Adissp activates insulin-independent glucose disposal and energy expenditure in white fat to treat diabetes and cardiometabolic disease

Abstract

Whether a pharmacological strategy can replicate the broad improvement of human cardiometabolic health associated with brown fat (BAT) remains an active area of investigation. Here, we show that adipokine Adissp activates both glucose disposal and energy expenditure within white fat, delivering pleiotropic metabolic benefits. Endogenous Adissp is essential for glucose homeostasis. Administration of recombinant Adissp (rAdissp) protein sustainably normalizes hyperglycemia in type 1 and type 2 diabetic mice by activating insulin-independent Akt signaling. Furthermore, rAdissp robustly induces a comprehensive thermogenic program, which not only reduces body weight but also independently ameliorates a wide range of cardiometabolic diseases. Thus, a single adipokine, Adissp, recapitulates the systemic metabolic benefits of BAT and essentially functions as a cold mimetic. These findings reveal an unanticipated insulin-independent glucose uptake pathway and offer mechanistic insights into the cardiometabolic protection linked to human BAT. Adissp and its analogs represent a promising class of therapeutic agents to concurrently and synergistically treat diabetes and cardiometabolic diseases.

Adipose-secreted protein Adissp has the therapeutic potential for diabetes and metabolic disease via unique mechanisms of action.

INTRODUCTION

Diabetes affects ~10% of the global population, with type 2 diabetes (T2D) constituting the majority of cases. Patients with T2D often need more than one medication to manage their glucose levels, and most of them will eventually resort to insulin therapy. Meanwhile, in type 1 diabetes (T1D), insulin remains the only established treatment. Insulin therapy is associated with body weight gain, glucose variability, hypoglycemia, and insulin resistance, which necessitates escalating insulin doses and significantly increases cardiometabolic risk. Now, only ~50% of individuals with T2D and 20% of those with T1D achieve optimal glycemic control (1, 2). Consequently, there is a need to develop additional antidiabetes drugs to provide more therapeutic options. Furthermore, given the high prevalence of obesity and other cardiometabolic conditions among individuals with insulin resistance and diabetes (3–11), a single-therapeutic agent that can concurrently target multiple cardiometabolic diseases, rather than focusing on a single entity, is highly desirable.

In this context, glucagon-like peptide 1 receptor (GLP-1R) agonists and dual GLP-1R and glucose-dependent insulinotropic polypeptide receptor (GIPR) agonists have emerged as revolutionary treatments for obesity and T2D, primally by stimulating insulin secretion and suppressing appetite (12–15). However, these therapies face several limitations, including loss of lean mass, adverse gastrointestinal events, plateaued efficacy, and reduced effectiveness on weight loss in individuals with both obesity and T2D. Increasing energy expenditure represents a promising complementary strategy to overcome these challenges, especially when combined with GLP-1–based therapies to achieve a sustainable, healthy weight loss (16, 17). Now, however, no therapies targeting energy expenditure have received regulatory approval.

BAT uniquely exhibits both thermogenic and glucose sink properties. Extensive studies in rodents have shown that BAT and/or WAT browning have both antiobesity and antidiabetes effects (18). In humans, BAT is associated with lower blood glucose and reduced risks for T2D, insulin resistance, dyslipidemia, fatty liver, cardiovascular disease, and hypertension, benefits that dissociate from body weight and fat distribution (19–21) and can be similarly achieved through cold exposure (22). Although BAT-selective adipokines are postulated to mediate these wide-ranging metabolic advantages (18–20, 23), the exact mechanisms, particularly those underlying BAT’s antidiabetic effects, remain poorly understood.

We previously identified an adipokine, Adissp (adipose-secreted signaling protein), which is abundantly and almost exclusively expressed in adipose tissue, induced by cold exposure, and highly enriched in BAT (24). Adissp is efficiently secreted by both mouse and human adipocytes (24, 25) and is present in human circulation (26, 27) (The human Protein Atlas). While Adissp is important for adipose thermogenesis (24), two key questions remain to be addressed. First, does Adissp play a direct signaling role in glucose homeostasis, independent of its role in adipose thermogenesis? Second, could pharmacological Adissp provide any therapeutic benefits for diabetes and other cardiometabolic diseases?

RESULTS

Endogenous Adissp is required for glucose homeostasis

Under ambient temperature and a regular chow diet, Adissp adipose-specific knockout (ADKO) mice showed elevated blood glucose levels both in the ad libitum state (Fig. 1A) and after a 3-hour fast (fig. S1A), whereas prolonged fasting glucose levels remained unaffected (fig. S1B), indicating impaired glucose uptake. Circulating insulin levels were unchanged (Fig. 1B). The elevated blood glucose was not linked to altered adipose thermogenesis, as thermogenic defect appeared only under conditions of cold exposure or high-fat diet feeding (24). Instead, ADKO mice showed significantly lower basal Akt activity, indicated by reduced phosphorylation at Thr³⁰⁸ and Ser⁴⁷³ in BAT, subcutaneous inguinal WAT (iWAT), and epididymal WAT (eWAT) (Fig. 1C), but not in the skeletal muscle, liver, or heart (fig. S1C). Tracer studies with [³H]2-deoxyglucose (2-DG) showed reduced basal glucose uptake in BAT and iWAT of ADKO mice, but not in eWAT (Fig. 1D, left). Insulin treatment fully restored Akt activity (fig. S1D) and glucose uptake (Fig. 1D, right), ruling out insulin resistance. Normal responses in insulin tolerance and glucose tolerance tests further suggested intact insulin signaling in ADKO mice (figs. S1, E and F). Notably, primary adipocytes isolated from ADKO mice also displayed reduced basal Akt activity (Fig. 1E), suggesting adipocyte-autonomous nature of these defects. The linear regression analysis of public microarray data (28) from subcutaneous adipose tissue samples in a cohort of 770 men showed that mRNA levels of Adissp were negatively correlated with Homeostatic Model Assessment for Insulin Resistance (HOMA-IR) (Fig. 1F). Together, these findings demonstrate that endogenous Adissp is required for systemic glucose homeostasis by maintaining basal Akt activity in adipose tissue independently of insulin action, in addition to its role in energy expenditure.

Fig. 1. Endogenous Adissp is essential for maintaining glucose homeostasis.

(A and B) Blood glucose (A) and insulin (B) levels in 4-month-old male Adissp ADKO mice (n = 11) and littermate controls (n = 12) under ad libitum feeding conditions. (C) Basal Akt activation in BAT, iWAT, and eWAT of 3-month-old male Adissp ADKO mice (n = 3) and littermate controls (n = 3) under ad libitum feeding conditions. (D) 2-DG uptake in BAT, iWAT, and eWAT of 6-month-old male Adissp ADKO mice (n = 3) and littermate controls (n = 3). Left: Ad libitum feeding conditions without insulin injection. Right: Three-hour fasting followed by insulin injection. (E) Basal Akt activation in primary adipocytes isolated from iWAT of 2-week-old Adissp ADKO mice (n = 3) and littermate controls (n = 3). (F) Linear regression analysis of HOMA-IR in relation to Adissp expression from adipose tissue of 770 men. Statistical significance was determined using a two-tailed Student’s t test. Western blot signals were quantified as the p-Akt/Akt ratio.

rAdissp activates Akt and stimulates glucose uptake independently of insulin signaling in cultured adipocytes

To further elucidate Adissp’s effects on glucose metabolism, we produced full-length, His-tagged recombinant Adissp (rAdissp) protein (Fig. 2A) from serum-free, protein-free culture medium of Expi293F cells. The treatment of differentiated brown adipocytes with rAdissp in vitro strongly induced Ucp1 expression but did not affect the expression of β3-adrenergic receptor (β3-AR) or common fat genes aP2 and Adiponectin (fig. S2, A and B). Strikingly, 30-min rAdissp treatment of adipocytes stimulated insulin-independent glucose uptake in a dose-dependent manner with an EC₅₀ (median effective concentration) of 166 nM (Fig. 2B); this rapid onset resembled the action of insulin and was distinct from the action of the hepatokine fibroblast growth factor 21 (FGF21) (29). Glucose transporter 4 (GLUT4) is responsible for glucose uptake into adipose tissue in response to insulin (30). In adipocytes transfected with Myc-GLUT4-GFP (green fluorescent protein) plasmids, the immunostaining of the exofacial Myc tag indicated that rAdissp stimulated GLUT4 trafficking to the plasma membrane (Fig. 2, C and D), albeit to a lesser extent compared to insulin.

