A New Selective Peroxisome Proliferator-Activated Receptor γ Antagonist with Antiobesity and Antidiabetic Activity

Abstract

Peroxisome proliferator-activated receptor γ (PPAR-γ) plays a key role in adipocyte differentiation and insulin sensitivity. Its synthetic ligands, the thiazolidinediones (TZD), are used as insulin sensitizers in the treatment of type 2 diabetes. These compounds induce both adipocyte differentiation in cell culture models and promote weight gain in rodents and humans. Here, we report on the identification of a new synthetic PPARγ antagonist, the phosphonophosphate SR-202, which inhibits both TZD-stimulated recruitment of the coactivator steroid receptor coactivator-1 and TZD-induced transcriptional activity of the receptor. In cell culture, SR-202 efficiently antagonizes hormone- and TZD-induced adipocyte differentiation. In vivo, decreasing PPARγ activity, either by treatment with SR-202 or by invalidation of one allele of the PPARγ gene, leads to a reduction of both high fat diet-induced adipocyte hypertrophy and insulin resistance. These effects are accompanied by a smaller size of the adipocytes and a reduction of TNFα and leptin secretion. Treatment with SR-202 also dramatically improves insulin sensitivity in the diabetic ob/ob mice. Thus, although we cannot exclude that its actions involve additional signaling mechanisms, SR-202 represents a new selective PPARγ antagonist that is effective both in vitro and in vivo. Because it yields both antiobesity and antidiabetic effects, SR-202 may be a lead for new compounds to be used in the treatment of obesity and type 2 diabetes.

THE PEROXISOME PROLIFERATOR-activated receptors (PPARs, NR1C1, NR1C2, NR1C3; Ref. 1) are members of the nuclear receptor superfamily. They function as heterodimers with the receptor of 9-cis-retinoic acid (RXR, NR2B; Ref. 1), and bind to specific peroxisome proliferator response elements (PPREs) to regulate transcription of their target genes. Three different PPAR genes have been characterized, which give rise to four distinct proteins (α, β/δ, γ1, and γ2). Although the PPARs were first cloned as orphan members of the nuclear receptor gene family, rapid progress has been made in their functional analysis. This research contributed to a better understanding of the importance of fatty acids as hormones and has established the PPARs as molecular targets for the development of drugs to treat human diseases (2).

PPARγ is expressed predominantly in adipose tissue, where it is known to play a critical role in adipocyte differentiation and fat deposition (3, 4). It can be activated by arachidonic acid-metabolites generated by the cyclooxygenase and lipooxygenase pathways (5–7) and by fatty acid-derived components released from oxidized low density lipoproteins (8). The antidiabetic thiazolidinediones (TZDs), currently used as insulin sensitizers, are the best synthetic PPARγ ligands in terms of specificity and affinity, although the mechanism by which activation of PPARγ leads to an improvement of insulin action is still debated (3, 4, 9). Moreover, deletion of one allele of PPARγ was recently shown to protect mice from high fat diet (HFD)-induced adipocyte hypertrophy and insulin resistance, underlying the complexity of the role of PPARγ in insulin sensitivity (10).

The most extensively studied therapeutic utility for PPARγ has been in the treatment of type 2 diabetes. TZDs were shown to enhance the sensitivity of target tissues to insulin and to reduce plasma glucose, lipid, and insulin levels in animal models of type 2 diabetes, as well as in human (11, 12). However, due to their ability to induce gene expression in adipocytes and to enhance adipocyte differentiation (13), TZDs also have negative side-effects. Indeed, they induce adipocyte differentiation and weight gain in patients with already serious metabolic disorders (14, 15). For this reason, important efforts are made to identify new PPARγ modulators having antidiabetic action, without promoting weight increase. Recently, a novel PPARγ ligand (GW 0072), which is a partial agonist in transactivation assays, was shown to inhibit adipocyte differentiation in cell culture (16). In addition, new antagonists for PPARγ are being characterized: bisphenol A diglycidyl ether (BADGE) inhibits adipocyte differentiation (17), PD 068235 also blocks adipocyte differentiation but does not revert the phenotype of terminally differentiated adipocytes (18), LG 100641 blocks adipocyte differentiation as well, and stimulates glucose uptake in 3T3-L1 adipocytes (19). So far, none of these inhibitors has been tested in vivo to verify whether they may prevent obesity and reduce insulin resistance, which would delay the onset of type 2 diabetes.

Here, we report that SR-202 [dimethyl α-(dimethoxyphosphinyl)-p-chlorobenzyl phosphate] is a novel PPARγ-specific antagonist that blocks adipocyte differentiation induced either by thiazolidinediones or by the combination of dexamethasone, insulin, and 3-isobutyl-1-methylxanthine (IBMX). We have investigated the activity of this antagonist in vivo and demonstrate that blocking PPARγ activity gives rise to a decrease in fat deposit and increase in insulin sensitivity. This new PPARγ antagonist may serve to develop compounds that will be more beneficial in the treatment of obesity and type 2 diabetes than thiazolidinediones because it yields both antiobesity and antidiabetic effects.

RESULTS

SR-202 Is a Specific Antagonist of PPARγ

Using a transcriptional reporter assay, we have identified a compound, SR-202 (Fig. 1A), that selectively modulates PPARγ transcriptional activity. HeLa cells were cotransfected with full-length PPARα, PPARβ, or PPARγ cDNA and a single PPRE-containing reporter plasmid. Alternatively, cells were also cotransfected with farnesoid X receptor (FXR) encoding plasmid and an FXR response element containing reporter plasmid. Cells were then treated with specific ligands of each of these nuclear receptors in the presence or absence of increasing concentrations of SR-202. As shown in Fig. 1, B–E, SR-202 itself did not cause any significant change in the basal transcriptional activity in the presence of PPARα, PPARβ, FXR, and PPARγ, but absence of agonists of these receptors. Similarly, SR-202 did not inhibit specific ligand-stimulated PPARα, PPARβ, and FXR transcriptional activities (Fig. 1, B–D). However, SR-202 showed a dose-dependent attenuation of troglitazone-induced PPARγ transcriptional activity, with an IC₅₀ of 140 μm (Fig. 1E). Thus, SR-202 shows selectivity both among the PPAR family members and other nuclear receptors because it specifically inhibits the PPARγ activity. This inhibition is not due to cytotoxic effects because, after a 36-h exposure of the cells to 400 μm SR-202, the lactate dehydrogenase (LDH) activity released in the medium never exceeded 6% of the total cellular LDH activity (data not shown).

