Sympathetic inputs regulate adaptive thermogenesis in brown adipose tissue through cAMP-Salt inducible kinase axis

Various physiological stimuli, such as cold environment, diet, and hormones, trigger brown adipose tissue (BAT) to produce heat through sympathetic nervous system (SNS)- and β-adrenergic receptors (βARs). The βAR stimulation increases intracellular cAMP levels through heterotrimeric G proteins and adenylate cyclases, but the processes by which cAMP modulates brown adipocyte function are not fully understood. Here we described that specific ablation of cAMP production in brown adipocytes led to reduced lipolysis, mitochondrial biogenesis, uncoupling protein 1 (Ucp1) expression, and consequently defective adaptive thermogenesis. Elevated cAMP signaling by sympathetic activation inhibited Salt-inducible kinase 2 (Sik2) through protein kinase A (PKA)-mediated phosphorylation in brown adipose tissue. Inhibition of SIKs enhanced Ucp1 expression in differentiated brown adipocytes and Sik2 knockout mice exhibited enhanced adaptive thermogenesis at thermoneutrality in an Ucp1-dependent manner. Taken together, our data indicate that suppressing Sik2 by PKA-mediated phosphorylation is a requisite for SNS-induced Ucp1 expression and adaptive thermogenesis in BAT, and targeting Sik2 may present a novel therapeutic strategy to ramp up BAT thermogenic activity in humans.


The cAMP production in brown adipocytes is required for cold-induced adaptive thermogenesis.
In order to investigate effects of βAR-cAMP signaling deficiency in brown adipocytes, we have generated Ucp1-Cre;Gnas f/f (Gnas BKO ) mice. Cre-negative Gnas f/f mice were used as controls. Gnas mRNA and protein levels were specifically reduced in interscapular brown adipose tissue (iBAT), but not in other tissues ( Supplementary  Fig. 1, Fig. 1B-C). Protein kinase A (PKA) is activated upon elevated cAMP levels. Consistently, PKA activity (determined by immunoblot of phosphor-PKA substrate antibody) was abolished in the iBAT of Gnas BKO mice (Fig. 1C). Similar to βless mice 23 , Gnas BKO mice had a pale and enlarged iBAT (Fig. 1A). Their brown adipocytes exhibited white adipocyte-like morphology, containing a single and large lipid droplet (Fig. 1A). They also showed reduced thermogenic gene expression, such as Ucp1, Pgc1α, Dio2, Cox8b, and Cidea (Fig. 1B). Ucp1 protein levels were also diminished in iBAT of Gnas BKO mice (Fig. 1C). We further examined the thermogenic capacity in Gnas BKO mice. In the indirect calorimetry experiment, murine-selective β3-AR agonist CL-316,243 (CL) failed to induce oxygen consumption in Gnas BKO mice (Fig. 1D), even though there were no differences in basal O2 consumption, respiratory exchange ratio (RER), food intake and physical activity (Supplementary Fig. 2A-E). Consequently, Gnas BKO mice could not maintain their core temperature under 4 °C cold challenge (Fig. 1E). Thus, cAMP signaling in brown adipocytes is required for BAT thermogenic function.
