The activin-βA/BMP-2 chimera AB204 is a strongstimulator of adipogenesisMeejung Kim1, Jong In Kim2, Jae Bum Kim2 and Senyon Choe1,3*1Joint Centre for Biosciences, Lee Gil Ya Cancer and Diabetes Research Institute, Gachon University of Medicine and Science, Incheon, Korea2National Creative Research Initiatives Centre for Adipose Tissue Remodelling, Department of Biological Sciences, Institute of MolecularBiology and Genetics, Seoul National University, Republic of Korea3Qualcomm Institute, University of California at San Diego, La Jolla, CA, USA

Abstract

Several of the bone morphogenetic proteins (BMPs) have been reported to induce white as well asbrown adipogenesis. Here, we characterized the adipogenic potential of AB204, a recombinant chi-meric protein of activin-βA and BMP-2, in in vitro, ex vivo and in vivo settings. BMP-2 is generallyknown to promote adipogenesis. When compared with BMP-2, which previously showed varying de-grees of adipogenesis, AB204 displayed superior in vitro adipogenic differentiation of mouse 3T3-L1pre-adipocytes and human adipose-derived stem cells (hASCs). Surprisingly, implantation of hASCs,preconditioned with AB204 for as short a time as 48h, into the subcutaneous space of athymic nudemice effectively produced fat pads, but not with BMP-2. When BMP-2 and AB204 were injected intra-peritoneally, AB204 promoted dramatic systemic adipogenesis of C57BL/6 mice on a high-fat dietvery effectively. The results implicate the novel clinical potential of AB204, including induction offat tissue ex vivo or in vivo for tissue re-engineering and regenerative medicinal purposes, more thanany known natural protein ligand. Copyright © 2015 John Wiley & Sons, Ltd.

Received 7 October 2014; Revised 28 April 2015; Accepted 29 April 2015

Keywords BMP; TGFβ; adipogenesis; adipose-derived stem cells

1. Background

Regulation of metabolic homeostasis is critical for organ-isms to maintain good health. Adipose tissue is a centralplayer in systemic energymetabolism,which serves as a res-ervoir to store excess energy, in the form of triglycerides,and to secrete various hormones and cytokines as anendocrine hotspot. On the other hand, another type of adi-pocytes called brown adipocytes functions in energy expen-diture by dissipating chemical energy as heat, providing amechanism of thermogenesis against hypothermia and obe-sity (Rosenwald and Wolfrum, 2014). Adipose tissues areknown to possess distinct genetic profiles and play variousmetabolic roles (Gesta et al., 2006; Kusminski et al., 2012;Sun et al., 2011). Although obesity is associated with an in-creased risk of metabolic disease, expansion of white

adipose tissue, even in obese individuals, has been sug-gested to improve insulin sensitivity in humans and rodents(Kim et al., 2007; Kusminski et al., 2012; Porter et al.,2009). Furthermore, peroxisome proliferator-activatedreceptor-γ (PPARγ), a ligand-activated transcription factorrequired for adipocyte differentiation and maintenance,has been shown to be crucial in glucose homeostasis (Imaiet al., 2004; Tontonoz and Spiegelman, 2008). Adipose-specific PPARγ knockout mice develop systemic insulin re-sistance, while activation of PPARγ improves whole-body in-sulin sensitivity (He et al., 2003; Sugii et al., 2009).

Bone morphogenetic proteins (BMPs) represent thelargest subset within the transforming growth factor-β(TGFβ) superfamily, which mediates a multitude of devel-opmental processes (Massague et al., 2005; Yoshida et al.,2013). We previously reported that structural similaritiesbetween members of the TGFβ superfamily can beexploited to create synthetic ligands (Synbiologics®)with modified activities (Allendorph et al., 2011). For in-stance, the AB2 chimera library consists of chimeras ofBMP-2 and Activin-βA ligands by ligating six sequence-

*Correspondence to: Senyon Choe, Qualcomm Institute, Univer-sity of California at San Diego, La Jolla, CA 92093, USA.E-mail: schoe@ucsd.edu

Copyright © 2015 John Wiley & Sons, Ltd.

JOURNAL OF TISSUE ENGINEERING AND REGENERATIVE MEDICINE RESEARCH ARTICLEJ Tissue Eng Regen Med 2015.Published online in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/term.2050

equivalent segments of BMP-2 and Activin-βA in all possi-ble sequential combinations. Briefly, AB204 consists ofthree Activin-βA segments and three BMP-2 segments inthe sequence order BABBAA, where A and B representActivin-βA and BMP-2, respectively.

