The Regulation of Fatty Acid Synthase by STAT5A

Dulbecco’s modified Eagle medium (DMEM) and leukemia inhibitory factor (LIF) were purchased from Invitrogen. Fetal bovine serum was purchased from Atlanta Biologicals, and calf serum was purchased from Biosource. Polyclonal phospho-specific STAT5 (Y⁶⁹⁴) antibody and FAS antibody were purchased from BD Transduction Laboratories. STAT1 antibody was purchased from Upstate Biotechnology. Porcine growth hormone, ovine PRL (both prepared from pituitaries), and cycloheximide were purchased from Sigma. STAT3 and STAT5A antibodies were purchased from Santa Cruz. [α-³²P] dCTP was purchased from Perkin-Elmer and Amersham Biosciences. Deoxynucleotide thymine triphosphate, dATP, and dGTP were purchased from Amersham Biosciences. Oligonucleotides were purchased from Integrated DNA Technologies. DNase polymerase I large (Klenow) fragment was purchased from Promega.

Murine 3T3-L1 preadipocytes were plated and grown to 2 days postconfluence in DMEM containing 10% bovine serum. Medium was changed every 48 h. Cells were induced to differentiate by changing the medium to DMEM containing 10% fetal bovine serum, 0.5 mmol/l 3-isobutyl-methylxanthine, 1 μmol/l dexamethasone, and 1.7 μmol/l insulin. After 48 h, this medium was replaced with DMEM supplemented with 10% fetal bovine serum, and the cells were maintained in this medium until used for experimentation.

Cell monolayers were rinsed with PBS and then harvested in a nondenaturing buffer containing 10 mmol/l Tris (pH 7.4), 150 mmol/l NaCl, 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 mU 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 by bicinchoninic acid analysis (Pierce) according to the manufacturer’s instructions.

Total RNA was isolated from cell monolayers with Trizol (Invitrogen), according to manufacturer’s instructions with minor modifications. For Northern blot analysis, 15 μg total RNA was denatured in formamide and electrophoresed through a formaldehyde/agarose gel. The RNA was transferred to Zeta Probe-GT (Bio-Rad) in a buffer containing 75 mmol/l sodium citrate tribasic, 10 mmol/l NaOH, and 750 mmol/l NaCl. The blots were cross-linked, hybridized, and washed as previously described (29). Probes were labeled by random priming using Klenow fragment and [α-³²P] dCTP.

The rat FAS promoter −250 to +65/luciferase construct was generously provided by Dr. Steve Clarke. The rat FAS promoter −1,594 to +65/luciferase and −700 to +65/luciferase constructs were generously provided by Dr. Peter Tontonoz. The FAS −1,594 to +65/luciferase construct was mutated at positions −901 and −902 within the STAT5A binding site using the QuikChange Site Directed Mutagenesis Kit, according to manufacturer’s instructions (Stratagene). The following oligonucleotide and corresponding antisense oligonucleotide were used to alter the STAT binding site with the altered bases underlined: GGG AGG GTG AGG GTC AAG GAA ACC AGC AAC TCA GG. Sequence analysis was performed to confirm the presence of the mutated bases using Big Dye Terminator Extension Reaction (ABI Prism). The minimum promoter thymidine kinase (TK) renilla vector was purchased from Promega.

3T3-L1 preadipocytes were transiently cotransfected with the various FAS promoter constructs and the TK/renilla vector to control for transfection efficiency, as previously described (30), using Polyfect (Qiagen) according to manufacturer’s instructions. Cell lysates were analyzed for firefly and renilla luciferase activity using the Dual Luciferase Reporter Assay System (Promega). Relative light units were determined by dividing firefly luciferase activity by renilla luciferase activity. Results are given as ±SD.

