STAT5A Promotes Adipogenesis in Nonprecursor Cells and Associates With the Glucocorticoid Receptor During Adipocyte Differentiation

Dulbecco’s modified Eagle’s medium (DMEM) was purchased from Life Technologies. Bovine and fetal bovine serum (FBS) were obtained from Sigma and Life Technologies, respectively. Murine tumor necrosis factor (TNF)-α was purchased from Biosource International. The non-phospho STAT antibodies were monoclonal IgGs purchased from Transduction Laboratories or polyclonal IgGs from Santa Cruz. A highly phospho-specific polyclonal antibody for STAT5 (tyrosine [Y⁶⁹⁴]) was purchased from Upstate Biotechnologies. Anti-PPAR-γ was a mouse monoclonal (Santa Cruz). Anti-GR was a rabbit polyclonal (Santa Cruz). Darglitizone was generously provided by Pfizer (Groton, CT).

Murine 3T3-L1 preadipocytes were plated and grown to 2 days post-confluence in DMEM with 10% bovine serum. Medium was changed every 48 h. Cells were induced to differentiate by changing the medium to DMEM containing a standard induction cocktail of 10% FBS, 0.5 mmol/l 3-isobutyl-1-methylxanthine, 1 μmol/l dexamethasone, and 1.7 μmol/l insulin (MIX, dexamethasone, insulin [MDI]). After 48 h, this medium was replaced with DMEM supplemented with 10% FBS, and cells were maintained in this medium. NIH-3T3 and BALB/c fibroblast cell lines were obtained from the American Type Culture Collection (ATCC). Fibroblast cell lines ectopically expressing STAT5 genes were induced to differentiate with the standard (MDI) induction cocktail, in the presence or absence of 2.5 μmol/l darglitizone.

Retroviral-mediated stable expression of pBabe parental vector, pBabePPAR-γ2, pBabeSTAT5A, pBabeSTAT5B, and pBabeSTAT5A/B was generated in NIH-3T3 and BALB/c cell lines. Recombinant retroviruses were produced as described (22,23). Briefly, BOSC-23 packaging cells were transiently transfected with 20 μg of each retroviral construct. STAT5A/B combination transfections were carried out using 10 μg of each construct. At 8 and 24 h after transfection, the media were changed to DMEM and 10% calf serum, and at 48 h, the virus-containing media were collected, filtered (0.45 μm), and supplemented with 4 μg/ml polybrene. Target cells were incubated for 10 h with the retrovirus-containing solution. Puromycin (2.5 μg/ml) selection was initiated at 48 h and continued for 2 weeks.

Cell monolayers were rinsed with PBS and harvested in a nondenaturing buffer containing 150 mmol/l NaCl, 10 mmol/l Tris, pH 7.4, 1 mmol/l EGTA, 1 mmol/l EDTA, 1% Triton X-100, 0.5% Nonidet P-40, 1 μmol/l phenylmethylsulfonyl fluoride, 1 μmol/l pepstatin, 50 trypsin inhibitory milliunits of aprotinin, 10 μmol/l leupeptin, and 2 mmol/l sodium vanadate. Samples were extracted for 30 min on ice and centrifuged at 15,000 rpm at 4°C for 15 min. Supernatants containing whole-cell extracts were analyzed for protein content using a BCA kit (Pierce) according to the manufacturer’s instructions.

Cells were harvested under nondenaturing conditions, and the protein content of the whole-cell extracts was analyzed as described above. Protein extracts (1 mg/sample) were preincubated with protein A-Sepharose (RepliGen), and the resulting supernatant was incubated with polyclonal anti-GR antibody (2 μg) for 1 h at 4°C followed by incubation with protein A-Sepharose for an additional hour. The beads were rinsed twice with PBS, and immunoprecipitates were analyzed by SDS-PAGE followed by Western blotting with either STAT5A or GR antibodies.

Oil Red-O staining was performed as described by Green and Kehinde (21).