Recently, there have been major advances in under-standing the role of different members of the BMP familyligands in control of adipocyte differentiation as well asenergy metabolism (James, 2013; Kim and Choe,2011; Kuo et al., 2014; Liu et al., 2014b; Mandrupet al., 1997; Rosen and MacDougald, 2006; Tang et al.,2004; Tseng et al., 2008; Whittle et al., 2012). We findthat AB204 displays superior in vitro adipogenic poten-tial, resulting in part from its insensitivity to Noggin,an inhibitor for BMP ligands. In addition, implantationof hASCs pretreated with AB204, but not with BMP-2,into the subcutaneous space produced fat pads very ef-fectively. Finally, intraperitoneal injection of BMP-2 orAB204 induces adipogenesis in vivo. These resultsimplicate the novel clinical potential of AB204 for thepurposes of tissue engineering and inducing fat tissuein regenerative medicine.

2. Methods

2.1. Materials

Recombinant human BMP-2, BMP-7, AB204 and Nogginwere obtained from Joint Protein Central (http://jointproteincentral.com). Recombinant human Activin-βAand BMP-4 were purchased from R&D Systems.

2.2. Cell culture

3T3-L1 cells weremaintained and induced to differentiate aspreviously described (Huang et al., 2009; Tang et al., 2004).The hASCs used for the results shownwere passages 5–12from one hASC cell line from one donor, purchased fromInvitrogen (STEMPRO® human adipose-derived stemcells), as previously used for the studies described byHwang et al. (2011). hASCs (Invitrogen) were main-tained in MesenPro RS Medium (Invitrogen). The cellswere induced to differentiate as previously described(Seale et al., 2011); they were either treated with ligandsthroughout the whole differentiation process orpreconditioned by adding ligands before differentiation.The ligands were treated with 10 and 50 ng/ml forFigure 1A; 50 ng/ml for Figures 1B, 2A, B, D and E, 3Aand C; 10, 50, 100 and 200 ng/ml for Figure 2C; and 50and 100ng/ml for Figure 3B. All cells used in this studywere cultured at 37 °C in 5% CO2 environment.

2.3. Oil red O staining

Dishes were washed with phosphate-buffered saline(PBS) and fixed with 10% buffered formalin. Cells were

then stained with filtered oil red O solution (0.5% oilred O in propylene glycol), washed twice with distilledwater and visualized.

2.4. Quantitative real-time PCR (qRT–PCR)analysis

Total RNAwas isolated with Trizol (Invitrogen) or TRIsure(Bioline), following the manufacturers’ instructions. Com-plementary DNA (cDNA) was prepared from 1mg RNAusing a QuantiTect Reverse Transcription Kit (Qiagen)and PCR was run with SYBR Premix Ex TaqII, ROX plus(TAKARA) in duplicate for each sample and quantified in7900HT Fast Real-Time PCR System (Applied Biosystems)or CFX96 (BioRad) (for primer sequences, see supplemen-tary material, Table S1). Data were expressed as arbitraryunits after normalization to levels of expression of the in-ternal controls cyclophillin (mouse) and 18S (human).

2.5. Western blot analysis

Cells were collected in cell lysis buffer (Cell Signaling)and clarified by centrifugation at 12 000× g for 20minat 4 °C. An equal amount of protein in the supernatant,as determined by a Bicinchoninic Acid (BCA) Protein As-say Kit (Sigma-Aldrich), was separated by sodium dodecylsulphate–polyacrylamide gel electrophoresis (SDS–PAGE)and transferred to nitrocellulosemembranes (GEHealthcare).The membranes were blocked, incubated with antibody for2h at room temperature and analysed with peroxidase-coupled secondary antibody and chemiluminescence(Animal Genetics Inc.). Antibodies used for immunoblot-ting were anti-β-actin (Sigma), anti-p-Smad1/5/8, anti-aP2, anti-PPARγ and anti-p-C/EBPb (Cell Signaling).

