Repressor transcription factor 7-like 1 promotes adipogenic competency in precursor cells

Abstract

The identification of factors that define adipocyte precursor potential has important implications for obesity. Preadipocytes are fibroblastoid cells committed to becoming round lipid-laden adipocytes. In vitro, this differentiation process is facilitated by confluency, followed by adipogenic stimuli. During adipogenesis, a large number of cytostructural genes are repressed before adipocyte gene induction. Here we report that the transcriptional repressor transcription factor 7-like 1 (TCF7L1) binds and directly regulates the expression of cell structure genes. Depletion of TCF7L1 inhibits differentiation, because TCF7L1 indirectly induces the adipogenic transcription factor peroxisome proliferator-activated receptor γ in a manner that can be replaced by inhibition of myosin II activity. TCF7L1 is induced by cell contact in adipogenic cell lines, and ectopic expression of TCF7L1 alleviates the confluency requirement for adipocytic differentiation of precursor cells. In contrast, TCF7L1 is not induced during confluency of non-adipogenic fibroblasts, and, remarkably, forced expression of TCF7L1 is sufficient to commit non-adipogenic fibroblasts to an adipogenic fate. These results establish TCF7L1 as a transcriptional hub coordinating cell–cell contact with the transcriptional repression required for adipogenic competency.

Adipose tissue is a highly specialized compartment of cells actively involved in maintaining global metabolic homeostasis through lipid synthesis and storage, adipokine secretion, and insulin responsiveness (1). Adipocytes compose the majority of cells in adipose tissue and play a critical role in normal physiology, but their dysfunction is also at the center of a diverse range of diseases, including obesity, diabetes, and lipodystrophies (2). Furthermore, primary preadipocytes and adipose-derived stem cells have shown promise in treating multiple conditions (3–5). Therefore, it is critical to understand the process by which spindly fibroblastic precursor cells undergo conversion into round lipid-laden fat cells.

In vitro models of adipogenesis, such as the extensively studied committed preadipocyte cell line 3T3-L1 cells, have elucidated two major phases of adipogenesis: commitment and terminal differentiation (6, 7). Terminal differentiation is characterized by the induction of metabolic genes, many of which are the direct targets of the transcription factors peroxisome proliferator-activated receptor γ (PPARγ) and C/CAAT-binding protein (C/EBP) α and β (8–14). Recent efforts have focused on identifying committed preadipocyte populations in vivo (15, 16), as well as on determining molecular factors that define the committed preadipocytes phenotype. Zinc finger protein 423 (Zfp423) is a critical preadipocyte factor upstream of PPARγ that is not present in non-adipogenic fibroblasts (17). However, Zfp423 also has been identified as a regulator of neurologic development (18), suggesting that other factors also may be involved in specifying adipogenic competency and commitment of precursor cells upstream of PPARγ.

Confluency could provide insight into other factors that confer adipogenic competency, because it promotes adipogenesis in many model systems (19, 20). This cell–cell contact is associated with substantial reorganization of the actomyosin as well as the microtubule cytoskeleton, providing permissive conditions for adipocyte differentiation (21–23). Moreover, several studies have found that cell shape regulation is essential for determining lineage decisions in mesenchymal stem cells (MSCs) (22, 24, 25). Interestingly, many of the genes repressed early after the addition of adipogenic stimuli to confluent preadipocytes are regulators of cell structure (26–28). The repressed cell structure genes are not enriched as genomic targets for PPARγ or C/EBPα (8, 9), suggesting a role for an as-yet unknown transcriptional repressor in regulation of cell shape during adipocyte differentiation.

Transcription factor 7-like 1 (TCF7L1, formerly known as TCF3) is an intriguing candidate for such a repressor. Transcription factor proteins play a role in the canonical Wnt pathway that regulates adipogenesis (29), MSC lineage commitment (30), and expression of cell structure genes (31). A dominant negative form of TCF7L2 promotes adipogenesis (29), and the transcription factor 7 family member motif is enriched at sites of histone modification in preadipocytes (26). TCF7L1 is of particular interest because it has been genetically linked to type 2 diabetes (32) and shown to be an important transcriptional repressor of canonical Wnt signaling targets (33–36). TCF7L1 regulates cell fate decisions in mouse embryonic stem cells (36, 37) and is a key regulator of terminal differentiation of other tissues (34, 38, 39). However, the extent to which TCF7L1 is important for mammalian cell differentiation remains unknown, because TCF7L1 null mice are early embryonic lethal (33).