Proteins were separated in 7.5% polyacrylamide (National Diagnostics) gels containing SDS according to Laemmli (24) and transferred to nitrocellulose (Bio-Rad) in 25 mmol/l Tris, 192 mmol/l glycine, and 20% methanol. After transfer, the membrane was blocked in 4% milk for 1 h at room temperature. Results were visualized with horseradish peroxidase-conjugated secondary antibodies (Sigma) and enhanced chemiluminescence (Pierce).

Total RNA was isolated from cell monolayers with TriZOL (Invitrogen) according to the manufacturer’s instructions, with minor modifications. For Northern blot analysis, 20 μg total RNA was denatured in formamide and electrophoresed through a formaldehyde-agarose gel. The RNA was transferred to Zeta Probe-GT (Bio-Rad), cross-linked, hybridized, and washed as previously described (25). Probes were labeled by random priming using the Klenow Fragment (Promega) and [α³²P]dATP (Perkin Elmer, Life Sciences).

To assess the role of STAT5 in adipogenesis, we generated a series of stable cell lines that ectopically express STAT5 proteins. We used retroviral-mediated transfection of nonprecursor fibroblast cells (BALB/c and NIH-3T3) to produce cell lines containing the pBabe vector alone, pBabeSTAT5A, pBabeSTAT5B, or the pBabe STATs 5A and 5B (STAT5A/B) expression vectors in combination. To test our hypothesis that STAT5 proteins are involved in adipogenesis, BALB/c cells ectopically expressing STAT5 genes were induced to differentiate at 2 days after confluence with the standard differentiation cocktail. Adipogenesis was judged by morphologic and biochemical criteria. As shown in Fig. 1, we assessed the adipogenesis of STAT5 BALB/c stable cell lines by examining the expression of PPAR-γ, C/EBPα, aP2, and glycerol phosphate dehydrogenase (GPD). Cells were induced to differentiate in the absence or presence of darglitizone, and total RNA was isolated at the indicated times. Darglitizone (TZD) was added to ensure activation of any PPAR-γ expressed in the BALB/c or NIH-3T3 stable cell lines. Previous studies have shown that exogenous activators of PPAR-γ are necessary to obtain significant levels of adipogenesis in nonprecursor cell lines (18). As shown in Fig. 1A, we observed an induction in PPAR-γ mRNA in STAT5B cells at 3 days after the initiation of differentiation. Yet, the induction of PPAR-γ mRNA occurred at only 2 days after initiation of differentiation in STAT5A cells and occurred within 1 day in STAT5A/B cells. A similar pattern was observed for the induction of C/EBPα mRNA (Fig. 1B). In addition to examining two adipogenic transcription factors, PPAR-γ and C/EBPα, we examined the expression of aP2, a fat-specific lipid binding protein. As shown in Fig. 1C, the induction of aP2 mRNA occurred in STAT5B cells at day 3 and was expressed in a TZD-dependent manner. In STAT5A cells, aP2 mRNA was induced at day 2 in a TZD-dependent manner, and in STAT5A/B cells, this mRNA was induced at day 1 in a similar manner. Interestingly, the expression of aP2 in STAT5B cells always depended on the presence of TZD. Yet, in 5A cells or 5A/5B cells, the expression of aP2 mRNA was independent of TZD in the later stages of adipogenesis. The expression of GPD, a late marker of adipogenesis, was induced 3 days after the initiation of differentiation in STAT5A cells and at 2 days after initiation of differentiation in STAT5A/B cells in the presence of TZD. However, GPD expression was independent of TZDs at the later stages of differentiation in STAT5A/B cells (days 5 and 6). GPD mRNA was never induced in 5B cells (data not shown).

The adipogenesis of BALB/c cells ectopically expressing STAT5 genes was also assessed by staining for neutral lipids. BALB/c cell lines were induced to differentiate in the absence or presence of darglitizone and subjected to Oil Red-O staining 96 h after the initiation of differentiation. As shown in Fig. 2, the presence of STAT5A alone or STAT5A and STAT5B in combination resulted in a substantial increase in Oil Red-O staining. We were able to detect some lipid accumulation in cells containing an “empty” vector or STAT5B in the presence of a TZD. However, there was no significant difference in Oil Red-O staining in these two cell types. In addition, the increased lipid accumulation observed in STAT5A cells and STAT5A/B cells was independent of TZDs.

