Adrenomedullin 2 Enhances Beiging in White Adipose Tissue Directly in an Adipocyte-autonomous Manner and Indirectly through Activation of M2 Macrophages*

Adrenomedullin 2 (ADM2) is an endogenous bioactive peptide belonging to the calcitonin gene-related peptide family. Our previous studies showed that overexpression of ADM2 in mice reduced obesity and insulin resistance by increasing thermogenesis in brown adipose tissue. However, the effects of ADM2 in another type of thermogenic adipocyte, beige adipocytes, remain to be understood. The plasma ADM2 levels were inversely correlated with obesity in humans, and adipo-ADM2-transgenic (tg) mice displayed resistance to high-fat diet-induced obesity with increased energy expenditure. Beiging of subcutaneous white adipose tissues (WAT) was more noticeably induced in high-fat diet-fed transgenic mice with adipocyte-ADM2 overexpression (adipo-ADM2-tg mice) than in WT animals. ADM2 treatment in primary rat subcutaneous adipocytes induced beiging with up-regulation of UCP1 and beiging-related marker genes and increased mitochondrial uncoupling respiration, which was mainly mediated by activation of the calcitonin receptor-like receptor (CRLR)·receptor activity-modifying protein 1 (RAMP1) complex and PKA and p38 MAPK signaling pathways. Importantly, this adipocyte-autonomous beiging effect by ADM2 was translatable to human primary adipocytes. In addition, M2 macrophage activation also contributed to the beiging effects of ADM2 through catecholamine secretion. Therefore, our study reveals that ADM2 enhances subcutaneous WAT beiging via a direct effect by activating the CRLR·RAMP1-cAMP/PKA and p38 MAPK pathways in white adipocytes and via an indirect effect by stimulating alternative M2 polarization in macrophages. Through both mechanisms, beiging of WAT by ADM2 results in increased energy expenditure and reduced obesity, suggesting ADM2 as a novel anti-obesity target.


Adrenomedullin 2 (ADM2) is an endogenous bioactive peptide belonging to the calcitonin gene-related peptide family.
Our previous studies showed that overexpression of ADM2 in mice reduced obesity and insulin resistance by increasing thermogenesis in brown adipose tissue. However, the effects of ADM2 in another type of thermogenic adipocyte, beige adipocytes, remain to be understood. The plasma ADM2 levels were inversely correlated with obesity in humans, and adipo-ADM2-transgenic (tg) mice displayed resistance to high-fat diet-induced obesity with increased energy expenditure. Beiging of subcutaneous white adipose tissues (WAT) was more noticeably induced in high-fat diet-fed transgenic mice with adipocyte-ADM2 overexpression (adipo-ADM2-tg mice) than in WT animals. ADM2 treatment in primary rat subcutaneous adipocytes induced beiging with up-regulation of UCP1 and beiging-related marker genes and increased mitochondrial uncoupling respiration, which was mainly mediated by activation of the calcitonin receptor-like receptor (CRLR)⅐receptor activitymodifying protein 1 (RAMP1) complex and PKA and p38 MAPK signaling pathways. Importantly, this adipocyte-autonomous beiging effect by ADM2 was translatable to human primary adipocytes. In addition, M2 macrophage activation also contributed to the beiging effects of ADM2 through catecholamine secretion. Therefore, our study reveals that ADM2 enhances subcutaneous WAT beiging via a direct effect by activating the CRLR⅐RAMP1-cAMP/PKA and p38 MAPK pathways in white adipocytes and via an indirect effect by stimulating alternative M2 polarization in macrophages. Through both mechanisms, beiging of WAT by ADM2 results in increased energy expenditure and reduced obesity, suggesting ADM2 as a novel anti-obesity target.
An imbalance between energy intake and energy expenditure is the cause for the development of obesity, which is a high risk factor for type 2 diabetes and related metabolic disorders. Adaptive thermogenesis in adipose tissue is an important contributor to overall energy expenditure; thus, enhancing thermogenesis in adipose tissue is considered one of the promising therapeutic strategies to improve energy homeostasis (1).
In contrast to white adipose tissue (WAT), 3 which stores energy as triglycerides, brown adipose tissue (BAT) dissipates energy as heat by uncoupling protein 1 (UCP1)-mediated uncoupling of the mitochondrial respiratory chain from ATP synthesis (2). Upon stimuli such as ␤-3 adrenergic agonists or cold challenge, some adipocytes within WAT can exhibit brown-like features (3,4) and have been identified as the third type of adipocytes, named "brite" (brown-in-white) or "beige" adipocytes (5). This biological process is referred to as WAT "browning" or "beiging" (3).
