Growth Differentiation Factor-5 Promotes Brown Adipogenesis in Systemic Energy Expenditure

Adipose tissue is a complex and essential organ highly responsible for the regulation of systemic energy homeostasis (1). The two types of adipose tissues are white adipose tissue (WAT) and brown adipose tissue (BAT). WAT functions as an energy storage depot and an endocrine organ that secretes adipokines, whereas BAT contains multilocular lipid droplets that generate heat through mitochondrial uncoupling of lipid oxidation (2–4). Brown adipocytes residing in the interscapular and perirenal regions are developmentally related to skeletal muscle (5), whereas brown fat-like adipocytes expressing uncoupling protein 1 (UCP1), which are also known as brite or beige cells, are sporadically found in the WAT of rodents and humans (6). Although the amount of BAT was thought to be minimal in human adults, recent studies have demonstrated that adult humans have substantial amounts of functioning BAT (7–9). On the basis of these findings, the promotion of BAT function is believed to be a potential approach in combating obesity and relevant metabolic diseases in humans (10).

Bone morphogenetic proteins (BMPs)/growth differentiation factors (GDFs) belong to the transforming growth factor-β (TGF-β) superfamily, which generates intracellular signals through a mechanism related to type I and type II serine/threonine kinase receptors expressed at the cell surface as multifunctional regulators of development and tissue homeostasis (11). In genetic studies in humans and mice, perturbed BMP signaling induces the pathogenesis of various disorders such as skeletal and vascular diseases as well as cancer (12,13). By contrast, BMP/GDF family members play different roles in adipogenesis (14). BMP2 and BMP4 have both been shown to accelerate white adipogenesis (15,16), whereas BMP7 has been demonstrated to promote brown adipocyte differentiation and thermogenesis (17). Moreover, BMP8B increases BAT thermogenesis through central and peripheral actions (18).

GDF5 is also a member of the BMP/GDF subfamily (19) and is sometimes referred to as cartilage-derived morphogenetic protein-1. In fact, GDF5 is expressed in the condensing mesenchyme of the cartilage primordium during early limb development, with predominant expression at the sites of joint formation referred to as the interzone region at the late developmental stage (20). Furthermore, GDF5 mutations lead to skeletal dysplasia and osteoarthritis in humans and mice (21,22). However, little attention has been paid to the role of GDF5 in adipogenesis and systemic energy homeostasis. In the current study, therefore, we attempted to demonstrate the possible involvement of GDF5 in the regulation of brown adipogenesis as well as in the pathogenesis of obesity and related metabolic diseases.

The expression and reporter vectors used in this study are listed in Supplementary Table 1. Antibodies were from different companies as follows: anti-mothers against decapentaplegic homolog-5 (Smad5) and anti–β-actin from Santa Cruz Biotechnology, anti-GDF5 from R&D Systems, anti-GAPDH and anti-phospho Smad1/5/8 from Cell Signaling Technology, anti-UCP1 from Abcam, anti-lamin B1 from Zymed Laboratories, and anti–β-tubulin from Sigma-Aldrich. Mouse GDF5 ELISA Kit was obtained from Blue Gene. DIO Rodent Purified Diet with 60% energy from fat was obtained from Japan SLC. Escherichia coli–derived recombinant mouse GDF5 (853-G5: <1.0 EU/1 μg cytokine, as determined by the Limulus amebocyte lysate method) was purchased from R&D Systems.

GDF5^Rgsc451 mice (M100451) were provided by the RIKEN BioResource Center through the National Bio-Resource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan (23). The ob/ob mice with the C57BL/6 background were obtained from Japan SLC. To generate aP2-GDF5 transgenic mice, mouse GDF5 cDNA was cloned into a plasmid containing a 5.4-kb mouse aP2 promoter, and the transgenic mice were generated by the pronuclear injection methods into C57BL/6 mice. Two independent transgenic founder lines were identified by Southern blot. The probe for Southern blotting was generated by PCR using the following primers: 5′-CAAACTCCTCACTCTTTTGCTGTG-3′ and 5′-GTTACCTCCCTTTCTGTCAGCATC-3′. Genotyping was performed using the following primers: 5′-AAGCATCTTCAAAAGCAGGATCT-3′ and 5′-GCCTTAATCTCATTAAAGAACAGGTC-3′.

