The Forkhead Transcription Factor FoxC2 Inhibits White Adipocyte Differentiation*

In this study, we show that expression of FoxC2 blocks the capacity of 3T3-L1 preadipocytes to undergo adipogenesis in the presence of dexamethasone, isobutylmethylxanthine, and insulin. This block is characterized by an extensive decrease in the expression of proteins associated with the function of the mature fat cell, most notably C/EBPα, adiponectin, perilipin, and the adipose-specific fatty acid-binding protein, FABP4/aP2. Since the expression of these proteins lies downstream of PPARγ, we overexpressed PPARγ in Swiss mouse fibroblasts to promote adipocyte differentiation. We show that FoxC2 blocks the ability of PPARγ to induce adipogenic gene expression in response to exposure of the cells to dexamethasone, isobutylmethylxanthine, insulin, and a PPARγ ligand. Interestingly, the expression of aP2 escapes the inhibitory action of FoxC2 under conditions that promote maximum PPARγ activity. In contrast, FoxC2 inhibits the expression of C/EBPα, perilipin, and adiponectin even in the presence of potent PPARγ ligands. Finally, we show that FoxC2 does not affect the ability of PPARγ to bind to or transactivate from a PPARγ response element. These data suggest that FoxC2 blocks adipogenesis by inhibiting the capacity of PPARγ to promote the expression of a subset of adipogenic genes.

Adipose tissue, once thought to be just a storage depot for excess energy supply, is now realized to be a fully functioning endocrine organ, secreting numerous factors that help to maintain normal whole body metabolism (1,2). Significant changes in adipose mass or perturbation in adipocyte signaling can result in the disruption of overall metabolic homeostasis (3,4). Changes in adipose mass can result from increased lipid storage in preexisting adipocytes as well as an increase in adipocyte number through the proliferation and differentiation of preadipocytes (5). The differentiation of preadipocytes into mature insulin-responsive adipocytes results from coordinated signaling cascades descending on many transcription factors that promote the expression of terminal adipogenic genes (4,6). Much of what is known regarding the transcriptional events of adipogenesis has been gained through the use of cell lines predisposed to the adipocyte lineage. The most frequently used are the 3T3-L1 and 3T3-F442A cell lines (7)(8)(9). Signaling events that induce adipogenesis promote both the increased expression of some factors as well as the down-regulation of others. Early events in adipogenesis include up-regulation of C/EBP␤ 1 and C/EBP␦, which in turn activate transcription of PPAR␥ and C/EBP␣ (4, 10 -13). These two factors work in a cooperative fashion to promote the expression of terminally adipogenic genes. In fact, the expression of both PPAR␥ and C/EBP␣ is necessary and sufficient for complete adipogenesis to occur (14 -17). In addition, the inducers of adipocyte differentiation promote the down-regulation of factors that maintain the undifferentiated state of preadipocytes. Members of both the GATA family as well as factors in the Wnt signaling pathway act as negative regulators of adipogenesis, and their maintained expression during differentiation blocks terminal adipogenesis (18 -21). More recently, members of the forkhead (FOX) family of transcription factors have been identified as playing both positive and negative roles in adipocyte differentiation.
Forkhead transcription factors have been shown previously to play varying roles during early development by regulating growth, differentiation, and apoptosis of different cell types (22)(23)(24). They bind as monomers to the consensus sequence 5Ј-TGACCTTTGACCT-3Ј, and selective binding is dictated by flanking sequences as well as the relative abundance of individual members. The affinity of FOX proteins to bind to sequences overlaps to a great extent between members, and in addition, some FOX proteins primarily act as transcriptional activators while others act as transcriptional repressors (24). Recently, it has been shown that specific FOX family members, FoxO1, FoxA2, and FoxC2, play a role in maintaining normal metabolic homeostasis in the adult organism. FoxO1 expression increases during adipogenesis, and its transcriptional activity is negatively regulated by insulin signaling through Akt phosphorylation during the early phase of adipogenesis. Overexpression of a constitutively active FoxO1 blocks adipocyte differentiation as a result of induced expression of the tumor suppressor, p21 CIP , and the concurrent block in clonal expansion (25,26). Expression of FoxA2 also inhibits adipocyte differentiation, and it is induced in an obese diabetic mouse, proposing a role for negative feedback against adipocyte differentiation. FoxA2 induces the expression of pref-1, a known inhibitor of adipogenesis (27), as well as genes that function in glucose and lipid metabolism (28). FoxC2 has been shown to have an effect on the formation of white and brown adipose tissue in rodents. Its expression in mice is restricted to adipose tissue, and overexpression of FoxC2 targeted to adipose tissue results in attenuation of white adipose tissue formation while brown adipose development is grossly enhanced (29). Data from those mice indicate an increase in insulin sensitivity and more efficient signaling though ␤-adrenergic receptors. Furthermore, the expression of FoxC2 is up-regulated in mature 3T3-F442A adipocytes upon exposure to insulin and tumor necrosis factor ␣ (30). The dramatic decrease in the white adipose depots in FoxC2 transgenic mice suggested that FoxC2 might act to suppress white adipogenesis. Consequently, our goal in this study is to determine whether FoxC2 blocks the differentiation of white preadipocytes and, if so, to begin to identify the molecular mechanisms regulating such an inhibitory action. We show that inducible overexpression of FoxC2 in 3T3-L1 preadipocytes inhibits adipogenesis and that, more specifically, it blocks the expression of several known PPAR␥ downstream targets that function in the mature adipocyte. These observations led us to create a cell line whose capacity to undergo adipogenesis is directly dependent on the expression and activity of PPAR␥. To this end, we employed Swiss mouse fibroblasts, which constitutively express PPAR␥. In these cells, we inducibly overproduced FoxC2 and found that its expression results in a significant attenuation in the capacity of these fibroblasts to undergo adipogenesis. Furthermore, the degree of attenuation appears to be selective for individual PPAR␥ downstream targets. Finally, we show that FoxC2 does not alter the ability of PPAR␥ to bind or transactivate a PPAR␥ response element, indicating that FoxC2 functions at a location downstream of PPAR␥.

