Loss of the conserved PKA sites of SIK1 and SIK2 increases sleep need

Although sleep is one of the most conserved behaviors, the intracellular mechanism regulating sleep/wakefulness remains unknown. We recently identified a protein kinase, SIK3, as a sleep-regulating molecule. Mice that lack a well-conserved protein kinase A (PKA) phosphorylation site, S551, showed longer non-rapid eye movement (NREM) sleep and increased NREMS delta density. S551 of SIK3 is conserved in other members of the SIK family, such as SIK1 (S577) and SIK2 (S587). Here, we examined whether the PKA phosphorylation sites of SIK1 and SIK2 are involved in sleep regulation by generating Sik1S577A and Sik2S587A mice. The homozygous Sik1S577A mice showed a shorter wake time, longer NREMS time, and higher NREMS delta density than the wild-type mice. The heterozygous and homozygous Sik2S587A mice showed increased NREMS delta density. Both the Sik1S577A and Sik2S587A mice exhibited proper homeostatic regulation of sleep need after sleep deprivation. Despite abundant expression of Sik1 in the suprachiasmatic nucleus, the Sik1S577A mice showed normal circadian behavior. Although Sik2 is highly expressed in brown adipose tissue, the male and female Sik2S587A mice that were fed either a chow or high-fat diet showed similar weight gain as the wild-type littermates. These results suggest that PKA-SIK signaling is involved in the regulation of sleep need.

. SIK family mRNA expression and mutant proteins. (a) Scheme of SIK1, SIK2, and SIK3. The serine residue in the PKA consensus sequence is conserved among the family. Although there are multiple protein isoforms of SIK3, this scheme shows the longest isoform. (b) Digital PCR results of Sik1, Sik2 and Sik3 mRNA of the cerebral cortex, hippocampus, hypothalamus, liver and brown adipose tissue (BAT) of the wild-type mice (n = 4). Each sample was measured in duplicate. One-way analysis of variance followed by Tukey's test. (c-e) In situ hybridization of Sik1 and Sik2. (c) Sik1 mRNA was strongly expressed in the suprachiasmatic nucleus (SCN) and broadly expressed in the forebrain. Scale bar, 500 μm. (d) In situ hybridization showed that Sik2 mRNA was broadly expressed in the forebrain. Scale bar, 500 μm. (e) Sik1 and Sik2 were expressed in the hippocampal dentate gyrus of the wild-type mice (upper and middle). Sik2 expression was not detected in the dentate gyrus of the Sik2-deficient mice (lower). Scale bars, 100 μm. (f-h) Sik1 (f), Sik2 (g) and Sik3 (h) mRNA expression in the cerebral cortex, hypothalamus, BAT, liver and adrenal gland after one week of high-salt diet feeding. Twotailed t-test with Bonferroni correction. (i) SIK2 protein was expressed in the BAT of the Sik2 +/+ and Sik2 S587A/ S587A mice. SIK2 was not detected in the BAT of the Sik2-deficient mice. (j) FLAG-SIK1 WT, FLAG-SIK1 S577A, FLAG-SIK2 WT, FLAG-SIK2 S587A, FLAG-SIK3 WT, and FLAG-SIK3 S551A transiently expressed in HEK293 cells were immunoprecipitated with an anti-FLAG antibody and then subjected to immunoblotting using www.nature.com/scientificreports www.nature.com/scientificreports/ wild-type mice 17,18 . Sik2 is highly expressed in brown adipose tissue (BAT) 8,19 , which is a specialized thermogenic organ 20 . Whereas the Sik2-deficient mice showed body weights similar to those of wild-type mice 21 , mice that express SIK2 with the serine-to-alanine substitution at the 587th residue specifically in BAT were susceptible to diet-induced obesity 22 . Given that mice lacking S551 of SIK3 showed normal circadian rhythm 1 , despite altered circadian behavior of the Sik3-deficient mice 23 , Sik1 S577A and Sik2 S587A mice may show metabolic and circadian phenotypes that are different from those in the Sik1-or Sik2-deficient mice.