The adipogenic capabilities of PPAR-γ2 and STAT5 proteins were compared by isolating RNA from BALB/c cells that stably expressed vector, PPAR-γ2, STAT5A, STAT5B, or STATs 5A and 5B. Each cell line was induced to differentiate in the absence or presence of darglitizone (TZD), and total RNA and whole-cell extracts were isolated 120 h later. The expression of adipsin, a late marker fat-specific gene, was examined by Northern blot analysis. As shown in Fig. 3A, adipsin mRNA was highly induced in PPAR-γ2, STAT5A, and STAT5A/B cells in the absence of TZD. In addition, adipsin mRNA was expressed at substantially lower levels in vector and 5B cells. The expression of adipsin has been shown to be repressed by TZDs (26).

Whole-cell extracts were also analyzed for the expression of PPAR-γ2 and STAT5A by Western blot analysis and compared with the levels of protein expression in 3T3-L1 adipocytes. As shown in Fig. 3B, PPAR-γ protein is readily detectable in all samples except the vector. In the PPAR-γ2 cells, there are three types of PPAR-γ proteins: PPAR-γ1, PPAR-γ2, and the ectopically expressed PPAR-γ2. The ectopic PPAR-γ2 protein migrates slowest, and the expression of endogenous PPAR-γ2 and PPAR-γ1 is induced in these cells, as we also observed in cells ectopically expressing STAT5 genes. The levels of PPAR-γ in the BALB/c cells are comparable to the level of PPAR-γ found in 3T3-L1 adipocytes (Fig. 3B). The expression of STAT5A protein is detectable in vector and 5B cells, but the expression of this protein is induced in STAT5A cells and STAT5A/B cells. Also, the expression of STAT5A is elevated in PPAR-γ2 cells that were exposed to TZD treatment.

Because background levels of adipogenesis were observed in the BALB/c cells (Fig. 2E), we also examined the effect of ectopically expressed STAT5 genes in NIH-3T3 cells. The expression of various adipocyte markers was examined in NIH-3T3 cell lines ectopically expressing vector, STAT5A, STAT5B, or both STATs 5A and 5B. As shown in Fig. 4A, ectopic expression of STAT5A alone or STAT5A/B induced PPAR-γ mRNA expression. However, ectopic expression of STAT5B was insufficient to induce PPAR-γ mRNA in the NIH-3T3 cell lines above levels observed in vector cells. The induction of aP2 mRNA was slightly increased in a TZD-dependent manner in vector and 5B cells. However, there was a substantial induction of aP2 mRNA in STAT5A and STAT5A/B cells. Two late markers of adipogenesis, GPD and adipsin, were undetectable in vector or 5B cells. However, GPD was induced in a TZD-dependent manner in 5A and 5A/B cells, and adipsin was also significantly induced in these cells. Moreover, expression of GLUT4, the insulin-sensitive glucose transporter, was induced in a TZD-dependent manner in STAT5A and STAT5A/B cells. As an additional indicator of adipogenesis, we measured PPAR-γ protein levels in the NIH-3T3 cell lines. As shown in Fig. 4B, STAT5A or STAT5A/B cells induce expression of PPAR-γ, whereas STAT5B alone does not support PPAR-γ expression.

