Exercise Training Promotes Sex-Specific Adaptations in Mouse Inguinal White Adipose Tissue

Obesity is a worldwide health crisis that is causally linked with numerous chronic diseases. Over the past decade, there have been increased efforts by the scientific community to investigate the role of adipose tissue in the etiology, prevention, and treatment of obesity. While both white adipose tissue (WAT) and brown adipose tissue (BAT) exist in humans and rodents, accumulation of WAT, particularly in visceral fat depots, is associated with increased risk of cardiovascular disease and chronic inflammatory diseases, including type 2 diabetes (1). In contrast, preferential accumulation of fat in subcutaneous WAT depots is considered metabolically favorable (1). WAT consists of multiple cell types, with the major component being white adipocytes that are specialized to store energy, whereas the major component of BAT is brown adipocytes, which dissipate stored energy as heat by uncoupling oxidative respiration in mitochondria via uncoupling protein-1 (UCP1). Various stimuli lead to trans-differentiation of white adipocytes and the formation of beige adipocytes, a distinct cell type having increased UCP1 and energy expenditure potential compared with white adipocytes (2). All three types of adipocytes can regulate tissue–tissue communication through the secretion of adipokines, and obesity has been shown to dysregulate normal adipokine secretion and function (1).

In contrast to obesity, exercise training decreases adipocyte cell size and lipid content (3,4), increases mitochondrial respiration in WAT, and induces beiging (5–7). We found that as little as 11 days of exercise training alters the expression of >1,500 genes in the subcutaneous inguinal WAT (iWAT) of male mice (7,8) and that iWAT from exercise trained mice has marked beneficial effects on whole-body metabolism (9). Furthermore, exercise training results in profound changes in adipokine profiles of mice, including adipokines that have favorable effects on metabolism (9–11). Thus, an emerging concept is that exercise training–induced adaptations to adipose tissue are an important component in the beneficial effects of exercise on metabolism.

The vast majority of rodent studies to date have been performed in males, presumably to avoid increased variability from hormonal changes associated with the estrous cycle in females. In addition to androgens and estrogens regulating distinct male and female phenotypes, these sex hormones are also important regulators of whole-body metabolic homeostasis (12,13). Testosterone is the most characterized androgen and is considered an important contributor to metabolic function and modulator of WAT metabolism (12–14), and low testosterone levels in men are associated with a significant increase in fat mass and risk for type 2 diabetes (15). Since sex steroid hormones play an essential role in iWAT biology, it is scientifically and clinically important to investigate the effects of exercise training on iWAT phenotype and function in both males and females. In this study, we find that exercise training–induced adaptations to iWAT are fundamentally different between the sexes and identify testosterone as a key regulator of this sexual dimorphism. We also discover a novel, exercise-induced, and sex-specific adipokine in Crisp1, in which expression is increased by testosterone and functions to increase glucose and fatty acid metabolism.

All experiments were conducted following National Institutes of Health guidelines, and protocols were approved by the Joslin Diabetes Center Institutional Animal Care and Use Committee. Ten-week-old male and female C57BL/6 mice (Charles River Laboratories) were housed at 23°C with lights on at 6:30 a.m. and off at 6:30 p.m. Standard mouse diet (9F5020 LabDiet; PharmaServ, Inc.) and water were provided ad libitum. Mice were individually housed with or without a running wheel (24-cm diameter; Nalgene) for 11 days (10). Daily running distance, duration, and speed were recorded by Columbus Instruments hall effect sensors (0297–0501) and multidevice interface software (0297–0051). Body weights were recorded before and after the exercise training period. In vivo whole-body composition was measured (DXA; Lunar PIXImus 2). Running wheels were locked at 6:00 a.m., and tissues were collected between 11:00 a.m. and 1:00 p.m. Mice were sacrificed under isoflurane anesthesia, and blood, triceps, iWAT, epigonadal WAT (eWAT), and interscapular BAT (iBAT) were collected.

Adipose tissue organ culture was performed as described (16) and used for ex vivo testosterone treatment. Briefly, iWAT was washed in PBS containing 5% penicillin/streptomycin, cut into 25-mg pieces, placed in a 12-well culture plate containing M199 supplemented with 1% penicillin/streptomycin (15140122; Thermo Fisher Scientific), 1 µmol/L insulin (11376497001; Sigma-Aldrich), and 250 nmol/L dexamethasone (D1756; Sigma-Aldrich), and maintained at 5% CO₂, 37°C, pH 7.4. After 24 h, the media was replaced with M199 containing 1% penicillin/streptomycin and DMSO (vehicle) or 10 µmol/L testosterone (Sigma-Aldrich).