To examine whether the phosphorylation of S577 of SIK1 and S587 of SIK2 is required for proper sleep/ wake behavior, we generated mutant mice in which S577 of SIK1 and S587 of SIK2 were substituted with alanine residues. Both the Sik1 and Sik2 mutant mice showed an increased NREMS delta, an indicator of sleep need. Consistent with the lower expression of Sik1 and Sik2 in the brain compared with Sik3, the sleep phenotypes of the Sik1 S577A mice and the Sik2 S587A mice were milder than those of the Sik3 mutant mice. The Sik1 S577A mice showed normal circadian behavior and re-entrainment to a new circadian rhythm. Additionally, the male and female Sik1 S577A mice showed similar body weights as the wild-type littermates, and the male and female Sik2 S587A mice fed either a chow or high-fat diet showed a similar body weight gain as the wild-type littermates. Thus, the conserved PKA sites of SIK1 and SIK2 are thought to be required for the proper regulation of sleep need and play a minor role in circadian and body weight regulation.

Results
Sik mRNA expression in the brain and other tissues. First, we examined the mRNA levels of Sik1, 2, and 3 in the cerebral cortex, hippocampus, hypothalamus, liver, and BAT. Sik3 was the most abundant Sik member in the brain (Fig. 1b). Sik1 mRNA was highly expressed in the SCN and broadly expressed in the cerebral cortex, hippocampus, thalamus, hypothalamus and brain stem (Fig. 1c,e). Sik2 was highly abundant in the BAT as previously reported 8,19 (Fig. 1b) and broadly expressed in the cerebral cortex, hippocampus, thalamus, and hypothalamus (Fig. 1d,e), consistent with a previous report 24 , while there was no expression in the Sik2-deficient mice (Fig. 1e).
Since SIK1 was originally identified as a molecule induced by a high-salt diet in the rat adrenal gland 7 , we examined whether a high-salt diet affects Sik1, 2 and 3 expression in the brain, BAT, liver and adrenal gland. One week of a high-salt diet did not affect the Sik1 mRNA expression in the cerebral cortex, hypothalamus, BAT, liver or adrenal gland (Fig. 1f). A high-salt diet increased the Sik2 mRNA expression in the liver and adrenal gland (Fig. 1g) and did not cause significant changes in the Sik3 mRNA expression in all tissues examined (Fig. 1h). We also confirmed the SIK2 protein expression in the BAT (Fig. 1i, S1a,c) and the brain (Fig. S1b,d).
The SIK1 S577A and SIK2 S587A proteins did not bind to 14-3-3. For characterization of the SIK1 S577A and SIK2 S587A proteins, FLAG-tagged SIK protein variants were transiently expressed in HEK293 cells. Since SIKs have a RRAS motif, a consensus sequence for PKA, immunopurified FLAG-tagged SIK1, 2, and 3 were detected with an anti-phospho-PKA substrate motif antibody (Fig. 1j). In contrast, the FLAG-tagged SIK1 S577A, SIK2 S587A, and SIK3 S551A showed only faint immunoreactivity to the anti-phospho-PKA substrate motif antibody 2,10 . The 14-3-3 protein was purified with the FLAG-tagged SIK1, 2, and 3 but not with the FLAG-tagged SIK1 S577A, SIK2 S587A and SIK3 S551A (Fig. 1j), as previously reported 2,10,11 . the Sik1 S577A mice showed an increased sleep need. We generated Sik1 S577A mice to examine whether the presence of the PKA phosphorylation site, S577, of SIK1 is required for normal sleep/wakefulness (Fig. 1a). The external appearance, such as coat color, whisker morphology and eye shape, of the heterozygous and homozygous Sik1 S577A mice was indistinguishable from that of the Sik1 +/+ mice. The heterozygous and homozygous Sik1 S577A mice did not show any abnormal locomotion or sensory response against touch or sound. The heterozygous and homozygous Sik1 S577A mice were fertile. The male and female Sik1 S577A and Sik1 S577A mice showed similar weights compared to the wild-type littermates at the age of 8 weeks (Fig. S1a).
