Withaferin A Promotes White Adipose Browning and Prevents Obesity Through Sympathetic Nerve–Activated Prdm16-FATP1 Axis

Obesity is defined as a medical condition of abnormal or excessive fat accumulation in adipose tissue, to the extent that the energy homeostasis is disrupted, which has become a major risk factor for the onset and progression of type 2 diabetes, nonalcoholic fatty liver disease, and cardiovascular disease (1,2). Over the last 30 years, the prevalence of obesity in many countries has doubled, or even quadrupled, fueled by excessive food intake, lack of physical activity, and adaptive genetic variation of human that facilitates energy storage (3,4). To date, obesity has become one of the leading causes of death worldwide (5). However, the effective pharmacotherapy for obesity is still lacking.

Withaferin A (WA) is a steroidal lactone derived from Withania somnifera, a traditional medicinal herb prescribed for a variety of ailments owing to its anti-inflammatory, antitumor, antidiabetic, and neuroprotective properties (6). WA is highly lipid-soluble and can cross the blood–brain barrier (7). Recently, it has been reported that WA enhances leptin sensitivity and prevents obesity by targeting hypothalamic neurons. Leptin is a key adipokine that plays crucial roles in the maintenance of energy balance and metabolic homeostasis (8). Leptin activates its signal pathway in hypothalamus, leading to increased fat utilization, elevated energy expenditure (EE), and promoted white adipose tissue (WAT) browning, and thus counteracts obesity development (9,10). The elevated plasma level of leptin coexists with the excessive adiposity in obese human and rodents, which is interpreted as evidence of leptin resistance (11). Leptin resistance impedes the clinical application of leptin for obesity treatment (12,13). Thus, sensitizing the leptin signaling to prevent obesity has attracted widespread attention, proposing a novel strategy for obesity treatment. WA potentiates leptin signaling in the arcuate nucleus, dorsomedial hypothalamic nucleus, and ventromedial hypothalamic nucleus in the hypothalamus to reduce fat mass and protect against high-fat diet (HFD)-induced obesity in mice (14). However, the mechanism underlying the adiposity-reducing effects of WA remain incompletely understood.

Beige adipocytes are brownlike adipocytes dispersed throughout the white fat depots, which have the capacity for uncoupled respiration and heat production (15). Stimulation of the beige fat biogenesis in WAT (also known as “WAT browning”) confers beneficial effects on metabolism, leading to elevated EE, increased thermogenesis, and reduced adiposity (16). In humans and rodents, the WAT depots show potential for WAT browning, although this potential is weakened in older and obese subjects (17,18). Emerging evidence shows that pharmacological agents, environmental stimuli, or physical exercise can induce WAT browning and reduce body weight (19–21). The hypothalamus, as a center for controlling energy balance and metabolic homeostasis, is tightly involved in the regulation of WAT browning through the sympathetic nervous system (SNS) (22,23). Activation of leptin signal in the hypothalamus leads to an increased sympathetic nervous activity (SNA) and an enhanced WAT browning (24). WA potentiates hypothalamic leptin signaling and reduces adiposity (14). However, it is unknown whether the sympathetic nerve–mediated WAT browning contributes to the antiobesity effects of WA.

To explore the neural and molecular mechanisms by which WA induces WAT browning and reduces adiposity, the chemical denervation, genome-wide deep-sequencing analysis, lipidomics analysis, and in vivo RNA interference technology were used in this study. Our findings identified that WA could be a promising agent for the pharmacotherapy of obesity and provided a therapeutically useful way of administering WA for treatment of obesity and associated comorbidities.

Animal experiments were performed in strict accordance with the guidelines of the Ethics Committee of Peking University Health Science Center (LA2016113) and approved by the Animal Care and Use Committee of Peking University. C57BL/6 male mice were obtained from the Department of Laboratory Animal Science of Peking University Health Science Center and the Charles River Laboratories Beijing Branch (Beijing Vital River Laboratory Animal Technology Co., Ltd.). Mice were housed under controlled light (12-h dark/light cycle, with the dark cycle encompassing 8 p.m. to 8 a.m.) and temperature (22 ± 2°C) conditions with free access to food and water. For diet-induced obesity studies, mice were placed on the 60 kcal% HFD (D12492i; Research Diets) at the age of 8 weeks. For thermoneutral studies, mice were housed at 30°C in a light-controlled climatic chamber. During all procedures of experiments, the number of animals and their suffering by treatments were minimized.