Besides Ucp1-mediated proton leak, adaptive thermogenesis also requires lipolysis to generate fatty acid and mitochondrial respiration to generate proton gradient across the mitochondrial membrane. Indeed, Gnas BKO mice exhibited reduced in vitro Forskolin (FSK)-induced lipolytic activity of iBAT ( Fig. 2A), which was consistent with reduced PKA activity (Fig. 1C). The in vitro FSK-induced lipolytic activity in epididymal WAT (eWAT) was not altered in Gnas BKO mice ( Fig. 2A). Additionally, CL-induced serum glycerol levels were lower in Gnas BKO mice, showing an attenuated lipolytic response in vivo (Fig. 2B). The cAMP signaling in brown adipocytes drives mitochondrial biogenesis through promoting transcription of the Peroxisome proliferator-activator gamma coactivator 1 alpha (Pgc1α); Pgc1α mRNA was reduced in the iBAT of Gnas BKO mice (Fig. 1B). Consistently, we observed reduced expression of most mitochondrial ETC genes encoded by both nuclear and mitochondrial genomes ( Fig. 2.C,D). Consistently, the mitochondrial DNA copy numbers were reduced by half in the iBAT of Gnas BKO mice (Fig. 2E). We further performed mass spectrometry analysis of isolated mitochondria from iBAT of control and Gnas BKO mice 38,39 . We identified more than 630 mitochondrial proteins (roughly 60% of the mitochondrial proteins listed in MitoCarta2.0) ( Supplementary Fig. 3A). However, mitochondrial proteome in isolated iBAT mitochondria was minimally affected by Gnas deficiency (Fig. 2G). Hexokinase 1 (Hk1) was the most upregulated protein in iBAT mitochondria from Gnas BKO mice in the mass spectrometry dataset, which was confirmed by immunoblots ( Supplementary Fig. 3B). Therefore, despite reduced mRNA levels of electron transport chain (ETC) subunits, the ETC proteome composition was not affected in iBAT of Gnas BKO mice. For example, the complex IV protein levels (mtDNA-encoded mt-Co1 and mt-Co2, nuclear-encoded Cox4, Cox5b, and Cox6b) in isolated mitochondria were not affected by Gnas deficiency (Fig. 2H). This data suggests that cAMP signaling controls lipolysis (though PKA activation) and mitochondrial biogenesis (through regulating Pgc1α transcription) in brown adipose tissue.
Animals at thermal neutral zone (~30 °C for mice) maintain their body temperature through its basal metabolism without additional thermogenesis and physical activity. There is minimal sympathetic flow to BAT for mice housed at thermoneutrality (~30 °C). However, housing mice at room temperature (RT, ~22 °C) results in the activation of the BAT-mediated adaptive thermogenesis. Interestingly, reductions in mRNA levels of ETC Gross view and H&E staining of dissected iBAT in ~6-8-week-old male CON and Gnas BKO mice. Scale bar: 200 μm. A single adipocyte was outlined by a dashed yellow line. Scale bar: 50 μm. (B) q-PCR analysis of Gnas, Ucp1, Pgc1α, Cox8b, Cidea and Dio2 mRNA levels in iBAT from 6-8 week-old male CON and Gnas BKO mice. Sample sizes: n = 7 for both genotypes. (C) Immunoblots showing amounts of Gnas, phosphor-PKA substrates, Ucp1 and Hsp90 in iBAT from 6-8 week-old male CON and Gnas BKO mice. (D) Oxygen consumption recordings in response to CL in 6-8 week-old male CON and Gnas BKO mice. Sample size: CON (n = 4) and Gnas BKO (n = 3). (E) Core temperature of 6-8 -week-old male CON and Gnas BKO mice upon 4 °C cold challenge. Sample size: CON (n = 12) and Gnas BKO (n = 9). subunits and Ucp1 proteins in iBAT were less profound in Gnas BKO mice housed at thermoneutrality (Fig. 2F,H), suggesting that sympathetic inputs regulate Ucp1 and Pgc1α transcription through Gnas-mediated cAMP generation in brown adipocytes. Glycerol released in vitro prior to and after Forskolin (10 uM FSK) stimulation from iBAT and eWAT from 6-8-week-old CON and Gnas BKO mice. Sample size: n = 10 per genotypes. (B) Serum glycerol levels prior to and one-hour after CL injection in 6-8-week-old CON and Gnas BKO mice. Sample size: CON (n = 5) and Gnas BKO (n = 4). q-PCR analysis of relative mRNA levels of nuclear (C) and mitochondrial (D) encoded ETC gene expression in 6-8-week-old CON and Gnas BKO mice. Sample size: n = 5 per genotypes. (E) Relative mitochondrial DNA (mtDNA) levels in the iBAT of 6-8-week-old CON and Gnas BKO mice. Sample size: n = 5 per genotypes. (F) Heatmap showing log2 fold changes of mRNA levels of nuclear and mitochondrial encoded ETC gene expression in 6-8-week-old CON and Gnas BKO mice housed at room temperature (RT) and thermoneutrality (30 °C). (G) Volcano plots showing significantly (p < 0.1) down-or up-regulated mitochondrial proteins (over 1.5 fold) in isolated iBAT mitochondria from 6-8-week-old male Gnas BKO mice. (H) Immunoblots showing amounts of Ucp1, C-IV subunits (mt-Co1, mt-Co2, Cox4, Cox5b, and Cox6b1) and Hsp60 in iBAT mitochondria from 6-8-week-old male CON and Gnas BKO mice housed at RT and 30 °C. The cAMP production in beige adipocytes is dispensable for cold-induced beige adipocyte renaissance. SNS-induced βAR activation is also critical for the formation of beige adipocytes, brown-like adipocytes with multilocular morphology and Ucp1-dependent thermogenic activity 40,41 . For example, cold failed to induce beige adipocyte formation in adult βless mice ( Supplementary Fig. 4D), highlighting the importance of βAR signaling in beige adipocyte formation in vivo. Additionally, chemical denervation with 6-hydroxydopamine (6-OHDA) prior to 3-week of age, or prior to 7-day 8 °C cold challenge in adult mice led to reduced Ucp1 mRNA expression ( Supplementary Fig. 4A). Western blot confirmed that 6-OHDA suppressed PKA activity and Ucp1 protein levels in 3-week-old pups and cold-treated adult mice ( Supplementary Fig. 4B,C), suggesting that sympathetic innervation was required for beige adipocyte genesis during postnatal development and cold-induced beige adipocyte formation in adult mice.
