FABP4 Attenuates PPARγ and Adipogenesis and Is Inversely Correlated With PPARγ in Adipose Tissues

Adiposity is closely correlated with important physiological parameters such as blood pressure, systemic insulin sensitivity, dyslipidemia, and serum triglyceride levels (1,2), rendering obesity to an independent risk factor for myocardial infarction, stroke, type 2 diabetes, and certain cancers (3). Among adipose tissues, visceral fat is more closely correlated with obesity-associated pathologies than overall adiposity (4–8).

The nuclear receptor peroxisome proliferator–activated receptor γ (PPARγ) is a master regulator of adipose cell differentiation, playing a critical role in systemic lipid and glucose metabolism (9). PPARγ is activated by natural or synthetic agonists such as the antidiabetic thiazolidinedione (TZD) (10). Activated PPARγ is a master regulator of adipogenesis, acting as a transcription factor of genes expressed in mature adipocytes, including fatty acid binding protein (FABP4), CD36, lipoprotein lipase (LPL), and adiponectin, all of which contain peroxisome proliferator response elements (PPREs) (11). PPARγ is also expressed in myeloid cells, and its activation promotes an anti-inflammatory phenotype (11). Disruption of PPARγ specifically in myeloid cells also predisposes mice to the development of diet-induced obesity, insulin resistance, and glucose intolerance (12), whereas activation of PPARγ within macrophages promotes lipid efflux, thereby stabilizing atherosclerotic lesions (13).

A major target gene of PPARγ is the lipid transporter FABP4, also known as aP2 (14). PPARγ induces FABP4 almost exclusively in adipocytes and macrophages. FABP4 acts as a fatty acids chaperone, which couples intracellular lipids to biological targets and signaling pathways (15,16). FABP4 has been implicated in several aspects of the metabolic syndrome in mice, including insulin resistance and atherosclerosis (17–22). In humans, a T87C polymorphism in the FABP4 promoter leads to reduced transcriptional activity, resulting in lower serum triglycerides and a reduced risk of atherosclerosis and type 2 diabetes (23). In addition, the level of circulating human FABP4 was proposed as an independent prognostic marker for future development of diabetes (24,25). Furthermore, PPARγ is considered a tumor suppressor gene (11), whereas FABP4 has a role in tumorigenesis (26,27).

Understanding the mechanisms by which obesity alters adipose tissue biology is of major importance because the fat tissue acts as an endocrine organ that regulates systemic metabolism (28). Visceral adipose tissues (VATs) and subcutaneous adipose tissues (SATs) are functionally distinct, and many clinical studies led to the conclusion that VAT, rather than SAT, is the key factor inducing the metabolic syndrome, also serving as a source of inflammatory and stress-promoting factors (8,29–31).

To date, the mechanisms by which FABP4 promotes insulin resistance and inflammation are not fully understood. The fact that FABP4 is induced by PPARγ but exerts metabolic and immunologic activities opposite to those of PPARγ led us to hypothesize that FABP4 might be involved in regulating PPARγ activity. Here we show that FABP4 negatively regulates PPARγ levels in macrophages and adipocytes, thereby attenuating adipocyte differentiation; we elucidate a mechanism involving proteasomal PPARγ degradation and further demonstrate elevated FABP4/PPARγ ratio in intra-abdominal fat compared with SAT in mice and VAT compared with SAT in humans.

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) were from the American Type Culture Collection. Immortalized wild-type (WT) and FABP4-null preadipocytes, as well as immortalized WT, FABP4-null, and FABP4-complemented macrophage cell lines were generated from WT and FABP4-null mice as previously described (22,32). 3T3-L1 cells and preadipocytes from FABP4 knockout and WT mice were grown in Dulbecco’s modified Eagle’s medium (DMEM) containing 10% bovine serum. THP-1 cells and immortalized macrophage cell lines were grown in RPMI 1640 medium containing 10% FBS, 2 mmol/L l-glutamine, 1.5 g/L sodium bicarbonate, 4.5 g/L glucose, 10 mmol/L HEPES, and 1 mmol/L sodium pyruvate. All other cell lines were grown in DMEM with 10% FBS (Gibco), 100 IU/mL penicillin, and 0.1 mg/mL streptomycin. The FABP4 inhibitor BMS309403 and the proteasome inhibitor MG262 were from Calbiochem. Rosiglitazone was from Cayman Chemical. JetPEI transfection reagent was from Polyplus Transfection. Small interfering RNA (siRNA) oligonucleotides were from Dharmacon. Cell Line Nucleofector Kit V was used for siRNA transfections of THP-1 cells (Amaxa GmbH, Cologne, Germany). All other reagents were from Sigma-Aldrich.