Oxygen consumption rate (OCR) of tissue was measured by a Seahorse Bioscience XF24 analyzer (17). iWAT was removed, and five pieces (<5 mg) of tissue per mouse were assayed in XF24 V7 (#100777–004) and XF Islet Capture Microplates (#101122–100). Glycolysis stress tests were performed on 3T3-L1 adipocytes using 10 mmol/L glucose, 1 μmol/L oligomycin, and 25 mmol/L 2-deoxy-d-glucose.

iWAT was processed to isolate mature adipocytes and adipose-derived stem cells (ASCs), as described (18). Cell proliferation was evaluated at passage 1 using an MTS Assay Kit (ab197010; Abcam). At passage 4, ASCs were differentiated into adipocytes, and adipogenesis was evaluated by Oil Red O staining (19).

Serum testosterone (ADI-900–065), dehydroepiandrosterone (DHEA) (DI-900–093) (Enzo Life Sciences, Inc.), and CRISP1 (DY4675–05; R&D Systems) were measured by ELISA. Protein concentrations were determined by BCA Protein Assay (23225; Thermo Fisher Scientific). Immunoblotting was performed as described (10) using primary antibodies from Santa Cruz Biotechnology (CRISP1 [Sc21280], UCP1 [Sc6529], and HKII [Sc6521]) or Cell Signaling Technology (GAPDH [2118S] and α-tubulin [2125]). The ChemiDoc Touch System and ImageLab v5.2.1 were used for analysis (Bio-Rad Laboratories). For immunohistochemistry, iWAT was fixed and embedded in paraffin, tissue slices stained with hematoxylin and eosin, and cell size measured (CellProfiler 3.0). Immunofluorescence was performed using UCP1 antibody (Sc6529; Santa Cruz Biotechnology) (20). Surface temperature of mice was determined using an infrared camera (T300; FLIR Systems) as described (21). Briefly, mice were removed from cages with or without wheels, anesthetized with 85 mg/kg body weight pentobarbital, shaved in the inguinal dorsal area, and thermal images taken. All images were analyzed using FLIR Tools software.

For quantitative RT-PCR, RNA was isolated from tissue using an RNA extraction kit (Zymo Research), reverse-transcribed (Applied Biosystems), and cDNA amplified. For each gene, mRNA expression was calculated relative to Actb and 18s. Primer sequences used in quantitative RT-PCR are presented in Supplementary Table 1. Androgen receptor transcription factor chromatin immunoprecipitation PCR was performed on iWAT treated with 10 μmol/L testosterone and 10 μmol/L 5α-dihydrotestosterone (DHT) (D-073; Sigma-Aldrich) for 24 h. ChIPAb+ Androgen Receptor (1710489; EMD Millipore) was used for immunoprecipitation as previous described (22).

In vitro glucose uptake was quantified using the accumulation of [³H]2-deoxyglucose in cells as described (10). In vitro fatty acid uptake was measured using [¹⁴C]palmitic acid, and fatty acid oxidation determined by conversion of ¹⁴C-labeled palmitic acid to ¹⁴CO₂ (10).

Data are means ± SEM. Sample sizes are indicated in the figure legends. All statistical analyses were performed using GraphPad Prism v7. Statistical significance was analyzed by two-way ANOVA followed by Tukey multiple comparisons test or unpaired two-tailed Student t test.

The data sets generated and analyzed during the current study are available from the corresponding authors upon reasonable request.

To determine if there are sex-specific differences in response to voluntary exercise, male and female mice were housed in cages containing wheels for 11 days. Male mice performed 5.9 ± 0.4 km/day of voluntary wheel running exercise, while female mice exercised significantly more at 6.9 ± 0.6 km/day (Fig. 1A). This difference was due to female mice exercising for a longer period of time during the dark phase (15%) (Fig. 1B), resulting in a higher duration of exercise (Fig. 1C). There was no difference in average running speed (Fig. 1D). Despite differences in running distance, triceps muscle HKII protein content, an established marker of exercise training (23), was robustly and similarly increased in both males and females (Fig. 1E and F). Food intake normalized to body weight was not different between males and females, but the increase associated with exercise was higher for females (Fig. 1G). Training did not change body weight, lean mass, or fat mass in females, whereas in males, there was a tendency for a decrease in body weight, an increase in lean mass, and a decrease in fat mass (Fig. 1H–K).

