Effects of Pioglitazone on Suppressor of Cytokine Signaling 3 Expression: Potential Mechanisms for Its Effects on Insulin Sensitivity and Adiponectin Expression

Monoclonal anti-phosphotyrosine antibodies (PY 20) were purchased from Transduction Laboratories (Lexington, KY). Anti-myc antibody was from Upstate Biotechnology (Lake Placid, NY). Anti–phospho-Akt, anti-STAT3, anti–phospho-STAT3, and anti-actin antibodies were from Cell Signaling Technology (Beverly, MA). Anti-SOCS3 antibody was from Immuno-Biological Laboratories (Gunma, Japan). Interleukin (IL)-6 and growth hormone (GH) were from Sigma (St. Louis, MO). AG490 was from Sigma. Mouse Adiponectin ELISA kit was from R&D Systems (Minneapolis, MN). IL-1α and TNF-α were provided by Dainippon Pharmaceutical (Osaka, Japan). Pioglitazone was a gift of Takeda Pharmaceutical (Osaka, Japan).

Murine 3T3-L1 cells were obtained from American Type Culture Collection (Manassas, VA). Briefly, after confluence, cells were left for 2 more days in Dulbecco’s modified Eagle’s medium/high glucose supplemented with 100 units/ml streptomycin and 10% fetal bovine serum in a 10% CO₂ environment. Differentiation was induced by changing the culture medium to the same one containing 0.5 mmol/l 3-isobutyl-1 methylxanthine, 1 μmol/l dexamethasone, and 1 μmol/l insulin for 3 days, followed by the culture in the medium containing 0.8 μmol/l insulin for another 3 days. The medium was changed every 3 days. The cells were used for experiments 14–16 days after the induction of differentiation.

Adenovirus vectors containing cDNAs encoding wild-type PPARγ and PPARγ S112A, in which Ser112 is replaced with Ala, were constructed as previously described (24). A virus vector encoding β-galactosidase (LacZ) was used as a control (24). Adenovirus vectors encoding myc-tagged SOCS1 and SOCS3 were provided by Dr. Naka (Osaka University, Osaka, Japan). Adenovirus vectors encoding dominant-negative STAT3 were provided by Dr. Kunisada (Osaka University, Osaka, Japan) (25). Differentiated 3T3-L1 adipocytes were infected with adenovirus 48 h before the experiments.

Total RNA was extracted from 3T3-L1 adipocytes or mouse tissues using an ISOGEN kit or a RNeasy kit, respectively. Quantitative RT-PCR was conducted according to the manufacturer’s protocol (Applied Biosystems, Foster City, CA). Briefly, we synthesized cDNA using oligo (dT) primers with the TaqMan Reverse Transcription Reagents. Reverse-transcribed cDNA was mixed with PCR Master Mix and gene-specific Assays-on-Demand Gene Expression Products and amplified on an ABI PRISM 7700 (Applied Biosystems). Results were normalized against glyceraldehyde-3-phosphate dehydrogenase (GAPDH) gene expression.

Six-week-old C57BLKS/J db/db mice, C57BLS/J db/+m mice, and C57BL/6J mice were purchased from Clea Japan (Tokyo, Japan). All animals were housed on a 12-h light-dark cycle. In the experiments, on high-fat–fed mice, C57BL/6J mice were fed a standard chow or a high-fat diet (Quick Fat; Clea Japan) for 12 weeks. Pioglitazone (10 mg · kg⁻¹ · day⁻¹) was mixed with the food and orally administered to db/db mice or high-fat–fed mice for 2 weeks. All protocols for animal use and euthanasia were approved by Guide for Animal Experiments of University of Toyama.