Fig. 1.

SR-202 Is a Specific PPARγ Antagonist A, Structure of SR-202. B, Effect of SR-202 on Wy14643-induced PPARα activity. C, Effect of SR-202 on bezafibrate-stimulated PPARβ activity. D, Effect of SR-202 on GW1358-induced FXR activity. E, Effect of SR-202 on troglitazone-induced PPARγ activity. HeLa cells were transfected with the reporter construct (AcoA.TK.CAT or pCAT-promIBABP), and full-length mPPARα, mPPARβ, mPPARγ, or hFXR expression plasmids. They were then treated for 36 h with specific ligands and increasing concentrations of SR-202 as indicated. Each bar represents the mean ± sem of three experiments. CAT values were normalized to the β-galactosidase activity resulting from the expression of a cotransfected vector.

We next examined whether SR-202 is a ligand of PPARγ using the coactivator-dependent receptor ligand assay (CARLA) described previously (7). CARLAs are based on the capacity of a ligand to stimulate the interaction of nuclear receptors with a coactivator, here steroid receptor coactivator-1 (SRC-1). As expected, the known potent ligand of PPARγ, BRL 49653 (Rosiglitazone) induced interaction of PPARγ with SRC-1 in a dose-dependent manner (Fig. 2A). SR-202 alone did not promote any interaction between PPARγ and SRC-1, indicating the absence of agonist activity of the compound (Fig. 2B). However, coincubation of BRL 49653 (0.5 μm) with increasing concentrations of SR-202 (50–400 μm) led to a dose-dependent decrease in the recruitment of SRC-1 to the receptor (Fig. 2B). At the maximal concentration of SR-202 (400 μm), BRL 49653-stimulated SRC-1 binding to PPARγ was decreased by 75%, suggesting that this compound could displace the activator and almost completely abolish the recruitment of SRC-1 to the receptor. The association of PPARα and PPARβ with SRC-1 was not affected by SR-202, confirming the specificity of SR-202 for PPARγ (data not shown). Thus, SR-202 is able to both interact specifically with PPARγ and inhibit its agonist-dependent interaction with the coactivator SRC-1.

Fig. 2.

SR-202 Inhibits BRL 49653-Mediated Recruitment of SRC-1 by PPARγ A, Representative gel and quantification by densitometry (CARLA values corresponding to SRC-1 recruitment) of radiolabeled SRC-1 pull down assay, using the hPPARγ ligand binding domain, in the presence of increasing concentrations of BRL 49653. B, Representative gel and quantification (SRC-1 recruitment) of the effect of SR-202 on BRL 49653-stimulated recruitment of SRC-1 by PPARγ. Each bar represents the mean ± sem of at least three experiments.

In conclusion, SR-202 is a selective PPARγ antagonist, which is able to antagonize TZD-induced transcriptional activity of PPARγ. This effect may result from its ability to inhibit TZD-stimulated recruitment of coactivators by PPARγ.

SR-202 Inhibits PPARγ-Dependent Differentiation of Adipocytes

A major role of PPARγ is to stimulate adipogenesis. We thus designed experiments to see whether SR-202 could also antagonize PPARγ activity in a functional assay, more particularly in adipocyte differentiation. For that purpose, preadipocyte 3T3-L1 cells were pretreated with different concentrations of SR-202 or vehicle for 24 h and then induced to differentiate with medium containing either BRL 49653 (25 nm) and insulin (5 μg/ml) or a mixture containing dexamethasone (1 μm), insulin (10 μg/ml), and IBMX (0.5 mm). As shown in Fig. 3, A and B, SR-202 was able to significantly inhibit BRL 49653- and hormone-induced adipocyte differentiation of 3T3-L1 cells in a dose-dependent manner, as shown by the decrease in lipid content revealed by Oil red O staining. The antiadipogenic effect of SR-202 in this experimental setting was also revealed by the reduced expression of an adipocyte differentiation marker, the adipocyte fatty acid binding protein (aP2) (Fig. 3C). The inhibition of adipogenesis at high concentrations of SR-202 was not due to a toxic effect of the antagonist because it could be attenuated by cotreating the cells with a high concentration of BRL 49653 (1 μm instead of 25 nm, data not shown). Furthermore, the lack of toxicity of SR-202 was again confirmed by quantification of LDH activities released in the culture media (reaching only 5% of total cellular LDH activity after a 48-h treatment with 400 μm SR-202) and by the total number of cells that remained unchanged at the end of the differentiating protocols (data not shown).

Fig. 3.

SR-202 Is a Potent Antagonist of PPARγ-Induced Adipocyte Differentiation A, Effect of SR-202 on BRL 49653-induced adipogenesis of 3T3-L1 cells. Cells were pretreated with SR-202 or vehicle (water) for 24 h and then treated with BRL 49653 (25 nm) and insulin (5 μg/ml), with or without SR-202. After 6 d of treatment, cells were stained with Oil Red O. B, Effect of SR-202 on hormone-induced adipocyte differentiation of 3T3-L1 cells. Cells were pretreated with SR-202 and then induced with dexamethasone/IBMX/Ins with or without SR-202 for 2 d. Thereafter, the medium was changed to insulin alone with or without SR-202 for an additional 2 d. Cells were then stained with Oil Red O. C, Effect of SR-202 on the mRNA levels of aP2 adipocyte differentiation marker. Ribonuclease protection assay was performed on 10 μg of total RNA from 3T3-L1 cells treated as indicated above. L27, Ribosomal protein.

