Birgit Gustafson,1,2 Ann Hammarstedt,1,2 Shahram Hedjazifar,1,2 Jenny M. Hoffmann,1,2

Per-Arne Svensson,2 Joseph Grimsby,3 Cristina Rondinone,3 and Ulf Smith1,2

BMP4 and BMP Antagonists RegulateHuman White and Beige AdipogenesisDiabetes 2015;64:1670–1681 | DOI: 10.2337/db14-1127

The limited expandability of subcutaneous adiposetissue, due to reduced ability to recruit and differentiatenew adipocytes, prevents its buffering effect in obesityand is characterized by expanded adipocytes (hypertro-phic obesity). Bone morphogenetic protein-4 (BMP4)plays a key role in regulating adipogenic precursor cellcommitment and differentiation. We found BMP4 to beinduced and secreted by differentiated (pre)adipocytes,and BMP4 was increased in large adipose cells. However,the precursor cells exhibited a resistance to BMP4 owingto increased secretion of the BMP inhibitor Gremlin-1(GREM1). GREM1 is secreted by (pre)adipocytes and isan inhibitor of both BMP4 and BMP7. BMP4 alone, and/orsilencing GREM1, increased transcriptional activationof peroxisome proliferator–activated receptor g andpromoted the preadipocytes to assume an oxidativebeige/brown adipose phenotype including markers ofincreased mitochondria and PGC1a. Driving white adi-pose differentiation inhibited the beige/brown markers,suggesting the presence of multipotent adipogenic pre-cursor cells. However, silencing GREM1 and/or addingBMP4 during white adipogenic differentiation reactivatedbeige/brown markers, suggesting that increased BMP4preferentially regulates the beige/brown phenotype.Thus, BMP4, secreted by white adipose cells, is an inte-gral feedback regulator of both white and beige adipogeniccommitment and differentiation, and resistance to BMP4by GREM1 characterizes hypertrophic obesity.

Obesity and its associated negative metabolic and healthconsequences are globally increasing at an epidemic rate.The subcutaneous adipose tissue (SAT) is the largestadipose depot of the body, and it is also the major sink for

excess fat storage. However, SAT has a limited expand-ability; when the limit is exceeded, fat accumulates indifferent ectopic sites, including the liver and visceraladipose tissue, and this is the major driver of themetabolic consequences of obesity (1). Studies have alsoshown that individuals with type 2 diabetes (T2D), com-pared with nondiabetic subjects, have increased amountof ectopic fat for the same amount of total body fat,supporting a reduced ability to expand the subcutaneousdepot (2). Consequently, ability to store excess fat in thesubcutaneous depot has a protective effect against theobesity-associated complications, a concept supported bythe ability of peroxisome proliferator–activated receptor(PPAR) g ligands to reduce ectopic fat while subcutaneousbody fat is increased (3). However, these ligands can onlyenhance the differentiation of already committed preadi-pocytes and cannot promote the commitment and recruit-ment of new adipose cells.

SAT contains a pool of preadipocytes and otherprecursor/stem cells that can differentiate into matureadipocytes (4). Regulation of adipogenesis is an importantquestion, since adipose cell expansion (hypertrophic adi-pocytes) is associated with a dysfunctional adipose tissuewith local and systemic insulin resistance and both in vivoand ex vivo studies have shown that hypertrophic obesityis characterized by a reduced recruitment of new cells(4–9). Importantly, we have recently shown that adiposecell size in the abdominal SAT is considerably larger inindividuals with a genetic predisposition (defined as beinga first-degree relative) for T2D than in matched individ-uals lacking a known heredity for diabetes (5,8). Thisnovel finding links hypertrophic obesity, ectopic fat accu-mulation, and associated insulin resistance with genetic

1Lundberg Laboratory for Diabetes Research, Sahlgrenska Academy at theUniversity of Gothenburg, Gothenburg, Sweden2Department of Molecular and Clinical Medicine, Sahlgrenska Academy at theUniversity of Gothenburg, Gothenburg, Sweden3MedImmune, LLC, Gaithersburg, MD

Corresponding author: Ulf Smith, ulf.smith@medic.gu.se.

Received 23 July 2014 and accepted 14 December 2014.

This article contains Supplementary Data online at http://diabetes.diabetesjournals.org/lookup/suppl/doi:10.2337/db14-1127/-/DC1.

© 2015 by the American Diabetes Association. Readers may use this article aslong as the work is properly cited, the use is educational and not for profit, andthe work is not altered.

1670 Diabetes Volume 64, May 2015

OBESITY

STUDIES

risk for T2D, a concept that has received strong recentsupport from large clinical studies (10,11).

The adipose tissue mesenchymal stem cells serve asa reservoir and allow a continued renewal of precursorcells that can differentiate into adipocytes (12–14). Thebone morphogenetic proteins (BMPs) are of particularinterest, since some members have been shown to recruitadipose precursor cells into the adipose lineage (6). BMP7is reported to be a regulator of brown adipogenesis (15),while BMP2 and -4 are related to white adipogenesis(6,12,16). We recently demonstrated that human SATpreadipocytes induce BMP4 during differentiation andthat BMP4 increased both commitment and differentia-tion of human precursor cells (6). In addition, recentstudies with human precursor cells found BMP4 also topromote the induction of a beige phenotype (17).

Adipogenic commitment of early precursor cells byBMP4 is mediated by the dissociation of an intracellularcomplex consisting of the PPARg transcriptional activa-tor zinc finger protein-423 (ZNF423) (18) and themesenchymal cell canonical WNT1-inducible signalingpathway protein-2 (WISP2), thereby allowing nuclearentry of ZNF423 and PPARg induction (19). Thus,BMP4 signaling and its cross-talk with canonical WNT/WISP2 is an essential component of the early inductionof adipogenesis. Consequently, inability to adequatelyincrease BMP4 in precursor cells would decrease adipo-genesis and, instead, promote enlargement of availablecells, i.e., hypertrophic obesity similar to what wasfound when the commitment factor, early B cell factor(Ebf )1, was genetically deleted in mice (20).

The amount of BMP available for signaling is tightlyregulated by the complex BMP receptor signaling path-ways including a number of structurally distinct BMPantagonists that alter the ability of BMPs to bind totheir receptors and regulate development of many dif-ferent cell types (21,22). Very little is currently knownabout the endogenous BMP antagonists and their effectson BMP4 action in human adipogenesis and hypertro-phic obesity.