Here we show that TCF7L1 represses structure-related genes during adipogenesis. Intriguingly, TCF7L1 is induced in a cell contact–dependent manner by confluency in preadipocytes and is required for adipocyte differentiation by repressing transcription of cell structure genes. TCF7L1 also is sufficient to bestow adipogenic potential on non-adipogenic cells. These results implicate TCF7L1 as an adipogenic competency factor that uniquely determines adipogenic fate through cell structure organization required for adipocyte gene activation.

Results

TCF7L1 Represses Cytostructural Genes During Adipocyte Differentiation.

To determine whether TCF7L1 was directly targeting the cell structure genes rapidly repressed on the addition of adipogenic stimuli to confluent cells (26, 27), we used ChIP early in adipogenesis with a validated TCF7L1 antibody (Fig. S1A), followed by deep sequencing (ChIP-seq). This analysis identified 556 high-confidence [i.e., 2% false discovery rate (FDR)] binding sites in the 3T3-L1 genome. A representative sample of these sites was confirmed by TCF7L1 ChIP-qPCR in control cells, with a loss of enrichment detected in TCF7L1-depleted cells (Fig. 1A). TCF7L1 binding was detected primarily in intergenic regions (Fig. 1B), and binding sites demonstrated marked conservation with other species (Fig. S1B), as has been found for other transcription factors (8, 9, 40). Furthermore, these binding sites are enriched in TCF7L1-related motifs (Table S1). Interestingly, previously described Wnt pathway targets in adipogenesis, such as cyclin D1, COUP-TFII, and FABP4 (41, 42), did not have any TCF7L1-binding sites (data available at the GEO Web site, www.ncbi.nlm.nih.gov/geo), suggesting that TCF7L1 may be binding at a subset of classic Wnt target genes or independent of this pathway, as has been described previously for skin stem cells (39). Using PANTHER to group related genes in pathways (43), we found that cell structure was the most enriched biological process in the 833 genes within 100 kb of TCF7L1-binding sites (Fig. 1C).

Fig. 1.

TCF7L1 represses cytostructural genes during adipocyte differentiation. (A) Validation of TCF7L1-binding sites from ChIP-seq. TCF7L1 ChIP-qPCR in control cells and TCF7L1-depleted 3T3-L1 cells at 24 h after the addition of adipogenic stimull (DMI; see SI Materials and Methods). (B) Distribution of TCF7L1-binding sites throughout the genome. (C) TCF7L1 binds near genes in cell structure pathways. PANTHER biological processes enriched with an FDR <5% for all genes within 100 kb of TCF7L1-binding sites. (D) HDAC1 colocalizes with TCF7L1. HDAC1 ChIP-qPCR in 3T3-L1 cells during early adipogenesis. (E) Reduced H3K9Ac at TCF7L1-binding sites. Average H3K9ac profile at TCF7L1-binding sites in day 0 preadipocytes, at 24 h after addition of DMI, and in day 10 mature adipocytes. The profile of input for each time point was determined as well. The one-tailed Wilcoxon rank-sum test was used to compare the difference in acetylation at TCF7L1-binding sites with acetylation at matched control regions in a 1-kb region. P < 0.05, preadipocyte > 24 h after DMI > adipocytes. (F) TCF7L1 binding is increased at genes repressed during adipogenesis. Percentage of TCF7L1 and PPARγ-binding sites within 100 kb of genes repressed (1,406 genes) or induced during adipogenesis (1,011 genes) or a random set of genes (1,366 genes). Fisher's exact test was used to compare the percentage of TCF7L1-binding sites near repressed or induced genes with the percentage near random genes. The χ² test was used to compare the percentage of PPARγ-binding sites near repressed or induced genes with the percentage near random genes. ***P < 0.001; *P < 0.05. For A and D, 20 TCF7L1-binding sites and five negative control sites were interrogated. Each point represents the percent input of one site. The lines represent mean ± SEM for TCF7L1 or control sites in each cell population, *** P < 0.001.