2.6. Animal experiments

All animal experiments were performed according to pro-cedures approved by the Institutional Animal Care andUse Committee (IACUC) of the Centre of Animal Careand Use (CACU) at Lee Gil Ya Cancer and Diabetes Insti-tute of Gachon University. The mice were maintained onstandard rodent chow or 60% high-fat diet (ResearchDiets) with 12h light and dark cycles; 6week-old male an-imals were used for the experiments. Crl:NU/NU-nuathymic nude mice were purchased from Orient Bio Inc.(Kyeonggi, Korea); these mice were divided into fourgroups, three mice/group: (a) injected with medium(vehicle only); (b) injected with 3×105 hASCs that werenot pretreated (untreated); (c) 3×105 hASCs pretreatedwith 50ng/ml BMP-2; (d) 3×105 hASCs pretreated with50ng/ml AB204. The cells were grown to near-confluence,with or without BMP-2 or AB204 (50ng/ml), trypsinized,suspended in medium, centrifuged into pellets, resus-pended and injected, as in Tang et al. (2004).

M. Kim et al.

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term

C57BL/6Jms mice were purchased from Central Lab.Animal Inc. (Seoul, Korea). When these mice reached11weeks of age, they were divided into three groups, 10mice/group: (a) injected with 1mM HCl as vehicle; (b)injected with 0.5μg BMP-2 dissolved in 1mM HCl; (c)injected with 0.5μg AB204 dissolved in 1mM HCl. Ligandswere injected without cells directly into the intraperito-neal space at the lower left abdomen three times/weekfor 10weeks. Before sacrifice, the fat mass ratio was deter-mined, sing a mq7.5 NMR Analyser (Bruker Optics).

2.7. Histological analysis

Fat pads from injected cells and epididymal fat were re-moved and the specimens were fixed in 4% paraformalde-hyde and embedded in paraffin. Serial sections (10mm)were mounted on slides and stained with haematoxylinand eosin (H&E).

2.8. Statistics

Statistical significance was assessed by two-tailed Stu-dent’s t-test; p<0.05 was considered significant.

3. Results

3.1. AB204 is a superior adipogenic inducer toBMP-2

Mouse pre-adipocytes (3T3-L1) were treated with10ng/ml BMP-2 or AB204 during adipogenic differentia-tion, with 1mM rosiglitazone as a positive control. AB204differentiated the cells into lipid-laden adipocytes to agreater degree than BMP-2-treated cells, and in a dosage-dependent manner judging from oil red O staining(Figure 1A). Likewise, expression of an array of adipogenic

Figure 1. BMP-2 and AB204 enhance adipogenic differentiation of mouse pre-adipocytes. (A) Oil red O staining of cells treated withvehicle (control), 1mM rosiglitazone, BMP-2 or AB204 (10 and 50 ng/ml) during adipogenic induction for 6 days. (B) qRT–PCR anal-ysis of cells treated with vehicle (control), BMP-2 or AB204 (50 ng/ml) during adipogenic induction for 6 days: PPARγ, peroxisomeproliferator-activated receptor-γ; SREBP1c, sterol regulatory element binding protein 1c; aP2, adipocyte protein 2; PPARα, peroxisomeproliferator-activated receptor-α; FAS, fatty acid synthase; GLUT4, glucose transporter type 4; ATGL, adipose triglyceride lipase; CD36,cluster of differentiation 36; Pref-1, pre-adipocyte factor 1; Ctl, control; data are presented as mean±SD (n=3); *statistically signif-icant differences vs control (*p< 0.05, **p< 0.005, ***p< 0.0005); #statistically significant differences vs BMP-2 (#p< 0.05,##p< 0.005, ###p< 0.0005)

Activin-βA/BMP-2 chimera AB204 stimulates adipogenesis

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term

markerswasmarkedly enhanced in cells treatedwith AB204in quantitative RT–PCR (qRT–PCR) analysis (Figure 1B; seealso supplementary material, Figure S1), while expressionof an adipogenic inhibitor, pre-adipocyte factor 1 (Pref-1),was inhibited with treatment of BMP-2 and AB204.

AB204 also enhanced adipogenic differentiation of hASCsto a greater degree than natural BMPs (BMP-2, -4, -7)and even the positive control treated with rosiglitazone(Figure 2A). BMP downstream signalling pathwayp-Smad1/5/8 as well as various adipogenic and lipogenicmarkers were significantly enhanced in cells treated withBMP-2 and AB204 throughout their differentiation, asanalysed by qRT–PCR and western blot (Figure 2B, C).Consistent with results using 3T3-L1 cells, AB204 showeda higher degree of adipogenic capacity in hASCs thanBMP-2. The adipogenic capability of AB204, unlike BMP-2,displayed no appreciable sensitivity to a BMP antagonist,Noggin (Figure 2D, E).

