Transketolase Deficiency in Adipose Tissues Protects Mice From Diet-Induced Obesity by Promoting Lipolysis

The adipose tissue is critical for energy balance and nutritional homeostasis. In mammals, the adipose tissue is classified into the brown adipose tissue (BAT) and white adipose tissue (WAT) (1–3). BAT mainly participates in thermogenesis (1–3). The major function of WAT is to store and release fat in response to energy needs, dependent on lipogenesis and lipolysis, respectively (4,5). Although de novo lipogenesis takes place in adipocytes and other cell types, lipolysis, the hydrolysis of triacylglycerol (TAG) to liberate fatty acids and glycerol, predominantly occurs in WAT (5–8). Usually, lipolysis is catalyzed by three enzymes in three consecutive steps. First, adipose triglyceride lipase (ATGL) converts TAG to diacylglycerol and fatty acid. Next, hormone-sensitive lipase (HSL) generates monoacylglycerol and fatty acid from diacylglycerol. Finally, monoglyceride lipase (MGL) cleaves monoacylglycerol into glycerol and fatty acid (7–10).

Obesity occurs when energy generation exceeds energy expenditure, resulting in excess TAG accumulation in adipocytes (11–13). Recently, obesity has been rising dramatically and become a major health problem due to its close association with type 2 diabetes, cardiovascular disease, cancer cachexia, and fatty liver disease (5,11,14).

Being an active metabolic pathway in adipose tissues, the pentose phosphate pathway (PPP) has been intensively studied in adipocytes (15–17). The PPP, which originates from glycolytic intermediate glucose-6-phosphate (G6P), consists of oxidative and nonoxidative arms. G6P dehydrogenase (G6PD) is the rate-limiting enzyme in the oxidative arm that generates NADPH to maintain redox homeostasis and support biosynthetic pathways in cells. Increasing or decreasing the expression or activity of G6PD has been shown to promote or inhibit adipogenesis and adipocyte differentiation, respectively (18,19). It has also been reported that G6PD deficiency in mice alleviates high-fat diet (HFD)–induced insulin resistance and adipose tissue inflammation (20).

Transketolase (TKT), a thiamine diphosphate–dependent enzyme, catalyzes two reversible reactions in the nonoxidative branch of the PPP (Supplementary Fig. 5A). One is the conversion from xylulose-5-phosphate (Xu5P) and ribose-5-phosphate (R5P) to glyceraldehyde-3-phosphate (G3P) and sedoheptulose-7-phosphate. The other is to produce G3P and fructose-6-phosphate (F6P) from Xu5P and erythrose-4-phosphate. Since G3P and F6P are intermediate metabolites of the glycolytic pathway, TKT serves as a bridge linking the PPP and glycolysis (21,22). TKT has been reported to play critical roles in promoting deregulated metabolism and genome instability in tumors (23–26). Moreover, TKT is indispensable for mammalian development (27,28). In mice, TKT-null embryos are not viable (28). Furthermore, disruption of one TKT allele results in ∼77% weight reduction of adipose tissues (28). These findings suggest that TKT may play an important role in adipose tissues.

Although the function of G6PD in obesity and obesity-associated metabolic disorders has been intensively studied (18–20), the role of TKT in obesity is unknown in spite of its abundant expression in adipocytes (22). In order to explore the role of TKT in adipose tissues, we generated mature adipocyte-specific TKT knockout mice by using the Cre-loxP strategy. We found that adipocyte-specific TKT ablation protected mice from HFD-induced obesity as well as obesity-related metabolic disorders. TKT deficiency in adipocytes led to accumulation of PPP metabolites and reduction of glycolytic metabolites. Loss of TKT reduced pyruvate levels in adipocytes; caused compensatory elevation of ATGL, HSL and MGL expression, and lipolysis; and subsequently enhanced fatty acid oxidation (FAO) and mitochondrial function. Supplementation of pyruvate abolished lipolysis induced by TKT deficiency. Collectively, our results demonstrate that TKT is a critical regulator of adipocyte metabolism and therefore could develop as a potential therapeutic target for treating obesity, diabetes, and other metabolic diseases.