To study how STAT5 proteins are acting to influence the development of adipocytes, we examined the tyrosine phosphorylation of STAT5 during 3T3-L1 differentiation. These preadipocytes have the ability to differentiate into adipocytes that accumulated lipid, are insulin sensitive, and have the endocrine properties of native adipocytes. Whole-cell extracts were isolated over a time course of differentiation in 3T3-L1 cells and examined for STAT5 phosphorylation with a phospho-specific antibody that recognizes STAT5A and STAT5B when tyrosine⁶⁹⁴ is phosphorylated. As shown in Fig. 5A, STAT5 proteins were tyrosine-phosphorylated early during adipogenesis and before the expression of PPAR-γ. The phosphorylation of both STATs 5A and 5B was confirmed by immunoprecipitation with STAT5 antibodies and immunoblotting with an anti-phosphotyrosine antibody (data not shown). These results suggest that the tyrosine phosphorylation of STAT5 proteins has a role in adipogenesis. To address this question, we induced the STAT5A/B BALB/c cells to differentiate under conditions where a component of the induction cocktail was missing. We have previously shown that methyl-isobutyl-xanthine (MIX), insulin, or dexamethasone do not induce the tyrosine phosphorylation of STAT5 proteins in these cells (27). However, FBS treatment results in the tyrosine phosphorylation and nuclear translocation of STAT5 proteins, likely due to the presence of GH in FBS (27). Previous studies have demonstrated that MIX, DEX, and FBS are required for complete adipogenesis of 3T3-L1 cells (16) and nonprecursor cells ectopically expressing PPAR-γ or C/EBPs (17–20). Therefore, we examined the requirement of components of the induction cocktail (FBS, MIX, DEX, and insulin) in the differentiation of STAT5A/B BALB/c cells. The results in Fig. 5B demonstrate that the standard induction cocktail resulted in the induction of PPAR-γ expression in STAT5A/B BALB/c cells. Yet, in the absence of MIX, DEX, or FBS, these cells did not express PPAR-γ protein. However, if FBS was replaced with GH in 1% calf serum, the ability of STAT5A/B BALB/c cells to undergo adipogenesis and express PPAR-γ was restored. This regulation of PPAR-γ was identical in STAT5A BALB/c cells (data not shown).

Finally, we examined the interaction of STAT5A with GR. We chose to examine GR because it is highly associated with obesity and because the GR has been previously shown to associate with STAT5 in other cell types (11,13). In addition, the requirement of dexamethasone for expression of PPAR-γ in the STAT5A/B BALB/c cells suggests a role for GR in STAT5-mediated adipogenesis. To determine whether GR and STAT5A interact during adipogenesis, we harvested whole-cell extracts at various times during adipogenesis and performed Western blot analysis on whole-cell extracts as well as lysates from immunoprecipitations carried out against GR. As shown in Fig. 6A, GR levels are not regulated during adipogenesis, whereas STAT5A levels increase at 48 h after the initiation of differentiation. The results from immunoprecipitation experiments with anti-GR indicate that STAT5A associates with GR at 72 h after the initiation of differentiation. This interaction is maintained at 96, 120, and 144 h after the initiation of differentiation, when cells are fully differentiated. Interestingly, this association was undetectable in confluent preadipocytes (time 0) or in the early stages of adipogenesis. To further examine the role of STAT5A/GR association during adipogenesis, confluent 3T3-L1 preadipocytes were induced to differentiate in the presence of TNF-α. TNF-α is known to inhibit adipogenesis (25). The results in Fig. 6B demonstrate that the STAT5A-GR interaction that occurs in the later stages of adipogenesis is inhibited by TNF-α. Furthermore, the levels of both GR and STAT5A decline in the presence of TNF-α.

These studies demonstrate that STAT5A promotes adipogenesis in nonprecursor cells. Moreover, the adipogenesis of STAT5A cells or STAT5A/B cells was TZD independent, suggesting these cells are capable of making an endogenous PPAR-γ ligand. Our results also indicate that STAT5A is sufficient to induce adipogenesis in nonprecursor cells, and this adipogenesis is accompanied by an induction of PPAR-γ and C/EBPα in BALB/c cells. Although PPAR-γ is expressed at comparable levels in the BALB/c STAT5B cells, STAT5B is not adipogenic on its own (Figs. 1–4) but does enhance the adipogenic properties of STAT5A (Fig. 1A–D). STAT5A differs from STAT5B by eight amino acids at the COOH-terminus, and our results suggest that the COOH-terminal regions of STAT5A and STAT5B confer separate roles for these proteins in adipogenesis.