The Sik1 S577A/S577A mice exhibited a reduced total wake time over 24 h and during the dark phase (Fig. 2a) and an increased total NREMS time over 24 h and during the dark phase compared with the Sik1 +/+ mice (Fig. 2b). The Sik1 S577A/S577A mice showed an increased total REMS time during the dark phase (Fig. 2c). There was no significant difference in the daily total time spent in any vigilance state between the Sik1 S577A/+ and Sik1 +/+ mice. For episode duration, the Sik1 S577A/S577A mice showed shorter wake episode durations during both the light and dark phases than the Sik1 +/+ mice (Fig. 2d), whereas there were no significant changes in the NREMS episode duration (Fig. 2e) and REMS episode duration (Fig. 2f). The circadian variations of each state showed a sustained tendency toward shorter wake time and longer NREMS and REMS time during the dark phase of the Sik1 S577A/S577A mice ( Fig. 2g-i). EEG spectral analysis revealed that the hourly NREMS delta density of the Sik1 S577A/S577A mice was higher than that of the Sik1 +/+ and Sik1 S577A/+ mice (Fig. 2j). The Sik1 S577A/+ and Sik1 S577A/S577A mice had a higher delta range (1-4 Hz) power during wakefulness than the wild-type littermates (Fig. 2k). There was no significant difference in the delta range power during NREMS among the genotypic groups (Fig. 2l). The Sik1 S577A/S577A mice had a higher delta range power during REMS than the Sik1 S577A/+ and wild-type littermates (Fig. 2m). Thus, the Sik1 S577A homozygous mice showed increased sleep need in terms of time and EEG spectrum.
anti-FLAG, anti-phospho-PKA substrate motif, and anti-14-3-3 antibodies. BNST, bed nucleus of the stria terminalis; MPO, medial preoptic area; och, optic chiasma; PVT, paraventricular thalamic nucleus. One-way ANOVA followed by Tukey's test. Data are presented as the mean ± SEM. group. Full blots are shown in the Supplementary Information. www.nature.com/scientificreports www.nature.com/scientificreports/ To investigate the homeostatic regulation of sleep/wakefulness, we performed sleep deprivation for 6 h with gentle handling. There was no difference in time spent in each vigilance state during the light phase after sleep deprivation among the genotypic groups ( Fig. 3a-c). The Sik1 S577A/S577A mice showed a shorter wake time during the dark phase than the Sik1 S577A/+ mice, and the Sik1 S577A/+ mice showed a shorter wake time than the Sik1 +/+ mice (Fig. 3a). Consistently, the Sik1 S577A/S577A mice showed longer NREMS time during the dark phase than the Sik1 S577A/+ and Sik1 +/+ mice (Fig. 3b). The Sik1 S577A/+ mice showed longer REMS time during the dark phase than the Sik1 +/+ mice (Fig. 3c). The change in NREMS delta power for 2 h after sleep deprivation was higher than that under the basal condition in all genotypic groups (Fig. 3d). Sleep deprivation increased the NREMS delta power (k-m) EEG power density of the Sik1 +/+ , Sik1 S577A/+ , and Sik1 S577A/S577A mice during wakefulness (k), NREMS (l) and REMS (m). 13 male mice for each group. One-way analysis of variance (ANOVA) followed by Tukey's test (a-f, k-m). One-way repeated-measures ANOVA followed by Tukey's test (j). Data are presented as the mean ± SEM. www.nature.com/scientificreports www.nature.com/scientificreports/ to a similar extent among the genotypic groups (Fig. 3e). The Sik1 S577A/S577A mice showed a higher NREMS delta density during the light phase than the Sik1 +/+ mice (Fig. 3f). Similar to basal sleep/wakefulness, the Sik1 S577A/+ and Sik1 S577A/S577A mice had a higher delta power during wakefulness than the wild-type littermates after sleep (g-i) EEG power spectra of the Sik1 S577A mice during two hours after 6-h sleep deprivation during wakefulness (g), NREMS (h), and REMS (i). One-way repeated-measures analysis of variance (ANOVA) followed by Tukey's test (a-c,f). Two-tailed paired T-test (d). Two-way ANOVA (e). One-way ANOVA followed by Tukey's test (g-i). 13 male mice for each group. Data are presented as the mean ± SEM. (2020) 10:8676 | https://doi.org/10.1038/s41598-020-65647-0 www.nature.com/scientificreports www.nature.com/scientificreports/ deprivation (Fig. 3g), and there was no significant difference in the delta range power during NREMS among the genotypic groups (Fig. 3h). The Sik1 S577A/S577A mice had a higher delta range power during REMS than the wild-type littermates (Fig. 3i). These results suggest normal homeostatic sleep regulation in the Sik1 S577A mice.
