Pivotal Role of O-GlcNAc Modification in Cold-Induced Thermogenesis by Brown Adipose Tissue Through Mitochondrial Biogenesis

Obesity develops when energy intake exceeds energy expenditure, leading to excess calorie storage in the adipose tissues. Obesity is highly correlated with the development of type 2 diabetes, the metabolic syndrome, and cardiovascular disease. Adipose tissues have considerable influence on metabolic homeostasis. Two functionally different types of adipose tissues are present in mammals: white adipose tissue (WAT), which is the primary site of energy storage, and brown adipose tissue (BAT), which is specific to thermogenic energy expenditure. BAT uniquely expresses uncoupling protein 1 (Ucp1) and is rich in mitochondria. Furthermore, it is responsible for active metabolism during cold- or diet-induced thermogenesis and uses glucose and fatty acids as fuel (1–4).

Cellular homeostasis between nutrient sensing and energy metabolism is coordinately regulated by complex molecular mechanisms (5). Growing evidence indicates that the hexosamine biosynthetic pathway and its end product, uridine diphosphate N- acetylglucosamine (UDP-GlcNAc), act as important nutrient sensors (6–8). In a branch of the glycolytic pathway, fructose-6-phosphate is converted into UDP-GlcNAc through multiple enzymes in the hexosamine biosynthetic pathway. UDP-GlcNAc also serves as the donor for O-linked N-acetylglucosamine (O-GlcNAc) modification, which is associated with glucose, amino acid, and fatty acid metabolism (9). O-GlcNAc modification of proteins on serine/threonine residues occurs in the nucleus, cytoplasm, and mitochondria, and its addition is catalyzed by O-GlcNAc transferase (Ogt) (10).

Cold-induced thermogenesis is a fundamental function required in mammals to survive in severe environmental changes. BAT is important for maintaining body temperature under cold conditions (11). Recent reports have revealed that, even in humans, significantly increased glucose and fatty acid flux are observed via positron emission tomography with ¹⁸F-fluorodeoxyglucose and ¹⁸F-fluoro-thiahepadecaoic acid (12). In addition, dramatic changes in both glucose and fatty acid flux occur in the fasting and postprandial states in WAT. Since O-GlcNAc modification is considered a “nutrient sensor” for glucose and fatty acids, O-GlcNAc modification may influence BAT and WAT function. However, little is known about the role of O-GlcNAc modification in adipose tissues.

Therefore, in this study, to reveal the physiological role of O-GlcNAc modification in adipose tissues, we analyzed the phenotypes of adipose tissue–specific Ogt knockout (Ogt-FKO) and BAT-specific Ogt knockout (Ogt-BKO) mice. Here, we demonstrated that O-GlcNAc modification in BAT is essential for thermogenesis and mitochondrial biogenesis.

Ogt-KO mice were generated using the Cre-LoxP system. We crossbred Ogt-flox (Ogt^f/f) female mice (The Jackson Laboratory, Bar Harbor, ME) with Adipoq-Cre (kindly provided by Evan Rosen [Beth Israel Deaconess Medical Center]) and Ucp1-Cre (The Jackson Laboratory) mice to generate Ogt-FKO and Ogt-BKO mice, respectively. All experiments and analyses were performed with male mice. Because Ogt is present on the X chromosome, the first generation of female mice is heterozygous with X-linked inheritance. Thus, we examined male mice as complete knockout mice (Figs. 1A and 4A). All animal handling and experimentation were conducted according to the guidelines of the Research Center for Animal Life Science at Shiga University of Medical Science or the Korea Mouse Metabolic Phenotyping Center at Gachon University. All experimental protocols were approved by the Gene Recombination Experiment Safety Committee and Research Center for Animal Life Science at Shiga University of Medical Science or the Gachon University Institutional Animal Care and Use Committee.

Mice were euthanized at various time points by intraperitoneal administration of 10% pentobarbital with sevoflurane inhalation before immediate tissue collection. Inguinal WAT (iWAT), epididymal WAT (eWAT), interscapular BAT, gastrocnemius muscle, liver, and pancreas were dissected immediately, snap frozen in liquid nitrogen, and stored at −80°C until analysis.

Blood glucose concentrations were measured with glucose dehydrogenase–pyrroloquinoline quinone glucose test strips (Glutest Sensor; Sanwa Kagaku Kenkyusho, Nagoya, Japan). Plasma insulin levels were measured by ELISA (Morinaga, Tokyo, Japan). Blood ketone levels were measured using a β-ketone monitoring system (Precision Xceed; Abbott Japan, Chiba, Japan).