For animal experiments, human insulin (5 units/kg) was injected intraperitoneally. Epididymal fat pad, liver, and quadriceps muscles were removed at 10 min after injection. These tissues were homogenized for 1 min at 4°C in lysis buffer containing 25 mmol/l Tris-HCl (pH 7.4), 10 mmol/l Na₃VO₄, 100 mmol/l NaF, 50 mmol/l Na₄P₂O₇, 10 mmol/l EGTA, 10 mmol/l EDTA, 5 mg leupeptin/ml, 5 mg aprotinin/ml, 2 mmol/l phenylmethylsulfonyl fluoride, and 1% Nonidet-P 40. 3T3-L1 adipocytes were lysed in a cell-solublizing buffer containing 30 mmol/l Tris, pH 7.4, 150 mmol/l NaCl, 10 mmol/l EDTA, 1% Nonidet-P40, 1 mmol/l phenylmethylsulfonyl fluoride, 10 μg/ml aprotinin, 1 μmol/l leupeptin, 1 mmol/l Na₃VO₄, and 50 mmol/l NaF. For immunoprecipitation, whole-cell lysates were centrifuged at 4°C for 20 min to remove the insoluble materials. Western blot and immunoprecipitation were performed as described previously (7).

The amounts of adiponectin in culture medium for 3T3-L1 adipocytes or in mouse serum were measured using Quantikine Mouse Adiponectin/Acrp30 Immunoassay kit according to the manufacturer’s instruction (R&D Systems). The absorbance was measured by Labsystem iEMS Reader MF (Labsystems, Franklin, MA).

All data are presented as means ± SE. The statistical comparison between the groups was carried out using ANOVA or Student’s t test. P values <0.05 were considered statistically significant.

We first examined the expression levels of SOCS3 mRNA in insulin target tissues of db/db mice. As recently reported in db/db mice and high-fat–fed mice (9,14,16), SOCS3 mRNA level in epididymal fat tissues of 8-week-old db/db mice was increased to 2.7-fold of that in control C57BL/6J mice. After the oral administration of pioglitazone for 2 weeks, SOCS3 mRNA was decreased by ∼50% of that in epididymal fat tissues of db/db mice without pioglitazone (Fig. 1A). In the liver, whereas baseline SOCS3 expression of db/db mice was not increased compared with control lean mice, it was decreased by pioglitazone (Fig. 1B). In quadriceps muscle, no changes were observed between pioglitazone-treated and nontreated animals (Fig. 1C). The expression of SOCS3 in high-fat–fed mice was also changed by pioglitazone, in a manner similar to the changes seen in the db/db mice. SOCS3 expression in epididymal fat of high-fat–fed mice was increased compared with that in control mice fed standard diet and was decreased by pioglitazone administration (Fig. 1D). No changes of SOCS3 expression were observed in either liver or skeletal muscle of high-fat–fed mice, and subsequent pioglitazone treatment did not alter SOCS3 mRNA in these tissues (data not shown). We confirmed the previous reports that the same treatment with pioglitazone for 2 weeks improved the glucose profile during intraperitoneal glucose tolerance test and intraperitoneal insulin tolerance test in db/db mice (data not shown). Pioglitazone treatment also restored Akt phosphorylation in fat, liver, and skeletal muscle of db/db mice (data not shown), as reported recently (26,27).

Small molecule mediators of insulin resistance may have as a common mechanism the ability to increase the expression of SOCS proteins in insulin target tissues. Thus, we examined the effects of the representative mediators for insulin resistance on the induction of SOCSs in cultured 3T3-L1 adipocytes. All the stimuli, including IL-6, GH, TNF-α, and insulin, enhanced the expression of mRNAs for SOCS1 and SOCS3. Interestingly, the induction patterns of SOCS3 were varied among the mediators, e.g., the maximal induction was observed at the different time points (data not shown).