TCF family members repress transcription by recruiting histone deacetyase 1 (HDAC1) (44), a chromatin-modifying enzyme shown to play critical roles in determining progression of the early adipogenic cascade (45–47). HDAC1 was significantly enriched at TCF7L1-binding sites relative to negative control sites (Fig. 1D) and to a greater extent than at enhancers without TCF7L1-binding sites, where HDAC1 nevertheless can be recruited by other transcription factors (Fig. S1C). Histone 3 lysine 9 acetylation (H3K9ac) positively correlates with active transcription at both transcription start sites and enhancer regions (48), and previous reports validated this histone mark as a marker of active transcription during adipogenesis and in mature adipocytes (8, 49). Indeed, H3K9ac enrichment was diminished at TCF7L1-binding sites during adipogenesis (Fig. 1E), specifically at genes repressed during differentiation (Fig. S1D).

Approximately one-quarter of TCF7L1-binding sites were found within 100 kb of genes down-regulated during adipogenesis, much more than at random genes (Fig. 1F). In comparison, a similar enrichment of PPARγ-binding sites was noted within 100 kb of genes induced during adipogenesis (Fig. 1F), consistent with the known role of PPARγ as a transcriptional activator (50, 51). TCF7L1 binding also was enriched at induced genes (Fig. 1F), at a similar number of average binding sites per gene (Fig. S1E), suggesting that TCF7L1 may have a function at those sites as well. However, H3K9ac did not increase at these sites during early adipogenesis, and thus it is unlikely that TCF7L1 functioned as a transcriptional activator (Fig. S1F). Taken together, the enrichment of TCF7L1 binding and reduced H3K9ac at sites neighboring down-regulated genes suggests that TCF7L1 indeed functions as a transcriptional repressor in early differentiation.

TCF7L1 Is Required for Adipogenesis.

Because TCF7L1 was found to bind near cytostructure-related genes, and cell structure regulation is critical for adipogenesis, we next asked whether TCF7L1 is required for adipogenesis by using two independent siRNA to deplete TCF7L1 from 3T3-L1 preadipocytes. Adipocyte differentiation was impaired in the TCF7L1-depleted cells but not in cells treated with nontargeting siRNA, as assessed by Oil Red O (ORO) staining for neutral lipids (Fig. 2A) and accumulation of adipocyte proteins (Fig. 2B). Moreover, adipocyte gene expression was reduced (Fig. 2C) and preadipocyte gene expression was increased (Fig. 2D) in TCF7L1-depleted cells subjected to the adipocyte differentiation protocol, and depletion of TCF7L1 in 3T3-F442A cells, an alternate adipogenic model, also led to a substantial reduction in lipid accumulation (Fig. S2), suggesting that TCF7L1 is required for adipogenesis.

Fig. 2.

TCF7L1 depletion abrogates adipogenesis. (A) Reduced lipid accumulation. ORO staining (Upper) and phase-contrast microscopy (Lower) of siControl, siTCF7L1 #1, and siTCF7L1 #2 3T3-L1 electroporated cells differentiated for 7 d. (B) Reduced adipocyte protein expression. TCF7L1 and RAN protein levels in preadipocytes before DMI treatment (Upper) and PPARγ, FABP4, and RAN protein levels in day 7 adipocytes (Lower). (C) Reduced expression of adipocyte genes. siControl, siTCF7L1 #1, and siTCF7L1 #2 cells at day 7. (D) Increased expression of preadipocyte genes. siControl, siTCF7L1 #1, and siTCF7L1 #2 cells at day 7. Graphed values represent mean ± SEM (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001.

TCF7L1 Indirectly Activates the PPARγ Gene Locus via Effects on Cell Structure.