THP-1 cells (1 × 10⁶/mL) were induced to differentiate by phorbol myristic acid (100 nmol/L, 24 h). For experiments in which BMS309403 was included, the medium was supplemented with 10% lipoprotein-deficient serum instead of 10% FBS. 3T3-L1 preadipocytes were differentiated as previously described (33). In brief, cells were subcultured at <70% confluence. To induce differentiation, confluent cell cultures were placed in medium containing 0.5 mmol/L 1-methyl-3-isobutylxanthine, 1 μmol/L dexamethasone, and 5 μg/mL insulin in 10% FBS/DMEM for 48 h. Next, the medium was changed, and 5 μg/mL insulin was added for another 48 h. The medium was then replaced with 10% FBS/DMEM and was renewed every 2 days until the cells became fully differentiated and lipid droplets were apparent.

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

Cells at 50–70% confluence were transfected using the jetPEI reagent and the indicated plasmids, according to the manufacturer's instructions. Wherever needed, an empty vector was added to maintain a constant amount of 5 µg DNA per well. pCS6 (empty vector), pPPARγ, and pFABP4 were from Open Biosystems. The following vectors were provided by the indicated person: pp53, Y. Shaul, Weizmann Institute, Rehovot, Israel; pPPRE-luc, the laboratory of the late M. Liscovitch, Weizmann Institute; and pHA-Ub, C. Kahana, Weizmann Institute. THP1 cells were transiently transfected using the Amaxa Nucleofection Technology (Amaxa GmbH, Cologne, Germany) before differentiation.

FABP4-null mice were generated as previously described (22). Biopsies from epididymal and subcutaneous depots of 12-week-old mice were removed, frozen in liquid nitrogen, and stored at –80°C. Fat tissues were homogenized 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 protease inhibitors (1:1,000; Sigma-Aldrich) and 1 mmol/L phenylmethylsulfonyl fluoride (PMSF). The institutional ethics committee approved in advance all protocols of the study.

Paired human VAT and abdominal SAT biopsies were obtained during elective abdominal surgery for gastric banding, weight reduction surgery, or exploratory laparotomy (with negative findings) as previously described (34). The institutional ethics committee approved in advance all protocols of the study, and all participants provided a written informed consent after all objectives and procedures were explained. Biopsies were delivered on ice, rinsed in saline, frozen in liquid nitrogen, powdered, and then homogenized in lysis buffer (150 mmol/L NaCl, 50 mmol/L Tris-HCl, pH 7.5, 1% by volume Nonidet P-40, 0.25% sodium deoxycholate, 10 mmol/L sodium β-glycerophosphate, 5 mmol/L sodium pyrophosphate, 1 mmol/L EGTA, 1 mmol/L sodium vanadate, and 1 mmol/L NaF), supplemented with protease inhibitor cocktail (1:1,000).

Cell pellets were resuspended on ice for 20 min in a buffer consisting of 100 mmol/L KCl, 0.5 mmol/L EDTA, 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 −80°C. Protein concentration was determined by BCA Protein Assay Reagent 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 Biomedicals), anti-FABP4, anti-PPARγ, anti-LPL (Santa Cruz Biotechnology), anti-HA (Covance), and anti-ubiquitinated proteins (clone FK2; BIOMOL International). Mouse monoclonal anti-p53 was a gift of C. Prives (Columbia University, New York, NY). Detection was performed using horseradish peroxidase–conjugated secondary antibodies (Jackson ImmunoResearch Laboratories), followed by SuperSignal chemiluminescence detection system (Pierce). Quantitative densitometry was performed with EZ-Quant software. Protein levels were normalized to values obtained for β-actin.