Studies in both rodents and humans indicated that beiging of WAT increases the whole-body metabolic rate and improves energy homeostasis in obesity and type 2 diabetes (3, 6 -10). Enhancing WAT beiging alone was sufficient to alleviate obesity in mice (6), and emerging evidence also suggests that human thermogenic adipocytes are more similar to mouse beige adipocytes than to mouse brown adipocytes (11,12). Therefore, promoting WAT beiging has attracted great interest as a potential therapeutic approach for metabolic disorders.
Adrenomedullin 2 (ADM2), also known as intermedin, is an endogenous peptide discovered in 2004 and belongs to the calcitonin gene-related peptide (CGRP)/calcitonin family (21,22). The homology of its mature peptide between rodents and humans is rather high (mouse versus rat, 97.9%; mouse versus human, 89.4%; human versus rat, 87.2%). ADM2 is ubiquitously expressed in various tissues, including adipose tissue (23,24), and has been reported to play protective roles in the cardiovascular and renal systems via multiple mechanisms, such as antiinflammation, inhibition of oxidative stress, and endoplasmic reticulum stress (25)(26)(27)(28). ADM2 can be synthesized and secreted from adipocytes, and its expression is down-regulated in adipose tissues of db/db mice and high-fat diet (HFD)-induced obese mice (29). Subcutaneous ADM2 dosing by minipump implantation could improve hyperhomocysteinemia or HFD-induced insulin resistance in mice (29,30). ADM2 was also reported to substantially inhibit adipocyte MHC II expression and thus ameliorate insulin resistance in adipose tissue (31). Previous studies from our group also demonstrated that ADM2 reduced obesity in mice associated with increased thermogenesis in BAT (29); however, the effects of ADM2 in the energy metabolism of beige fat remain largely unknown.
In this study, we investigated the roles of ADM2 in energy homeostasis and WAT beiging and explored the underlying mechanisms. Our results suggest that ADM2 is a novel endogenous beiging activator with the potential to become a new therapeutic target for obesity and related metabolic disorders.

Plasma ADM2 Levels Are Inversely Correlated with Obesity in
Humans-To evaluate endogenous ADM2 levels in humans with different metabolic status, plasma ADM2 levels were measured in 74 Chinese individuals. Body weight and body mass index (BMI) were observed to be inversely correlated with ADM2 plasma levels with statistical significance (p ϭ 0.0014 for body weight and p ϭ 0.0176 for BMI) (Fig. 1, A and B). Based on the diagnostic criteria for obesity in the Chinese population (32), the plasma ADM2 levels were lower by 32% in obese subjects (BMI Ͼ 28) than in subjects with normal weight (BMI Յ 24) (Fig. 1C), implying a strong association of ADM2 with obesity. Neither age nor gender showed a noteworthy correlation with plasma ADM2 levels ( Fig. 1, D and E).
Adipocyte-ADM2 Overexpression Improves Mitochondrial Respiration and Thermogenesis in scWAT-Our previous studies have shown that ADM2 expression in adipose tissues is decreased after HFD treatment (29). Here we also observed a significant inverse relationship between the relative Adm2 mRNA levels in WAT and animal body weight in C57BL/6J male mice (p Ͻ 0.01) ( Fig. 2A). Transgenic mice overexpressing the human ADM2 gene specifically in adipocytes (adipo-ADM2-tg mice) have been generated to study the roles of ADM2 in metabolic homeostasis regulation (supplemental Fig.  S1) (29). In the transgenic mice, ADM2 expression was only increased in adipose tissues at both the transcriptional and translational levels, with a very minimal increase in macrophages at the mRNA but not at the protein level and no increase in other tissues examined (supplemental Fig. S1, B-D). These mice displayed less body weight and fat mass gain during 12-week HFD feeding compared with WT littermates (supplemental Fig. S2, A and B), which was in consistent with our previous observation on an 8-week HFD study (29). Adipo-ADM2-tg mice also displayed increased whole-body energy expenditure without significant differences in either food intake or ambulatory activity (supplemental Fig. S2, C-H), which was previously mainly attributed to the activation of thermogenesis in BAT (29). Here we further explored whether heat production from beige adipocytes additionally contributed to this elevation in systematic energy expenditure. Because scWAT is highly prone to beiging induction (5, 6), mitochondrial metabolic respiration in subcutaneous adipose depots was measured by Seahorse analyzer. There was an increase in the basal and FCCP-stimulated maximal respiration rate of oxygen consumption (OCR) in scWAT, indicating an acceleration of energy metabolism in ADM2-overexpressed adipose tissue (Fig. 2, B and C). The comparable ATP production OCR and declined coupling efficiency (Fig. 2, C and D) showed that the increased energy metabolized in the respiratory