Mice were maintained at 23–24°C on a 12-h light/dark cycle with free access to food and water. Male mice were used throughout the experiments. The protocol used here met the guidelines of the Japanese Society for Pharmacology and was approved by the Kanazawa University Committee for Ethical Use of Experimental Animals.

For glucose tolerance tests, glucose (2 g/kg) was intraperitoneally injected into animals after overnight fasting, and blood glucose was monitored with blood glucose strips and the Accu-Check glucometer at the indicated times, as described previously (24). To measure food intake, mice were individually housed for a week before food intake was measured. Mice were fed powder diets for 1 week to measure the accumulated amount of food intake (25). Open-field tests were performed to determine locomotor activity, as described previously (25). Metabolic rates were measured by indirect calorimetry in mice using the O₂/CO₂ metabolism measuring system for small animals (Muromachi Kikai). After a 24-h acclimatization period, Vo₂ and Vco₂ data were collected for 24 h.

Brown preadipocyte cell lines were derived from newborn wild-type (WT) mice (17). Stromovascular fraction (SVF) cells were prepared from adipose tissues as described previously (26). In brief, adipose tissues dissected from WT mice at 6–8 weeks old were minced and digested with 1 mg/mL collagenase for 45 min in Dulbecco's modified Eagle's medium (DMEM) containing 1% BSA. Tissues were then filtered through a 250-μm nylon mesh to remove undigested fragments. The suspension was centrifuged at 250g for 5 min, and the pellets were used as the SVF cells.

To examine the effect of GDF5 on adipogenesis, in vitro adipocyte differentiation was determined using two different methods, as described previously with slight modifications (17,26). To test the effect on differentiation processes to brown adipocytes, brown preadipocytes were cultured in DMEM containing 20 nmol/L insulin and 1 nmol/L triiodothyronine (T3) in the presence or absence of recombinant GDF5 for 7 days after confluence. For evaluation of the effect on differentiated brown adipocytes, alternatively, brown preadipocytes and SVF cells were individually cultured in DMEM to confluence, followed by further culture in DMEM containing the differentiation induction cocktail composed of 20 nmol/L insulin, 1 nmol/L T3, 0.125 mmol/L indomethacin, 0.5 μmol/L dexamethasone, and 0.5 mmol/L isobutyl methylxanthine for 2 days and subsequent culture in DMEM supplemented with 20 nmol/L insulin and 1 nmol/L T3 in the presence or absence of recombinant GDF5 for an additional 7 days. Cells were stained with Oil Red O using standard procedures as described elsewhere (25).

For luciferase assays, cells were transiently transfected with vectors by the lipofection method. Expression vectors of BMP7 and GDF5, in addition to BMP responsive element-luc used, are summarized in Supplementary Table 1. Serial 5′-deletion constructs of the peroxisome proliferator–activated receptor-γ (PPAR-γ) coactivator 1α (Ppargc1a) promoter were generated by the PCR-based cloning method using specific primers (Supplementary Table 2). The oligonucleotides for Smad5 short hairpin RNA (5′-GATCCGGTGTTCATCTATACTACGTTCAAGAGACGTAGTATAGATGAACACCTTTTTTG-3′ and 5′- AATTCAAAAAAGGTGTTCATCTATACTACGTCTCTTGAACGTAGTATAGATGAACACCG-3′) were synthesized, annealed, and inserted into the RNAi-Ready pSIREN-RetroQ vector. Retroviral vectors were transfected into PLAT-E cells (27) using the calcium carbonate method. Viral supernatants were collected 48 h after transfection, and the cells were infected with the supernatants for 72 h in the presence of 4 μg/mL polybrene. Cells were then subjected to selection by culture with 1 μg/mL puromycin for 3 days before being used in experiments.

Total RNA was extracted from cells or tissues, followed by synthesis of cDNA with reverse transcriptase and oligo-dT primers. The cDNA samples were then used as templates for real-time PCR analysis using specific primers for each gene (Supplementary Table 3). Expression levels of the genes examined were normalized using the 36b4 expression levels as an internal control for each sample. Chromatin immunoprecipitation (ChIP) experiments were performed following the protocol provided with the ChIP assay kit (Millipore). The following primers were used: 5′-CTGACTATATAAAGTCAGACTGAG-3′ and 5′-CAGTGAGAAATGCACAGGGGACCT-3′.

