All rat studies were approved by the Ethics Committee of Animal Experiments at Shanghai Tongren Hospital, Shanghai Jiao Tong University School of Medicine. The 21-day-old female Sprague–Dawley (SD) rats were purchased from China Three Gorges University Laboratory Animal Center, and all rats were allowed to adapt to the environment for 1 week. All rats were randomly divided into two groups: a control group (Ctrl) and a DHEA-induced PCOS model group (DHEA). Rats in the DHEA group were treated daily with a subcutaneous injection of DHEA (cat. no. SJ-HS0488, XXJL) for 21 days (6 mg/100 g bodyweight, dissolved in oil). The Ctrl group was injected with the same amount of oil. The successful PCOS model was determined by a significantly increased anti-Müllerian hormone (AMH) and luteinizing hormone/follicle-stimulating hormone (LH/FSH) ratio (8, 22, 23) and disordered estrous cycles. To verify the successful PCOS-like model, we randomly chose 5 control rats and 10 DHEA rats to test the levels of LH, FSH, and AMH. The estrous cycles were monitored for 10 days with a vaginal smear from day 11 to day 21 after DHEA or oil injection. In addition, two PCOS model rats and two control rats were randomly sacrificed to observe the ovarian morphology using hematoxylin and eosin (H&E) staining.

The PCOS model rats were randomly divided into two groups: a sham-operated (DHEA+Sham) group and a BAT transplantation (DHEA+BAT) group. Donor rats within 14 days after birth were operated on to take BAT (0.5 g of scapula), the peripheral white fat removed, and washed with sterile phosphate-buffered saline (PBS) at the same time. The recipient rats were intraperitoneally anesthetized. BAT was subcutaneously transplanted into the back of the recipient. For the DHEA+Sham group, the same procedure was used, except receiving donor tissues. These rats were kept for 3 weeks after operation. After 11 days of operation, the stages of the estrus cycle were determined by vaginal smear for consecutive 10 days. Finally, the rats were sacrificed to collect the BAT and ovarian tissues for further study (H&E staining, immunohistochemistry, and detection of metabolites).

Approximately 50 mg of BAT was added into 0.5 ml of solvent (methanol/water = 8:2), containing 4 µg/ml 2-chloro-l-phenylalanine as an internal standard, and then ground, ultrasonicated at room temperature (25–28°C) for 10 min, and finally stored at −20°C for 30 min. After centrifugation at 13,000 rpm at 4°C for 10 min, 200 µl of the supernatant was taken for subsequent metabolomics analysis. Ten microliters of the supernatant from all samples was saved and mixed for quality control.

The BAT and ovaries were fixed in 4% formaldehyde and embedded in paraffin; slices of 5 μm thickness were sectioned. After deparaffinization and rehydration, the sections were processed for blocking of endogenous peroxidase activity and antigen retrieval pretreatment, followed by blocking in 5% bovine serum albumin at room temperature for 20 min. The sections were then incubated overnight at 4°C with primary rabbit polyclonal uncoupling protein 1 (UCP1) antibodies (1:300, RRID: 72298; Cell Signaling Technology, Danvers, MA, USA). The second antibody was incubated for 30 min after washing three times with PBS. The signals were visualized with DAB incubation. Images were taken with the Digital Pathology Slide Scanner (KF-PRO-120, Ningbo Jiangfeng Medical Technology).

The rats were fasted for 16 h (1700–0900 hours) with free access to drinking water and then injected intraperitoneally with d-glucose (2.0 g/kg body weight) for glucose tolerance test (GTT). Blood glucose levels were measured before the injection and at 15, 30, 60, 90, and 120 min after injection using an Accu-Chek glucose monitor (Roche Diagnostics Corp., Indianapolis, IN, USA). Female rats were fasted for 4 h (0900–1300 hours), with free access to drinking water, and injected intraperitoneally with insulin (1 U/kg body weight; Humulus; Eli Lilly, Indianapolis, IN, USA) for the insulin tolerance test (ITT). Blood glucose levels were measured before the injection and at 15, 30, 60, 90, and 120 min after insulin injection.

Ultra-performance liquid chromatography (Ultimate 3000) combined with the Thermo Orbitrap Elite Mass Spectrometer was used for liquid chromatography–mass spectrometry (LC-MS) analysis. The flow rate was set to 0.4 ml/min with mobile phase A of 0.1% formic acid solution and mobile phase B of acetonitrile (0.1% formic acid). The column temperature was 25°C. Post time was set to 5 min to balance the system. MS uses the positive ion mode combined with the negative ion mode.

