SIK2 Is Critical in the Regulation of Lipid Homeostasis and Adipogenesis In Vivo

White adipose tissue (WAT) plays a critical role in the maintenance of energy homeostasis in mammals. Under feeding conditions, excessive nutrients are converted into the triacylglycerol in the liver and transported primarily as a form of VLDL to WAT, a preferred fuel depot in the mammalian system (1). At the same time, pancreatic hormone insulin promotes GLUT4 to the plasma membrane of WAT, thus increasing glucose uptake of this cell (2). Glucose is then converted into glycerol, fused with the free fatty acid (FFA) taken up from the VLDL to form triglycerides. In time of starvation or prolonged fasting, after the depletion of glycogen in most tissues, increased concentration of circulating catecholamine promotes lipolysis from WAT as an FFA, providing a new fuel source for the peripheral tissue (3–6).

Catecholamine binds to and activates β-adrenergic receptor in the fat cell, triggering a cAMP-dependent pathway in the cell. cAMP-dependent protein kinase (PKA) is responsible for the phosphorylation and activation of hormone-sensitive lipase (HSL), thus providing the molecular mechanism for cAMP-dependent increase of lipolysis (7). PKA is also critical in activating cAMP-dependent transcription by promoting Ser¹³³ phosphorylation of CREB (8–10). A CREB-dependent pathway is shown to be involved in the repression of GLUT4 and adiponectin expression not only during fasting conditions but also by insulin resistance, providing a potential molecular mechanism for the chronic transcriptional regulation of this metabolic phenotype (11–15).

Transcriptional regulation of CREB requires involvement of transcriptional coactivators including CBP/p300, which specifically binds to phospho-CREB on the promoter. In addition, another class of CREB coactivator, CREB-regulated transcription coactivator (CRTC), has been delineated recently (16,17). In the liver and pancreatic β-cells, CRTC2, a major isoform of CRTCs found in these cell types, is regulated in a PKA-dependent manner (18,19). In the context of the low cAMP concentration such as feeding conditions in the liver, activation of salt-inducible kinases (SIKs) leads to the increased phosphorylation of Ser¹⁷¹ of CRTC2, which in turn leads to the increased association of this factor with 14-3-3, resulting in the cytoplasmic retention. Fasting hormone glucagon-mediated activation of PKA leads to the activation of Ser/Thr phosphatases such as protein phosphatase 2B and protein phosphatase 4 and the repression of SIK kinases, resulting in the dephosphorylation and nuclear localization of CRTC2 and activation of CREB target genes (20,21). The role of CREB in response to catecholamine-dependent activation of PKA pathway in the fat cells was also recently delineated (22). CREB has been shown to play a role in activating activating transcription factor (ATF)3–dependent transcriptional pathway in white adipocytes; the exact role of CRTC2 or SIK2 in the control of signaling pathways in this tissue, however, has not been delineated to date.

Here, we provided the in vivo evidence regarding the crucial role of SIK2 in the regulation of adipocyte function and systemic insulin signaling. SIK2 knockout (SIK2 KO) mice displayed impaired glucose and insulin tolerance, with higher blood glucose and triglyceride levels compared with the wild-type (WT) mice, without affecting total adiposity of the animals. Unlike the expectation, the chronic depletion of SIK2 did not affect glucose metabolism in the liver or in the isolated primary hepatocytes. On the other hand, SIK2 KO adipocytes displayed the increases in cell size, with concomitant increases in macrophage infiltration, showing a sign of insulin resistance in this tissue. Indeed, depletion of SIK2 promoted the CRTC2-CREB transcriptional pathway, which led to the increased expression of ATF3 and subsequent downregulation of GLUT4 and adiponectin in fat cells both in vitro and in vivo. On the other hand, CRTC2 knockout adipocytes displayed improved glucose uptake that was associated with restoration of GLUT4 and adiponectin expression. We suggest that SIK2 is critical in controlling a CRTC2-CREB–dependent transcriptional circuit in the white adipocytes to control whole-body insulin sensitivity.