Tissues and cultured cells were solubilized in lysis buffer containing 1% Nonidet P-40. Samples were then subjected to SDS-PAGE, followed by transfer to polyvinylidene fluoride membranes and subsequent immunoblotting.

Tissues were fixed with 10% formalin and embedded in paraffin. Sections were dissected at a thickness of 5 μm and stained with hematoxylin and eosin. For immunohistochemical analysis, the VECTASTAIN ABC kit (Vector Laboratories) was used.

Results are expressed as the mean ± SEM, and the statistical significance was determined by the Student t test and one- or two-way ANOVA with post hoc test.

To evaluate the physiological and/or pathophysiological importance of GDF5 in adipose tissues in vivo, we examined whether obesity affected GDF5 expression in adipose tissues using genetic and acquired obesity models. In obese ob/ob mice, a drastic and significant increase in GDF5 expression was preferentially seen in interscapular BAT but not in epididymal visceral WAT (vWAT) or inguinal subcutaneous WAT (sWAT) or in other tissues such as bone, muscle, liver, and pancreas (Fig. 1A). Similarly, GDF5 protein was upregulated in BAT, but not in vWAT or sWAT, in ob/ob mice (Fig. 1B), which was confirmed by the quantification of Western blot data (Fig. 1C). Significantly increased GDF5 levels were found in the BAT of ob/ob mice by ELISA (Fig. 1D), whereas circulating levels of GDF5 were not significantly altered between mice with the two genotypes (Fig. 1E).

We next examined whether acquired obesity similarly affected GDF5 expression in BAT as seen in genetically obese mice. For this purpose, 8-week-old C57BL/6J mice were maintained with normal chow (NC) or a high-fat diet (HFD) for 3 consecutive months as a model of diet-induced obesity (DIO). As seen in the genetically obese ob/ob mice, GDF5 expression was preferentially upregulated in BAT, but not in vWAT and sWAT, of mice fed the HFD compared with those fed NC (Fig. 1F). Moreover, immunoreactive GDF5 was highly detected in the BAT of mice fed the HFD compared with BAT of mice fed NC (Fig. 1G).

We next attempted to generate transgenic mice with the C57BL/6J background under the control of the 5.4-kb promoter of mouse aP2 (28) (Supplementary Fig. 1A), which drives predominant expression in murine adipose tissues (such as sWAT, vWAT, and BAT) and macrophages (29). Among the different transgenic mice with the integrated transgene, we selected the two lines, #4 and #5, according to the data from Southern blot analysis (Supplementary Fig. 1B). In these aP2-GDF5 transgenic mice, GDF5 protein levels were markedly elevated in sWAT, vWAT, and BAT compared with levels in WT mice (Supplementary Fig. 1C and D). The GDF5 levels in peritoneal macrophages were also increased in aP2-GDF5 mice (Supplementary Fig. 1E), whereas transgene expression was preferentially seen in sWAT, vWAT, BAT, and peritoneal macrophages (Supplementary Fig. 1F). However, circulating GDF5 levels were not significantly increased in aP2-GDF5 mice (Supplementary Fig. 1G). The bone volume of vertebrae was comparable between aP2-GDF5 and WT mice (Supplementary Fig. 1H).

The body weight of aP2-GDF5 mice began to significantly diverge from that of WT mice by 16 weeks after birth during the maintenance on NC (Fig. 2A). The tissue weight relative to body weight was significantly lower in vWAT and sWAT, but not in BAT, from aP2-GDF5 mice than in those from WT mice, with no significant changes in other organs such as the liver, pancreas, spleen, kidney, and heart (Fig. 2B). Moreover, the nasoanal length was comparable between aP2-GDF5 and WT mice (Fig. 2C). No significant change was found in the amount of daily food intake between aP2-GDF5 and WT mice (Fig. 2D). Histological analysis revealed that the aP2-GDF5 mice had adipocytes with a significantly smaller size compared with WT mice (Fig. 2E), which was quantitatively supported by counting the numbers of cells with different diameters (Fig. 2F). Glucose tolerance tests showed significantly lower levels of blood glucose in aP2-GDF5 mice than in WT mice when determined 15–60 min after the intraperitoneal injection of glucose (Fig. 2G).