EXPERIMENTAL PROCEDURES
Plasmids and Cell Lines-Stocks of 3T3-L1 cells expressing the Tet-Off regulator protein (3T3L1-TET) were a kind gift of Jacques Pairault (UMR7079 CNRS UPMC, Paris, France). The FoxC2-pLNCX plasmid (mfh1) was a kind gift of Dr. Brigid Hogan (Vanderbilt University, Nashville, TN). FoxC2 cDNA was amplified out of FoxC2-pLNCX by PCR and then cloned into His-Topo-pCDNA3-V5 (Invitrogen). The FoxC2 cDNA now tagged with the V5 epitope sequence was then cloned into pBI-G (Clontech, Palo Alto, CA) using blunt end ligation. The resulting FoxC2-V5-pBI-G plasmid, along with a puromycin selection plasmid (pBABE-puro), was transfected into 3T3-L1-TET cells using FuGENE 6 (Roche Applied Science). Colonies of cells resistant to 2 g/ml puromycin were selected and analyzed for expression of the pBI-G vector on the basis of tetracycline-responsive ␤-galactosidase production. The initial selection gave rise to several non-homogeneous colonies with only 10 -20% of the cells expressing ␤-galactosidase activity. One of these colonies was subjected to serial dilution single-cell cloning. Colonies were selected on the basis that nearly the entire population of cells expressed ␤-galactosidase activity in a tetracyclineresponsive manner. The pBABE-PPAR␥ plasmid was a kind gift of Dr. Bruce Spiegelman (Dana Farber Institute, Harvard Medical School, Boston, MA). The pBABE-PPAR␥ plasmid was transfected into Swiss mouse 3T3 fibroblasts constitutively expressing the Tet-Off activator protein (Clontech). Primary colonies were selected using 100 g/l hygromycin. This initial selection gave rise to several non-homologous colonies, one of which was subjected to serial dilution. Colonies were selected based on their ability to undergo adipogenesis when exposed to the standard induction mixture along with 5 M troglitazone. Several colonies were selected and analyzed. A single colony, termed Swiss-P␥, was chosen on the basis that it was representative of the other populations and was subjected to further manipulation. The FoxC2-V5-pBI-G plasmid, along with pBABE-puro, was transfected into Swiss-P␥ cells using FuGENE 6. Colonies of cells resistant to 2 g/ml puromycin were selected and analyzed for the expression of the pBI-G vector on the basis of tetracycline-responsive ␤-galactosidase production. The initial selection gave rise to several non-homogeneous colonies with only 10 -20% of the cells expressing ␤-galactosidase activity. One of these colonies was subjected to serial dilution single-cell cloning. Colonies were selected on the basis that nearly the entire population of cells expressed ␤-galactosidase activity in a tetracycline-responsive manner.
Cell Culture-Stocks of 3T3-L1 preadipocytes (7) and 3T3L1-FoxC2 cells were maintained in DMEM containing 25 mM D-glucose and supplemented with 10% donor calf serum. For experiments, cells were grown in DMEM containing 10% fetal calf serum. Induction of differentiation was achieved by treatment with the standard induction mixture (DMI) containing dexamethasone (1 mM), 3-isobutyl-1-methylxan-thine (0.5 mM), and insulin (1.67 M) on 2-day post-confluent cells. Cells were maintained in this differentiation mixture for 48 h after which time the media was changed to DMEM with 10% FBS and insulin (835 nM). 48 h prior to induction, 3T3-L1-FoxC2 cells were exposed to 3 mM sodium butyrate (Sigma) for 24 h to enhance ectopic FoxC2 expression. The Swiss-P␥-FoxC2 cells were differentiated using the previously mentioned mixture of inducers along with 5 M troglitazone (Roche-Davis/Warner Lambert, Ann Arbor, MI). Additional ligands were also used at 1 M of a specific PPAR␥ agonist, GW7845 (GlaxoSmithKline), and 1 M rosiglitazone (Cayman Chemicals, Ann Arbor, MI). The PPAR␥ antagonist T0070907 was obtained from AdipoGenix, Inc. (Boston, MA) and used at 10 M as described previously (31). Cyclohexamide (Sigma) and MG132 (Calbiochem) were used at 5 g/ml and 12.5 M, respectively.
