fabp4 attenuates pparg and adipogenesis and is inversely ...adipogenesis and is inversely correlated...

Tali Garin-Shkolnik,1 Assaf Rudich,2 Gökhan S. Hotamisligil,3 and Menachem Rubinstein1

FABP4 Attenuates PPARg andAdipogenesis and Is InverselyCorrelated With PPARg inAdipose Tissues

Fatty acid binding protein 4 (FABP4, also known asaP2) is a cytoplasmic fatty acid chaperoneexpressed primarily in adipocytes and myeloid cellsand implicated in the development of insulinresistance and atherosclerosis. Here wedemonstrate that FABP4 triggers the ubiquitinationand subsequent proteasomal degradation ofperoxisome proliferator–activated receptor g(PPARg), a master regulator of adipogenesis andinsulin responsiveness. Importantly, FABP4-nullmouse preadipocytes as well as macrophagesexhibited increased expression of PPARg, andcomplementation of FABP4 in the macrophagesreversed the increase in FABP4 expression. TheFABP4-null preadipocytes exhibited a remarkablyenhanced adipogenesis compared with wild-typecells, indicating that FABP4 regulates adipogenesisby downregulating PPARg. We found that the FABP4level was higher and PPARg level was lower inhuman visceral fat and mouse epididymal fatcompared with their subcutaneous fat. Furthermore,FABP4 was higher in the adipose tissues of obesediabetic individuals compared with healthy ones.Suppression of PPARg by FABP4 in visceral fatmay explain the reported role of FABP4 in thedevelopment of obesity-related morbidities,

including insulin resistance, diabetes, andatherosclerosis.Diabetes 2014;63:900–911 | DOI: 10.2337/db13-0436

Adiposity is closely correlated with important physio-logical parameters such as blood pressure, systemic in-sulin sensitivity, dyslipidemia, and serum triglyceridelevels (1,2), rendering obesity to an independent riskfactor for myocardial infarction, stroke, type 2 diabetes,and certain cancers (3). Among adipose tissues, visceralfat is more closely correlated with obesity-associatedpathologies than overall adiposity (4–8).

The nuclear receptor peroxisome proliferator–activated receptor g (PPARg) is a master regulator ofadipose cell differentiation, playing a critical role in sys-temic lipid and glucose metabolism (9). PPARg is acti-vated by natural or synthetic agonists such as theantidiabetic thiazolidinedione (TZD) (10). Activated PPARgis a master regulator of adipogenesis, acting as a transcrip-tion factor of genes expressed in mature adipocytes,including fatty acid binding protein (FABP4), CD36, lipo-protein lipase (LPL), and adiponectin, all of which containperoxisome proliferator response elements (PPREs) (11).PPARg is also expressed in myeloid cells, and its activationpromotes an anti-inflammatory phenotype (11). Disruptionof PPARg specifically in myeloid cells also predisposes

1Department of Molecular Genetics, Weizmann Institute of Science, Rehovot,Israel2Department of Clinical Biochemistry, Ben-Gurion University of the Negev, Beer-Sheva, Israel3Department of Genetics and Complex Diseases, Harvard School of PublicHealth, Boston, MA

Corresponding author: Menachem Rubinstein, [email protected].

Received 18 March 2013 and accepted 1 December 2013.

© 2014 by the American Diabetes Association. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.

900 Diabetes Volume 63, March 2014

METABOLISM

mailto:[email protected]

mailto:[email protected]

http://creativecommons.org/licenses/by-nc-nd/3.0/

http://creativecommons.org/licenses/by-nc-nd/3.0/

mice to the development of diet-induced obesity, insulinresistance, and glucose intolerance (12), whereas activa-tion of PPARg within macrophages promotes lipid efflux,thereby stabilizing atherosclerotic lesions (13).

A major target gene of PPARg is the lipid transporterFABP4, also known as aP2 (14). PPARg induces FABP4almost exclusively in adipocytes and macrophages.FABP4 acts as a fatty acids chaperone, which couplesintracellular lipids to biological targets and signalingpathways (15,16). FABP4 has been implicated in severalaspects of the metabolic syndrome in mice, includinginsulin resistance and atherosclerosis (17–22). Inhumans, a T87C polymorphism in the FABP4 promoterleads to reduced transcriptional activity, resulting inlower serum triglycerides and a reduced risk of athero-sclerosis and type 2 diabetes (23). In addition, the level ofcirculating human FABP4 was proposed as an in-dependent prognostic marker for future development ofdiabetes (24,25). Furthermore, PPARg is considereda tumor suppressor gene (11), whereas FABP4 has a rolein tumorigenesis (26,27).

Understanding the mechanisms by which obesityalters adipose tissue biology is of major importance be-cause the fat tissue acts as an endocrine organ thatregulates systemic metabolism (28). Visceral adiposetissues (VATs) and subcutaneous adipose tissues (SATs)are functionally distinct, and many clinical studies led tothe conclusion that VAT, rather than SAT, is the keyfactor inducing the metabolic syndrome, also serving asa source of inflammatory and stress-promoting factors(8,29–31).

To date, the mechanisms by which FABP4 promotesinsulin resistance and inflammation are not fully un-derstood. The fact that FABP4 is induced by PPARg butexerts metabolic and immunologic activities opposite tothose of PPARg led us to hypothesize that FABP4 mightbe involved in regulating PPARg activity. Here we showthat FABP4 negatively regulates PPARg levels in macro-phages and adipocytes, thereby attenuating adipocytedifferentiation; we elucidate a mechanism involvingproteasomal PPARg degradation and further demon-strate elevated FABP4/PPARg ratio in intra-abdominalfat compared with SAT in mice and VAT compared withSAT in humans.