Fig. 2. rAdissp activates a PI3K-PDK1/mTORC2-Akt signaling cascade independently of insulin to stimulate glucose uptake in cultured adipocytes.

(A) Coomassie blue staining of rAdissp protein. (B) 2-DG uptake in mature brown adipocytes treated for 30 min with either rAdissp or insulin (n = 2). (C) Myc-GLUT4-GFP–expressing mature brown adipocytes were treated with either rAdissp (300 nM) or insulin (100 nM) for 30 min followed by anti-Myc staining. Scale bars, 50 μm. (D) Quantification of surface Myc-GLUT4-GFP (n = 108 cells per group). (E) Phosphorylation of Akt and AS160 in mature brown adipocytes treated with either rAdissp or insulin for 30 min. (F) 2-DG uptake in mature brown adipocytes pretreated with inhibitors for 2 hr followed by 300 nM rAdissp for 30 min (n = 3 per group). (G) Mature human SGBS adipocytes treated with 300 nM rAdissp for 30 min (p-Akt) or 12 hours (Ucp1). (H) PDK1 activation in mature brown adipocytes treated 300 nM rAdissp for 30 min. N.S., nonspecific. (I) Akt activation in mature brown adipocytes treated with GSK2334470 for 2 hours followed by 300 nM rAdissp for 30 min. (J) Akt activation in mature brown adipocytes treated with inhibitors for 2 hours followed by either rAdissp (300 nM) or insulin (100 nM) for 30 min. (K) Bulk of tyrosine phosphorylation in mature brown adipocytes treated with either rAdissp (300 nM) or insulin (100 nM) for 30 min. (L) Western blot analysis of Akt phosphorylation, its substrates, and the downstream target AS160 in mature brown adipocytes treated with rAdissp (300 nM) for 30 min, with or without a 2-hour pretreatment with LDC1267 (n = 2 per group). DAPI, 4′,6-diamidino-2-phenylindole.

On the basis of decreased basal Akt activity in Adissp-deficient cells, we examined whether rAdissp-stimulated glucose uptake is mediated by Akt. rAdissp robustly activated Akt and induced phosphorylation of its substrate AS160, key regulator of GLUT4 trafficking (30), in a dose-dependent manner, with appreciable activity at 25 nM and substantial effect around 300 nM (Fig. 2E). This activation was not attributable to the presence of the His tag or any potential contaminants inherent in the purification process, as rAdissp lacking the His tag similarly activated Akt (fig. S2C), whereas adipose-secreted regulator of appetite (ASRA) (31), another adipokine purified through the same approach, did not (fig. S2D). The Akt inhibitor MK2206 effectively blocked rAdissp-stimulated glucose uptake (Fig. 2F). rAdissp signaling, including Ucp1 induction and Akt activation, was also conserved in human adipocytes (Fig. 2G and fig. S2E).

We investigated how Akt is activated by rAdissp. rAdissp activated 3-phosphoinositide–dependent protein kinase-1 (PDK1) (Fig. 2H and fig. S2F). Pretreatment with the PDK1 inhibitors GSK2334470 (Fig. 2I and fig. S2G) or MP7 (fig. S2H) or the phosphatidylinositol 3-kinase (PI3K) inhibitors Wortmannin (Fig. 2J and fig. S2I) or GDC-0941 (fig. S2J) abolished rAdissp-induced Akt activation, whereas insulin receptor antagonist S961 and G protein–coupled receptor (GPCR) inhibitor Melittin and Gallein had no impact (Fig. 2J). rAdissp induced bulk tyrosine phosphorylation in a pattern largely overlapping with that of insulin (Fig. 2K) but failed to induce insulin receptor substrate-1 (IRS1) phosphorylation (fig. S2K). LDC1267, a pan-receptor tyrosine kinase (RTK) inhibitor (32) blocked Akt activation and phosphorylation of Akt substrate AS160 (Fig. 2L and fig. S2L). As anticipated, LDC1267, Wortmannin, and GSK2334470 each suppressed rAdissp-stimulated glucose uptake (Fig. 2F). Our data suggest that rAdissp-stimulated glucose uptake is mediated by a PI3K-PDK1/mechanistic target of rapamycin complex 2 (mTOC2)-Akt signaling axis, likely through activation of a putative RTK. The rAdissp did not activate mitogen-activated protein kinases (p38 and ERK1/2) (fig. S2M).

Acute rAdissp administration has a long-lasting glucose-lowering effect without stimulating insulin secretion

We next evaluated whether the in vitro findings were recapitulated in vivo. To assess direct effects, euglycemic, lean mice were intraperitoneally injected with a single dose of rAdissp. rAdissp treatment for 3 hours activated PDK1 and Akt and induced phosphorylation of AS160 in BAT, iWAT, and eWAT (Fig. 3A and fig. S3A), with no effects in the skeletal muscle, heart, or liver (fig. S3B). This was accompanied by significant reduction in blood glucose levels (Fig. 3B), independent of Ucp1-mediated thermogenesis (fig. S3C) and food intake (fig. S3D). Notably, acute rAdissp treatment, either alone or during glucose loading, did not stimulate insulin secretion (Fig. 3, C and E), distinguishing it from incretins. It also had no effect on glucagon secretion under either basal or insulin-induced hypoglycemic conditions (Fig. 3, D and F).

Fig. 3. Acute rAdissp induces long-lasting glucose-lowering without stimulating insulin secretion.