Uncoupling protein 1 (Ucp1), peroxisome proliferative–activated receptor, γ, coactivator 1 α (Pgc1α), tyrosine hydroxylase (TH), fatty acid transport protein 1 (FATP1), PR domain containing 16 (Prdm16), and β-actin–specific antibodies were purchased from Abcam (ab-10983; San Francisco, CA), Santa Cruz Biotechnology (sc-13067; Santa Cruz, CA), Millipore (AB152; Billerica, MA), Signalway Antibody (45112 and 25030; College Park, MD), and Sigma-Aldrich (A5316; St. Louis, MO). Anti-rabbit IgG/Alexa Fluor 488 were purchased from Bioss (bs-02950-AF488; Beijing, China). TransZol up, TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix, and TransStart Top Green qPCR SuperMix were purchased from TransGen Biotech (Beijing, China). RNaseZap was purphased from Invitrogen (Beijing, China). Triglycerides, cholesterol, HDL-cholesterol, and LDL-cholesterol reagent kits were purchased from Biosino Bio-Technology and Science Inc. (Beijing, China). Enhanced chemiluminescence assay kits were purchased from Bio-Rad Laboratories (Hercules, CA). WA was purchased from ChromaDex (Irvine, CA). DMSO was purchased from AMRESCO Inc. (Solon, CA). The neurotoxin (6-hydroxydopamine [6-OHDA]) was purchased from Sigma-Aldrich (H4381).

Administration of WA was carried out as previously described (14). For i.p. treatment, mice received 25 μL of vehicle (DMSO) for 4 days as acclimation before WA treatment. Then, WA was dissolved in DMSO (25 μL; 0.2, 2, 20, and 200 μg/kg) and administered to mice once a day for 7, 14, and 21 days or 2 months. Vehicle groups received 25 μL of DMSO during the course of the experiments. All treatments were performed within 90 min of the dark cycle.

Food intake and body weight were measured daily, and the percent increase of body weight was calculated by the following equation: 100 × (body weight of WA or vehicle injected group − initial body weight)/(initial body weight).

Infrared thermal imaging was performed using a FOTRIC thermal camera. Each mouse was placed on a cage top at a fixed distance away from the camera lens. Serial 1-s images (10 Hz) were taken in triplicate at baseline and at 15-min intervals for 1 h after i.p. injection of WA or vehicle. Infrared thermal imaging images were analyzed using AnalyzIR software. For analysis, a constantly sized circular region of interest was drawn over inguinal WAT (iWAT) and brown adipose tissue, and the average temperature was recorded.

After an adaptation to single caging, mice were placed in metabolic chambers on day 6 of WA treatment. During and after the 24-h acclimation, we administered vehicle or WA (2 μg/kg) within 90 min of the dark cycle. Indirect calorimetry recording was performed using an indirect open-circuit calorimeter Oxylet Physiocage System (LE1305 Physiocage 00, LE405 O₂/CO₂ Analyzer, and LE400 Air Supply and Swithching; Panlab, Cornellà, Spain). Room air flowed through each chamber at a rate of 450 mL/min. The O₂ and CO₂ levels were measured during 3-min sampling periods every 30 min, and data were analyzed with METABOLISM software (v2.2.01). Locomotor activity were measured using a two-dimensional infrared light beam. The VO₂ and VCO₂ were expressed in milliliters per minute per kilogram. The respiratory exchange ratio (RER) was determined by the ratio VCO₂/VO₂. The EE was calculated with the Weir equation (25). The mean values for VO₂, VCO₂, RER, and EE of the dark cycle and light cycle were compared for each group.

Sympathetic denervation was carried out as previously described (24). Eight-week-old male mice received 20 microinjections of 6-OHDA (1 μL/injection, 9 mg/mL in 0.15 mol/L NaCl containing 1% [w/v] ascorbic acid) throughout the right or both inguinal fat pads. Sham-operated fat pads received an equal volume of vehicle. Two weeks (unilateral) or 5 weeks (bilateral) after 6-OHDA injections, mice were i.p. injected with WA (2 μg/kg) once a day for 7 days. Body weights were monitored throughout the duration of the experiment. iWATs were harvested for histological/immunofluorescence assessment or processed for Western blot analysis, real-time quantitative PCR (qPCR), and whole-genome sequencing analysis.