We next sought to determine whether TCF7L1 regulates early adipogenesis. C/EBPβ is induced within 24 h of adipogenic stimulation of 3T3-L1 cells (52), and PPARγ2 also was induced at this time point (Fig. 3A), although this induction represented only a fraction of the eventual induction of PPARγ2 in mature adipocytes (Fig. S3A). Depletion of TCF7L1 had little effect on C/EBPβ induction (Fig. S3B), but markedly attenuated early PPARγ2 induction (Fig. 3A). Consistent with a role for TCF7L1 in the upstream regulation of PPARγ, ectopic expression of PPARγ was able to rescue the adipogenic defect in TCF7L1-depleted preadipocytes (Fig. 3 B and C). Evaluation by FAIRE (Formaldehyde-Assisted Isolation of Response Elements) (53) of an enhancer located ∼182 kb upstream of the PPARγ2 promoter that is functional early in adipogenesis and whose activation is correlated with PPARγ2 expression (49) revealed that depletion of endogenous TCF7L1 in preadipocytes led to a more repressive, closed chromatin structure (Fig. 3D), which may contribute to the reduced activation of the PPARγ gene. Of note, the −182-kb enhancer was not bound by TCF7L1 early in adipogenesis (Fig. S3C), suggesting that the positive effect of TCF7L1 on enhancer loci accessibility is indirect. Although TCF7L1-binding sites were found neighboring PPARγ start sites, H3K9ac did not change at these sites during early adipogenesis (Fig. S3D), and the binding was markedly weaker than that at other TCF7L1 sites validated by ChIP-qPCR (Fig. S3E).

Fig. 3.

TCF7L1 indirectly regulates PPARγ enhancer in early adipogenesis. (A) TCF7L1 depletion prevents PPARγ2 induction. PPARγ2 mRNA levels in siControl and siTCF7L1 3T3-L1 cells at 0 h and 24 h after DMI addition. (B) Ectopic PPARγ2 rescues adipogenic defects in TCF7L1-depleted cells. ORO (Upper) and phase-contrast microscopy (Lower) at 7 d after the addition of adipogenic stimuli to 3T3-L1 preadipocytes infected with Empty or PPARγ2 virus and electroporated with siTCF7L1 or siControl. (C) Ectopic PPARγ2 rescues adipocyte-specific protein expression in TCF7L1-depleted cells. TCF7L1, PPARγ, and RAN protein levels in 3T3-L1 cells before DMI treatment (Upper) and PPARγ, FABP4, and RAN protein levels at 7 d after adipogenic stimuli (Lower). (D) TCF7L1 depletion decreases PPARγ2 gene enhancer accessibility. FAIRE-qPCR at the 36b4 and PPARγ2 −182-kb enhancer in siControl and siTCF7L1 3T3-L1 cells treated for 24 h with DMI. (E) TCF7L1 depletion increases levels of cell structure–related genes that neighbor TCF7L1-binding sites, but not Dlk1, which lacks TCF7L1-binding sites. Col1α2, Dcn, Dlk1, Dpt, Ptn, Tes, Thbs1, Thbs2, and Tiam2 mRNA levels in siControl and siTCF7L1 3T3-L1 cells were treated for 24 h with DMI. In A, D, and E, graphed values represent mean ± SEM (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001.

We first considered whether TCF7L1 regulated Dlk1/Pref-1, whose down-regulation is required for adipogenesis (54, 55). However, our ChIP-seq experiment did not identify a TCF7L1-binding site near the Dlk1 gene (Fig. S4A; and complete dataset at GEO, www.ncbi.nlm.nih.gov/geo), and Dlk1/Pref-1 down-regulation during early adipogenesis was unaffected by TCF7L1 depletion (Fig. 3E), indicating that the effects of TCF7L1 are independent of Dlk1/Pref-1. In contrast, during early adipogenesis, TCF7L1 depletion did induce several cell structure genes with nearby TCF7L1-binding sites (Fig. 3E and Fig. S4 B and C).

Adipogenesis in TCF7L1-Depleted Cells Is Rescued by Inhibiting Myosin Activity.