Adipocyte-specific TKT knockout mice were generated by intercrossing Tkt^fl/fl mice (23) and Adiponectin-Cre mice (stock no. 028020; The Jackson Laboratory [29]) or Fabp4-Cre mice (Shanghai Biomodel Organism Science & Technology Development Co., Ltd. [30]). For diet-induced obesity, male mice were fed with HFD (60% kcal fat; Research Diets). All animal experiments were approved by the Shanghai Jiao Tong University School of Medicine Animal Care and Use Committee.

Whole-body fat and lean mass were measured by ¹H-nuclear magnetic resonance spectroscopy. O₂ consumption, CO₂ release, and physical activities were monitored with an OxyMax Comprehensive Laboratory Animal Monitoring System (CLAMS) (Columbus Instruments).

For glucose tolerance tests (GTT), mice were fasted overnight before intraperitoneal administration of glucose at 2 g/kg body wt. For insulin tolerance tests (ITT), normal chow diet (NCD)-fed mice or HFD-fed mice fasted for 4 h were intraperitoneally injected with insulin at 0.5 units/kg body wt or 0.8 units/kg body wt, respectively. Tail blood glucose concentrations were monitored with a glucometer.

Fresh epididymal WAT (eWAT) (100 mg) was isolated from mice and cut into pieces, followed by incubation in 1 mL serum-free medium containing 1% fatty acid–free BSA for 4 h in the absence or presence of 10 μmol/L isoproterenol (ISO) (I5627; Sigma-Aldrich) and/or 40 mmol/L sodium pyruvate. Glycerol or free fatty acid (FFA) release was measured using the kits shown in Supplementary Table 1.

Levels of hepatic TAG, serum TAG, insulin, adiponectin, leptin, glycerol, and FFA were determined using the kits shown in Supplementary Table 1.

Fresh tissues were fixed in 4% paraformaldehyde. For hematoxylin-eosin (H-E) staining, tissues were embedded in paraffin, cut into 4-μm sections, and stained with H-E. For oil red O staining, samples were embedded in the optimal cutting temperature (OCT) compound, cut into 10-μm sections, and stained with 0.5% oil red O. For dihydroethidium (DHE) staining, fresh tissues were immediately embedded in the OCT compound, cut into 10-μm sections, and stained with 10 μmol/L DHE.

Total protein was extracted from tissues, separated by SDS-PAGE, and probed with primary antibodies listed in Supplementary Table 1.

RNA extracted with TRIzol was reverse transcribed using the PrimeScript RT reagent kit (Takara). Quantitative PCR (qPCR) was performed using SYBR Green PCR reagents (Takara). Sequences of primers are listed in Supplementary Table 2.

mtDNA content was measured as previously described (31). Sequences of the primers used were listed in Supplementary Table 2.

Total RNA was extracted from eWATs. The mRNA was isolated with Oligo Magnetic Beads and randomly interrupted using divalent cations in NEB Fragmentation Buffer for cDNA synthesis. Libraries were generated using the NEBNext Ultra RNA Library Prep Kit (New England Biolabs, Ipswich, MA) for the Illumina system following the manufacturer’s instructions. Sequencing was conducted using the Illumina HiSeq X Ten platform.

eWAT (50 mg) was homogenized in precold extractant of 80% methanol with 1 μg/mL 2-chloro-d-phenylalanine, followed by centrifugation at 12,000 rpm for 10 min at 4°C. The supernatant was subjected to liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis. LC-MS/MS was performed using ultra-performance liquid chromatography coupled with tandem mass spectrometry. For each analysis, 2 μL sample was injected. A Waters ACQUITY UPLC HSS T3 C18 column (1.8 μm, 2.1 × 100 mm internal dimensions) was used for separation at 40°C. Mobile phase consisted of A (water with 0.04% ammonium acetate) and B (acetonitrile with 0.04% ammonium acetate).