iWATs were homogenized with a Polytron in ice-cold RIPA buffer (1% Triton X-100, 10 mmol/L Na₂HPO₄, 150 mmol/L NaCl, 1% sodium desoxycholate (DOC), 5 mmol/L EDTA, 5 mmol/L NaF, and 0.1% SDS) supplemented with protease and phosphatase inhibitors (catalog nos. P8340 and P2850; Sigma-Aldrich), sonicated, and cleared by centrifugation (10,000g, 10 min, at 4°C). Protein concentration in the supernatant was determined by bicinchoninic acid (BCA) assay (PP01; Aidlab). Protein (5 μg) in 1× sample buffer (62.5 mmol/L Tris HCl [pH 6.8], 2% [w/v] SDS, 5% glycerol, and 0.05% [w/v] bromophenol blue) was denatured by boiling at 100°C for 5 min, separated on 8% SDS-PAGE gels, and transferred onto nitrocellulose membrane (T60327; Pall Corporation) by electrophoresis. After electrophoresis, blots were blocked in 5% nonfat milk in Tris-buffered saline and Tween 20 (TBST) for 2 h at room temperature and probed with primary antibody in 5% BSA-TBST overnight at 4°C. After primary incubation, the blots were washed three times in TBST for 15 min, followed by incubation with horseradish peroxidase–conjugated secondary antibody in TBST with 5% nonfat milk for 2 h at room temperature. Then, the membranes were washed three times for 15 min with TBST and developed using an enhanced chemiluminescence assay (Bio-Rad Laboratories), and the band intensities were quantified using ImageJ software (National Institutes of Health).

Total RNA for real-time qPCR was extracted from fat pads and hypothalamus using TRIzol reagent (TransGen Biotech) following the manufacturer’s protocol. Quantification and integrity analysis of total RNA were performed by running 1 µL of each sample on a NanoDrop 5500 (Thermo Fisher Scientific, Waltham, MA). cDNA was prepared by reverse transcription (TransScript One-Step gDNA Removal and cDNA Synthesis Super Mix; TransGen Biotech). The relative expression of mRNAs was determined by the SYBR Green PCR System (Bio-Rad Laboratories). The relative expression of genes of interest was calculated by comparative threshold cycle method, and GAPDH was used as an endogenous control. Sequences of the primers used for real-time qPCR are available in Supplementary Table 1.

iWATs were collected from mice treated with WA (2 μg/kg) or vehicle for 7 days and immediately fixed in 4% paraformaldehyde solution for 48 h. Then, the samples were incubated sequentially with 20% sucrose and 30% sucrose in PBS for 2 d and frozen in OCT compound (Sakura Finetek, Tokyo, Japan). Tissue sections of 10-μm thickness were taken on a cryostat and allowed to air dry on slides, followed by processing or preservation at −80°C according to standard procedure. Tissue sections of 10-μm thickness were stained with hematoxylin-eosin (H-E). Stained slides were analyzed at the indicated magnification, and images were captured by a digital camera (Olympus).

iWAT sections were stained with H-E. Three mice per group were randomly selected. Adipocyte images were acquired using a microscope (Olympus) at ×200 original magnification. The cellular size was quantified by using ImageJ software, and then the number of adipocytes falling into each field with intervals of 5 μm was counted.

iWATs were collected from mice treated with WA (2 μg/kg) or vehicle for 7 days and immediately fixed in 4% paraformaldehyde solution for 48 h. Then, the samples were incubated sequentially with 20% sucrose and 30% sucrose in PBS for 2 d and frozen in OCT compound (Sakura Finetek). Tissue sections of 10-μm thickness were taken on a cryostat and allowed to air dry on slides, followed by processing or preservation at −80°C according to standard procedure. Frozen sections of tissues were subjected to TH staining. Sections were washed in PBS for 10 min, followed by incubation in blocking solution (10% normal goat serum, 0.2% Triton X-100, 2% BSA, and PBS) for at least 1 h at room temperature. Primary antibodies (antibody anti-TH [1:200]) were applied in blocking solution and incubated overnight at 4°C. Sections were washed at least three times with 5-min incubations in PBS plus 0.2% Triton X-100. Then, an Alexa Fluor 488–conjugated secondary antibody (1:200) was applied in blocking solution and incubated at room temperature for 2 h, followed by five washes with PBS plus 0.2% Triton X-100, and nuclei were stained with DAPI. Sections were mounted with VECTASHIELD medium (Vector Laboratories) and analyzed on a microscope (Histology Facility of the Department of Anatomy, Histology and Embryology, Peking University).