Oil Red O Staining-Staining was performed following the procedure described previously (32). The cells then were photographed using phase-contrast microscopy.
Gel Electrophoresis and Immunoblotting-Proteins were analyzed by SDS-12% polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (PerkinElmer Life Sciences) in 25 mM Tris, 192 mM glycine, and 10% methanol. Following transfer, membranes were blocked with 5% nonfat dry milk in phosphate-buffered saline-0.1% Tween 20 and probed with the antibodies specified below. Horseradish peroxidase-conjugated secondary antibodies (Sigma) and ECL substrate kit (PerkinElmer Life Sciences) were used for detection of specific proteins.
Antibodies-The perilipin antibody was a kind gift of Dr. Andrew Greenberg (Tufts University, Boston, MA). Anti-aP2 serum was kindly provided by Dr. David Bernlohr (University of Minnesota). Antibodies against PPAR␥, C/EBP␣, and C/EBP␤ were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). Anti-V5 antibody was purchased from Invitrogen, and the adiponectin antibody was purchased from Affinity BioReagents (Golden, CO).
Analysis of RNA-Total RNA was extracted with TRIzol (Invitrogen) according to manufacturer's instructions. Following quantification, messenger RNA was assayed in equivalent amounts of total RNA by reverse transcriptase-PCR (RT-PCR) as described previously (33). Primer sequences used for amplification were synthesized by Integrated DNA Technologies, Inc., based on the published sequences in the GenBank TM (sequences are available upon request). All of the PCR reactions were performed in the linear range of cycle number for each set of primers, and the corresponding products were analyzed by 1.5% agarose gel electrophoresis.
Reporter Assays-Swiss-P␥-FoxC2 cells were plated and grown to ϳ50% confluency at which time they were changed to media containing the appropriate tetracycline condition. Two days later, at ϳ85% confluency, the cells were transfected with 4 g of DR-1-LUC (a gift from Dr. Jackie Stephens, Louisiana State University) and 0.04 g of pRL-CMV (Promega, Madison, WI) using FuGENE 6. 24 h post-transfection, the cells were split into the standard induction mixture of DMI and their appropriate ligand and tetracycline condition. 48 h later, the medium was changed to the established tetracycline and ligand condition. The cells were harvested the following day and analyzed for luciferase activity and protein expression. Luciferase activity was measured using the DLRII kit (Promega, Madison, WI).

Overexpression of FoxC2 Blocks Adipogenesis in 3T3-L1
Preadipocytes-Overexpression of FoxC2 in adipose tissue results in decreased white adipose mass as shown by Cederberg et al. (29). In addition, the exposure of 3T3-F422A preadipocytes to insulin or tumor necrosis factor ␣ has been shown to both induce FoxC2 expression and inhibit adipocyte differentiation (30). To better understand the molecular mechanisms by which FoxC2 inhibits white adipose tissue formation, we employed the 3T3-L1 preadipocyte cell line (7). Using the tetracycline-inducible expression system, we conditionally overexpressed FoxC2 tagged with a V5 epitope in 3T3-L1 preadipocytes expressing the Tet-Off regulator protein. Multiple clones were isolated and subjected to preliminary experiments. Of those clones, a single clone representative of the pool was chosen on the basis of regulated FoxC2 expression and its ability to differentiate into mature adipocytes. To monitor the expression of both the endogenous and ectopic FoxC2 mRNAs in this cell line, we adopted a RT-PCR screen using different primers to select for the respective mRNAs. The total and ectopic FoxC2 mRNAs were assayed using a common upstream primer and downstream primers targeted against an internal FoxC2 sequence and the V5 epitope sequence, respectively. Other groups have shown that endogenous FoxC2 expression is significantly up-regulated in differentiated 3T3-F442A cells and to a lower extent in differentiated 3T3-L1 cells when they are exposed to insulin for 2 h. Fig. 1a demonstrates that the culture of 3T3-L1-FoxC2 cells in the absence of tetracycline induces the expression of ectopic mRNA to levels expressed by the endogenous gene when cells are exposed to insulin. These data indicate that the inducible expression system is producing physiological levels of FoxC2 in 3T3-L1 preadipocytes. Fig. 1b shows that the expression of the ectopic mRNA results in the production of a V5-tagged FoxC2 protein only in cells cultured in the absence of tetracycline.
To determine how FoxC2 expression affects adipogenesis, 3T3-L1-FoxC2 cells were induced to differentiate in the presence or absence of tetracycline for 4 days using the standard differentiation mixture of dexamethasone (DEX), isobutylmethylxanthine (MIX), and insulin. Samples were analyzed by Western blot analysis for various transcription factors and markers of terminal adipogenesis (Fig. 2). In the absence of FoxC2, both PPAR␥ and C/EBP␣ are induced by day 2 and abundantly expressed by day 3, leading to increased expression of perilipin, adiponectin, and aP2 by day 4. Culture in the absence of tetracycline results in abundant expression of the V5-tagged FoxC2 protein, which leads to the inhibition of the adipogenic proteins including PPAR␥ and C/EBP␣.