Fixed specimens embedded in paraffin were sectioned (3-µm thicknesses). Antibodies against UCP1 (U6382; Sigma-Aldrich, St. Louis, MO), cytochrome oxidase subunit 4 (COX4) (Novus Biologicals, Cambridge, U.K.), and perilipin (category no. 9349; Cell Signaling Technology, Tokyo, Japan) were used. Transmission electron microscopic analysis was performed with a Hitachi H-7500 (Hitachi, Tokyo, Japan). The adipocyte number and area were calculated using a BZ-H3C (Keyence, Osaka, Japan).

Ogt-FKO, Ogt-BKO, and control mice were deprived of food for 12 h before experiments. The mice were placed into individual cages with bedding and water in a room maintained at 4°C for 3 h. Mouse rectal body temperature was measured using a type T thermocouple rectal probe (RET-3; Physitemp Instruments, Inc., Clifton, NJ). After cold exposure, mice were euthanized by sevoflurane before immediate interscapular BAT and iWAT collection as described above.

Ogt-FKO, Ogt-BKO, and control mice were deprived of food for 18 h before experiments. Mice were placed into individual cages with bedding, food, and water in a room maintained at 4°C for 3 h. The mice were fed a normal (protein:fat:carbohydrate = 27.5:12.5:60 kcal%, CE-2; CLEA Japan, Inc., Tokyo, Japan) or ketogenic (protein:fat:carbohydrate = 10.4:89.5:0.1 kcal%, Very Low Carbohydrate Ketogenic Rodent Diet; Research Diets, Inc., New Brunswick, NJ) diet during cold exposure. Before and at 1, 2, and 3 h after cold exposure, mouse rectal body temperatures were assessed. Before and at 3 h after cold exposure, blood glucose concentrations and blood ketone levels were measured.

An intraperitoneal glucose tolerance test (IPGTT) in Ogt-FKO, Ogt-BKO, and control mice was performed in overnight-fasted mice after an intraperitoneal injection of glucose (1 g/kg body wt) at room temperature or 4°C. Blood glucose levels were determined at 15, 30, 60, 90, and 120 min after injection.

An intraperitoneal insulin tolerance test (IPITT) in Ogt-FKO and control mice was performed after an intraperitoneal injection of insulin (0.5 units/kg body wt) at 6 h after fasting at room temperature. Blood glucose levels were determined at 15, 30, 60, 90, and 120 min after injection.

Ogt-FKO, Ogt-BKO, and control mice were deprived of food for 18 h before experiments. The mice were placed into individual cages with bedding and water in a room maintained at 4°C for 3 h. Glucose (20%) or olive oil was administered orally (0.01 mL/g body wt) before and at 1 and 2 h after cold exposure. Before and at 1, 2, and 3 h after cold exposure, mouse rectal body temperatures were measured. Before and at 3 h after cold exposure, blood glucose concentrations and blood ketone levels were measured.

Fat and lean body masses were measured in mice using the ¹H Minispec system (LF90II; Bruker Optics, Ettlingen, Germany). Basal energy balance, including oxygen consumption, carbon dioxide production rate, respiratory quotient, energy expenditure, and food intake, was measured (2 days of acclimation followed by 2 days of measurement) using Comprehensive Lab Animal Monitoring Systems (CLAMS; Columbus Instruments, Columbus, OH).

Total RNA was extracted from tissues using an RNeasy Kit (Qiagen, Valencia, CA). cDNA was synthesized using reverse transcription reagents (Takara Bio, Otsu, Japan). Transcript abundance was assessed by real-time PCR on an Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific K.K., Yokohama, Japan) with SYBR Green (Bio-Rad Laboratories, Hercules, CA). Analytical data were normalized to GAPDH mRNA expression as an internal control. Primer sequences can be found in Supplementary Data.

DNA primers were designed to detect cytochrome oxidase 2 (Cox2) and uncoupling protein 2 (Ucp2) for mitochondrial DNA (mtDNA) and nuclear DNA, respectively. The ratio of Cox2 to Ucp2 within the samples was used to calculate the mtDNA copy number. Primer sequences can be found in Supplementary Data.