Total RNA was extracted from iWATs of WA or vehicle-treated mice using TRIzol reagent (TransGen Biotech). The quality of the RNA was determined with the NanoDrop 5500 (Thermo Fisher Scientific). For library preparation, 3 µg of total RNA/sample was used. Sequencing libraries were generated with NEBNext Ultra RNA Library Prep Kit for Illumina (New England Biolabs, Ipswich, MA). RNA molecules were selected using poly-T oligo-attached magnetic beads, fragmented, and reverse transcribed with the Elute, Prime, Fragment Mix. Then, end repair, A-tailing, adaptor ligation, and library enrichment were performed according to the manufacturer’s instructions. RNA libraries were assessed for quality using the Agilent 2100 Bioanalyzer. The clustering of the index-coded samples was performed on a cBot Cluster Generation System using TruSeq PE Cluster Kit v3-cBot-HS (Illumina). Then, the RNA libraries were sequenced as 100-bp/50-bp paired-end runs on an Illumina HiSeq 2000/2500 platform. Differential expression analysis of two conditions/groups (two biological replicates per condition) was performed using the DESeq R package (1.10.1). The resulting P values were adjusted using the Benjamini-Hochberg approach for controlling the false discovery rate. Genes with an adjusted P value <0.05 found by DESeq were assigned as differentially expressed. Gene Ontology (GO) enrichment analysis of differentially expressed genes (DEGs) was performed using the GOseq R package, in which gene length bias was corrected. GO terms with corrected P values <0.05 were considered significantly enriched by DEGs. Samples were measured and analyzed in Novogene Bioinformatics Technology Co., Ltd.

Prdm16 knockdown, FATP1 knockdown, and control recombinant adeno-associated virus–GFP vectors were constructed in Vigene Biosciences. Vector injection procedures were performed as previously described (26). Briefly, 8-week-old male mice were injected at multiple sites of the iWAT pads (virus diluted in sterile PBS: 1 × 10¹⁰ PFU/100 μL). Two weeks after vector injections, mice were i.p. injected with WA (2 μg/kg) once a day for 7 days. Body weights were monitored throughout the duration of the experiment. iWATs were harvested for histological/immunofluorescence assessment or processed for Western blot analysis and real-time qPCR.

Whole blood was collected by cardiac puncture and transferred to ice-cold EP tubes. The tubes were centrifuged at 2,000g for 30 min at 4°C and stored at −80°C. The serum was used for lipid measurement.

For lipidomics analysis, the extraction of lipids was carried out using the liquid–liquid extraction protocol. Briefly, 50 mg frozen iWAT tissues was dissected followed by addition of 2.5 mL of mixture of chloroform and methanol (v/v = 2:1) on ice, and the mixture was homogenized followed by addition of 1.25 mL distilled water, vortex for 30 s, and then incubation on ice for 5 min. Then, the sample was centrifuged at 12,000g at 4°C for 10 min. The lower chloroform layer was collected and dried by nitrogen. Samples were reconstituted in chloroform and methanol solution (v/v = 2:1), separated with CORTECS C18 columns (2.1 × 100 mm; Waters Corporation), and analyzed by a Q Exactive orbitrap mass spectrometer that was operated in both positive and negative mode using Xcalibur 3.0 software. A full scan followed by 10 data-dependent tandem mass spectrometry scans were acquired using higher energy collisional dissociation with stepped normalized collision energy of 15%, 30%, and 45%. Samples were measured and analyzed in the Technology Center for Protein Sciences, Tsinghua University. The data analysis was performed on LipidSearch software (Thermo Fisher Scientific).

A subcutaneous adipose tissue transcriptomic data set associated with WAT browning and obesity was identified (GSE116801). Log2-transformation was applied as needed. Heat maps of the top 120 significantly changed genes of subcutaneous adipose tissue from humans under the condition of WAT browning induction were carried out using the R/GeneMeta package.

All data are expressed as mean ± SEM. Statistical significance was determined using Student t test (two-tailed) or one-way or two-way ANOVA, as indicated in the figure legends. Numbers of cohorts and n values for each experiment are indicated in the figure legends. All data were analyzed using the appropriate statistical analysis methods with SPSS software (version 19.0). The analysis of EE, VO₂, and VCO₂ in Supplementary Figs. 2, 3, and 4 was based on ANCOVA. Significance was accepted at *P < 0.05, **P < 0.01, or ***P < 0.001.

The data set generated in this study is available from the corresponding author on reasonable request. Accession codes will be available before publication.