RESEARCH DESIGN AND METHODS

Cells and Reagents

Human THP-1 (TIB-202) cells, human HEK 293T cells,HeLa (CCL-2.1) cells, mouse NIH-3T3 fibroblasts (CRL-1658), and mouse 3T3-L1 preadipocytes (CL-173) werefrom the American Type Culture Collection. Immortal-ized wild-type (WT) and FABP4-null preadipocytes, aswell as immortalized WT, FABP4-null, and FABP4-complemented macrophage cell lines were generatedfrom WT and FABP4-null mice as previously described(22,32). 3T3-L1 cells and preadipocytes from FABP4knockout and WT mice were grown in Dulbecco’s

modified Eagle’s medium (DMEM) containing 10% bo-vine serum. THP-1 cells and immortalized macrophagecell lines were grown in RPMI 1640 medium containing10% FBS, 2 mmol/L L-glutamine, 1.5 g/L sodium bi-carbonate, 4.5 g/L glucose, 10 mmol/L HEPES, and 1mmol/L sodium pyruvate. All other cell lines were grownin DMEM with 10% FBS (Gibco), 100 IU/mL penicillin,and 0.1 mg/mL streptomycin. The FABP4 inhibitorBMS309403 and the proteasome inhibitor MG262 werefrom Calbiochem. Rosiglitazone was from CaymanChemical. JetPEI transfection reagent was from PolyplusTransfection. Small interfering RNA (siRNA) oligonu-cleotides were from Dharmacon. Cell Line NucleofectorKit V was used for siRNA transfections of THP-1 cells(Amaxa GmbH, Cologne, Germany). All other reagentswere from Sigma-Aldrich.

Differentiation of THP-1 Cells and Pre-adipocytes

THP-1 cells (1 3 106/mL) were induced to differentiateby phorbol myristic acid (100 nmol/L, 24 h). For exper-iments in which BMS309403 was included, the mediumwas supplemented with 10% lipoprotein-deficient seruminstead of 10% FBS. 3T3-L1 preadipocytes were differ-entiated as previously described (33). In brief, cells weresubcultured at ,70% confluence. To induce differentia-tion, confluent cell cultures were placed in mediumcontaining 0.5 mmol/L 1-methyl-3-isobutylxanthine,1 mmol/L dexamethasone, and 5 mg/mL insulin in 10%FBS/DMEM for 48 h. Next, the medium was changed,and 5 mg/mL insulin was added for another 48 h. Themedium was then replaced with 10% FBS/DMEM andwas renewed every 2 days until the cells became fullydifferentiated and lipid droplets were apparent.

Oil Red O Staining

Cells were washed, fixed with formaldehyde (4%,20 min), washed three times with PBS, stained with OilRed O (5% in isopropanol, freshly diluted 2:3 with water,30 min), and washed.

Transient Transfection Assays

Cells at 50–70% confluence were transfected using thejetPEI reagent and the indicated plasmids, according tothe manufacturer’s instructions. Wherever needed, anempty vector was added to maintain a constant amountof 5 mg DNA per well. pCS6 (empty vector), pPPARg, andpFABP4 were from Open Biosystems. The followingvectors were provided by the indicated person: pp53,Y. Shaul, Weizmann Institute, Rehovot, Israel; pPPRE-luc, the laboratory of the late M. Liscovitch, WeizmannInstitute; and pHA-Ub, C. Kahana, Weizmann Institute.THP1 cells were transiently transfected using the AmaxaNucleofection Technology (Amaxa GmbH, Cologne, Ger-many) before differentiation.

Preparation of Adipose Tissue Lysates

FABP4-null mice were generated as previously described(22). Biopsies from epididymal and subcutaneous depots

diabetes.diabetesjournals.org Garin-Shkolnik and Associates 901

of 12-week-old mice were removed, frozen in liquid ni-trogen, and stored at –80°C. Fat tissues were homoge-nized in lysis buffer (10 mmol/L Na2HPO4, pH 7.5,5 mmol/L EDTA, 100 mmol/L NaCl, 1% Triton, 0.5%sodium deoxycholate, 0.1% SDS) containing proteaseinhibitors (1:1,000; Sigma-Aldrich) and 1 mmol/Lphenylmethylsulfonyl fluoride (PMSF). The institutionalethics committee approved in advance all protocols of thestudy.

Paired human VAT and abdominal SAT biopsies wereobtained during elective abdominal surgery for gastricbanding, weight reduction surgery, or exploratory lapa-rotomy (with negative findings) as previously described(34). The institutional ethics committee approved inadvance all protocols of the study, and all participantsprovided a written informed consent after all objectivesand procedures were explained. Biopsies were deliveredon ice, rinsed in saline, frozen in liquid nitrogen, pow-dered, and then homogenized in lysis buffer (150 mmol/LNaCl, 50 mmol/L Tris-HCl, pH 7.5, 1% by volumeNonidet P-40, 0.25% sodium deoxycholate, 10 mmol/Lsodium b-glycerophosphate, 5 mmol/L sodium pyro-phosphate, 1 mmol/L EGTA, 1 mmol/L sodium vanadate,and 1 mmol/L NaF), supplemented with protease inhibitorcocktail (1:1,000).

Immunoblot Analysis

Cell pellets were resuspended on ice for 20 min ina buffer consisting of 100 mmol/L KCl, 0.5 mmol/LEDTA, 20 mmol/L HEPES, pH 7.6, 0.4% Nonidet P-40,20% glycerol, protease inhibitors, and 1 mmol/L PMSF.The clarified lysate was stored at 280°C. Protein con-centration was determined by BCA Protein Assay Re-agent Kit (Pierce, Rockford, IL) using BSA as a standard.Samples were subjected to 15% or 6–18% gradient SDS-PAGE and immunoblotted with anti-actin (MP Bio-medicals), anti-FABP4, anti-PPARg, anti-LPL (Santa CruzBiotechnology), anti-HA (Covance), and anti-ubiquitinatedproteins (clone FK2; BIOMOL International). Mousemonoclonal anti-p53 was a gift of C. Prives (ColumbiaUniversity, New York, NY). Detection was performedusing horseradish peroxidase–conjugated secondaryantibodies (Jackson ImmunoResearch Laboratories),followed by SuperSignal chemiluminescence detectionsystem (Pierce). Quantitative densitometry was per-formed with EZ-Quant software. Protein levels werenormalized to values obtained for b-actin.

Immunoprecipitation Assays

Cells were washed with ice-cold PBS and lysed with PBScontaining 1% Triton X-100, 0.5% sodium deoxycholate,10 mmol/L sodium pyrophosphate, 2 mmol/L sodiumvanadate, 100 mmol/L sodium fluoride, protease inhib-itors mixture, and 1 mmol/L PMSF. Samples (500 mgprotein) were precleared by adding 60 mL proteinG-Sepharose (1 h, 4°C; GE Healthcare, Uppsala, Sweden).Anti-PPARg antibody (1:100 in lysis buffer) was added

for 17 h at 4°C, and protein G-Sepharose was then addedfor another hour. The beads were then washed four timeswith 1% Nonidet P-40 in PBS, extracted with samplebuffer, and analyzed by SDS-PAGE and immunoblotting,as described above.

