Report

Inhibition of Adipogenesis by Wnt Signaling

See allHide authors and affiliations

Science  11 Aug 2000:
Vol. 289, Issue 5481, pp. 950-953
DOI: 10.1126/science.289.5481.950

Abstract

Wnts are secreted signaling proteins that regulate developmental processes. Here we show that Wnt signaling, likely mediated by Wnt-10b, is a molecular switch that governs adipogenesis. Wnt signaling maintains preadipocytes in an undifferentiated state through inhibition of the adipogenic transcription factors CCAAT/enhancer binding protein α (C/EBPα) and peroxisome proliferator- activated receptor γ (PPARγ). When Wnt signaling in preadipocytes is prevented by overexpression of Axin or dominant-negative TCF4, these cells differentiate into adipocytes. Disruption of Wnt signaling also causes transdifferentiation of myoblasts into adipocytes in vitro, highlighting the importance of this pathway not only in adipocyte differentiation but also in mesodermal cell fate determination.

Adipocytes arise from mesodermal stem cells, which have the capacity to differentiate into a variety of other cell types, including myocytes (1). Once committed to the adipocyte lineage, preadipocytes can remain quiescent, multiply, or undergo differentiation and become adipocytes. 3T3-L1 and 3T3-F442A cells are established mouse preadipocyte models. Both cell lines can be induced to differentiate in cell culture, but 3T3-F442A cells are thought to be arrested at a later point in development (2). Studies of these cellular models have revealed some of the molecular events that orchestrate adipogenesis, including the role of C/EBPs and PPARγ in mediating the expression of adipocyte-specific genes (3, 4).

Wnts are a family of paracrine and autocrine factors that regulate cell growth and cell fate (5). Signaling is initiated when Wnt ligands bind to transmembrane receptors of the Frizzled family. In the canonical Wnt signaling pathway, Frizzleds signal through Dishevelled to inhibit the kinase activity of a complex containing glycogen synthase kinase 3 (GSK3), Axin, β-catenin, and other proteins. This complex targets β-catenin for rapid degradation through phosphorylation. Thus, once hypophosphorylated due to Wnt signaling, β-catenin is stabilized and translocates to the nucleus where it binds the TCF/LEF family of transcription factors to regulate the expression of Wnt target genes (5–7).

To examine the role of Wnt signaling in adipogenesis, we tested whether Wnt expression in 3T3-L1 preadipocytes affected their ability to differentiate. We used Wnt-1 in these experiments because related cell lines, such as NIH-3T3 cells, respond to this ligand (8). In addition, we used two approaches to activate Wnt signaling downstream of the receptor: (i) treatment of cells with lithium, which inhibits GSK3 activity (9), and (ii) expression of a β-catenin mutant (β-catS33Y), which increases β-catenin stability (10). The 3T3-L1 preadipocytes were infected with a retrovirus vector alone (pLXSN) or retroviruses carrying either the genes for Wnt-1 (11) or β-catS33Y. After selection, cells were induced to differentiate in 10% fetal calf serum with methylisobutylxanthine, dexamethasone, and insulin (MDI). Control cells differentiated into adipocytes, as assessed by Oil Red-O staining and by immunoblot detection of the adipocyte fatty acid binding protein, 422/aP2 (12), whereas cells expressing Wnt-1 failed to differentiate (Fig. 1A). Neither lipid droplets nor 422/aP2 were detected in Wnt-1–expressing cells (12). Inhibition of differentiation was also observed when Wnt signaling was activated by lithium or β-catS33Y (Fig. 1A). Thus, Wnt signaling appears to inhibit adipogenesis in vitro.