FoxC2 Inhibits Adipogenesis at a Step Downstream of PPAR␥2-It is generally accepted that terminal differentiation of 3T3-L1 preadipocytes depends to a great extent on the activity of PPAR␥. In fact, the expression of PPAR␥ itself requires its continual activation through a positive feedback loop involving C/EBP␣ (34). Consequently, it is possible that the effects of FoxC2 on adipogenic gene expression observed in Fig. 2 may result from an inhibition of PPAR␥ activity. To better understand the mechanisms by which FoxC2 inhibits differentiation of 3T3-L1 preadipocytes, we employed a system in which adipogenesis in Swiss mouse fibroblasts is driven by ectopic PPAR␥ expression and activity. In this cell line, we conditionally overexpressed FoxC2 using the Tet-Off system, giving rise to a cell line termed Swiss-P␥-FoxC2. RT-PCR analysis of RNA isolated from these cells cultured in the presence or absence of tetracycline demonstrates that the expression of FoxC2 is tightly regulated by tetracycline (Fig. 3a). Furthermore, the ectopic FoxC2 tagged with a V5 epitope can bind to a FOX response element and this complex can be supershifted with the addition of a V5 antibody (Fig. 3b). Swiss-P␥-FoxC2 cells were induced to differentiate by exposure to DEX, MIX, and preadipocytes. a, post-confluent 3T3-L1 cells were induced to differentiate as noted under "Experimental Procedures" for 10 days. At day 6, the medium was changed to 10% fetal bovine serum for the remainder of the time course. At day 10, cells were exposed to 1.67 M insulin for 2 h after which time they were harvested for total RNA and subjected to RT-PCR analysis for both total and ectopic FoxC2 as well as glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control. 3T3-L1-FoxC2 cells were cultured in the presence or absence of tetracycline for 2 days and 3 mM sodium butyrate for the first 24 h. The cells were harvested for total RNA and subjected to RT-PCR analysis as above. b, 3T3-L1-FoxC2 cells were cultured in the presence or absence of tetracycline for 2 days after which time they were harvested. Extracts were subjected to Western blot analysis to measure ectopic FoxC2 expression using an anti-V5 antibody.
FIG. 2. Early expression of ectopic FoxC2 blocks adipogenesis in 3T3-L1 cells. 3T3-L1-FoxC2 cells were maintained in DMEM plus 10% calf serum until confluent, at which time (day Ϫ2) they were cultured in DMEM plus 10% calf serum in the presence or absence of tetracycline. In addition, at day Ϫ2, cells were exposed to 3 mM sodium butyrate for 24 h. Two days post-confluence (day 0), the cells were induced to differentiate by exposure to the standard induction mixture and their established tetracycline condition. Cells were harvested for total protein every day for 4 days, and extracts were subjected to Western blot analysis.
insulin for 4 days in the presence or absence of tetracycline, and total cell proteins were harvested for Western blot analysis. In the absence of FoxC2, PPAR␥ promotes the expression of several markers of adipogenesis including perilipin, adiponectin, and aP2 (Fig. 3c). Under these culture conditions (i.e. exposure to DEX, MIX, and insulin), ϳ20% of the cells become round and accumulate lipid droplets (Fig. 4a, ϪTROG, ϩTET). Ectopic expression of FoxC2 blocked this low level of morphological differentiation (Fig. 4a, ϪTROG, ϪTET) and also blocked the expression of adipogenic mRNAs (Fig. 3c). The small percentage of lipid laden cells following exposure to DEX, MIX, and insulin suggests that the activity of PPAR␥ is below optimal levels when compared with the level of differentiation in 3T3-L1 cells. To assess whether FoxC2 is capable of inhibiting adipogenesis when PPAR␥ is fully active, we exposed the cells to a potent PPAR␥ ligand in addition to the standard adipogenic mixture. In fact, in the absence of FoxC2, differentiation of the cells in the presence of 5 M troglitazone (PPAR␥ ligand) results in greater than 90% of the population accumulating lipid droplets (Fig. 4a,ϩTROG,ϩTET) compared with only 20% of the population in the absence of the exogenous PPAR␥ ligand (Fig. 4a, ϪTROG, ϩTET). More importantly, FoxC2 is capable of inhibiting lipid accumulation under conditions (ϩ troglitazone) promoting optimal PPAR␥ activity ( Fig.  4a, ϩTROG, ϪTET). To analyze the relationship between PPAR␥ activity and the inhibitory effect of FoxC2 on adipogenic gene expression, cells were stimulated to differentiate in the presence or absence of tetracycline with increasing concentrations of troglitazone for 6 days and proteins were analyzed by Western blot (Fig. 4b). Fig. 4b demonstrates that, in the absence of FoxC2 (ϩ tetracycline), there is a troglitazone dosedependent increase in the expression of perilipin, adiponectin, C/EBP␣, and aP2 and a corresponding decrease in the abundance of the PPAR␥2 protein. In contrast, the ectopic expression of FoxC2 significantly attenuates the expression of the adipogenic genes at all concentrations of troglitazone tested. Interestingly, higher doses of troglitazone appear to partially overcome the inhibitory effect of FoxC2 on aP2 expression (Fig.  4b, lanes 9 and 11). Previous studies have shown that the proteasomal-mediated degradation of PPAR␥2 correlates with PPAR␥ activity (35,36). However, even in the presence of FoxC2, troglitazone is still capable of facilitating the downregulation of PPAR␥2 protein, suggesting that FoxC2 is not blocking this process. However, it is important to point out

FIG. 3. FoxC2 blocks the expression of adipogenic genes in mouse fibroblasts expressing an ectopic PPAR␥2.