So far, it has not been clarified whether thiazolidinedione can prevent increases in SOCS expression by insulin-desensitizing agents. To examine this issue, we evaluated the effect of pioglitazone on SOCS3 and SOCS1 induction in 3T3-L1 adipocytes. SOCS3 mRNA induced by the stimulations with 1-h IL-6, 1-h GH, 8-h TNF-α, and 2-h insulin were suppressed by the pretreatment with 10 μmol/l pioglitazone by 35, 49, 53, and 61%, respectively (Fig. 2A). In contrast, pioglitazone did not change the SOCS1 mRNA level induced by the same stimuli (Fig. 2B). Figure 2C shows a dose-dependent effect of pioglitazone on the suppression of SOCS3 mRNA level induced by TNF-α. Although the suppression by pioglitazone was dose dependent, the maximal suppressive effect was partial. Because of this partial response, we wondered whether the level of PPARγ protein present in the target tissue influenced that the response of the tissue to pioglitazone treatment. Pioglitazone is reported to exert its various functions mainly through the PPARγ-dependent mechanisms (23,28). Therefore, we next examined the involvement of PPARγ in reducing SOCS3 expression by pioglitazone by overexpressing exogenous PPARγ proteins. Expression of wild-type PPARγ alone did not affect TNF-α–induced SOCS3 expression, but wild-type PPARγ clearly enhanced the ability of pioglitazone to suppress SOCS3 induction at an almost complete level (90% of suppression) (Fig. 2D). Mitogen-activated protein (MAP) kinase is known to phosphorylate Ser112 on PPARγ to decrease PPARγ transcriptional activity, when it is activated by TNF-α stimulation. Thus, we used an adenovirus vector encoding a mutant PPARγ, in which Ser112 is replaced by Ala, which cannot be phosphorylated by MAP kinase (S112A-PPARγ). In contrast to wild-type PPARγ, S112A-PPARγ expression inhibited TNF-α–induced SOCS3 expression by itself, which was further suppressed by pioglitazone. This result suggests that the reason for the failure of wild-type PPARγ to suppress SOCS3 expression may be the phosphorylation on Ser112 by TNF-α. Taken together, these results further suggest the involvement of active PPARγ in the inhibition of SOCS3 induction by pioglitazone.

Adiponectin is an adipokine that is secreted from fat tissue and antagonizes pro-inflammatory signals. To test the hypothesis that SOCS3 expression in fat cells may affect insulin sensitivity in the other tissues, we examined whether altered expression of SOCS3 affects adiponectin production in 3T3-L1 adipocytes. The concentration of adiponectin in the original culture media with 10% FCS was negligible, which became detectable by our enzyme-linked immunosorbent assay detection kit after culturing with fully differentiated 3T3-L1 adipocytes for >24 h. Overexpression of SOCS3 by adenovirus vectors significantly lowered the concentration of adiponectin in the media cultured with the cells for 72 h to ∼70% of the control without SOCS3 overexpression (Fig. 3A). In contrast, adenovirus-mediated SOCS1 overexpression, which was much more than the maximal induction of endogenous SOCS1 by TNF-α (data not shown), did not affect the concentration of adiponectin (Fig. 3A). Quantitative RT-PCR analysis confirmed that SOCS3 overexpression suppressed adiponectin expression at mRNA level (Fig. 3B). However, overexpression of SOCS3 did not affect STAT3 protein level, but it completely inhibited STAT3 phosphorylation (Fig. 3C).

Because SOCS3 is an endogenous inhibitor of JAK2-STAT3 pathway, these findings suggested that inactivation of JAK2-STAT3 pathway may be involved in the partial inhibition of adiponectin production by SOCS3 overexpression. To test this possibility, we next examined the effect of tyrphostin AG490, a specific inhibitor of JAK2, on adiponectin level. As expected, AG490 completely inhibited phosphorylation of JAK2 and STAT3 (Fig. 4C). AG490 also suppressed the expression of adiponectin at the protein level (Fig. 4A) and at the mRNA level (Fig. 4B) in a dose-dependent manner. To directly investigate the roles of STAT3, the effects of adenovirus-mediated overexpression of a dominant-negative STAT3 on adiponectin production was examined. Secreted adiponectin and adiponectin mRNA were reduced by dominant-negative STAT3 overexpression in an multiplicity of infection-dependent manner (Fig. 4D and E). Expression of dominant-negative STAT3 almost completely inhibited STAT3 phosphorylation (Fig. 4F), but the inhibitory effects of dominant-negative STAT3 on adiponectin production were less potent than those of SOCS3 overexpression or JAK2 inhibitor on adiponectin production (Figs. 3 and 4A–C).