(A) Phosphorylation of PDK1, Akt, and AS160 from 3-month-old male lean mice after rAdissp (70 nmol/kg, 3 hours) or insulin (0.75 U/kg, 45 min) (n = 4 per group). (B) Blood glucose in 2-month-old male lean mice treated with rAdissp (70 nmol/kg) for 3 hours (n = 4 per group). (C and D) Circulating insulin (C) and glucagon (D) from mice in (B) (n = 4 per group). (E) After a 6-hour fast, male lean mice received an oral glucose load (2 g/kg) plus intraperitoneal rAdissp (50 nmol/kg). Insulin was measured. n = 5 mice per group. (F) Ad libitum-fed male lean mice received intraperitoneally insulin (0.5 U/kg) plus rAdissp (50 nmol/kg). Glucagon was measured. n = 5 mice per group. (G) Human rAdissp (60 nmol/kg per day) was administrated to ob/ob mice, and blood glucose was measured. rAdissp, n = 14 (7 male and 7 female mice); vehicle, n = 15 (7 male and 8 female mice). (H) Circulating insulin (n = 7 per group) from male ob/ob mice. (I) Blood glucose measured 6 hours after rAdissp injection in 7-week-old BTBR ob/ob male mice. n = 5 to 6 per group. (J) Representative polyuria images. (K) Twelve-hour fasting, 2-month-old male lean mice received rAdissp (80 nmol/kg) and fasting glucose was measured. n = 4 per group. (L) Blood glucose in male DIO mice before and 3 hours after rAdissp (60 nmol/kg) injection following a 3-day pretreatment with MK2206 (50 mg/kg). Vehicle, n = 5; MK2206 + Vehicle, n = 5, Vehicle + rAdissp, n = 5; MK2206 + rAdissp, n = 8. (M and N) 2-DG uptake in indicated tissues from lean (M; rAdissp 80 nmol/kg, 3 hours) and ob/ob mice [(N), 45 nmol/kg, 3 hours]. n = 3 per group. Two-tailed Student’s t test for [(B), (H), (K), (M), and (N)] and two-way ANOVA for [(G), (I), and (L)].

To examine the acute effects of rAdissp on hyperglycemia, ob/ob mice were intraperitoneally injected with a single dose of either human rAdissp protein or mouse rAdissp protein at 60 nmol/kg (1.14 mg/kg), a dose comparable to that used for other biologics in diabetes and metabolic disease studies (33–36). Two hours postinjection, significant glucose lowering was observed in the treated group and was maintained for over 20 hours (Fig. 3G and fig. S3E), in contrast to the brief (<3-hour) effect of regular insulin. In contrast to this prolonged pharmacodynamic effect, most circulating rAdissp protein was cleared by 6 hours, as estimated by Western blot analysis (fig. S3F). The glucose lowering of rAdissp was not caused by any difference in food intake, as there was no food provided during the first 8 hours. Strikingly, following the initial injection, normal glucose levels could be maintained with one dose of rAdissp every 48 hours (Fig. 3G and fig. S3E), accompanied by a rapid improvement in hyperinsulinemia (Fig. 3H). In addition, the injection of conditioned medium containing nontagged rAdissp (50 nmol/kg) into ob/ob mice once every other day likewise lowered blood glucose (fig. S3G), confirming a rAdissp-specific effect. Next, we assessed the potency of rAdissp in the more severely diabetic BTBR ob/ob mouse model. A single dose of 58, 116, and 174 nmol/kg reduced blood glucose levels by 36, 50, and 60%, respectively (Fig. 3I), along with a marked improvement in polyuria (Fig. 3J). Across all experiments, no hypoglycemia was observed, even in lean, euglycemic mice fasted overnight, although fasting glucose levels were significantly lowered by rAdissp (Fig. 3K). Overall, the acute administration of rAdissp has a potent and long-lasting glucose-lowering activity without the risk of hypoglycemia.

Acute glucose lowering by rAdissp is Akt dependent and driven by enhanced glucose uptake into adipose tissue

We next determined whether Akt activation is essential for the glucose-lowering effect of rAdissp. High-fat diet-induced obese (DIO) mice were treated with the Akt inhibitor MK2206, followed by a single dose of rAdissp. MK2206 treatment elevated glucose levels as anticipated. While rAdissp effectively lowered glucose in control mice, it had no effect in MK2206-pretreated mice (Fig. 3L). As expected, rAdissp-induced activation of Akt and phosphorylation of AS160 were abolished by MK2206 (fig. S3H). These results demonstrate that Akt signaling mediates the acute glucose-lowering effects of rAdissp.

To ascertain that adipose tissue is responsible for the glucose-lowering effect, 2-DG was intraperitoneally administrated into lean, euglycemic mice pretreated with a single dose of rAdissp. rAdissp treatment led to a more than twofold increase in 2-DG accumulation in BAT, iWAT, and eWAT, but not in other tissues (Fig. 3M). Similar results were observed even in severely obese ob/ob mice with body weights reaching 70 g (Fig. 3N and fig. S3I). Notably, although glucose uptake per gram was lower in adipose tissue compared to skeletal muscle, the markedly expanded iWAT in these obese mice would accommodate substantial glucose uptake. These findings confirm that rAdissp primarily targets adipose tissue, particularly WAT, to mediate its glucose-lowering effect, underscoring the remarkable glucose-handling capacity of activated adipose tissue, which was also supported by other studies (35).

Chronic administration of rAdissp produces a sustained normalization of hyperglycemia and restores insulin sensitivity in DIO and ob/ob mice

The acute glucose-lowering efficacy of rAdissp led us to investigate its chronic effects. DIO mice were intraperitoneally injected with rAdissp once a day for 33 days, and ad libitum glucose levels were measured 24 hours post each injection. Consistent with Akt activation (fig. S4A), hyperglycemia was completely normalized and the effects were more pronounced over time (Fig. 4A), without changes in food intake (Fig. 4B). Similar data were obtained in an independent cohort of mice when glucose was measured 6, 12, and 24 hours postinjection (fig. S4B). Partly through improvement in overall metabolic health (see below), Adissp treatment resolved hyperinsulinemia (Fig. 4C) and restored both insulin sensitivity (Fig. 4D) and glucose tolerance (Fig. 4E and fig. S4C). To further assess glycemic control, we treated ob/ob mice with rAdissp once a day for 30 days. As observed in DIO mice, rAdissp led to a sustained normalization of hyperglycemia (Fig. 4F), without altering daily food intake (Fig. 4G). The treatment also significantly reduced hyperinsulinemia (Fig. 4H) and fully restored insulin sensitivity (Fig. 4I) and glucose tolerance (Fig. 4J and fig. S4D). Notably, although rAdissp shares downstream signaling pathways with insulin, it enhances rather than competes with insulin action. Collectively, these findings demonstrate that rAdissp administration reverses hyperglycemia and insulin resistance in two distinct T2D mouse models. Chronic rAdissp treatment also indirectly lowered glucagon levels (fig. S4, E and F), possibly via feedback mechanisms driven by improved hyperinsulinemia, consistent with the known antagonism between insulin and glucagon.

Fig. 4. Chronic rAdissp treatment sustainably normalizes hyperglycemia and restores insulin sensitivity in DIO and ob/ob mice.

(A to E) Male DIO mice received daily rAdissp (68 nmol/kg) for 33 days. Blood glucose was measured 24 hours after each injection under ad libitum conditions (day 1 included a 2-hour fast) (n = 19 to 20 per group) (A). Daily food intake (B) and circulating insulin (C) were assessed. Insulin tolerance tests (ITTs; insulin, 0.75 U/kg) (D) and glucose tolerance tests (GTTs; glucose, 0.1 g per mouse) (E) were performed. (F to J) Twelve-week-old ob/ob mice were treated daily with rAdissp (60 nmol/kg) for 30 days. Blood glucose was measured 24 hours after each injection. Vehicle group, n = 14 (6 male and 8 female mice); treated group, n = 16 (8 male and 8 female mice) (F). Daily food intake was recorded (G). Circulating insulin was measured in male mice (n = 6 to 8 per group) and female mice (n = 8 per group) (H). ITT (insulin, 0.75 U/kg) (I) and GTT (glucose, 0.06 g per mouse) (J) were conducted (vehicle group, n = 14; treated group, n = 16). (K to N) Following high-dose STZ, male mice received a single rAdissp dose, and glucose was measured over time after an 8- or 6-hour fast (n = 9 to 10 per group) (K), with representative polyuria images (L). After low-dose STZ, male mice were treated daily with rAdissp (72 nmol/kg) for 13 days; glucose was measured 24 hours postinjection (day 1 included a 2-hour fast) (n = 8 per group) (M), with representative polyuria images (N). Two-tailed Student’s t test for [(C) and (H)] and a two-way ANOVA for [(A), (D), (E), (F), (I), (J), (K), and (M)].