Reporter Gene Assay

NIH-3T3 cells (10–17 3 103 cells/well) in 96-well plateswere transfected using the jetPEI reagent with the fol-lowing plasmids: 33 PPRE-TK-luciferase reporter (67ng), pPPARg (67 ng), pFABP4 (67 ng), and 10 ng pRL-TK(Renilla luciferase reporter vector; Promega). After 24 h,the medium was replaced by DMEM/10% FBS plus 10mmol/L rosiglitazone or vehicle (DMSO) for 48 h. Celllysates were then analyzed for luciferase activity andnormalized to Renilla activity using the Dual-LuciferaseReporter Assay Kit (Promega), according to the manu-facturer’s instructions.

Quantitative Real-Time PCR

Total RNA (1 mg), isolated using PerfectPure RNACultured Cell Kit (5 Prime), was reverse transcribedusing random hexamers and High Capacity RNA-to-cDNA Kit (Applied Biosystems). The reaction mixtureswere diluted 15-fold, and 4.5 mL was then used as thetemplate for real-time PCR. The following assays wereperformed: mFABP4, Mn00445880_m1; mPPARg,Mn00440945_m1; and mTBP, Mm00446973_m1, asa loading control. These probes were designed usingRoche ProbeFinder Software (Roche Diagnostics).Quantitative real-time PCR (qPCR) was performedusing a Roche LightCycler 480 Real-Time PCR Systemand TaqMan Universal PCR Master Mix (AppliedBiosystems). The amplification program was asfollows: initial denaturation at 95°C for 15 min,followed by 45 cycles of 95°C for 15 s, 60°C for 1 min,and 40°C for 30 s. Gene expression was normalizedto TBP mRNA and was calculated using LightCyclerSoftware (Roche). The results are means of foldchange 6 SEM of triplicates.

Inhibition of FABP4 Gene Expression by RNAInterference

THP-1 cells were transfected using the Cell Line Nucleo-fector Kit V (Amaxa) with either siFABP4 siGENOMESMARTpool (140 nmol/L) or Non-Targeting siRNA Pool(140 nmol/L; Dharmacon), using the Amaxa Nucleo-fection Technology, program V-01. After 24 h, the cellswere induced to differentiate to macrophages and thelevels of FABP4 and PPARg mRNA were determinedby qPCR.

Statistical Analysis

ANOVA and Student t test were performed for statis-tical analysis, as appropriate. Statistical significance wasdesignated at P , 0.05. Values are expressed asmean 6 SEM.

902 FABP4 Downregulates PPARg and Adipogenesis Diabetes Volume 63, March 2014

RESULTS

Inhibition of FABP4 Enhances PPARg Level andActivity

THP-1 cells were induced to differentiate into macrophage-like cells and then incubated with either vehicle orthe FABP4 inhibitor BMS309403. Inhibition of FABP4did not affect the level of PPARg mRNA (Fig. 1A) butelevated the basal level of PPARg protein by 2.6 6 0.8–fold (P , 0.005, n = 3), as determined by immunoblot-ting and densitometry (Fig. 1B). To further study theeffect of FABP4 on PPARg, we knocked down FABP4mRNA in THP-1 cells by FABP4-targeting siRNA ornontargeting control siRNA. After 24 h, the cells weredifferentiated to macrophages, and after an additional48 h, they were analyzed for the protein and the mRNAlevels of both FABP4 and PPARg. FABP4 siRNA effec-tively inhibited the expression of FABP4 mRNA andprotein (Fig. 1C and D) and had only a minor effect onthe expression of PPARg mRNA (Fig. 1C ). In contrast,FABP4 knockdown increased the protein level of PPARgby 170 6 20% (n = 3, P , 0.01) (Fig. 1D). These resultssuggested that PPARg is negatively regulated by FABP4predominantly at the posttranscriptional level.

PPARg is induced during the differentiation of pre-adipocytes into adipocytes and in turn induces a setof differentiation-dependent genes, including FABP4,LPL, and adiponectin (9,35). Despite the fact that FABP4is a robust marker of mature adipocytes, addition ofits inhibitor, BMS309403, to the adipogenic cocktailincreased PPARg protein expression by 3.9 6 0.2–fold(n = 3, P , 0.05), as well as expression of the PPARggene products FABP4, LPL, and adiponectin (Fig. 1E).These findings suggested that FABP4 negativelyregulates PPARg-mediated gene activation duringadipogenesis.

We then studied the interrelationships of FABP4 andPPARg in vivo, comparing WT mice with FABP4-nullmice (17). We found that white adipose tissue of FABP4-null mice had a 2.7-fold higher level of PPARg comparedwith that of WT mice (Fig. 2A and B). Furthermore,immunoblotting of extracts from immortalized WT andFABP4-null mouse macrophages, as well as immortalizedFABP4-null macrophages, in which FABP4 expressionwas complemented, demonstrated that FABP4 deficiencyled to higher expression of PPARg, and that comple-mentation of FABP4 expression reduced the expression

Figure 1—FABP4 inhibition increases PPARg protein expression and activity in macrophage-like cells. A: Relative PPARg mRNA levels,isolated from differentiated macrophage-like THP-1 cells incubated with or without the FABP4 inhibitor BMS309403 (BMS, 25 mmol/L,48 h) and measured by qPCR (n = 6). B: THP-1 macrophage-like cells were similarly treated and whole cell lysates were immunoblottedwith antibodies against PPARg. The results shown are representative of three independent experiments. C: THP-1 macrophage-like cellswere transfected with nontargeting control (open bars) or FABP4 siRNA (closed bars) and then induced to differentiate with phorbolmyristic acid for 48 h. Shown are the relative levels of FABP4 mRNA and PPARg mRNA, isolated from the cell extracts and measured byqPCR (n = 6). D: Whole extracts from THP-1 cells treated as in C and immunoblotted with antibodies against FABP4, PPARg, and b-actin.E: Immunoblots of the indicated proteins isolated from control 3T3-L1 preadipocytes (Undiff.) or 3T3-L1 cells induced to differentiate for8 days (Diff.) or induced to differentiate in the presence of 25 mmol/L BMS309403 (Diff./BMS). The results shown are representative ofthree independent experiments. The lanes were run on the same gel but were noncontiguous. Molecular mass markers (kDa) are indicatedon the right sides of panels B, D, and E. Data are presented as means 6 SEM. NS, not significant. *P < 0.05; ***P < 0.001.


level of PPARg (Fig. 2C ). In contrast with the proteinlevel, FABP4 had no significant effect on the level ofPPARg mRNA in these cells (Fig. 2D). These resultssuggested that FABP4 regulates PPARg at the proteinlevel.