Adult aP2-GDF5 mice were then fed the HFD for 3 consecutive months for subsequent determination of body weight, tissue weight, and glucose tolerance. After being fed the HFD for at least 6 weeks, the gain in body weight was significantly lower in aP2-GDF5 than in WT mice (Fig. 2H). In aP2-GDF5 mice fed the HFD, the relative tissue weight was significantly lower in vWAT and sWAT than in those of WT mice (Fig. 2I), with no significant changes in the other tissues examined (Supplementary Fig. 2), as seen in the mice maintained on NC. Moreover, aP2-GDF5 mice displayed a marked increase in glucose tolerance compared with WT mice under HFD feeding conditions (Fig. 2J).

Physical activity and Vo₂ were monitored in aP2-GDF5 mice. Vo₂ (Fig. 3A) and energy expenditure (Fig. 3B) were significantly higher in aP2-GDF5 mice than in WT mice when measured as per animal according to recent recommendations (30). No significant difference was observed in spontaneous locomotor activity between mice with different genotypes (Fig. 3C). To understand how aP2-GDF5 mice showed increased Vo₂, we determined the endogenous level of the mitochondrial protein responsible for cellular energy expenditure, UCP1, in adipose tissue. UCP1 expression was significantly increased in the sWAT, but was not detected in the vWAT, of aP2-GDF5 mice compared with WT animals (Fig. 3D), whereas a slight but statistically significant increase was also seen in UCP1 expression in the BAT of aP2-GDF5 mice (Fig. 3E).

The histological analysis showed sWAT contained numerous clusters of smaller UCP1-expressing adipocytes interspersed within the larger unilocular white adipocytes in aP2-GDF5 mice, in contrast to WT mice, which predominantly displayed UCP1-negative large unilocular adipocytes (Fig. 3F). In the sWAT of aP2-GDF5 mice, increased expression was seen with Ucp1, Ppargc1a, and PRD1-BF-RIZ1 homologous domain–containing protein-16 (Prdm16), which are all markers for brown adipocytes, compared with WT mice (Fig. 3G).

Core body temperature was monitored after animals were maintained at 4°C for 6 h. In aP2-GDF5 mice, a significantly smaller decrease was seen in the core body temperature after cold exposure than in WT mice (Fig. 3H), whereas the baseline core temperature was not significantly altered between aP2-GDF5 and WT mice (data not shown).

Heterozygous GDF5^Rgsc451 mutant mice, characterized as dominant-negative GDF5 mutant (DN-GDF5) mice (23), were maintained for 6 months from the age of 8 weeks old on NC or the HFD. Their body weight, food intake, and metabolic efficiency were subsequently determined. Although no significant changes were found in body weight between DN-GDF5 and WT mice during 12 to 32 weeks after birth (Supplementary Fig. 3), the HFD was significantly more effective in increasing body weight in DN-GDF5 mice (53%) than in WT mice (34%) at the end of the HFD regimen compared with that in mice fed NC (Fig. 4A). Food intake was comparable between DN-GDF5 and WT mice on both NC and HFD feeding programs (Fig. 4B). Judging from the weight gain per gram of food consumed, however, metabolic efficiency was significantly higher in DN-GDF5 mice than in WT mice fed the HFD, but not NC (Fig. 4C).

We next examined physical activity and energy expenditure 7 days after the initiation of the HFD feeding program. No significant difference was seen in the spontaneous locomotor activity between DN-GDF5 and WT mice fed the HFD (Fig. 4D), whereas Vo₂ (Fig. 4E) and energy expenditure (Fig. 4F) were both significantly reduced in DN-GDF5 mice compared with WT mice fed the HFD, but not NC, when measured as per animal. DN-GDF5 mice exhibited significantly lower core body temperature than WT mice (Fig. 4G). Histological analysis revealed that larger lipid droplets were observed in the BAT of DN-GDF5 mice than WT mice on HFD programs (Fig. 4H), whereas no morphological changes were seen in the BAT of mice fed NC, irrespective of the animal genotype (Fig. 4I). Moreover, there was a significantly decreased expression of Ucp1, Ppargc1a, and Prdm16 in the BAT of DN-GDF5 mice compared with WT mice under obesogenic conditions (Fig. 4J), whereas Ucp1 expression in sWAT and vWAT was comparable between DN-GDF5 and WT mice (Fig. 4K).