Tissues were homogenized in precold extractant of 80% acetonitrile. After centrifugation, the supernatant was subjected to liquid chromatography–mass spectrometry analysis. Acylcarnitine analysis was performed as previously described (32).

All the replicate experiments are biological replicates, which were repeated at least three times. Data are represented as mean ± SEM or mean ± SD. Statistical significance was assessed by two-tailed Student t test or one-way ANOVA. P < 0.05 was considered statistically significant.

The data sets generated or analyzed during the current study are available from the corresponding author on reasonable request. The resource generated during or analyzed during the current study is available from the corresponding author on reasonable request.

To directly investigate the physiological functions of TKT in adipose tissues, we crossed Tkt^fl/fl mice with Adiponectin-Cre or Fabp4-Cre mice to generate adipocyte-specific TKT knockout mice (TKT^AKO and TKT^FKO, respectively) (Supplementary Fig. 1A). Different genotypes were identified by PCR (Supplementary Fig. 1B). Expression of TKT was notably decreased in eWAT, inguinal WAT (iWAT), and BAT in TKT^AKO mice compared with wild-type (WT) mice (Fig. 1A and B). Moreover, TKT expression was not affected in nonadipose tissues, including liver, kidney, lung, heart, and muscle, of TKT^AKO mice (Fig. 1B). Consistently, TKT was efficiently deleted in the eWAT, iWAT, and BAT of TKT^FKO mice (Supplementary Fig. 1C and D).

Compared with WT littermates, 5-month-old TKT^AKO mice showed no significant difference in body weight or body composition under NCD (Fig. 1C–E). However, when fed with HFD, TKT^AKO mice exhibited less body weight gain compared with WT littermates (Fig. 1F–H). Notably, the difference in body weight was obvious at as early as 5 weeks on HFD and became more significant afterward (Fig. 1F). ¹H-nuclear magnetic resonance analysis indicated that the difference in body weight was mainly due to reduced fat mass, although there was a mild reduction in the lean mass of TKT^AKO mice (Fig. 1H). In addition, TKT^AKO mice had smaller iWAT and eWAT depots and reduced adipocyte size (Fig. 1I–K). Likewise, TKT^FKO mice also gained less weight and had lower fat mass as well as smaller adipocytes than WT controls on HFD (Supplementary Fig. 1E–I). Taken together, these data indicate that sufficient expression of TKT in adipocytes is critical for HFD-induced obesity.

We next assessed the impact of adipocyte TKT deletion on hepatic lipid deposition. On NCD, there was no difference in liver morphology and histology, liver weight, liver TAG level, or levels of serum TAG, FFA, and glycerol between TKT^AKO and WT mice at the age of 5 months (Fig. 2A, B, and D–H). However, after 10-week HFD feeding, although WT littermates developed severe hepatic steatosis, TKT^AKO mice displayed much lower liver mass with less lipid accumulation (Fig. 2A–E). The serum TAG level was decreased, while levels of FFA and glycerol showed little change in TKT^AKO mice (Fig. 2F–H).

We found that loss of TKT in adipocytes decreased HFD-induced hyperleptinemia, although the mRNA level of leptin was not altered (Fig. 2I and Supplementary Fig. 2K). Additionally, TKT^AKO mice on HFD showed higher levels of serum adiponectin and elevated adiponectin mRNA expression in the eWAT (Fig. 2J and Supplementary Fig. 2K). Furthermore, other adipokines including apelin, resistin, and RBP4 were transcriptionally upregulated, but the mRNA level of visfatin was not altered in the eWAT of TKT^AKO mice on HFD (Supplementary Fig. 2K).

Similar phenotypes were observed in HFD-fed TKT^FKO mice compared with WT controls except that levels of serum FFA and glycerol were increased in TKT^FKO mice (Supplementary Fig. 2A–J and L).