Because cytostructure has been shown to regulate adipogenesis and adipogenic gene expression (21–25), we hypothesized that TCF7L1-dependent induction of PPARγ2 is an indirect result of cytostructural regulation. TCF7L1-depleted cells demonstrated an increase in myosin fiber formation (Fig. 4A). To test whether TCF7L1 regulation of myosin is required for adipogenesis, we treated TCF7L1-depleted preadipocytes with compounds that inhibit specific cytoskeletal components and have been shown to promote adipogenesis (21, 22, 25). Blebbistatin, a myosin II ATPase inhibitor, rescued PPARγ2 levels in TCF7L1-depleted cells during the first 24 h of differentiation (Fig. 4B). In contrast, disrupters of actin filaments (cytochalasin D) or microtubules (nocododazole) did not rescue PPARγ2 in TCF7L1-depleted cells (Fig. S5). These results suggest that the TCF7L1-dependent induction of PPARγ2 is mediated by myosin-dependent changes of cell organization. Furthermore, blebbistatin rescued adipogenesis in TCF7L1-depleted cells (Fig. 4C). Together, these data suggest that TCF7L1-dependent cell shape regulation indirectly remodels the chromatin conformation of an adipogenic enhancer of PPARγ2 expression, promoting adipocyte differentiation.

Fig. 4.

Myosin inhibition rescues TCF7L1-depleted cells. (A) TCF7L1-depleted cells display an increase in myosin fiber formation. Myosin IIa immunofluorescence in siControl and siTCF7L1 3T3-L1 cells treated with DMI for 24 h. Nuclei are counterstained with DAPI. (Scale bar: 20 μm.) Graphed values represent mean ± SEM (n = 3). *P < 0.05; **P < 0.01; ***P < 0.001. (B) Inhibition of myosin contraction rescues PPARγ2 expression in TCF7L1-depleted preadipocytes. PPARγ2 expression levels in siControl and siTCF7L1 3T3-L1 cells were treated for 24 h with DMI and DMSO or 50 μM blebbistatin. The one-tailed Student t test was used to determine significance. (C) Increased lipid accumulation in blebbistatin-treated, TCF7L1-depleted cells. ORO (Upper) and phase-contrast microscopy (Lower) at 7 d after addition of adipogenic stimuli to 3T3-L1 preadipocytes electroporated with siTCF7L1 or siControl. During the first 4 d of the differentiation protocol, cells were treated with DMSO or 50 μM blebbistatin as indicated.

TCF7L1 Is Permissive for Adipogenesis of Preconfluent Preadipocytes.

Because confluency is associated with important cytostructural changes in adipogenesis (22, 28), we hypothesized that the transcriptional repressor of cell structure genes must be present in early adipogenesis and may be induced by confluency. TCF7L1 was present in confluent preadipocytes and early adipogenesis, but diminished as terminal differentiation progressed (Fig. 5A). However, TCF7L1 was induced by confluency in multiple models of adipogenesis before the addition of adipogenic stimuli (Fig. 5B). The confluency-dependent induction of TCF7L1 was specific; neither isoform of the related TCF7L2 changed on confluency (Fig. 5B). Induction of TCF7L1 mRNA with confluence required cell contact, and mRNA levels were not affected by the treatment of preconfluent cells with conditional media from confluent cells (Fig. S6). TCF7L1 protein induction during confluency was blocked by the calcium chelator EGTA, which was rescued by addition of calcium chloride, but not magnesium chloride, to the cells (Fig. 5C). This suggests a role for cadherins, calcium-dependent cell junction proteins implicated in a wide variety of developmental processes (56). Indeed, treatment of confluent cells with a peptide that inhibits type 1 cadherins (57), but not a control peptide, decreased levels of TCF7L1 in confluent cells (Fig. 5D). Therefore, induction of TCF7L1 by confluency likely is cadherin-dependent.

Fig. 5.

TCF7L1 promotes adipogenesis in preconfluent cells. (A) TCF7L1 during adipogenesis. TCF7L1, PPARγ, FABP4, and RAN protein levels during 3T3-L1 adipogenesis after addition of DMI at day 0. (B) TCF7L1 is induced by the confluency of preadipocytes. TCF7L1, TCF7L2, and RAN protein levels in preconfluent and confluent 3T3-L1 preadipocytes, 3T3-F442A preadipocytes, and primary MEFs. (C) TCF7L1 induction by confluency is calcium-dependent. TCF7L1 and RAN protein levels in confluent 3T3-L1 preadipocytes were cultured in the presence of 2 mM EGTA, 2 mM CaCl₂, or 2 mM MgCl₂ for 16 h. (D) TCF7L1 induction is cadherin-dependent. Confluent preadipocytes were treated for 3 d with vehicle (H₂O), 500 μg/mL of control cyclic peptide, or 500 μg/mL of pan type 1 cadherin blocking cyclic peptide. (E) Increased lipid accumulation. Phase-contrast microscopy of preconfluent preadipocytes infected with empty or TCF7L1 virus before addition of DMI (Left). ORO (Center) and phase-contrast (Right) microscopy of same cells 7 d after addition of adipogenic stimuli. (F) Increased adipocyte protein expression. TCF7L1 and RAN protein levels in preconfluent and confluent preadipocytes before DMI treatment (Upper) and PPARγ, FABP4, and RAN protein levels at 7 d after addition of adipogenic stimuli (Lower).