To investigate the effects of training on specific fat depots, we measured fat mass in the three main fat depots in the mouse: iBAT, eWAT, and iWAT. Macroscopically, iBAT, eWAT, and iWAT all appeared darker and smaller in the trained males compared with sedentary, but there was no effect of training in females (Fig. 1L). Training resulted in a pronounced decrease in the mass of iBAT (–65%), eWAT (–33%), and iWAT (–33%) in males, whereas in females, there was only a tendency for a reduction in iBAT (–31%; P = 0.09) (Fig. 1M–O).

Based on our previous study showing that transplantation of exercise-trained iWAT into male C57BL6 sedentary mice results in exercise improved systemic metabolism (9), for subsequent experiments, we focused our studies on the iWAT depot. Immunohistochemical analysis showed that training decreased the diameter of adipocytes from iWAT in males, but not females (Fig. 2A–C). This decrease in adipocyte size was associated with a decrease in the expression of Leptin and an increase in expression of the two major rate-determining enzymes for lipolysis in adipocytes, Lipe (Hsl) and Pnpla2 (Atgl), in males (Fig. 2D–F). In females, training had a tendency to decrease Leptin mRNA expression but did not affect the expression of lipolytic genes (Fig. 2D–F).

Since fat mass is regulated both by preexisting adipocyte volume and the generation of new adipocytes (24), we evaluated the proliferation rate and adipogenesis of ASCs isolated from iWAT. Exercise training reduced the proliferation rate (–18.9%) and differentiation (–34%) of ASCs isolated from the iWAT of male mice (Fig. 2G–I), whereas there was no effect in females. Taken together, these findings show that training significantly reduces iWAT mass, increases gene expression of lipolytic enzymes, and alters the proliferation rate and adipogenicity of ASCs only in male mice.

Previous studies demonstrated that exercise training increases thermogenic brownlike adipocytes in male mouse WAT (5,7,11,25). To determine if this beiging effect is different between sexes, we evaluated UCP1 expression and other beiging markers (26,27) in iWAT of trained male and female mice (Fig. 3A–D). Exercise training increased Ucp1/UCP1 expression in males, but not in females. We next determined if the lack of effect of training on beiging in females also occurred with cold exposure, a well-established beiging stimulus. Interestingly, 11 days of cold exposure (5°C) resulted in similar increases in UCP1 in iWAT in males and females, with the increase in females occurring in the absence of decreases in body weight (Supplementary Fig. 1A–C).

Similar to UCP1, exercise training increased Cidea in males, but not females (Fig. 3E). The Pparg cofactor Prdm16 did not change with training in either sex, but interestingly, baseline levels of Prdm16 were significantly higher in male iWAT (Fig. 3G). Tle3 is a white adipocyte–selective Pparg cofactor that antagonizes Prdm16 function, suppressing the expression of brown-selective genes in adipose tissue (26,28). Exercise training decreased Tle3 expression in iWAT from males but had no effect in females (Fig. 3H). In contrast to Prdm16, in which baseline levels were higher in males, baseline expression of Tle3 was nearly threefold higher in females compared with males. In contrast to the sex-specific expression of UCP1, Prdm16, and Tle3, Dio2 was similarly increased by training in both male and female mice (Fig. 3F).

Since UCP1 generates heat by dissipating the mitochondrial proton gradient, we tested whether exercise training changes the thermogenic capacity of iWAT in vivo using infrared thermography (Fig. 3I and J). Interestingly, thermal images, obtained in mice 6 h after removal from wheel cages, showed that exercise training increased temperature in males but not females (Fig. 3J), consistent with the concept that beiging results in activation of thermogenesis. These striking findings demonstrate that exercise training induces beiging in the iWAT of male, but not female mice. This lack of response in female mice may be specific to exercise training, given that cold exposure increases UCP1 expression in the iWAT of female mice. The sexual dimorphism of the Pparg cofactors Prdm16 and Tle3 may play an important role in the beiging of iWAT induced by exercise training.