Male 8-week-old C57BL/6 mice were purchased from Charles River Laboratories International. SIK2 KO (SIK2^−/−) mice were obtained from Lexicon Pharmaceuticals, and liver kinase B1 (LKB1)-floxed (LKB1 fl/fl) mice were obtained from National Cancer Institute (NCI) mouse repository. CRTC2 knockout (CRTC2^−/−) mice were generated by Macrogen by using CRTC2 knockout ES cell clones from Sanger Institute. Mice were backcrossed with C57BL/6 five times before being used for the experiment and housed in a specific pathogen-free animal facility at the Sungkyunkwan University School of Medicine (12:12-h light-dark cycle). For induction of obesity and insulin resistance, male 4- to 8-week-old mice were fed a high-fat diet (60% fat diet, measured in kcal, cat. no. D12492; Research Diets) for 16–20 weeks. For animal experiments involving adenoviruses, mice were tail vein injected with recombinant adenovirus (0.1–0.5 × 10⁹ pfu per mice) as previously described (23). For measurement of fasting blood glucose level, animals were fasted for 16 h or 6 h with free access to water. For the glucose tolerance test (GTT), 16-h-fasted mice were injected with glucose (2 g/kg body wt i.p. for normal chow diet and 1 g/kg body wt i.p. for high-fat diet). For the insulin tolerance test, 6-h-fasted mice were injected with 0.5 units of insulin/kg body wt i.p. (normal chow diet) or 1 unit of insulin/kg body wt i.p. (high-fat diet). Blood glucose and insulin levels were measured from tail vein blood collected at the designated times. For activation of the insulin signaling pathway, either PBS (for control) or insulin (0.5 units per mouse) was injected intraperitoneally for 10 min before the collection of liver and WAT for further analyses. All procedures were approved by the Sungkyunkwan University School of Medicine Institutional Animal Care and Use Committee or by the Korea University Institutional Animal Care and Use Committee.

Primary hepatocytes were prepared from mice by the collagenase perfusion method as previously described (24). Cells were plated in medium 199 (Sigma-Aldrich) supplemented by 10% FBS, 10 units/mL penicillin, 10 mg/mL streptomycin, 23 mmol/L HEPES, and 10 nmol/L dexamethasone. After attachment, cells were treated with drugs indicated in the figure legends.

The subcutaneous fat pads of mice were minced and digested for 30 min at 37°C in collagenase buffer (pH 7.4, 1 mg/mL type 1 collagenase, 100 mmol/L HEPES, 125 mmol/L NaCl, 5 mmol/L KCl, 1.3 mmol/L CaCl₂, 5 mmol/L d-glucose, and 2% BSA). After filtration of the digested tissues through a 300-µm nylon mesh, the filtrate was centrifuged at 500g for 10 min to separate floating adipocytes from the stromal vascular fraction pellet. The supernatant was discarded, and the pellet containing stromal vascular fraction cells was filtrated through a 40-µm cell strainer. After centrifugation, the pellet was resuspended in red blood cell lysis buffer (pH 7.4, 0.16 mol/L NH4Cl, 0.017 mol/L Tris, and 0.01 mol/L EDTA) and incubated for 10 min at room temperature. The pellet was centrifuged, and then cells were plated in DMEM (Hyclone) supplemented by 10% FBS, 10 units/mL penicillin, and 10 mg/mL streptomycin as described previously (25).

For the differentiation, cells were incubated in DMEM with 10% FBS for 48 h after confluence. Cells were then cultured in DMI media (DMEM, 10% FBS, 10 units/mL penicillin, 10 mg/mL streptomycin, 5 μg/mL insulin, 1 μmol/L dexamethasone, and 115 μg/mL IBMX) for 2 days and were maintained in fresh insulin media (DMEM, 10% FBS, 10 units/mL penicillin, 10 mg/mL streptomycin, and 5 μg/mL insulin) that were replenished every 2 days for the duration of the study. Cells were differentiated for total of 8–10 days.

All recombinant lentiviruses were produced by transient transfection of 293T cells according to standard protocols (26). Briefly, subconfluent 293T cells were cotransfected with 4 μg pLKO.1-US, pLKO.1-SIK2 RNAi, pLKO.1-CRTC2 RNAi, pLPCX, or pLPCX-HA SIK2, together with 3 μg pCMV-ΔR8.91 and 1 μg pMD2G-VSVG. After 16 h, 4 mmol/L sodium butyrate was added into the fresh media, and recombinant lentivirus was harvested 48 h posttransfection.

For stable knockdown or overexpression, 3T3 L1 cells were infected with unspecific control (US), SIK2 RNAi, CRTC2 RNAi, overexpression control, or SIK2 WT lentiviruses and then selected with puromycine. Single clones were analyzed for the reduced expression of SIK2 for the subsequent studies. For the acute knockdown and overexpression of SIK2, 3T3 L1 cells expressing mouse coxsackievirus and adenovirus receptor were used (27).

Total RNA from cells and tissues were extracted using an RNA extraction kit (Intron). cDNA generated by reverse transcriptase (GenDEPOT) was analyzed by quantitative PCR using an SYBR green PCR kit and a TP 800 Thermal Cycler Dice model (Takara). Primers were designed using Oligoperfect Designer (Invitrogen). All data were normalized to ribosomal L32 expression.