We next addressed the importance of TCF7L1 in mediating the permissive effect of confluency on adipogenesis. Preadipocytes transduced with empty retrovirus did not undergo adipogenesis when exposed to differentiation medium before confluency (Fig. 5 E and F), although they eventually become confluent during the differentiation process. In contrast, under the same conditions, ectopic expression of TCF7L1 was sufficient to support robust adipogenesis of preconfluent preadipocytes (Fig. 5 E and F). Cell structure and motility pathways were the only categories of repressed genes enriched in preconfluent preadipocytes expressing TCF7L1 (Fig. S7), confirming the specific role of TCF7L1 in repressing cytostructural genes. Thus, TCF7L1 acts specifically as a mediator of the confluency requirement for adipogenesis. Although TCF7L1 overcame the requirement for confluency, adipogenesis did not occur without exposure to adipogenic stimuli (Fig. S8 A and B), suggesting that induction of TCF7L1 functions specifically as a mediator of adipogenic competency, but not terminal differentiation.

TCF7L1 Is Sufficient to Confer Adipogenic Competency.

Although NIH 3T3 cells are non-adipogenic, ectopic expression of such factors as PPARγ (58) and Zfp423 (17) is sufficient to convert these cells to adipocytes in the presence of differentiation medium. NIH 3T3 cells lack robust TCF7L1 expression (Fig. 6A). Remarkably, ectopic expression of TCF7L1 in NIH 3T3 cells rendered these cells competent to undergo nearly complete adipogenesis at the levels of lipid accumulation and morphology (Fig. 6B) as well as gene expression (Fig. 6 C–E). Consistent with the hypothesis that TCF7L1 promotes a preadipocyte-like phenotype, ectopic expression of TCF7L1 in NIH 3T3 cells dramatically induced PPARγ2 expression (Fig. 6F) and increased accessibility of the PPARγ gene enhancer after adipogenic stimulation (Fig. 6G). TCF7L1 promotion of adipogenic competency did not require induction of Zfp423 (Fig. 6H), implying that TCF7L1 may be downstream of Zfp423 or can promote adipogenesis in a parallel pathway.

Fig. 6.

TCF7L1 confers adipogenic competency to NIH 3T3 cells. (A) TCF7L1 is not induced by confluency of non-adipogenic NIH 3T3 cells. TCF7L1 and RAN protein levels in confluent 3T3-L1 cells and preconfluent, confluent, and DMI plus rosiglitazone (TZD) treated NIH 3T3 cells. (B) Adipogenesis of NIH 3T3 cells expressing TCF7L1. ORO (Upper) and phase-contrast microscopy (Lower) of NIH 3T3 cells infected with Empty or TCF7L1 virus 7 d after addition of adipogenic stimuli. (C) Adipocyte protein expression in NIH 3T3 cells expressing TCF7L1. TCF7L1 and RAN protein levels in NIH 3T3 cells before DMI plus TZD treatment (Upper) and PPARγ, FABP4, and RAN protein levels at 7 d after adipogenic stimuli (Lower). (D) Increased adipocyte gene expression in NIH 3T3 cells expressing TCF7L1. (E) Decreased preadipocyte gene expression in NIH 3T3 cells expressing TCF7L1. (F) TCF7L1 promotes PPARγ2 expression in NIH 3T3 cells. Empty and TCF7L1 virus-infected NIH 3T3 cells at 0 h and 24 h after DMI plus TZD addition. (G) TCF7L1 promotes PPARγ2 gene enhancer accessibility in NIH 3T3 cells. Empty and TCF7L1 NIH 3T3 cells treated for 24 h with DMI plus TZD. (H) Zfp423 expression levels. 3T3-L1 preadipocytes and NIH 3T3 cells infected with TCF7L1 or empty virus, For D–H, graphed values represent mean ± SEM (n = 3). **P < 0.01; **P < 0.01; ***P < 0.001. ns, not significant.