We next investigated the effects of TKT deficiency in adipocytes on glucose tolerance and insulin sensitivity. On NCD, TKT^AKO mice showed lower blood glucose levels during the GTT but a similar response to insulin during the ITT compared with WT mice at 5 months of age (Fig. 2K and M). When challenged with 5-week and 10-week HFD, TKT^AKO mice showed better glucose tolerance and higher insulin sensitivity compared with WT littermates (Fig. 2L and N). Furthermore, loss of TKT in adipocytes decreased hyperinsulinemia after 10-week HFD, although there was no difference on NCD and 5-week HFD (Fig. 2O), which also indicated improved insulin sensitivity.

To explore the mechanism by which TKT ablation improved glucose and insulin homeostasis, we compared the level of Akt phosphorylated at Ser⁴⁷³, an indicator of the activity of the insulin signaling pathway in HFD-fed mice. Phosphorylation of Akt at Ser⁴⁷³ was elevated in eWATs both under the basal condition and after insulin injection in TKT^AKO mice (Fig. 2P), suggesting improved insulin signaling activity.

Consistently, compared with WT controls, TKT^FKO mice exhibited alleviated glucose intolerance and insulin resistance induced by HFD (Supplementary Fig. 2M and N). The level of serum insulin was reduced in TKT^FKO mice (Supplementary Fig. 2O). Western blotting analysis revealed the insulin signaling pathway was more activated in eWATs of the TKT^FKO mice with or without insulin stimulation (Supplementary Fig. 2P). Collectively, these results suggest that TKT deletion in adipocytes ameliorates HFD-induced metabolic disorders and leads to systemically metabolic benefits.

We next explored whether adipocyte TKT abrogation affects energy balance. Indirect calorimetry analysis showed that VO₂ and VCO₂ were similar between TKT^AKO and WT mice on NCD (Fig. 3A–D). Nevertheless, after HFD, both TKT^AKO mice and TKT^FKO mice exhibited significantly increased O₂ consumption and CO₂ production, compared with controls (Fig. 3E–H and Supplementary Fig. 3A–D). However, physical activities, indicated by x-axis total (XTOT) counts and y-axis total (YTOT) counts, and daily food intake were not changed in TKT^AKO mice (Fig. 3I–K). In addition, TKT^FKO mice showed greater physical activities than controls, although they had similar food consumption (Supplementary Fig. 3E and F). Furthermore, TKT deficiency did not affect respiratory exchange ratio (RER) (Supplementary Fig. 3G and H). In conclusion, loss of TKT in adipocytes increases systemic energy expenditure.

To obtain a molecular view of the effect of TKT deficiency on adipose tissues, we performed RNA sequencing (RNA-seq) with eWATs from TKT^AKO and WT mice on HFD for 10 weeks as well as TKT^FKO mice and their WT littermates on HFD for 14 weeks, respectively. Ablation of TKT resulted in significantly differential expression of 2,386 genes with P_adjusted <0.05 and |log2 fold change| >1 (Fig. 4A). The Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis of RNA-seq data revealed that the genes associated with lipolysis regulation were most significantly changed among the upregulated genes (Fig. 4B). Specifically, lipolytic enzyme genes including ATGL, HSL, and MGL were transcriptionally upregulated in TKT-deficient adipocytes (Fig. 4C). To confirm that, we performed qPCR and Western blotting analysis and found mRNA and protein levels of ATGL, HSL, and MGL were increased in eWATs and iWATs of TKT^AKO mice on HFD (Fig. 4D and E and Supplementary Fig. 4E and F). Especially, at 3 weeks of HFD, when there was no difference in body weight between TKT^AKO and WT mice, the lipolytic enzyme genes were all upregulated (Supplementary Fig. 4B–D). Similarly, mRNA and protein levels of ATGL, HSL, and MGL were augmented in TKT^FKO mice after 14-week HFD (Supplementary Fig. 4G and H). However, the expression of ATGL, HSL, and MGL was unchanged in TKT^AKO mice at the age of 5 months on NCD (Fig. 4F and G), which is consistent with similar body weight and fat mass between TKT^AKO and WT mice on NCD (Fig. 1C–E). Furthermore, expression of lipolytic enzyme genes of WT mice was downregulated upon HFD feeding (Supplementary Fig. 4A), as previously reported (33). However, the downregulation was minimized by TKT deletion (Supplementary Fig. 4I).