chain was not used for ATP synthesis but mostly dissipated for heat production. Consistent with this, the mRNA levels of the key thermogenesis gene Ucp1 and mitochondrial respiration genes Ndufb8, Uqcrc2, and Cox8b were up-regulated in the scWAT of adipo-ADM2-tg mice (Fig. 2E). Moreover, the beige-selective markers Tmem26, Cd137, and Tbx15 were also markedly upregulated, whereas the white adipocyte-related markers Igfbp3 and Lep were markedly down-regulated by adipocyte-ADM2 overexpression (Fig. 2, F and G). To rule out possible effects from other cell types in crude adipose tissues, adipocytes were specifically isolated from scWAT, and the subsequent flow cytometry analysis showed a striking increase in the beiging marker CD137 in adipocytes from adipo-ADM2-tg mice (Fig.  2H). Collectively, these results suggested that adipocyte-ADM2 overexpression activated beiging in scWAT with the enhancement of metabolic respiration and UCP1-mediated thermogenesis, which might make a substantial contribution to increased whole-body energy expenditure and decreased body weight in adipo-ADM2-tg mice.
Cold-induced Beiging Was Markedly Activated in scWAT of Adipo-ADM2-tg Mice-Beige adipocytes can be activated by cold exposure (7,13). We next examined whether cold-induced beiging could be further promoted by adipocyte-ADM2 overexpression. After 48-h cold challenge at 4°C, the subcutaneous adipocytes of adipo-ADM2-tg diet-induced obese mice showed a much greater decrease in cell size than those of WT mice (Fig.  3, A and B). The protein levels of both UCP1 and the beige marker TMEM26 in scWAT were substantially increased by adipocyte-ADM2 overexpression (Fig. 3, C and D). Additionally, adipo-ADM2-tg mice maintained thermal homeostasis at a higher body temperature than WT mice challenged with cold exposure (Fig. 3E). Taken together, these results indicate that adipocyte-ADM2 overexpression could enhance cold-induced beiging and thermogenesis in scWAT.
ADM2 Treatment Promotes Beiging in Vitro in an Adipocyteautonomous Manner-To investigate how ADM2 induced beiging of white adipocytes from scWAT, rat scWAT-derived SVCs were isolated and differentiated into mature adipocytes in vitro. After treatment with ADM2 in cell culture, rat adipocytes were subjected to RNA sequencing (RNAseq) profiling analysis. A general up-regulation of mitochondrial respiration and thermogenesis-related genes was observed in ADM2treated adipocytes, suggesting a white-to-beige switch at the molecular level in adipocytes by ADM2 treatment (Fig. 4A). qPCR analysis further verified those transcriptional changes. Ucp1 expression was robustly increased more than 10-fold by ADM2 treatment, and other mitochondrial respiration-related genes, including Pgc1a, Ndufb8, Sdhb, Cox5b, and Cox8b, were also markedly up-regulated (Fig. 4B). Additionally, beige-selective markers, including Cd137, Tmem26, Fgf21, and Cited, were all up-regulated, whereas the mRNA expression of white adipocyte markers, including Retn and Igfbp3, was decreased by ADM2 treatment (Fig. 4C). As a key thermogenesis protein in beige adipocytes (2), UCP1 was also up-regulated at the protein level, as shown by Western blotting and immunofluorescent staining (Fig. 4, D and E). Moreover, ADM2 treatment resulted in a marked increase in mitochondrial respiration in primary subcutaneous adipocytes, as measured by Seahorse analyzer (Fig. 4, F and G). The basal and maximal respiration OCRs of adipocytes were improved by 62% and 49%, respectively (Fig.  4G). More importantly, the uncoupling OCR was substantially elevated by nearly 2-fold, and the coupling efficiency declined from ϳ70% to 50% after ADM2 treatment (Fig. 4, G and H), suggesting enhanced mitochondrial uncoupling along with heat generation. Consistently, the mitochondrial membrane potential was decreased after ADM2 treatment, as shown by reduced TMRM fluorescence intensity (Fig. 4I). Altogether, these results indicate that ADM2 could promote white adi-FIGURE 1. Plasma ADM2 levels are inversely correlated with obesity in humans. Plasma samples were collected from a total of 74 Chinese adults, and the ADM2 levels were measured by radioimmunoassay. A, negative correlation of body weight (BW) versus plasma ADM2 levels. B, inverse correlation of BMI and plasma ADM2 levels. C, comparison of the plasma ADM2 levels in normal (BMI Յ 24), overweight (24 Ͻ BMI Յ 28), and obese (BMI Ͼ 28) human subjects. D, no significant correlation of age versus plasma ADM2 levels. E, similar plasma ADM2 levels in the male (n ϭ 38) and female (n ϭ 36) human subjects. For group comparison (C and E), the data are represented as mean Ϯ S.E. One-way ANOVA with Newman-Keuls test (C) or two-tailed Student's t test (E): *, p Ͻ 0.05 versus control. For linear regression (A, B, and D), the slope was considered significantly non-zero when p Ͻ 0.05. pocytes beiging and increased UCP1-mediated thermogenesis in a cell-autonomous manner.