the Sik2 S587A mice showed a mild increase in sleep need. Next, we generated Sik2 S587A mice to investigate the role of the phosphorylation of SIK2 S587 in sleep/wake behavior (Fig. 1a). We confirmed that a SIK2 protein of the same size as wild-type protein was expressed in the homozygous Sik2 S587A mice (Fig. 1i). The SIK2 levels in the brains and BAT of the Sik2 S587A/S587A mice were lower than those of the Sik2 +/+ mice (Fig. S1). The external appearance of the heterozygous and homozygous Sik2 S587A mice was indistinguishable from that of the Sik2 +/+ mice. Although there was no significant difference in body weight at the age of 8 weeks of the male mice with the Sik2 S587A genotype (Fig. S2b), the body weight of the female heterozygous and homozygous Sik2 S587A mice was lower than that of the wild-type littermates. The heterozygous and homozygous Sik2 S587A mice did not show any abnormal locomotion or sensory response. The heterozygous and homozygous Sik2 S587A mice were fertile.
Both the Sik2 S587A/+ and Sik2 S587A/S587A mice had a total time spent in each vigilance state similar to that of the wild-type mice (Fig. 4a-c). There was no significant difference in episode duration of each vigilance state among the genotypic groups ( Fig. 4d-f). EEG spectral analysis revealed a higher NREMS delta density of the Sik2 S587A/+ and Sik2 S587A/S587A mice than that of the Sik2 +/+ mice throughout the entire day (Fig. 4g). There was no difference in the delta range power in each vigilance state among the genotypic groups ( Fig. 4h-j). Thus, the Sik2 S8587A homozygous mice showed increased sleep need in terms of EEG spectrum.
Normal circadian behavior of the Sik1 S577A mice. Since Sik1 is expressed in the SCN (Fig. 1c) and was reported to be involved in the phase shift of circadian rhythm based on deficiency studies 16 , we examined the circadian behavior of the Sik1 S577A mice. Under constant darkness, the Sik1 S577A/+ and Sik1 S577A/S577A mice showed a circadian period similar to that of the wild-type littermates (Fig. 6a-c). Furthermore, the Sik1 S577A/+ and Sik1 S577A/S577A mice showed re-entrainment to a new circadian time with a similar phase shift to the wild-type mice ( Fig. 6d-f). Thus, Sik1 S577A mutant mice showed normal circadian behavior in terms of endogenous circadian period length and re-entrainment.
Normal body weight growth of the Sik2 S587A mice. Since Sik2 is highly expressed in BAT (Fig. 1b), which regulates energy metabolism, we examined the body weight growth of the Sik2 S587A mice. At the age of 6 weeks, the body weights of the male Sik2 S587A mice fed a chow or a high-fat diet were similar to those of the wild-type littermates (Fig. 7a). The body weights of the female Sik2 S587A/S587A mice on chow were smaller than those of the Sik2 +/+ mice (Fig. 7b). However, there was no difference between the female Sik2 +/+ and Sik2 S587A/ S587A mice on a high-fat diet (Fig. 7b). At the age of 20 weeks, both male and female Sik2 S587A mice fed a chow or a high-fat diet showed body weights similar to those of their wild-type littermates (Fig. 7c,d). There was no significant difference in the body weight gain from 6 weeks to 20 weeks of age among the genotypic groups for both sexes (Fig. 7e,f). Thus, male and female Sik2 S587A mice have normal body weight regulation.

Discussion
In this study, we examined the sleep/wake behavior of mice lacking conserved PKA phosphorylation sites of SIK1 and SIK2. Both the Sik1 S577A mice and the Sik2 S587A mice showed an increased NREMS delta density, an index of sleep, which is consistent with the results of the Sik3 Slp mice that lack an exon 13-encoded region and the Sik3 S551A mice that lack a conserved PKA phosphorylation site 1,2 .