Activin A, a secreted homodimer of inhibin-bA (INHBA)subunits (23,24), follistatin, (25) and the pseudoreceptorBMP and activin membrane-bound inhibitor (BAMBI) (26)are all expressed in the adipose tissue. Follistatin is se-creted by adipose tissue explants (27), and noggin isa well-established and secreted inhibitor that binds tothe BMP receptors, but its biological functions aremostly undetermined (6,28). Chordin and chordin-like-1(CHRDL1) are secreted proteins that bind BMP2, -4, and-7 (29,30), and CHRDL1 and has been shown to enhancethe proliferation of human mesenchymal stem cells (31).Gremlin-1 (GREM1) is a potent extra- and intracellularinhibitor of BMP4 (32) and is involved in fibrosis andarthritis development (33).

Brown adipose tissue (BAT) is specialized for energyexpenditure and maintaining body temperature. Untilnow, the physiological significance of BAT for whole-body

metabolism in adult man has been unclear, but coldexposure in men increased resting energy expenditure,glucose oxidation, and insulin sensitivity (34). An inter-mediate kind of brown adipocytes, the beige cells, hasbeen demonstrated in SAT (35,36). Increasing the activa-tion of beige cells in mice was associated with reducedweight gain and improved glucose tolerance (37).

In the current study, we examined the effects of BMP4and several BMP inhibitors during adipogenic differentia-tion of human subcutaneous preadipocytes. Our resultsprovide evidence for the concept that hypertrophic obesityis a condition of preadipocyte resistance to BMP4 as aconsequence of increased secretion of GREM1. In addition,GREM1 regulates the ability of BMP4 to induce beige/brown adipogenesis, making it an interesting target inobesity.

RESEARCH DESIGN AND METHODS

Human SubjectsGenes/proteins were studied in isolated mature adiposecells and adipose tissue needle biopsies of the SAT from33 individuals. The subjects were between 26 and 52 yearsold, with mean BMI 24.4 6 2.3 kg/m2 (range 19.5–27.5)and adipose cell size 92.8 6 9.7 mm (range 71.5–118.4).Additional SAT was obtained from 24 individuals by nee-dle biopsy (n = 23) or bariatric surgery (n = 1) for theadipogenic studies. The subjects were between 27 and 66years of age and had a mean BMI of 27.5 6 7.1 kg/m2

(range 19.3–54.8) and adipose cell size 95.9 6 17.0 mm(range 52.8–122.5). All subjects had normal glucose levelsand had no known chronic diseases. The ethics committeeof the University of Gothenburg approved the study de-sign, and written informed consent was received fromparticipants prior to inclusion in the study.

Digestion of Adipose Tissue Biopsies andPreadipocyte DifferentiationSAT was digested with collagenase and cell size measuredas previously described (6,7). The remaining stromal-vascular progenitor cells were isolated and differentiatedas previously described (6). When indicated, 100 ng/mL(3 nmol/L) BMP4 was added.

Quantitative Real-Time PCRDetails of real-time PCR assays were described previously(6). Gene-specific primers and probes were designed usingPrimer Express software or purchased as Assay-on-Demand (Life Technologies, Stockholm, Sweden).

Overexpression of CHRDL1For overexpression of CHRDL1, cells were transfectedwith an myc-DDK–tagged ORF clone of CHRDL1 obtainedfrom Origene (TrueORF Gold RC202635; BioNordikaSweden) using Lipofectamine 2000 (Life Technologies)according to the manufacturers’ protocol. An empty vec-tor was used as negative control and green fluorescentprotein expression to monitor transfection efficiency. Adi-pogenic differentiation was induced 24 h after transfection.

diabetes.diabetesjournals.org Gustafson and Associates 1671

Small Interfering RNAHuman isolated preadipocytes were transfected withCHRDL1, noggin, or GREM1 small interfering RNA(siRNA) using RNAiMAX (Life Technologies) accordingto the manufacturer’s instructions. After 48 h, mediumwas changed to adipocyte differentiation media with andwithout BMP4 (10–100 ng/mL) as stated. RNA wasextracted with an EZNA total RNA kit (VWR, StockholmSweden) after 6 days of culture.

Immunohistochemistry and Confocal Imaging of BMP4and MitochondriaHuman adipose tissue stromal cells were grown on glassslides, fixed with 4% formaldehyde for 15 min, andpermeabilized in 0.1% Triton X-100 for 5 min. Cellswere then blocked with 20% FBS for 30 min followed byincubation with anti-BMP4 antibody (MAB1049; MerckMillipore, Solna, Sweden) and UCP1 (ab10983; Abcam,Cambridge, U.K.) for 3h. After washing in PBS andincubation with secondary antibody conjugated withAlexa-594 for 1 h, confocal images were collected by theLeica SP5 confocal system. Cells were also stained withMitoTracker Red according to the manufacturer’s instruc-tions (Life Technologies).

Whole-Cell Extracts and ImmunoblottingProtein lysates and immunoblotting were performed aspreviously described (38) using the following antibodies:SMAD1/5/8 (sc-6031 Santa Cruz; AH Diagnostics, Solna,Sweden), pSMAD1/5/8 (Cell Signaling 9511; BioNordika,Stockholm, Sweden), pSMAD3 (Cell Signaling 8769), AKT(Cell Signaling 9272), SMAD3 (sc-8332 Santa Cruz),BMP4 (ab93939), and BMP7 (sc-53917).

Detection of Secreted GREM1 and BMP4AlphaLISA was used to measure secreted human GREM1(MedImmune, Gaithersburg, MD), and secreted BMP4was detected with an ELISA from Abcam (ab99982)according to the manufacturer’s instructions.

Statistical AnalysisThe experimental data are presented as means 6 SEM.Data analyzes were carried out with the PASWstatistics(SPSS, Inc.) for Macintosh. For cells differentiated invitro, Student paired t test was used for comparison ofgene expression with basal samples within groups, andone-way ANOVA (post hoc Tukey test) was used for directcomparisons between groups, correcting for multiple varia-bles where applicable. Differences were considered statisti-cally significant at P , 0.05 level.