It is known that PPAR-γ promotes adipogenesis in C/EBPα-deficient cells, yet C/EBPα cannot promote adipogenesis in the absence of PPAR-γ (28). These recent findings suggest that C/EBPα and PPAR-γ participate in a single pathway of fat cell development, with PPAR-γ being the proximal effector of adipogenesis (28). We surmise that STAT5 proteins participate in this identified pathway of fat cell development and hypothesize it is likely that STAT5A is an early effector of adipogenesis. This is supported by our data demonstrating that ectopic STAT5A expression results in the induction of PPAR-γ expression in both BALB/c and NIH-3T3 fibroblasts (Figs. 1, 3, and 4) as well as in the early activation of STAT5 preceding the expression of PPAR-γ in 3T3-L1 adipocytes (Fig. 5A). The ability of STAT5A to induce adipogenesis in nonprecursor cells may be mediated by its ability to directly regulate PPAR-γ expression. Alternatively, the role of STAT5A in regulating adipogenesis could be mediated by its association with GR or other unidentified co-activators.

Interestingly, STAT5A and GR can associate in NIH-3T3 fibroblasts (13), but we have shown these two transcription factors are not associated in 3T3-L1 preadipocytes (Fig. 6A). The association of STAT5A and GR occurs late in adipogenesis, coincident with increasing levels of STAT5A, and may have an inhibitory role because STAT5 has been shown to inhibit GR-mediated regulation of transcription in other cell types (15). It seems unlikely that the interaction of STAT5A with GR depends on the concentration of STAT5A because NIH-3T3 and 3T3-L1 have similar levels of STAT5A expression (data not shown). Hence, we hypothesize that other factors contribute to the GR/STAT5A association that occurs in the later stages of adipogenesis.

Additionally, our data demonstrate that GR expression is not regulated during adipogenesis, but the association between GR and STAT5A is highly regulated during fat cell differentiation and that inhibition of differentiation is accompanied by an inhibition in the STAT5A/GR interaction (Fig. 6B). These results indicate that the association of STAT5A and GR may be important in regulating adipocyte gene expression, and future studies will focus on understanding this relationship. Furthermore, the effect of TNF-α on the association of STAT5A and GR suggests that the interaction of these transcription factors in adipocytes may play a role in insulin sensitivity. There is long-standing evidence that GR in the nervous system affects energy balance (29), but recent studies suggest that peripheral GR in adipose tissue also contributes to abdominal obesity (30). In particular, the effect of glucocorticoids produced in adipose tissue has been examined in animal studies. Transgenic mice overexpressing 11β-hydroxysteroid dehydrogenase type I selectively in adipose tissue exhibited pronounced insulin-resistant diabetes, hyperlipidemia, and hyperphagia (31). Taken together, these studies strongly suggest that there are GR-dependent pathways in adipocytes that contribute to systemic biology.

In summary, our findings suggest that STAT5A and STAT5B have nonredundant roles in adipogenesis and that STAT5A, alone or in the presence of STAT5B, is adipogenic in nonprecursor cells. These findings support recent studies demonstrating that STAT5 proteins are highly regulated in adipocytes (5,7), that they have the ability to influence adipogenesis (8,9), and that transgenic mice lacking either STAT5 gene have a reduction in fat pad size (4). Hence, it is not unexpected that STAT5A is adipogenic in nonprecursor cells. However, GH, a potent STAT5A and STAT5B activator, inhibits the expression of many adipocyte genes such as fatty acid synthase and GLUT4 in fully differentiated adipocytes (32). We predict that STAT5 proteins can exert both negative and positive effects on adipocyte gene expression, perhaps via their association with GR.

This work was supported by grant R01DK52968-02 from the National Institutes of Health (to J.M.S.).

We would like to thank Patricia Arbour-Reily and Kyle J. Waite for technical assistance and Dr. Evanna Gleason for expert help with the microscopy.