Given these findings, we next determined whether the trained iWAT exhibits enhanced mitochondrial function by assessing mitochondrial bioenergetics in intact iWAT from each group of mice (Fig. 4A). Basal OCR (Fig. 4B), proton leak, which is basal respiration not coupled to ATP production (Fig. 4C), and maximal OCR (Fig. 4D) were all increased by training in iWAT from male, but not female mice. Consistent with these data, training in iWAT from male mice increased the expression of genes related to mitochondrial biogenesis and function, including AMPKα1 (Prkaa1/AMPK), Ppargc1a, Esrra (Fig. 4E–G), and the mitofusion genes Mfn1, Mfn2, and Opa1 (Fig. 4H–J). In female iWAT, training did not change expression of these mitochondrial genes, with the exception of Mfn2 (Fig. 4I). Thus, exercise training induces sex-specific metabolic adaptations and responses in terms of iWAT mass, gene expression of beiging markers, thermogenic activity, and mitochondrial uncoupling.

We next explored possible mechanisms by which exercise training induces the striking metabolic adaptations in male, but not female, iWAT. To determine the molecular effects of exercise training on iWAT in male mice, we performed an enrichment network motif analysis for the most upregulated genes (false discovery rate [FDR] <0.01) in our previously published microarray data set derived from iWAT from male mice trained for 11 days (GSE68161) (7). This analysis identified a pattern of interaction that involved estrogen-related receptor α Esrra (Err1), nuclear receptor subfamily 5 group A1 Nr5a1 (Sf1), and androgen receptor (Ar) as the most active transcription factors (Fig. 5A). The presence of steroidogenesis-related transcription factor (Sf1) and androgen receptor prompted us to explore the androgen metabolism pathway in iWAT with exercise training. Sex hormones contribute to the regulation of adipose tissue metabolism (29–31), and in males, testosterone has antiobesity properties associated with reduced fat accumulation, inhibition of adipogenesis (32,33), and increased lipolysis (34). We investigated exercise training–induced concentration changes in testosterone and DHEA, a sex hormone precursor that is mediated by exercise training (35–37). Exercise training had no effect on testosterone concentrations in male or female mice (Supplementary Fig. 2B). In contrast, exercise training increased DHEA concentrations in males and decreased DHEA in females (Supplementary Fig. 2C). Adipose tissue expresses enzymes required for the production of active androgens, like testosterone from C₁₉ circulating precursors (14). In males, training increased gene expression of steroid sulfatase (Sts), a key player in steroid biosynthesis, and steroid 5-α-reductase 2 (Srd5a2), which catalyzes the conversion of testosterone into the more potent DHT (13,17,36,38). Conversely, in female mice, training reduced Sts and did not change Srd5a2 expression (Supplementary Fig. 2D–F). While training did not change gene expression of sex hormone receptors (Ar and Esr1/2) in the iWAT of either sex (Supplementary Fig. 2G–I), in male mice only, exercise training increased mRNA expression of the Ar coactivators Ncoa1 and Ncoa4 (Supplementary Fig. 2J and K) and downregulated the expression of corepressors Pias1 and Daxx (Supplementary Fig. 2L–N). Based on these data showing that male mice respond to training by increasing Ar coactivators and decreasing Ar corepressors, we used the RegNetwork database (39) to perform integration analysis. We compared genes differentially expressed by training in iWAT (FDR <0.25) (7) with a comprehensive set of experimentally observed genes regulated by the Ar (39). This analysis yielded 16 genes, 13 of which were upregulated and 3 downregulated by exercise training and the Ar (Fig. 5B). Subsequent gene set enrichment analysis confirmed a strong association with biological processes such as fatty acid metabolism and mitochondrial compartmentalization (Fig. 5C and E). Remarkably, we identified two androgen response elements in the promoter region of the UCP1 gene (−3,297 and −3,263) where androgen receptor (AR) can bind and initiate UCP1 transcription (Fig. 5F–G).

Based on these findings, we determined whether testosterone phenocopies the beiging effects induced by exercise training in male mice. iWAT was isolated from sedentary male and female mice and incubated with 1 µmol/L testosterone for 24 h. Testosterone incubation of iWAT from male mice significantly increased Ucp1, Prkaa1, Ppargc1a, and Essra mRNA expression (Fig. 5G–J). In contrast, incubation of iWAT from female mice with testosterone had no effect on gene expression (Fig. 5G–J).