3.2. Preconditioning of AB204 for 2days issufficient to drive ex vivo differentiation

To address the enhanced adipogenic potency of AB204, weinvestigated whether constant presence of the ligandwas re-quired for adipogenic differentiation. BMP-2 and BMP-4

have been previously demonstrated to induce commitmentof mouse pluripotent stem cells into pre-adipocytes bytreating only before adipogenic induction (Tang and Lane,2012; Tang et al., 2004). If the effects of AB204 also resultedfrom stimulating the stem cells to commit to the adipogeniclineage, pretreatment of hASCs before any induction wouldbe sufficient to drive enhanced adipogenesis. In our experi-ments, hASCs were treated with 50ng/ml BMP-2 andAB204 for 2days during the cell proliferation stage, beforeadipogenic induction. Cells were then transferred to thedifferentiation-inducing medium. This preconditioning ofhASCs with AB204 for just 2days resulted in markedlyboosted differentiation compared to BMP-2 and BMP-4 aswell as BMP-7 (Figure 3A). Pretreatment with AB204 trig-gered adipogenesis of otherwise virtually non-adipogenichASCs to a similar degree as treatingwith BMP-2 throughoutthe entire differentiation process (see supplementary mate-rial, Figure S2). Preconditioning of AB204 also increasedadipogenic markers such as PPARγ and CCAAT/enhancerbinding protein-β (C/EBPb) in addition to phosphorylated-Smad1/5/8 in a dose-dependent manner (Figure 3B).

To discover whether the preconditioning of hASCs withAB204 in vitro was sufficient to commit hASCs to theadipogenic lineage in vivo, hASCs pretreated with 50ng/mlof the ligands were injected, without scaffold, into the sub-cutaneous space adjacent to the sternum of athymic nude

Figure 2. BMP-2 and AB204 stimulate adipogenic differentiation of hASCs. (A) Oil red O staining of hASCs treated with vehicle (con-trol), 2mM rosiglitazone, BMP-2, BMP-4, BMP-7 or AB204 (50 ng/ml) during adipogenic induction for 6 days. (B) Time-dependentqRT–PCR analysis of hASCs treated with vehicle (control), BMP-2 or AB204 (50 ng/ml) from 2 days of culture before induction andduring adipogenic induction for 6 days: data are presented as mean±SD; *statistically significant differences vs control (*p< 0.05,**p< 0.005, ***p< 0.0005); #statistically significant differences vs BMP-2 (#p< 0.05, ##p< 0.005, ###p< 0.0005). (C) Western blotof hASCs treated with vehicle (control) or BMP-2 and AB204 (10, 50, 100 and 200 ng/ml) during adipogenic induction for 6 days.(D) Oil red O staining of hASCs treated with vehicle (control), BMP-2 or AB204 (50 ng/ml) in addition to Noggin (1mg/ml) duringadipogenic induction for 6 days. (E) qRT–PCR analysis of hASCs treated with BMP-2 or AB204 (50 ng/ml) in addition to Noggin(at the indicated concentrations) during adipogenic induction for 6 days: data are presented as mean±SD; ID-1, inhibitor of DNAbinding 1

M. Kim et al.

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term

mice, a site normally devoid of adipose tissue (Mandrupet al., 1997; Tang et al., 2004). hASCs preconditionedex vivo by AB204 rapidly differentiated into mature adipo-cytes after implantation (Figure 3C). Mice transplantedwith just hASCs had an outwardly visible pad, but underhistological examination this was found to be a mass ofundifferentiated cells. hASCs preconditioned by AB204became vascularized and differentiated into mature adipo-cytes indistinguishable from endogenous epididymaladipose tissue of the same animal. In contrast to resultsfrom in vitro experiments, hASCs preconditioned byBMP-2 failed to develop into mature adipocytes by2weeks.

Figure 3D shows the gross appearance of the sternumafter 4weeks. The fat pad is visible as a lump on theoutside, as well as a distinct lump of cells attached tothe inside of the skin when dissected, in contrast to noth-ing under the skin for mice injected with medium alone.Although we do not have an exact explanation for whatkeeps the differentiated adipocytes in the subcutaneousspace without dispersing, we observed that undifferenti-ated cells dispersed with time, while a differentiated fatpad remained intact, with formation of new blood vesselsvisible around the pad by 4weeks. Since the fat pad is

formed from differentiated hASCs injected into a sitedevoid of endogenous adipose tissue, the origin of thefat pad is believed to be hASCs, but this hypothesisrequires further experimental verification.