We further compared lipolytic activity by analyzing glycerol and fatty acid release from TKT-deficient and control eWAT explants. As expected, treating isolated eWAT explants with β-adrenergic receptor agonist ISO, which stimulates lipolysis by activating the protein kinase A (PKA) pathway, increased the rate of glycerol and fatty acid release (Fig. 4H and Supplementary Fig. 4J). Under the basal condition and after ISO treatment, the rate of glycerol and fatty acid release was higher in TKT-deficient eWAT explants compared with WT controls (Fig. 4H and Supplementary Fig. 4J). Therefore, these findings demonstrate that loss of TKT in adipose tissues promotes lipolysis.

TKT is an important enzyme linking the nonoxidative PPP and glycolysis (Supplementary Fig. 5A). In order to comprehensively elucidate the impact of TKT deficiency on metabolic pathways in adipocytes, we performed metabolomics with eWATs of 5-week HFD–fed TKT^AKO mice and WT controls. TKT deficiency led to accumulation of ribulose-5-phosphate, R5P, and Xu5P in the nonoxidative PPP, whereas levels of glycolytic metabolites including G6P, F6P, G3P, dihydroxyacetone phosphate, phosphoenolpyruvate, and pyruvate were decreased (Fig. 5A and B and Supplementary Fig. 5A), which was consistent with our previous findings in TKT-deficient liver tissues (23). Of note, expression of many glycolytic genes, especially Hk3, was decreased in TKT-deficient eWATs (Fig. 5C and D and Supplementary Fig. 5F). Hk3, more abundant than Hk1 and Hk2 in adipocytes (Supplementary Fig. 5G), catalyzes the reaction converting glucose to G6P. Therefore, Hk3 downregulation may explain the reduction of G6P in TKT-deficient adipocytes. In addition, we found that R5P-driven purine synthesis and uric acid production were enhanced in TKT-deficient adipocytes (Supplementary Fig. 5B). Furthermore, TKT deficiency led to reduction in glutathione (GSH) and increase in reactive oxygen species (Supplementary Fig. 5C and D), indicating that TKT ablation disturbed redox homeostasis.

Although the level of pyruvate decreased in TKT^AKO adipocytes, metabolites of TCA cycle remained stable (Supplementary Fig. 5E), suggesting an alternative carbon source feeding into the TCA cycle. Studies have shown that glucose utilization and fat catabolism have an inverse relationship (34,35). Therefore, we hypothesized that decreased glycolysis due to TKT deficiency would result in compensatory increase of lipolysis in TKT^AKO adipocytes. In order to test the hypothesis, we performed ex vivo lipolysis assay in the absence or presence of pyruvate. Without pyruvate supplementation, with or without ISO treatment, the rate of glycerol and fatty acid release was higher in the TKT-deficient eWAT explants compared with WT controls (Fig. 5E). Interestingly, pyruvate not only effectively inhibited lipolysis but also minimized the difference in lipolysis due to TKT deficiency (Fig. 5E). Collectively, these data suggest that TKT ablation causes metabolic switch by decreasing glycolysis and increasing lipolysis in adipocytes.

Since another alternative source of carbon for TCA cycle is branched-chain amino acids (BCAAs) in adipose tissues (36), we assessed the effects of TKT deficiency on BCAA catabolism. Levels of leucine and isoleucine were significant decreased in eWATs of TKT^AKO mice on HFD, whereas the level of valine remained unchanged (Supplementary Fig. 5H). Moreover, expression of several enzyme genes for BCAA catabolism was elevated in eWATs of TKT^AKO mice on HFD (Supplementary Fig. 5I and J). These findings suggest that leucine and isoleucine are probably more catabolized in TKT-deficient adipocytes.