To address whether cAMP in beige adipocytes themselves is necessary for beige adipocyte formation in iWAT, we compared Ucp1 expression and beige adipocyte abundance in iWAT in Gnas BKO and adipocyte-specific Gnas knockout mice (Adiponectin-Cre;Gnas f/f ; Gnas AKO ). The postnatal beige adipocyte developed normally in iWAT from 3-week-old Gnas BKO and Gnas AKO pups, indicating that adipocyte cAMP signaling was dispensable for de novo beige adipocyte formation during postnatal development ( Supplementary Fig. 4E). However, 7-day 8 °C cold treatment robustly induced Ucp1 transcription in iWAT of adult Gnas BKO and control mice, but had no effect in Gnas AKO mice ( Supplementary Fig. 4E). Histology analysis confirmed that multilocular beige adipocytes reappeared in iWAT of cold-treated Gnas BKO mice ( Supplementary Fig. 4F), demonstrating that cAMP signaling in beige adipocytes themselves was dispensable for their maintenance in vivo at adult stage ( Supplementary Fig. 4G).
Gnas BKO mice are not obese under HFD despite thermogenic defects in iBAT. We then accessed whether defective adaptive thermogenesis in BAT was linked to metabolic dysfunctions in Gnas BKO mice. At room temperature (RT), the Gnas BKO mice had normal body weight, lean and fat mass under normal chow feeding, their visceral fat mass was specifically reduced at the expense of enlarged iBAT (Supplementary Fig. 5A-B). This fat redistribution was not due to a secondary adaptive response triggered by the defective adaptive thermogenesis, because it was also present in Gnas BKO mice housed at thermoneutrality ( Supplementary Fig. 5B). After 6-week high-fat diet (HFD), the Gnas BKO mice showed no differences in body weight, and lean and fat mass ( Fig. 3A-B), and levels of fasting serum TG and glucose remained unchanged ( Fig. 3C-D). The iBAT in Gnas BKO mice had a three-fold increase in size and contained unilocular lipid-filled adipocytes ( Fig. 3E-G). In contrast, their eWAT mass was reduced by half ( Fig. 3E-G), although their lipolytic activity or adipocyte size was not altered (Fig. 3G-H, Fig. 2A). The genomic content in the eWAT was reduced in the Gnas BKO mice (Fig. 3I), suggesting that reduction of adipocyte numbers may account for smaller eWAT mass in the Gnas BKO mice. Pro-adipogenic Cebpa and Pparg gene expressions (and Pparg protein levels) were diminished in the eWAT of Gnas BKO mice, without the change of the abundance of Pdgfra + Sca1 + progenitors ( Fig. 3J-K, Supplementary  Fig. 6). Notably, the fat redistribution between iBAT and eWAT was not observed in Gnas AKO mice 42 . Collectively, diminished thermogenic capacity in Gnas BKO mice was not associated with significant metabolic abnormalities under normal chow and HFD.