RESULTS

BMP4 Is Secreted by Adipose Cells and Increased inHypertrophic ObesityTo address the potential role of BMP4 in hypertrophicobesity, we first asked whether BMP4 is expressed inmature adipose cells and, if so, whether it is reduced inenlarged cells. However, BMP4 transcript levels were highin isolated mature adipocytes and correlated positively

with the cell size (Fig. 1A), albeit not with BMI overthis limited range (19.5–27.5 kg/m2) (Fig. 1B).

We have previously shown that BMP4 mRNA levelsare increased during human preadipocyte differentiation(6). To ensure that BMP4 protein also is present in adi-pose cells, we examined differentiated and undifferenti-ated preadipocytes with both immunofluorescence andWestern blot. BMP4 protein was clearly induced afterdifferentiation (Fig. 1C). Furthermore, intermediate–and low–molecular weight secreted BMP4 protein inadipose tissue biopsies was also positively correlatedwith cell size (Fig. 1E). To validate that BMP4 also wassecreted from differentiated human adipocytes, we col-lected cell culture medium from day 0 and days 0–3 and9–12 from preadipocytes undergoing differentiation andmeasured BMP4 protein secretion with ELISA. BMP4 pro-tein was also secreted by differentiated (pre)adipocytes(Fig. 1F).

BMP Inhibitors Are Increased in Hypertrophic ObesityA possible explanation for the unexpected finding ofincreased BMP4 in hypertrophic obesity could be thatthe endogenous BMP inhibitors also are increased,thereby inducing a cellular resistance to BMP4 and itsproadipogenic effect. All measured BMP4 inhibitors wererobustly expressed in isolated mature adipose cells aswell as in undifferentiated preadipocytes and adipocytesdifferentiated in vitro (Supplementary Table 1). Further-more, transcript levels of GREM1 and CHRDL1 correlatedpositively with those of BMP4 in both mature adipocytesand intact adipose tissue biopsies (Supplementary Table2), supporting the concept of a BMP4 resistance. Inaddition, GREM1 and CHRDL1 mRNA levels in the hu-man adipose tissue biopsies correlated positively andsignificantly (P = 0.02 and P = 0.003, respectively) withadipose cell size of the donors, while there was no sig-nificant correlation for noggin or follistatin (data notshown). These results show that both BMP4 and certainBMP4 inhibitors are increased in adipose tissue charac-terized by expanded adipose cells, supporting the possi-bility that hypertrophic obesity is a condition of cellularBMP4 resistance.

Of note, CHRDL1 is the most highly expressed BMPinhibitor in differentiated (pre)adipocytes and, surpris-ingly, the highest expression is seen in mature adiposecells (Supplementary Table 1). Human tissue expressionpattern also shows CHRDL1 to be most highly expressedin the SAT (Supplementary Fig. 1A) and to be higher inobese than in lean individuals in this depot (Supplemen-tary Fig. 1B).

Regulation of BMP4 and BMP4 Inhibitors DuringAdipogenesisTo characterize the potential role of the BMP4 inhibitors,we examined their expression after differentiation ofsubcutaneous adipogenic precursor cells. BMP4 increasedas expected, and all but two of the inhibitors (CHRDL1and noggin) were reduced in differentiated cells (Fig. 2A).

1672 BMP4 Regulates Human White and Beige Adipogenesis Diabetes Volume 64, May 2015

We also analyzed BMP2 and BMP7 after differentiation.However, BMP2 transcript levels decreased (Fig. 2A),while BMP7 was generally not expressed in human pre-adipocytes or differentiated adipose cells and we could notdetect BMP7 protein in Western blots (data not shown).Thus, these were not further studied.

CHRDL1 increased gradually during differentiation,while noggin showed a transient early increase followedby inhibition (Fig. 2B and C). The expression of theseinhibitors was accentuated by the presence of BMP4,probably due to the enhanced differentiation induced byBMP4 (Fig. 2B and C). We characterized the time coursefor these inhibitors in relation to PPARg transcript levels,and both the increase in CHRDL1 and the decrease inGREM1 followed a time course similar to that of PPARginduction (Fig. 2D). The other BMP inhibitors (CHRD,

FST, and BAMBI) only showed minor differences inmRNA levels after preadipocyte differentiation, suggest-ing that they played a less important role for the apparentBMP4 resistance (Supplementary Table 1).

To validate the potential importance of CHRDL1 andnoggin in regulating human white adipogenesis, we exam-ined their expression after 9 days in preadipocytes un-dergoing poor or very good adipogenic differentiation. Wealso examined factors in the differentiation cocktail,which affected their expression. Table 1 shows thatinduction of CHRDL1 was positively related to that ofPPARg as well as ability of the cells to undergo differen-tiation. This is consistent with an overall positive corre-lation between CHRDL1 and PPARg expression in fullydifferentiated (pre)adipose cells (Fig. 2E). Thus, CHRDL1is a good marker of adipogenic differentiation consistent

Figure 1—BMP4 mRNA levels in mature adipose cells correlate with adipocyte cell size. Human adipocytes were isolated from SATbiopsies after collagenase digestion, and RNA was extracted. BMP4 mRNA levels correlate significantly with adipose cell size (P < 0.01,n = 33) (A) but not with BMI (B) over this limited range. C: BMP4 is induced during differentiation of human preadipocytes. Preadipocytesfrom two individuals were isolated from human adipose tissue, differentiated or not into adipocytes, and BMP4 protein was detected withimmunofluorescence. Double staining with anti-BMP4 and DAPI. D: Human adipose tissue biopsies were analyzed by immunoblot forintermediate– and low–molecular weight BMP4. Results from seven different donors from the same immunoblot. AKT was used as loadingcontrol, and recombinant human BMP4 (rhBMP4) was included as control. E: Ratio of total BMP4 to AKT, P = 0.01. F: Secretion of BMP4 tothe medium of preadipocytes undergoing differentiation from day 0 to days 0–3 and 9–12. Gene expressions were first normalized to 18SrRNA and then normalized to one individual (= 1). Spearman rank correlation was used to analyze data that were not normally distributed.

diabetes.diabetesjournals.org Gustafson and Associates 1673

with the very high expression in mature adipose cells(Supplementary Table 1).

In contrast to CHRDL1, there was no clear differencein noggin expression. This finding, together with the lackof correlation with cell size/BMI, suggests that noggin isan unlikely contributor to the preadipocyte BMP4 resis-tance in hypertrophic obesity. In contrast, the ability toinhibit GREM1 after differentiation was positively associated

with PPARg transcript activation, suggesting that GREM1could be an important endogenous regulator of BMP4resistance and adipogenesis (Fig. 2F).