Cell monolayers were rinsed with PBS and then harvested in a nuclear homogenization buffer containing 20 mmol/l Tris (pH 7.4), 10 mmol/l NaCl, 3 mmol/l MgCl₂, 1 μmol/l dithiothreitol, 1 μmol/l phenylmethylsulfonyl fluoride, 1 μmol/l pepstatin, 50 trypsin inhibitory mU aprotinin, 10 μmol/l leupeptin, and 2 mmol/l sodium vanadate. Igepal CA-630 (Nonidet P-40) was added to a final concentration of 0.15%, and cells were homogenized with 16 strokes in a Dounce homogenizer. The homogenates were centrifuged at 1,500 rpm for 5 min. Supernatants were saved as cytosolic extract, and the nuclear pellets were resuspended in one-half volume of nuclear homogenization buffer and were centrifuged as before. The pellet of intact nuclei was resuspended again in one-half of the original volume of nuclear homogenization buffer and centrifuged again. The majority of the pellet (intact nuclei) was resuspended in an extraction buffer containing 20 mmol/l HEPES (pH 7.9), 420 mmol/l NaCl, 1.5 mmol/l MgCl₂, 0.2 mmol/l EDTA, 1 μmol/l dithiothreitol, 1 μmol/l phenylmethylsulfonyl fluoride, 1 μmol/l pepstatin, 50 trypsin inhibitory mU aprotinin, 10 μmol/l leupeptin, 2 mmol/l sodium vanadate, and 25% glycerol. Nuclei were extracted for 30 min on ice. The samples were subjected to centrifugation at 10,000 rpm at 4°C for 10 min. Supernatants containing nuclear extracts were analyzed for protein content, using a bicinchoninic acid protein assay kit (Pierce).

Proteins were separated in 6% polyacrylamide (National Diagnostics) gels containing SDS according to Laemmli (31) and transferred to nitrocellulose (Bio-Rad) in 25 mmol/l Tris, 192 mmol/l glycine, and 20% methanol. Following transfer, the membrane was blocked in 4% milk overnight at 4°C. Results were visualized with horseradish peroxidase–conjugated secondary antibodies (Jackson ImmunoResearch Laboratories) and enhanced chemiluminescence (Pierce).

Double-stranded oligonucleotides were end labeled with [α-³²P] dCTP using Klenow. Binding reactions were performed with nuclear extracts, according to Ritzenthaler et al. (32). Protein-DNA complexes were resolved and visualized, as previously described (30). For supershift analysis, nuclear extracts were preincubated with 4 μg antibody for 1 h at room temperature.

GH is known to reduce adipose tissue in vivo (11) and decrease the expression of FAS in adipocytes (10). GH is also a potent activator of STAT5 in adipocytes (27). Hence, we hypothesized that STAT5 proteins directly modulate FAS expression. Therefore, we examined the regulation of FAS by two STAT5 activators, GH and PRL. As shown in Fig. 1A, we observed that both GH and PRL resulted in a decrease in FAS mRNA in fully differentiated 3T3-L1 adipocytes. GH decreased FAS mRNA by 12 h of treatment, and PRL treatment resulted in decreased expression within 6 h. In an independent experiment, we observed that both GH and PRL also resulted in a decrease in expression of FAS protein levels (Fig. 1B). To demonstrate the specificity of this effect, fully differentiated 3T3-L1 adipocytes were treated with PRL for 8 h with the various doses indicated in Fig. 1C. There was a notable decrease in FAS protein level with 8.8 nmol/l PRL treatment, and the decrease of FAS protein levels by PRL treatment was dose dependent. An acute treatment of 1.3 μmol/l PRL was included as a positive control for STAT5 phosphorylation. The level of STAT5A expression was unchanged by GH or PRL (Fig. 1B and C) and is shown to indicate even loading of protein samples.

Clearly, the analysis of mRNA and whole-cell extracts demonstrated that activators of STAT5 decreased expression of FAS protein and mRNA. Yet, it was unclear if the effects of GH and PRL were mediated by affecting FAS transcription and/or protein turnover. To assess whether the effects of PRL on FAS expression could be attributed to changes in the turnover of FAS protein, we examined the loss of FAS in the presence of cycloheximide (5 μmol/l) or ethanol, a vehicle control. Whole-cell extracts were collected at various times and used for Western blot analysis. As shown in Fig. 2, either cycloheximide or PRL treatment caused a decrease in FAS protein. In the presence of cycloheximide, a loss of FAS was observed at 8 h, regardless of the presence of PRL. PRL treatment alone decreased the level of FAS within 12 h of treatment. These results indicate that PRL does not affect the turnover of FAS protein in 3T3-L1 adipocytes and suggest that PRL may exert its effects on FAS at the level of transcription.