The homozygous Sik1 S577A mice showed a shorter total wake time and longer total NREMS time than the wild-type mice in addition to an increased NREMS delta density. Thus, the Sik1 S577A mice had a sleep phenotype similar to that of the Sik3 mutant but to a lesser degree, which is explained by the lower Sik1 expression than Sik3 expression in the brain. Increased delta power during wakefulness and REMS of the Sik1 S577A mice may indicate sleep need during wakefulness and REMS, which was also observed in the Sik3 Slp mice 1 . The sleep-deprived Sik1 S577A mice showed a longer NREMS time and higher NREMS delta density than the sleep-deprived wild-type www.nature.com/scientificreports www.nature.com/scientificreports/ littermates. Sleep deprivation increased the NREMS delta density to a similar extent, suggesting that the Sik1 S577A mice showed proper homeostatic sleep regulation.
Both the heterozygous and homozygous Sik2 S587A mice exhibited an increased NREMS delta density, but the Sik2 S587A mice did not alter the time spent in wakefulness and sleep. Lower brain expression of Sik2 than Sik3 accounts for the milder sleep phenotype of the Sik2 S587A mice than that of the Sik3 Slp and Sik3 S551A mice. Differential expression patterns between Sik2 and Sik3 may cause normal sleep time of the Sik2 S587A mice. Thus, the current results as a whole support the idea that PKA-SIK signaling is involved in the regulation of sleep need. Importantly, KIN-29, the Caenorhabditis elegans homologue of SIK is involved in the regulation of sleep 1,25 .
Consistent with previous studies 2,10,11 , the SIK1 S577A, SIK2 S587A, and SIK3 S551A proteins showed reduced immunoreactivities for phospho-PKA substrate motifs. The remaining immunoreactivities for the phospho-PKA motif of SIK1 S577A, SIK2 S587A, and SIK3 S551A are attributed to other PKA recognition sites, such as SIK1 T475, SIK2 T484 and SIK3 T469 11 . In addition, the SIK1 S577A, SIK2 S587A, and SIK3 S551A proteins showed decreased binding to 14-3-3, as previously reported 2,10,11 . Importantly, each PKA phosphorylation site of SIKs has differential effects on binding to 14-3-3. Berggreen et al. showed much lower 14-3-3 binding of the SIK3 T469A and SIK3 S551A proteins than the wild-type SIK3 using HEK293 cells 10   www.nature.com/scientificreports www.nature.com/scientificreports/ is consistent with the decreased 14-3-3 binding observed in the present study. However, Henriksson et al. showed that the alanine substitution of S343 or S358 of SIK2 decreased 14-3-3 binding, but the alanine substitution of SIK2 S587 did not affect binding to 14-3-3 in HEK293 cells 14 . Although the reason for this discrepancy is unknown, the effect of PKA on SIKs may be omplex depending on the context and the presence of other upstream kinases 4 .
Consistent with a previous report indicating that SIK1 in the SCN is necessary for proper circadian re-entrainment 16 , SIK1 is highly expressed in the SCN. However, the Sik1 S577A mice showed normal circadian rhythm and re-entrainment to an advanced light-dark cycle, suggesting that the functional change of SIK1 has little effect on circadian behavior. Similarly, the Sik3 Slp mice showed an enhanced sleep need and normal circadian behavior 1 , while SIK3 deficiency caused abnormal circadian behavior in mice and fruit flies 23,32 . However, since systemic SIK3 deficiency caused severe growth retardation and metabolic and skeletal abnormalities 33,34 , neuron-specific SIK3 deficiency is necessary to examine the role of brain SIK3 in circadian behavior in detail.
Male Sik2 S587A mice exhibited body weights similar to those of the wild-type littermates fed either a high-fat diet or chow. Although the female Sik2 S587A mice showed lower body weights than the wild-type controls at the ages of 6 and 8 weeks, they showed a normal body weight gain as they grew. These results are not consistent with a previous report demonstrating that mice with specific Sik2 S587A overexpression in BAT showed high-fat diet-induced obesity 22 . Thus, the SIK2 S587A protein in the BAT may promote weight gain, but SIK2 S587A outside the BAT may counteract the effect on the BAT. Although we did not examine this issue in our study, the Sik2 S587A mice may show altered glucose metabolism since higher blood glucose was found in Sik2-deficient mice 35 and those with pancreatic beta cell specific Sik2 ablation 36 , but not liver-specific SIK2 deficiency, than wild-type mice 19 . Although the mice deficient in Sik1 showed lower body weights under both chow and high-fat diets 17,18 , the male and female Sik1 S577A mice did not show any significant change in body weight. Thus, PKA-SIK1/2 signaling does not play a major role in energy metabolism.