Recent data indicate that exercise training promotes the expression and secretion of adipokines (10,40), but whether there is sex specificity in this response is not known. In this study, we focused on identifying adipokines that were upregulated in response to exercise training and modulated by androgens. Of the 16 candidate genes identified by integration analysis (Fig. 5B), Crisp1 emerged as an important protein for investigation. Crisp1 was the only gene of the 16 candidates that encodes as a secreted protein, determined by annotation as “extracellular space” in gene ontology (41). CRISP1 is a cysteine-rich glycoprotein expressed primarily in the epididymis and salivary glands that has been studied in the context of sperm maturation, capacitation, and gamete interaction (42–44). We validated the in silico analysis of CRISP1 in male and female iWAT and found that CRISP1 protein increased with training only in male mice (Fig. 6A). The association with training in male mice was further established by a positive correlation between running distance and CRISP1 mRNA expression in male but not female mice (Fig. 6B). CRISP1 protein increased with training in male, but not female, iWAT (Fig. 6C and D). CRISP1 was detected in iWAT-conditioned media only from male mice, and exercise training in male mice resulted in a significant increase in CRISP1 concentrations in the media (Fig. 6E). CRISP1 serum concentrations were 20-fold higher in male compared with female mice and did not change with training (Fig. 6F).

We found that CRISP1 was expressed in mature adipocytes and not stromal vascular fraction (SVF) (Fig. 6G) and that training increased CRISP1 mRNA expression in the mature adipocytes from male mice only (Fig. 6H). Interestingly, we found that in addition to CRISP1 being expressed in adipocytes from C57BL6 mice, using National Center for Biotechnology Information Gene Expression Omnibus 2R, we found that CRISP1 is also expressed in four out of five additional mouse strains (Supplementary Fig. 3A). Cold exposure had no effect on Crisp1 mRNA and protein expression in iWAT of male and female mice (Supplementary Fig. 3B and C).

Although there was no CRISP1 expression in SVF, when we selected for ASCs, we found increased Crisp1 expression at day 6 of differentiation, but only in male ASCs, and this increase followed Ar expression (Fig. 6I and J). To determine whether testosterone regulates CRISP1 expression and secretion in iWAT, adipose tissue organ cultures of iWAT from both sexes were incubated with testosterone. Testosterone treatment for 24 h induced a robust increase in Crisp1 expression in male iWAT, but not in female iWAT (Fig. 6K). Furthermore, testosterone treatment of iWAT increased CRISP1 concentrations only in the conditioned media from male iWAT (Fig. 6L). Incubating iWAT with testosterone in the presence of flutamide, a selective and competitive androgen receptor antagonist, showed that CRISP1’s androgen-mediated expression in male mice occurs through the Ar (Fig. 6I). Taken together, these data show that CRISP1 is a secreted protein that is released from iWAT of male mice in response to exercise training in vivo and with testosterone incubation in vitro.

Since training alters the expression of adipokines that regulate energy expenditure activating thermogenesis (10,45), we determined whether CRISP1 promotes beiging in cultured primary ASCs. ASCs were isolated from male and female mice iWAT and incubated for 6 days throughout differentiation with 10 μmol/L or 100 μmol/L CRISP1 for males and females, respectively, representing physiological serum concentrations for each sex. CRISP1 treatment induced Ucp1 mRNA and protein expression in differentiated adipocytes (Fig. 7A–C) and other beiging markers such as Ppargc1a (Fig. 7D).

Oil Red O images on day 6 postdifferentiation show how the differentiated adipocytes treated with CRISP1 were more differentiated and took on a multilocular appearance, containing several small lipid droplets (Fig. 7G and Supplementary Fig. 3D). CRISP1 increased expression of adiponectin (Supplementary Fig. 3E) and lipolysis-related genes such as Lipe and Pnpla2 (Fig. 7E and F) in ASCs of both sexes. Based on these data, we also investigated the effects of CRISP1 on fatty acid metabolism and found that incubation with CRISP1 increased fatty acid uptake and oxidation in 3T3-L1 adipocytes (Fig. 7H and I). CRISP1 incubation of 3T3-L1 adipocytes also increased glucose uptake and glycolytic flux (Fig. 7J and K) and increased the expression of Glut4 (Fig. 7L). Taken together, we identify CRISP1 as a novel exercise-induced adipokine that demonstrates sexual dimorphism, promotes thermogenic programming, and regulates fatty acid and glucose metabolism in cultured adipose cells.