Finally, we examined the effects of pioglitazone on adiponectin production and SOCS3 expression in vitro and in vivo. In contrast to the effect of JAK2 inhibitor or expression of dominant-negative STAT3, pioglitazone increased adiponectin expression at the protein level (Fig. 5A) and at the mRNA level (Fig. 5B) in a dose-dependent manner, which was paralleled with the increased STAT3 phosphorylation (Fig. 5C) in 3T3-L1 adipocytes. Furthermore, oral administration of pioglitazone for 2 weeks increased serum adiponectin concentration (Fig. 5D) and mRNA of adiponectin in epididymal fat (Fig. 5E) of db/db mice with increased tyrosine phosphorylation of STAT3 in the fat tissue (Fig. 5F).

Recent studies have shown that SOCS proteins may impair insulin signaling and glucose metabolism in experiments with cultured cells (8,9) or mouse models (10) overexpressing SOCS proteins. In addition, when the expression of SOCS is suppressed in knockout mice or in mouse models treated with an antisense oligonucleotide for SOCS gene, insulin sensitivity is improved (10,19). In this study, we now show that the expression of endogenous SOCS3 is increased in fat tissues of insulin-resistant mice (Fig. 1) and in 3T3-L1 adipocytes treated with cytokines or hormones known to induce insulin resistance (Fig. 2) as previously reported (9,14,16,29). The synthetic insulin sensitizer, pioglitazone, suppresses the expression of SOCS3 and improves insulin sensitivity in these in vivo and in vitro insulin-resistant models via PPARγ-dependent mechanisms. These results indicate that altered expression levels of endogenous SOCS3 in fat cells may modulate insulin sensitivity in conditions associated with insulin resistance and that SOCS3 is a target gene, whose expression is modified by pioglitazone.

In the second half of this study, we have shown that SOCS3 in fat cells inhibits adiponectin production via JAK2-STAT3–dependent mechanisms. These results are very important when we understand how SOCS3 in the fat tissue regulates whole-body insulin sensitivity, i.e., STAT3 phosphorylation was inhibited in experiments in which SOCS3 overexpression decreased adiponectin production (Fig. 3). Additionally, a chemical inhibitor of JAK2 and expression of dominant-negative STAT3 also inhibited adiponectin production (Fig. 4). These results suggest that the physiological activation of JAK2 and, at least in part, STAT3 may be necessary for the production of adiponectin in fat cells and that increased expression of SOCS3 may decrease adiponectin production by inhibiting this pathway. The mechanism of how the activation of JAK2-STAT3 is related to adiponectin production is currently unclear. The consensus sequence of the STAT3 binding site has not yet been found in the promoter region of adiponectin gene, whereas those of several transcription factors, such as PPARγ, SREBP1-c, and C/EBP, have been reported (30–32). One of the possibilities is that activation of STAT3 may increase adiponectin mRNA at the step other than the transcription level, e.g., increasing the stability of mRNA. Alternatively, STAT3 may indirectly enhance adiponectin production via enhancing the activity of other transcription factors. It should also be noted that the inhibitory effects of SOCS3, the JAK2 inhibitor, and dominant-negative STAT3 on adiponectin production were all partial, as shown in Figs. 3 and 4. This result suggests the involvement of the transcription factors other than STAT3, such as PPARγ, in the regulation of gene expression of adiponectin. SOCS3 may not inhibit the activation of all these transcription factors. If so, this not only helps explain the partial inhibition of adiponectin production by SOCS3, dominant-negative STAT3, or JAK2 inhibitor (Figs. 3 and 4), but also may account for the small increase in adiponectin production by pioglitazone, which decreased SOCS3 expression and enhanced STAT3 phosphorylation (Fig. 5D). The precise mechanisms by which active JAK2-STAT3 pathway enhances adiponectin production should be more thoroughly investigated.