3.3. AB204 stimulates adipose tissue expansion

When BMP-2 and AB204were directly injected intraperito-neally into high-fat diet (HFD)-induced obese C57BL/6mice, the mice grew heavier. They showed a higher totalfat mass while maintaining their lean mass, exhibiting anincreased fat mass ratio (Figure 4A, B). These physiologicalchanges were not induced by the same ligands in C57BL/6mice fed with normal chow diet (data not shown). Bothsubcutaneous and mesenteric fat depots expanded due totreatment with AB204 more effectively than with BMP-2,whereas epididymal fat expansion was equally effectiveby both ligands (Figure 4C–F). Consistent with increasedfat mass, epididymal and subcutaneous expression ofadipogenic markers, such as PPARγ, adipocyte protein 2(aP2) and Adiponectin, were also increased in both theBMP-2- and AB204-injected groups (Figure 4G, H).

Figure 3. Preconditioning of AB204 for 48 h is sufficient to drive ex vivo differentiation of hASCs into mature adipocytes. (A) Oil red Ostaining of hASCs pretreated with vehicle (control), BMP-2, -4, -7 or AB204 (50 ng/ml) for the last 2 days of proliferation, then in-duced to differentiate for 6 days in the absence of any ligands. (B) Western blot of hASCs pretreated with vehicle (control), BMP-2or AB204 (50 and 100 ng/ml) for 2 days during proliferation before adipogenic induction without any ligands for 6 days. (C) H&Estaining of hASCs pretreated with BMP-2 or AB204 (50 ng/ml) for 2 days during proliferation and transplanted into the subcutaneousspace of athymic nude mice: the mice were injected with medium (vehicle only), hASCs that were not pretreated (untreated), BMP-2-pretreated hASCs or AB204-pretreated hASCs; 2weeks post-injection, the mice were sacrificed and the injected areas were examined;the skin above the injected site and any masses of cells underneath were fixed and stained (magnification=×100 and×200); epidid-ymal fat tissue was used as a positive control for endogenous fat (magnification=×100); the images are representative of three mice/group. (D) Gross appearance of the sternum injected with vehicle (control), and cells pretreated with AB204: the images are repre-sentative of four mice/group; CEBPb, CCAAT/enhancer binding protein β

Activin-βA/BMP-2 chimera AB204 stimulates adipogenesis

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term

4. Discussion

We report here that BMP-2 and AB204, a chimeric proteinof BMP-2 and Activin-βA, possess adipogenic potential.In particular, AB204 effectively stimulates mouse pre-adipocytes as well as human adipose-derived mesenchy-mal stem cells (hASCs) to differentiate into adipocytes,even by preconditioning for only 48h prior to adipogenicinduction. Such unprecedented adipogenic potency ofAB204 may be in part due to amplified Smad1/5/8 signal-ling, as well as its reduced sensitivity to endogenous BMPinhibitors such as Noggin (Allendorph et al., 2011). It isunknown whether AB204 triggers alternative pathwaysdownstream of Smad signalling to enhance adipogenesis.

The effects of BMP-2 on adipogenesis have not beenrigorously studied. Its effect has been reported as beingvariable, depending on concentration, type of cell andpresence of other signalling molecules (Liu et al., 2014a;Rosen and MacDougald, 2006), making it difficult to eval-uate the adipogenic potential of BMP-2 alone. However,the effects of BMP-2 and AB204 were significantly consis-tent in our study with both 3T3-L1 cells and hASCs. Ournext goal is to replicate this investigation using primarycells from different donors and expand our data poolusing a variety of adipose depots from a wider range ofdonor age and BMI.

A technical difficulty in evaluating the role of BMP sig-nalling in vivo is that mice lacking major BMP genes die inearly embryogenesis, before any adipocytes develop(Rosen and MacDougald, 2006). Hence, we employedtwo other approaches to investigate the adipogenic effectsof BMP-2 and AB204. First, we employed an ex vivo celltransplantation approach to achieve full adipocyte differ-entiation in vivo. Transplanted cells differentiated into

mature adipocytes under appropriate conditions, compa-rable to endogenous adipose tissue (Mandrup et al.,1997). Second, we injected the ligands directly into miceto determine their systemic effects under physiological cir-cumstances. This approach allowed examination of the li-gands’ long-term effects on various endogenous adiposedepots.