Effect of Silencing or Overexpressing CHRDL1CHRDL1 was silenced by ;90% with siRNA (Fig. 3A); how-ever, this reduction did not increase, but markedly reduced,transcriptional activation of PPARg, adiponectin, and FABP4

Figure 2—Induction of BMP4 and BMP4 inhibitors during differentiation of human preadipocytes. Isolated human stromal cells from SATwere induced to differentiate. mRNA was extracted and analyzed with quantitative real-time PCR. A: mRNA levels of BMP2 and -4 and theBMP4 inhibitors at day 14 of differentiation (n = 24). The BMP4 inhibitors CHRDL1 (B) and noggin (C ) are rapidly induced duringdifferentiation. B and C: Gray bars, controls; black bars, BMP4 stimulation (n = 3). D: Time course for changes in CHRDL1, PPARg,and GREM1 mRNA levels during initiation of differentiation. Black bars, PPARg; gray bars, CHRDL1; dark-gray bars, GREM1. Log scalewas used owing to large differences between genes. There was a positive correlation between changes in PPARg and CHRDL1mRNA, P<0.01, even when the individual with high CHRDL1 expression was omitted (P < 0.01) (E), while suppression of GREM1 correlated with theincrease of PPARg mRNA levels after differentiation (F ), P < 0.01 (n = 38, differentiation day 14). Results were first normalized to 18S rRNAand then normalized to expression levels in the control sample/DMEM (= 1) at day 0. Results are means6 SEM. (*)P< 0.1, *P< 0.05, **P<0.02, ***P < 0.002.

Table 1—Identifying factors in the differentiation cocktail that regulated CHRDL1 and noggin

;5% degree of differentiation ;50% degree of differentiation

CHRDL1 Noggin PPARg2 CHRDL1 Noggin PPARg2

DMEM 0.85 0.91 1.15 1.05 0.94 1.00

IBMX 0.58 0.89 1.35 0.88 0.77 1.11

Pio 0.70 1.08 1.21 1.17 0.76 2.75

IBMX/pio 0.99 0.69 10.2 8.15 0.68 573

IBMX/dexa 2.65 3.05 20.2 5.18 1.62 268

Cocktail 6.16 1.93 274 45.2 1.28 3,243

CHRDL1 and noggin and their relation to preadipocyte differentiation potential, single additions or combinations thereof were added toDMEM/10% FBS. Data are from differentiation day 9 when the cells had started acquiring lipids. IBMX/pio/dexa/insulin represents a fulldifferentiation cocktail (dexa, dexamethasone; pio, pioglitazone), shown in boldface. DMEM/FBS is used as reference. Results from twoindividuals with different degrees of differentiation. Differentiation in percent relates to percent surface that was covered with lipid droplets.

1674 BMP4 Regulates Human White and Beige Adipogenesis Diabetes Volume 64, May 2015

during differentiation (Fig. 3B). Furthermore, addition ofBMP4 did not rescue the inhibitory effect of small interferenceCHRDL1 (siCHRDL1) on adipogenic differentiation (Fig. 3B).

Interestingly, cells transfected with CHRDL1 siRNAshowed increased ACTA2 (aSMA), a marker of the myofi-broblast phenotype, as well as INHBA mRNA levels (Fig.3C), suggesting that CHRDL1 may cross-talk with othersignaling pathways, possibly transforming growth factor(TGF)b, which is known to inhibit adipogenesis and toincrease these markers of fibrosis (23,39–41). Thus,

CHRDL1 is not an endogenous inhibitor of BMP4 in hu-man preadipocytes, and we also saw no inhibitory or pos-itive effect on pSMAD1/5/8 activation by BMP4 afterCHRDL1 overexpression in human preadipocytes (Fig.3D and E).

Consistent with the concept that CHRDL1 cross-talkswith other signaling pathways, we did not find thatoverexpressing CHRDL1 in preadipocytes directly en-hanced differentiation (Fig. 3F); instead, it decreasedACTA2, CTGF, and INHBA, further supporting potential

Figure 3—Silencing the BMP4 inhibitor CHRDL1 in preadipocytes (preadip) elicits negative effect on adipogenesis. Human stromal cellswere induced to differentiate into adipocytes for 5 days with scrambled or specific siRNA. A: CHRDL1 and GREM1 mRNA expression indifferentiated cells was reduced by 87% and 92%, respectively, after silencing, and GREM1 expression in preadipocytes was reduced by92%. B: PPARg, adiponectin (APM1), and FABP4 mRNA levels were reduced by CHRDL1 siRNA. C: Silencing CHRDL1 increases ACTA2and INHBA mRNA levels. D and E: Phosphorylation of SMAD1/5/8 is not changed in cells transfected with CHRDL1. The cells wereincubated for 2 h with 40 ng/mL BMP4. Noggin, 100 ng/mL, was used as BMP4 inhibitor. Results are means 6 SEM of three experiments.F: Human preadipocytes were transfected with CHRDL1 or empty vector. Differentiation was initiated 48 h after transfection, and mRNAlevels were analyzed at differentiation day 6. CHRDL1 expression in cells transfected with an empty vector was ;10-fold induced atdifferentiation day 6, while transfection with CHRDL1 further increased CHRDL1 ;20-fold, i.e., 150- to 200-fold. G: CHRDL1 overexpres-sion reduces ACTA2, CTGF, and INHBA mRNA levels in preadipocytes. Results were first normalized to 18S rRNA and then normalized toexpression levels in the control sample (scrambled = 1) or vector. Results are means 6 SEM of four experiments. (*)P < 0.1, *P < 0.05,**P < 0.02, ***P < 0.002.

diabetes.diabetesjournals.org Gustafson and Associates 1675

cross-talk with TGFb (Fig. 3G). We also examined whethersilencing CHRDL1 altered initial upstream signaling ofTGFb, measured as pSMAD3 increase by TGFb after120 min, but saw no such direct upstream effect (datanot shown).

GREM1 Is Increased in Hypertrophic ObesityGREM1 mRNA levels in whole SAT biopsies were posi-tively correlated with adipose cell size of the donors(Fig. 4A), and ability to inhibit GREM1 during differenti-ation of preadipocytes was also markedly reduced in hy-pertrophic obesity and correlated with the cell size of thedonors (Fig. 4B). This is consistent with the negative cor-relation between GREM1 and PPARg transcriptional acti-vation seen in differentiated cells (Fig. 2F).