We have proposed that increased adiponectin production, after pioglitazone-mediated suppression of SOCS3 expression, plays an important role in improved insulin sensitivity. But in general, SOCS3 regulates signal transduction of many other factors related to insulin sensitivity. Therefore, improved insulin sensitivity after reduced SOCS3 expression may not be explained only by increased adiponectin production. For example, insulin signaling is the most important target directly affected by SOCS3 expression among such factors. As reported in several previous papers, SOCS3 directly inhibits insulin signaling via the association to insulin receptor or the degradation of IRS proteins (8,9,14–16). Thus, decreased expression of SOCS3 by pioglitazone directly leads to the enhancement of insulin signaling in insulin-target tissues. Furthermore, some adipokines secreted from adipocytes should also be considered. We examined the effects of forced expression of SOCS3 on the production of adiponectin, TNF-α, IL-1β, and IL-6. Only the adiponectin production was significantly changed by the forced expression of SOCS3, but the others were not (data not shown). Finally, it is highly possible that SOCS3 negatively regulated leptin signaling in our system of db/db mice as already reported (33,34). Unfortunately, we could not test the effects of SOCS3 on leptin signaling in the current study. It should be examined in the near future. Moreover, the improvement of insulin sensitivity by pioglitazone is not fully regulated by the expression level of SOCS3, but some other possible mechanisms are suggested, such as decreases of adipocyte size, free fatty acid, resistin, and TNF-α (35,36). An increase of adiponectin observed in our db/db mouse model (Fig. 5D) is a part of such complex mechanisms by which pioglitazone ameliorates insulin sensitivity.

Adiponectin exists in plasma creating three major oligometric forms: low molecular weight, middle molecular weight, and a high molecular weight. Thiazolidinediones (TZDs) are reported to upregulate high–molecular weight adiponectin predominantly (27). Unfortunately, we could not examine the effects of pioglitazone on each form of adiponectin. But it is highly possible that pioglitazone may increase high–molecular weight adiponectin predominantly in our experimental system as in the previous reports. At least two possible mechanisms are recently reported by which TZDs increase high–molecular weight adiponectin, i.e., pioglitazone directly facilitates the generation of high–molecular weight adiponectin in addition to the activation of adiponectin gene transcription (35). We have demonstrated that pioglitazone increases adiponectin production accompanied by the enhancement of STAT3 activation via the reduced expression of SOCS3. It is not clear whether the present data on STAT3 and SOCS3 are related to the facilitated generation of high–molecular weight adiponectin. Further studies are necessary to address this issue.

A recent study investigated the role of SOCS3 in fat tissue by creating fat-specific SOCS3 transgenic mice (17). In these mice, overexpression of SOCS3 in adipose tissue inhibited local insulin action but improved systemic glucose metabolism with high-fat diet conditions. Adiponectin production was increased, which was accompanied by the decreased amount of fat tissue. Their data seem quite different from our current results, in which suppression of SOCS3 by pioglitazone was observed together with improved glucose metabolism and increased adiponectin production. The improvement of systemic glucose metabolism and the increment of adiponectin observed in these transgenic mice may be associated with decreased adipocyte cell size as the authors suggested. In their transgenic mice, insulin sensitivity and adiponectin production may be regulated by the factors associated with altered adipocyte differentiation. Because the systems are very different, their results are not necessarily inconsistent with our current results.

In conclusion, our results indicate that SOCS3 levels are increased in the pathological conditions of insulin resistance and that pioglitazone suppresses SOCS3 expression through the activation of PPARγ. Some parts of the insulin-sensitizing effect of pioglitazone involve its effect on SOCS3 expression in fat tissue, which is associated with the enhanced JAK2/STAT3 activity and increased production of adiponectin, leading to improved whole-body insulin sensitivity.

M.K. has received a Grant-in-Aid for Scientific Research from the Ministry of Education, Science, Sports, and Culture, Japan (17590918).