Adissp markedly lowers blood glucose in STZ-induced T1D mice

Our data suggest that while continuous rAdissp treatment improves insulin sensitivity, its primary glucose-lowering mechanism operates independently of insulin. To validate this and evaluate its therapeutic potential for T1D, we administered a single dose of rAdissp to mice with severe T1D induced by a high dose of streptozotocin (STZ). rAdissp rapidly reduced glucose levels by 40%, with effects lasting over 52 hours (Fig. 4K), and was accompanied by a reduction in polyuria (Fig. 4L). Next, chronic studies were conducted in mice with milder T1D induced by low doses of STZ. Daily rAdissp injections for 13 days consistently lowered glucose levels by 40 to 50% (Fig. 4M) and significantly improved polyuria (Fig. 4N). These results strongly support the insulin-independent mechanism underlying rAdissp’s glucose-lowering action in vivo, underscoring its potential as a therapeutic option for T1D.

Administration of rAdissp strongly induces WAT browning and a comprehensive thermogenic program

We set to investigate the impacts of pharmacological Adissp on adipose thermogenesis, initially assessing its potency in promoting WAT browning in lean mice. After only two doses of rAdissp, the iWAT depots acquired a visibly brownish appearance comparable to that seen after 6 hours of cold exposure at 4°C (Fig. 5A). This phenotypic shift was supported by a marked upregulation of Ucp1 expression (Fig. 5B). Similar results were obtained at thermoneutral conditions (fig. S5, A and B). Together, our data suggest that rAdissp is a powerful activator of both glucose uptake and WAT browning, mirroring the action of cold.

Fig. 5. rAdissp treatment activates a comprehensive thermogenic program.

(A and B) iWAT browning and Ucp1 expression in lean mice treated with daily rAdissp (70 nmol/kg) for up to 3 days or exposed to cold (4°C) for 6 hours. Representative images from three mice per group are shown. (C) Induction of Ucp1 and key thermogenic transcriptional activators and coactivators in iWAT by daily rAdissp treatment for 33 days. (D) Genes significantly up-regulated (left) and down-regulated (right) in iWAT following rAdissp treatment were categorized on the basis of their differential expression between BAT and eWAT. (E) Distribution of BAT-selective (left) and WAT-selective (right) genes according to their regulation in iWAT following rAdissp treatment. (F) Concurrent induction of genes involved in fatty acid oxidation and de novo lipogenesis in iWAT by rAdissp treatment. (G) Induction of genes involved in the TCA cycle in iWAT following rAdissp treatment. (H) Induction of genes encoding mitochondrial oxidative phosphorylation complexes in iWAT following rAdissp treatment. (I) Suppression of genes encoding chemokines, cytokines, and their receptors in iWAT following rAdissp treatment. (J) Induction of genes encoding glycolytic enzymes in iWAT by rAdissp treatment. (K) rAdissp activates coordinated metabolic pathways. In (C) and (F) to (J), gene expression values from RNA-seq are presented as log₂ (fold change of rAdissp versus vehicle). n = 3 mice per group. CoA, coenzyme A.

In DIO mice chronically treated with rAdissp, browning occurred in both iWAT and eWAT depots, with notably smaller adipocytes (fig. S5C). To gain a molecular understanding of the thermogenic effects of rAdissp, we conducted bulk RNA sequencing (RNA-seq) on iWAT samples from these DIO mice. A total of 2053 genes were up-regulated and 2269 were down-regulated [false discovery rate (FDR) < 0.05; log₂ cutoff = 0.585] (tables S1 and S2). Notably, Ucp1 emerged as the second most induced gene, showing a 44-fold increase. Key transcriptional drivers of WAT browning—including Pgc-1α (Ppargc1a); Pgc-1β (Ppargc1b); Prdm16; Hlx; Pparα; and estrogen-related receptor (ERR) α, β, and γ (Esrra, b and g)—were significantly up-regulated (Fig. 5C; shown as log₂ [rAdissp/vehicle]).

To examine whether rAdissp induces a global shift toward a BAT-like phenotype, we compared these differentially expressed genes (DEGs) with our previously published RNA-seq datasets from BAT and eWAT in wild-type mice (37). Strikingly, 1069 of the 2053 rAdissp–up-regulated genes had significantly higher expression in BAT than in eWAT, while 769 of the 2269 down-regulated genes showed significantly lower expression in BAT. Only 67 up-regulated and 31 down-regulated genes showed the opposite trend (Fig. 5D). Next, using a fivefold differential gene expression cutoff, we identified 323 BAT-selective genes and 465 WAT-selective genes (37). Of these, 252 BAT-selective genes were induced by Adissp, with none suppressed. Conversely, 160 WAT-selective genes were down-regulated by Adissp, with only 10 showing increased expression (Fig. 5E). Notably, most of the fatty acid oxidation (FAO) genes (Fig. 5F), all tricarboxylic acid (TCA) cycle genes (Fig. 5G), and 56 of the 67 core oxidative phosphorylation genes (Fig. 5H) were up-regulated by rAdissp. Overall, rAdissp treatment broadly reprograms WAT toward a BAT-like molecular signature. In addition, chemokine/cytokine genes and their receptors involved in pro-inflammatory response were down-regulated (Fig. 5I), implying that WAT browning is associated with improved adipose tissue health. Notably, rAdissp also induced thermogenic gene expression in BAT, but to a much lesser extent (fig. S5D). Given the large mass of WAT depots, the cardiometabolic benefits described below are likely driven primarily by WAT browning.

Cold stimulates the formation of glycolytic adipocytes to enhance glucose utilization (38). We found that every gene encoding a glycolytic step was significantly induced (Fig. 5J), suggesting that rAdissp-stimulated glucose uptake is closely connected to glycolysis, which feeds into the TCA cycle and fuels thermogenesis. This observation aligns with the enhanced glucose-lowering effects seen during chronic rAdissp treatment. Cold exposure also triggers de novo lipogenesis (DNL) in both BAT and WAT (39–42), in which Akt signaling has been implicated (41). The simultaneous induction of genes for lipid catabolism and synthesis was thought to constitute a futile cycle to maximalize energy expenditure and heat production (43). Likely due to its activation of Akt (fig. S4A), rAdissp treatment strongly up-regulated key lipogenic genes (Fig. 5F), suggesting the presence of an FAO-DNL futile cycle. While we cannot fully exclude a change in energy substate utilization, our data suggest that rAdissp mimics the action of cold by simultaneously activating multiple coordinated metabolic pathways that use both carbohydrates and fatty acids to amplify WAT thermogenesis (Fig. 5K). To our knowledge, Adissp represents the first described peptide cold mimetic.

Administration of rAdissp produces pleiotropic improvements in cardiometabolic diseases

Consistent with activation of a broad thermogenic program, DIO mice chronically treated with rAdissp appeared noticeably warmer upon handling. Infrared camera imaging confirmed an increase in body surface temperature (Fig. 6A), especially in the tail, a key site for heat dissipation and body temperature regulation in rodents (44–46), while core body temperature remained normal (Fig. 6B). Despite the initial morbid obesity, which is difficult to treat, and continued high-fat feeding, rAdissp treatment significantly blocked body weight gain, resulting in a 10% difference by the end of the 5-week study (Fig. 6C). EcoMRI revealed a reduction in fat mass without loss of lean mass (Fig. 6D).