ADM2 Up-regulates UCP1 Expression in Adipocytes Mainly through the CRLR⅐RAMP1-cAMP/PKA and p38 MAPK Pathways-We next explored the underlying signal transduction mechanisms for the adipocyte-autonomous beiging effects by ADM2. The receptor complexes of ADM2 are composed of a calcitonin-receptor-like receptor (CRLR) and one of the three receptor activity-modifying proteins (RAMPs) (21). Ucp1 up-regulation by ADM2 in adipocytes was markedly attenuated by the CRLR non-selective antagonist ADM17-47 (65%) and by the CRLR-RAMP1-selective antagonist CGRP8-37 (58%) but not by the CRLR-RAMP2-and CRLR-RAMP3-selective antagonist ADM22-52 (Fig. 5A), implying that CRLR-RAMP1 was the major receptor for the beiging effects of ADM2. G s -mediated activation of adenylate cyclase represents the major signaling pathway coupled with the CRLR-RAMP receptors (33). The cellular cAMP content in adipocytes was increased by ϳ2-fold after ADM2 treatment (Fig. 5B). cAMP-dependent PKA was also activated, with PKA-mediated substrate phosphorylation increased as early as 1 h after ADM2 treatment (Fig. 5C). Furthermore, the cAMP antagonist Rp-cAMPS and the PKA inhibitors H89 and PKI markedly attenuated up-regulation of Ucp1 by ADM2 by 67%, 87%, and 76%, respectively (Fig. 5E). The activated phosphorylation of PKA substrate by ADM2 was also suppressed by those inhibitors at the same concentration (supplemental Fig. S3A). The p38 MAPK pathway is considered a key pathway mediating beiging for several other endogenous peptides (14,17,18). Immunoblotting revealed a time-dependent elevation of phosphorylated p38 MAPK. Its phosphorylation level began to increase from 1-2 h and reached a peak 4 -8 h after ADM2 treatment (Fig. 5D). Consistently, Ucp1 up-regulation and p38 MAPK phosphorylation after ADM2 stimulation were both markedly attenuated by pretreatment with the p38 MAPK inhibitor SB202190 (Fig. 5E and supplemental Fig. S3B). The PI3K-Akt inhibitor LY294002 showed no inhibitory effects (Fig. 5E). Neither inhibitor suppressed Ucp1 expression in adipocytes in the absence of ADM2 (supplemental Fig. S3C). Therefore, the direct effects of ADM2 on UCP1 up-regulation and beiging in white adipocytes were mainly mediated by the CRLR⅐RAMP1-cAMP/PKA and p38 MAPK pathways.
ADM2 Treatment Activates Beiging of Primary Human White Adipocytes-To address whether ADM2 was also capable of beiging activation in humans, primary human subcutaneous white adipocytes were subjected to ADM2 treatment.
Consistent with the findings in rodent cells, ADM2 substantially induced the transcription of UCP1 and the beiging-related genes CD137 and MTUS1 (Fig. 6A) and down-regulated the mRNA levels of the white adipocyte-related markers LEP and IGFBP3 (Fig. 6B) in human primary adipocytes. UCP1 was also markedly elevated at the protein level by ADM2 treatment, as determined by both immunofluorescent staining and Western blotting (Fig. 6, C and D). Furthermore, ADM2 treatment improved the mitochondrial respiration capacity, as demonstrated by a 42% increase in maximal respiration OCR (Fig. 6, E  and F). These results indicate that the adipocyte-autonomous action underlying the effect of ADM2 on WAT beiging is translatable to humans.