The results of the transplantation experiment demon-strated that mere preconditioning of the cells with AB204in vitro is sufficient to drive adipogenic differentiation.Since this procedural difference may be critically instru-mental in tissue re-engineering through cell transplanta-tion, we re-emphasize that the presence of the ligand isnot a requirement for inducing adipogenesis post-cell trans-plantation. We did not try to determine how the ligand suc-cessfully triggers commitment at this stage, except thatAB204 seems to induce a stronger commitment signal thanBMP-2 and BMP-4. Cells differentiated into mature adipo-cytes in vivo indistinguishable from those of endogenousadipose tissue. The ex vivo conditioning of hASCs could beparticularly instrumental for clinical applications becausedifferentiated adipocytes are technically cumbersome tohandle. Further, ex vivo treatment eliminates any immuno-genic issues for clinical applications of Synbiologics, ashASCs are washed prior to transplantation to remove anyresidual ligands or non-human materials, such as fetalbovine serum, that is used in preconditioning media. Weenvision that Synbiologics could become novel ex vivo orin vivo components for the ex vivo cell conditioning neces-sary for tissue re-engineering and regenerative medicine.

Intraperitoneal injection of AB204, and to a lesser extentof BMP-2, into HFD-fed mice revealed that the ligands areactually highly adipogenic in vivo. Overall fat mass, includ-ing subcutaneous, mesenteric and epididymal adipose tis-sues, expanded, indicating that AB204 and BMP-2 are

Figure 4. BMP-2 and AB204 stimulate adipose tissue expansion. C57BL/6 mice were intraperitoneally injected three times/week withvehicle (V), BMP-2 (B2) or AB204 (AB) (0.5mg/injection) for 10weeks after 5weeks of high-fat diet (n=10 mice/group). (A) Bodyweight gain was determined by the difference in weight before injection and 10weeks after. (B) Fat mass ratio was determined atthe end of 10weeks, using an NMR Analyser; after sacrifice of the animals, subcutaneous (C), mesenteric (D) and epididymal (E)fat pads were weighed. (F) Gross image of sacrificed animals, dissected to reveal epididymal adipose tissue; a representative exampleis shown. (G, H) qRT–PCR analysis of PPARγ, aP2 and Adiponectin from epididymal (G) and subcutaneous (H) adipose tissue of thesame animals described above; data are presented as mean±SE (A–E, G, H); n ≥ 7

M. Kim et al.

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term

potent stimulators of adipogenesis that act in a systemicmanner. Based on the results of induced stem cell differen-tiation to adipocytes and expansion of endogenous adiposetissue, we speculate that AB204, and to a lesser extentBMP-2, played a crucial role in stimulating the formationof new adipocytes, thereby increasing the metabolic capac-ity of the essential adipose tissues and the induction ofPPARγ and other functional regulators of metabolic homeo-stasis. Increased expression of these adipogenic markergenes in adipose tissue signifies enhanced adipogenic char-acteristics of the tissue in terms of its potential for novel ad-ipogenesis. Therefore, we can speculate that BMP-2 andAB204 ligands have caused these adipose tissues to obtainsuperior self-sustainability by forming new adipocytesthrough hyperplasia, more than increasing its size throughhypertrophy. It has been discovered that BMP-2 can di-rectly induce PPARγ expression through Smad signalling.Although both aP2 and Adiponectin are target genes ofthe transcription factor PPARγ, there aremany other factorsthat may influence the expression of these genes directly orindirectly (Tan et al., 2012). Also, post-translational regula-tion of PPARγ through various downstream BMP signallingpathways should be taken into consideration (James,2013). Therefore, we cannot ascertain that all PPARγ targetgenes will be controlled in a predictable pattern, especiallyin such a complicated in vivo setting.

It should be noted that AB204 actually consists of threeActivin-βA segments and three BMP-2 segments in the se-quence. Throughout this study, however, we used onlyBMP-2 to compare to AB204, because upon the initialscreening test, AB204 acted as an adipogenic inducer, likeBMP-2, rather than as an inhibitor, like Activin-βA (seesupplementary material, Figure S3). This was partly antici-pated, since AB204 was already known to possess strongersignalling activity thanBMP-2 in previous studies (Allendorphet al., 2011; Yoon et al., 2014), while Activin-βA is known toinhibit adipogenesis. We cannot specifically define themechanistic reasoning behind this behaviour, but AB204not only displayed almost no sensitivity to Noggin, a natu-ral antagonist for BMP-2, but also exhibited higher affinity(KD=0.381nM) to Activin receptor II (ActRII) than BMP-2(KD=36.8nM), closer to Activin-βA (KD=0.203nM), andeight-fold reduced affinity (KD=16.4nM) to BMP receptorIa (BMPRIa) than BMP-2 (KD=2.09nM) in the previousstudy. We conclude that enhanced SMAD-mediated signal-ling ability of AB204 does not result simply from theenhanced binding affinity of receptor pairs utilized in

nature by BMP-2, but likely from a mix of signalling path-ways through different pairs of receptors.