SR-202 Inhibits Adipose Tissue Accumulation in Vivo

Because SR-202 inhibits PPARγ-dependent adipocyte differentiation, we investigated the effect of SR-202 in vivo, on body and fat composition in wild-type (wt) mice under standard diet (SD) or HFD. SR-202 treatment given as food admixture was started at 3 wk of age, i.e. at the weaning period that corresponds to the beginning of important white adipose tissue (WAT) accumulation (20), and prolonged during a consecutive 10 wk. Wt mice treated with SR-202 for 10 wk gained significantly less body weight than untreated wt mice under both SD and HFD (Fig. 4A, left panel). Consistent with this result, the treatment of wt mice by SR-202 significantly decreased WAT mass under SD and protected them from HFD-induced increase in WAT mass (Fig. 4B, left panel, compare CTL SD mice with SR-202 HFD mice). This effect of SR-202 on adiposity was not specific of epididymal WAT because SR-202-treated wt mice had also a lower brown adipose tissue (BAT) mass than untreated wt mice, under both SD and HFD (Fig. 4C, left panel). In addition to the decrease in fat mass development, the size of white adipocytes was modulated in parallel with the WAT mass. SR-202-treated wt mice had smaller adipocytes than untreated wt mice on SD (Fig. 5A, a and b). Quantification of cell numbers indicated that there was more cells per microscopic field in WAT of SR-202-treated wt mice than in WAT of untreated wt mice, confirming the smaller size of adipocytes (Fig. 5B, a and b). Importantly, SR-202-treated wt mice were protected from HFD-induced adipocyte hypertrophy (Fig. 5, A and B, c and d). To assess whether the effects of SR-202 on WAT involved the modulation of PPARγ activity, we measured the mRNA levels of PPARγ target genes including lipoprotein lipase (LPL), fatty-acid translocase (FAT)/CD36, aP2 and sterol regulatory element-binding protein 1c (SREBP-1c). As shown in Fig. 6, SR-202-treated wt mice exhibited a significant decrease in PPARγ activity as demonstrated by the reduction of LPL, aP2, CD36 and SREBP-1c mRNA levels. Altogether, these data show that a decrease of PPARγ activity by treatment with a specific antagonist is associated with decreased adiposity.

Fig. 4.

SR-202 Prevents Weight Gain and Adipose Tissue Deposit Content in Mice Body weight (A) and fat weight (B and C) of wt and PPARγ +/− male mice treated or not with SR-202 (400 mg/kg) under both SD and HFD. SR-202 was given as food admixture and the treatment was started just after weaning during a consecutive 10 wk. Each bar represents the mean ± sem of at least 10 mice per group (* and $, P < 0.05; NS, no significative difference). *, Comparison between CTL and SR-202-treated mice; $, comparison between SD and HFD.

Fig. 5.

SR-202 Treatment Decreases Adipocyte Size A, Histological analysis of epididymal WAT from wt and PPAR γ+/− male mice treated or not with SR-202 (400 mg/kg) under both SD and HFD. Twenty-micrometer sections were stained with both eosin and hematoxylin and with Oil Red O. B, Quantification of adipocyte number (mean ± sem) in at least three microscopic fields of WAT sections per mouse from three different mice per group. *, Comparison between CTL and SR-202-treated mice; $, comparison between SD and HFD (* and $, P < 0.05).

Fig. 6.

SR-202 Decreases PPARγ Activity in Vivo Amounts of the mRNA of the following PPARγ target genes were determined in WAT of wt and PPARγ +/− mice under HFD treated or not with SR-202: LPL, FAT/CD36, aP2, and SREBP-1c. The mRNA level of each target was determined by reverse transcription followed by real-time PCR using the Light-cycler-FastStart DNA Master SYBR Green I as described in Materials and Methods. Each bar represents the mean ± sem (n = 5). *, P < 0.05 (comparison with CTL wt mice).

An alternative way to inhibit PPARγ activity is to invalidate its own gene in mice. We have generated PPARγ mutant mice by homologous recombination (data to be published elsewhere). Early embryonic lethality of PPARγ null mice was observed, as also reported by others (10, 21). Heterozygous PPARγ mice (PPARγ +/−) developed normally, were fertile and apparently healthy. We challenged the PPARγ +/− mice with HFD, in the presence or not of SR-202 and analyzed the expression of PPARγ target genes. Compared with wt animals, untreated PPARγ +/− mice under HFD showed, as expected, a decrease in PPARγ activity as illustrated by the reduction of LPL, CD36, aP2, and SREBP-1c mRNA levels in WAT (Fig. 6). However, SR-202-treated PPARγ +/− mice under HFD did not show a further decrease of PPARγ activity compared with untreated PPARγ +/− mice because both groups of mice had similar levels of LPL, CD36, aP2, and SREBP-1c mRNA levels (Fig. 6). Analyses of body and fat composition revealed that untreated PPARγ +/− mice had a lower body weight than untreated wt mice, under both SD and HFD (Fig. 4A). In addition, untreated PPARγ +/− mice on SD, showed reduced WAT and BAT mass compared with untreated wt mice (Fig. 4, B and C, respectively). Moreover, they were protected from HFD-induced hypertrophy of WAT and BAT (Fig. 4, B and C, respectively). Consistent with the difference in WAT mass, adipocytes of untreated PPARγ +/− mice were smaller than those of untreated wt mice (Fig. 5, A and B, e and a). Similarly, untreated PPARγ +/− mice were protected from HFD-induced adipocyte hypertrophy (Fig. 5, A and B, g and c). Interestingly, treatment of PPARγ +/− mice with SR-202 did not further decrease body and adipose tissue weight (Fig. 4, A–C, right panels), whereas it further decreased the size of white adipocytes, under both SD and HFD (Fig. 5B, e–h).

Thus, decreasing PPARγ activity, either by treatment with a specific antagonist or by invalidation of one allele of the PPARγ gene, leads to decreased adiposity in mice. These data indicate that the phenotype of SR-202-treated mice is a bona fide consequence of the reduction of PPARγ activity in vivo.