FABP4 Attenuates Adipogenesis

We then studied the role of FABP4 in adipogenesis. Uponinduction of differentiation, we found that the FABP4-null adipocytes accumulated a larger number and a largervolume of lipid droplets compared with those accumu-lated in WT adipocytes (Fig. 3A and B). The enhancedadipogenesis in the FABP4-null adipocytes was apparentwhen the differentiation was performed both in the ab-sence and presence of rosiglitazone, a proadipogenicTZD. Furthermore, FABP4-null adipocytes expressedhigher levels of PPARg compared with their WT coun-terparts (Fig. 3C and D). PPARg activity was also higherin the FABP4-null adipocytes, as evaluated by the level ofadiponectin, both in the absence and presence ofrosiglitazone (Fig. 3C, bottom). We therefore propose thatFABP4 attenuates the formation of fat cells by down-regulating the master regulator of adipogenesis, PPARg.We noticed that the FABP4-null adipocytes exhibitedstronger Oil Red O staining compared with rosiglitazone-treated WT adipocytes (Fig. 3B), suggesting that FABP4inhibition might be more effective in promoting insulinsensitivity than this antidiabetic drug.

Overexpression of FABP4 Reduces PPARg ProteinExpression and Activity

To further show the impact of FABP4 on PPARg levels,we overexpressed PPARg in NIH-3T3 cells, alone or in

combination with FABP4. Immunoblotting showedthat coexpression of FABP4 with PPARg reduced thelevel of PPARg by 70 6 2% (n = 3, P , 0.05) (Fig. 4A).A similar result was obtained upon transfection of thecells with an R126Q mutant of FABP4, which lacks theability to bind fatty acids (data not shown). This resultsuggested that the fatty acid binding capacity of FABP4is not required for its effect on PPARg levels. The re-duced expression of PPARg upon its coexpression withFABP4 took place only at the protein level as no signif-icant change in the expression of PPARg mRNA wasobserved (Fig. 4B).

We next tested whether the FABP4-mediated re-duction in PPARg expression also affected its trans-activation potential. NIH-3T3 cells were transfected withpPPRE-luc, a reporter gene containing three PPREsupstream of a luciferase gene, and either a control vector(pCS6) or the FABP4 expression vector pFABP4. Thecells were then incubated either with the PPARg ligandrosiglitazone or with a vehicle. Overexpression of FABP4significantly reduced the basal as well as the rosiglitazone-induced transcriptional activity of PPARg (Fig. 4C ). Thisresult indicates that FABP4 interferes with PPARgactivation both by its endogenous ligands and by itspharmacological ligands.

FABP4 Accelerates the Proteasomal Degradation ofPPARg

Because PPARg mRNA levels were not affected by FABP4silencing or inhibition (Fig. 1A and C ), we studied thepossible regulation of PPARg by FABP4 at the proteinlevel. To this end, we followed the rate of PPARg decay in

Figure 2—PPARg expression is higher in FABP4-null mice. A: Immunoblots of subcutaneous white adipose tissue lysates from FABP4-null (KO) and WT mice (showing four of eight mice per group) with antibodies to FABP4, PPARg, and b-actin. B: Densitometric analysis ofthe immunoblots of PPARg of the above adipose tissues. Data are presented as means 6 SEM of the densitometric readings (n = 8 miceper group). *P< 0.05. C: Immunoblots of PPARg and FABP4 in immortalized macrophage cell lines fromWT and FABP4-null mice and theFABP4-null macrophage cell line in which FABP4 expression was complemented. Data are representative of three independent experi-ments. D: Relative expression level of PPARgmRNA in FABP4-null and FABP4-complemented immortalized mouse macrophage cell lines(n = 3). NS, not significant; P = 0.48.


the presence of cycloheximide, an inhibitor of proteinsynthesis. Human 293T cells were transfected withpPPARg and either control vector (pCS6) or pFABP4.Cycloheximide was then added; whole cell extracts wereprepared at various times and immunoblotted witha PPARg antibody. Coexpression of FABP4 with PPARgreduced the half-life of PPARg at all time points (Fig. 5Aand B). Thus, it is likely that FABP4 accelerates the rateof PPARg degradation.

We then examined PPARg levels in the presence ofthe proteasome inhibitor MG262. Human HeLa andmouse NIH-3T3 cells were transfected with pPPARgand either pFABP4 or the control vector pCS6. MG262abolished the FABP4-mediated reduction of PPARglevels in both human and mouse cells (Fig. 5C and D).MG262 increased the level of PPARg even in theabsence of FABP4 (Fig. 5D), supporting an earlierreport that the basal turnover of PPARg is mediated by

proteasomal degradation (36). These experiments sug-gest that FABP4 accelerates the proteasomal degrada-tion of PPARg.