Figure 1

(A) Wnt signaling blocks adipogenesis in vitro. 3T3-L1 preadipocytes were infected with a retrovirus vector alone (pLXSN) or retroviruses carrying the genes for Wnt-1 (11) or β-catS33Y (10) using standard methods (25). Two days post-confluence, cells were induced to differentiate with MDI (26). A second pLXSN-infected plate was induced to differentiate in the continuous presence of 25 mM LiCl (Lithium). Treatment of cells with 25 mM NaCl does not inhibit differentiation (27). Three weeks later, cells were stained with Oil Red-O (27) to visualize the degree of lipid accumulation. In vitro data in this and subsequent figures are representative of at least three independent experiments. (B) Wnt-1 blocks adipogenesis in vivo. 3T3-F442A preadipocytes were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the Wnt-1 gene. After selection, cells were treated with trypsin and were injected subcutaneously into athymic mice (13, 14). Seven weeks later, the resulting tissue was dissected, fixed in formalin, and embedded in paraffin wax. Sections were stained with hematoxylin and eosin. Bar, 16 μm. In vivo data here and in Fig. 3E are representative of at least two or three animals. (C) Wnt signaling blocks induction of C/EBPα and PPARγ. 3T3-L1 preadipocytes were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the Wnt-1 gene. Two days post-confluence, cells were induced to differentiate with MDI and were lysed at the times indicated for immunoblot analyses of C/EBPα (26) and PPARγ (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). There are three forms of C/EBPα (28): p42C/EBPα (top), p40C/EBPα (middle), and p30C/EBPα (bottom), and two forms of PPARγ (29): γ2 (top) and γ1 (bottom).

Next, we used a model of adipocyte differentiation in which 3T3-F442A preadipocytes are injected subcutaneously into athymic mice (13, 14). Over several weeks, these cells differentiate into adipocytes, forming a discrete fat pad. Seven weeks after implantation, we found that control pLXSN-infected cells differentiated into unilocular and multilocular adipocytes, whereas cells expressing Wnt-1 formed pads composed entirely of undifferentiated, fibroblast-like cells (Fig. 1B). These results indicate that Wnt signaling can repress adipocyte differentiation in a more physiological model.

To gain mechanistic insight into the inhibition of adipogenesis, we examined the effect of Wnt-1 on expression of regulatory genes in the adipogenic program. After induction of differentiation, control and Wnt-1–expressing cells showed a similar pattern of C/EBPβ and C/EBPδ expression (12). Likewise, Wnt-1 had no effect on the expression of c-Myc (15) and CHOP (12). In contrast, C/EBPα and PPARγ, two proteins normally required for adipocyte differentiation (16, 17), were undetectable in Wnt-1–expressing cells at either the protein level (Fig. 1C) or the RNA level (12).

To test whether expression of C/EBPα and PPARγ could rescue differentiation of Wnt-1–expressing cells, we infected 3T3-L1 cells with retroviruses carrying the genes for C/EBPα or PPARγ (18). After selection, cells were induced to differentiate with MDI. Three weeks later, control cells (Fig. 2A) and those expressing C/EBPα alone (Fig. 2B) differentiated into adipocytes that stained vividly with Oil Red-O. Cells expressing PPARγ also differentiated into adipocytes, although these cells contained smaller fat droplets that stained less intensely (Fig. 2C). Cells expressing Wnt-1 alone remained fibroblastic (Fig. 2D). However, when C/EBPα or PPARγ was coexpressed, ∼25% of the cells differentiated into adipocytes (Fig. 2, E and F). Expression of the adipocyte marker, 422/aP2, correlated with the degree of differentiation (12). These results suggest that Wnt-1 represses differentiation by inhibiting the expression of the C/EBPα and PPARγ transcription factors.

Figure 2

C/EBPα and PPARγ partially rescue adipocyte differentiation of Wnt-1–expressing cells. Control (pLXSN) or Wnt-1-expressing cells were reinfected with a retrovirus vector alone (pTS13) (A and D) or retroviruses carrying the genes for C/EBPα (B andE) or PPARγ (C and F) (18) and were coselected with G418 and hygromycin. Two days post-confluence, cells were induced to differentiate with MDI. PPARγ-infected cells were induced with MDI and 5 μM troglitazone. Three weeks later, cells were stained with Oil Red-O. Plates of cells (100 mm) and micrographs are shown. Bar, 20 μm.