Fig. 6. Administration of rAdissp produces pleiotropic improvements in cardiometabolic diseases.

(A) Representative images of body surface temperature in male DIO mice after five weeks of daily rAdissp (68 nmol/kg) treatment. (B) Core body temperature in the mice from (A) (n = 19 to 20 per group). (C and D) Body weight gain and EchoMRI in the mice from (A) (n = 19 to 20 per group). (E and F) Representative liver images (E) and hematoxylin and eosin (H&E) staining (F) from the mice in (A). Scale bar, 200 μm. (G and H) Liver triglyceride content [(G), n = 10 per group] and weight [(H), n = 10 to 19] from mice in (A). (I) Serum levels of AST and ALT from mice in (A) (n = 19 to 20 per group). (J to M) Heart triglyceride levels [(J), n = 10 per group], weight [(K), n = 19 to 20], and representative images [(L) and (M)] from the mice in (A). (N and O) Serum triglyceride (n = 10 per group) and uric acid levels (n = 19 to 20 per group) from the mice in (A). (P to Z) ob/ob mice received daily rAdissp (60 nmol/kg) for 1 month. Representative images of body surface temperature (P). Body weight gain [(Q), vehicle: n = 14, 6 male and 8 female mice; rAdissp: n = 16, 8 male and 8 female mice]. In male ob/ob mice, liver images (R), H&E staining [(S), scale bars = 200 μm], triglycerides [(T), n = 6 to 8 per group], weight [(U), n = 6 to 8 per group], and serum AST/ALT [(V), n = 6 to 8 per group] were determined. Heart triglycerides [(W), n = 6 to 8 per group], weight [(X), n = 6 to 8 per group], H&E staining (Y), and serum triglycerides [(Z), n = 6 to 8 per group] were measured. Two-tailed Student’s t test was performed, except for (C), where a two-way ANOVA was performed. Scale bar, 200 μm.

Strikingly, while livers from vehicle-treated mice appeared pale, indicative of severe lipid accumulation, livers from the treatment group were markedly darker (Fig. 6E), and steatosis was effectively resolved, as shown by hepatic histology (Fig. 6F), triglyceride content (Fig. 6G), and liver weight (Fig. 6H). Serum levels of aspartate aminotransferase (AST) and alanine aminotransferase (ALT) returned to near-normal (Fig. 6I), indicating an almost complete reversal of steatosis-induced liver damage. Likewise, cardiac steatosis (Fig. 6J), hypertrophy (Fig. 6, K and L), inflammation, and injury (Fig. 6M) were substantially ameliorated. The depletion of liver and cardiac lipid stores did not result in elevated circulating lipids; instead, hypertriglyceridemia was corrected (Fig. 6N). These metabolic benefits also led to a remarkable improvement of hyperuricemia (Fig. 6O). Whereas recombinant FGF21 protein, a hepatokine with similar metabolic advantages (29, 47–49), and antidiabetic drug TZDs, have been linked to bone loss (50, 51), rAdissp treatment did not adversely affect bone density (fig. S6A).

Comparable improvements in metabolic health were observed in ob/ob mice treated for 1 month. Increased thermogenesis (Fig. 6P) by rAdissp treatment significantly slowed down but did not completely prevent, the body weight gain (Fig. 6Q), most likely due to the extreme hyperphagia. Treated ob/ob mice also showed marked improvements in liver steatosis (Fig. 6, R to U, and fig. S6, B and C) and injury (Fig. 6V and fig. S6D) and cardiac steatosis (Fig. 6, W and X, and fig. S6E) and injury (Fig. 6Y), along with lower circulating triglyceride levels (Fig. 6Z and fig. S6F) and reduced hyperuricemia (fig. S6G). Note that these beneficial effects along with restoration of insulin sensitivity occurred, although the mice continued to gain weight, a dissociation that may have important clinical implications. In aggregate, our findings demonstrate that in both DIO and ob/ob mice, chronic rAdissp treatment delivers pleotropic cardiometabolic benefits without the need to suppress appetite or reduce body weight. Notably, rAdissp did not affect mouse locomotor activity (fig. S6H), further supporting a model in which rAdissp-induced energy expenditure in WAT, although not yet directly measured, is the primary driver of the observed weight loss and cardiometabolic improvements.

DISCUSSION

Identifying additional therapeutic agents with unique mechanisms of action is crucial for addressing the many dimensions of cardiometabolic disease. Our studies reveal that endogenous Adissp is a physiological activator of both glucose disposal and energy expenditure in adipose tissue, effectively harnessing the specialized properties of BAT. By integrating these two distinct yet synergistic functions, pharmacological Adissp delivers robust glycemic control and metabolic benefits in various mouse models of metabolic disease, without relying on insulin signaling or appetite suppression. To our knowledge, Adissp is the first adipokine to demonstrate this potent dual metabolic action and one of the few biologics capable of producing broad metabolic improvements in preclinical rodent studies. WAT, particularly the expanded WAT associated with obesity, is the primary pharmacological site of Adissp action. Coupled with conserved Adissp signaling in human adipocytes, these findings support Adissp’s potential clinical utility. Mechanistically, Adissp operates distinctly from GLP-1R agonists and GLP-1R/GIPR agonists, positioning it as a promising agent of a class of metabolic therapeutics.

Our studies shed light on the body weight– and thermogenesis-independent glucose-handling function of BAT in mice (18) and the lower prevalence of insulin resistance and type 2 diabetes among individuals with higher BAT mass (19–21). In lean Adissp ADKO mice, we observed reduced basal Akt activity and glucose uptake in adipose tissue, leading to elevated blood glucose despite normal body weight and adipose thermogenesis. Consistent with this physiological role, the acute administration of rAdissp induced a striking reduction in blood glucose that was entirely Akt dependent and thermogenesis independent. The more severe the hyperglycemia, the greater the glucose-lowering response is; a single Adissp dose could reduce blood glucose by up to 60%. Chronic rAdissp treatment further enhanced glycemic control, producing sustained normalization of hyperglycemia and alleviating hyperinsulinemia and restoring insulin sensitivity and glucose tolerance. Notably, dosing every other day was sufficient to maintain normal glycemia, an extraordinarily long-lasting effect compared to native insulin or GLP-1. The future optimization of Adissp via protein engineering could likely extend dosing intervals while retaining efficacy, supporting its clinical development. In this regard, it is worth mentioning that, although recombinant FGF21 similarly normalizes hyperglycemia in mice apparently through a different mechanism (29, 52), its efficacy in human trials has been more limited (53). In contrast to Adissp (and insulin), FGF21 deficiency in lean mice does not elevate blood glucose (54), nor does FGF21 injection lower glucose in euglycemic animals (29, 55). These inherent differences, combined with Adissp’s insulin-independent action and insulin-sensitizing effects, underscore its unique clinical promise for T1D, T2D, and diabetes caused by rare insulin receptor disorders (56–59).