Western blot analyses on 10–60 μg protein extracts from cells and tissue were performed as previously described (28). SIK2 antibody was purchased from R&D system. Antibodies for LKB1, phospho-Thr¹⁷² AMPKα, AMPKα, phospho-Ser⁴⁷³ AKT, phospho-Thr³⁰⁸ AKT, AKT, phospho-Thr⁶⁴² AS160, and insulin receptor substrate 1 were obtained from Cell Signaling Technology. Antibodies for suppressor of MEK null 1 (SMEK1) and SMEK2 were provided from Novus Biologicals. Antibody for phospho-Tyr^1162/1163 insulin receptor (IR) and CRTC2 was from Calbiochem. IR antibody was purchased from Santa Cruz. Phospho-Tyr⁶¹² IRS-1 antibody was purchased from Sigma-Aldrich. HSP90α/β (Santa Cruz), β-actin, and α-tubulin (Sigma Aldrich) were used to assess equal loading.

Lipolysis was measured as rate of glycerol release over a 2-h period. For experiments, the differentiated 3T3 L1 cells or primary adipocytes were incubated with serum-free DMEM for 8 h and then washed two times with 1× PBS. Cells were further incubated with Krebs-Ringer buffer containing 2% BSA for 2 h with or without forskolin (20 μmol/L). Media were collected, heated at 65°C for 8 min, and assayed by using a free glycerol assay kit (Abcam).

The differentiated 3T3 L1 cells and primary adipocytes were incubated with serum-free DMEM for 6 h and then washed two times with KRPH buffer (pH 7.4, 20 mmol/L HEPES, 5 mmol/L KH2PO4, 1 mmol/L MgSO4, 1 mmol/L CaCl2, 136 mmol/L NaCl, and 4.7 mmol/L KCl). They were further incubated for 30 min with KRPH buffer containing 2% BSA and treated with or without 1 μmol/L insulin for 20 min, after which 1 mmol/L 2-deoxy-D-glucose (2DG) was incubated for 20 min. Cells were collected in 10 mmol/L Tris-HCl (pH 8.0) containing 1% Triton X-100 and heated at 95°C for 15 min. After centrifugation, the supernatant was analyzed for 2DG-6-phosphate (2DG6P) content using a 2DG uptake assay kit (Cosmo Bio Co., Ltd.). For measurement of 2DG uptake in the skeletal muscle, the muscles were isolated from nonfasted WT and SIK2 KO mice. Isolated muscles were individually incubated with KRPH buffer containing 2 mmol/L sodium pyruvate for 30 min and then treated with or without 1 mU/mL insulin for 15 min. Afterward, 8 mmol/L 2DG was added into the media for an additional 30 min at the end of the incubation period, and the muscles were removed from the buffer, gently rinsed with 1× PBS, and immediately frozen in liquid nitrogen. Each frozen muscle was homogenized and prepared for the measurement of 2DG as previously described (29). The content of 2DG6P was measured by using a 2DG uptake assay kit (Cosmo Bio Co., Ltd.).

3T3 L1 cells or primary adipocytes were differentiated for 6 days and then the cells were maintained to DMEM/FBS alone for a further 4 days. The differentiated cells were incubated with serum-free DMEM for 1 h before treatment with or without 1 μmol/L insulin for 30 min. The same volume of QBT fatty acid uptake solution (QBT Fatty Acid Uptake Assay kit; Molecular Devices) as is found in serum-free DMEM was added to the well, and the plates were immediately transferred to a fluorescence microplate reader (excitation 488 nm/emission 515 nm) for kinetic reading (every 20 s for 30 min). Data represent the mean of the area under under the curve (0–90 min).

4-h-fasted WT and SIK2 KO mice were injected with 500 mg/kg i.p. Triton WR1339 (Tyroxapol; Sigma Aldrich) (15 g/dL in 0.9% NaCl solution). Blood samples were collected at 30, 60, and 90 min after injection. LDL/VLDL levels were measured by a colorimetric assay kit (Abcam).

Primary hepatocytes isolated from WT and SIK2 KO mice were seeded at 2 × 10⁴ cells per well on XF-24 cell culture plates (Seahorse Bioscience) coated with matrigel (BD Biosciences) in M199 medium (Sigma). After 4-h incubation, cells were equilibrated with KRB buffer (pH 7.4, 110 mmol/L NaCl, 4.7 mmol/L KCl, 2 mmol/L MgSO₄, 1.2 mmol/L Na₂HPO₄, 2.5 mmol/L glucose, and 0.5 mmol/L carnitine) and incubated at 37°C for 1 h without CO₂. For induction of fatty acid oxidation, palmitate-BSA complex was injected at a final concentration of 100 μmol/L into XF-24 cartridge (Seahorse Bioscience). Fatty acid oxidation capacity was represented as increased oxygen consumption rate in response to the palmitate-BSA complex.