Since GREM1 is a secreted BMP inhibitor, we measuredGREM1 in the culture medium of human preadipocytes

undergoing differentiation. Secretion of GREM1 corre-lated positively with GREM1 mRNA levels (Fig. 4C), andit remained essentially stable after day 6 of differentia-tion (Fig. 4D). This is in accordance with the rapiddownregulation of GREM1 during initiation of differen-tiation and the partial rebound effect seen at later timepoints (Fig. 4E). Furthermore, GREM1 secretion by dif-ferentiated preadipocytes correlated positively with BMIof the donors and also tended to correlate with cell size(Fig. 4F and G). We also verified that GREM1 protein isa bona fide inhibitor of BMP4 (as well as BMP7 [datanot shown]) signaling and pSMAD1/5/8 activation (Fig.4H and I).

We also analyzed the expression of GREM1 in differenthuman tissues. It is highly expressed in the adipose tissueand has a higher expression in omental tissue than in SAT(Supplementary Fig. 2).

Figure 4—GREM1 mRNA levels are positively correlated with cell size of the donors. A: GREM1 mRNA and cell size in whole adiposetissue, P < 0.05, n = 29. B: GREM1 mRNA and cell size in fully differentiated adipocytes, P < 0.05, n = 34. C: Secretion of GREM1 bydifferentiated preadipocytes at day 15 correlates with GREM1 mRNA levels, P < 0.01, n = 15. D: Secretion of GREM1 protein to the cellculture medium during adipogenic differentiation of human preadipocytes. Mean values of 15 individuals. E: GREM1 is rapidly reducedduring initiation of adipogenesis but increases partially at later time points. Secretion of GREM1 (at day 15 of preadipocyte differentiation) ispositively correlated with BMI, P < 0.05 (F ) but not significantly with the cell size of these donors (G) (P = 0.156, n = 15). H: GREM1 is aninhibitor for BMP4 signaling and inhibits pSMAD1/5/8. Preadipocytes were incubated with/without combinations of recombinant BMP4(10–40 ng/mL) and GREM1 protein (50–200 ng/mL) for 120 min. I: Ratio of pSMAD1/5/8 and SMAD1/5/8 protein (n = 3). mRNA results werefirst normalized to 18S rRNA and then normalized to expression levels in the undifferentiated sample (= 1). *P < 0.05.

1676 BMP4 Regulates Human White and Beige Adipogenesis Diabetes Volume 64, May 2015

Effect of Silencing GREM1 on White and BeigeAdipogenesis

We then examined the effect of silencing GREM1 (.90%inhibition [Fig. 3A]) in undifferentiated preadipocytesand found PPARg induction to be significantly (P ,0.05) increased to an extent similar to that seen in con-trol cells incubated with BMP4, and the effect of piogli-tazone was also significantly higher (P , 0.02) (Fig. 5A).Similarly, the PPARg transcriptional activator ZNF423(42) was significantly increased (Fig. 5B), and the effectof both BMP4 and pioglitazone was also higher insiGREM1 cells (Fig. 5B). These effects of silencingGREM1 indicate that it is an important regulator ofthe proadipogenic effect of endogenous BMP4 in humanpreadipocytes.

BMP4 has also recently been shown to enhance beigeadipogenesis in human precursor cells (17). We also foundaddition of BMP4 to increase the beige adipose markerTMEM26 in the preadipocytes (P , 0.05), and this effectwas markedly increased in siGREM1 cells, while addingthe PPARg ligand pioglitazone alone had no effect onTMEM26 under any of these conditions (Fig. 5C). Inter-estingly, and as also previously noted (17), the beige ad-ipose cell marker TBX1 was expressed at low levels, but itincreased significantly after addition of BMP4 (data notshown). Nevertheless, the expression of other markers ofbeige cells, TMEM26 and CD137 (TNFRSF9), was closelycorrelated in human preadipocytes (Fig. 5D).

The brown adipose cell marker ZIC1, but not PRDM16(data not shown), was clearly increased in siGREM1 cells(P , 0.02), and this was further enhanced by BMP4 (Fig.5E). Unexpectedly, we also found BMP8B, a regulator ofthermogenesis and BAT activation in mice (43), to berobustly expressed in human SAT as well as in the pre-adipocytes but only slightly increased by the addition ofBMP4 and/or by silencing GREM1 (data not shown).However, BMP8B expression correlated closely with thatof TMEM26, suggesting that it is a marker of beige adi-pogenesis in human adipose tissue (Fig. 5F). The mRNAlevels of UCP1 in control human preadipocytes were low,but the expression was increased and became measurablein most siGREM1 cells whether or not cAMP or BMP4was present (data not shown). Importantly, silencingGREM1 in human preadipocytes increased the mitochon-drial content as measured by the MitoTracker Red, andthis was further increased by BMP4 (Fig. 5G and H).Furthermore, induction of UCP1 was also seen underthese conditions (Fig. 5G and H). PGC1a was also in-creased by both silencing GREM1 and adding BMP4and further markedly increased by adding pioglitazone(Fig. 5I).

Taken together, GREM1 is an attractive target forovercoming the BMP4 resistance in hypertrophic obesity,since silencing GREM1 enhances endogenous BMP4signaling and action and increases ZNF423 and PPARginduction as well as expression of markers of an oxidativebeige/brown adipose cell phenotype.

White Adipogenic Differentiation Reduces Beige/Brown AdipogenesisWe also examined the beige/brown adipose cell markersCD137, TMEM26, and ZIC1 after induction of white adi-pose cell differentiation of the preadipocytes. All beigemarkers were dramatically reduced, suggesting the pres-ence of multipotent precursor cells, which could undergoeither beige/brown or white differentiation depending onthe ambient signals (Fig. 5J). Similarly, BMP8B was alsoreduced after induction of white adipogenesis (Fig. 5J).However, silencing GREM1 before white adipogenic dif-ferentiation reactivated TMEM26 (Fig. 5K) and ZIC1 (datanot shown), suggesting that GREM1 and/or the enhancedBMP4 signaling exerted a particularly prominent effect inpromoting the beige/brown phenotype of human preadi-pocytes. Ongoing studies are aimed at clarifying detailedmolecular mechanisms for this.