Cells were washed with ice-cold PBS and lysed with PBS containing 1% Triton X-100, 0.5% sodium deoxycholate, 10 mmol/L sodium pyrophosphate, 2 mmol/L sodium vanadate, 100 mmol/L sodium fluoride, protease inhibitors mixture, and 1 mmol/L PMSF. Samples (500 μg protein) were precleared by adding 60 μL protein G-Sepharose (1 h, 4°C; GE Healthcare, Uppsala, Sweden). Anti-PPARγ antibody (1:100 in lysis buffer) was added for 17 h at 4°C, and protein G-Sepharose was then added for another hour. The beads were then washed four times with 1% Nonidet P-40 in PBS, extracted with sample buffer, and analyzed by SDS-PAGE and immunoblotting, as described above.

NIH-3T3 cells (10–17 × 10³ cells/well) in 96-well plates were transfected using the jetPEI reagent with the following plasmids: 3× PPRE-TK-luciferase reporter (67 ng), pPPARγ (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 10 μmol/L rosiglitazone or vehicle (DMSO) for 48 h. Cell lysates were then analyzed for luciferase activity and normalized to Renilla activity using the Dual-Luciferase Reporter Assay Kit (Promega), according to the manufacturer’s instructions.

Total RNA (1 μg), isolated using PerfectPure RNA Cultured Cell Kit (5 Prime), was reverse transcribed using random hexamers and High Capacity RNA-to-cDNA Kit (Applied Biosystems). The reaction mixtures were diluted 15-fold, and 4.5 μL was then used as the template for real-time PCR. The following assays were performed: mFABP4, Mn00445880_m1; mPPARγ, Mn00440945_m1; and mTBP, Mm00446973_m1, as a loading control. These probes were designed using Roche ProbeFinder Software (Roche Diagnostics). Quantitative real-time PCR (qPCR) was performed using a Roche LightCycler 480 Real-Time PCR System and TaqMan Universal PCR Master Mix (Applied Biosystems). The amplification program was as follows: 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 normalized to TBP mRNA and was calculated using LightCycler Software (Roche). The results are means of fold change ± SEM of triplicates.

THP-1 cells were transfected using the Cell Line Nucleofector Kit V (Amaxa) with either siFABP4 siGENOME SMARTpool (140 nmol/L) or Non-Targeting siRNA Pool (140 nmol/L; Dharmacon), using the Amaxa Nucleofection Technology, program V-01. After 24 h, the cells were induced to differentiate to macrophages and the levels of FABP4 and PPARγ mRNA were determined by qPCR.

ANOVA and Student t test were performed for statistical analysis, as appropriate. Statistical significance was designated at P < 0.05. Values are expressed as mean ± SEM.

THP-1 cells were induced to differentiate into macrophage-like cells and then incubated with either vehicle or the FABP4 inhibitor BMS309403. Inhibition of FABP4 did not affect the level of PPARγ mRNA (Fig. 1A) but elevated the basal level of PPARγ protein by 2.6 ± 0.8–fold (P < 0.005, n = 3), as determined by immunoblotting and densitometry (Fig. 1B). To further study the effect of FABP4 on PPARγ, we knocked down FABP4 mRNA in THP-1 cells by FABP4-targeting siRNA or nontargeting control siRNA. After 24 h, the cells were differentiated to macrophages, and after an additional 48 h, they were analyzed for the protein and the mRNA levels of both FABP4 and PPARγ. FABP4 siRNA effectively inhibited the expression of FABP4 mRNA and protein (Fig. 1C and D) and had only a minor effect on the expression of PPARγ mRNA (Fig. 1C). In contrast, FABP4 knockdown increased the protein level of PPARγ by 170 ± 20% (n = 3, P < 0.01) (Fig. 1D). These results suggested that PPARγ is negatively regulated by FABP4 predominantly at the posttranscriptional level.

PPARγ is induced during the differentiation of preadipocytes into adipocytes and in turn induces a set of differentiation-dependent genes, including FABP4, LPL, and adiponectin (9,35). Despite the fact that FABP4 is a robust marker of mature adipocytes, addition of its inhibitor, BMS309403, to the adipogenic cocktail increased PPARγ protein expression by 3.9 ± 0.2–fold (n = 3, P < 0.05), as well as expression of the PPARγ gene products FABP4, LPL, and adiponectin (Fig. 1E). These findings suggested that FABP4 negatively regulates PPARγ-mediated gene activation during adipogenesis.