Previous studies have shown that ATGL-mediated lipolysis promotes mitochondrial function (7). Therefore, we assessed the effect of TKT deletion on mitochondrial mass and function in eWATs after HFD feeding. We found that mtDNA content was increased in both TKT^AKO and TKT^FKO mice compared with their WT controls (Fig. 6A). The gene ontology analysis of RNA-seq data revealed that many genes encoding mitochondrial components including the electron transport chain (ETC) were upregulated in TKT-deficient eWATs (Fig. 6B and C). Moreover, qPCR analysis confirmed that genes of mitochondrial respiratory chain complexes (CI, CIII, CIV, and CV) were transcriptionally elevated in eWATs of TKT^AKO mice fed with HFD for 5 weeks and 10 weeks (Fig. 6D). Consistently, protein levels of ETC components including mt-Co1, mt-Co2, mt-Co3, and mt-Atp6 were increased (Fig. 6E). In addition, mRNA and protein levels of ETC genes were increased in iWAT of 10-week HFD–fed TKT^AKO mice (Supplementary Fig. 6A and B). Similarly, in HFD-fed TKT^FKO mice, these genes were elevated (Supplementary Fig. 6C).

For provision of sufficient energy for cells, fatty acids liberated by lipolysis are oxidized in the mitochondria. RNA-seq data suggested that genes of FAO were broadly increased in TKT-deficient adipocytes (Supplementary Fig. 6D). To confirm this, we conducted qPCR and found that enzyme genes of FAO were upregulated in both eWATs and iWATs of HFD-fed TKT^AKO mice (Supplementary Fig. 6E–G). Consistently, TKT^FKO mice also exhibited higher mRNA levels of FAO enzyme genes compared with their WT controls (Supplementary Fig. 6H). As previously reported (37), low levels of long-chain acylcarnitine species can indicate high activity of FAO. We found that levels of C14-C18 long-chain acylcarnitine species in eWATs of 10-week HFD–fed TKT^AKO mice were decreased when compared with WT controls, suggesting increased FAO due to TKT deletion (Supplementary Fig. 6I and J). Taken together, our data suggest that TKT is an important regulator for lipolysis, FAO, and mitochondrial function.

Here, we uncovered an unexpected role of TKT in balancing the carbon supply from glucose and TAG in adipose tissues. Deletion of TKT in adipocytes prevented HFD-induced obesity in mice by triggering lipolysis and FAO. TKT-deficient adipose tissues accumulated nonoxidative PPP metabolites and decreased glycolytic products, which caused insufficient energy supply from carbohydrate. As a consequence, lipolysis and FAO were forced to be elevated to meet energy demands (Fig. 7).

TKT is highly expressed in adipose tissues, and its haploinsufficiency causes a significant reduction of fat mass (28). Given the critical role of TKT in development, it is hard to determine whether reduced fat mass of the TKT heterozygous mice on NCD is due to growth retardation or deregulated metabolism or both. By using the Adiponectin-Cre– and Fabp4-Cre–mediated conditional knockout mouse models, we are able to analyze the function of TKT in adipose tissues. Both Adiponectin-Cre and Fabp4-Cre demonstrate efficient gene deletion in adipocytes (38). Although Fabp4 was originally identified as an adipocyte-specific protein, more and more evidence has shown that Fabp4 is also expressed in other cell types or organs, including macrophages, heart, and intermyofibrillar cells in the skeletal muscle (38–40). Adiponectin-Cre is more adipocyte specific, which shows little expression in other nonadipose tissues (38,39). In our study, the phenotypes of both TKT^AKO and TKT^FKO mice were largely similar, except that TKT^FKO mice showed higher serum FFA and glycerol levels (Supplementary Fig. 2G and H) and more physical activities (Supplementary Fig. 3E) compared with WT mice. Expression of Fabp4 in other cell types could probably account for the discrepancy.