DISCUSSION

Hypertrophic obesity is associated with a dysregulatedadipose tissue, inflammation, and local and systemicinsulin resistance (4,5,8,9), and the degree of insulin re-sistance is positively correlated with adipose cell size (16)as well as future risk of developing T2D (44). Thus, un-derstanding adipose precursor cell recruitment and differ-entiation in the large SAT can open new possibilities forpreventing ectopic fat accumulation and the metaboliccomplications of obesity. Furthermore, recent animaldata have shown that cells in SAT have the greatest plas-ticity in terms of inducing beige/brown adipose cells andthat these cells play an important role in total body en-ergy expenditure and weight gain (45). BMP4 regulatesadipose precursor cell commitment into the white adiposelineage (6,13,46) and is induced in human preadipocytesundergoing differentiation (16), overexpression in the ad-ipose tissue in mice leads to an increased beige/brownphenotype in SAT (46), and BMP4 can also activate beigeadipose cell development in human precursor cells (17).

Since hypertrophic obesity is associated with an im-paired subcutaneous adipogenesis (16), we postulated thatthis could be due to reduced BMP4 in the precursor cells.However, we found that cellular BMP4 transcript and pro-tein levels are increased in hypertrophic obesity, and BMP4is secreted by the adipose cells, supporting a functionalfeedback regulation promoting the recruitment of new ad-ipose cells when needed rather than just expanding existingcells. These results suggest that the precursor cells are re-sistant to secreted BMP4, possibly due to increased activityof the endogenous BMP inhibitors.

The time course for the induction of the BMP inhibitorsduring preadipocyte differentiation showed that all butnoggin and CHRDL1 were markedly decreased. CHRDL1 in-creased during differentiation and was a positive markerof PPARg induction and adipogenesis, and in fact, it is notan inhibitor of BMP4 in human preadipocytes. ExpressingCHRDL1 in the undifferentiated preadipocytes had nodirect effect on PPARg but reduced several markers of

diabetes.diabetesjournals.org Gustafson and Associates 1677

fibrosis and TGFb activation. Silencing CHRDL1 reducedPPARg and increased ACTA2 (aSMA), CTGF, and INHBAas markers of a myofibroblast phenotype. These findingsindicate that CHRDL1 cross-talks with, and inhibits, the

profibrotic TGFb pathway, which is consistent with thefinding that genetic deletion of CHRDL1 in mice enhancesrenal fibrosis (31). Interestingly, CHRDL1 has been foundin the human adipose tissue secretome (47), indicating

Figure 5—Silencing GREM1 in preadipocytes induced a beige/brown adipose phenotype. PPARg2 was increased by silencing GREM1 andfurther induced by 1 mmol/L pioglitazone (pio) (A), comparable with the PPARg transcriptional activator ZNF423 (B). C: Similarly, TMEM26 wassignificantly increased, but pioglitazone had no direct effect. D: TMEM26 and CD137 mRNA levels were correlated in undifferentiatedpreadipocytes, P < 0.001, n = 23. The correlation remained significant when the outlier was omitted, P < 0.05. Results were first normalizedto 18S rRNA and then to one subject = 1. E: The brown adipose marker ZIC1was induced in siGREM1 cells and further increased by BMP4. F:TMEM26 correlated also with the transcript level of BMP8B, P < 0.01, n = 14, in undifferentiated preadipocytes. Results were first normalizedto 18S rRNA and then to one subject = 1. G: Silencing GREM1 increased the mitochondrial content in human preadipocytes. Mitochondriawere stained with MitoTracker Red. Silencing GREM1 also induced expression of UCP1, which was further enhanced by BMP4. Doublestaining with DAPI. H: Quantification of MitoTracker Red (four subjects) and UCP1 staining (three subjects). I: PGC1a was induced by silencingGREM1, and an additive effect was seen with pioglitazone and BMP4 in combination. J: Inducing white adipogenesis reduced TMEM26,CD137, BMP8B, and ZIC1 mRNA levels. Results from two to five individuals. BMP8B was only analyzed days 0 and 9. K: Silencing GREM1before induction of adipogenesis increased TMEM26 in partly differentiated cells. Results from six individuals at differentiation day 9. Resultswere first normalized to 18S rRNA and then normalized to expression levels in the undifferentiated sample at 0 h or the control sample(scrambled = 1) (A–C, E, and H–J). Results are means 6 SEM of 6–9 experiments. (*)P < 0.1, *P < 0.05, **P < 0.02, ***P < 0.002.

1678 BMP4 Regulates Human White and Beige Adipogenesis Diabetes Volume 64, May 2015

that it may be a circulating protein, but nothing is knownabout potential endocrine effects.

Noggin is a well-established inhibitor of BMP4, and asexpected, antagonizing noggin enhances endogenousBMP4-stimulated preadipocyte differentiation to boththe white (6) and beige (data not shown) phenotype.However, its temporal expression, lack of associationwith adipose cell size, and precursor cells differentiationcapacity suggest that it is not specifically related to thereduced adipogenesis or BMP4 resistance in hypertrophicobesity.

In contrast, GREM1 is a secreted and powerful inhibitorof BMP4 and BMP7 signaling as well as of adipogenicdifferentiation. The time course for GREM1 is consistentwith an important regulatory effect on adipogenesis. It issecreted by human (pre)adipocytes, secreted protein andmRNA levels were positively correlated, and transcript lev-els were closely correlated with adipose cell size. SilencingGREM1 also increased the effect of BMP4 on PPARg in-duction. Thus, GREM1 is an important regulator of whiteadipogenesis and the ability of BMP4 to induce commit-ment and differentiation.

Interestingly, GREM1 also regulates BMP4-inducedbeige/brown adipogenesis as well as the effect of BMP4on markers of mitochondrial content and PGC1a induc-tion. Unexpectedly, we also found BMP4 and GREM1 tobe involved in the regulation of BMP8B. BMP8B activatesthermogenesis in BAT in mice, but it is not expressed inwhite adipose tissue (43). However, BMP8B is robustlyexpressed and closely associated with TMEM26, suggest-ing that it is a marker of beige adipose cells in humanSAT.

Induction of white adipogenesis markedly reducedTMEM26, CD137, BMP8B, and ZIC1 in the human cells,

suggesting the presence of multipotent precursor cellsthat could enter both white and beige/brown adipogene-sis. This concept is similar to the proposal by Lee et al.(48) in murine SAT, and lineage-tracing experiments haveshown that beige adipose cells can revert to white cells(49). Importantly, we also found that silencing GREM1 inpartly differentiated white adipocytes reactivated TMEM26and ZIC1, suggesting a transdifferentiation potential andthat GREM1/BMP4 strongly promotes a beige/brownphenotype of already committed and differentiating pre-adipocytes. Our findings are schematically illustrated inFig. 6.