The cAMP signaling inhibits Salt-inducible kinases (SIKs) in BAT in response to sympathetic inputs.
Mammalian SIK family contains three members: Sik1, 2 and 3 [43][44][45] . Both Sik1 and Sik2 are expressed in mature adipocytes compared to stromal-vascular fraction (SVF) cells, although Sik2 is an adipose-enriched SIK isoform, which is abundantly expressed in many fat depots (Supplementary Fig. 7) 37,45 . To test whether sympathetic nerves regulate SIK activity and whether this regulation is required for thermogenic gene expression, we analyzed CL's effect on Sik2 activity in differentiated brown adipocyte. We had previously shown that Sik2 S587 phosphorylation is a negative indicator of its kinase activity, because hyper-phosphorylated Sik2 was accompanied with de-phosphorylation of its known substrates (CRTCs and HDAC4) 32,34 . In in vitro differentiated brown adipocytes, CL treatment robustly induced PKA signaling (Fig. 4A). Sik2 was hyper-phosphorylated at Ser587 in response to CL in differentiated brown adipocytes, and consequently, its substrate Hdac4 was hypo-phosphorylated (Fig. 4A). These data suggested that Sik2 activity was inhibited by cAMP-PKA signaling in brown adipocytes in vitro.
We then determined whether Sik2 activity in brown adipose tissue was differentially regulated at RT and thermoneutrality in vivo. Ucp1 expression and PKA activity were reduced at thermoneutrality in iBAT from C57bl/6 J mice due to reduced sympathetic inputs (Fig. 4B). Along with reduced PKA activity, Sik2 was hypo-phosphorylated at Ser587 and Hdac4 was hyper-phosphorylated at Ser245 in iBAT at thermoneutrality (Fig. 4B). On the other hand, acute 4 °C cold stimulation for half hour robustly increased cAMP signaling in iBAT, which led to Sik2 Ser587 hyper-phosphorylation and Hdac4 Ser245 hypo-phosphorylation (Fig. 4C). The Gnas BKO mice had no PKA signaling in iBAT (Fig. 1C), hence, Sik2 was hypo-phosphorylated and active in the iBAT of Gnas BKO mice at RT (Fig. 4D). Collectively, we conclude that Sik2 activity is negatively correlated with sympathetic activity and Ucp1 expression in iBAT.

SIKs inhibits thermogenic gene expression in brown adipocytes in vitro and in vivo.
Using adenoviral-mediated knockdown in differentiated brown adipocytes, we found that combinational knockdown of Sik1 and Sik2 led to elevated Ucp1 mRNA levels along with other thermogenic genes (such as Pgc1α and Dio2) ( Supplementary Fig. 8A). Additionally, using SIK specific small molecule inhibitors, we found that HG-9-91-01 and MRT199665 potently inhibited HDAC4 phosphorylation and promoted Ucp1 expression in differentiated brown adipocytes ( Fig. 5A-B) 46 . All data here suggests that SIKs suppress thermogenic gene expression in differentiated brown adipocytes in vitro.
Then we determined whether SIK deficiency affected thermogenic gene expression in vivo. Sik1 and Sik2 are two major SIK isoforms in iBAT 37 . Both Sik1 and Sik2 single knockout mice exhibited similar Ucp1 expression compared with their littermate controls ( Supplementary Fig. 8B-E), which was possibly due to redundant roles of Sik1 and Sik2 in regulating Ucp1 gene expression in brown adipose tissue. Transcription of Sik1 in BAT was robustly upregulated by thermal stress, as Sik1 mRNA level at RT was ~7-fold higher than that at thermoneutrality ( Supplementary Fig. 9). In order to minimize the compensatory effect of Sik1 on thermogenic gene expression in iBAT of Sik2 KO mice, we did all the experiments at thermoneutrality, where Sik1 expression was greatly reduced. Indeed, at thermoneutrality Sik2 KO mice exhibited reduced Hdac4 Ser245 phosphorylation, elevated Ucp1 protein levels and thermogenic gene expression (Ucp1 and Dio2) (Fig. 5C,D). Thus, we conclude that Sik2 suppresses thermogenic gene expression in BAT at thermoneutrality. Notably, PKA-mediated activation of hormone-sensitive lipase (HSL) was not affected by inhibition of Sik2 (Fig. 5.A,C).