FABP4 Enhances the Ubiquitination of PPARg

Most proteasome substrates are ubiquitinated prior totheir degradation. To investigate whether FABP4 affectsthe ubiquitination of PPARg, we transiently expressedvarious combinations of pPPARg, a plasmid encodingHA-tagged ubiquitin (pHA-Ub), and pFABP4 in 293Tcells for 48 h, followed by addition of the proteasomeinhibitor MG262 for 17 h to preserve short-livedubiquitin conjugates. Immunoprecipitation with anti-PPARg antibody followed by immunoblotting with ananti-HA tag antibody revealed that cells transfectedwith pPPARg together with pHA-Ub expressed freePPARg, as well as high-molecular-weight HA-positivebands (Fig. 6A). Upon coexpression of PPARg and

Figure 3—FABP4-null mouse preadipocytes (from embryonic fibroblasts) undergo enhanced adipogenesis and express higherlevels of PPARg. WT and FABP4-null (KO) cells were induced to differentiate into adipocytes in the absence or presence ofrosiglitazone (Rosi, 10 mmol/L) for 9 days. A: Light microscopic images of undifferentiated WT preadipocytes and differentiatedWT and FABP4-null adipocytes. Notice the difference in the number and size of the oil droplets. B: Oil Red O staining of neutral lipidsin WT and FABP4-null adipocytes after differentiation as above in the absence or presence of rosiglitazone (Rosi). C: Immunoblotsof cell lysates derived from WT or FABP4-null adipocytes. Data shown in B and C are representative of three independentexperiments, each one performed in triplicate. D: Densitometric analysis of the immunoblots of PPARg, FABP4, and b-actin as shownin C. Open bars, WT adipocytes; closed bars, FABP4-null adipocytes. Data are presented as means 6 SEM of the densitometricreadings (n = 4). *P < 0.05; **P < 0.01.


FABP4, the level of free PPARg was reduced and ex-pression of the ubiquitin-conjugated PPARg practicallydisappeared (Fig. 6A). Overexpression of FABP4 ro-bustly increased the conjugation of HA-ubiquitin to

PPARg, as observed following inhibition of its pro-teasomal degradation with MG262 (Fig. 6A, comparelanes 5 and 6).

We were also able to detect endogenous untaggedubiquitin conjugates of PPARg using anti-ubiquitin an-tibody. Human 293T cells were cotransfected withpPPARg and either pFABP4 or a control vector. In theabsence of FABP4, monoubiquitinated PPARg was themain conjugate as determined by immunoprecipitationwith anti-PPARg antibody followed by immunoblottingwith anti-ubiquitin antibody (Fig. 6B, lane 2). Uponcoexpression of FABP4 and PPARg, the level of themonoubiquitinated PPARg was reduced, whereas thelevel of the polyubiquitinated forms of PPARg increased(Fig. 6B, compare lanes 2 and 3). As expected, MG262considerably elevated the polyubiquitination of PPARgand further reduced the level of the monoubiquitinatedPPARg (Fig. 6B, compare lanes 4 and 5 to 2 and 3).These experiments suggest that FABP4 triggers theproteasomal degradation of PPARg by increasing itspolyubiquitination. Unlike the case of PPARg, coex-pression of FABP4 with p53, a protein extensivelycharacterized as a target of ubiquitin-dependentproteasomal degradation (37), had no effect on theprotein level of p53, thereby demonstrating thespecific effect of FABP4 on PPARg degradation(Fig. 6C).

FABP4 Is Increased and PPARg Is Reduced in MouseEpididymal Fat Compared With Subcutaneous Fat

Because both FABP4 and visceral fat are implicated inobesity-related morbidities, we compared the expressionof FABP4 and PPARg in adipose tissues of mouse sub-cutaneous fat and epididymal fat, the latter resemblingvisceral fat in humans. In both WT and FABP4-nullmouse strains, expression of FABP4 was higher in epi-didymal fat compared with subcutaneous fat, in corre-lation with reduced amounts of PPARg (Fig. 7A and B).Similarly, PPARg expression in adipose tissues of WTmice was reduced compared with its level in FABP4-nullfat (Fig. 7A and B). These findings further confirm therole of FABP4 as a negative regulator of its master reg-ulator PPARg and demonstrate how these two genes aredifferentially expressed in the functionally distinct fatdepots. Although the level of PPARg in FABP4-null epi-didymal fat appears to be somewhat lower comparedwith that in subcutaneous fat, analysis of paired samplesdid not show a statistically significant difference betweenthe levels of PPARg in the mouse epididymal and sub-cutaneous fat (n = 4 pairs, P = 0.602).

FABP4 Is Upregulated in Human Visceral FatCompared With Subcutaneous Fat

We then measured the relative amount of FABP4 andPPARg in paired samples of human VAT and SATobtained from biopsies of lean (6), obese (11), and obesediabetic (9) individuals (Fig. 7C and D). VAT FABP4 of

Figure 4—Coexpression of PPARg and FABP4 reduces PPARglevels and activity in NIH-3T3 cells. A: Immunoblots of extractsfrom NIH-3T3 cells transiently transfected with the indicated ex-pression plasmids for 48 h. Wherever required, control vectorpCS6 was added to obtain identical amounts of DNA vector.Immunoblots were probed with antibodies to FABP4, PPARg, andb-actin. B: Relative expression level of PPARg mRNA in NIH-3T3cells cotransfected with the indicated combinations of controlpCS6 vector, pFABP4, and pPPARg (n = 3). NS, not significant;P = 0.865. C: Normalized luciferase activity of extracts from NIH-3T3 cells transfected for 72 h with pPPRE-luc, pRL-TK Renilla luc,and either pPPARg+pCS6 (open bars) or pPPARg+pFABP4(closed bars). Vehicle (control) or rosiglitazone (10 mmol/L) wereadded 24 h after initiation of the transfection. Data are presentedas means 6 SEM (n = 3). ***P < 0.001.


each individual was normalized to its SAT FABP4 level.No significant difference in the VAT/SAT ratio of FABP4was observed upon comparing lean and healthy obeseindividuals. However, the VAT/SAT ratio of FABP4 wassignificantly higher in obese diabetic patients comparedwith lean or healthy obese individuals (Fig. 7D). Consis-tent with our previous results, PPARg was negativelycorrelated with FABP4 and was higher in the human SAT,compared with the human VAT (Fig. 7E).

DISCUSSION

The fatty acid chaperone FABP4 is induced by PPARg,the master regulator of adipogenesis, yet these two reg-ulators exert opposite effects on various metabolicparameters such as insulin resistance and inflammation(14). Here we demonstrate a previously unrecognizednegative feedback loop, whereby FABP4 specifically trig-gers proteasomal degradation of PPARg and conse-quently inhibits PPARg-related functions, therebyproviding a possible mechanism that explains their op-posite effects. Our observations are consistent with anearlier finding that PPARg activity is elevated in FABP4-null macrophages (20). FABP4 was reported to physicallyinteract with PPARg (38); hence, it is likely that sucha physical interaction triggers the ubiquitination and thesubsequent proteasomal degradation of PPARg. This in-teraction took place through a site distinct from the fattyacid binding pocket of FABP4 (38), and therefore theability of the fatty acid binding inhibitor BMS309403 toblock the effect of FABP4 on PPARg might be the out-come of an indirect allosteric effect.