Our data indicate that PRL directly regulates the expression of FAS. Recent studies by Yin, Clarke, and Etherton (28) have shown that GH abrogates the induction of FAS by insulin and suggested that the −112 to +65 region of the rat FAS promoter was sensitive to the regulation by GH. Therefore, we hypothesized that a STAT5 recognition element may be present in this region. To address this question, 3T3-L1 preadipocytes were transiently transfected with a luciferase construct containing the FAS promoter fragment of −250 to +65. After 48 h, cells were stimulated with PRL for the times indicated in Fig. 3Aand were then analyzed for luciferase activity as described in research design and methods. We observed that PRL treatment had no significant effects on the relative luciferase activity (Fig. 3A), clearly demonstrating that the −250 to +65 region of the FAS promoter is not sensitive to PRL. Therefore, we examined the PRL responsiveness of two additional rat FAS promoter/luciferase constructs that incorporated larger regions, −1,594 to +65 and −700 to +65, of the promoter. Transfection of these constructs into 3T3-L1 cells revealed a PRL-responsive region present between −1,594 and −700 of the rat FAS promoter. As shown in Fig. 3B, treatment with PRL resulted in a 64% decrease in luciferase activity of the rat FAS promoter (−1,594 to +65)/luciferase construct. Although the basal level of luciferase activity for the rat FAS promoter (−700 to +65)/luciferase construct was similar to that of the −1,594 to +65 construct, PRL had no effect on the level of luciferase activity (Fig. 3B). These data demonstrate that a PRL-sensitive region exists between −1,594 and −700 of the FAS promoter and support our hypothesis that FAS is transcriptionally regulated by STAT5 activators.

Since PRL is a potent activator of STAT5, we hypothesized that the −1,594 to −700 region of the FAS promoter may contain a STAT5 binding site that conferred PRL responsiveness. Therefore, we examined the rat FAS promoter (GenBank X62889) for the presence of STAT consensus sites (TTCNNNGAA). An examination of the FAS promoter did not result in the identification of any sequences that precisely matched the STAT consensus site. However, as shown in Table 1, we identified four regions that were similar to the consensus sequence. To evaluate these potential STAT5 binding sites, we performed a series of electromobility shift assays (EMSA). For these experiments, nuclear and cytosolic extracts were prepared from 3T3-L1 adipocytes acutely treated with PRL for 15 min. As shown in Fig. 4A, PRL did not induce the binding of nuclear protein complexes to the −951 to −933, −1,226 to −1,214, or −4,639 to −4,623 regions of the FAS promoter. The induction of binding by a PRL-activated protein complex to the rat β-casein STAT5 binding site (1) is included as a positive control. However, we did observe PRL-dependent binding by a nuclear protein complex to the −908 to −893 region of the FAS promoter (Fig. 4B). To determine the specificity of binding, an EMSA was performed using a mutant version of the −908 to −893 oligonucleotide, in which two nucleotides were changed (Table 1). As shown in Fig. 4B, there was no detectable binding to the mutant form of the binding site following PRL stimulation. In addition, binding of the PRL-activated protein complex was successively competed with increasing concentrations of the unlabeled −908 to −893 oligonucleotide (Fig. 4C). Further evidence of specificity is shown in Fig. 4D, in which an excess of the −908 to −893 oligonucleotide competed away binding induced by PRL treatment (lane 4), whereas the mutant form of the oligonucleotide did not (lane 5). Although we did not detect binding to the −951 to −933 site, an excess of this oligonucleotide moderately diminished binding to the radiolabeled −908 to −893 probe (lane 6). Yet, there was no appreciable competition of binding by the −1,226 to −1,214 oligonucleotide (lane 7). As anticipated, binding was competed away with the STAT5 binding site of the rat β-casein promoter. Interestingly, a STAT3 binding site, −168 to −148 from the rat α₂-macroglobulin promoter (33), also resulted in binding competition. However, binding was not competed with a STAT1 binding site that is present at −221 to −207 of the peroxisome proliferator–activated receptor γ2 promoter (34).

To determine whether the protein complex binding the −908 to −893 contained STAT proteins, we performed supershift analysis using antibodies directed against the STAT proteins expressed in adipocytes. As shown in Fig. 5, the protein complex induced by PRL was fully supershifted with a STAT5A antibody (lane 5) and weakly supershifted with an STAT5B antibody (lane 6). STAT1 and STAT3 antibodies had no effect on the mobility of the complex (lanes 3 and 4). We also investigated the specificity of binding by STAT proteins by comparing the binding induced by LIF (0.5 nmol/l), a cytokine that activates STAT1 and STAT3 but not STAT5 in 3T3-L1 adipocytes (35). LIF stimulated binding to the −908 to −893 site by a protein complex that exhibited highly reduced binding affinity and slightly faster mobility than the complex induced by PRL (Fig. 5). Moreover, the LIF-induced complex was supershifted by a STAT1 antibody (lane 8) but not with antibodies for STAT3 or STAT5A (lanes 9 and 10). We used three other STAT3 antibodies capable of supershifting STAT3 complexes and did not observe any STAT3 LIF-stimulated binding to the −908 to −893 site (data not shown).