SR-202 Decreases Adipocytokine Secretion by WAT

White adipocytes are not only forming a passive tissue mass devoted to the storage of excess energy but are also endocrine secretory cells (22). For example, adipsin, TNFα, leptin, and plasminogen activator inhibitor-1 are all secreted by fat cells, under control of nutritional (feeding and fasting) and pathological (obesity) states (22). In particular, regulatory roles in energy homeostasis and/or systemic insulin sensitivity have been shown for leptin and TNFα (22). In addition, the amount of circulating leptin and TNFα correlates with body fat stores and/or hyperinsulinemia, suggesting a potential involvement of these cytokines in obesity-induced insulin resistance (23, 24). Because SR-202 treatment is associated with decreased adiposity, we tested the effects of SR-202 treatment on leptin and TNFα secretion by the adipose tissue. Interestingly, SR-202-treated wt mice, under SD, had normal plasma leptin levels but lower plasma TNFα levels (Fig. 7, A and B, left panels). As expected, we found a strong increase of plasma leptin and TNFα levels in untreated wt mice under HFD (Fig. 7, A and B, left panels; compare CTL SD mice with CTL HFD mice). Under SR-202 treatment, these mice were protected from HFD-induced hyperleptinemia and HFD-related increase in TNFα levels (Fig. 7, A and B, left panels). These data provide further evidence that the secretion levels of leptin and TNFα are closely related to the size of adipocytes. In addition, it suggests that inhibiting PPARγ activity by treatment with an antagonist can prevent from obesity-induced secretion of these cytokines.

Fig. 7.

SR-202 Decreased Secretion of Adipocytokines Fed levels of plasma leptin (A) and TNFα (B) in wt and PPAR γ+/− male mice treated or not with SR-202 (400 mg/kg) under both SD and HFD. Decreased PPARγ activity, by SR-202 treatment or by deletion of one allele of the PPARγ gene, results in a decrease of leptin and TNFα plasma levels. Each bar represents the mean ± sem of at least 10 mice per group (* and $, P < 0.05; ** or $$, P < 0.01; NS, no significant difference). *, Comparison between untreated and treated mice; $, comparison between SD and HFD.

Similar analyses were performed in PPARγ +/− mice. Untreated PPARγ +/− mice, under SD, had similar levels of leptin but lower plasma levels of TNFα compared with untreated wt mice (Fig. 7, A and B). Furthermore, untreated PPARγ +/− mice were protected from HFD-induced increase in both leptin and TNFα levels (Fig. 7, A and B, right panels). Finally, treating PPARγ +/− mice with SR-202 further decreased plasma levels of leptin and TNFα under HFD, whereas only leptin was further decreased under SD (Fig. 7, A and B, right panels).

In conclusion, decreasing PPARγ activity, either by treatment with a specific antagonist or by invalidation of one of the two alleles of the PPARγ gene, leads to decreased leptin and TNFα plasma levels, which is consistent with the smaller size of adipocytes. Treating PPARγ +/− mice with SR-202, further decreased plasma levels of these signaling molecules, in agreement with a more pronounced decrease in the size of adipocytes (see Fig. 5).

SR-202 Protects against HFD-Induced Insulin Resistance

Both reducing the adipocyte size and modulating the secretion of two major adipocytokines, leptin and TNFα, are known to influence insulin sensitivity (3, 25). We thus investigated whether SR-202 treatment could increase insulin sensitivity in mice. For that purpose, we measured glucose and insulin levels, in addition to plasma free fatty acid (FFA) levels, which are known to be linked to insulin sensitivity. Treatment of wt mice on SD with SR-202 did not affect the plasma glucose levels but strongly decreased the plasma insulin levels (Fig. 8, A and B, left panels). No change in FFA plasma levels was observed in SR-202-treated wt mice under SD (Fig. 8C, left panel). As expected, HFD feeding in untreated wt mice was associated with an increase in FFA levels (Fig. 8C, left panel), a high hyperinsulinemia (Fig. 8B, left panel), and a slight, however statistically not significant, increase in glucose levels (Fig. 8A, left panel), indicating a compensated state of insulin resistance. In contrast, SR-202 treated mice were protected from HFD-induced insulin resistance, as measured by the maintenance of relatively low insulin and FFA plasma levels (Fig. 8, B and C, left panels). These data demonstrate that decreasing PPARγ activity by treatment with an antagonist potentiates insulin sensitivity and improves lipid parameters under HFD.

Fig. 8.

SR-202 Treatment Protects against HFD-Induced Insulin Resistance Fed levels of plasma glucose (A), insulin (B), and FFA (C) in wt and PPAR γ+/− male mice treated or not with SR-202 (400 mg/kg) under both SD and HFD. Decreased PPARγ activity, by SR-202 treatment or by deletion of one allele of the PPARγ gene, was associated with increased insulin sensitivity. Each bar represents the mean ± sem of at least 10 mice per group (* and $, P < 0.05; $$, 0.01; NS, no significant difference). *, Comparison between untreated and treated mice; $, comparison between SD and HFD.

We next investigated insulin sensitivity in PPARγ +/− mice. Consistent with the result obtained in SR-202-treated wt mice, untreated PPARγ +/− mice under SD showed lower plasma insulin levels than untreated wt mice, without any significant change in FFA plasma levels (Fig. 8, B and C, right panels). However, PPARγ +/− animals were protected from HFD-induced hyperinsulinemia and HFD-related increase of FFA plasma levels (Fig. 8, B and C, right panels). Finally, treatment of PPARγ +/− mice with SR-202 did not further improve insulin sensitivity, as indicated by the lack of change in metabolic parameters (Fig. 8, A–C, right panels).

In conclusion, a partial decrease in PPARγ activity leads to increased insulin sensitivity and protects mice from HFD-induced obesity and insulin resistance. However, a combined pharmacological and genetic action to reduce PPARγ activity, such as in SR-202-treated PPARγ +/− mice, did not further improve insulin sensitivity compared with untreated PPARγ +/− mice.

SR-202 Increases Insulin Sensitivity in ob/ob Mice

After demonstrating the ability of SR-202 to prevent the development of an insulin resistance as seen upon a HFD challenge, we next tested whether SR-202 could improve insulin sensitivity in the context of an already developed obesity with hyperglycemia and hyperinsulinemia. For that purpose, we used ob/ob mice, a well characterized strain of obese and diabetic mice. Treatment of the ob/ob mice with SR-202 was started at 8 wk of age (experimental d 0), when mice are already overtly diabetic (see glucose and insulin levels, d 0, Fig. 9, A and B). The treatment as food admixture was maintained for 20 d, after which metabolic parameters were measured (experimental d 20).

Fig. 9.