Sik2 suppresses Ucp1-dependent adaptive thermogenesis at thermoneutrality. We then determined whether Sik2 deficiency affected BAT thermogenic capacity in vivo. Consistent with elevated Ucp1 expression, Sik2 KO mice exhibited increased norepinephrine-induced oxygen consumption (~1.5 fold) upon norepinephrine injection at thermoneutrality (Fig. 6A, Supplementary Fig. 10F). We further examined whether Sik2 KO mice can maintain their core temperature upon 4 °C acute cold challenge. Mice acclimated at 30 °C were singly housed in a 4 °C chamber and their core body temperatures were monitored every hour and up to 6 hours. We observed that the core temperatures of wild-type (WT) mice dropped rapidly upon 4 °C cold challenge (from 37 °C to 29 °C in ~4 hours), while Sik2 KO mice maintained their core temperatures at ~35 °C for up to 6-8 hours at 5 °C (Fig. 6B). Half of WT mice dropped their core temperature lower than 28 C after 6-hour cold challenge, while all Sik2 KO mice sustained theirs (Fig. 6C). WT and Sik2 KO mice at 6-8-week-old of age have similar body weight and fat content, therefore there will be no difference in body fat insulation from heat loss. Also, there were no significant differences in other metabolic parameters, such as basal oxygen consumption, RER, food intake  and physical activity ( Supplementary Fig. 10A-E). Hormone-induced lipolytic activity and most mitochondrial gene expression (except for Atp5b and mt-Cyb) were not affected in iBAT of Sik2 KO mice at thermoneutrality (Supplementary Fig. 10G-I). These data indicates that Sik2 KO mice, at thermoneutrality, have increased Ucp1 expression and thermogenic capacity without affecting lipolysis and mitochondrial biogenesis in iBAT.
CL promotes BAT adaptive thermogenesis in vivo through a cAMP-and Ucp1-dependent mechanism 47 . Eight consecutive days of CL injection increased cold resistance in WT mice previously housed at thermoneutrality; however, CL administration had no additive effect on Ucp1 expression and cold resistance in Sik2 KO mice (Fig. 6E-F), which suggested that Sik2 inhibition may be the key downstream event of βAR-cAMP signaling to promote adaptive thermogenesis. To further determine whether this cold-resistance phenotype in Sik2 KO mice was due to enhanced adaptive thermogenesis, not by the other means (such as shivering thermogenesis), we generated Sik2;Ucp1 double KO mice to examine whether cold-resistance in Sik2 KO mice is through an Ucp1-dependent mechanism. Ucp1 is indispensable for BAT-mediated adaptive thermogenesis 9,10 , and both Ucp1 KO mice and Sik2;Ucp1 double KO mice were unable to maintain their core temperatures upon cold challenge (Fig. 6D), indicating that Ucp1 is necessary for enhanced adaptive thermogenesis in Sik2 KO mice at thermoneutrality. Despite elevated thermogenic capacity, Sik2 KO mice gained similar body weight under HFD at thermoneutrality (data not shown). Since the Sik2 global knockout mouse model was employed in this study, we cannot rule out the possibility that Sik2 expression in non-adipose tissues could regulate adiposity through different mechanisms.

Hdac4 deficiency alone in brown adipocytes does not affect adaptive thermogenesis.
Previously we have demonstrated that class IIa histone deacetylases (class IIa HDACs) and CREB regulated transcription coactivator (CRTCs) were functional SIK substrates and represented two cAMP-dependent transcriptional responses 32,34 . We then investigated whether class IIa HDACs can activate Ucp1 expression and adaptive thermogenesis in BAT. In in vitro differentiated brown adipocytes, FSK robustly elevated Ucp1 mRNA levels, which was blocked by co-treatment with a class IIa HDAC inhibitor, LMK235 (Supplementary Fig. 11A). Hdac4 activity was inhibited by LMK235, since Glut4, a glucose transport suppressed by class IIa HDACs in adipocytes 48 , was increased upon LMK235 treatment ( Supplementary Fig. 10A). This data suggested that Hdac4 activity was required for cAMP-induced Ucp1 expression in vitro. We have showed that Hdac4 in iBAT were hypo-phosphorylated and active in response to sympathetic inputs and in Sik2 KO mice (Fig. 4B-D). However, mice with BAT-specific deletion of Hdac4 (Ucp1-Cre;Hdac4 f/f ,Hdac4 BKO mice) showed no change in thermogenic gene expression in iBAT ( Supplementary Fig. 11B,C). Furthermore, Hdac4 BKO mice showed normal basal and CL-induced oxygen consumption, RER, food intake and physical activity ( Supplementary Fig. 11D-I). These data may indicate that other class IIa HDACs (Hdac5/7/9) and/or CRTCs are needed for optimal cAMP-induced adaptive thermogenesis in BAT. Further studies are warranted to address the roles of these cofactors in adaptive thermogenesis in BAT.