While the glucose-lowering effects of this nonoptimized rAdissp are robust, the body weight reduction in morbidly obese mice is, understandably, less pronounced compared to that achieved with appetite-suppressing GLP-1–based therapies but is in line with FGF21 treatment at comparable daily doses (49), which also acts independently of appetite. rAdissp-induced WAT browning, while not confined to specific depots, occurred more gradually in obese mice compared to lean mice, particularly during the early stages of treatment, which likely limited immediate body weight loss. We anticipate that extended treatment beyond our current scheme would accelerate WAT browning and thereby expedite weigh loss. Alternatively, initiating treatment at a stage of mild obesity—when WAT is more responsive to browning—may yield more pronounced effects. Nonetheless, our data suggest that the energy-expending activity of Adissp may offer a potential viable addition to body weight management. Future efforts could explore combining Adissp with GLP-1–based therapies to achieve more potent, durable, and healthy weight loss with lean mass sparing, particularly in sedentary individuals with obesity. Moreover, Adissp may serve as a standalone option for individuals with loss-of-function variants of GLP-1R (60) or as a means to maintain weight loss following GLP-1–based therapies. Perhaps more importantly, regardless of its effect on body weight reduction, rAdissp robustly restored insulin sensitivity and resolved hepatic and cardiac steatosis, normalized hypertriglyceridemia, and improved hyperuricemia. Although we cannot fully exclude potential direct actions in other tissues, the data suggest preferential shuttling of lipids to adipose tissue for thermogenesis. In rAdissp-treated ob/ob mice, these benefits occur even as body weight increased, consistent with the body weight–independent metabolic advantages conferred by BAT in humans (19–21). This dissociation may point to a shortcut for treating obesity-associated comorbidities without necessarily requiring weight loss.

Cold exposure simultaneously promotes glucose uptake, glycolysis, mitochondrial oxidative metabolism, Ucp1-mediated uncoupling, and a futile cycle of fatty acid oxidation and synthesis, thereby maximizing fuel utilization and heat production in adipose tissue. Although the underlying mechanisms may differ, pharmacological Adissp replicates the functional outcome of cold stimulation by activating and synchronizing these seemingly opposing metabolic pathways. It is remarkable that a single adipokine, Adissp, can harness the specialized features of BAT and serve as a bona fide cold mimetic.

Several limitations of this work should be noted. First, while Adissp signaling is conserved in human adipocytes, it remains uncertain whether Adissp will confer similar cardiometabolic benefits in humans. Second, the cognate receptor(s) of Adissp have yet to be identified. Third, it is unclear whether the dual actions of Adissp on glucose uptake and energy expenditure are mediated through a single integrated molecular pathway or through two distinct, unconnected pathways. Nevertheless, our findings reveal an insulin-independent mechanism for glucose uptake and broaden our understanding of how brown and beige adipose tissues contribute to systemic metabolic health. We propose that Adissp and its analogs may represent a promising class of therapeutics for concurrently and synergistically treating insulin resistance, diabetes, obesity, and associated comorbidities.

MATERIALS AND METHODS

Animals

All mice were housed under a 12-hour light/12-hour dark cycle at 23°C with ad libitum access to a standard chow containing 4% (w/w) fat and water, unless otherwise specified. C57BL/6J wild-type mice (stock no. 000664), B6.Cg-Lep^ob/J mice (stock no. 000632), and BTBR.Cg-Lep^ob/WiscJ (stock no. 004824) were obtained from the Jackson Laboratory. ADKO mice were previously generated (24). Briefly, mice with exon 3 of Adissp alleles flanked by loxP sites were crossed with Adiponectin-Cre mice (61) to generate Adissp ADKO mice. DIO mice were established by feeding male C57BL/6J mice a high-fat diet containing 36% (w/w) fat (Bioserv, catalog no. S3282) for at least 16 weeks. STZ-induced diabetic mice were generated as described (62). Briefly, male wild-type mice received either a single intraperitoneal injection of STZ (200 mg/kg) or daily injections of STZ (40 mg/kg) for five consecutive days. Blood glucose levels were measured 1 week postinjection to confirm hyperglycemia. Age-matched littermates were used for all experiments. Animal studies were conducted in accordance with protocol no. 202000027 approved by the Institutional Animal Care and Use Committee at the University of Massachusetts Chan Medical School.

Expression and purification of rAdissp protein

Expi293F cells (Thermo Fisher Scientific, catalog no. A14527), derived from the human embryonic kidney (HEK) 293 cell line, were cultured in suspension in serum-free, protein-free Expi293 Expression Medium (Thermo Fisher Scientific, catalog no. A1435101). Mouse and human Adissp cDNAs were cloned into the pHL-sec vector (Addgene, catalog no. 99845) with a C-terminal 6 × His tag. Plasmids were purified using an endotoxin-free plasmid extraction kit (QIAGEN, catalog no. 12362) and transiently transfected into Expi293F cells cultured in the serum-free, protein-free medium. Five days posttransfection, the culture medium (1000 ml) was harvested and incubated with 5-ml bed volume of Ni–nitrilotriacetic acid (NTA) affinity beads (QIAGEN, catalog no. 30230) for 2 hours at 4°C and then passed through a 50-ml column. The column was washed twice with 10 ml of Ni-NTA binding/wash buffer (1 mM tris (2-carboxyethyl) phosphine hydrochloride (TCEP), 20 mM tris (pH 7.5), 40 mM imidazole, and 1000 mM NaCl), and rAdissp protein was eluted using 20 ml of elution buffer [8% glycerol (v/v), 1 mM TCEP, 20 mM tris (pH 7.5), 500 mM imidazole, and 500 mM NaCl]. Eluted protein was concentrated and buffer-exchanged into Hepes buffer [20 mM Hepes (pH 7.5) and 150 mM NaCl] using an Amicon Ultra-15 centrifugal filter unit with a 3-kDa molecular weight cutoff (Millipore, catalog no. UFC900324). The buffer exchange achieved a 20,000-fold dilution of the elution buffer, resulting in a final residual imidazole concentration of 25 μM. The purified protein was sterile-filtered through a 0.22-μm Spin-X centrifuge tube filter (COSTAR). Protein concentration was determined using a bicinchoninic acid (BCA) assay, and endotoxin levels were measured using a commercial assay (Thermo Fisher Scientific, catalog no. 88282), confirmed to be ≤0.01 EU/μg. Aliquots were stored at −80°C and subjected to no more than three freeze-thaw cycles.

In vivo rAdissp treatment

Mice were divided into two groups matched with body weight and glucose levels. rAdissp was intraperitoneally administered either daily or every other day, with doses and treatment durations specified for each experiment. The control group received Hepes buffer. Blood glucose levels were measured under ad libitum feeding conditions unless otherwise indicated. Food intake was measured daily in group-housed mice: four to five DIO mice/cage (five cages per group) and two ob/ob mice/cage (seven to eight cages per group). Body weight was measured weekly. Fat mass and lean mass were assessed using a Body Composition Analyzer (EchoMRI, Echo Medical Systems). Body surface temperature was monitored using an infrared camera, and core body temperature was measured with a MicroTherma 2 rectal probe (ThermoWorks) according to the manufacturer’s instructions.

To assess whether the glucose-lowering effect of rAdissp is Akt-dependent, the Akt inhibitor MK2206 (MedChemExpress, catalog no. HY-10358) was administered to DIO mice via intraperitoneal injection at 50 mg/kg daily for three consecutive days, followed by a single dose of rAdissp (60 nmol/kg). Blood glucose was measured 3 hours after rAdissp administration.