We then studied the interrelationships of FABP4 and PPARγ in vivo, comparing WT mice with FABP4-null mice (17). We found that white adipose tissue of FABP4-null mice had a 2.7-fold higher level of PPARγ compared with that of WT mice (Fig. 2A and B). Furthermore, immunoblotting of extracts from immortalized WT and FABP4-null mouse macrophages, as well as immortalized FABP4-null macrophages, in which FABP4 expression was complemented, demonstrated that FABP4 deficiency led to higher expression of PPARγ, and that complementation of FABP4 expression reduced the expression level of PPARγ (Fig. 2C). In contrast with the protein level, FABP4 had no significant effect on the level of PPARγ mRNA in these cells (Fig. 2D). These results suggested that FABP4 regulates PPARγ at the protein level.

We then studied the role of FABP4 in adipogenesis. Upon induction of differentiation, we found that the FABP4-null adipocytes accumulated a larger number and a larger volume of lipid droplets compared with those accumulated in WT adipocytes (Fig. 3A and B). The enhanced adipogenesis in the FABP4-null adipocytes was apparent when the differentiation was performed both in the absence and presence of rosiglitazone, a proadipogenic TZD. Furthermore, FABP4-null adipocytes expressed higher levels of PPARγ compared with their WT counterparts (Fig. 3C and D). PPARγ activity was also higher in the FABP4-null adipocytes, as evaluated by the level of adiponectin, both in the absence and presence of rosiglitazone (Fig. 3C, bottom). We therefore propose that FABP4 attenuates the formation of fat cells by downregulating the master regulator of adipogenesis, PPARγ. We noticed that the FABP4-null adipocytes exhibited stronger Oil Red O staining compared with rosiglitazone-treated WT adipocytes (Fig. 3B), suggesting that FABP4 inhibition might be more effective in promoting insulin sensitivity than this antidiabetic drug.

To further show the impact of FABP4 on PPARγ levels, we overexpressed PPARγ in NIH-3T3 cells, alone or in combination with FABP4. Immunoblotting showed that coexpression of FABP4 with PPARγ reduced the level of PPARγ by 70 ± 2% (n = 3, P < 0.05) (Fig. 4A). A similar result was obtained upon transfection of the cells with an R126Q mutant of FABP4, which lacks the ability to bind fatty acids (data not shown). This result suggested that the fatty acid binding capacity of FABP4 is not required for its effect on PPARγ levels. The reduced expression of PPARγ upon its coexpression with FABP4 took place only at the protein level as no significant change in the expression of PPARγ mRNA was observed (Fig. 4B).

We next tested whether the FABP4-mediated reduction in PPARγ expression also affected its transactivation potential. NIH-3T3 cells were transfected with pPPRE-luc, a reporter gene containing three PPREs upstream of a luciferase gene, and either a control vector (pCS6) or the FABP4 expression vector pFABP4. The cells were then incubated either with the PPARγ ligand rosiglitazone or with a vehicle. Overexpression of FABP4 significantly reduced the basal as well as the rosiglitazone-induced transcriptional activity of PPARγ (Fig. 4C). This result indicates that FABP4 interferes with PPARγ activation both by its endogenous ligands and by its pharmacological ligands.

Because PPARγ mRNA levels were not affected by FABP4 silencing or inhibition (Fig. 1A and C), we studied the possible regulation of PPARγ by FABP4 at the protein level. To this end, we followed the rate of PPARγ decay in the presence of cycloheximide, an inhibitor of protein synthesis. Human 293T cells were transfected with pPPARγ and either control vector (pCS6) or pFABP4. Cycloheximide was then added; whole cell extracts were prepared at various times and immunoblotted with a PPARγ antibody. Coexpression of FABP4 with PPARγ reduced the half-life of PPARγ at all time points (Fig. 5A and B). Thus, it is likely that FABP4 accelerates the rate of PPARγ degradation.

We then examined PPARγ levels in the presence of the proteasome inhibitor MG262. Human HeLa and mouse NIH-3T3 cells were transfected with pPPARγ and either pFABP4 or the control vector pCS6. MG262 abolished the FABP4-mediated reduction of PPARγ levels in both human and mouse cells (Fig. 5C and D). MG262 increased the level of PPARγ even in the absence of FABP4 (Fig. 5D), supporting an earlier report that the basal turnover of PPARγ is mediated by proteasomal degradation (36). These experiments suggest that FABP4 accelerates the proteasomal degradation of PPARγ.