Discussion
Defective adaptive thermogenesis is often associated with obesity. Several mouse models with defective thermogenesis, for example, the βless mice, were prone to HFD-induced obesity and hepatosteatosis 23 . Although Gnas BKO mice showed similar thermogenic defects as the βless mice, they didn't show accelerated obesity under HFD. It has been reported that total adipocyte-specific Gnas knockout mice (Gnas AKO ) also showed defective thermogenesis without the development of obesity 42 . Thus, the metabolic abnormalities in the βless mice might be not due to cAMP signaling deficiency in adipose tissues. Although cAMP deficiency in BAT does not lead to drastic obesity, it is plausible that augmenting cAMP signaling in BAT on the other settings may beneficially affect energy homeostasis. Nevertheless, our study clearly demonstrates that cAMP signaling is vital for BAT thermogenic activity.
The beige adipocytes scattered within WAT also require cAMP signaling for their formation, maintenance and function 49 , despite differences in anatomical structures, developmental origins, and gene signatures compared to classical brown adipocytes in iBAT [50][51][52][53] . The brown adipocytes in iBAT were directly innervated by SNS, but the WAT is sparsely innervated; only 6% adipocytes are in contact with sympathetic nerves 54 . The sympathetic nerve runs along with capillary and may be in contact with various cell types within adipose tissues, such as white adipocytes, cells within capillary (pericyte and endothelial cell), adipocyte progenitors, patrolling immune cells and others 40,55,56 . Many non-adipocyte cell types, such as endothelial cells, regulatory T cells, and macrophages, may respond to SNS-released catecholamine and synthesize more catecholamines in WAT, functioning as an amplifier to augment cold-induced catecholamine production and consequently beige adipocyte biogenesis in WAT [57][58][59][60] . Another model to propagate sympathetic neuronal signaling in WAT is through cAMP intercellular transfer through connexin 43-mediated gap junction in adipocytes 61 . We noticed significant difference in beige adipocyte formation in iWAT between Gnas BKO and Gnas AKO mice. Cold-induced beige adipocyte formation is abolished in adipocyte-specific Gnas knockout mice 42 , but not in Gnas BKO mice, suggesting the presence of a white adipocyte-beige adipocyte communication mechanism. This is consistent with our recent report that Liver kinase b1-class IIa Hdac4 signaling in white adipocytes can regulate beige adipocyte renaissance non-cell autonomously 37 .
This study has also illustrated a core genetic program, consisting of cAMP and SIK, in brown adipocytes that mainly controls Ucp1 transcription and thermogenic capacity in response to cold stimulation (Fig. 7). This program does not affect the acute response of cAMP signaling, such as lipolysis (mediated by PKA-dependent activation of HSL) in brown adipocytes. Many stimuli may activate adaptive thermogenesis in brown adipose tissue through this mechanism. For example, fasting inducible hepatokine, fibroblast growth factor 21 (Fgf21), can promote adaptive thermogenesis through sympathetic activation 62 , and serum Fgf21 levels are positively correlated with brown fat activity in humans 63 . Purinergic signaling, particularly, the ATP released from SNS can be converted to adenosine, and then increase adaptive thermogenesis via engaging the adenosine A2A receptor and cAMP signaling in brown adipocytes 64 . It is tempting to speculate that many of these stimuli, if not all, can suppress SIK activity in brown adipocytes to promote Ucp1 expression and adaptive thermogenesis.