To investigate whether GDF5 indeed promotes brown adipogenesis in vitro, SVF cells derived from BAT, sWAT, and vWAT were individually cultured with the induction cocktail for 2 days, followed by further culture for an additional 7 days in growth medium containing insulin and T3 in the presence or absence of GDF5 or an equal amount of PBS as control. Upregulation of Ucp1 by GDF5 was seen in SVF cells isolated from BAT and sWAT, but not in those from vWAT (Fig. 5A), with no significant changes in Pparg expression by GDF5 in SVF cells from all three adipose tissues tested (Fig. 5B). Moreover, brown preadipocyte clonal cells prepared from newborn WT mice (17) were cultured in the differentiation induction cocktail for 2 days, followed by further culture for an additional 7 days in growth medium containing insulin and T3 in the presence or absence of GDF5. The numbers of cells positive for Oil Red O were markedly increased to a similar extent irrespective of the presence of GDF5 under these experimental conditions, whereas GDF5 significantly increased expression of Ucp1 but not Pparg (Fig. 5C and D). However, expression of Ucp1 was not significantly altered in differentiated cells acutely treated with GDF5 for 6 h (Supplementary Fig. 4).

We next evaluated the possible mechanism underlying the promotion by GDF5 of brown adipogenesis. For this purpose, brown preadipocytes were cultured without the induction cocktail for determination of the expression of brown adipocyte markers. Brown preadipocytes were cultured with GDF5 in the presence or absence of different activin-like kinase (ALK) inhibitors. The inhibitor of ALK2/3, LDN193189, almost completely abrogated the GDF5-mediated expression of both Ucp1 and Ppargc1a, whereas SB431542, an inhibitor of ALK4/5/7, failed to significantly affect Ucp1 and Ppargc1a expression (Fig. 5E and F) at the concentrations used, suggesting that GDF5 could promote brown adipogenesis through a BMP receptor (BMPR) complex containing the BMP type I receptor (BMPRI) composed of ALK2 or ALK3 and the BMP type II receptor (BMPRII). Expression of several BMPRs, such as ALK2/3 (Acvr1 and Bmpr1a) and BMPRII (Bmpr2), however, was not significantly altered in any WATs from aP2-GDF5 and WT mice (Supplementary Fig. 5A–C).

Smad signaling is the major intracellular pathway of the BMP/GDF signal transduction scheme (31), whereas the transcriptional coactivator PGC-1α is a major regulator of oxidative metabolism and mitochondrial biogenesis and is a critical positive regulator of UCP1 expression (32). To further investigate how GDF5 regulates brown adipogenesis, Ppargc1a promoter activity was monitored in brown preadipocytes transfected with a variety of Smad expression vectors. Although Smad5 alone significantly increased Ppargc1a promoter activity in brown preadipocytes, cointroduction of Smad4 and Smad5 dramatically accelerated luciferase activity, with a less potent increase by the Smad4/Smad1 cointroduction (Fig. 6A). Serial deletion analysis revealed the importance of the region between 966 and 1563 bp upstream of the Ppargc1a transcriptional start site for the activation of the Ppargc1a promoter by Smad4/Smad5 (Fig. 6B), whereas ChIP assays showed the binding of Smad5 to the Ppargc1a promoter in brown preadipocytes overexpressing Smad5 (Fig. 6C). Significant upregulation in Ppargc1a expression was seen in brown preadipocytes transfected with Smad4/Smad5 (Fig. 6D). Although GDF5 markedly induced the phosphorylation of Smad5 in the nucleus of brown preadipocytes in addition to Smad1/5/8 (Fig. 6E), GDF5-induced phosphorylation of Smad5 was almost completely inhibited by the ALK2/3 inhibitor LDN193189 (Fig. 6F and G). We investigated whether Smad5 was involved in GDF5-induced Ppargc1a expression. Retroviral introduction of short hairpin Smad5 significantly prevented GDF5-induced Ppargc1a expression (Fig. 6H) and clearly inhibited Smad5 expression (Fig. 6I).