SR-202 Is an Insulin Sensitizer in Vivo, Decreasing Hyperglycemia and Hyperinsulinaemia in ob/ob Mice Treatment of 8-wk-old ob/ob mice was started at d 0 up to d 20. Serum glucose (A) and insulin (B) levels at the fed state in ob/ob mice on d 0 (white bars) or on d 20 after the treatment with vehicle (gray bars) or SR-202 (black bars). Plasma glucose (C) and insulin (D) levels during an ip glucose tolerance test in ob/ob mice performed at experimental d 20, in untreated mice (white circles) or in mice treated with SR-202 (black circles). E, Intraperitoneal insulin tolerance test in ob/ob mice treated or not with SR-202. F, amounts of the mRNAs of LPL, FAT/CD36, and aP2 in WAT of ob/ob mice treated or not with SR-202 (400 mg/kg) during 20 d. Values are expressed as the mean ± sem (n = 10). *, P < 0.05, **, P < 0.01.

Although hyperglycemia and hyperinsulinemia continued to increase in the untreated ob/ob mice (compare d 0 and d 20, Fig. 9, A and B), daily treatment with SR-202 prevented the time-dependent increase in glucose concentrations. After 20 d of treatment, fed glucose concentrations were 67% of those of untreated animals (P < 0.01) (Fig. 9A). This decreased hyperglycemia was associated with decreased insulin levels to 70% of control values (P < 0.05) (Fig. 9B). The decrease in glucose and insulin was not a secondary response to changes in body weight because all animals gained weight at comparable rates (data not shown). Thus, the PPARγ antagonist SR-202 induced a coordinated decrease in both glucose and insulin concentrations, suggesting an overall improvement in glucose homeostasis.

We then conducted glucose and insulin tolerance tests to determine whether SR-202 could improve glucose disposal and insulin sensitivity in ob/ob mice. As shown in Fig. 9C, SR-202-treated ob/ob mice showed an enhanced glucose disposal compared with untreated ob/ob mice despite a strikingly reduced insulin induction (Fig. 9D), suggesting an increase of insulin sensitivity. In agreement with these data, ip administration of insulin resulted in a stronger glucose lowering effect in SR-202-treated ob/ob mice compared with untreated mice (Fig. 9E).

Finally, SR-202 treatment directly affected the expression of PPARγ target genes. 20 d of treatment of ob/ob mice with SR-202 induced a decrease of PPARγ activity as illustrated by the decrease of LPL, CD36, and aP2 mRNA levels in WAT (Fig. 9F).

In conclusion, SR-202 functions as an insulin sensitizer, markedly decreasing hyperglycemia and hyperinsulinemia in ob/ob mice, a mouse model of obesity and insulin resistance.

DISCUSSION

This study describes a new synthetic antagonist of PPARγ, which has the capacity to antagonize both TZD-stimulated recruitment of SRC-1 by PPARγ and TZD-induced transcriptional activity of this receptor. We have tested the activity of this antagonist both in vitro and in vivo, in the whole animal. In vitro, we have demonstrated that the inhibition of PPARγ activity by SR-202 indeed inhibits adipocyte differentiation of 3T3-L1 cells. In vivo, treatment of mice with SR-202, from weaning onwards, prevents the full development of WAT and BAT. Furthermore, it protects mice from HFD-induced adiposity. In addition, SR-202 improves insulin sensitivity, by decreasing HFD-induced FFA, leptin and TNFα secretion in association with a prevention of adipocyte hypertrophy. Finally, SR-202 is an insulin sensitizer, markedly reducing hyperglycemia and hyperinsulinemia in ob/ob mice. Thus, we report on a PPARγ antagonist compound that is effective in vitro and in vivo and offers the potential for an improved therapy for metabolic diseases.

Two Animal Models of PPARγ Decreased Activity

The strength of our study is based on the use of two mouse models in which PPARγ activity is decreased. In the first model, we use an antagonist of PPARγ, SR-202, which we have found to be specific and functional in vitro. In the second model, decreased PPARγ activity was achieved by deletion of one of the two alleles of the PPARγ gene. Comparative analyses of both groups of mice allows us to propose that the SR-202 effects reflect a reduced PPARγ activity. Moreover, treatment of PPARγ +/− mice with SR-202 was performed to investigate the effects of severely reduced PPARγ activity. We have demonstrated that reduction of PPARγ activity was achieved in SR-202-treated wt mice and PPARγ +/− mice because they showed a similar decrease in expression levels of known PPARγ target genes. Moreover, they are similarly protected from HFD-induced adipocyte hypertrophy and insulin resistance. However, treatment of PPARγ heterozygous mice with SR-202 did not further reduce PPARγ activity and did not amplify the insulin sensitizing phenotype. However, a further decrease in the size of adipocytes and plasma levels of TNFα and leptin was observed. Several explanations can be proposed as to why PPARγ activity in not further decreased in SR-202-treated PPARγ +/− mice, whereas some of the SR-202 effects are amplified (see above). First, SR-202 is a low affinity compound, and it is likely that the inability of SR-202 to further improve insulin sensitivity in PPARγ +/− mice is caused by the difficulty in obtaining a complete shut-down of PPARγ activity. Particularly, the reduction of PPARγ expression in PPARγ +/− mice may increase this difficulty. Second, transgenic mice may use compensatory mechanisms that are not present in wt mice, masking the effects of SR-202. In both cases, it is possible that SR-202 has some PPARγ independent effect, activating or antagonizing other factors, and leading to additive effects in PPARγ +/− mice. Nevertheless, the fact that SR-202 amplifies some phenotypes in PPARγ +/− mice (reduction of TNFα and leptin levels) and not others (insulin sensitivity) underlines the different contributions of PPARγ in different processes. It indicates that PPARγ is incontestably an essential actor of adipocyte differentiation, with a participation in insulin action that is more complex. Other pathways are likely to be involved in insulin signaling as well. However, we can not exclude that treatment with higher concentrations of SR-202 could improve insulin sensitivity in PPARγ +/− mice.