Muraoka M et al. has demonstrated that overexpression of Sik2.S587A, a mutant that is refractory of cAMP-mediated suppression, suppressed expressions of thermogenic genes in brown adipocyte cell line T37i. Furthermore, transgenic mice expressing in brown adipocytes had lower Ucp1 and Pgc1α expression in the iBAT and exhibited defective adaptive thermogenesis at room temperature 65 . This Sik2 gain-of-function mouse model resembles the Gnas BKO mouse model regarding their thermogenic phenotypes; they both show reduced Ucp1 expression, mitochondrial biogenesis, and impaired adaptive thermogenesis. Sik2 loss-of-function mouse model, such as Sik2 global KO mice, did not exhibit any significant differences in thermogenic gene expression and activity at RT, likely due to compensation from Sik1. But at thermoneutrality (without sympathetic inputs to brown adipocytes), Sik2 deficiency alone is sufficient to promote transcription of thermogenic genes. Similarly, Sik2 deficiency in A y /a mice rescued the melanogenesis defect in melanocytes 66 , suggesting that hyperactivation of Sik2 might be a causal factor for abnormalities caused by cAMP signaling deficiency in different tissues. SIK belongs to the AMPK-related kinases and shares similar substrates as AMPK. However, SIK has different activity profile as AMPK in different physiological conditions. For example, glucagon during fasting can suppress SIK through cAMP signaling in the liver within minutes 34,67 . But activation of AMPK occurs at the later point of fasting due to nutrient depletion and the increased AMP/ATP ratio 68,69 . Similarly, SIK activity is acutely suppressed by cold stimulation, while AMPK is activated under chronic cold exposure in BAT 70,71 . Indeed, adipocyte-specific AMPK knockout mice exhibited reduced Ucp1 expression and defective thermogenic response to CL 72 , which is opposite to the phenotype observed in SIK deficient mice. Therefore, it is plausible that SIK and AMPK regulate two distinct processes needed for optimal adaptive thermogenesis in BAT. Upon acute cold stimulation, SIK is rapidly inactivated by PKA-mediated phosphorylation to promote Ucp1 expression to boost up thermogenic capacity. Then AMPK activation is needed to maintain mitochondrial homeostasis (independently of Pgc1α) to cope with the sustained cold environment. Therefore, whether inhibiting SIK (particularly Sik2 in adipose tissue) alone or in a combination of AMPK activators may potentially regulate energy balance in an obesogenic environment requires further investigations.

Materials and Methods
Mouse models. All animal experiments were approved by the UCSF Institutional Animal Care and Use Committee in adherence to US National Institutes of Health guidelines and policies. Adiponectin-Cre mice were obtained from The Jackson Laboratory (#028020). Hdac4 f/f , and Ucp1-Cre mice in C57bl/6 J background were kindly provided by Drs. Eric Olson and Evan Rosen, respectively. Gnas f/f mice in 129S6/SvEvTac Black Swiss background were provided by Dr. Lee S Weinstein 42 . The βless mice were provided by Dr. Shingo Kajimura. Sik1 null mice in C57bl/6 N background were obtained from UC Davis KOMP repository (Sik1 tm1(KOMP)Vlcg ), Sik2 null mice in C57bl/6 J background were provided by Dr. Hiroshi Takemori 66 . Mice were housed in a temperature-con- Metabolic studies. ~6-8-week-old male mice were fed with a 60% fat diet (Research Diets, D12492) for additional 6 weeks at room temperature (RT) or thermoneutrality (30 °C). For HFD at thermoneutrality, 5-week-old mice were housed in a 30 °C rodent chamber for 3-4-weeks prior to starting HFD. Body weight was monitored once a week. To measure in vitro lipolysis, mice were fasted for 6 hours and 20 mg of fat tissue was incubated at 37 °C in modified Krebs-Ringer buffer (121 mM NaCl, 5 mM KCl, 0.5 mM MgCl, 0.4 mM NaH 2 PO 4 , and 1 mM CaCl 2 ) supplemented with 1% fatty acid free BSA, 0.1% glucose, and 20 mM HEPES. Glycerol content in the buffer before and after 20 µM Forskolin (FSK) was determined using Infinity Triglycerides Reagent (Thermo, TR22421). To measure in vivo lipolysis, mice were fasted for 6 hours and serum glycerol levels were measured before and after 1 mg kg −1 CL injection.