Exercise training has remarkable benefits on metabolic health, and recent studies suggest that adipose tissues can play an important role in these adaptations (3,4,40). Surprisingly, we discover that exercise training induces adaptations in the subcutaneous iWAT of male mice, but not female mice. Even though male mice performed less training than female mice, only male mice had pronounced exercise training–induced histological adaptations, increased expression of beiging genes, and increased respiration in their iWAT (Fig. 8). These striking findings suggest that exercise-induced beiging of iWAT is exclusive to males.

While the iWAT in female mice did not show changes in beiging and mitochondrial biogenesis gene expression after exercise training, chronic cold exposure induced a robust increase in the expression of UCP1 protein in both male and female iWAT. These results indicate that the female iWAT is capable of beiging. They also suggest that exercise- and cold-induced beiging occur through distinct mechanisms and that the physiological function of a beige iWAT phenotype in response to exercise and cold may be different. Why exercise-induced beiging of iWAT occurs and why an increased body temperature would be advantageous to a trained animal are issues that are still poorly understood in the field, but our current findings showing that these responses are sex-specific may help address these questions in future studies.

Diet-restricted weight loss has been shown to increase Ucp1 and beiging in mice (46), raising the possibility that the lower body weights in trained males, but not females, are a mechanism for the sex-specific beige phenotype. While we cannot entirely rule out body weight as a contributing mechanism in the trained male mice, we believe it is unlikely that this is the only factor involved in the beiging response. This is because we find no correlation between absolute body weight (Supplementary Fig. 1D and E) and the change in body weight (Supplementary Fig. 1F and G) with Ucp1 expression in male mice. Another example of increased Ucp1 independent of body weight is shown by our finding that cold exposure, which did not change body weight in male mice and actually increased body weight in female mice, resulted in a significant increase in Ucp1/UCP1 expression in iWAT from both males and females.

Sex steroid hormones are likely contributors to the distinct pattern of exercise-induced WAT responses that we observe in male and female mice. We hypothesize that in males, the increase in circulating DHEA with training enhances the synthesis of androgens in iWAT, functioning in an autocrine manner to stimulate Ar signaling and leading to activation of the beiging program. Previous studies in male rodents demonstrate that exercise can increase circulating concentrations of the testosterone/estrogen precursor DHEA (35,37). Our results corroborate these data in males and surprisingly also show a reduced DHEA concentration in female mice. DHEA’s conversion to testosterone and estrogens usually occurs in the gonads, but also occurs to a lesser degree in WAT (14,37). The gene expression data further support our hypothesis, showing that in males, but not females, there is increased expression of the enzymes Sts and Srd5a2, both known to enhance androgen concentrations (13,17,36). Increases in these enzymes in males may upregulate androgen metabolism within iWAT after exercise training, leading to more robust adaptations in the males. Differences in WAT Ar metabolism between the sexes are well documented. While the Ar is expressed in WAT of both sexes, in females, testosterone acts on the Ar in WAT to promote fat accumulation, while in males, testosterone acts on the Ar to promote lipolysis and fatty acid oxidation (30,31,47). Taken together, exercise training, by modulating androgens with iWAT in a sex-specific manner, may control Ar signaling and regulate metabolic processes, including lipolysis and beiging in iWAT. While it is known that both males and females benefit from the effects of exercise training on glucose homeostasis and that under some conditions, males may have greater responses (48), it is likely that some of the mechanisms that mediate these effects may be different between the sexes, and understanding potential interactions among Ar signaling, fat metabolism, and glucose tolerance will be an important area for future investigation.