In vivo administration of conditioned medium containing rAdissp protein

HEK293 cells were transfected with pHL-sec plasmids expressing nontagged Adissp or empty vector plasmids. Conditioned medium was collected and concentrated, and the rAdissp concentration was quantified by Western blot using purified rAdissp as a standard. The rAdissp-containing conditioned medium was intraperitoneally injected into 3-month-old male ob/ob mice at days 1, 3, and 5 at a dose of 50 nmol/kg. Control mice received an equivalent volume of vector-conditioned medium. Blood glucose levels were measured at indicated time points.

In vivo glucose uptake assay

In vivo glucose uptake was measured as previously described (63). To assess basal glucose uptake, 6-month-old male Adissp ADKO mice and their littermate controls were injected with 2-DG (Revvity, catalog no. NET549A250UC) at 100 μCi/kg body weight. One hour postinjection, mice were euthanized, and tissues were collected, weighed, and snap-frozen in liquid nitrogen. To evaluate insulin-stimulated glucose uptake, 6-month-old male mice were fasted for 3 hours and then injected intraperitoneally with insulin (0.75 U/kg). Thirty min later, 2-DG at 100 μCi/kg was administered, and mice were euthanized 1 hour after the radioactive glucose injection. Tissues were harvested as described above.

To determine rAdissp-stimulated glucose uptake, 3-month-old male wild-type lean mice and 4-month-old male ob/ob mice were treated with rAdissp with indicated doses for 3 hours, followed by an injection of 2-DG at 100 μCi/kg. One hour after the radioactive glucose injection, mice were euthanized and tissues were collected, weighed, and snap-frozen in liquid nitrogen.

Tissues were digested in 500 μl of 1 M NaOH at 60°C for 60 min and then neutralized with 500 μl of 1 M HCl. A 300-μl aliquot of the neutralized lysate was mixed with 1 ml of 6% perchloric acid, vortexed, and centrifuged at 13,000g for 5 min. One milliliter of the resulting supernatant was combined with 4 ml of scintillation cocktail, and total radioactivity was measured in counts per minute (c.p.m.) using liquid scintillation counting.

Glucose and insulin tolerance tests

DIO mice fasted for 3 hours were intraperitoneally injected glucose at 0.1 g per mouse and insulin at 0.75 U/kg body weight for glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs), respectively. ob/ob mice fasted for 5 hours were intraperitoneally injected glucose at 0.06 g per mouse and insulin at 0.75 U/kg body weight for GTT and ITT, respectively. Blood glucose levels were measured at 0, 30, 60, 90, and 120 min postinjection. GTT and ITT were performed at treatment days 22 and 29, respectively.

Measurement of circulating insulin, glucagon, AST, ALT, and uric acid levels

Commercial ELISA kits were used to measure circulating levels of insulin (Crystal Chem, catalog no. 90080) and glucagon (Crystal Chem, catalog no. 81518), with concentrations calculated using a four-parameter logistic curve fit. In Fig. 3F, insulin levels were measured in a randomly selected subset of serum samples. To measure insulin levels in the context of glucose loading, mice were prefasted for 6 hours. Mice were gavaged with glucose (2 g/kg) along with intraperitoneal injection of rAdissp (50 nmol/kg) at 0 min, and blood was drawn at 0 (before glucose and rAdissp administration), 15, 30, 60, and 120 min. To measure glucagon secretion at insulin-induced hypoglycemia condition, mice under ad libitum feeding were injected with insulin (0.5 U/kg) and rAdissp (50 nmol/kg) at 0 min, and blood was drawn at 0 (prior to insulin and rAdissp injection), 30, 60, 90, and 120 min. Commercial kits were used to measure serum levels of ALT (Abcam, catalog no. ab105134), AST (Abcam, catalog no. ab105135) and uric acid (Cayman, catalog no. 700320).

Measurement of triglyceride levels

A portion of liver (100 mg) and heart (100 mg) was harvested, washed with cold phosphate-buffered saline (PBS), and homogenized in 1 ml of 5% NP-40/ddH₂O. The homogenate was heated at 80° to 100°C until the solution became cloudy and then cooled to room temperature; this cycle was repeated to ensure complete triglyceride solubilization. Samples were centrifuged to remove insoluble material, and the supernatant was diluted 10-fold with ddH₂O before analysis. Commercial kit (Abcam, catalog no. ab65336) was used to measure levels of triglycerides in the liver, heart, and serum.

Histology

iWAT, eWAT, liver and heart tissues were fixed in 10% formalin and embedded in paraffin. Hematoxylin and eosin (H&E) staining was performed using standard protocols by the Morphology Core Facility at the University of Massachusetts Chan Medical School.

Open field test

Mice received a single intraperitoneal injection of rAdissp (50 nmol/kg). Two hours later, individual mice were placed in the middle of the open field apparatus and illuminated by light-emitting diodes placed 60 cm above the apparatus floor. The subject mice were allowed to move free and uninterrupted throughout the open field box for a single 10-min period, during which time the movement of mice was recorded with an overhead camera. EthoVision XT software was used to calculate the distance moved and the mean velocity.

Adipocyte culture and differentiation

Differentiated adipocytes were used in in vitro experiments. The immortalized brown preadipocyte cell line was previously generated (64). On day −2 of differentiation, brown preadipocytes at 70% confluence were cultured in Dulbecco’s modified Eagle’s medium (DMEM; catalog no. 11965-092, Gibco) supplemented with 10% fetal bovine serum (FBS; catalog no. S11550, Atlanta Biologicals), 20 nM insulin (Sigma-Aldrich, catalog no. I6634), 1 nM 3,3′,5-triiodo-l-thyronine (Sigma-Aldrich, catalog no. T0281), penicillin (50 U/ml), and streptomycin (50 mg/ml; differentiation medium). To induce adipocyte differentiation on day 0, the cells were cultured in differentiation medium supplemented with 0.125 mM indomethacin (Alfa Aesar, catalog no. A19910-06), 0.5 μM dexamethasone (Sigma-Aldrich, catalog no. d4902), and 0.5 mM isobutylmethylxanthine (Sigma-Aldrich, catalog no. I7018) for 48 hours and were then returned to the differentiation medium for 4 days.

The primary iWAT preadipocyte culture and differentiation were described previously (37). Preadipocytes were isolated from 2-week-old mice and cultured until confluence. Differentiation was initiated by culturing the confluent cells in DMEM/F12 medium (Gibco, catalog no. 11320-033) containing 10% FBS, 850 nM insulin, 1 nM triiodothyronine, 0.5 mM isobutylmethylxanthine, 0.5 μM dexamethasone, and 0.125 mM indomethacin. After 2 days, the cells were maintained in DMEM/F12 medium supplemented with 10% FBS, 850 nM insulin, and 1 nM triiodothyronine, with the medium being changed every 2 days. On day 6, primary iWAT adipocytes were fully differentiated, and cells with at least 95% differentiation efficiency were used in experiments. Human preadipocyte SGBS cells were previously generated (65) and were differentiated following an established protocol (66).

In vitro glucose uptake assay

Differentiated brown adipocytes were serum-starved overnight and then treated with rAdissp at concentrations of 0, 20, 40, 80, 160, 320, 640, or 1000 nM, or with 100 nM insulin, for 30 min. 2-DG was added to the adipocyte culture and incubated for an additional 30 min. Glucose uptake wea measured using a Glucose Uptake Assay Kit (Abcam, catalog no. 136955) according to the manufacturer’s protocol. In inhibitor studies, serum-starved adipocytes were pretreated for 2 hours with RTK inhibitor LDC1267 (1 μM), PI3K inhibitor Wortmannin (1 μM), PDK1 inhibitor GSK2334470 (5 μM), or Akt inhibitor MK2206 (2 μM), followed by addition of rAdissp (300 nM) and 2-DG as described above.