To explore the mechanism underlying the antiobesity effects of WA, we first examined whether WA reduces body weight or adiposity by promoting WAT browning. We found that WA treatment at dosage levels of 2, 20, or 200 μg/kg for 7, 14, or 21 days reduced body weight and fat pad weights, without altering food intake in HFD-fed mice (Supplementary Fig. 1A–L). WA at dosage levels of 2, 20, or 200 μg/kg led to an increased expression of WAT browning–associated genes in the iWAT (Supplementary Fig. 1M–O). Thus, WA treatment at a dosage level of 2 μg/kg was selected for further study. We performed WA treatment (2 μg/kg i.p. once daily for 2 months) to examine the long-term effects of WA (Fig. 1A). We found that WA treatment led to a significant and sustained reduction in body weight, equivalent to a total weight loss of 16.1% at week 9 (Fig. 1B). WA reduced the weights of inguinal, epididymal, and perirenal fat pads, but did not affect food intake and the weight of brown adipose tissue (Fig. 1C and D). WA increased expression levels of WAT browning–associated genes in iWAT (Fig. 1E). Notably, WA reduced body weight and increased WAT browning–associated gene expressions at as early as week 1 (Fig. 1B and E). Therefore, the i.p. injection of WA at 2 μg/kg for 7 days was selected for further study. The thermoneutrality (30°C for mice) is critical for examining the probrowning effects of chemical compounds; thus, we performed the experiments at either room temperature or thermoneutrality (Fig. 1F). We observed that WA reduced body weight and fat pad weights at either room temperature or thermoneutrality, with no change in food intake (Fig. 1G–J). The canonical WAT browning phenotypes, such as “brown-like” adipocytes with smaller and multilocular lipid droplets in iWAT, were observed in WA-treated mice (Fig. 1K). WA increased the expression levels of the genes involved in WAT browning, SNA, and adaptive thermogenesis in iWAT (Fig. 1L). The expression levels of UCP1 and PGC1α were significantly elevated in iWAT (Fig. 1M and Supplementary Fig. 2A). The thermogenesis of iWAT was increased during the light phase (when mice were inactive) in WA-treated mice (Fig. 1N). WA led to enhanced whole-body EE and decreased RER (Fig. 1O and P and Supplementary Fig. 3A), suggesting that WA may promote a predominant utilization of fat as a fuel source. WA did not affect the daily locomotor activity (Supplementary Fig. 3B). Together, these findings show that WA induces WAT browning, enhances EE, and reduces body weight, suggesting that WA may prevent HFD-induced obesity by promoting WAT browning, and the SNS may contribute to mediating the WAT browning–inducing effects of WA.

To determine the role of sympathetic nerves in mediating the WAT browning–inducing effects of WA, we assessed the leptin signal activity in the hypothalamus and also examined the SNA in iWAT. We found that WA treatment upregulated the expression levels of the genes associated with activation of the classic leptin signal pathway (LepR, JAK2, Stat3/5, PI3K, AKT, mTOR, S6K1, PDE3b, AMPK, Mchr1, and Pomc) and also increased the expression levels of the positive regulators of leptin signaling (Rock1, Sitr1, MFN2, and BDNF) (Fig. 2A). TH, a key synthase of norepinephrine and an indicator of the activity of sympathetic nerves, was significantly increased in iWAT of the mice treated with WA (Fig. 2C and D and Supplementary Fig. 4A). Of note, the unilateral sympathetic denervation of iWAT by 6-OHDA restored the size of adipocytes in iWAT, as compared with contralateral sham operation (Fig. 2E and F). Moreover, the WA-increased expression levels of WAT browning–associated genes were also normalized by unilateral sympathetic denervation (Fig. 2G and H and Supplementary Fig. 4B). Taken together, these findings indicate that WA enhances leptin sensitivity in the hypothalamus and increases the sympathetic activity in iWAT, and the sympathetic innervation is required for the WA-induced WAT browning.

To examine the extent to which the sympathetic nerve–mediated WAT browning may contribute to the elevation of EE as well as the reduction in body weight in WA-treated mice, we performed bilateral denervation of iWATs (Fig. 3A) (24). We found that the reduced body weight, elevated EE, as well as decreased RER were diminished by bilateral denervation of iWATs in WA-treated mice (Fig. 3B–E and Supplementary Fig. 5A). The motor activity was not affected by bilateral denervation (Supplementary Fig. 5B). Notably, EE was elevated by 10.1% in the bilaterally sympathetic denervated mice treated with WA compared with 25.3% in sham-operated mice treated with WA (Fig. 3D), highlighting that the activity of sympathetic nerves in WAT is essential for mediating the obesity-preventive effects of WA. The lipidomic analysis showed that WA treatment increased the levels of oleic and arachidonic acid and decreased the level of a saturated fatty acid (tricosanoic acid) in iWAT, and these were dampened by iWAT bilateral sympathetic denervation (Fig. 3F), suggesting that WA may modify the fatty acid composition in iWAT via SNS, which may also be implicated in the promotion of the biological process of WAT browning. Taken together, these observations suggest that the sympathetic denervation abrogates WAT browning by blocking the sympathetic signal from the hypothalamus to the WA, and demonstrate that WA-induced WAT browning is mediated by the hypothalamus-SNS-WAT axis.