For Western blot analysis, proteins were resolved by SDS-PAGE and transferred to nitrocellulose membranes. The membranes were incubated with O-GlcNAc antibody (RL2) (MA1-072; Thermo Scientific, Waltham, MA) as well as antibodies against the following molecules: OGT (O6264; Sigma-Aldrich); β-tubulin (H-235; Santa Cruz Biotechnology, Santa Cruz, CA); UCP1 (U6382; Sigma-Aldrich); peroxisome proliferator–activated receptor γ coactivator-1α (PGC-1α) (ab54481; Abcam plc, Cambridge, U.K.); CREBP (sc-186; Santa Cruz Biotechnology); phosphorylated CREBP (sc-7978; Santa Cruz Biotechnology); phosphorylated (Ser/Thr) protein kinase A substrate (PKA) (category no. 9621; Cell Signaling Technology, Tokyo, Japan); phosphorylated hormone-sensitive lipase (HSL) Ser⁵⁶³ (category no. 4139; Cell Signaling Technology); phosphorylated HSL Ser⁶⁶⁰ (category no. 4126; Cell Signaling Technology); HSL (category no. 4107; Cell Signaling Technology); cytochrome c oxidase subunit 1 (MTCO1) (ab14705; Abcam plc); COX4 (ab14744; Abcam plc); mitochondrial transcription factor A (TFAM) (LS-C30495; LifeSpan BioSciences, Seattle, WA); succinate dehydrogenase complex, subunit A (ab14715; Abcam plc); pyruvate dehydrogenase (Novus Biologicals, Cambridge, U.K.); ubiquitin (Cell Signaling Technology); autophagy-related gene (ATG) 5 (Cell Signaling Technology); ATG7 (Cell Signaling Technology); microtubule-associated protein 1 light chain 3 (LC3) (Novus Biologicals); β-actin (A5316; Sigma-Aldrich); pan-actin (sc-1616; Santa Cruz Biotechnology); long-chain acyl-CoA dehydrogenase (LCAD) (ACADL; Proteintech Group, Rosemont, IL); and medium-chain acyl-CoA dehydrogenase (MCAD) (ACADM; GeneTex, Inc., Irvine, CA). After additional washing, the membranes were incubated with horseradish peroxidase–linked secondary antibodies followed by chemiluminescence detection.

The interscapular brown fat pad (eight fat pads for each group) was dissected from 8-week-old Ogt-BKO and Ogt-flox mice as previously described (13). Briefly, tissues were minced and digested with 1.5 units/mL collagenase D (1108874103; Roche) in 10 mmol/L CaCl₂ and 2.4 units/mL dispase II (04942078001; Roche) for 40–50 min while shaking at 37°C. Digestion was stopped by adding complete DMEM/F12 containing 10% FBS and penicillin/streptomycin (stromal vascular [SV] culture medium). Cells were collected by centrifugation at 700g for 10 min, resuspended, and strained through a 70 μmol/L cell strainer (BD Biosciences). Cells were further filtered through a 40-mm cell strainer to remove clumps and large adipocytes. After further centrifugation as mentioned above, SV cells were resuspended in SV culture medium and plated onto sixwell collagen-coated dishes. At confluency (day 0), cells were exposed to a differentiation cocktail including 0.25 µmol/L dexamethasone, 10 µg/mL insulin, 1 nmol/L T3, 0.5 μmol/L rosiglitazone, 0.5 mmol/L isobutylmethylxanthine, and 125 μmol/L indomethacin in SV culture medium. At 48 h after induction, the cells were maintained in SV culture medium containing 5 mg/mL insulin and 1 µmol/L rosiglitazone for 6 days. Cells were fully differentiated on day 8. For examination of the role of the proteasomal pathway, proteasomal inhibitor MG132 (25 µmol/L) was applied at 3 h before sample collection. All chemicals for cell culture were obtained from Sigma-Aldrich unless otherwise indicated.

Results are expressed as means ± SEM. Student t tests were used to assess differences between two groups. A P value of <0.05 was considered statistically significant.