BMP7 has been shown to promote brown fat formation (17), which is mediated by the orchestration of a heterodimer with BMP2 or BMP4 rather than their individual homodimers with a lower efficiency (33,34). To clarify the possible relation between GDF5 and BMP7, we investigated whether GDF5 and BMP7 could interact with each other in the regulation of brown adipogenesis by using a luciferase reporter plasmid with BMP responsive element. Single transfection with the expression vector of BMP7 or GDF5 significantly increased luciferase activity, whereas cotransfection of both BMP7 and GDF5 did not synergistically or additively enhance the reporter activity (Supplementary Fig. 6).

Our findings demonstrated that transgenic mice overexpressing GDF5 in adipose tissues showed a lean phenotype and were less prone to HFD-induced obesity due to increased systemic energy expenditure. In addition, we observed an impairment of energy expenditure and thermogenesis in mutant mice harboring the dominant-negative form of GDF5 under obesogenic conditions. Moreover, GDF5 promoted brown adipogenesis through a mechanism related to the BMPR, Smad5, and PGC-1α pathway (Supplementary Fig. 7). To our knowledge, this is the first direct demonstration of the positive regulation of brown adipogenesis for systemic energy expenditure by GDF5 upregulated in response to obesity in BAT using in vitro and in vivo experimental techniques. The present findings that transgenic overexpression of GDF5 in adipose tissues facilitated the development of brown fat-like cells, often called brite or beige cells, together with the expression of UCP1 in sWAT, but not in vWAT, despite the equally increased ectopic GDF5 expression in sWAT and vWAT, give rise to an idea that leanness and beneficial glucose homeostasis are at least partly mediated by elevated systemic energy expenditure as a consequence of the conversion of sWAT into brown fat-like cells in aP2-GDF5 transgenic mice. In fact, the transformation of WAT into brown fat-like cells is shown to prominently occur in the inguinal subcutaneous adipose tissue and less frequently in the epididymal/perigonadal adipose tissue (35,36). However, given that UCP1 expression was also significantly increased in the BAT of aP2-GDF5 transgenic mice, the relative importance of WAT and BAT in increased whole-body energy expenditure in aP2-GDF5 mice is not clear.

The reason GDF5 upregulation failed to generate more BAT in mice under obesogenic conditions is not clear so far. One possible but unproven speculation is that GDF5 would be upregulated to at least in part participate in the compensatory promotion for BAT to preserve the functionality toward energy expenditure even in obese mice. This idea could be supported by our current observations that DN-GDF5 mice displayed lower energy expenditure and core temperature along with larger lipid droplets in BAT than WT mice under obesogenic conditions. The present findings that GDF5 significantly upregulated Ucp1 expression in SVF cells derived from sWAT and BAT and in brown preadipocytes treated with the differentiation induction cocktail give rise to an idea that GDF5 could directly promote brown adipogenesis from adipose progenitor cells rather than stimulate differentiated brown adipocytes in vitro. The possibility that GDF5 may accelerate brown adipogenesis through a mechanism irrelevant to adipose progenitors in vivo is, however, not ruled out. To prove the direct promotion by GDF5 of brown adipogenesis from adipose progenitors in vivo, for instance, transplantation studies using GDF5-treated white adipose progenitors are appreciable.

It should be noted that GDF5, at least partly, promoted brown adipogenesis through a mechanism associated with the Smad pathway. The BMP/GDF signaling pathway is shown to involve the transcription factors Smad1/5/8, which interact with the universal co-Smad, Smad4, to form heterodimers capable of recognizing the Smad binding element on upstream promoter regions of different target genes (31). The current results that Smad1/5/8-dependent transcriptional activity was not additively or synergistically enhanced by cointroduction of GDF5 and BMP7 do not argue in favor of an idea that GDF5 functionally interacts with BMP7 in terms of activation of the Smad pathway in brown adipocytes. In addition to this Smad pathway, BMP family members also use the p38 mitogen-activated protein kinase (MAPK) pathway to regulate a variety of biological activities (12). BMP7 accelerates a full program of brown adipogenesis through activation of p38 MAPK (17), while BMP8B centrally or peripherally enhances BAT thermogenesis by the activation of AMP-activated protein kinase or p38 MAPK (18). Because BMP7 (17) and BMP8B (18) can phosphorylate Smad1/5/8, the Smad/PGC-1α pathway might be partly responsible for a variety of functional alterations mediated by BMP7 and/or BMP8B in brown adipocytes in addition to p38 MAPK or AMP-activated protein kinase. A p38 inhibitor is shown to completely block the BMP7-induced UCP1 expression in brown preadipocytes (17), whereas several p38 inhibitors failed to significantly affect the upregulation of Ucp1 expression despite marked phosphorylation of p38 in brown preadipocytes exposed to GDF5 in our preliminary experiments (data not shown). These previous and present observations are thus unfavorable for an idea that the p38 MAPK pathway would play a pivotal role in the brown adipogenesis mediated by GDF5 in terms of UCP1 upregulation compared with BMP7-mediated brown adipogenesis. To further elucidate the role of GDF5 in relation to BMP7, accordingly, immunoprecipitation assays are appreciable for the direct demonstration of the heterodimerization between BMP7 and GDF5.