To dissect the molecular mechanism underlying the WA-induced WAT browning, we performed a genome-wide transcriptomic sequencing analysis for iWATs of both denervated mice and sham-operated mice. We visualized the gene expression profiles as a heat map and observed that WA remarkably increased expression levels of the genes associated with WAT browning (Ucp1, Prdm16, and Ppargc1a), mitochondrial biogenesis (Mtfp1 and Slc25a25), lipid metabolism (FATP1, Fabp4, and Pdk4), glucose metabolism (Sik2, Irs1, and Pfkfb1/3), and nervous system function (Negr1, Nr4a3, Trim67, Unc13a, and Dok6), while reduced expression levels of the genes associated with immunity (Ccr5, Oas1g, Oas1a, and Orm2), neural inhibition (Slc6a2 and Mt3), and apoptosis (Cadm1 and Gzma) (Fig. 4A and B). These pleiotropic effects of WA were abrogated by the iWAT sympathetic denervation (Fig. 4A and B). GO and Kyoto Encyclopedia of Genes and Genomes analyses showed that WA activated the pathways associated with adaptive thermogenesis, insulin signaling, and kinase activity regulation, while they inhibited the pathway relative to inflammatory response, and these were dampened by sympathetic denervation (Fig. 4C–E). Volcano plot and Venn diagram analysis showed that WA markedly increased expression levels of FATP1, Ucp1, Prdm16, and Pparγ in iWAT, which were also abolished by sympathetic denervation (Fig. 4F–H). GeneMANIA prediction showed a potential interaction and cross talk between Prdm16 and FATP1 in iWAT treated with WA (Fig. 4I). By analyzing a human subcutaneous adipose tissue data set, we found that the WAT browning–promoting condition, such as physical exercise, also robustly increased the expression levels of Prdm16 and FATP1, demonstrating that Prdm16 and FATP1 may be important factors in the regulation of WAT browning in humans (Fig. 4J). Taken together, these findings provide a comprehensive insight into the transcriptomic modifications induced by WA and demonstrate that Prdm16 and FATP1 may be critical mediators in the biological process of WA-induced WAT browning, which contributes to understanding of the potential mechanism by which WA induces WAT browning and prevents obesity.

To evaluate the roles of Prdm16 and FATP1 in WA-induced WAT browning and weight loss, in vivo RNA interference technology was used (Fig. 5A). Prdm16 or FATP1 knockdown restored the adipocytic sizes and expression levels of WAT browning–associated genes (Fig. 5B–D) and also diminished the reduction in both body weight and fat pad weight of WA-treated mice (Fig. 5E and F). The food intake was not affected by Prdm16 or FATP1 knockdown (Fig. 5G). Knockdown of Prdm16 or FATP1 dampened the increased EE as well as the decreased RER under the condition of WA treatment (Fig. 5H and I and Supplementary Fig. 6A). Intriguingly, we observed that Prdm16 knockdown remarkably downregulated FATP1 expression, with or without WA treatment (Fig. 5J and K). In contrast, FATP1 knockdown did not change Prdm16 expression level (Fig. 5J and K). Collectively, these findings suggest that FATP1 may be a downstream effector of Prdm16, and this Prdm16-FATP1 axis may mediate the WAT browning–inducing and body weight–reducing effects of WA.

To determine whether Prdm16-FATP1 axis may also be associated with the sympathetic activity in human adipocytes, we performed a correlation analysis among Prdm16, FATP1, PGC1α, and DβH (the enzyme converting dopamine to norepinephrine) using the GTEXv5 human subcutaneous and visceral adipose tissue databases. We found that Prdm16 and FATP1 mRNA levels were positively correlated with both PGC1α and DβH (Supplementary Fig. 5A–H). This analysis verified that Prdm16 and FATP1 may also be closely correlated with the SNA and mitochondrial function in the human adipocytes. In summary, the findings in this study point to the model illustrated in Fig. 6.

In conclusion, this study demonstrates the therapeutic potential of WA for obesity treatment and reveals that Prdm16 may be the main transcription factor responsible for WA-induced WAT browning through the sympathetic nerve–adiposity axis, and FATP1 may be the main fatty acid transporter downstream of Prdm16. These findings indicate that WA promotes WAT browning and prevents HFD-induced obesity through sympathetic nerve–mediated activation of the Prdm16-FATP1 pathway.