Our findings show that FABP4 expression is not justan outcome of adipogenesis. Rather, FABP4 negativelyregulates PPARg and preadipocyte differentiation. Ourobservation that FABP4-null adipocytes express higherPPARg levels may explain their improved differentiationcapacity. Expression array analysis of many cell typessuggests that the changes in gene expression induced byTZDs take place mostly in adipocytes (39) as part of a denovo adipogenesis (40). Evidently, the influence ofFABP4 on adipogenesis was far more extensive than thatof rosiglitazone.

The higher level of FABP4 and the lower level ofPPARg in VAT compared with SAT suggest a causativelink between VAT FABP4 and the metabolic syndromeand may explain some morphological and functionaldifferences between the adipocyte subtypes comprisingthese two types of fat tissue. For instance, visceral pre-adipocytes proliferate and differentiate to mature adipo-cytes less readily than their subcutaneous counterparts.Visceral adipocytes are smaller and less responsive toTZD compared with subcutaneous adipocytes (41), pos-sibly due to the reduced PPARg. Furthermore, the VATsecretes higher levels of proinflammatory cytokines(8,42,43).

Previous studies reported similar levels of FABP4mRNA in human VAT and SAT (6,44,45). Our findingsthat FABP4 protein is higher in both human VAT andmouse intra-abdominal fat suggest that FABP4 may alsobe regulated posttranscriptionally. PPARg mRNA levelswere reported in paired samples of human VAT and SAT,

Figure 5—FABP4 accelerates proteasomal degradation of PPARg. A: Human 293T cells were transfected with pPPARg and eithercontrol pCS6 vector or pFABP4. After 48 h, the cells were treated with cycloheximide (350 mmol/L) and harvested at the indicated timesafter cycloheximide addition. Immunoblots of cell extracts with antibodies directed against PPARg, FABP4, and b-actin are shown.B: Densitometric analysis of the bands shown in A, relative to time 0. Data are presented as means 6 SEM of the densitometric readings(n = 3). *P < 0.05. C: HeLa cells were transfected with pPPARg and either pCS6 or pFABP4. After 48 h, the proteasome inhibitor MG262(0.1 mmol/L) was added for 17 h, and cell extracts were then prepared and immunoblotted with antibodies to PPARg, FABP4, and b-actin.D: Mouse NIH-3T3 cells were transfected and treated with MG262, and extracts were immunoblotted as in C. Data shown arerepresentative of three independent experiments.


but the data are rather inconsistent. In contrast, and inline with our observations, comparative data at theprotein level consistently showed reduced levels and ac-tivity of PPARg in human VAT (46–48), further sup-porting the notion of posttranslational regulation ofPPARg by FABP4. Our finding of higher FABP4 levels inVAT relative to their levels in SAT in obese diabeticindividuals may explain why obesity-associated pathologycorrelates better with visceral fat mass than with overalladiposity (4–8).

Our finding that FABP4 interferes with level andhence activity of PPARg in adipocytes and macrophagesmay be critical to understanding the development of themetabolic syndrome and its related pathologies, such astype 2 diabetes and atherosclerosis. We hypothesize thatlipid accumulation elevates FABP4, thereby reducingPPARg. In view of the impact of PPARg on insulin sen-sitivity, our findings may explain the progression frominsulin resistance to an overt diabetic state when FABP4

levels increase, thereby compromising PPARg levelsand activities. FABP4 also promotes a proinflammatorystate, further advancing the development of metabolicsyndrome–associated pathologies. A unique populationof VAT-resident regulatory T cells was recently impli-cated in control of the inflammatory state of adiposetissue (49). PPARg is a crucial regulator of this regulatoryT-cell population, necessary for complete restoration ofinsulin sensitivity by TZD in obese mice.

Pharmacologic activators of PPARg, such as TZDs,significantly improve insulin sensitivity in type 2 di-abetes, and the net effect of PPARg activation in murinemacrophages is atheroprotective (13). However, TZDstrigger rapid degradation of PPARg (36), possibly ac-counting for the recent reports implicating them in in-creased risk of cardiovascular death (50). We proposethat similarly to TZD, FABP4 lowers PPARg levels inmacrophages, thereby triggering its adverse physiologicaleffects (11). In this respect, inhibition of FABP4,

Figure 6—FABP4 enhances ubiquitination of PPARg. A: 293T cells were transfected with the indicated expression vectors. Whereverrequired, control vector pCS6 was added to obtain identical amounts of vector DNA. After 48 h, the cells were treated with vehicle orMG262 (0.1 mmol/L; 17 h), cell lysates were immunoprecipitated with an anti-PPARg antibody, and the immunoprecipitates wereimmunoblotted with antibodies directed against HA-ubiquitin (top) or PPARg (bottom). The lanes were run on the same gel but werenoncontiguous. B: 293T cells were transiently transfected with pPPARg and either pFABP4 or an empty vector. After 48 h, the cells weretreated with MG262 as above, cell lysates were immunoprecipitated with an anti-PPARg antibody, and the immunoprecipitates wereimmunoblotted with the anti-ubiquitin antibody. C: 293T cells were transiently transfected with pPPARg or pp53 and either pFABP4 or anempty vector. After 48 h, cell lysates were immunoblotted with the anti-PPARg or anti-p53 antibodies. Data shown in A–C are repre-sentative of three independent experiments.


combined with activation of PPARg by TZDs could at-tenuate the harmful effects of TZDs, while maintainingtheir beneficial activities.

Acknowledgments. The authors acknowledge the help and advice ofDaniela Novick and Sara Barak of the Weizmann Institute.

Funding. This work was supported by research grants from the estates ofFannie Sherr and Helena Barkman Schramm and from the Nissim Center. M.R.is the Edna and Maurice Weiss Professor of Cytokine Research.

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

Author Contributions. T.G.-S. designed and performed experimentsand wrote the manuscript. A.R. and G.S.H. provided samples, analyzedresults, and reviewed the manuscript. M.R. designed experiments, analyzedresults, and wrote the manuscript. T.G.-S. is the guarantor of this work and, assuch, had full access to all the data in the study and takes responsibility for theintegrity of the data and the accuracy of the data analysis.