We also examined whether a high-salt diet affects Sik expression. Although a high-salt diet for seven days strongly induced Sik1 mRNA in the adrenal gland of rats 7 , we did not observe a significant increase in Sik1 mRNA in the adrenal gland of mice. This unexpected result may be due to species differences and/or some technical differences. Instead, we found that a high-salt diet increased the Sik2 mRNA levels in the adrenal gland and liver. Sik1 mRNA showed a higher individual difference than Sik2 and Sik3 mRNA.
One limitation of our study is the protein level characterization of mutant mice. Although several antibodies for SIK1 are commercially available, all the SIK1 antibodies we tested did not work. Therefore, we generated polyclonal antibodies in rabbits for SIK1. However, our SIK1 antibodies are not sensitive enough to detect endogenous SIK1, which may be expressed at a low level according to the digital PCR results. For SIK2, a good anti-SIK2 antibody (Cell Signaling Technology #6919) detected SIK2 in the BAT of the wild-type mice, and the corresponding bands were not detected in the SIK2-deficient mice (Fig. S4a). However, this SIK2 antibody showed a band similar in size to SIK2 in brains, which was found for both the wild-type and SIK2 KO mice (Fig. S4b). Then, we used a different antibody (Merck #07-1378), which detects SIK2 in brains (Fig. S4d) and BAT (Fig. S4c). Although identification of endogenous proteins interacting with SIKs in the brain is crucial for the mechanistic understanding of how the SIK family regulates sleep need, the low abundance of SIK2 protein in the brain makes immunoprecipitation difficult. This issue needs to be addressed in future studies using mice that have endogenous SIKs with epitope tags.
We found that SIK2 (S587A) protein level in the brains of the Sik2 S587A/S587A mice was lower than the level of SIK2 protein in the wild-type mice. Additionally, the amount of SIK2 (S587A) protein in BAT of the Sik2 S587A/S587A mice was lower than the amount of SIK2 protein in the wild-type mice. This result suggests that phosphorylation at S587A affects the protein amount and/or stability. Since 14-3-3 has been reported to stabilize the binding protein 37,38 , reduced binding of SIK2 S587A to 14-3-3 may result in reduced SIK2 S587A levels. Thus, PKA may regulate SIK2 in multiple aspects, including kinase activity and protein stability, via 14-3-3 binding, which leads to altered phosphorylation of substrates and sleep need.
In summary, the current study broadened and confirmed the hypothesis that PKA-SIK signaling constitutes an intracellular signaling pathway regulating sleep/wakefulness, as was originally conceived based on findings in Sik3 mutant mice. Elucidation of downstream signals is a challenge for the near future.

Animals.
All animal experiments were conducted in accordance with the Guidelines for Animal Experiments of the University of Tsukuba and were approved by the Institutional Animal Care and Use Committee of the University of Tsukuba (approved protocol ID #180094). Sik1 S577A and Sik2 S587A mice and littermate wild-type mice were maintained on the C57BL/6 background. Male C57BL/6 mice were obtained from CLEA Japan. Sik2deficient mice 21,39 (#nbio071) were obtained from the Animal Resource Bank of National Institutes of Biomedical Innovation, Health and Nutrition, Japan. Mice were housed under humidity-and temperature-controlled conditions on a 12:12 h light/dark cycle. Food was provided ad libitum. Mice were weaned at 4 weeks of age and were housed in groups of four or five.
High-salt diet feeding. For a high-salt diet, NaCl was added to MF (containing 0.3% NaCl, Oriental Yeast) to a concentration of 6.6% NaCl by weight [43][44][45] , and then water was added to mix NaCl and MF. Mice were fed a high-salt diet or an MF chow diet for one week. One week of feeding with a high-salt diet did not affect the body weight but increased water intake and urination.