The Compound Discoverer software (Thermo Scientific, San Jose, CA, USA) was used to analyze the data. Post-editing was performed in EXCEL 2007 software. In order to obtain consistent differential variables, the resulting matrix was further optimized by removing all peaks with ion intensity = 0 in more than 80% of the samples. The data were normalized to the peak area of the corresponding internal standard, and the internal standard was used for reproducibility. Finally, the ion peaks generated by the internal standard were eliminated. Then, the edited data matrix was imported into Simca-P software (version 11.0). Before the multivariate statistical analysis, the data were mean centered and Pareto scaled. Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were conducted to analyze the dissimilarity tendency among groups. Variable importance in projection (VIP) >1.0 and p-values <0.05 were selected as statistically significant according to the PLS-DA model. The Kyoto Encyclopedia of Genes and Genomes (KEGG) online database pathway enrichment analysis and ingenuity pathway analysis (IPA) were applied to understand the functions and interactions of the genes and metabolites in biological systems.

The steroidogenic human granulosa cell-like tumor cell line (KGN) was maintained in Dulbecco’s modified Eagle’s medium (DMEM)/F12 supplemented with 10% fetal bovine serum (FBS) (both from Gibco, Amarillo, TX, USA), 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified atmosphere at 37°C with 5% CO₂. The growth medium was changed every 2–3 days.

Apoptosis was detected using an annexin V-FITC apoptosis detection kit (RRID: 556547; BD, Franklin Lakes, NJ, USA). KGN cells were seeded into six-well plates (3 × 10⁵ cells per well) and starved for 4 h in FBS-free medium. After stimulation with dihydrotestosterone (DHT) (Solarbio, Beijing, China) for 4 h, sphingosine was added (10 μm). The cells were detached using trypsin, washed with cold PBS twice, and then each well was incubated with 5 μl annexin V-FITC and 5 μl propidium iodide (PI) at room temperature for 15 min in the dark. Cells were detected by flow cytometry.

All the results were presented as the mean ± standard error of the mean (SEM). Univariate analysis of variance (ANOVA, with post-hoc Scheffe test) was applied to determine the significance of the observed differences among the Ctrl, DHEA+Sham, and DHEA+BAT groups using SPSS 26 for Windows (IBM, Armonk, NY, USA), whereas two groups were compared using unpaired Student’s t-test in the Ctrl and DHEA groups. GraphPad Prism 9.0 was used for other data analyses. A p < 0.05 was considered statistically significant.

In order to determine the active adipokines of BAT related to PCOS, BAT transplantation in PCOS model rats was established. The experiment design is illustrated in Figure 1A . The PCOS model validation, estrous cycle monitoring, and PCOS-like phenotype observation were conducted in the scheme.

PCOS is always accompanied by altered plasma gonadotropin concentrations. The levels of LH, FSH, and AMH were tested to verify whether the PCOS model was a success. Compared with those in the Ctrl group, the concentrations of LH, FSH, and AMH and the LH/FSH ratio were significantly higher in the DHEA group ( Figure 1B ). The vaginal smear results of the rats showed disordered estrous cycles in the DHEA group ( Figure 1C ). Two PCOS model rats and two control rats were randomly sacrificed for H&E staining. H&E staining was performed to determine the alteration of ovarian pathology. In the DHEA group, multiple cystic follicles appeared with thinner granulosa cell layers, which were vacuolated and disordered in structure with corpus luteus ( Figure 1D ).

UCP1 is the critical regulator of thermogenesis, as a specific marker of BAT (24). The BAT of donor rats transplanted to rats in the DHEA+BAT group was verified by immunohistochemistry ( Figure 1E ). The results of UCP1 immunohistochemistry indicated that the transplanted tissues were BAT. The results mean the successful establishment of BAT transplantation in PCOS model rats.

Irregular menstruation is one of the diagnostic criteria of PCOS. To explore whether BAT transplantation could recover the ovarian performance of PCOS rats, vaginal smear was performed to observe the effect of BAT transplantation on the estrous cycle in PCOS model rats ( Figure 2A ). In our study, we found that the Ctrl group had an ordered and complete estrous cycle, whereas a disordered estrous cycle manifested in the DHEA+Sham group. BAT transplantation rescued the phenotype of disordered estrous cycles in PCOS rats. Compared with the DHEA+Sham group, follicles of different stages and multiple luteal bodies reappeared with BAT transplantation ( Figure 2B ). Together, these results indicate that BAT transplantation can improve the ovarian performance of PCOS model rats.