To investigate the role of endogenous Wnt signaling in preadipocytes, we inhibited the Wnt pathway in 3T3-L1 cells by expression of dominant-negative TCF4 (dnTCF4) (10), which cannot be activated by β-catenin (19) but does bind to TCF/LEF consensus binding sites. Cells were infected with a retrovirus carrying the dnTCF4 gene and were maintained in 10% calf serum. Whereas control cells failed to differentiate under these conditions, ∼35% of cells expressing dnTCF4 underwent adipogenesis (Fig. 3A), accompanied by expression of C/EBPα, PPARγ, and 422/aP2 (Fig. 3B). Ectopic expression of dnTCF4 also induced adipogenesis of C3H10T1/2 cells and NIH-3T3 fibroblasts (12), suggesting that many cell types with adipogenic potential are restrained from this fate by Wnt signaling.

Figure 3

(A) DnTCF4 causes spontaneous adipogenesis of 3T3-L1 cells in vitro. Preadipocytes were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the dnTCF4 gene and were maintained in 10% calf serum. Two weeks post-confluence, these cells and (for comparison) 3T3-L1 adipocytes (Ads) were stained with Oil Red-O. (B) DnTCF4-induced adipogenesis is accompanied by expression of C/EBPα, PPARγ, and 422/aP2, as assessed by immunoblot analyses. Flag-tagged dnTCF4 was analyzed with antibody to Flag (Sigma). (C) Phase contrast micrograph showing Axin-induced adipogenesis of 3T3-L1 cells. The 3T3-L1 preadipocytes were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the gene for Axin (20) and were maintained in the absence of differentiation media. Many Axin-expressing cells differentiated into adipocytes within a few days, whereas control cells did not. Bar, 20 μm. (D) Free cytosolic β-catenin decreases during adipocyte differentiation. Cytosolic fractions from confluent 3T3-L1 preadipocytes (Preads) and fully differentiated 3T3-L1 adipocytes (Ads) were analyzed by immunoblot for β-catenin (11). (E) DnTCF4 causes spontaneous adipogenesis of 3T3-L1 cells in vivo. Preadipocytes were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the dnTCF4 gene and were injected subcutaneously into athymic mice (13, 14). Seven weeks later, the resulting tissue was dissected and fixed. Sections were stained with hematoxylin and eosin. Bar, 16 μm.

We also tested the effects of a different inhibitor of the Wnt signaling pathway, Axin, which acts by facilitating the phosphorylation (and therefore degradation) of β-catenin (6). Ectopic expression of Axin (20) in 3T3-L1 preadipocytes promoted their differentiation into adipocytes in the absence of inducing agents (Fig. 3C). Consistent with the hypothesis that endogenous Wnt signaling inhibits adipogenesis, free cytosolic β-catenin was found to decrease dramatically over the course of adipocyte differentiation (Fig. 3D).

Unlike 3T3-F442A cells, 3T3-L1 preadipocytes do not undergo adipogenesis when injected into athymic mice (14). To test whether Wnt signaling inhibits differentiation of 3T3-L1 cells in this in vivo model, we examined the effect of ectopic dnTCF4 expression. Seven weeks after implantation, control cells remained undifferentiated, whereas ∼25% of the implanted dnTCF4-expressing cells had differentiated into adipocytes (Fig. 3E). Thus, the inability of 3T3-L1 cells to differentiate in vivo appears to be due to endogenous Wnt signaling.

Other laboratories have shown that Wnt signaling is a positive regulator of myogenesis (21, 22). Because adipocytes and myocytes originate from the same precursor cell, we speculated that Wnt signaling is important for the cell fate decision leading to muscle or fat. To test this idea, we expressed dnTCF4 in C2C12 myoblasts and found that these cells differentiated into adipocytes, as assessed by Oil Red-O staining (Fig. 4A) and expression of an adipocyte marker, 422/aP2 (Fig. 4B). Similar results were observed with a second myoblast model, G8 cells (12). The transdifferentiation of these myoblasts suggests that active Wnt signaling is required for continued commitment to the myocyte lineage.