Cells were washed twice with PBS. Monolayer cells were fixed on dishes with 3.7% formaldehyde in PBS for 30 min. Cells were stained with 40% isopropanol containing 0.5% Oil red O for 1 h, followed by rinsing in 60% isopropanol and dH₂O to remove unbound dye. For quantification of lipid accumulation, Oil red O was extracted with 100% isopropanol, and the optical density of the solution was detected at 500 nm.

The epididymal fat pads were isolated from mice and stained with hematoxylin-eosin (H-E) for morphological analysis. Adipocyte size was measured with the use of MACSCOPE software (Mitani Sangyo Co. Ltd.).

For measurement of macrophage infiltration, adipose tissues from mice fed a high-fat diet were fixed for 24–48 h with 10% formalin, dehydrated, and embedded in paraffin. After blocking with PBS containing 1% BSA, sections were subjected to immunohistochemical staining overnight at 4°C with a 1:150 dilution of a rat monoclonal antibody to F4/80 (Abcam). A biotin-conjugated rabbit polyclonal to rat IgG was used at 1:100 dilution as secondary antibody. Slides were observed with a light microscope.

Body composition of 8-week-old male, normal diet–fed WT and SIK2^−/− mice was determined using MRI or DEXA as previously described (30–32). Food intake, locomotor activity, respiratory quotient, and energy expenditure of normal diet–fed WT and SIK2^−/− mice were analyzed using a metabolic cage (Harvard Apparatus; Panlab). For the measurement of metabolites, 4-h-fasted mice were killed for the terminal blood collection. Blood glucose for basal conditions was monitored from tail vein blood using an automatic glucose monitor (OneTouch; LifeScan, Inc.). Plasma and liver triglyceride (TG) levels and plasma nonesterified fatty acids were measured by colorimetric assay kits (Waco). Insulin, adiponectin, and cytokines (tumor necrosis factor [TNF]-α, interleukin [IL]-1β) were measured by mouse ELISA kits (insulin and adiponectin, ALPCO Diagnostics, cytokine, Invitrogen).

Total liver lipids were extracted with chloroform:methanol (2:1, v/v) mixture according to Folch method (33).

Results are shown as mean ± SD for quantitative PCR and quantitation of protein levels and as mean ± SEM for measuring metabolites, as indicated in the figure legends. The comparison of different groups was carried out using two-tailed unpaired Student t tests, and a P value <0.05 was considered statistically significant and reported as in legends.

SIK2 has been implicated in various signaling cascades such as glucose and fat metabolism in the liver, neuroprotective effect in the brain, and the eumelanin synthetic pathway in melanocytes (34–38). To invariably delineate the functional consequences of the chronic depletion of SIK2 in vivo, we generated knockout mice for SIK2 by deleting two critical exons (exons 2 and 3) from the genomic sequences, resulting in >90% depletion of SIK2 in most tissues as confirmed by quantitative PCR and Western blot analysis (Fig. 1A and B and Supplementary Fig. 1A). As expected, SIK2 KO mice displayed a slight increase in blood glucose levels and a mild increase in serum insulin levels that were associated with impaired glucose tolerance and insulin tolerance without changes in body weight in the context of both normal diet and high-fat diet feeding (Fig. 1C–H and Supplementary Fig. 1B–D). The plasma insulin levels during the GTT were not different between SIK2 KO mice and the control, excluding the potential changes in insulin secretion from the pancreatic β-cells that could affect the result of GTT (Supplementary Fig. 1E). To further analyze the potential metabolic phenotypes in mice that were associated with SIK2 depletion, we conducted indirect calorimetry. During the assay period, no differences in body weight were observed between WT and SIK2 KO mice, although a slight decrease in food intake and an increase in locomotor activity were observed in SIK2 KO mice compared with the control (Supplementary Fig. 2A–C). In spite of no significant changes in UCP1 or peroxisome proliferator–activated receptor (PPAR)γ coactivator-1α expression in SIK2 KO brown adipocytes compared with the control (Supplementary Fig. 2D), we observed decreased respiratory quotient and decreased energy expenditure in SIK2 KO mice compared with control (Supplementary Fig. 2E and F), providing the explanation for the lack of changes in body weight between the two groups.