Cold tolerance test (CTT). ~6-8-week-old male and female mice were single-housed with free-access to food and water during CTT. Rectal temperature was measured hourly with a BAT-12 Microprobe Thermometer (Physitem Instruments) during 4 °C cold challenge.
The homogenates were filtered through cheesecloth to remove residual particulates. The intact mitochondria were isolated by differential centrifugation. The mitochondrial pellet was resuspended in 25 μL of isolation buffer and mitochondrial protein content was quantitated using the Bradford assay.
Mass spectrometry. The pellets of purified BAT mitochondria from 10-12-week old male mice housed at RT or thermoneutrality (n = 3 for each genotype/condition) were resuspended in 8 M urea, 50 mM Tris, 5 mM CaCl 2 , 100 mM NaCl, and protease inhibitors. Mitochondria were lysed by probe sonication on ice, and proteins were reduced by the addition of 5 mM DTT for 30 min at 37 °C, followed cysteine alkylation by the addition of 15 mM iodoacetamide at RT for 45 min in the dark. The reaction was then quenched by the addition of 15 mM DTT for 15 minutes at RT. Proteins were first digested by the addition of endoproteinase LysC (Wako LC) at a 1:50 substrate:enzyme and incubated for 2 h at RT. Next, samples were further digested by the addition of trypsin (Promega) at 1:100 substrate:enzyme, and incubated overnight at 37 °C. Protein digests were then acidified by the addition of 0.5% trifluoracetic acid, and samples desalted on C18 stage tips (Rainin). Peptides were resuspended in 4% formic acid and 3% acetonitrile, and approximately 1 μg of digested mitochondria proteins was loaded onto a 75 μm ID column packed with 25 cm of Reprosil C18 1.9 μm, 120 Å particles (Dr. Maisch). Peptides were eluted into a Q-Exactive Plus (Thermo Fisher) mass spectrometer by gradient elution delivered by an Easy1200 nLC system (Thermo Fisher). The gradient was from 4.5% to 31% acetonitrile over 165 minutes. All MS spectra were collected with oribitrap detection, while the 15 most abundant ions were fragmented by HCD and detected in the orbitrap. All data were searched against the Mus musculus uniprot database (downloaded July 22, 2016). Peptide and protein identification searches, as well as label-free quantitation were performed using the MaxQuant data analysis algorithm, and all peptide and protein identifications were filtered to a 1% false-discovery rate 38,39 . Flow cytometry. The eWAT were minced and then digested in 2 ml of digestion buffer (2 mg ml −1 at 250U/mg, Worthington, and 30 mg ml −1 bovine serum albumin in Hams F-10 medium) at 37 °C for ~60 minutes. The homogenates were washed with PBS and filtered through a cell strainer ( Isolation of genomic DNA. Total DNA was isolated using QIAamp DNA mini kit (Quiagen) from fifty miligrams of eWAT frozen tissue from HFD mice as previously described 73 . Quantification was performed using a Fisher spectrophotometer at 260 nm. mtDNA Quantification. The relative mtDNA content was measured using real-time qPCR. The β2 microglobulin gene (B2M) was used as the nuclear gene (nDNA) normalizer for calculation of the mtDNA/ nDNA ratio. The relative mtDNA content was calculated using the formula: mtDNA content = 1/2 ΔCt , where ΔC t = C t mtDNA − C t B2M .
Statistical analysis. We used GraphPad Prism 6.0 to assess data for normal distribution and similar variance between groups. Data were presented as the mean ± s.e.m. Statistical significance was determined using a unpaired two-tailed Student's t test with unequal variance, or one-way ANOVA between multiple groups: ns: not significant, *p < 0.1, **p < 0.05 and ***p < 0.01. We selected sample size for animal experiments based on numbers typically used in similar published studies. We did not perform randomization of animals or predetermine sample size by a statistical method. In vitro measurements of glycerol and FFA were done with 3 technical replicates.
Data availability. Mass spectrometry dataset of BAT mitochondrial proteome from control and Gnas BKO mice was deposited to the ProteomeXchange Consortium (http://proteomecentral.proteomexchange.org) via the PRIDE partner repository under accession number PXD009262.