Figure 4

(A) DnTCF4 causes transdifferentiation of C2C12 myoblasts into adipocytes. C2C12 myoblasts were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the dnTCF4 gene and were maintained in 10% fetal calf serum. Within 2 weeks, some dnTCF-4–infected cells acquired lipid. Bar, 20 μm. (B) DnTCF4-mediated transdifferentiation is accompanied by expression of the adipocyte marker 422/aP2, as assessed by immunoblot analysis. Trace signal in control C2C12 myoblasts (pLXSN) is likely due to cross reactivity of the 422/aP2 antibody with a muscle isoform of the fatty acid binding protein. (C) Wnt-1 and Wnt-10b stabilize free cytosolic β-catenin in 3T3-L1 cells. Preadipocytes were infected with a retrovirus vector alone (pLXSN) or a retrovirus carrying the genes for Wnt-1 or Wnt-10b (30). Cytosolic fractions were analyzed for β-catenin by immunoblot (11). (D) Wnt-10b blocks adipogenesis of 3T3-L1 preadipocytes in vitro. Two days post-confluence, control (pLXSN) and Wnt-10b–infected cells were induced to differentiate with MDI (26). Two weeks later, cells were stained with Oil Red-O. Bar, 20 μm. Representative micrographs are shown. (E) Wnt-10b is repressed during differentiation of 3T3-L1 preadipocytes. RNA was isolated from cells that were 70% confluent (70%), 2 days post-confluent (Day 0), and cells that had been induced with MDI for the number of days indicated. Ribonuclease protection analysis (31) with probes for mouse Wnt-10b [amplified by PCR (24)] and TCF4 (32). Yeast tRNA served as the control for background protection. (F) Wnt-10b is expressed at higher levels in 3T3-L1 cells than in 3T3-F442A cells, as assessed by ribonuclease protection analysis of RNA from confluent cells.

The finding that inhibition of Wnt signaling in preadipocytes and myoblasts induces adipogenesis suggests that these mesodermal cells resist entry into the adipogenic program by secreting a member of the Wnt family. We therefore determined which Wnts were expressed in 3T3-L1 cells. Using a combination of reverse transcriptase–polymerase chain reaction (RT-PCR) with pan-specific Wnt primers (23) and ribonuclease protection analyses, we identified Wnt-5a and Wnt-10b. Of these, only Wnt-10b was able to stabilize free cytosolic β-catenin (Fig. 4C) and block adipogenesis when ectopically expressed (Fig. 4D). Analysis of Wnt-10b mRNA over the course of 3T3-L1 adipocyte differentiation revealed that its expression was highest in growing and confluent cells, but decreased upon MDI treatment (Fig. 4E), which is consistent with the idea that a decrease in Wnt-10b is required for adipogenesis to occur. Wnt-10b may act through TCF4, which is constitutively expressed throughout adipocyte differentiation (Fig. 4E). In accordance with our suggestion that Wnt signaling underlies the different adipogenic tendencies of 3T3-L1 and 3T3-F442A cells, Wnt-10b is expressed at much higher levels in 3T3-L1 preadipocytes than 3T3-F442A cells (Fig. 4F). Collectively, these data suggest that Wnt-10b is an endogenous regulator of adipogenesis.

In summary, we report that Wnt signaling functions as an adipogenic switch. When it is on, adipogenesis is repressed; when it is off, adipogenesis is initiated. The crucial role of Wnt signaling in the adipogenic program is emphasized by the finding that in its absence, myoblasts are reprogrammed to the adipocyte lineage and undergo spontaneous differentiation.

  • * To whom correspondence should be addressed. E-mail: macdouga{at}umich.edu

REFERENCES AND NOTES

View Abstract

Navigate This Article