We first clarified the role of O-GlcNAc modification in WAT and BAT using Ogt-FKO mice (Fig. 1A). OGT protein expression was decreased in both WAT and BAT of Ogt-FKO mice compared with those of control mice (Fig. 1B). Similarly, O-GlcNAc modifications analyzed by the RL2 antibody were also decreased in these tissues (Fig. 1B). Ogt-FKO mice developed normally (Fig. 1C), and there were significant reductions in eWAT and iWAT masses (Fig. 1D and E). In contrast, BAT in Ogt-FKO mice displayed an increased fat mass. Upon dissection, it was visibly lipid laden and displayed a milky appearance compared with the dark brown color of BAT in control mice (Fig. 1F). Histologically, BAT cells in Ogt-FKO mice were substantially enlarged with a WAT-like appearance resulting from accumulation of lipid droplets (Fig. 1G and H; Supplementary Fig. 1A). Quantitative analysis of cell size revealed that Ogt-FKO mice had significantly smaller adipocytes in eWAT and iWAT but larger adipocytes in BAT (Fig. 1I–K).

We next examined the effects of adipose tissue–specific O-GlcNAc modification deficiency in glucose metabolism and thermogenesis. IPGTTs and IPITTs revealed no significant differences between the two genotypes (Fig. 2A and B). For assessment of thermogenesis, a cold exposure test was performed at 4°C for 3 h. Surprisingly, although all mice visibly shivered, cold intolerance was evident only in Ogt-FKO mice, with the mice reaching critical hypothermia within 3 h (Fig. 2C). During cold exposure, there were no differences in blood glucose or plasma insulin levels between the two genotypes (Fig. 2D and E). eWAT and iWAT masses in Ogt-FKO mice were reduced compared with those in control mice (Fig. 2F and G). In contrast, BAT in Ogt-FKO mice was heavier than that in control mice and displayed substantial lipid droplet accumulation even after cold exposure (Fig. 2F and G). Therefore, we focused on the detailed molecular mechanism underlying O-GlcNAc modification deficiency–mediated cold intolerance.

Interestingly, BAT Ucp1 gene expression was significantly low at room temperature in Ogt-FKO mice (Fig. 2H). Moreover, Ogt-FKO mice displayed impaired cold-stimulated Ucp1 mRNA expression compared with control mice (Fig. 2H). UCP1 protein expression was also significantly decreased in Ogt-FKO mice (Fig. 2I). Similarly, immunohistochemical analysis revealed decreased UCP1 expression both at room temperature and after cold exposure (Fig. 2J). To test the differentiation level of BAT, we analyzed mRNA expression of Cox7a1, Cidea, and Prdm16. There was no significant difference in the expression of these genes between the two genotypes (Fig. 2K). Taken together, Ogt-FKO mice displayed significant intolerance to cold exposure compared with control mice, partially because of the lack of Ucp1 expression.

Induction of the cold-induced thermogenesis program in BAT in vivo is strongly influenced by the sympathetic nervous system and the resulting action of the β-adrenergic receptor/PKA/CREB axis. Thus, we next evaluated changes in the expression levels of genes associated with the β-adrenergic receptor/PKA/CREB axis. First, we examined PKA, HSL, and CREB phosphorylations and found no significant differences between Ogt-FKO and control mice (Supplementary Fig. 2). However, we observed a significant reduction in protein level of PGC-1α, a key regulator of thermogenesis, although there were no significant changes in mRNA expression of the gene encoding PGC-1α, Ppargc1a (Fig. 3A and B). Because PGC-1α is also known as a master regulator of mitochondrial biogenesis, we next evaluated the expression of genes associated with mitochondria. Although the mtDNA copy number was similar in the two genotypes (Fig. 3C), mRNA expression of mtDNA-encoded genes was significantly decreased in Ogt-FKO mice, suggesting decreased transcription of mtDNA-encoded genes, probably due to the decreased TFAM protein levels (Fig. 3D). Moreover, immunohistochemical analysis of COX4 in BAT revealed decreased expression in Ogt-FKO mice (Fig. 3E). In addition, the protein levels of MTCO1, COX4, TFAM, and succinate dehydrogenase complex, subunit A, were significantly decreased in BAT (Fig. 3F and G). These results suggested that Ogt deletion in adipose tissues resulted in impaired cold-induced thermogenesis in BAT through dysregulation of PGC-1α.

For evaluation of whether BAT is the sole organ responsible for the dysregulation of cold-induced thermogenesis after Ogt deletion, Ogt-BKO mice were generated by crossbreeding Ogt-flox mice with Ucp1-Cre mice (Fig. 4A). In Ogt-BKO mice, OGT protein expression was decreased only in BAT, resulting in decreased O-GlcNAc modification as measured using the RL2 antibody (Fig. 4B). Similar to Ogt-FKO mice, Ogt-BKO mice developed normally (Fig. 4C). There were no differences in body composition, food intake, energy expenditure, resting energy expenditure, oxygen consumption, carbon dioxide production, or locomotor activity between Ogt-BKO and control mice (Fig. 4D–H; Supplementary Fig. 3). Although there were no differences in eWAT or iWAT masses, a difference in BAT mass was observed (Fig. 4I). Consistent with Ogt-FKO mice, Ogt-BKO mice displayed accumulation of enlarged lipid droplets in BAT cells (Fig. 4J; Supplementary Fig. 1B).