Tissue–tissue communication, including adipokine secretion to regulate other metabolic tissues, has emerged as an important biological process in the control of systemic metabolism. We previously reported that transplanting iWAT from trained mice into sedentary recipient mice improves glucose tolerance and results in metabolic improvements in other tissues, including skeletal muscle, leading us to hypothesize that exercise-trained iWAT has endocrine effects, releasing adipokines that mediate tissue-to-tissue communication and contributing to improved metabolism (9). Consistent with this, transforming growth factor-β2 was demonstrated to be an exercise-regulated adipokine that controls glucose homeostasis and fatty acid uptake in mice (10), but we hypothesize that there are many more undiscovered adipokines that regulate metabolism. In this study, we discover another novel exercise-induced adipokine in CRISP1, a cysteine-rich secreted protein that has primarily been studied in the context of epididymal maturation (42,49). Our findings clearly demonstrate that exercise training increases CRISP1 in adipose tissue and that CRISP1 regulates many aspects of adipocyte metabolism, promoting glucose and fatty acid metabolism, inducing the expression of UCP1 and mitochondrial genes, and increasing the formation of multilocular lipid droplets in male mice. Given the effects of CRISP1 on adipocytes, it is likely that this adipokine can function in an autocrine and paracrine manner. These discoveries open a new avenue of investigation to determine the role of CRISP1 in mammalian metabolism and how exercise training–induced increases in CRISP1 function in metabolic changes with exercise.

As with the effects of exercise training to increase markers of beiging, the exercise-induced increases in CRISP1 were sex-specific, only showing adaptations in male mice. These dramatic differences in CRISP1 between males and females in both adipose tissue and circulating blood CRISP1 concentrations indicate that this secreted protein is a sex-specific adipokine that may have a key role in mediating the exercise-induced sex-specific adaptions of iWAT in mice. The finding that testosterone mediates CRISP1 expression and secretion in WAT through its binding to Ar further supports the concept that CRISP1 in a sex-specific adipokine. As exercise training regulates the concentrations of sex hormones, and sex hormones control fat mass distribution and function (35), it will be important to further investigate the role of CRISP1 in these processes.

Our current findings demonstrate the critical importance of investigating both male and female rodents in preclinical studies of adipose tissue, as there can be dramatically different adaptations in response to exercise and possibly other stimuli. Further studies in multiple mouse strains will be useful to broadly understand sex differences in exercise training adaptations to iWAT. Studies investigating the effects of exercise training in men and women on beiging have been conflicting (50,51), and thus, translating the current mouse findings to human studies will be important to determine if adipose tissue from men and women respond differently to exercise training and other stimuli. The dramatic differences in adaptations to adipose tissue with exercise training in males versus females are critical for understanding the sex-specific molecular mechanisms through which exercise training regulates WAT metabolism and may produce whole-body improvements in metabolism and metabolic disease. In an era of personalized medicine and the development of targeted therapies for obesity and metabolic disease, uncovering sex-specific adipokines induced by exercise training may play an important role in developing new pharmacological targets for obesity and metabolic disease.

Acknowledgments. The authors thank Dr. Tara MacDonald, Dr. Min Young Lee, Royce Conlin, Jeppe K. Larsen, and Romane Potin for technical assistance, Dr. Hui Pan and Dr. Jonathan M. Dreyfuss for statistical analysis support, and Afsah Dean and Allen Clermont for animal technical support (Joslin Diabetes Center and Harvard Medical School).

Funding. This work was supported by National Institutes of Health/National Institute of Diabetes and Digestive and Kidney Diseases grants R01DK099511 and R01DK101043 (to L.J.G.), K23DK114550 (to R.J.W.M.), and 5T32DK00726042 and 1F32DK12643201 (to M.V.), American Diabetes Association grant 1-17-PMF-009 (to A.B.A.-W.), and the Joslin Diabetes Center DRC (National Institute of Diabetes and Digestive and Kidney Diseases grant P30 DK36836).

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

Author Contributions. P.N., R.J.W.M., and C.R.R.A. designed experiments, performed experiments, analyzed the data, and wrote the manuscript. S.R.-L., A.B.M., L.A.R., H.T., A.B.A.-W., M.V., N.S.M., B.G.A., and M.F.H. performed experiments. L.J.G. designed experiments, analyzed the data, and wrote the manuscript. All authors participated in the manuscript review. All authors approved the final manuscript. L.J.G. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at Cell Symposia: Exercise Metabolism, Sitges, Spain, 5–7 May 2019, the 78th Scientific Sessions of the American Diabetes Association, Orlando, FL, 22–26 June 2018, and the 77th Scientific Sessions of the American Diabetes Association, San Diego, CA, 9–13 June 2017.