M2 Macrophage Activation Also Contributes to the Beiging Effects of ADM2-The anti-inflammatory M2 macrophages resident in WAT are considered an important mediator for beige adipocyte recruitment according to the latest studies (13,19,20). ADM2 has been reported to be anti-inflammatory and atherosclerosis-protective by regulating macrophage functions in the cardiovascular system (34,35), which prompted us to ask whether ADM2 could induce beiging of white adipocytes in adipose tissues by cross-talk with M2 macrophages. qPCR results showed that the mRNA levels of M2 macrophage markers, including Cd206 and Arginase 1 (Arg1), were markedly up-regulated in scWAT in adipo-ADM2-tg mice, whereas the expression of M1 macrophage markers (Cd11c and iNOS) was not changed (Fig. 7A). Flow cytometry analysis consistently demonstrated a marked increase in the percentage of M2 macrophages (F4/80ϩ, CD206ϩ, CD11cϪ) within the stromal vascular fraction of scWAT in adipo-ADM2-tg mice compared with WT animals (Fig. 7B), displaying an accumulation of M2 macrophages in scWAT by ADM2 overexpression. Mouse peritoneal macrophages were isolated and treated with ADM2 in vitro. Up-regulation of the M2 marker Arg1 in macrophages at both the mRNA and protein levels after 12-h ADM2 treatment showed that ADM2 could directly induce M2 polariza-tion in macrophages (Fig. 7, C and D). Local catecholamine secretion from M2 macrophages in WAT is a crucial inducer for beige fat recruitment through the direct stimulation of ␤3 adrenergic receptor on adipocytes (13,20). Tyrosine hydroxylase (Th), dopa decarboxylase (Ddc), and dopamine ␤-hydroxylase (Dbh) are three key enzymes for catecholamine synthesis, and ADM2 treatment resulted in a substantial increase in the protein levels of Th and a small but significant increase in Ddc in macrophages (Fig. 7E). Furthermore, norepinephrine (NE) secretion for 8 h from ADM2-treated macrophages was more than 2-fold that from control cells (Fig. 7F).
To determine whether the catecholamine released by ADM2-treated macrophages was sufficient to enhance beiging of adipocytes, conditional medium collected from macrophages was transferred to incubate adipocytes (Fig. 8A). Ucp1 expression in adipocytes was markedly elevated by ϳ2-fold with the conditional medium from ADM2-treated macrophages compared with that from control macrophages. Moreover, this increase in Ucp1 expression could be blocked by propranolol, a ␤-adrenergic receptor antagonist (Fig. 8B). Taken together, these results suggested that M2 polarization along with increased catecholamine secretion from macrophages is an additional important mechanism underlying the beiging effects of ADM2 scWAT.

Discussion
Our previous researches regarding the metabolic benefits of ADM2 mainly focused on the activation of BAT (29). In this study, we further demonstrated the beneficial effects of ADM2 on energy homeostasis and highlighted its roles in enhancing WAT beiging. The results of our study collectively point to a working model for the effects of ADM2 on WAT beiging, as illustrated in Fig. 8C. On one hand, ADM2 interacts with the CRLR-RAMP1 receptor complex on white adipocytes and stimulates G s -mediated activation of adenylate cyclase, resulting in the cAMP-dependent activation of PKA. p38 MAPK is also activated in response to ADM2 stimulation. These two kinases further drive the up-regulation of thermogenesis-and beiging-related genes, including UCP1, and lead to beiging of white adipocytes directly. On the other hand, ADM2 acts on resident macrophages in WAT and stimulates M2 polarization along with catecholamine secretion, which, in turn, activates the ␤3 adrenergic receptor on adipocytes and enhances beiging indirectly. Beige adipocytes activated by the above two mechanisms dissipate excess energy through metabolic thermogenesis, thus improving systematic energy homeostasis and alleviating obesity and related metabolic disorders.
Because scWAT is highly prone to beiging induction compared with visceral WAT (5, 6), we chiefly focused on fat pads from subcutaneous fat depots in animal models and adopted differentiated scWAT adipocytes for all in vitro experiments. In addition to the expressional changes in thermogenesis-and beiging-related genes, this study further demonstrated ADM2mediated improvement of mitochondrial respiration in both ex vivo isolated subcutaneous fat tissues as well as in vitro differ- entiated subcutaneous adipocytes. These results provide direct functional evidence to support the beiging effects of ADM2 in scWAT.