Most proteasome substrates are ubiquitinated prior to their degradation. To investigate whether FABP4 affects the ubiquitination of PPARγ, we transiently expressed various combinations of pPPARγ, a plasmid encoding HA-tagged ubiquitin (pHA-Ub), and pFABP4 in 293T cells for 48 h, followed by addition of the proteasome inhibitor MG262 for 17 h to preserve short-lived ubiquitin conjugates. Immunoprecipitation with anti-PPARγ antibody followed by immunoblotting with an anti-HA tag antibody revealed that cells transfected with pPPARγ together with pHA-Ub expressed free PPARγ, as well as high-molecular-weight HA-positive bands (Fig. 6A). Upon coexpression of PPARγ and FABP4, the level of free PPARγ was reduced and expression of the ubiquitin-conjugated PPARγ practically disappeared (Fig. 6A). Overexpression of FABP4 robustly increased the conjugation of HA-ubiquitin to PPARγ, as observed following inhibition of its proteasomal degradation with MG262 (Fig. 6A, compare lanes 5 and 6).

We were also able to detect endogenous untagged ubiquitin conjugates of PPARγ using anti-ubiquitin antibody. Human 293T cells were cotransfected with pPPARγ and either pFABP4 or a control vector. In the absence of FABP4, monoubiquitinated PPARγ was the main conjugate as determined by immunoprecipitation with anti-PPARγ antibody followed by immunoblotting with anti-ubiquitin antibody (Fig. 6B, lane 2). Upon coexpression of FABP4 and PPARγ, the level of the monoubiquitinated PPARγ was reduced, whereas the level of the polyubiquitinated forms of PPARγ increased (Fig. 6B, compare lanes 2 and 3). As expected, MG262 considerably elevated the polyubiquitination of PPARγ and further reduced the level of the monoubiquitinated PPARγ (Fig. 6B, compare lanes 4 and 5 to 2 and 3). These experiments suggest that FABP4 triggers the proteasomal degradation of PPARγ by increasing its polyubiquitination. Unlike the case of PPARγ, coexpression of FABP4 with p53, a protein extensively characterized as a target of ubiquitin-dependent proteasomal degradation (37), had no effect on the protein level of p53, thereby demonstrating the specific effect of FABP4 on PPARγ degradation (Fig. 6C).

Because both FABP4 and visceral fat are implicated in obesity-related morbidities, we compared the expression of FABP4 and PPARγ in adipose tissues of mouse subcutaneous fat and epididymal fat, the latter resembling visceral fat in humans. In both WT and FABP4-null mouse strains, expression of FABP4 was higher in epididymal fat compared with subcutaneous fat, in correlation with reduced amounts of PPARγ (Fig. 7A and B). Similarly, PPARγ expression in adipose tissues of WT mice was reduced compared with its level in FABP4-null fat (Fig. 7A and B). These findings further confirm the role of FABP4 as a negative regulator of its master regulator PPARγ and demonstrate how these two genes are differentially expressed in the functionally distinct fat depots. Although the level of PPARγ in FABP4-null epididymal fat appears to be somewhat lower compared with that in subcutaneous fat, analysis of paired samples did not show a statistically significant difference between the levels of PPARγ in the mouse epididymal and subcutaneous fat (n = 4 pairs, P = 0.602).

We then measured the relative amount of FABP4 and PPARγ in paired samples of human VAT and SAT obtained from biopsies of lean (6), obese (11), and obese diabetic (9) individuals (Fig. 7C and D). VAT FABP4 of each individual was normalized to its SAT FABP4 level. No significant difference in the VAT/SAT ratio of FABP4 was observed upon comparing lean and healthy obese individuals. However, the VAT/SAT ratio of FABP4 was significantly higher in obese diabetic patients compared with lean or healthy obese individuals (Fig. 7D). Consistent with our previous results, PPARγ was negatively correlated with FABP4 and was higher in the human SAT, compared with the human VAT (Fig. 7E).