Similar to that in Ogt-FKO mice, cold intolerance was evident in Ogt-BKO mice, with mice reaching critical hypothermia within 3 h (Fig. 5A). Ogt-BKO mice showed no significant changes in expression levels of OGT or O-GlcNAc modification detected by the RL2 antibody in iWAT after 3 h of cold exposure (Supplementary Fig. 4A–C). The BAT UCP1 protein level was severely diminished in Ogt-BKO mice compared with control mice at room temperature (Fig. 5B). Moreover, PGC-1α protein level was significantly lower, although mRNA expression was unchanged compared with control mice, suggesting posttranslational modification (Fig. 5C and D). To further explorer this mechanism, we evaluated primary brown adipocytes from Ogt-BKO mice. PGC-1α expression was significantly reduced in primary brown adipocytes from Ogt-BKO mice compared with those from control mice. Treatment with proteasomal inhibitor MG132 partially but significantly increased PGC-1α protein levels in Ogt-BKO mice, suggesting that O-GlcNAc modification protects PGC-1α from proteasomal degradation (Fig. 5E–G). Similar to findings in Ogt-FKO mice, immunohistochemical analysis showed that COX4 was decreased in BAT (Fig. 5H), and other mitochondrial protein levels were significantly decreased in Ogt-BKO mice (Fig. 5I and J). There was no difference in blood glucose levels, but there was a significant decrease in plasma insulin levels in Ogt-BKO mice (Fig. 5K and L). In other glucose metabolism–related tissues such as liver, muscle, and pancreas, no apparent difference was found in OGT or mitochondrial-related protein expression, and O-GlcNAc modification was observed in Ogt-BKO mice compared with control mice (Supplementary Fig. 5B). This phenomenon was consistent in Ogt-FKO mice (Supplementary Fig. 5A).

During cold exposure, increased glucose and fatty acid oxidation appears to maintain core temperature by stimulating heat production in BAT. To evaluate glucose metabolism during cold exposure, we performed IPGTTs in Ogt-BKO and control mice. Surprisingly, body temperatures during an IPGTT at 4°C were sustained even in Ogt-BKO mice (Fig. 6A). However, Ogt-BKO mice displayed slightly but significantly lower glucose levels compared with control mice at 30 and 60 min of cold exposure (Fig. 6B). Compared with glucose levels at room temperature, substantially lowered glucose levels were observed during cold exposure in both genotypes, suggesting increased glucose utilization (Fig. 6B). These data suggest that diet-induced thermogenesis by glucose rescued hypothermia in Ogt-BKO mice. To test this hypothesis, we performed glucose- and oil-loading tests during cold exposure. There was no apparent difference in shivering between glucose- and oil-loaded groups (data not shown). Oral administration of oil failed to maintain body temperature in Ogt-BKO mice during cold exposure (Fig. 6C). After cold exposure, oil administration in Ogt-BKO mice resulted in significantly lower glucose levels compared with control mice (Fig. 6D). In addition, Ogt-BKO mice fed a normal diet could maintain body tempeature, but Ogt-BKO mice fed a low-carbohydrate ketogenic diet could not, although there were no changes in blood glucose or ketone body levels between Ogt-BKO and control mice (Fig. 6E–G). Because a ketogenic diet is deficient in carbohydrates but high in fat, these data suggest that Ogt-BKO mice preferentially use glucose, and fat utilization is defective during cold exposure. To further explorer the mechanism, we next investigated whether Ogt deletion affects lipolysis and fatty acid oxidation in BAT. The protein expression of ATGL and HSL, key enzymes of lipolysis, were similar between Ogt-BKO and control mice (Supplementary Fig. 6A and B). Protein expression of LCAD and MCAD, fatty acid oxidation gene known PGC-1α targets, were significantly lower in Ogt-BKO mice (Fig. 6H–J). These data strongly support the hypothesis that O-GlcNAc modification deficiency impairs fatty acid oxidation in BAT.