a, Swiss-P␥-FoxC2 cells were cultured in the presence or absence of tetracycline for 2 days and harvested for total RNA. Samples were subjected to RT-PCR analysis using oligonucleotide primers against ectopic FoxC2 and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as a control. b, Swiss-P␥-FoxC2 cells were induced to differentiate using the standard induction mixture in the presence or absence of tetracycline. Extracts were subjected to gel shift analysis using a FOXRE consensus oligonucleotide. Prior to incubation with the radiolabeled oligonucleotide, the ϪTET sample was incubated with an anti-V5 antibody shown by the supershift in the far right lane. An aliquot of the ϪTET sample was also incubated with a 100-fold excess of the unlabeled oligonucleotide prior to the addition of the radiolabeled probe (cold oligo). c, cells were manipulated as in b and were harvested for total protein. Extracts were subjected to Western blot analysis.
that, in the absence or at low doses of troglitazone, FoxC2 appears to increase the abundance of PPAR␥2 (Figs. 3b and 4b,  lanes 5 and 6), which is consistent with the notion that at suboptimal PPAR␥2 activity (low doses of ligand) FoxC2 can directly inhibit PPAR␥2 activity resulting in its extended lifespan. Furthermore, at of these same low doses of troglitazone, FoxC2 significantly attenuates aP2 expression. To determine whether PPAR␥ ligands other than troglitazone are capable of counteracting the inhibitory effect of FoxC2, Swiss-P␥-FoxC2 cells were induced to differentiate in the presence of GW347845, rosiglitazone, or troglitazone (Fig. 5a). The inhibitory effect of FoxC2 is evident in the presence of all three ligands as shown by a FoxC2-dependent inhibition of perilipin, C/EBP␣, and adiponectin. Again, aP2 seems to escape this inhibitory action of FoxC2. A plausible explanation for this apparent absence of effect on aP2 expression is that the aP2 protein may be significantly more stable than the other markers of terminal adipogenesis. To determine whether FoxC2 is FIG. 6. FoxC2 blocks adipogenesis in Swiss-P␥ cells during an early phase of the differentiation process. a and b, Swiss-P␥-FoxC2 cells were differentiated using the standard induction mixture in the presence of troglitazone with or without tetracycline for 6 days. Cells were harvested at the indicated times for total protein and subjected to Western blot analysis.

FIG. 5. FoxC2 blocks expression of adipogenic genes at the level of transcription even in the presence of potent PPAR␥ ligands. a, Swiss-P␥-FoxC2
cells were differentiated using the standard induction mixture in the presence of different PPAR␥ ligands as shown and in the presence or absence of tetracycline for 5 days. Cells were harvested for total protein and subjected to Western blot analysis. b, Swiss-P␥-FoxC2 cells were differentiated using the standard induction mixture supplemented with 5 M troglitazone in the presence or absence of tetracycline for 5 days at which time samples were harvested for total RNA and analyzed by RT-PCR analysis. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.
inhibiting transcription of the aP2 gene as well as the other genes, we measured the abundance of the corresponding mRNAs by RT-PCR analysis. Fig. 5b demonstrates that, whereas perilipin, GLUT4, C/EBP␣, and adiponectin gene expression are all blocked by the FoxC2, aP2 gene expression is completely unaffected. The analyses presented in Fig. 5 were performed in differentiated cells; therefore, to gain a better understanding of the sequence of events that lead to the inhibitory action of FoxC2, we analyzed protein expression in Swiss-P␥-FoxC2 cells during the course of the differentiation process. As expected from previous data, FoxC2 blocks expression of the adipogenic markers, C/EBP␣, perilipin, and adiponectin; however, it has little effect on aP2. Driven by the overexpression and activity of PPAR␥, these adipogenic genes are all expressed and attenuated by FoxC2 by 2-3 days post-induction (Fig. 6a).