Moreover, AB204 has been shown to induce both oste-ogenesis and adipogenesis in vitro as well as in vivo. Thismulti-function potential is characteristic of many otherBMPs, including BMP-2, BMP-4, BMP-7 and BMP-9. BMPshave many functions throughout the entire body apartfrom osteogenesis, which include, but are not limited to,embryonic patterning and development, tissue homeosta-sis and regeneration. As hallmark feature of BMPs, thefinal outcome of their signal transduction is highly depen-dent on spatiotemporal circumstances, in addition to theendocrine properties of those that are secreted into circu-lation, displaying both pro-osteogenic and pro-adipogeniceffects (Kim and Choe, 2011; James, 2013). Therefore, inand around bone tissue, AB204 stimulates osteogenesis(Yoon et al., 2014), whereas in an adipogenic environmentAB204 is able to induce adipogenesis, as indicated by thisstudy. Treatment of AB204 or BMP-2 to hASCs was notable to induce any kind of differentiation in hASC culturemedium, without the addition of either adipogenic orosteogenic differentiation cocktail. It is not well under-stood precisely how BMP signalling controls adipogenesisvs osteogenesis, except that it may depend on the dosageof ligand, ligand–receptor interaction and the type of cell(James, 2013).

Conflict of interest

The authors have declared that there is no conflict ofinterest.

Acknowledgements

We thank Su Bin Kim and Jee Hyung Sohn for their excellenttechnical assistance. We also thank CACU of Gachon Universityfor their assistance in making tissue sections and H&E staining,and Su Sung Kim for conducting NMR analysis. We acknowledgeProfessor Hee Sook Jun for the gift of hASCs. This work was sup-ported by the World Class University Programme (Grant NosR31-100320 and R32-10215), (Incheon Free Economy Zone,joint Center for Biosciences), the National Research Foundation(NRF), Korea (Grant No. 2011-0029948) and the NationalCreative Research Initiative Programme of Korea (Grant No.2012-0001241).

References

Allendorph GP, Read JD, Kawakami Y et al.2011; Designer TGFbeta superfamily li-gands with diversified functionality. PLoSOne 6: e26402.

Gesta S, Bluher M, Yamamoto Y et al. 2006; Ev-idence for a role of developmental genes inthe origin of obesity and body fat distribution.Proc Natl Acad Sci U S A 103: 6676–6681.

He W, Barak Y, Hevener A et al. 2003;Adipose-specific peroxisome proliferator-

activated receptor-γ knockout causes insu-lin resistance in fat and liver but not inmuscle. Proc Natl Acad Sci U S A 100:15712–15717.

Huang H, Song TJ, Li X et al. 2009; BMPsignaling pathway is required forcommitment of C3H10T1/2 pluripotentstem cells to the adipocyte lineage.Proc Natl Acad Sci U S A 106: 12670–12675.

Hwang OK, Park EY, Mai LD et al. 2011; Insu-lin-producing cells differentiated from hu-man adipose tissue-derived stem cellsameliorate hyperglycemia in diabetic mice.J Tissue Eng Regen Med 8: 482–488.

Imai T, Takakuwa R, Marchand S et al. 2004;Peroxisome proliferator-activated receptor-γis required in mature white and brown adi-pocytes for their survival in the mouse. ProcNatl Acad Sci U S A 101: 4543–4547.

Activin-βA/BMP-2 chimera AB204 stimulates adipogenesis

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term

James AW 2013; Review of signaling pathwaysgoverningMSCosteogenic and adipogenic dif-ferentiation. Scientifica (Cairo) 2013: 684736.

Kim JY, van de Wall E, Laplante M et al. 2007;Obesity-associated improvements in meta-bolic profile through expansion of adiposetissue. J Clin Invest 117: 2621–2637.

Kim M, Choe S 2011; BMPs and their clinicalpotentials. BMB Rep 44: 619–634.

Kuo MM, Kim S, Tseng CY et al. 2014; BMP-9as a potent brown adipogenic inducer withanti-obesity capacity. Biomaterials 35:3172–3179.