The PPARγ Antagonist SR-202 Inhibits Adipocyte Differentiation

There is now strong evidence supporting a key role of PPARγ in adipogenesis based on both gain or loss of function experiments. Indeed, it has been shown that the ectopic expression and activation of PPARγ in fibroblasts are sufficient to induce an adipogenic response (26). Moreover, genetic studies using chimeric mice have confirmed that PPARγ is required for adipogenesis in vivo (27). Finally, it was recently demonstrated that reducing PPARγ activity with a selective PPARγ antagonist, BADGE, inhibits adipocyte differentiation in vitro (17). Herein, we describe a new synthetic PPARγ antagonist belonging to the phosphonophosphate family with antiadipogenic activity. In vitro, SR-202 inhibits both TZD and hormonally induced adipocyte differentiation of 3T3-L1 cells, as shown by the decrease in lipid accumulation and aP2 mRNA levels. Hormone-induced aP2 expression was even more dramatically reduced indicating that SR-202 is more or less effective depending on the differentiation protocol applied. Recently, it was suggested that there may be at least two pathways for 3T3-L1 adipocyte differentiation (19). The first one would be regulated by PPARγ and its ligands and accordingly could be blocked by the PPARγ antagonist LG 100641. The second pathway would be independent of PPARγ action. Indeed, it was insensitive to LG 100641 (19). The fact that SR-202 antagonized both hormone- and TZD-induced adipocyte differentiation reveals that SR-202 has effects on both pathways. In addition to antagonizing PPARγ activity, SR-202 would inhibit another actor of adipocyte differentiation that responds to the IBMX/dexamethasone/insulin cocktail. A similar double antiadipogenic effect has also been described for two other PPARγ antagonists, BADGE (17) and PD 068235 (18). The in vitro observations are corroborated by in vivo experiments where SR-202 treated wt mice have decreased PPARγ activity, together with a decreased WAT and BAT mass, and are protected from HFD-induced adipocyte hypertrophy.

Treatment of Mice with SR-202 Improves Insulin Sensitivity

In addition to regulating adipocyte differentiation, PPARγ plays a key role in insulin signaling and agonists of PPARγ are used in the treatment of type 2 diabetes (11). However, the mechanism for this role remains obscure.

We demonstrate herein that the reduction of PPARγ activity by partial gene targeting improves insulin sensitivity in mice, under HFD, as previously reported by others (10). Importantly, we show that treatment of wt mice with SR-202 reproduces the insulin-sensitizing phenotype, confirming that reduction of PPARγ activity is associated with an increased insulin sensitivity. This observation suggests that antagonists of PPARγ could be useful to explore the signal transduction pathways involved in PPARγ-mediated insulin sensitization. Although our experiments do not provide a molecular mechanism to explain the increase of insulin sensitivity associated to the reduction of PPARγ activity, they suggest that this increased sensitivity is a consequence of the diminution of adipose tissue mass. Indeed, HFD feeding, which increases adipocyte size, also increases secretion of molecules causing insulin resistance such as FFA and TNFα. In contrast, SR-202-treated wt mice and untreated PPARγ heterozygous mice, which are protected from HFD-induced adipocyte hypertrophy, showed a significant decreased FFA and TNFα levels, and a concomitant improved insulin sensitivity. This is associated with a decreased expression in WAT of genes involved in fatty acid transport and lipogenesis, which participate in preventing adipocyte hypertrophy and therefore obesity under HFD. We thus can conclude that reduction of PPARγ activity increases insulin sensitivity, presumably by preventing adipocyte hypertrophy and consequently decreasing the secretion of FFA and TNFα. These results are in agreement with human genetic studies. One rare mutation renders PPARγ more active, leading to increased adipocyte differentiation and obesity (28). An other more common mutation impedes PPARγ activity and is associated with a lower body mass index, improved insulin sensitivity, and higher levels of plasma high density lipoprotein cholesterol (29).

The role of leptin secretion in participating to the modulation of insulin sensitivity remains unclear. Indeed, although lipoatrophic animals without leptin are insulin resistant, there is evidence that leptin can interfere with insulin signaling in some cell types in vitro (30, 31). Previous reports have shown that activation of PPARγ also results in a decrease of leptin secretion (32–34) and that PPARγ +/− mice under HFD overexpress and hypersecrete leptin (10). Thus, at first sight, our results concerning the concomitant decrease in PPARγ activity and plasma levels of leptin may appear surprising. However, the fact that we observe a similar reduced leptin secretion in two models of decreased PPARγ activity (deletion of one allele of the PPARγ gene or treatment with an antagonist of PPARγ) and a further decrease in SR-202 treated PPARγ +/− mice are clearly consistent. In addition, SR-202 treatment also modulates two other major regulators of leptin secretion, the size of adipocytes and plasma insulin levels, in favor of a decrease in leptin plasma levels. As mentioned above, how to relate this effect on leptin secretion and insulin sensitivity remains unclear. In particular, the fact that SR-202 increased insulin sensitivity in ob/ob mice, a mouse model of obesity without leptin, indicates that the action of the drug is not mediated by leptin.

PPARγ Antagonists Like SR-202 Could Be Clinically Useful in the Treatment of Obesity and Type 2 Diabetes

Our data and other recent reports (10, 35) underline the interesting paradox that both PPARγ overactivity, due to TZD, and PPARγ underactivity, due to either haploinsufficiency or treatment with a PPARγ antagonist, protect against obesity-induced insulin resistance. One explanation of this paradox may reside in the number and size of adipocytes. Indeed, TZD-mediated activation of PPARγ in rodents results in an increase in the number of small adipocytes and a decrease in the number of large, hypertrophic fat cells (25). In parallel, expression of two cytokines that influence energy homeostasis and insulin sensitivity, namely leptin and TNFα, is repressed by PPARγ agonists in adipocytes (30, 32, 36). Remarkably, untreated PPARγ +/− mice or SR-202-treated wt mice have smaller adipocytes than untreated wt mice, and they secrete less leptin and TNFα and release less fatty acids, contributing to enhanced insulin sensitivity. Although the ways of modulating PPARγ activity are distinct, the outcome appears similar: generation of small adipocytes and attenuated secretion of cytokines that interfere with insulin sensitivity. Consequently, not only agonists but also antagonists of PPARγ could be clinically useful in the treatment of obesity and type 2 diabetes. As a proof of concept, the treatment of ob/ob mice clearly establishes the efficacy of SR-202 in improving glucose disposal and insulin sensitivity, confirming that SR-202 is an insulin sensitizer in a model of preexisting obesity and insulin resistance. In addition, if SR-202, like BADGE, has a low affinity for PPARγ, it presents the specific advantages to be soluble in water and nontoxic in vivo, which is a significant bonus from a therapeutic point of view. Furthermore, it is possible that higher affinity PPARγ antagonists can be derived based on the structure of SR-202.