Unexpectedly, both pyruvate challenge test and hyperinsulinemic-euglycemic clamp studies did not provide information regarding changes in hepatic glucose production, suggesting that chronic depletion of SIK2 might not promote changes in glucose metabolism in the liver (data not shown). Even though we detected changes in gluconeogenic gene expression in the liver of SIK2 KO mice compared with control both fed a normal diet and a high-fat diet, primary hepatocytes from SIK2 KO mice showed no basal or forskolin-induced changes in expression of gluconeogenic genes (Supplementary Fig. 3A–C). As acute depletion of SIK2 in the liver promotes enhanced activation of CRTC2, a critical transcriptional regulator for hepatic gluconeogenesis, we wanted to detect whether chronic depletion of SIK2 also leads to the activation of this coactivator. Surprisingly, phosphorylation status of CRTC2 was not greatly perturbed upon chronic depletion of SIK2 both in the liver and in primary hepatocytes—in stark contrast to the results obtained from acute adenovirus-mediated knockdown studies (34) (Supplementary Fig. 3D–F). On the other hand, chronic depletion of LKB1, an upstream kinase for AMPK and its related kinase (AMPKRK) families including SIK2, led to the increased appearance of dephosphorylated CRTC2 that was associated with increased expression of gluconeogenic genes, suggesting that other AMPKRK might be more important in the regulation of CRTC2 and hepatic gluconeogenesis (Supplementary Fig. 4A and B). A slight increase in the expression of SIK1 or SIK3 in SIK2 KO liver could also suggest that this compensatory increase in other AMPKRK might prohibit dysregulation of the CRTC2-dependent pathway in the absence of SIK2 (Supplementary Fig. 4C and D).

We observed increased plasma triglyceride levels in SIK2 KO mice compared with control, which suggests that lipid metabolism might be perturbed by depletion of SIK2 (Fig. 2A and B). Expression of genes involved in lipogenesis such as SREBP-1c and FAS was increased in the liver of SIK2 KO mice compared with control, suggesting that this kinase is more closely connected with fatty acid biosynthesis as previously described (35) (Fig. 2C and D). However, the TG levels in the liver of WT mice and those of SIK2 KO mice were not different when fed the normal diet or the high-fat diet, showing that changes in lipogenic gene expression per se might not lead to the lipid accumulation in the liver (Fig. 2E and F). While LDL/VLDL secretion from the livers of SIK2 KO mice and WT mice was not different (Fig. 2G), we observed decreased fatty acid uptake and increased lipolysis from primary adipocytes of SIK2 KO mice compared with those of control (Supplementary Fig. 5A and B), drawing an explanation for the resultant elevation of plasma triglyceride levels in SIK2 KO mice. Consistent with the results showing no change in body weight between SIK2 KO mice and WT mice, we did not observe changes in either adiposity or total fat mass (by MRI and DEXA analysis) (Supplementary Fig. 6) in SIK2 KO mice compared with control.

When we analyzed the morphology of white adipocytes, however, we observed the increased population of larger fat cells for SIK2 KO mice compared with control both under normal diet and high-fat diet feeding conditions (Fig. 3A and Supplementary Fig. 7A). While expression of lipogenic genes was generally increased, expression of PPARγ was rather decreased, suggesting that the insulin sensitivity in this tissue might be compromised (Fig. 3B) (39). Furthermore, increased infiltration of macrophage into the fat cells was apparent with SIK2 KO, as evidenced by immunohistochemistry using antibody against F4/80 and quantitative PCR analysis measuring expression of macrophage-specific genes (Fig. 3C and D). In conjunction with the data, we observed increased expression of genes for proinflammatory cytokines such as MCP-1 and IL-1β, suggesting that increased association of macrophage could lead to the inflammatory response in white adipocytes from SIK2 KO mice (Fig. 3E). The plasma cytokine levels were only slightly elevated in SIK2 KO mice compared with the control, hinting that the inflammatory response, and perhaps its effect on the insulin signaling, might be localized to the white adipocytes (Fig. 3F). In line with this notion, we observed decreased phosphorylation of Akt and AS160 in white adipocytes, but not in the liver and the skeletal muscle, of SIK2 KO mice compared with control (Fig. 3G and Supplementary Fig. 7B and C). We observed a slight increase in tribbles homolog 3 (TRB3) expression in SIK2 KO adipocytes compared with WT control, which could account for changes in AKT signaling without apparent changes in tyrosine phosphorylation of either insulin receptor or insulin receptor substrate 1 in this tissue (Supplementary Fig. 8A). In line with this observation, TRB3 expression in the skeletal muscle was not different between the two groups (Supplementary Fig. 8B).

Next, we wanted to delineate the molecular mechanism by which deletion of SIK2 promoted an increase in fat cell size in SIK2 KO mice. Recently, it was reported that chronic inhibition of CREB activity in fat cells would promote decreased expression of ATF3, preventing the adipogenesis and the formation of larger fat cells, resulting in the improvement of insulin signaling, suggesting that activation of the CREB-dependent transcriptional pathway might cause the current phenotypes of SIK2 KO adipocytes (22). Indeed, we were able to observe the increased appearance of dephosphorylated CRTC2, an active form of CREB coactivator, in white adipocytes from SIK2 KO mice (Fig. 4A and B). In line with the increased transcriptional potential of a CREB-dependent pathway, we were able to observe the increased expression of ATF3 (Fig. 4C). In addition, expression levels of GLUT4 and adiponectin, two genes that were shown to be downregulated by ATF3, were significantly reduced in white adipocytes from SIK2 KO mice both under normal diet and under high-fat diet feeding conditions (Fig. 4C and Supplementary Fig. 8C).