The current study was designed to clarify the role of protein O-GlcNAc modification in adipose tissues. We demonstrated that this modification was essential for cold-induced thermogenesis in BAT. Advances in brown adipose cell biology during the last decade have increased our understanding of the cellular origin (14–16), function (4,16,17), and adult human tissue distribution (3,18–20) of BAT. However, a comprehensive overview of BAT biology has not been fully clarified. One of the most prominent findings from this study was that Ogt-KO mice displayed impaired thermogenesis during cold exposure. Compared with that in control mice, the BAT of both Ogt-FKO and Ogt-BKO mice showed an increased number of large lipid droplets. To reveal the detailed mechanism underlying impaired thermogenesis by O-GlcNAc modification deficiency, we compared the expression levels of mitochondrial proteins in the BAT of Ogt-KO and control mice. In the BAT of Ogt-FKO mice, protein and mRNA expression levels of Ucp1 and mtDNA-encoded proteins were decreased significantly and accompanied by decreased PGC-1α expression. PGC-1α–dependent Ucp1 expression and mitochondrial biogenesis are essential for cold exposure–induced thermogenesis in BAT (21). Thus, O-GlcNAc modification deficiency in BAT may impair cold-induced thermogenesis by modulating PGC-1α.

PGC-1α is a key transcriptional coactivator that regulates mitochondrial biogenesis (21). PGC-1α stability and activity are regulated by posttranslational mechanisms such as acetylation and phosphorylation (22–24). PGC-1α degradation can also be regulated by ubiquitin-dependent proteasomal degradation (25,26). We found no difference in PGC-1α mRNA expression between Ogt-KO and control mice, whereas PGC-1α protein levels were decreased in Ogt-KO mice (Figs. 3B and 5D). In addition, primary adipocytes from Ogt-BKO mice showed lower PGC-1α protein levels compared with control mice (Fig. 5E and G). Treatment with the proteasomal inhibitor resulted in a partial but significant increase in PGC-1α protein levels, indicating the ubiquitin-proteasomal pathway in the stability of PGC-1α (Fig. 5E and G). These results suggest that O-GlcNAc modification may regulate the PGC-1α degradation process in BAT. This hypothesis is in agreement with a previous study that demonstrated that O-GlcNAc modification of PGC-1α protein in the liver antagonizes ubiquitination and subsequent degradation (27). Taken together, O-GlcNAc modification of PGC-1α protein may stabilize PGC-1α by inhibiting ubiquitination in BAT.

An alternative explanation for the decreased mitochondrial proteins is activation of autophagy. It is known that mitochondrial homeostasis is regulated by the balance between mitochondrial biogenesis and degradation (28). LC3, a regulatory protein essential for induction of autophagy, localizes to autophagosome membranes during autophagy activation. We found that LC3-I protein expression was decreased and LC3-II protein expression was increased in Ogt-BKO mice, suggesting acceleration of autophagy (Supplementary Fig. 7A and B). Moreover, protein expressions of ATG5 and ATG7, essential for autophagy induction, were increased in Ogt-BKO mice compared with control mice (Supplementary Fig. 7A, C, and D). In transmission electron microscopy, autophagosomes were found in Ogt-BKO mice but not in control mice (Supplementary Fig. 7E). These data suggest that autophagy is enhanced in Ogt-BKO mice. It has been supported that autophagy has a role in remodeling mitochondrial contents and regulating adipose mass and differentiation (28,29). It is also possible that the enhancement of autophagy is a consequence of damaged mitochondria due to Ogt depletion. Further examinations are necessary to reveal the relationship of dysregulated mitochondrial proteins and autophagy in the context of Ogt deletion.