Our data clearly demonstrate that the −908 to −893 region of the rat FAS promoter binds nuclear PRL-activated STAT5 proteins in vitro. To determine whether this region of the FAS promoter contributed to the regulation of FAS by PRL in living cells, we performed site-specific mutagenesis to alter two basepairs at positions −902 and −901 within the rat FAS promoter (−1,594 to +65)/luciferase construct. We have shown that this mutation abolished binding of PRL-induced proteins to this site (Fig. 4B and D). Transfection of the wild-type and mutant constructs into 3T3-L1 cells revealed that the basal level of luciferase activity was unaffected by mutation of the −902 and −901 bp of the FAS promoter (Fig. 6). However, the 60% decrease in luciferase activity induced by PRL for the wild-type construct was eliminated with the mutation. Thus, these data clearly indicate that the −908 to −893 site of the FAS promoter is sensitive to PRL and suggest that this site confers the negative regulation of FAS by PRL-activated STAT5A protein complexes.

The novel findings in this study include data demonstrating that FAS levels are decreased following stimulation with activators of STAT5 in 3T3-L1 adipocytes, the identification of a PRL-responsive region of the rat FAS promoter, and the characterization of a STAT5 binding site in this region. These results strongly suggest that STAT5A directly represses the expression of FAS in adipocytes. Moreover, our data indicate that STAT5A has an antilipogenic function in murine adipocytes, in addition to an adipogenic role previously described by our laboratory (8) and others (4,6,7).

Previous studies have shown that GH, an activator of STAT5, attenuated the induction of FAS in adipocytes by insulin in 3T3-F442A adipocytes (10) and decreased expression of FAS in vivo in pigs (11). We observed that two activators of STAT5 in adipocytes, GH and PRL, decreased protein and mRNA expression of FAS in 3T3-L1 adipocytes. These results are consistent with the findings that GH in cows (36) and PRL in rats (24) can inhibit the activity of FAS in adipose tissue. Surprisingly, our results revealed that the effects of GH on FAS are transient and reversible, whereas the effects of PRL are more pronounced with time. The differences in the action of these two STAT5 activators may be due to receptor levels or specific signal proteins that have yet to be identified. Alternatively, the differences we observe in GH and PRL effects on FAS in our murine cells may also be affected by different species of hormones we used (sheep and pig) on murine cells. Our results show that PRL did not affect the protein turnover of FAS but indicate that changes in FAS levels are mediated at the transcriptional level. Our experiments are supported by other studies (25,28,37–48) that demonstrate that FAS can be modulated at the transcriptional level. However, increased turnover of FAS mRNA in 3T3-F442A adipocytes has been demonstrated as one of the means through which GH attenuates the induction of FAS by insulin (10). Nonetheless, our results strongly suggest that the PRL-induced repression of FAS is mediated by an inhibition of transcription.

To elucidate the mechanism of transcriptional regulation by PRL, we investigated the regulation of the FAS promoter. Although a previous study (28) indicated that the −112 to +65 region of the rat FAS promoter was sensitive to regulation by GH in murine cells, we did not observe PRL-mediated regulation of the rat FAS promoter within this region in murine cells (Fig. 3A). However, our analysis of larger regions of the rat FAS promoter clearly indicated that a PRL-responsive region existed between −1,594 and −700 of the rat FAS promoter (Fig. 3B). Yin, Clarke, and Etherton (28) have previously shown that GH attenuated the stimulation by insulin of the rat FAS promoter (−112 to +65)/luciferase construct, a region of the promoter that does not contain a STAT consensus site. Furthermore, in another study (37), it was demonstrated that staurosporine, an inhibitor of JAK/STAT signaling, did not block the effect of GH on insulin-stimulated FAS expression in murine cells. Thus, it is unlikely that STAT5 proteins mediate the inhibitory effects of GH on insulin regulation of FAS. Yet, in light of our current findings, we postulate that the repression of basal levels of FAS by PRL is directly regulated by STAT5 proteins via binding to a site within the −1,594 to −700 region of the rat FAS promoter. We also hypothesize that the delayed effects we observe on FAS mRNA and protein are due to the stability of the FAS mRNA and protein. Our studies indicate that FAS protein is very stable (Fig. 2), and preliminary studies suggest that the FAS mRNA is more stabile than other adipocyte mRNAs (data not shown).