EEG/EMG electrode implantation surgery. At 8-12 weeks of age, male mice were implanted with EEG/ EMG electrodes with 4 EEG electrode pins and 2 flexible stainless EMG wires under anesthesia using isoflurane (4% for induction, 2.5% for maintenance) 46  3 mm) and attached to the skull using dental cement. Then, the EMG wires were inserted into the neck muscles. All mice were allowed at least 4 to 7 days for recovery from the surgery. After the recovery period, all mice were attached to a tether cable and then allowed to habituate to the recording conditions for 7 days. EEG/EMG recording and analysis. EEG/EMG recordings were analyzed as previously described 46 . The recording room was kept under a 12:12 h light/dark cycle and a constant temperature (24-25 °C). The EEG/EMG data were visualized and semiautomatically analyzed by MATLAB-based software. The sleep/wake state in each 20-s epoch was classified as NREMS, REMS or wakefulness. Wakefulness was scored based on the criteria of the presence of fast EEG, high amplitude and variable EMG. NREMS was staged based on high amplitude, delta (1-4 Hz) frequency EEG and low EMG tonus. REMS was characterized by theta (6-9 Hz)-dominant EEG and EMG atonia. For analysis of baseline sleep/wake behavior, EEG/EMG signals were recorded for 2 consecutive days from the onset of the light phase, ZT0. The total amounts of time spent in wakefulness, NREMS and REMS were derived by summing the total number of 20-s epochs in each state. Mean episode durations were determined by dividing the total time spent in each state by the number of episodes of that state. EEG signals were subjected to fast Fourier transform analysis from 1 to 30 Hz with 1-Hz bins using MATLAB-based custom software. For normalization of the EEG power density, the EEG power density in each frequency bin was expressed as a percentage of the mean EEG power in a relatively constant higher frequency range (16-30 Hz) 2 . The hourly delta density during NREMS indicates the hourly averages of delta density, which is the ratio of delta power (1-4 Hz) to total EEG power  at each 20-s epoch or all epochs. For sleep deprivation, the mice were sleep-deprived for six hours from the onset of light phase ZT0 by gentle handling 47 . During that time, food and water were available. For evaluation of the effect of sleep deprivation, NREMS delta power during the first hour after sleep deprivation was expressed relative to the same ZT of the basal recording or relative to the mean of the basal recording.
In situ hybridization. In situ hybridization was performed as previously described with some modifications 48 . Fragments of the coding regions for Sik1 and Sik2 0.7-0.8 kb in length were generated by PCR using mouse brain and BAT cDNA as templates, respectively. The primers used were Sik1-Fw (5′-ATAGA CTGTG ATCTC CACAG CTCAC TT-3′), Sik1-Rv (5′-ACAGG GAGCA AGCAC ATAGG-3′), Sik2-Fw (5′-AACCC CTCCC TTGAG AGTGT-3′) and Sik2-Rv (5′-GGAAG AGTCG CTTCT GTTGG-3′). Sik1 and Sik2 cDNAs were inserted into pGEM-T easy (Promega) and used for digoxigenin (DIG)-labeled probe synthesis. Mice were perfused with PBS followed by 4% paraformaldehyde (PFA), and harvested brains were postfixed in 4% PFA overnight. Forty μm-thick brain sections were treated with 0.3% Triton X-100, digested with 1 μg/ml proteinase K, treated with 0.75% glycine, and then treated with 0.25% acetic anhydride in 0.1 M triethanolamine. After overnight incubation with a DIG-labeled probe at 60 °C, the sections were washed and then incubated with alkaline phosphatase-conjugated anti-DIG Fab fragments (Roche). The reactions were visualized with a Quantitative RT-PCR and digital PCR. RNA was prepared from the mouse brain, liver, BAT and adrenal gland using the RNeasy Lipid Tissue Mini Kit (Qiagen) and QIAzol Lysis Reagent (Qiagen). cDNA was synthesized using oligo dT primer and RNAs as a template with the PrimeScript Reverse Transcriptase Kit (TaKaRa). To quantitate Sik mRNA in each tissue, we used the QX200 Droplet Digital PCR system (Bio-Rad). We prepared a mixture containing cDNA, primers, and ddPCR EvaGreen (Bio-Rad), and processed the mixture for the QX200 Droplet Generator to create droplets. Then, we performed digital PCR and read duplicated droplets using a QX200 Reader according to the manufacturer's instructions.