WA is a liposoluble natural compound with an ability to cross the blood–brain barrier. WA exerts profound neuropharmacological effects, including promotion of neurite outgrowth, mitigation of neuritic atrophy, and facilitation of synapse reconstruction (7,27). As a newly recognized leptin sensitizer, WA prevents obesity by potentiating hypothalamic leptin signaling (14). However, the mechanism by which WA prevents obesity remains to be elucidated.

In this study, we found that the i.p. treatment of WA promoted WAT browning, elevated EE, and prevented HFD-induced obesity, without affecting food intake. Our long-term study showed that WA exerted sustained antiobesity effects over a period of 2 months. Emerging evidence indicates that WA exerts its unique antiobesity effects by enhancing the hypothalamic leptin signaling. WA treatment leads to an increased leptin sensitivity and reduced body weight in diet-induced obesity mice, but not in lean mice, leptin-deficient (ob/ob) mice, or leptin receptor–deficient (db/db) mice (14). WA increases the level of phosphorylated STAT3 (Tyr705) and reduces the level of phosphorylated PERK (Thr980) in hypothalamus, demonstrating that WA relieves endoplasmic reticulum stress and sensitizes leptin signaling (14). In this study, we also observed that WA upregulated expression levels of the genes associated with leptin signaling in the hypothalamus, confirming that the enhancement of leptin sensitivity contributes to the antiobesity effects of WA.

In this study, we observed that high dosage of WA reduced food intake, whereas lower dosages of WA prevented weight gain without affecting daily food intake in HFD-fed mice. Several lines of evidence show that WA exerts multiple effects in a dose-dependent manner in vitro and in vivo (28–31). For example, a low dosage of WA protects against palmitic acid–induced insulin resistance and increases cell viability of human umbilical vein endothelial cells (30), whereas a high dosage of WA induces apoptosis and cell death (28,29). These findings suggest that the discrepancy of food intake in WA-treated mice might also be due to the different dosage of WA.

Multiple lines of evidence show that the hypothalamus-SNS axis plays crucial roles in the regulation of WAT browning (32,33). A series of antiobesity hormones (leptin and adiponectin), pharmacological agents (butein and phyllodulcin), environmental stimuli (cold exposure), and physical exercise promote WAT browning through the hypothalamus-SNS axis (19–21,24). The visualization of the sympathetic neuro–adipose connections strengthens the notion that sympathetic nerves are essential for beige adipogenesis (34–36). In this study, we found that WA elevated EE by 10.1% in the bilaterally sympathetic denervated mice, whereas WA elevated EE by 25.3% in sham-operated mice. Together, these observations suggest that WA-induced WAT browning by the sympathetic nerve–adiposity connection essentially contributes to the antiobesity effects of WA.

Several lines of evidence show that the chemical denervation may serve as a reliable surgical procedure for investigating the mechanism underlying the WAT browning (24,37), although the chemical denervation may affect adipocytes and also blood vessels (38). In this study, we evaluated the vascular responses in iWAT under the condition of chemical denervation through adoption of several measures. We observed that the denervated iWAT showed normal volume and gloss without adipose atrophy or ischemic necrosis, revealing that the blood flow in iWAT was not obviously affected by the chemical denervation. Thus, these observations suggest that the diminished WAT browning in the denervated mice treated with WA might be mainly caused by sympathetic denervation of adipocytes.

The sympathetic nerves regulate the adipocytic factors associated with WAT browning. Stimulation of sympathetic nerves increases the level of connexin 43 in WAT (39); and WAT-specific connexin 43 overexpression promotes WAT browning (37). In this study, we observed that Prdm16 and FATP1 in WAT were upregulated by WA, and this was blocked by sympathetic denervation. The inhibition of Prdm16 or FATP1 in WAT diminished the WAT browning–inducing and body weight–reducing effects of WA. Moreover, the gene expression profile analysis of human subcutaneous fat (40) also revealed that Prdm16 and FATP1 may be critical factors for the maintenance of fat homeostasis in human.