Since PRL is a potent activator of STAT5 (Fig. 1B), we examined the rat FAS promoter for sites that resembled the STAT consensus sequence, TTCNNNGAA. We identified four sites that were similar to the STAT consensus, but our analysis by EMSA indicated that only one site, at position −908 to −893, was bound by a PRL-induced nuclear protein complex in a highly specific manner (Fig. 4). Similar results were obtained with GH (data not shown). In our experiments, the PRL-induced protein complex was fully supershifted by a STAT5A antibody (Fig. 5). The functional significance of the −908 to −893 site was determined by mutating two nucleotides at positions −902 and −901, within the STAT5 binding site of the rat FAS promoter (−1,594 to +65)/luciferase construct. Mutation of this site completely abrogated the downregulation by PRL that was observed with the wild-type construct (Fig. 6). Taken together, these data strongly suggest that −908 to −893 of the rat FAS promoter is a STAT5 binding site, which confers the negative transcriptional regulation of FAS by PRL. These results support our hypothesis that STAT5 directly represses expression of FAS in adipocytes.

The association of increased transcription of FAS in rodent obesity (49) and the inhibition of adipogenesis by an allosteric inhibitor of FAS (26) are highly indicative that regulation of FAS expression and activity in adipocytes is an important control of energy homeostasis. The characterization of a STAT5 binding site in the rat FAS promoter identifies a novel mechanism for repression of FAS expression. The −908 to −893 site of the rat FAS promoter is also present in the murine FAS promoter (AL663090) at position −895 to −887. Interestingly, the human FAS promoter (AF250144) contains a site at −1,091 to −1,083 (TTCGAGGAA) that is different from the sequence identified from the rodent promoters but complements the STAT consensus sequence TTCNNNGAA. Hence, although the precise sequence of the STAT5 binding site is not conserved across species, the binding by STAT5 to these sites in the FAS promoter may be an evolutionarily retained mechanism of regulating FAS expression in adipose tissue.

Our observation that STAT5A, not other adipocyte STATs, preferentially binds to the −908 to −893 site (Fig. 5) is consistent with previous studies (50,51) demonstrating that STAT proteins bind similar sequences but exhibit subtle differences in affinity for nucleotides between and beyond the half-sites of the palindrome. This specificity in STAT binding may account for the distinct repertoire of target genes regulated by each STAT protein (51). Our data suggest that the regulation of transcription mediated by the −908 to −893 region of the rat FAS promoter is primarily regulated by STAT5A binding.

PRL has been shown to have antilipogenic effects in adipose tissue through inhibition of LPL expression and the repression of FAS (24) and LPL (18) activity. At this time, our studies and others suggest that both PRL and GH repress FAS expression. However, these growth factors appear to have different effects (transient versus sustained) on FAS regulation. These differences may be attributable to expression levels or actions of specific GH and PRL signaling proteins and/or specific cell types. Nonetheless, our results strongly suggest that PRL modulates the expression of FAS in adipocytes through STAT5A and supports a role for STAT5 as a regulator of energy balance in adipocytes. PRL appears to have dual functions, positively and negatively affecting adipocyte gene expression. This hypothesis is supported by another study (13) showing that STAT5 can act as a modulator of GH-induced inhibition of aP2 expression. The association of STAT5 proteins with coactivators, corepressors, and other transcription factors likely affects the ability of STAT5A to have adipogenic and antiadipogenic effects. Recent work from our laboratory has demonstrated that the association between STAT5A and the glucocorticoid receptor is highly regulated during fat cell differentiation (8). Hence, cooperation between STAT5A and glucocorticoid receptor may occur in the modulation of FAS, since glucocorticoids have been demonstrated to affect FAS transcription and activity in adipose tissue (52,53). We are currently investigating the cross talk between these pathways in adipocytes. In summary, we have observed that PRL represses expression of FAS in adipocytes and negatively regulates the rat FAS promoter. Our identification of a STAT5 binding site in the promoter of FAS characterizes a novel mechanism of regulating FAS expression. In summary, we hypothesize that the regulation of FAS by STAT5 is likely an important contribution to the maintenance of energy homeostasis.

This work was supported by grant R01DK52968-05 from the National Institutes of Health to J.M.S.

We thank James E. Baugh and Patricia Arbour-Reily for technical assistance with this project.