PKA and p38 MAPK are both reported key signaling mediators for white adipocyte beiging (14,17,18,36). PKA drives beiging-related transcriptional responses via the activation of cAMP-response element-binding protein and PGC-1␣ (7). p38 MAPK is necessary for UCP1 expression through activating ATF2 and PGC-1␣ (36) and is activated in response to cAMP/ PKA signaling (37). In our study, phosphorylation of p38 MAPK also followed PKA induction, suggesting that p38 MAPK might be a downstream effecter of PKA in the ADM2activated signaling pathway, but the precise relationship between these two signals needs to be further clarified. TSC1-mTORC1 signaling has recently been identified as a linker between extracellular signals and transcriptional factors to initiate the brown-to-white adipocyte phenotypic switch (38). Whether this pathway is also down-regulated upon ADM2 treatment is worth further investigation.
It has been well recognized that anti-inflammatory M2 macrophages serve as an important source of catecholamine for beiging activation in scWAT (13,19,20). Endogenous secretors such as IL-4, IL-13, and adiponectin have been reported to promote beiging of white adipocytes via cross-talk with resident macrophages in WAT (8,13,19). In this study, we also observed a marked activation of beiging accompanied by an accumulation of M2 macrophages in scWAT in adipo-ADM2-tg mice. In vitro studies also displayed M2 polarization of primary mouse macrophages after ADM2 treatment. It has been shown before that the CRLR-cAMP/PKA-pPTEN pathway is responsible for ADM2 inhibition of acetylated low-density lipoprotein uptake in macrophages (25), but the receptor and transduction signals involved in its M2 activation need to be addressed in the future. This study indicates that M2 macrophages also serve as a crucial contributor to the ADM2-mediated beiging effect in scWAT.
The cellular origin of beige adipocytes has not been completely elucidated. A bipotential precursor population capable of de novo beige and white differentiation has been identified (39). Beige fat biogenesis can be enhanced in multiple ways, such as stimulation of bipotential precursor proliferation, promotion of their subsequent commitment to the beige lineage (40), and induction of phenotype switch in differentiated white adipocytes (41). Our in vitro results collectively showed that ADM2 could exert its effects on differentiated adipocytes, thus possibly induce beiging by stimulating white-to-beige conversion. Whether other mechanisms related to beige precursors are also involved requires further investigation. A significant increase in carbon dioxide production (VCO 2 ) in adipo-ADM2-tg mice was observed during daytime but not during nighttime (supplemental Fig. S2E). Both adipocyte metabolism and thermogenesis are under the control of circadian rhythms (42,43). CGRP, another member of the same peptide family as ADM2, has been found to have a circadian profile of plasma concentration (44). Thus it is possible that ADM2 could function in a circadian manner to regulate beiging and metabolic thermogenesis.
To identify the effects of ADM2 in adipose tissues, an adipocyte-specific transgenic mouse model was developed using the aP2 gene promoter in this study. It has been claimed previously that the aP2 promoter is expressed in activated macrophages in addition to adipocytes (35). In adipo-ADM2-tg mice, the overexpression efficiency of the human ADM2 gene was dramatically lower in macrophages than in adipose tissues (supplemental Fig. 1, B and C), which is consistent with other reports in which the efficiency of the FIGURE 8. ADM2 promotes beiging in scWAT through cross-talk between adipocytes and macrophages. A and B, conditioned medium test. A, the conditional medium system for adipocytes and macrophages. After treatment with 20 nM ADM2 for 12 h, fresh medium was given to macrophages for catecholamine accumulation. Then the conditioned medium of macrophages was collected and given to adipocytes. After incubation with the medium for 8 h, adipocytes were collected for Ucp1 mRNA analysis. B, relative mRNA levels of Ucp1 in adipocytes treated with conditioned medium from ADM2-treated or control macrophages for 8 h with or without 1-h pretreatment of the ␤-adrenergic receptor antagonist propranolol (10 M), and adipocytes treated with normal control medium were used as a negative control. ␤-Actin was used as a normalization gene. Data are represented as mean Ϯ S.E. n ϭ 3 independent experiments (3-6 samples/experiment). One-way ANOVA with Newman-Keuls test: **, p Ͻ 0.01 versus normal control medium; ##, p Ͻ 0.01 versus control conditioned medium; &&, p Ͻ 0.01 versus ADM2-treated conditioned medium. C, proposed model of the effects of ADM2 on WAT beiging and energy homeostasis. As an endogenous beiging activator, ADM2 interacts with the CRLR-RAMP1 receptor on adipocytes and activates beiging of white adipocytes directly. Additionally, ADM2 acts on resident macrophages in scWAT and promotes M2 polarization and catecholamine production. The catecholamine secreted locally by M2 macrophages in turn activates the ␤3 adrenergic receptor on adipocytes and enhances white adipocyte beiging. The induced beige adipocytes utilize UCP1 to uncouple oxidative respiration and dissipate excess energy by thermogenesis, thus improving systematic energy homeostasis and attenuating HFD-induced obesity and related metabolic disorders. aP2 promoter in macrophages was much lower than that of adipocytes (45)(46)(47).