This work on SR-202, which directly acts on PPARγ, might be compared with studies on RXR, the obligate partner of PPARs. Very recently, mice lacking RXR in adipocytes (37) or mice treated with an antagonist of RXR, HX531 (38), have been reported to be resistant to HFD-induced obesity and insulin resistance. However, treatment of PPARγ +/− mice with HX531 severely depleted the WAT, leading to a reemergence of insulin resistance (38). This biphasic effect may raise concerns for the potential clinical use of molecules such as HX531, more particularly in patients with the variant PPARγ Ala12 allele. HX531 might worsen insulin resistance in patients that have lower PPARγ activity (29). The lack of effect of SR-202 in PPARγ +/− mice suggest that such an effect would not occur in this situation. The use of SR-202 type compounds could be considered for patients with normal PPARγ activity (Pro12 allele) or for patients with the more active variant Ala 12 PPARγ. Finally, PPARγ antagonists seem more adapted for clinical use than RXR antagonists that can also act on RXR homodimers or on other RXR containing heterodimers, triggering undesired side effects.

In conclusion, we have characterized a first PPARγ antagonist compound, SR-202, which is effective both in vitro and in vivo. We propose that, by preventing adipocyte differentiation and lipid accumulation, SR-202 protects mice from HFD-induced adipocyte hypertrophy and insulin resistance. SR-202 also improves insulin sensitivity in a model of established obesity, confirming that SR-202 is an insulin sensitizer. Treatment with compounds having the characteristics of SR-202 could be extremely interesting if they behave in human like in mouse. They would prevent adipogenesis and obesity while increasing insulin sensitivity, in contrast to TZDs, which induce weight gain. Consequently, SR-202 offers a potential for the development of compounds that could improve the therapy of metabolic diseases.

MATERIALS AND METHODS

Reagents

Dimethyl α-(dimethoxyphosphinyl)-p-chlorobenzyl phosphate (SR-202) was synthesized at Ilex onc. Rosiglitazone (BRL 49653) was a gift from Park Davis Pharmaceuticals (Ann Arbor, MI). IBMX, dexamethasone, and insulin were purchased from Sigma (St. Louis, MO).

CARLA

The CARLAs were performed as previously described with small modifications (7). Briefly, bacteria expressing GST-human (h) PPARγ LBD were incubated at 37 C until the culture reached A₆₀₀ of 0.7. At that point, isopropyl β-d-1-thiogalactopyranoside was added at the final concentration of 0.5 mm and the cells were incubated for a further 3 h at 28 C. Then, bacteria were lysed by three freeze cycles in liquid nitrogen. Sepharose 4B beads equilibrated in NETN (20 mm Tris-HCl, pH 8; 100 mm NaCl; 1 mm EDTA; 0.5% Nonidet P-40; 1 mm dithiothreitol) were added to the cell extract (250 μl for PPARγ culture). GST-PPARγ fusion protein was allowed to bind to the beads overnight at 4 C. SRC-1 (7) was produced using the TNT quick coupled transcription/translation system of Promega Corp. (Madison, WI) and labeled with methionine ³⁵S. Beads, SRC-1, and the compounds were mixed overnight at 4 C. After three centrifugation-wash cycles with NETN, the beads were loaded on a 10% SDS-PAGE gel. Autoradiography was used to localize SRC-1 and the amount of this protein was determined by image analysis on an image master Amersham Pharmacia Biotech (Buckinghamshire, UK).

Cell Culture

3T3-L1 cells were cultured in DMEM supplemented by 10% bovine calf serum. Two days after reaching confluency, differentiation was induced in DMEM supplemented with 10% fetal calf serum and dexamethasone (1 μm)/IBMX (0.5 mm)/insulin (10 μg/ml). After 48 h, the medium was changed with only the addition of insulin for an additional 2 d. Thereafter, the cells were grown in DMEM supplemented with 10% fetal calf serum. Alternatively, cells were also differentiated with BRL 49653 (25 nm) and insulin (5 μg/ml). SR-202 or vehicle (H₂O) was added 24 h before induction of differentiation. Medium was replenished with ligands every 2 d. Adipogenesis was determined by the staining of lipids with Oil Red O and by measuring the expression of adipocyte markers. RNA was isolated using the Trizol reagent (Life Technologies, Inc.).

Transfections

Human HeLa cells were transiently transfected using the phosphate calcium method with expression vectors for murine (m) PPARγ (39), mPPARβ (40), or mPPARα (41), the single PPRE-containing reporter plasmid AcoA.TK.CAT (chloramphenicol acetyl transferase) and a β-galactosidase encoding plasmid as control. Cells were also transfected with a hFXR encoding plasmid and pCAT-promIBABP reporter plasmid. They were exposed to the ligands for 36 h, lysed, and assayed for CAT and β-galactosidase activity. Transfections were performed in quadruplicates.

Measurement of LDH Activity

Viability of the cells was estimated by measuring the LDH activity released in the culture medium. Because LDH is a cytosolic enzyme, the activity of the enzyme found in the medium reflects cellular lysis. The LDH activity is measured by adding 15 μl of medium samples in 1 ml of the following reaction medium (200 mm phosphate buffer, pH 6.8; 20 mm NADH; and 1 mm pyruvate). LDH activity was determined with a spectrophotometer. Individual values are then expressed as percentage of the total quantity of LDH present in cells, determined after sonication of the cells.

Animals

Wild-type and PPARγ heterozygous mice, on a mixed background (sv129/C56Bl6), were maintained at 20 C with a 12-h light, 12-h dark cycle. Generation of PPARγ heterozygous mice will be described elsewhere. Animal experiments were approved by the animal authorization committee of the canton of Vaud (Switzerland). At 3 wk of age, the animals were separated by sex, genotyped for the PPARγ gene, and started to be fed with SD or HFD (D12451, 45 kcal% fat, Research Diets, Inc.) supplemented or not with SR-202 (400 mg/kg). All mice were weighted on a weekly basis, starting at 3 wk of age. Ten weeks later, animals were killed by cervical dislocation and tissues were removed, weighed, and frozen. Eight-week-old male ob/ob mice were purchased from Janvier Breeding (Le Genest Saint Isle, France). These mice were fed with a SD supplemented or not with SR-202 (400 mg/kg) for 20 d. Blood was withdrawn at the fed state on the indicated days from the tip of the tail for the measurement of plasma glucose and insulin levels. Glucose and insulin tolerance tests were performed on 6-h-fasted mice. Animals were injected ip with 2 mg/g body weight of glucose or 0.75 mU/g body weight of insulin. Blood was taken by tail puncture immediately before and 15, 30, 60, and 120 min after injection for measurements of plasma glucose or insulin levels.