We also observed decreases in plasma levels of total and high-molecular weight adiponectin in SIK2 KO mice compared with the control, partially explaining the glucose and insulin intolerance that is associated with these mice (Fig. 4D). In line with this result, we observed reduced phosphorylation of AMPK in the liver and the skeletal muscle of SIK2 KO mice compared with the control (Fig. 5A). Reduced AMPK activity was often associated with the reduction in fatty acid oxidation (40). We observed reduced expression of genes involved in the fatty acid β-oxidation (ACOX1, CPT-1, MCAD, and PPARα) both in the liver and in the skeletal muscle in SIK2 KO mice (Fig. 5B) and the resultant decrease in fatty acid oxidation in primary hepatocytes from SIK2 KO mice compared with the control (Fig. 5C). We also observed the reduction in glucose uptake from the muscle of SIK2 KO mice compared with the control, which could contribute to the pronounced glucose phenotype that was associated with the depletion of SIK2 (Fig. 5D).

Having seen the effect of systemic SIK2 depletion on the cellular signaling in the white adipocytes, we wanted to confirm whether it directly stems from the changes in the property of fat cells. Thus, we prepared 3T3 L1 adipocytes that were stably infected with lentivirus expressing either short hairpin RNA (shRNA) for SIK2 or control shRNA. Indeed, SIK2 knockdown 3T3 L1 cells (SIK2 KD cells) were differentiated into fat cells at a much faster rate than the control cells, as evidenced by the Oil red O staining and increased expression of SREBP-1c and FAS (Fig. 6A and B). As in the case for SIK2 KO mice, SIK2 KD cells showed impaired insulin response, as shown by decreased Akt phosphorylation and 2-deoxyglucose uptake when treated with insulin in the culture media (Fig. 6C and D). SIK2 KD cells exhibited reduced fatty acid uptake and increased lipolysis, traits that resembled those of SIK2 KO primary adipocytes (Supplementary Fig. 9). Furthermore, SIK2 knockdown also greatly enhanced dephosphorylation of CRTC2, which led to the increased expression of ATF3 and the decreased expression of GLUT4 and adiponectin in 3T3 L1 adipocytes (Fig. 6E and F). Conversely, 3T3 L1 cells stably expressing SIK2 showed reduced Oil red O staining and enhanced phosphorylation of AKT and AS160 compared with the control, resulting in increased glucose uptake (Supplementary Fig. 10A–C). In addition, we also observed reduced expression of ATF3 and subsequent increased expression of GLUT4 and adiponectin, showing a direct role of SIK2 in regulating metabolism and insulin signaling in WAT (Supplementary Fig. 10D). These data further support the hypothesis that the depletion of SIK2 directly affects the CRTC2-CREB pathway in the adipocytes to display the peculiar phenotypes that are associated with SIK2 KO mice. The role of SIK2 in adipocyte function is critical in the early stage of differentiation for neither knockdown nor overexpression of SIK2 greatly affected the gene expression or Akt phosphorylation in 3T3 L1 cells after the differentiation (Supplementary Fig. 11A–D).

Next, we wanted to verify whether the intrinsic adipogenic potential of preadipocytes was perturbed upon depletion of SIK2. Thus, we isolated preadipocytes from either WT mice or SIK2 KO mice and grew them under the culture media that could promote adipogenesis in vitro. As in the case for SIK2 knockdown 3T3 L1 cells, preadipocytes from SIK2 KO mice displayed increased expression of lipogenic genes and could differentiate into the adipocytes in a rate that was faster than that from WT mice (Fig. 7A and B). In addition, we were also able to confirm that the resultant adipocytes displayed reduced insulin signaling as shown by decreased Akt phosphorylation and AS160 phosphorylation together with reduced 2-deoxyglucose uptake in response to insulin (Fig. 7C and D). As well, we also observed increased expression of CRTC2, showing indeed that enhancement of the CRTC2-CREB pathway is critical in promoting the formation of insulin-resistant adipocytes (Fig. 7E). Activation of the CREB-CRTC2 transcriptional pathway resulted in the increased expression of ATF3, triggering the reduced expression of both GLUT4 and adiponectin, which would contribute to the observed perturbation in insulin signaling (Fig. 7F).

