograms enlarged around the area plotted in the graph are indicated. Error bars = SEM. See also Figures S4 and S5.

Figure S5

Effect of Sik1 Knockdown on Period Length, Related to Figure 5 (A) Representative traces from Per2::Luc U2OS cells transfected with siSik1, siNT or siCry1. siSik1 (inducing > 90% knockdown of Sik1) induces period lengthening while siCry1 (>90% knockdown of Cry1) induces period shortening. Data summated in (B) ^∗∗∗ = p < 0.001, Student’s t test, n = 8. (C) Representative actograms from C57Bl/6 mice injected with either Sik1 siRNA (siSik1, 50% knockdown of Sik1) or nontargeting siRNA (siNT). 4 days after injection of siRNA into the 3V the animals were maintained in constant dark in order for their free-running period to be determined. (D) Mice treated with siSik1 show greater period length than control mice injected with siNT (p = 0.016, Student’s t test, n = 11, error bars = SEM).