Discussion

We have shown that confluency of committed preadipocytes induces the transcriptional repressor TCF7L1, which binds to specific sites in chromatin to silence genes involved in cell structure after addition of adipogenic stimuli. TCF7L1-mediated cell structure changes contribute to PPARγ2 induction, providing synchrony between adipogenic commitment and terminal differentiation. This critical role of TCF7L1 is reflected in its requirement for adipocyte differentiation and its sufficiency as an adipogenic competency factor for non-adipogenic fibroblasts.

To date, mammalian studies of TCF7L1 have focused on its role in regulating pluripotency (37) and skin stem cell differentiation (34, 39). The present study identifies a role for TCF7L1 in adipogenesis, where it likely functions as a transcriptional repressor of cell structure–related genes. This study also provides evidence that a downstream transcription factor is directly involved in critically regulating cell structure during the transition from adipogenic commitment to terminal differentiation, and illustrates the predictive power of ChIP-seq data to discover novel cell type–specific functions for transcription factors during differentiation. Nevertheless, it is possible that TCF7L1 also may have indirect effects on other factors affecting adipogenesis.

In addition, this study has identified a molecular mechanism by which confluency signals transcriptional events, mediated by TCF7L1, allowing adipogenic stimuli-dependent regulation of cytostructural reorganization. Thus, TCF7L1 acts as a hub between two well-known but unexplained phenomena that occur during adipocyte differentiation: the requirement for cell–cell contact and the subsequent repression of genes controlling cell structure during adipogenesis. Elevated PPARγ level is a hallmark of in vivo adipocyte precursor populations (16); thus, the cytostructural regulator TCF7L1 confers a preadipocyte-like phenotype by promoting PPARγ enhancer accessibility and expression during early adipogenesis. Importantly, in TCF7L1-depleted cells, PPARγ2 levels and adipocyte differentiation could be rescued by chemical inhibition of myosin ATPase activity. Thus, these data also suggest that TCF7L1 provides a link between structural changes during adipogenic commitment and subsequent gene induction during terminal differentiation.

Our findings indicate that TCF7L1 induction by preadipocyte confluency may be dependent on cadherin-mediated cell contact. Cadherins, functioning as calcium-dependent cell–cell junctions, have been previously associated with a wide variety of developmental processes (56), including cell fate decisions (24), but their role in adipocyte differentiation has not been explored extensively. Cadherin composition or function may be a distinguishing feature of adipogenic and non-adipogenic fibroblasts, given that cell contact does not induce TCF7L1 in NIH 3T3 cells.

Recently, Zfp423 was reported to be present in preadipocytes, but not non-adipogenic fibroblasts, and to be required for adipogenesis (17). However, TCF7L1 does not require robust Zfp423 expression in NIH 3T3 cells to bestow adipogenic competency to these cells, demonstrating that induction of TCF7L1 is also a major determinant of the adipocyte precursor phenotype. Other stem cell populations have been shown to require a core group of transcription factors that cooperatively maintain precursor characteristics, such that disruption of any single component may alter the differentiation potential (37). Because both TCF7L1 and Zfp423 are capable of promoting a commitment to adipogenesis in NIH 3T3 cells, it seems likely that preadipocyte populations subscribe to this paradigm. Moreover, the role of TCF7L1 in adipogenesis provides general mechanistic insights into how cells can coordinate cell contact, structure, and transcriptional regulation during the process of differentiation.

Materials and Methods

Further details on the materials and methods used in this study are provided in SI Materials and Methods.

Cell Culture.

Murine 3T3-L1, 3T3-F442A, NIH 3T3, MEFs, and BOSC cells were maintained in DMEM (Invitrogen) supplemented with 10% FBS (Tissue Biologicals) and 1% penicillin/streptomycin (Invitrogen). 3T3-L1, 3T3-F442A, and NIH 3T3 were differentiated as described in SI Materials and Methods.