Prdm16 is a key adipocytic transcription factor governing the development of beige adipocytes (41–44). Stimulation of sympathetic nerves activates Prdm16 in white adipocytes, which further activates the Pgc1α/Pparα/Pparγ/retinoid X receptor pathway to upregulate UCP1 and lead to WAT browning (41,45). The fatty acid transporters (FATPs) in white adipocytes are activated in response to the conditions that induce WAT browning. FATPs are the main fatty acid suppliers in beige adipocytes (46–48). The activation of FATP1 enhances WAT browning and protects against HFD-induced obesity (49,50). In humans, the polymorphisms and mutations in FATP genes (FATP1, FATP4, and FATP5) are associated with the pathological lipid metabolism and insulin signaling, showing that FATPs may be potential drug targets for obesity and type 2 diabetes therapies (51–54). These findings suggest that the Prdm16-FATP1 axis may be essential for mediating the WAT browning induced by WA.

It has been reported that Prdm16 and FATP1 in adipocytes can be robustly induced by several natural agents for obesity treatment, such as berberine (55,56), suggesting that Prdm16 and FATP1 may be the mediators of some natural products in the processes of WAT browning induction and body weight reduction. Intriguingly, FATP1 or Prdm16 in iWAT cannot be induced by some browning-inducing conditions, such as cold exposure (57) (GSE153091). In this study, the deep-sequencing analysis showed that Prdm16 and FATP1 were the top genes in iWAT of the mice treated with WA; and Prdm16 or FATP1 knockdown weakened the WAT browning induced by WA. These findings demonstrate that Prdm16 and FATP1 may play critical roles in mediating the effects of WA on WAT browning.

Notably, in this study, we observed a 1- to 2-g difference in body weight between short hairpin green fluorescent protein (shGFP-WA) and shPrdm16-WA/shFATP1-WA groups accompanied with an ∼0.1-g increase in fat-pad weight in iWAT (Fig. 5E and F). This discrepancy might be due to the weight change of other adipose tissues, such as subcutaneous WAT, other organs that contain white fat, or other possible factors. Thus, the body composition analysis could be essential for deciphering the mechanisms underlying the anti-obesity effects of natural compounds.

Mounting evidence indicates that external stimuli that promote WAT browning may also lead to an alteration in fatty acid composition in adipose tissues (58). Importantly, the modification of fatty acid composition in WATs contributes to the promotion of WAT browning, which has been considered a novel strategy for obesity treatment (59–61). The unsaturated fatty acids, such as oleic acid and arachidonic acid, facilitate the mitochondrial uncoupling and oxidative phosphorylation in adipocytes, contributing to the prevention of obesity, fatty liver disease, and cardiovascular disease (62–64). In this study, we found that WA increased the levels of oleic acid and arachidonic acid and reduced the level of tricosanoic acid in iWAT, and these were normalized by sympathetic denervation. Therefore, we cannot rule out that the modification of fatty acid composition caused by WA may also be implicated in the antiobesity and the metabolic beneficial effects of WA.

In summary, this study demonstrates that WA prevents HFD-induced obesity by promoting WAT browning through the sympathetic–adipose axis. The Prdm16-FATP1 pathway in adipocytes mediates the WA-induced WAT browning. These findings suggest that WA could be a candidate of antiobesity agents and provide a new insight into obesity treatment and the maintenance of human health.

Acknowledgments. The authors thank G. Dodd and T. Tiganis (Department of Biochemistry and Molecular Biology, Monash University, Melbourne, VIC, Australia) for excellent technical support in the chemical denervation; L. Tian, J. Wang, and K. Wang (Histology Facility of the Department of Anatomy, Histology and Embryology, Peking University) for technical support; and Y. Wan (Key Laboratory for Neuroscience, Peking University), S. Yu (Peking University School of Pharmaceutical Sciences), S. Zhu, X. Wang (Department of Physiology and Pathophysiology, Peking University), and M. Ye (Peking University School of Pharmaceutical Sciences) for kindly providing access to necessary equipment. The authors also thank the team of the Biomedical Sequencing Facility at Novogene for support with next-generation sequencing and data analysis and X. Liu (Center of Biomedical Analysis, Tsinghua University) for support with lipidomics analysis.

Funding. This work was supported by grants from the National Key Research and Development Program of China (2017YFC1700402 to R.Z.), the National Natural Science Foundation of China (81471064, 81670779, and 81870590 to R.Z.), the Beijing Municipal Natural Science Foundation (7162097 and H2018206641 to R.Z.), and the Peking University Research Foundation (BMU20140366 to R.Z.).

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

Author Contributions. B.G. and J.L. performed the experiment and analyzed the data. B.W., C.Z., Z.S., and M.Z. participated in experiments. J.L. and R.Z. wrote the manuscript. L.Q. and W.Z. edited the manuscript. R.Z. conceived the idea of the study and contributed to manuscript writing. All authors reviewed and approved the manuscript for submission. B.G., J.L., and R.Z. are the guarantors of this work and, as such, had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.