References1. Berk-Planken II, Hoogerbrugge N, Stolk RP, Bootsma AH, Jansen H; DALI

Study Group. Atorvastatin dose-dependently decreases hepatic lipase ac-

tivity in type 2 diabetes: effect of sex and the LIPC promoter variant.

Diabetes Care 2003;26:427–432

2. Schuelke M, Krude H, Finckh B, et al. Septo-optic dysplasia associated

with a new mitochondrial cytochrome b mutation. Ann Neurol 2002;51:

388–392

3. Calle EE, Rodriguez C, Walker-Thurmond K, Thun MJ. Overweight, obesity,

and mortality from cancer in a prospectively studied cohort of U.S. adults.

N Engl J Med 2003;348:1625–1638

4. Goodpaster BH, Krishnaswami S, Resnick H, et al. Association between

regional adipose tissue distribution and both type 2 diabetes and impaired

glucose tolerance in elderly men and women. Diabetes Care 2003;26:

372–379

5. Rodríguez A, Catalán V, Gómez-Ambrosi J, Frühbeck G. Visceral and

subcutaneous adiposity: are both potential therapeutic targets for

Figure 7—FABP4 and PPARg levels are inversely correlated in mouse and human adipose tissues. A: Immunoblots of PPARg, FABP4,and b-actin from paired subcutaneous (S) and epididymal ( E ) fat of WT and FABP4-null mice. Data shown are representative of fourindependent experiments. B: Densitometric analysis of the immunoblots of PPARg and FABP4 from paired SAT (open bars) andepididymal adipose tissue (closed bars) from WT mice. Data shown are means 6 SEM of the densitometric readings (n = 4).C: Immunoblots of PPARg and FABP4 from paired human subcutaneous (S) and visceral fat (V). Shown are immunoblots of pairedSAT and VAT lysates representing 28 individuals. D: FABP4 ratios in SAT and VAT as determined by densitometric analysis ofthe FABP4 immunoblots of adipose tissues from lean (6), obese (11), and obese diabetic (9) individuals. Data are means 6 SEM ofthe densitometric readings. E: PPARg levels in SAT and VAT as determined by densitometric analysis of the PPARg immunoblots(n = 28). Densitometric values in B–E were normalized to b-actin. All data are presented as means 6 SEM of the densitometricreadings. ***P < 0.001.


tackling the metabolic syndrome? Curr Pharm Des 2007;13:2169–2175

6. van Beek EA, Bakker AH, Kruyt PM, Hofker MH, Saris WH, Keijer J. Intra-and interindividual variation in gene expression in human adipose tissue.Pflugers Arch 2007;453:851–861

7. Wajchenberg BL. Subcutaneous and visceral adipose tissue: their relationto the metabolic syndrome. Endocr Rev 2000;21:697–738

8. Yang X, Smith U. Adipose tissue distribution and risk of metabolic disease:does thiazolidinedione-induced adipose tissue redistribution provide a clueto the answer? Diabetologia 2007;50:1127–1139

9. Tontonoz P, Hu E, Spiegelman BM. Stimulation of adipogenesis in fibro-blasts by PPAR gamma 2, a lipid-activated transcription factor. Cell 1994;79:1147–1156

10. Lehmann JM, Moore LB, Smith-Oliver TA, Wilkison WO, Willson TM,Kliewer SA. An antidiabetic thiazolidinedione is a high affinity ligand forperoxisome proliferator-activated receptor gamma (PPAR gamma). J BiolChem 1995;270:12953–12956

11. Tontonoz P, Spiegelman BM. Fat and beyond: the diverse biology ofPPARgamma. Annu Rev Biochem 2008;77:289–312

12. Odegaard JI, Ricardo-Gonzalez RR, Goforth MH, et al. Macrophage-specificPPARgamma controls alternative activation and improves insulin resis-tance. Nature 2007;447:1116–1120

13. Chawla A, Boisvert WA, Lee CH, et al. A PPAR gamma-LXR-ABCA1 path-way in macrophages is involved in cholesterol efflux and atherogenesis.Mol Cell 2001;7:161–171

14. Tontonoz P, Graves RA, Budavari AI, et al. Adipocyte-specific transcrip-tion factor ARF6 is a heterodimeric complex of two nuclear hormonereceptors, PPAR gamma and RXR alpha. Nucleic Acids Res 1994;22:5628–5634

15. Furuhashi M, Hotamisligil GS. Fatty acid-binding proteins: role in metabolicdiseases and potential as drug targets. Nat Rev Drug Discov 2008;7:489–503

16. Boord JB, Fazio S, Linton MF. Cytoplasmic fatty acid-binding proteins:emerging roles in metabolism and atherosclerosis. Curr Opin Lipidol 2002;13:141–147

17. Maeda K, Cao H, Kono K, et al. Adipocyte/macrophage fatty acid bindingproteins control integrated metabolic responses in obesity and diabetes.Cell Metab 2005;1:107–119

18. Hotamisligil GS, Johnson RS, Distel RJ, Ellis R, Papaioannou VE,Spiegelman BM. Uncoupling of obesity from insulin resistance througha targeted mutation in aP2, the adipocyte fatty acid binding protein. Sci-ence 1996;274:1377–1379

19. Makowski L, Boord JB, Maeda K, et al. Lack of macrophage fatty-acid-binding protein aP2 protects mice deficient in apolipoprotein E againstatherosclerosis. Nat Med 2001;7:699–705

20. Makowski L, Brittingham KC, Reynolds JM, Suttles J, Hotamisligil GS. Thefatty acid-binding protein, aP2, coordinates macrophage cholesterol traf-ficking and inflammatory activity. Macrophage expression of aP2 impactsperoxisome proliferator-activated receptor gamma and IkappaB kinaseactivities. J Biol Chem 2005;280:12888–12895

21. Furuhashi M, Fucho R, Görgün CZ, Tuncman G, Cao H, Hotamisligil GS.Adipocyte/macrophage fatty acid-binding proteins contribute to metabolicdeterioration through actions in both macrophages and adipocytes in mice.J Clin Invest 2008;118:2640–2650