Because of their inducible nature and relevance to human health, thermogenic beige adipocytes have been considered a potentially new therapeutic target for energy imbalance, and the discovery of endogenous beiging activators will provide new strategies to defend against metabolic diseases. Our results revealed a direct beiging effect of ADM2 in human primary adipocytes at both the expressional and functional levels. The inverse correlation between human plasma ADM2 and BMI also suggests that ADM2 might exert an anti-obesity role in humans. These results together suggest that the mechanism of action of ADM2 revealed in rodent studies might be translatable to humans and provide a strong scientific rationale to further explore ADM2 as a potential new therapeutic target for obesity and metabolic disorders in future studies. Human Subjects-Blood samples were obtained from 74 human subjects (38 men and 36 women) from the outpatient department of Beijing Chao-Yang Hospital. Basic characteristics of the human subjects are listed in supplemental Table 1. Approximately 2-3 g of adipose tissue was excised from abdominal subcutaneous fat depots of metabolically healthy human subjects undergoing abdominal surgical operations at Peking University Third Hospital. Approval from the Local Research Ethics Committee was granted for human tissue use, and the procedures used were in accordance with institutional guidelines and the Code of Ethics of the World Medical Association (Declaration of Helsinki). All patients gave their informed consent before the procedures.

Materials-Human
Animals-The transgenic mouse line with adipocyte-specific overexpression of the ADM2 gene (adipo-ADM2-tg mice) was generated on the C57BL/6J background with the human ADM2 gene driven by the fatty acid-binding protein (aP2) gene promoter as described previously (29). Male adipo-ADM2-tg mice at 6 -8 weeks of age and their WT littermates were housed in a temperature-controlled room (22°C) with a 12-h light/dark cycle. Animals were fed ad libitum with either a normal chow diet or high-fat diet (D12492, Research Diets, New Brunswick, NJ) for 12 weeks. For the cold challenge experiment, mice were housed individually in a 4°C chamber for 48 h. The body temperature was measured by an electric rectal thermometer. All studies followed the guidelines of the Animal Care and Use Committee of Peking University and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.
Cell Culture-Primary stromal vascular cells (SVCs) were isolated from the scWAT of male Sprague-Dawley rats (160 -180 g body weight) or from human scWAT biopsy samples as described previously (48,49). The isolated SVCs were plated and cultured to confluence in DMEM/F12 (1:1) medium containing 10% FBS. For rat adipocytes, SVCs were differentiated into adipocytes in serum-free DMEM/F12 (1:1) supplemented with 5 g/ml insulin, 33 M biotin, and 200 pM tri-iodothyronine. After 2 days of differentiation, the differentiation mixture was removed, and cells were maintained in serum-free DMEM/ F12 (1:1) for another 2-3 days before experiments. For human adipocytes, SVCs were differentiated initially in 10% FBS-DMEM/F12 (1:1) supplemented with 500 M isobutylmethylxanthine, 5 g/ml insulin, 0.1 M dexamethasone, and 33 M biotin. After 2 days of differentiation, the differentiation mixture was removed, and cells were induced for an additional 2 days with 5 g/ml insulin and 33 M biotin in 10% FBS-DMEM/ F12 (1:1) and then maintained in 10% FBS-DMEM/F12 for further experiments. The drugs used for adipogenic differentiation were all purchased from Sigma-Aldrich (St. Louis, MO). Mouse peritoneal macrophages were isolated as described previously (25). Eight-week-old C57BL/6J male mice were injected intraperitoneally with 2 ml of 4% thioglycollate broth (BD Biosciences Clontech). Three days later, macrophages were obtained by peritoneal lavage with 8 ml of cold PBS with 10 mM EDTA and 10% FBS. Cells were plated at 1.0 ϫ 10 6 /ml of RPMI 1640 with 10% FBS. After incubation for 3 h at 37°C, nonadherent cells were washed away, and adherent cells were used for treatment.