The fatty acid chaperone FABP4 is induced by PPARγ, the master regulator of adipogenesis, yet these two regulators exert opposite effects on various metabolic parameters such as insulin resistance and inflammation (14). Here we demonstrate a previously unrecognized negative feedback loop, whereby FABP4 specifically triggers proteasomal degradation of PPARγ and consequently inhibits PPARγ-related functions, thereby providing a possible mechanism that explains their opposite effects. Our observations are consistent with an earlier finding that PPARγ activity is elevated in FABP4-null macrophages (20). FABP4 was reported to physically interact with PPARγ (38); hence, it is likely that such a physical interaction triggers the ubiquitination and the subsequent proteasomal degradation of PPARγ. This interaction took place through a site distinct from the fatty acid binding pocket of FABP4 (38), and therefore the ability of the fatty acid binding inhibitor BMS309403 to block the effect of FABP4 on PPARγ might be the outcome of an indirect allosteric effect.

Our findings show that FABP4 expression is not just an outcome of adipogenesis. Rather, FABP4 negatively regulates PPARγ and preadipocyte differentiation. Our observation that FABP4-null adipocytes express higher PPARγ levels may explain their improved differentiation capacity. Expression array analysis of many cell types suggests that the changes in gene expression induced by TZDs take place mostly in adipocytes (39) as part of a de novo adipogenesis (40). Evidently, the influence of FABP4 on adipogenesis was far more extensive than that of rosiglitazone.

The higher level of FABP4 and the lower level of PPARγ in VAT compared with SAT suggest a causative link between VAT FABP4 and the metabolic syndrome and may explain some morphological and functional differences between the adipocyte subtypes comprising these two types of fat tissue. For instance, visceral preadipocytes proliferate and differentiate to mature adipocytes less readily than their subcutaneous counterparts. Visceral adipocytes are smaller and less responsive to TZD compared with subcutaneous adipocytes (41), possibly due to the reduced PPARγ. Furthermore, the VAT secretes higher levels of proinflammatory cytokines (8,42,43).

Previous studies reported similar levels of FABP4 mRNA in human VAT and SAT (6,44,45). Our findings that FABP4 protein is higher in both human VAT and mouse intra-abdominal fat suggest that FABP4 may also be regulated posttranscriptionally. PPARγ mRNA levels were reported in paired samples of human VAT and SAT, but the data are rather inconsistent. In contrast, and in line with our observations, comparative data at the protein level consistently showed reduced levels and activity of PPARγ in human VAT (46–48), further supporting the notion of posttranslational regulation of PPARγ by FABP4. Our finding of higher FABP4 levels in VAT relative to their levels in SAT in obese diabetic individuals may explain why obesity-associated pathology correlates better with visceral fat mass than with overall adiposity (4–8).

Our finding that FABP4 interferes with level and hence activity of PPARγ in adipocytes and macrophages may be critical to understanding the development of the metabolic syndrome and its related pathologies, such as type 2 diabetes and atherosclerosis. We hypothesize that lipid accumulation elevates FABP4, thereby reducing PPARγ. In view of the impact of PPARγ on insulin sensitivity, our findings may explain the progression from insulin resistance to an overt diabetic state when FABP4 levels increase, thereby compromising PPARγ levels and activities. FABP4 also promotes a proinflammatory state, further advancing the development of metabolic syndrome–associated pathologies. A unique population of VAT-resident regulatory T cells was recently implicated in control of the inflammatory state of adipose tissue (49). PPARγ is a crucial regulator of this regulatory T-cell population, necessary for complete restoration of insulin sensitivity by TZD in obese mice.

Pharmacologic activators of PPARγ, such as TZDs, significantly improve insulin sensitivity in type 2 diabetes, and the net effect of PPARγ activation in murine macrophages is atheroprotective (13). However, TZDs trigger rapid degradation of PPARγ (36), possibly accounting for the recent reports implicating them in increased risk of cardiovascular death (50). We propose that similarly to TZD, FABP4 lowers PPARγ levels in macrophages, thereby triggering its adverse physiological effects (11). In this respect, inhibition of FABP4, combined with activation of PPARγ by TZDs could attenuate the harmful effects of TZDs, while maintaining their beneficial activities.

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

Funding. This work was supported by research grants from the estates of Fannie 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 this article were reported.

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