One of the interesting findings in this study is the lack of marked reciprocal abnormalities in heterozygous GDF5^Rgsc451 mutant mice maintained on NC in contrast to the phenotypes seen in aP2-GDF5 transgenic mice. The GDF5^Rgsc451 allele product is secreted normally for homodimerization but inhibits the function of native GDF5 proteins in a dominant-negative fashion. This allele carries an amino acid substitution (W408R) at the highly conserved residue in the BMPRI binding domain of the ligand GDF5 molecule (37). The current data could be accounted for by taking into consideration that less GDF5 expression was seen in the BAT of mice fed NC compared with those fed the HFD. Because brachypodism and ankylosis are both seen in DN-GDF5 (GDF5^Rgsc451) mice (23), the possibility that skeletal abnormalities could affect the metabolic phenotypes of DN-GDF5 mice under obesogenic conditions has not been ruled out. Although DN-GDF5 mice showed phenotypes opposite those seen in aP2-GDF5 transgenic mice in body weight gain and energy expenditure under obesogenic conditions, future analysis on phenotypes in mice with a conditional deletion of GDF5 in adipose tissues should be performed.

In addition to β-adrenergic agonists (38), several factors can trigger the development of brown-like adipocytes in WAT. Pharmacological activation of PPAR-γ by synthetic drugs induces the transformation of white adipocytes into brown adipocytes (39), whereas increased prostaglandin levels sensitize WAT to the stimulatory effects of catecholamines, leading to their conversion into brown adipocytes (40). Moreover, fibroblast growth factor-21 facilitates the conversion of white into brown fats (41). Nevertheless, compounds that can activate β-adrenergic signaling in BAT have not been successfully proved to be effective in treating metabolic syndromes in humans. This is probably due to inappropriate bioavailability and pharmacokinetics or by intolerable side effects resulting from overactivation of β-adrenergic receptors in other tissues (42). From this point of view, manipulation of the GDF5/BMPR/Smad/PGC-1α pathway is a plausible strategy to promote the BAT activity required for protection against obesity and a variety of obesity-related metabolic diseases, with a risk of joint and skeletal malformation in childhood. Because GDF5 is initially synthesized as a monomeric proform that undergoes post-translational processing to produce a dimeric mature form for exocytotic release (23), promotion of GDF5 activity would be an alternative approach for the discovery and development of novel strategies beneficial for activating BAT functions to combat against obesity in human beings.

Acknowledgments. The authors are grateful to Drs. C. Ronald Kahn (Joslin Diabetes Center, Boston, MA), Bruce Spiegelman (Dana-Farber Cancer Institute, Boston, MA), Shiro Ikegawa (RIKEN, Yokohama, Japan), and Toshio Kitamura (Tokyo University, Tokyo, Japan) for generously providing brown preadipocyte cell lines, the 5.4-kb mouse aP2 promoter construct, GDF5^Rgsc451 mice, and PLAT-E cells, respectively. The authors are also grateful to Dr. Takeshi Sakurai (Kanazawa University, Kanazawa, Japan) for his kind guidance in measuring energy expenditure.

Funding. This work was supported in part by grants-in-aids for Scientific Research to E.H. from the Ministry of Education, Culture, Sports, Science and Technology, Japan, and in part by research grants to E.H. from the ONO Medical Research Foundation, the Uehara Memorial Foundation, and the Mochida Memorial Foundation for Medical and Pharmaceutical Research, Japan.

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

Author Contributions. E.H., Y.N., S.Takad., H.F., T.I., S.H., S.Takah., Y.O., and T.W. researched data. Y.Y. contributed to discussion and wrote the manuscript. Y.Y. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.