Taken together, our results show that BMP4 is animportant endogenous regulator of human white andbeige adipogenesis. However, the powerful BMP4/7 in-hibitor GREM1, which is specifically increased in hyper-trophic obesity, antagonizes these effects and is aninteresting target in human obesity/T2D.

Funding. The authors thank the Swedish Research Council, the SwedishDiabetes Association, the Novo Nordisk Foundation, the Torsten SöderbergsFoundation, and the West Sweden ALF program for financial support.Duality of Interest. J.G. and C.R are employees of MedImmune, LLC.Financial support was also received from MedImmune (MA-414379) as a re-search agreement with the University of Gothenburg. No other potential conflictsof interest relevant to this article were reported.Author Contributions. B.G. performed experiments, analyzed data,and wrote the manuscript. A.H., S.H., and J.M.H. performed experiments.P.-A.S. contributed with data in human tissues. J.G. and C.R. contributedwith discussion and reviewed the manuscript. U.S. designed the experiments,did the data analysis and interpretation, and wrote the manuscript. U.S. is theguarantor of this work and, as such, had full access to all the data in the studyand takes responsibility for the integrity of the data and the accuracy of thedata analysis.

Figure 6—Schematic view of the regulation of commitment and differentiation of human subcutaneous adipose precursor cells by BMP4and GREM1. BMP4 increases during differentiation of human preadipocytes and can exert autocrine/paracrine effects as a secretedmolecule. This effect of BMP4 enhances commitment and differentiation of new preadipocytes and therefore strives to prevent adiposecell hypertrophy. The effect of BMP4 is regulated by the endogenous BMP inhibitors, and in human preadipocytes, GREM1 is a keyregulator of the effect of BMP4 to promote both white and beige/brown differentiation.

diabetes.diabetesjournals.org Gustafson and Associates 1679

Prior Presentation. Parts of this study were presented in abstract form atthe 74th Scientific Sessions of the American Diabetes Association, San Francisco,CA, 13–17 June 2014.

References1. Neeland IJ, Turer AT, Ayers CR, et al. Dysfunctional adiposity and the risk ofprediabetes and type 2 diabetes in obese adults. JAMA 2012;308:1150–11592. Gallagher D, Kelley DE, Yim JE, et al.; MRI Ancillary Study Group of the LookAHEAD Research Group. Adipose tissue distribution is different in type 2 diabetes.Am J Clin Nutr 2009;89:807–8143. Gastaldelli A, Harrison SA, Belfort-Aguilar R, et al. Importance of changes inadipose tissue insulin resistance to histological response during thiazolidinedionetreatment of patients with nonalcoholic steatohepatitis. Hepatology 2009;50:1087–10934. Arner E, Westermark PO, Spalding KL, et al. Adipocyte turnover: relevanceto human adipose tissue morphology. Diabetes 2010;59:105–1095. Arner P, Arner E, Hammarstedt A, Smith U. Genetic predisposition for type 2diabetes, but not for overweight/obesity, is associated with a restricted adipo-genesis. PLoS ONE 2011;6:e182846. Gustafson B, Smith U. The WNT inhibitor Dickkopf 1 and bone morpho-genetic protein 4 rescue adipogenesis in hypertrophic obesity in humans. Di-abetes 2012;61:1217–12247. Isakson P, Hammarstedt A, Gustafson B, Smith U. Impaired preadipocytedifferentiation in human abdominal obesity: role of Wnt, tumor necrosis factor-alpha, and inflammation. Diabetes 2009;58:1550–15578. Jansson PA, Pellmé F, Hammarstedt A, et al. A novel cellular marker ofinsulin resistance and early atherosclerosis in humans is related to impaired fatcell differentiation and low adiponectin. FASEB J 2003;17:1434–14409. Weisberg SP, McCann D, Desai M, Rosenbaum M, Leibel RL, Ferrante AWJr. Obesity is associated with macrophage accumulation in adipose tissue. J ClinInvest 2003;112:1796–180810. Cederberg H, Stan�cáková A, Kuusisto J, Laakso M, Smith U. Family historyof type 2 diabetes increases the risk of both obesity and its complications: is type2 diabetes a disease of inappropriate lipid storage? J Intern Med. 12 July 2014[Epub ahead of print]11. Scott RA, Fall T, Pasko D, et al.; RISC Study Group; EPIC-InterAct Consor-tium. Common genetic variants highlight the role of insulin resistance and bodyfat distribution in type 2 diabetes, independent of obesity. Diabetes 2014;63:4378–438712. Bowers RR, Kim JW, Otto TC, Lane MD. Stable stem cell commitment to theadipocyte lineage by inhibition of DNA methylation: role of the BMP-4 gene. ProcNatl Acad Sci U S A 2006;103:13022–1302713. Bowers RR, Lane MD. A role for bone morphogenetic protein-4 in adipocytedevelopment. Cell Cycle 2007;6:385–38914. Tang QQ, Otto TC, Lane MD. Commitment of C3H10T1/2 pluripotent stemcells to the adipocyte lineage. Proc Natl Acad Sci U S A 2004;101:9607–961115. Tseng YH, Kokkotou E, Schulz TJ, et al. New role of bone morphogeneticprotein 7 in brown adipogenesis and energy expenditure. Nature 2008;454:1000–100416. Gustafson B, Hammarstedt A, Hedjazifar S, Smith U. Restricted adipo-genesis in hypertrophic obesity: the role of WISP2, WNT, and BMP4. Diabetes2013;62:2997–300417. Elsen M, Raschke S, Tennagels N, et al. BMP4 and BMP7 induce the white-to-brown transition of primary human adipose stem cells. Am J Physiol CellPhysiol 2014;306:C431–C44018. Gupta RK, Arany Z, Seale P, et al. Transcriptional control of preadipocytedetermination by Zfp423. Nature 2010;464:619–62319. Hammarstedt A, Hedjazifar S, Jenndahl L, et al. WISP2 regulates pre-adipocyte commitment and PPARg activation by BMP4. Proc Natl Acad Sci U S A2013;110:2563–256820. Gao H, Mejhert N, Fretz JA, et al. Early B cell factor 1 regulates adipocytemorphology and lipolysis in white adipose tissue. Cell Metab 2014;19:981–992