However, it appears that the PPAR␥-dependent induction of C/EBP␣ expression during the initial 1-2 days of differentiation is not as sensitive to the inhibitory action of FoxC2 as the other proteins. Consequently, we performed a more detailed time course to assess whether FoxC2 is acting downstream of C/EBP␣ rather than acting directly on PPAR␥. The data presented in Fig. 6b demonstrate that, in the absence of FoxC2, perilipin expression is induced as early as 8 -16 h and that this event is blocked by overexpression of FoxC2. In contrast, the induction of C/EBP␣ during this same 8 -16-h time period is completely resistant to the action of FoxC2 (Fig. 6b). In fact, the down-regulation of C/EBP␣ expression by FoxC2 does not become apparent until 44 -48 h post-induction. Adiponectin expression is induced several hours later than perilipin and C/EBP␣ (40 -44 h) and only under conditions where FoxC2 expression is repressed. The expression of aP2 occurs to a limited extent prior to treatment with troglitazone (0 h), but it is enhanced many fold in response to the PPAR␥ ligand and is unaffected by the presence of FoxC2. We conclude from these data that FoxC2 inhibits the capacity of PPAR␥ to promote the expression of select downstream target genes, which might result from a direct effect on PPAR␥ as well as inhibition of other factors. In this regard, we have recently demonstrated that C/EBP␣ is required for the optimum expression of adiponectin (37); thus, the FoxC2-associated down-regulation of adiponectin might be due to the block in C/EBP␣ expression.
Of additional interest is the observation that FoxC2 appears to enhance the abundance of the ectopically expressed PPAR␥. Other studies have shown that ligand activation of PPAR␥ results in a destabilization of the protein and its subsequent proteasomal degradation (35,36). Therefore, it is conceivable that FoxC2 might be enhancing PPAR␥ expression by blocking its activity and consequently preventing its degradation by the proteasome. To determine whether the expression of FoxC2 was causing a stabilization of PPAR␥, a short 24-h time course was performed where Swiss-P␥-FoxC2 cells were differentiated using the standard protocol in the presence or absence of tetracycline (Fig. 7). 8 h post-stimulation, the cells were treated with cyclohexamide to inhibit protein synthesis and control samples were also treated with the proteasomal inhibitor MG132. Western blot analysis shows that, when cells are cultured in the absence of FoxC2, the turnover of PPAR␥ over a 16-h time period is signif-icantly faster than when cells express FoxC2. Additionally, inhibiting the proteasome with MG132 arrests the decay of PPAR␥ in the presence or absence of FoxC2. These data are consistent with the notion that FoxC2 may be inhibiting the capacity of PPAR␥ to promote transcription of adipogenic genes by decreasing its transcriptional activity. FoxC2 Does Not Interfere with the Ability of PPAR␥ to Bind to or Transactivate from a PPAR␥ Response Element-These variable changes in adipogenic gene expression and PPAR␥ stability led us to further investigate what effect FoxC2 has on the ability of PPAR␥ to bind and transactivate from a PPRE. Gel shift analysis of differentiating Swiss-P␥-FoxC2 cells show that the expression of FoxC2 does not alter the ability of PPAR␥ to bind to a consensus PPRE (Fig. 8). In addition, the PPAR␥⅐PPRE complex cannot be supershifted with an anti-V5 antibody, suggesting that FoxC2 does not associate directly with PPAR␥ within its DNA binding complex. To assay the transcriptional activity of PPAR␥, we employed a luciferase reporter gene driven by a PPRE regulated promoter (DR1-LUC). Swiss-P␥-FoxC2 cells maintained in the presence or absence of tetracycline to control FoxC2 expression were transfected with the DR1-LUC construct along with a control Renilla plasmid to account for variations in transfection efficiency. 24 h post-transfection, cells were split into conditions of the standard induction mixture (DEX, MIX, and insulin) and increasing doses of troglitazone. In addition, one condition was supplemented with a specific PPAR␥ antagonist. Cells were maintained in this condition for 2 days after which time the induction mixture was removed and cells were cultured in their established condition of TET, troglitazone, and PPAR␥ antagonist. Samples were harvested and subjected to luciferase and Western blot analysis (Fig. 9). As expected, the expression of the luciferase reporter gene increases in response to increasing doses of troglitazone (Fig. 9a). Interestingly, FoxC2 expression appears to have no inhibitory effect on DR-1 luciferase reporter gene transcription in cells where it is clearly blocking adipogenic gene expression. To confirm that this DR-1 reporter is capable of monitoring a decrease in PPAR␥ activity, we also challenged the system with a potent PPAR␥ antagonist. The exposure to T0070907 (PPAR␥ antagonist) completely blocked PPAR␥-driven adipogenic gene expression including aP2 in the presence or absence of FoxC2. In addition, this drug suppressed troglitazone-dependent transcription of the reporter gene to levels below those expressed in the absence of the PPAR␥ ligand (Fig. 9b, lanes 13 and 14). DISCUSSION Targeted overexpression of FoxC2 in adipose tissue has previously been shown to result in a significant decrease in the mass of the major white adipose depots in the mouse. An explanation for this observation is that FoxC2 blocks the differentiation of preadipocytes into mature white fat cells in vivo. The goal of the present study was to uncover potential mechanisms through which FoxC2 may block adipogenesis. To this end, we employed both the 3T3-L1 preadipocyte cell model as well as a fibroblast cell line in which adipogenesis is driven by PPAR␥. Initially, we