22. Furuhashi M, Tuncman G, Görgün CZ, et al. Treatment of diabetes andatherosclerosis by inhibiting fatty-acid-binding protein aP2. Nature 2007;447:959–965

23. Tuncman G, Erbay E, Hom X, et al. A genetic variant at the fatty acid-binding protein aP2 locus reduces the risk for hypertriglyceridemia, type 2diabetes, and cardiovascular disease. Proc Natl Acad Sci USA 2006;103:6970–6975

24. Stejskal D, Karpisek M. Adipocyte fatty acid binding protein in a Caucasianpopulation: a new marker of metabolic syndrome? Eur J Clin Invest 2006;36:621–625

25. Kaess BM, Enserro DM, McManus DD, et al. Cardiometabolic correlatesand heritability of fetuin-A, retinol-binding protein 4, and fatty-acid bindingprotein 4 in the Framingham Heart Study. J Clin Endocrinol Metab 2012;97:E1943–E1947

26. Hancke K, Grubeck D, Hauser N, Kreienberg R, Weiss JM. Adipocyte fattyacid-binding protein as a novel prognostic factor in obese breast cancerpatients. Breast Cancer Res Treat 2010;119:367–377

27. Nieman KM, Kenny HA, Penicka CV, et al. Adipocytes promote ovariancancer metastasis and provide energy for rapid tumor growth. Nat Med2011;17:1498–1503

28. Kershaw EE, Flier JS. Adipose tissue as an endocrine organ. J Clin En-docrinol Metab 2004;89:2548–2556

29. Tchernof A. Visceral adipocytes and the metabolic syndrome. Nutr Rev2007;65:S24–S29

30. Masuzaki H, Paterson J, Shinyama H, et al. A transgenic model of visceralobesity and the metabolic syndrome. Science 2001;294:2166–2170

31. Després JP, Lemieux I. Abdominal obesity and metabolic syndrome. Nature2006;444:881–887

32. Sethi JK, Xu H, Uysal KT, Wiesbrock SM, Scheja L, Hotamisligil GS.Characterisation of receptor-specific TNFalpha functions in adipocyte celllines lacking type 1 and 2 TNF receptors. FEBS Lett 2000;469:77–82

33. MacDougald OA, Lane MD. Transcriptional regulation of gene expressionduring adipocyte differentiation. Annu Rev Biochem 1995;64:345–373

34. Harman-Boehm I, Blüher M, Redel H, et al. Macrophage infiltration intoomental versus subcutaneous fat across different populations: effect ofregional adiposity and the comorbidities of obesity. J Clin Endocrinol Metab2007;92:2240–2247

35. Rosen ED, Sarraf P, Troy AE, et al. PPAR gamma is required for thedifferentiation of adipose tissue in vivo and in vitro. Mol Cell 1999;4:611–617

36. Hauser S, Adelmant G, Sarraf P, Wright HM, Mueller E, Spiegelman BM.Degradation of the peroxisome proliferator-activated receptor gamma islinked to ligand-dependent activation. J Biol Chem 2000;275:18527–18533

37. Leng RP, Lin Y, Ma W, et al. Pirh2, a p53-induced ubiquitin-protein ligase,promotes p53 degradation. Cell 2003;112:779–791

38. Adida A, Spener F. Adipocyte-type fatty acid-binding protein as inter-compartmental shuttle for peroxisome proliferator activated receptorgamma agonists in cultured cell. Biochim Biophys Acta 2006;1761:172–181

39. Way JM, Harrington WW, Brown KK, et al. Comprehensive messengerribonucleic acid profiling reveals that peroxisome proliferator-activatedreceptor gamma activation has coordinate effects on gene expression inmultiple insulin-sensitive tissues. Endocrinology 2001;142:1269–1277

40. Memon RA, Tecott LH, Nonogaki K, et al. Up-regulation of peroxisomeproliferator-activated receptors (PPAR-alpha) and PPAR-gamma messen-ger ribonucleic acid expression in the liver in murine obesity: troglitazoneinduces expression of PPAR-gamma-responsive adipose tissue-specificgenes in the liver of obese diabetic mice. Endocrinology 2000;141:4021–4031


41. Adams M, Reginato MJ, Shao D, Lazar MA, Chatterjee VK. Transcriptionalactivation by peroxisome proliferator-activated receptor gamma is inhibitedby phosphorylation at a consensus mitogen-activated protein kinase site.J Biol Chem 1997;272:5128–5132

42. Xu H, Barnes GT, Yang Q, et al. Chronic inflammation in fat plays a crucialrole in the development of obesity-related insulin resistance. J Clin Invest2003;112:1821–1830

43. Wellen KE, Hotamisligil GS. Inflammation, stress, and diabetes. J ClinInvest 2005;115:1111–1119

44. Poulain-Godefroy O, Lecoeur C, Pattou F, Frühbeck G, Froguel P. Inflam-mation is associated with a decrease of lipogenic factors in omental fat inwomen. Am J Physiol Regul Integr Comp Physiol 2008;295:R1–R7

45. Vural B, Atalar F, Ciftci C, et al. Presence of fatty-acid-binding protein 4expression in human epicardial adipose tissue in metabolic syndrome.Cardiovasc Pathol 2008;17:392–398

46. Tchkonia T, Giorgadze N, Pirtskhalava T, et al. Fat depot origin affectsadipogenesis in primary cultured and cloned human preadipocytes. Am JPhysiol Regul Integr Comp Physiol 2002;282:R1286–R1296

47. Sauma L, Franck N, Paulsson JF, et al. Peroxisome proliferator activatedreceptor gamma activity is low in mature primary human visceral adipo-cytes. Diabetologia 2007;50:195–201

48. Walker GE, Marzullo P, Verti B, et al. Subcutaneous abdominal adiposetissue subcompartments: potential role in rosiglitazone effects. Obesity(Silver Spring) 2008;16:1983–1991

49. Cipolletta D, Feuerer M, Li A, et al. PPAR-g is a major driver of theaccumulation and phenotype of adipose tissue Treg cells. Nature 2012;486:549–553

50. Nissen SE, Wolski K. Effect of rosiglitazone on the risk of myocardialinfarction and death from cardiovascular causes. N Engl J Med 2007;356:2457–2471


fabp4 attenuates pparg and adipogenesis and is inversely ...adipogenesis and is inversely correlated...

Documents