Mitochondrial Respiration Measurements-SVCs were plated into XF24 cell culture microplates and differentiated into adipocytes. Tissue samples were freshly excised from scWAT and placed into XF24 islet capture microplates as described previously (50). Mitochondrial energetics were determined by Seahorse XF flux analyzer. The OCR was measured in the basal state and in response to respiratory chain modulators injected sequentially as follows: oligomycin (1 M for cells and 10 M for tissues), FCCP (carbonyl cyanide p-trifluoromethoxyphenylhydrazone, 1 M for cells and 8 M for tissues), rotenone (1 M for cells and 3 M for tissues), and antimycin A (1 M for cells and 12 M for tissues). Basal, uncoupling (proton leak), or maximal respiration OCR was calculated by subtracting the OCR measured after antimycin A and rotenone addition from the OCR measured in the basal state, measured after oligomycin addition, or measured after FCCP addition, respectively. ATP production OCR was calculated as the decline in OCR in the basal state after oligomycin addition. Coupling efficiency was calculated as the ratio of the ATP production OCR to basal respiration OCR. The respiratory chain modulators were all purchased from Sigma-Aldrich. The microplates and analyzer were from Seahorse Bioscience (North Billerica, MA).
Flow Cytometry Analysis-For adipocyte staining, primary mature adipocytes isolated from scWAT were stained with anti-PE-CD137 antibody or isotype matched IgG-labeled PE. For adipose tissue macrophage staining, SVCs isolated from scWAT were stained with a mixture of anti-FITC-CD45, anti-APC-F4/80, anti-PE-CD11c, and anti-BV421-CD206 antibod-ies. To set proper compensation and population gates, singlecolor positive cells were stained with each antibody alone, and cells were incubated with isotype-matched IgG labeled with FITC, APC, PE, or BV421 as negative controls. The stained cells then underwent analysis with a FACS Aria flow cytometer, and the data were analyzed using CellQuest software. The anti-BV421-CD206 antibody was from BioLegend (San Diego, CA), and other antibodies and instruments used were from BD Biosciences.
Histology of Adipose Tissues and Adipocytes-H&E staining and immunohistochemistry of scWAT paraffin sections and immunofluorescent staining of adipocytes were conducted as described previously (13,49). Adipocytes were stained with tetramethylrhodaminemethyl ester (TMRM) for membrane potential. Lipid droplets were stained by LipidTOX, and nuclei were stained with Hoechst 33258. Immunofluorescent signals were observed with a Leica confocal microscope. All fluorescent dyes were from Invitrogen.
RNAseq Profiling-Total RNA from differentiated adipocytes was isolated using TRIzol reagent (Promega, Madison, WI). RNAseq profiling and data analysis were conducted by the Beijing Genomics Institute.
ELISA and RIA-The cellular cAMP content in differentiated adipocytes and NE released into medium by macrophages were detected using ELISA kits from Cloud-Clone Corp. (Houston, TX). The ADM2 levels in plasma or WAT homogenates were measured using an intermedin/adrenomedullin-2 RIA kit from Phoenix Pharmaceuticals.
Quantitative RT-PCR and Immunoblotting-Quantitative RT-PCR (qPCR) reactions were performed according to previous protocols (48). Relative target mRNA levels were normalized to 18S rRNA or ␤-actin. The primers used for qPCR analysis are listed in supplemental Table 2.
Statistical Analysis-The data are expressed as mean Ϯ S.E. For group comparison, one-way ANOVA followed by Newman-Keuls test for post testing (multiple-group comparison) or unpaired two-tailed Student's t test (two-group comparison) was used for statistical analysis. p Ͻ 0.05 was considered statistically significant. For correlation analysis, linear regression was used, and the slope was considered significantly non-zero when p Ͻ 0.05. The value of r represents the goodness of fit.
Author Contributions-Y. L., S. Z., X. L., H. Z., Z. X., B. L., M. X., C. J., J. S., and X. W. made substantial contributions to acquisition of data or analysis and interpretation of data. Y. L., C. J., J. S., and X. W. contributed to the design of the experiments and writing of the manuscript. All the authors approved the final version of the paper. C. J. and J. S. are the guarantors of this work and, as such, had full access to all data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.