21. Gazzerro E, Canalis E. Bone morphogenetic proteins and their antagonists.Rev Endocr Metab Disord 2006;7:51–6522. Walsh DW, Godson C, Brazil DP, Martin F. Extracellular BMP-antagonistregulation in development and disease: tied up in knots. Trends Cell Biol 2010;20:244–25623. Zaragosi LE, Wdziekonski B, Villageois P, et al. Activin a plays a critical rolein proliferation and differentiation of human adipose progenitors. Diabetes 2010;59:2513–252124. Carlsson LM, Jacobson P, Walley A, et al. ALK7 expression is specific foradipose tissue, reduced in obesity and correlates to factors implicated in met-abolic disease. Biochem Biophys Res Commun 2009;382:309–31425. Nakatani M, Takehara Y, Sugino H, et al. Transgenic expression ofa myostatin inhibitor derived from follistatin increases skeletal muscle massand ameliorates dystrophic pathology in mdx mice. FASEB J 2008;22:477–48726. Luo X, Hutley LJ, Webster JA, et al. Identification of BMP and activinmembrane-bound inhibitor (BAMBI) as a potent negative regulator of adipo-genesis and modulator of autocrine/paracrine adipogenic factors. Diabetes 2012;61:124–13627. Flanagan JN, Linder K, Mejhert N, et al. Role of follistatin in promotingadipogenesis in women. J Clin Endocrinol Metab 2009;94:3003–300928. Wang G, Zhang H, Zhao Y, et al. Noggin and bFGF cooperate to maintain thepluripotency of human embryonic stem cells in the absence of feeder layers.Biochem Biophys Res Commun 2005;330:934–94229. Zhang D, Ferguson CM, O’Keefe RJ, Puzas JE, Rosier RN, Reynolds PR. Arole for the BMP antagonist chordin in endochondral ossification. J Bone MinerRes 2002;17:293–30030. Fernandes H, Dechering K, van Someren E, et al. Effect of chordin-like 1 onMC3T3-E1 and human mesenchymal stem cells. Cells Tissues Organs 2010;191:443–45231. Lin J, Patel SR, Cheng X, et al. Kielin/chordin-like protein, a novel enhancerof BMP signaling, attenuates renal fibrotic disease. Nat Med 2005;11:387–39332. Topol LZ, Bardot B, Zhang Q, et al. Biosynthesis, post-translation modifi-cation, and functional characterization of Drm/Gremlin. J Biol Chem 2000;275:8785–879333. Costello CM, Cahill E, Martin F, Gaine S, McLoughlin P. Role of gremlin inthe lung: development and disease. Am J Respir Cell Mol Biol 2010;42:517–52334. Chondronikola M, Volpi E, Børsheim E, et al. Brown adipose tissue improveswhole-body glucose homeostasis and insulin sensitivity in humans. Diabetes2014;63:4089–409935. Jespersen NZ, Larsen TJ, Peijs L, et al. A classical brown adipose tissuemRNA signature partly overlaps with brite in the supraclavicular region of adulthumans. Cell Metab 2013;17:798–80536. Waldén TB, Hansen IR, Timmons JA, Cannon B, Nedergaard J. Recruited vs.nonrecruited molecular signatures of brown, “brite,” and white adipose tissues.Am J Physiol Endocrinol Metab 2012;302:E19–E3137. Seale P, Conroe HM, Estall J, et al. Prdm16 determines the thermogenicprogram of subcutaneous white adipose tissue in mice. J Clin Invest 2011;121:96–10538. Gustafson B, Smith U. Cytokines promote Wnt signaling and inflammationand impair the normal differentiation and lipid accumulation in 3T3-L1 pre-adipocytes. J Biol Chem 2006;281:9507–951639. Du B, Cawthorn WP, Su A, et al. The transcription factor paired-relatedhomeobox 1 (Prrx1) inhibits adipogenesis by activating transforming growthfactor-b (TGFb) signaling. J Biol Chem 2013;288:3036–304740. Ohgo S, Hasegawa S, Hasebe Y, Mizutani H, Nakata S, Akamatsu H.Bleomycin inhibits adipogenesis and accelerates fibrosis in the subcutaneousadipose layer through TGF-b1. Exp Dermatol 2013;22:769–77141. Wei J, Ghosh AK, Sargent JL, et al. PPARg downregulation by TGFß in fi-broblast and impaired expression and function in systemic sclerosis: a novelmechanism for progressive fibrogenesis. PLoS ONE 2010;5:e13778

1680 BMP4 Regulates Human White and Beige Adipogenesis Diabetes Volume 64, May 2015

42. Gupta RK, Mepani RJ, Kleiner S, et al. Zfp423 expression identifies com-mitted preadipocytes and localizes to adipose endothelial and perivascular cells.Cell Metab 2012;15:230–23943. Whittle AJ, Carobbio S, Martins L, et al. BMP8B increases brown adiposetissue thermogenesis through both central and peripheral actions. Cell 2012;149:871–88544. Lönn M, Mehlig K, Bengtsson C, Lissner L. Adipocyte size predicts incidenceof type 2 diabetes in women. FASEB J 2010;24:326–33145. Cohen P, Levy JD, Zhang Y, et al. Ablation of PRDM16 and beige adiposecauses metabolic dysfunction and a subcutaneous to visceral fat switch. Cell2014;156:304–316

46. Qian SW, Tang Y, Li X, et al. BMP4-mediated brown fat-like changes inwhite adipose tissue alter glucose and energy homeostasis. Proc Natl Acad SciU S A 2013;110:E798–E80747. Dahlman I, Elsen M, Tennagels N, et al. Functional annotation of the humanfat cell secretome. Arch Physiol Biochem 2012;118:84–9148. Lee YH, Petkova AP, Mottillo EP, Granneman JG. In vivo identification ofbipotential adipocyte progenitors recruited by b3-adrenoceptor activation andhigh-fat feeding. Cell Metab 2012;15:480–49149. Wang QA, Tao C, Gupta RK, Scherer PE. Tracking adipogenesis during whiteadipose tissue development, expansion and regeneration. Nat Med 2013;19:1338–1344

diabetes.diabetesjournals.org Gustafson and Associates 1681