showed that adipogenesis in 3T3-L1 preadipocytes is inhibited by FoxC2. This is most clearly demonstrated in Fig. 2, which shows that conditional ectopic expression of FoxC2 inhibits the expression of genes associated with FIG. 9. FoxC2 has no effect on the capacity of PPAR␥ to promote transcription from a PPRE under conditions where it blocks adipogenesis in Swiss-P␥ cells. a, Swiss-P␥-FoxC2 cells in either the presence or absence of tetracycline were transfected with the PPAR␥ reporter gene, DR1-LUC, and control plasmid, CMV-Renilla using FuGENE 6. Transfections were split the following day into the standard induction mixture with an increasing dose of troglitazone in the presence or absence of a PPAR␥ antagonist (T0070907) maintaining the established tetracycline condition. Two days later, the medium was changed maintaining the established troglitazone, PPAR␥ antagonist, and tetracycline condition. 24 h later, the cells were harvested and analyzed for luciferase activity. b, samples from a were harvested and analyzed by Western blot analysis. terminal differentiation. Studies during the last several years have positioned PPAR␥ as a principal regulator of terminal adipogenesis. In fact, it is likely that many signaling pathways responsible for initiating the adipogenic process converge on PPAR␥. Additionally, many transcriptional events regulating the expression of proteins responsible for establishing the mature fat cell phenotype are themselves regulated by PPAR␥. With this in mind, we questioned whether FoxC2 is inhibiting the differentiation of preadipocytes by blocking the activity of PPAR␥. Using fibroblasts that constitutively express PPAR␥2, we demonstrate that FoxC2 blocks the capacity of these cells to differentiate in the presence of an exogenous PPAR␥ ligand regardless of its potency (Figs. 3b and 4a). It is interesting, however, that the expression of aP2 escapes the inhibitory action of FoxC2 particularly under conditions that promote maximum PPAR␥ activity. In addition, the capacity of FoxC2 to inhibit the expression of other adipogenic proteins is somewhat selective. Specifically, Fig. 6b shows a rapid induction of both perilipin and C/EBP␣ during the initial 8 -16 h following activation of PPAR␥ but FoxC2 only blocks perilipin expression during this time period. The eventual inhibition of C/EBP␣ by FoxC2 occurs at 44 -48 h post-induction, suggesting that the PPAR␥-associated expression of C/EBP␣ versus perilipin is regulated by different mechanisms. It is interesting that the induction of adiponectin expression lags behind that of perilipin and C/EBP␣, since it is not detected by Western blot analysis until 40 -44 h post-induction. These data suggest that PPAR␥ alone is not a direct activator of adiponectin but requires the cooperation of another factor that itself is induced by PPAR␥. The most likely factor is C/EBP␣, because recent studies have demonstrated that the induction of adiponectin expression during terminal adipogenesis requires both PPAR␥ and C/EBP␣ (37). Consequently, the FoxC2-associated block in adiponectin expression may result from a corresponding inhibition of C/EBP␣ that occurs at approximately the same time as the induction of adiponectin expression (40 -44 h). It is also noteworthy that FoxC2 appears to result in the stabilization of PPAR␥, which could be attributed to decreased transcriptional activity (Fig. 7). These results suggest that FoxC2 may affect the transcriptional activity of PPAR␥ but only on specific promoter environments. Interestingly, FoxC2 has no effect on the capacity of PPAR␥ to bind to or transactivate from a PPRE (Fig. 8), supporting the idea that the capacity of FoxC2 to block PPAR␥-driven transcription is promoter-selective.
Mechanisms by Which FoxC2 Inhibits Adipogenesis in 3T3-L1 Preadipocytes-The data presented in Fig. 2 demonstrate that the expression of FoxC2 at the commencement of the differentiation process in 3T3-L1 preadipocytes blocks the production of both PPAR␥2 and C/EBP␣. The expression of these two transcription factors is regulated by several upstream factors, which potentially could be targets of FoxC2. The induction of adipogenesis in 3T3-L1 preadipocytes involves the exposure of a confluent, quiescent population of cells to a mixture of adipogenic hormones including DEX, MIX, and insulin. This treatment activates a burst of proliferation (referred to as clonal expansion), which occurs during the initial 2-3 days of the differentiation process. Various cell cycle-associated proteins expressed during this time appear to play a direct role in regulating the expression of PPAR␥ and C/EBP␣ as the preadipocytes exit clonal expansion to undergo terminal differentiation (4, 38 -40). It is possible that FoxC2 somehow interferes with this well coordinated transition from clonal expansion to terminal adipogenesis. In fact, Nakae et al. (25) have shown that the overexpression of a constitutively active form of another member of the forkhead family, FoxO1, inhibits the differentiation of 3T3-F442A preadipocytes by preventing the cells from undergoing clonal expansion. This event appears to result from the FoxO1-associated induction of the cyclin-dependent kinase inhibitor, p21 CIP . Other studies have shown that the expression of FoxA2 also blocks adipogenesis, but the mechanism appears to be somewhat different from that of FoxO1. In this case, FoxA2 activates the expression of pref1, a known inhibitor of differentiation. Expression of pref-1 is normally down-regulated during differentiation, but in the presence of FoxA2, its expression is sustained resulting in a block in the adipogenic process (28). Therefore, it is conceivable that FoxC2 may alter the expression of cell cycle factors, which in turn regulate expression specific terminal markers.