Another interesting finding of this study was that impaired thermogenesis in Ogt-KO mice was observed under fasting conditions. This effect was recovered by oral intake of a normal diet or after glucose injection but not by consumption of a fatty acid–rich ketogenic diet deficient in glucose (Fig. 6C and E; Supplementary Fig. 8). Moreover, we found a marked decrease in the protein levels of acyl-CoA dehydrogenases, such as MCAD and LCAD, in Ogt-BKO mice (Fig. 6H–J). Therefore, the metabolic rate of fatty acid was probably decreased. A previous study in mice with adipose-specific deletion of carnitine palmitoyltransferase 2 demonstrated similar changes because of the lack of fatty acid oxidation (30). In addition, it has been reported that PGC-1α is related to regulation of gene expression involved in fatty acid oxidation (31,32), and acute RNA interference–mediated PGC-1α knockdown leads to profound downregulation of fatty acid gene expression (33). Hence, we speculate that fatty acid oxidation was impaired because of the decrease of PGC-1α protein in Ogt-KO mice. Collectively, O-GlcNAc modification is essential for fat oxidation in BAT. Because glucose does not need a β-oxidation process to form acetyl-CoA, glucose may be preferentially used in Ogt-KO mice to maintain body temperature and generate heat even under the reduced UCP1 condition. This hypothesis is partially supported by our results showing which enzymes were intact in the glycolysis pathway (Supplementary Fig. 6C). In addition, blood glucose levels in Ogt-BKO mice during cold exposure were lower than those in control mice (Fig. 6B and D). Thus, enhanced glucose uptake and use in BAT of Ogt-KO mice were likely to be associated with heat production during cold exposure. Thus, O-GlcNAc modification may act as a “metabolic switch” in BAT and plays a crucial role in whole-body glucose and lipid homeostasis in cold environments.

Ogt deletion from both WAT and BAT resulted in reduced WAT and increased BAT masses (Fig. 1D–F). In contrast, Ogt-BKO mice displayed similar WAT and increased BAT masses compared with control mice (Fig. 4D). Because both Ogt-FKO and Ogt-BKO mice displayed intolerance to cold exposure, the differences in WAT have little effect on this phenotype.

There are two limitations to this study. First, we did not identify the direct target of Ogt in BAT. Although we speculate that PGC-1α may be a potential direct target of Ogt, further experiments are necessary to test this hypothesis. Second, energy expenditure during hypothermia was not measured as a result of a technical issue. Because the sympathetic nervous system plays a significant role in cold-induced thermogenesis, further experiments such as β3 adrenergic stimulation may predict energy expenditure during hypothermia.

In conclusion, the posttranslational O-GlcNAc modification plays a pivotal role in cold-induced thermogenesis in BAT. The current findings provide novel insights into BAT biology.

Acknowledgments. The authors are indebted to Keiko Kosaka, Yoshiko Asano, Takefumi Yamamoto (Shiga University of Medical Science), and the Central Research Laboratory of Shiga University of Medical Science for expert technical assistance in this study. The authors thank Evan Rosen (Harvard University) and Wataru Ogawa and Tetsuya Hosooka (Kobe University) for providing the Adipoq-Cre mice. The authors appreciate constructive discussion from Hiroshi Sakaue (Tokushima University) and Shingo Kajimura (University of California, San Francisco).

Funding. This study was supported by Grants-in-Aid for Scientific Research (KAKENHI) from the Japan Society for the Promotion of Science (15K09383 to O.S. and 16K09743 to S.U.). This work was also supported by a grant from the Bio & Medical Technology Development Program of the National Research Foundation, which is funded by the Ministry of Science, ICT and Future Planning (NRF-2014M3A9D5A01073886). This study was also funded by the Shiga University of Medical Science.

Duality of Interest. The Department of Medicine, Shiga University of Medical Science, has received research promotion grants (shogaku kifukin) from Astellas; Boehringer Ingelheim; Daiichi Sankyo; Kowa Pharmaceuticals; Kyowa Hakko Kirin; Mitsubishi Tanabe Pharma; Merck Sharp & Dohme; Ono Pharmaceutical Co., Ltd.; Sanofi; Sanwa Kagaku Kenkyusho; Shionogi; Taisho Toyama Pharmaceutical Co., Ltd.; Takeda; and Teijin Pharma. However, the research topics of these donation grants are not restricted. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. N.O., K.M., S.I., S.U., and H.M. designed the study. N.O., S.I., S.-Y.P., C.S.C., and M.L. conducted the research. N.O. analyzed data. N.O., K.M., and H.M. wrote the manuscript. S.K. gave constructive comments regarding the study concept. K.M., O.S., S.K., S.U., and H.M. reviewed and edited the manuscript. K.M. 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. All authors read and approved the final manuscript.

Prior Presentation. Parts of this study were presented in abstract form at the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017; at the 2015 Keystone Symposia on Molecular and Cellular Biology, Kyoto, Japan, 25–29 October 2015; at the 35th Congress of Japan Society for the Study of Obesity, Nagoya, Japan, 2–3 October 2015; and at the 60th Annual Meeting of the Japan Diabetes Society, Nagoya, Japan, 18–20 May 2017.