Water and food were sufficiently provided for ad libitum intake. All rats were weighed every week, for a total of 6 weeks. The results showed that the mean weight of the DHEA group had significantly decreased compared to that of the Ctrl group on days 7, 14, and 21 before BAT transplantation ( Figure 3A ). The PCOS model rats were randomly divided into two groups on day 21: a DHEA+Sham group and a DHEA+BAT group. We compared the effects of BAT transplantation on the mean weight of PCOS model rats, and it was found that the mean weights of rats in the DHEA+BAT group increased compared with those in the DHEA+Sham group on days 28, 35, and 42. However, a significant difference was only observed on day 35 when comparing the DHEA+Sham group and the DHEA+BAT group.

PCOS is also caused by a metabolic disorder, which is characterized by impaired glucose tolerance and insulin tolerance. The GTT was performed on day 39 and the ITT performed on day 42 to observe the metabolic changes. The GTT results showed a delayed glucose clearance in the DHEA+Sham group, and BAT transplantation partially reversed it ( Figure 3B ). For the ITT, the DHEA+Sham group had higher glucose levels ( Figure 3C ). In addition, the UCP1 levels decreased in the DHEA+Sham group and were recovered in the DHEA+BAT group ( Figure 3D ). These results indicate that BAT transplantation ameliorates insulin resistance and corrects the metabolic abnormality in PCOS model rats.

BAT transplantation could activate the recipient’s BAT function; thus, there are metabolic differences between the DHEA+Sham and DHEA+BAT groups. In order to identify the active metabolites of the recipient’s BAT, a series of multivariate variable pattern recognition analyses were carried out using LC-MS. Firstly, PCA was established to determine the separation tendency between the DHEA+Sham and DHEA+BAT groups ( Supplementary Figure S1 ). A two-component PCA model was obtained with the following parameters: R ² X = 0.506, Q ² = 0.284 (positive mode) and R ² X = 0.517, Q ² = 0.364 (negative mode). To further specify the metabolic variations associated with PCOS, a supervised PLS-DA model was established with two predictive components (positive mode: R ² X = 0.424, R ² Y = 0.925, Q ² = 0.312; negative mode: R ² X = 0.467, R ² Y = 0.925, Q ² = 0.607). As shown in Figure 4A , a clear separation was obtained in the scores plot, with all the DHEA+Sham in the left half and DHEA+BAT in the right half. To further validate the established model, a 200-time permutation test was performed for the PLS-DA model. The Y-axis intercept for Q ² was below 0 [Q ₂ intercept (0, −0.0521) (positive mode) and (0, −0.166) (negative mode)] ( Figure 4B ). These results validate the current supervised model. The PLS-DA results demonstrated significant metabolic differences between the DHEA+Sham and DHEA+BAT groups. Moreover, significant biochemical changes were induced by BAT transplantation.

A total of 321 metabolites were identified in this study. Further metabolomics analysis identified a total of 22 differential metabolites ( Table 1 ), including 9 upregulated and 13 downregulated metabolites, when comparing the DHEA+BAT group with the DHEA+Sham group. The differential metabolites with VIP > 1 and p < 0.05 were clustered and shown as a heatmap between the DHEA+Sham and DHEA+BAT groups ( Figure 5A ). The results were visualized in a volcano plot of all metabolites, as shown in Figure 5B . KEGG pathway enrichment analysis of the differential metabolites showed their connection with the adipocytokine signaling pathway, insulin resistance, etc. ( Figure 5C ). Important analysis of the metabolic pathways in the bubble diagram showed that the differential metabolites were associated with amino acid metabolism, glycerophospholipid metabolism, and pyruvate metabolism ( Figure 5D ).

The differential metabolites were roughly categorized into 12 distinct classes, mainly including glycerophosphoinositols, glycerophosphocholines, and sphingolipids ( Figure 6A ). IPA of the differential metabolites revealed several related signaling pathways, such as the PI3K/AKT, MAPK, and Wnt signaling pathways, which have been proven to be closely associated with PCOS (25–27) ( Figure 6B ). These altered metabolites might explain the positive effect of BAT transplantation on PCOS model rats.

The relative differences among the metabolite classes revealed significant changes in the abundance of metabolites. Glycerophosphoinositols, sphingolipids, aldehydes, and amino acids were decreased ( Figures 7A, B, D, E ), while glycerophosphocholines and peptides were increased in the DHEA+BAT group ( Figures 7C, F ).

To further explore the biological roles of metabolites, we randomly chose the differential metabolite sphingosine. KGN cells were pretreated with DHT to mimic the pathophysiological status of PCOS. DHT has been confirmed to be capable of inducing rodent models to exhibit similar reproductive and metabolic features to PCOS patients (28) and has been used in KGN cells at a dosage of 25 nM to imitate the physiological characteristics of PCOS (29). Our results showed that sphingosine significantly increased the apoptosis of DHT-treated KGN cells ( Figure 8 ).