Kusminski CM, Holland WL, Sun K et al.2012; MitoNEET-driven alterations inadipocyte mitochondrial activity reveal acrucial adaptive process that preserves in-sulin sensitivity in obesity. Nat Med 18:1539–1549.

Liu J, Chen L, Zhou Y et al. 2014a; Insulin-likegrowth factor-1 and bone morphogeneticprotein-2 jointly mediate prostaglandinE2-induced adipogenic differentiation ofrat tendon stem cells. PLoS One 9: e85469.

Liu Y, Liu Y, Zhang R et al. 2014b; All-trans-retinoic acid modulates bone morphogenicprotein 9-induced osteogenesis andadipogenesis of preadipocytes throughBMP/Smad and Wnt/β-catenin signalingpathways. Int J Biochem Cell Biol 47:47–56.

Mandrup S, Loftus TM, MacDougald OA,et al. 1997; Obese gene expression atin vivo levels by fat pads derived from s.c.implanted 3 T3-F442A preadipocytes. ProcNatl Acad Sci U S A 94: 4300–4305.

Massague J, Seoane J, Wotton D 2005;Smad transcription factors. Genes Dev 19:2783–2810.

Porter SA, Massaro JM, Hoffmann U et al.2009; Abdominal subcutaneous adiposetissue: a protective fat depot? DiabetesCare 32: 1068–1075.

Rosen ED, MacDougald OA 2006; Adipocytedifferentiation from the inside out. NatRev Mol Cell Biol 7: 885–896.

Rosenwald M, Wolfrum C 2014; The originand definition of brite versus white andclassical brown adipocytes. Adipocyte 3:4–9.

Seale P, Conroe HM, Estall J et al. 2011;Prdm16 determines the thermogenic pro-gram of subcutaneous white adipose tissuein mice. J Clin Invest 121: 96–105.

Sugii S, Olson P, Sears DD, et al. 2009;PPARγ activation in adipocytes is suffi-cient for systemic insulin sensitization.Proc Natl Acad Sci U S A 106: 22504–22509.

Sun K, Kusminski CM, Scherer PE 2011; Adi-pose tissue remodeling and obesity. J ClinInvest 121: 2094–2101.

Tan CK, Chong HC, Tan EHP et al. 2012; Get-ting ‘Smad’ about obesity and diabetes.Nutr Diabetes 2: e29–35.

Tang QQ, Lane MD 2012; Adipogenesis: fromstem cell to adipocyte. Annu Rev Biochem81: 715–736.

Tang QQ, Otto TC, Lane MD 2004; Commit-ment of C3H10T1/2 pluripotent stem cellsto the adipocyte lineage. Proc Natl Acad SciU S A 101: 9607–9611.

Tontonoz P, Spiegelman BM 2008; Fat andbeyond: the diverse biology of PPARγ.Annu Rev Biochem 77: 289–312.

Tseng YH, Kokkotou E, Schulz TJ et al. 2008;New role of bone morphogenetic protein 7in brown adipogenesis and energy expen-diture. Nature 454: 1000–1004.

Whittle AJ, Carobbio S, Martins L et al. 2012;BMP8B increases brown adipose tissuethermogenesis through both central andperipheral actions. Cell 149: 871–885.

Yoon BH, Esquivies L, Ahn C et al. 2014; Anactivin A/BMP2 chimera, AB204, displaysbone-healing properties superior to thoseof BMP2. J Bone Miner Res 29: 1950–1959.

Yoshida H, Kanamori Y, Asano H et al. 2013;Regulation of brown adipogenesis by theTGFβ family: involvement of Srebp1c inTGFβ- and Activin-induced inhibition ofadipogenesis. Biochim Biophys Acta 1830:5027–5035.

Supporting informationThe following supporting information may be found in the online version of this article:

Figure S1. BMP-2 and AB204 enhance adipogenic differentiation of mouse pre-adipocytesFigure S2. Preconditioning of AB204 for 48 h triggers hASCs to commit to the adipogenic lineage to a degree similar totreating BMP-2 throughout the differentiation processFigure S3. Activin-βA inhibits adipogenic differentiation in comparison to vehicle (control), BMP-2 and AB204Table S1. qRT–PCR primer sequences

M. Kim et al.

Copyright © 2015 John Wiley & Sons, Ltd. J Tissue Eng Regen Med 2015.DOI: 10.1002/term