Finally, we wanted to confirm whether the CRTC2-dependent transcriptional pathway is indeed essential for the SIK2-mediated phenotypes in white adipocytes. Thus, we prepared 3T3 L1 adipocytes that were stably infected with lentivirus expressing either shRNA for CRTC2 or control shRNA. We observed lower lipid contents in CRTC2 knockdown-3T3 L1 cells (CRTC2 KD cells) compared with the control (Fig. 8A). Furthermore, we observed enhanced insulin signaling (as shown by increased phosphorylation of AKT and AS160) as well as elevated glucose uptake in CRTC2 KD cells compared with the control (Fig. 8B and C). In addition, we also prepared the preadipocytes from CRTC2 KO mice and grew them under the media to promote adipogenesis. We observed that the depletion of CRTC2 led to the reduction in expression of lipogenic genes such as SREBP-1c, FAS, and ACC (Supplementary Fig. 12A and B). Furthermore, 2-deoxyglucose uptake tended to be higher in CRTC2 knockout cells compared with control cells, in agreement with the tendency of increase in Akt phosphorylation upon CRTC2 knockout (Supplementary Fig. 12C). Finally, depletion of CRTC2 led to the decreased expression of ATF3 and prompted a reciprocal increase in expression of GLUT4 and adiponectin genes (Supplementary Fig. 12D). These data underscore the importance of SIK2 in the determination of the property of white adipocytes by primarily controlling activity of the CREB-CRTC2–dependent transcriptional pathway.

Obesity is highly associated with the progression of numerous metabolic diseases including type 2 diabetes and atherosclerosis (41,42). Previously, it was shown that the increased levels of plasma TGs or FFA during obesity could be the primary cause for the progression of metabolic diseases by promoting peripheral insulin resistance. Identification of various adipocyte-derived cytokines, termed adipokines, added to the complexity of this simple paradigm (43). Leptin is known to primarily act upon hypothalamus to control feeding behavior by stimulating the JAK-STAT pathway, thus modulating the overall metabolic phenotypes in mammals (44,45). On the other hand, adiponectin exerts its insulin-sensitizing effects by controlling AMPK and PPARα-dependent pathways both in the central and in the peripheral tissues (46–49). More classical adipose tissue–derived cytokines such as TNF-α, IL1-β, and IL-6 would promote insulin resistance in multiple peripheral tissues by eliciting inflammatory response in target tissues (rev. in 43).

Catecholamine-dependent activation of cAMP signaling in the fat cells is known to activate PKA-mediated acute posttranslational modifications. For example, fasting-mediated activation of PKA results in the phosphorylation and activation of HSL, thus promoting increased lipolysis of triglycerides in the fat cells and the release of FFA to the bloodstream (7). A recent report by Montminy and colleagues (22) suggested that the transcriptional pathway is also critical in mediating chronic effects in response to catecholamine in the fat cells. Mice expressing the CREB-inhibitor A-CREB in an adipocyte-specific manner display an insulin-sensitive phenotype that is associated with reduced fat cell size, increased glucose uptake, and increased plasma adiponectin levels. Indeed, they identified that decreased expression of ATF3, a CREB target gene, in the white adipocytes is responsible for the increased expression of GLUT4 and adiponectin in this tissue. These data are in total agreement with our experiments using CRTC2 KD 3T3 L1 cells or CRTC2 knockout adipocytes (Fig. 8 and Supplementary Fig. 12). Conversely, we observed that activation of the CREB-CRTC2 pathway in the white adipocytes by depletion of SIK2 expression leads to increased expression of ATF3 that is associated with decreased expression of GLUT4 and adiponectin genes (Fig. 7). These data strongly support the previous report showing that catecholamine-dependent activation of cAMP-PKA pathway is not only critical in promoting phosphorylation of HSL and the resultant lipolysis but also important in eliciting CREB-CRTC2-dependent transcriptional mechanisms that govern the fate of the white adipocytes.

We observed an increase in expression of proinflammatory cytokine genes in WAT from SIK2 KO mice compared with the control (Fig. 3E), suggesting that either the adipocytes or the nonadipocytes including macrophages in the tissue are responsible for the observed changes. Unlike the case for the WATs, we failed to detect the expression of IL-1β in differentiated preadipocytes from either WT mice or SIK2 KO mice. While we observed an increase in expression of MCP-1 in differentiated SIK2 KO preadipocytes, we also observed a similar increase in expression of MCP-1 in differentiated CRTC2 knockout preadipocytes, suggesting that the changes in expression of proinflammatory cytokine genes might not be due to the transcriptional changes occurring in white adipocytes themselves. Rather, we suspect that the increased expression of proinflammatory cytokine genes was mostly derived from the macrophages of WATs, in which we observed the increased association of M1 macrophages with the deletion of SIK2 in mice (J.P. and S.-H.K., unpublished observations). Further study will be necessary as to whether SIK2 and/or CREB-CRTC2 might be directly involved in the regulation of proinflammatory cytokine expression in the macrophages.