For quantitation of the effect of a high-salt diet, cDNAs were subjected to real-time PCR (ViiA7; ThermoFisher) using SYBR PreMix Ex Taq (TaKaRa) and gene-specific primers. The expression of Sik1, 2, and 3 in each tissue, was normalized to the glyceraldehyde-3-phosphate dehydrogenase (Gapdh) gene 49  High-fat diet and body weight. Body weight was measured weekly from the age of 6 weeks to 20 weeks.

Circadian behavior.
Male mice were housed individually in a cage containing a wireless running wheel (Med Associate #ENV-047). The rotation numbers of the wheels were collected by Wheel Manager software (Med Associate) via a receiving device (Med Associate #DIG807). The mice were entrained on the running wheels on a 12 h: 12 h light-dark cycle for 7 days and then allowed to run freely for 14 days under constant darkness. For analysis of re-entrainment to a new light-dark cycle, the mice were maintained on a 12 h: 12 h light-dark cycle, and then, the cycle was advanced by six hours. Ten days later, the cycle was advanced again by six hours. The circadian period and re-entrainment were analyzed by a Python-based program. The onset of activity on each day was used to measure the phase relative to the light-dark cycle 16 . The free-running period was calculated with linear regression analysis of activity onset using a Python-based program. The circadian activity amplitude was normalized to the mean amplitude of the wild-type group 1 .
Immunoprecipitation and western blot. Used primary antibodies were anti-DYKDDDDK tag antibody (mouse, Monoclonal, Clone #1E6, Wako) as anti-FLAG antibody, anti-phospho-PKA substrate (RRXS/T) antibody (rabbit, monoclonal, 100G7E; Cell Signaling Technology #9624), anti-14-3-3 (pan) antibody (rabbit, polyclonal, Cell Signaling Technology #8312), anti-β-tubulin antibody (rabbit, monoclonal, Cell Signaling Technology #2128), anti-GAPDH (D16H11) XP antibody (rabbit, monoclonal, Cell Signaling Technology #5174) and anti-SIK2 antibody (rabbit, polyclonal, Cell Signaling Technology #6919 and Merck #07-1378). Since the anti-SIK2 antibody (Cell Signaling Technology #6919) showed a band similar in size to SIK2 in brains, which was found for both the wild-type and SIK2 KO mice (Fig. S4b), we used the anti-SIK2 antibody (Merck #07-1378) for densitometric analysis of the brain and BAT (Fig. S4c,d, S1). Mice were sacrificed by cervical dislocation under deep anesthesia with sodium pentobarbital (50 mg/kg body weight). Then, BAT and brains were rapidly removed, frozen in liquid nitrogen and stored at −80 °C until use. The BAT and brain were homogenized using a rotor-stator homogenizer (Polytron) in ice-cold lysis buffer (20 mM HEPES pH 7.5, 100 mM NaCl, 10 mM Na 4 P 2 O 7 , 1.5% Triton X 100, 15 mM NaF, 1X PhosSTOP (Roche), 5 mM EDTA, 1X Protease Inhibitor (Roche)), and then centrifuged for 12 min at 14,000 g at 4 °C. For detection of SIK2, brain homogenates were processed for methanol/chloroform precipitation. The pellets were resolved in SDS-PAGE sample buffer. For immunoprecipitation (IP), transiently expressed 3xFLAG-tagged SIK1, SIK2, and SIK3 proteins were extracted with lysis buffer and then centrifuged for 12 min at 14,000 g at 4 °C. The extracted proteins were incubated with a precipitating antibody at 4 °C for 2 h. The beads were washed three times with IP buffer, and the final precipitates were subjected to immunoblotting analysis. The supernatants were separated by SDS-PAGE and transferred to PVDF membranes. The blots were subsequently washed in PBS plus 0.1% Tween-20 (PBS-T) and incubated overnight at 4 °C with primary antibody (1:1000) in PBS-T with 5% bovine serum albumin. The blots were then washed and incubated with horseradish peroxidase-conjugated donkey anti-rabbit IgG (Jackson Immuno Research; 1:5,000