Glucose and lipids both serve as fuels to supply energy in adipose tissues. Glucose-derived pyruvate either converts to lactate or enters the mitochondrial TCA cycle. The catabolism of TAG is mainly dependent on lipolysis, followed by FAO in the mitochondria. In fact, the selection of glucose or lipids as fuels can influence physiological status and functions of cells (34,35,41–44). In addition, our findings that leucine and isoleucine levels decreased in TKT-deficient adipose tissues suggest that BCAA catabolism might also increase due to TKT deletion. Here, our findings showed that in adipose tissues, TKT deficiency caused a significant reduction in conversion from glucose to pyruvate, leading to insufficient carbohydrate-derived energy supply—eventually switching to reliance on lipolysis and possibly BCAA to meet energy demands. However, we found that adipocyte TKT ablation did not alter RER (Supplementary Fig. 3G and H), which often serves as an indicator of which fuel is metabolized to supply energy. We deduced that since RER reflects whole-body fuel consumption, the metabolic changes in adipose tissues due to TKT deficiency may not be sufficient to alter RER, which was consistent with previous studies (45–47).

We show here that TKT deletion increased lipolysis by upregulating the expression of ATGL, HSL, and MGL as early as 3 weeks on HFD, when there was no difference in body weight (Supplementary Fig. 4B–D). This finding indicates that increased lipolysis was a cause of resistance to HFD-induced obesity rather than a secondary result. More importantly, we found that increased lipolysis was caused by decreased glycolysis, as supplementation of pyruvate abolished TKT deficiency–induced lipolysis (Fig. 5E).

Lipolysis liberates a large amount of fatty acids, which not only serve as energy substrates and precursors of other lipids but also are directly involved in cellular signaling pathways and regulation of gene transcription (7). For instance, provision of fatty acids for PPAR activation requires the involvement of lipolysis. In the liver, ATGL regulates the rate of FAO and PPARα activity (48). In the cardiac muscle, ATGL deficiency suppresses cardiac mitochondrial respiration and decreases rates of substrate oxidation and oxidative phosphorylation (49). Consistently, our data suggested that TKT deficiency increased lipolysis by elevating ATGL level and enhanced mitochondrial function by upregulating the expression of ETC genes. Additionally, FAO was also increased in TKT-deficient adipocytes. Furthermore, we found that WAT browning was probably enhanced, whereas BAT thermogenesis was not altered after TKT deletion (Supplementary Fig. 7).

In summary, our study discovered an important role of TKT in regulating lipolysis in adipose tissues. These findings may provide new insights into clinical therapy for obesity and obesity-related metabolic diseases.

Acknowledgments. The authors thank Dr. Jiqiu Wang and Dongqin Gu at the Department of Endocrinology and Metabolism, China National Research Center for Metabolic Diseases, Ruijin Hospital, Shanghai Jiao Tong University School of Medicine for providing reagents and helping with metabolic cages. The authors thank Dr. Haipeng Sun at Department of Pathophysiology, Shanghai Jiao Tong University School of Medicine for providing reagents.

Funding. This work was supported by grants from National Natural Science Foundation of China (81672322, 81972210, and 31601118), the Shanghai Municipal Science and Technology Major Project (19JC1410200), National Key R&D Program of China (2019YFA0906100), National Key Research and Development Program of China (2016YFC1304800), the Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning, and Construction Plan of Laboratory Technical Team in Shanghai Universities (SYjdyx18007).

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

Author Contributions. N.T., L.W., and X.T. designed the research. N.T., Q.L., Y. Li, L.T., Y. Lu, Y.Z., L.H., J.M., M.F., and M.L. performed the experiments. H.C., L.Z., B.L., T.X., and X.T. provided reagents and scientific input. N.T., P.Z., L.W., and X.T. analyzed the data. N.T. and X.T. wrote the manuscript. X.T. 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.