SIK2 belongs to the subset (SIK1–3) of AMPKRKs that are activated by upstream kinases such as LKB1 or calcium/calmodulin-dependent protein kinase kinase-β (50,51). In vitro study as well as in vivo analysis using mice with acute SIK2 depletion revealed that SIK2 could be involved in the regulation of gluconeogenesis and lipogenesis in the liver (35). Previous global knockout mice for SIK2 exhibited a neuroprotective effect in brain by activating CREB target genes such as brain-derived neurotrophic factor (BDNF), PPARγ coactivator-1α, and B cell lymphoma 2 (Bcl-2) and displayed the increased eumelanin synthesis in their melanocytes (36,38). In this study, we investigated the role of SIK2 in the determination of metabolic phenotype in vivo by using chronic knockout mice. Unlike the case for the acute knockdown, we only observed a slight effect of SIK2 depletion on hepatic gluconeogenesis in mice. Furthermore, we did not observe changes in expression of gluconeogenic genes in primary hepatocytes prepared from SIK2 KO mice compared with that from control mice, suggesting that the observed effects on hepatic gene expression in SIK2 KO mice might be indirectly driven by dysregulation of adipocyte signaling or changes in plasma levels of adipokines such as adiponectin. Alternatively, the lack of ostensible effect of chronic SIK2 KO on hepatic gene expression might be potentially due to the compensatory increases in other SIK isoforms in the liver. The latter hypothesis is consistent with the results obtained using primary hepatocytes from LKB1 knockout mice, in that we observed higher expression of gluconeogenic genes and reduced phosphorylation of CRTC2 in LKB1 knockout cells compared with those in control cells. Indeed, expression of both SIK1 and SIK3 are significantly induced upon SIK2 KO, supporting a potential role of these isoforms in the hepatic metabolism in the longer-term absence of SIK2 in mice. On the other hand, depletion of SIK2 led to the increased expression of lipogenic genes both in mouse liver and in primary hepatocytes from SIK2 KO mice, showing the critical role of SIK2 in regulating hepatic lipogenesis, presumably by regulating SREBP-1c or ChREBP activity (23,35). The role of SIK2 in brown adipocytes was also suggested by a study using mice stably expressing aP2-driven transgene for dominant-negative SIK2, suggesting that SIK2 might block the thermogenic potential for the brown adipocytes (52). On the other hand, chronic depletion of SIK2 in brown adipocytes did not perturb expression of thermogenic genes, suggesting that the effect of a dominant negative transgene might not be SIK2 specific but could broadly affect functions of other SIK kinases or AMPK-related kinases.

In summary, we delineated for the first time the chronic effect of SIK2 depletion on glucose and lipid metabolism using a knockout mouse model. SIK2-null mice displayed higher plasma glucose and triglyceride levels that are associated with impaired glucose and insulin tolerance under normal diet as well as high-fat diet conditions. We found that the depletion of SIK2 leads to the accumulation of larger fat cells due to the hyperactivation of a CREB-CRTC2–dependent transcriptional program, leading to the increased infiltration of macrophage and decreased plasma high–molecular weight adiponectin levels that contribute to the reduced insulin signaling in white adipocytes, providing a potential mechanism for the observed metabolic phenotype of SIK2 KO mice. Though we observed a dramatic effect of SIK2 depletion on adipogenesis and cAMP-dependent transcriptional program in fat cells, we cannot completely rule out the contribution of other SIK isoforms on specific aspects of the CREB-CRTC2 pathway or lipid metabolism, since all three isoforms are highly expressed in adipocytes. Further study is necessary to delineate the relative contribution of other SIK isoforms in the regulation of signaling pathways in fat cells.

Acknowledgments. The authors thank Sun Myung Park (Korea University) for technical assistance.

Funding. This work was supported by the National Research Foundation of Korea (grant nos. NRF-2010-0015098, NRF-2010-0019513, and NRF-2012M3A9B6055345), funded by the Ministry of Science, ICT & Future Planning, Republic of Korea; a grant of the Korea Health Technology R&D Project (grant no. A111345), Ministry of Health and Welfare, Republic of Korea; and a Korea University Grant, Republic of Korea.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. J.P. and Y.-S.Y. researched data and reviewed and edited the manuscript. H.-S.H. and Y.-H.K. researched data. Y.O., K.-G.P., C.-H.L., and S.-T.K. reviewed and edited the manuscript. S.-H.K. conceived/designed the experiments and wrote the manuscript. S.-H.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.