The data presented in Figs. 5 and 6 demonstrate that FoxC2 is capable of inhibiting the expression of proteins normally regulated by PPAR␥. It is possible that the principal target of FoxC2 is PPAR␥, even in 3T3-L1 cells in which FoxC2 inhibits PPAR␥2 expression. The activation of PPAR␥2 in 3T3-L1 cells occurs at 2 days post-induction, coinciding with the cessation of clonal expansion. This event involves an earlier expression of C/EBP␤, C/EBP␦, and PPAR␥1. Our previous studies have identified C/EBP␤ as a potent inducer of PPAR␥2, but we have also shown that the activation of PPAR␥1 can also initiate adipogenesis by mechanisms that involve establishing a positive feedback loop comprising C/EBP␣ (41). It is conceivable that FoxC2 may prevent the induction of PPAR␥2 by blocking the activity of PPAR␥1. Clearly, in mouse fibroblasts whose adipogenic potential depends exclusively on the activation of an ectopic PPAR␥2, the forced expression of FoxC2 blocks adipogenesis, indicating that PPAR␥ is a potential target of FoxC2.
Mechanisms by Which FoxC2 Inhibits Expression of Select PPAR␥2 Target Genes-The data presented in Fig. 6 show that FoxC2 inhibits the expression of proteins that are induced in response to activation of an ectopic PPAR␥2. FoxC2 may be acting directly at the level of transcriptional complexes responsible for mediating the adipogenic activity of PPAR␥2. This may involve a direct interaction with PPAR␥ as has been suggested for FoxO1 (42) or an association of FoxC2 with coactivators such as CREB-binding protein/p300. Data presented in this paper indicates that FoxC2 blocks adipogenesis by inhibiting a subset of genes that ultimately prevents the acquisition of the mature fat cell phenotype. Although the details are poorly understood, PPAR␥ promotes transcription through the assembly of a complex of coactivators and chromatin-remodeling factors. The nature of these complexes is probably unique for a given promoter, and the factors involved most probably change over the course of differentiation. The data in this study suggest that FoxC2 is disrupting one or more components of these transcriptional complexes, altering PPAR␥ transcriptional activity on select promoters that results in decreased adipogenic potential. It is also possible that FoxC2 acts directly to repress the transcription of select PPAR␥ 2 target genes that are required for the expression of the other genes associated with terminal adipogenesis. For instance, C/EBP␣ is induced in direct response to the activation of PPAR␥2 and its presence appears to be critical for the differentiation of preadipocytes into mature fat cells. Specifically, the activation of PPAR␥2 in fibroblasts lacking C/EBP␣ gives rise to immature adipocytes that do not express GLUT4 or ␤ 3 -adrenergic receptors and that express low levels of other proteins including adiponectin (34,(43)(44)(45). Furthermore, sustained expression of PPAR␥2 during terminal adipogenesis depends on the availability of C/EBP␣ (34). Consequently, decreased C/EBP␣ expression through inhibition of PPAR␥ by FoxC2 would have a significant impact on adipogenic gene expression. It is possible that FoxC2 may act directly on the promoter/enhancer of the C/EBP␣ gene, thereby antagonizing the action of PPAR␥2, which promotes C/EBP␣ expression.
The forkhead family of transcription factors is rapidly coming into light as important regulators of metabolic homeostasis. FoxC2 expression is restricted to adipose tissue in mouse and human (29) and is increased in adipose tissue of ob/ob mice (28). Previous studies have shown that targeted overexpression of FoxC2 in adipose tissue results in a significant ablation of white fat depots and hypertrophy of brown adipose depots (29). Furthermore, the adipocytes that do form in the white adipose depots have a brown-like phenotype. Although humans lack distinct brown depots, it is evident that white depots have brown adipocytes dispersed throughout the tissue (46 -48). In addition, it has also been shown that white and brown adipocytes can transdifferentiate in response to external conditions (49 -52). More recently, FoxC2 has been shown to be down-regulated in insulinresistant human subjects that coincided with the downregulation of brown adipogenic genes (53). Although we could not find any evidence for FoxC2-induced transdifferentiation in our system, FoxC2 may act to suppress white adipogenesis in favor of enhancing brown fat cell formation. In this regard, the engineered fibroblasts employed in this study may lack the positive effectors required to induce the expression of brown adipogenic genes. Further study of FoxC2 and its role in regulating a possible switch from white to brown adipogenesis may provide important information leading to therapies for obesity-associated disorders and metabolic syndrome.