GLUT4 translocation

Differentiated brown adipocytes transfected with Myc-GLUT4-GFP plasmids (67) were serum-starved overnight. Cells were then treated with either rAdissp (300 nM) or insulin (100 nM) for 30 min. Following treatment, cells were fixed with 3.7% formalin at 37°C for 20 min at room temperature without permeabilization. After fixation, cells were blocked in PBS containing 5% normal goat serum for 60 min at room temperature. Subsequently, cells were incubated overnight at 4°C with a Myc-tag antibody (Cell Signaling Technology, catalog no. 2278S) diluted 1:100 in blocking buffer. The following day, cells were gently washed twice with PBS and incubated with Alexa Fluor 488–conjugated goat anti-rabbit IgG (Thermo Fisher Scientific, catalog no. A-11008) for 2 hours at room temperature. Nuclei were then stained with DAPI (Sigma-Aldrich, catalog no. D9542) for 30 min at room temperature. Fluorescent images were acquired using a Nikon Eclipse Ti2 inverted confocal microscope (Nikon Instruments/Nikon Corp) under identical settings for all samples.

The ratio of surface-to-total Myc-GLUT4-GFP signal for individual cells was determined using ImageJ. An outline was drawn around each individual cell to define the region of interest (ROI). The total integrated GFP intensity within this ROI was measured. Background fluorescence was measured from a cell-free area within the same field of view and subtracted from the total cell intensity measurement. The plasma membrane was identified based on the Myc-tag immunostaining. The measurement of cell surfaces tool in ImageJ was used to specifically quantify Myc intensity at the plasma membrane. The ratio of surface Myc fluorescence to total background-corrected GFP fluorescence was calculated for each cell. This quantification was performed for at least 100 individual cells per experimental condition.

Western blotting analysis

For in vitro studies, mature brown adipocytes were serum-starved overnight and then treated with rAdissp, rAdissp-containing conditioned medium, recombinant (r)ASRA, or insulin at the indicated concentrations for 30 min. In certain experiments, mature adipocytes were pretreated for 2 hours with the following inhibitors: PDK1 inhibitors GSK2334470 (5 μM) or MP7 (5 μM), insulin receptor antagonist S961 (5 μM), PI3K inhibitors Wortmannin (1 μM) or GDC-0941 (1 μM), GPCR inhibitor Melittin (Gαs subunit inhibitor; 1 μM), Gallein (Gβγ inhibitor; 5 μM), and RTK inhibitor LDC1267 (1 μM). For in vivo studies, rAdissp was intraperitoneally administered with doses and treatment durations specified for each experiment.

Equal amounts of cultured mature adipocytes or adipose tissue were homogenized in lysis buffer [100 mM NaCl, 50 mM tris (pH 7.5), 0.5% Triton X-100, and 5% (w/v) glycerol] supplemented with protease inhibitor cocktail (Complete Mini, EDTA-free; Roche, catalog no. 11836170001). Lysates were centrifuged at 13,000 rpm for 10 min at 4°C. Equal amounts of protein were resolved by SDS–polyacrylamide gel electrophoresis, transferred onto polyvinylidene difluoride membranes, and immunoblotted with a primary antibody against Ucp1 (Sigma-Aldrich, catalog no. U6382), p-PDK1-Ser²⁴¹ (Cell Signaling Technology, catalog no. 3438S), total PDK1 (Cell Signaling, catalog no. 5662S), p-Akt-Thr³⁰⁸ (Cell Signaling, catalog no. 13038S), p-Akt-Ser⁴⁷³ (Cell Signaling, catalog no. 4060S), total Akt (Cell Signaling, catalog no. 4691S), p-AS160-Thr⁶⁴² (Cell Signaling Technology, catalog no. 8881S), total AS160 (Cell Signaling Technology, catalog no. 2670S), p-IRS1-Tyr⁸⁹⁵ (Cell Signaling Technology, catalog no. 3070S), total IRS1 (Cell Signaling Technology, catalog no. 3282S), p-p38-Thr¹⁸⁰/Tyr¹⁸² (Cell Signaling Technology, catalog no. 4511S), total p38 (Cell Signaling Technology, catalog no. 8690S), p-ERK-Thr²⁰²/Tyr²⁰⁴ (Cell Signaling Technology, catalog no. 4370S), total ERK (Cell Signaling Technology, catalog no. 9102S), Adiponectin (Cell Signaling Technology, catalog no. 2789S), phospho-tyrosine (P-Tyr-1000) MultiMab (Cell Signaling Technology, catalog no. 8954S), or β-actin (Proteintech, catalog no. 66009-1-Ig).

Real-time quantitative PCR

Total RNA was extracted from mature brown adipocytes or tissues using TRIzol reagent (Invitrogen, catalog no. 15596-018) following the manufacturer’s instructions. An equal amount of RNA was used for reverse transcription. Quantitative real-time PCR (qRT-PCR) was performed using SYBR green fluorescent dye (Bio-Rad, catalog no. 1725272) on an ABI7300 PCR instrument. The ribosomal 36B4 (U36) gene was used as an internal control. The relative mRNA expression levels were calculated using the ΔΔ^-Ct method. Primer sequences will be provided upon request.

RNA-seq and data analysis

The DIO mice were treated with rAdissp (68 nmol/kg) or vehicle for 33 days. Following treatment, mice were euthanized, and iWAT depots were harvested. Total RNA (n = 3 per group) was extracted using TRIzol reagent (Invitrogen, catalog no. 15596018). RNA-seq libraries construction and sequencing were performed by the Yale Center for Genome Analysis. One microgram of total RNA per sample was used to generate RNA-seq libraries in biological triplicates using the KAPA RNA HyperPrep Kit with RiboErase (HMR) (Roche, catalog no. KK8561). The amplified libraries were sequenced on an Illumina NovaSeq X platform.

The RNA-seq analysis was performed with OneStopRNAseq (68). Paired-end reads were aligned to the mouse genome (mm10). Aligned exon fragments with mapping quality higher than 20 were counted toward gene expression. Differential expression (DE) analysis was performed with DESeq2 (69), and log₂ fold change shrinkage for each comparison was created. Significant DEGs were filtered with the criteria FDR < 0.05. These DEGs were coanalyzed with published BAT and eWAT RNA-seq dataset (GEO accession number GSE56367) of wild-type mice (37). BAT-selective genes and WAT-selective genes were identified with a fivefold DE cutoff and fragments per kilobase of transcript per million (FPKM) value > 5 in at least one tissue.

The raw RNA-seq data can be accessed in GEO under accession number GSE302416 (www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE302416).

Statistical analysis

The sample size for this study was determined on the basis of prior experience and existing literature in the field. The investigators conducting the mouse experiments were not blinded to the genotypes. The number of biological samples (n) was provided for each figure panel. Unless otherwise specified, the data were presented as means ± SEM, and individual data points were plotted. Statistical analyses were performed using GraphPad Prism 8.0 software. Analysis between two groups was performed using two tailed Student’s t test. Time-course analysis was done by two-way analysis of variance (ANOVA) followed by a post hoc test using Bonferroni’s method for individual time points. Statistical significance was considered when P < 0.05.

Supplementary Materials

The PDF file includes:

Figs. S1 to S6

Legends for tables S1 and S2

Other Supplementary Material for this manuscript includes the following:

Tables S1 and S2