Plasma Metabolites Measurement

Whole blood was withdrawn at the fed state, at the end of the dark cycle, from the orbital sinus by using heparinized microcapillary tubes. After withdrawal, the blood samples were immediately centrifuged, the serum removed, placed in a fresh tube and immediately frozen. Blood glucose levels were measured with a glucometer (Roche Molecular Biochemicals, Mannheim, Germany) on whole blood. Serum insulin, leptin, and TNFα levels were assayed using the murine ELISA kits (CrystalChem, Downers Grove, IL; or R&D Systems, Minneapolis, MN). Serum levels of nonesterified FFA were measured using a colorimetric assay (Roche Molecular Biochemicals).

Histological Analysis

Adipose tissue was removed from three animals per group, fixed in 10% paraformaldehyde/PBS and maintained at 4 C until used. Fixed specimens were embedded in tissue-freezing medium (Leica Instruments, Nussloch, Germany) and frozen in dry ice and isopentane. Twenty-micrometer cryostat tissue sections were mounted on slides and fixed. After washing in PBS, sections were used for histological staining (hematoxylin-eosin and Oil Red O). The number of adipocytes was manually quantified on at least three different microscopic fields from three different animals per group.

Measurement of mRNA

LPL, CD36/FAT, aP2, and SREBP-1c mRNA levels were determined by reverse transcription followed by real-time PCR using the Light-cycler-FastStart DNA Master Sybr Green I (Roche Molecular Biochemicals). Total RNA (1 μg) was reverse-transcribed in a 20-μl reaction containing 1× TR buffer (50 mm Tris-HCl, 75 mm KCl, 3 mm MgCl₂), 10 mm dithiothreitol, 0.5 mm deoxy-NTPs, 0.25 μg of random and oligo(deoxythymidine) primers (Promega Corp.) and 100 U of Superscript II enzyme (Invitrogen, Basel, Switzerland), after the conditions recommended by the manufacturer. PCRs were carried out in duplicate by adding 5 μl of diluted first-strand reaction in 15 μl of Hot Start reaction mix containing FastStart Taq DNA polymerase, Sybr green as dye, 2–3 mm MgCl₂ and 0,5 μm of each specific primers. The following primer combinations were used: 5′-CAGTGGGGCATGTTGACATT-3′ and 5′-TGAGAGCGAGTCTTCAGGTA-3′ for LPL; 5′-AAGATCCAAAACTGTCTGTA-3′ and 5′-GTCCTGGCTGTGTTTGGAGG-3′ for CD36/FAT; 5′-CTTGTCTCCAGTGAAAACTT-3′ and 5′-GTGGAAGTCACGCCTTTCAT-3′ for aP2; and 5′-ACGGAGCCATGGATTGCACA-3′ and 5′-AAGGGTGCAGGTGTCACCTT-3′ for SREBP-1c. After 10 min at 95 C, the tubes were subjected to 40 cycles of amplification including denaturation for 10 sec at 95 C, hybridization for 5 sec at 58–60 C, and elongation for 10 sec at 72 C. For each target, a standard corresponding to a fragment of the cDNA, cloned in pGEM-T easy vector (Promega Corp.), was serially diluted and quantified in parallel to generate a eight-point serial standard curve for the PCR analysis. Results were expressed in fmol (10⁻¹⁵ mol) or amol (10⁻¹⁸ mol)/μg of total RNA.

Statistical Analysis

Values are reported as mean ± sem. Statistical significance was determined by the unpaired Student’s t test.

Acknowledgments

We thank Patrick Gouait for valuable technical help in animal handling and Hubert Vidal for the access to the IFR 62 (Institut Fédéral de Recherche) platform.

This work was supported by grants to B.D. and W.W. from the Swiss National Science Foundation, the Etat de Vaud, l’Association Française de Nutrition—Société de Nutrition et de Diététique de Langue Française, the National Center of Competence in Research “Frontiers in Genetics,” the Human Frontier Science Program, and by Ilex.

^*

J.R. and F.T. contributed equally to this work.

^†

Present address: Institut National de la Santé et de la Recherche Médicale Unité 449, Faculté de médecine R.T.H. Laennec, rue Guillaume Paradin, 69372 Lyon cedex 08, France.

^‡

Present address: Department of Physiology/Neurobiology, University of Basel, Klingelbergstrasse 50/70, CH-4056 Basel, Switzerland.

Abbreviations:

 
• aP2,

  Adipocyte fatty acid binding protein;

 
• BADGE,

  bisphenol A diglycidyl ether;

 
• BAT,

  brown adipose tissue;

 
• CARLA,

  coactivator-dependent receptor ligand assay;

 
• CAT,

  chloramphenicol acetyl transferase;

 
• FAT/CD36,

  fatty acid translocase;

 
• FFA,

  free fatty acids;

 
• FXR,

  farnesoid X receptor;

 
• HFD,

  high fat diet;

 
• h,

  human;

 
• IBMX,

  3-isobutyl-1-methylxanthine;

 
• LDH,

  lactate deshydrogenase;

 
• LPL,

  lipoprotein lipase;

 
• m,

  murine;

 
• PPARγ,

  peroxisome proliferator-activated receptor γ;

 
• PPREs,

  peroxisome proliferator response elements;

 
• RXR,

  receptor of 9-cis-retinoic acid;

 
• SD,

  standard diet;

 
• SR-202,

  dimethyl α-(dimethoxyphosphinyl)-p-chlorobenzyl phosphate;

 
• SRC-1,

  steroid receptor coactivator-1;

 
• SREBP-1c,

  sterol regulatory element-binding protein 1c;

 
• TZDs,

  thiazolidinediones;

 
• WAT,

  white adipose tissue;

 
• wt,

  wild-type.