Research ArticleStem Cells

The transcription factor Lef1 switches partners from β-catenin to Smad3 during muscle stem cell quiescence

See allHide authors and affiliations

Science Signaling  24 Jul 2018:
Vol. 11, Issue 540, eaan3000
DOI: 10.1126/scisignal.aan3000

Muscle stem cells swap coactivators

After birth, skeletal muscle stem cells (MuSCs) transition from a proliferative to a quiescent state. These quiescent stem cells, also known as satellite cells, retain the ability to self-renew and, upon activation, re-enter the cell cycle to generate myoblasts for muscle repair. The entry of MuSCs into quiescence is characterized by the expression of target genes that are activated by the Tcf and Lef (Tcf/Lef) family of transcription factors. Aloysius et al. found that, although Tcf/Lef transcription factors promoted gene expression by cooperating with β-catenin in response to Wnt signaling in proliferating myoblasts, β-catenin was not required for the induction of Tcf/Lef target genes during quiescence. Instead, Lef1 partnered with the transcriptional coactivator Smad3. Smad3 mediates transcriptional responses to transforming growth factor–β (TGF-β) signaling, implicating cross-talk between the Wnt and TGF-β pathways in controlling MuSC quiescence.

Abstract

Skeletal muscle stem cells (MuSCs), also known as satellite cells, persist in adult mammals by entering a state of quiescence (G0) during the early postnatal period. Quiescence is reversed during damage-induced regeneration and re-established after regeneration. Entry of cultured myoblasts into G0 is associated with a specific, reversible induction of Wnt target genes, thus implicating members of the Tcf and Lef1 (Tcf/Lef) transcription factor family, which mediate transcriptional responses to Wnt signaling, in the initiation of quiescence. We found that the canonical Wnt effector β-catenin, which cooperates with Tcf/Lef, was dispensable for myoblasts to enter quiescence. Using pharmacological and genetic approaches in cultured C2C12 myoblasts and in MuSCs, we demonstrated that Tcf/Lef activity during quiescence depended not on β-catenin but on the transforming growth factor–β (TGF-β) effector and transcriptional coactivator Smad3, which colocalized with Lef1 at canonical Wnt-responsive elements and directly interacted with Lef1 specifically in G0. Depletion of Smad3, but not β-catenin, reduced Lef1 occupancy at target promoters, Tcf/Lef target gene expression, and self-renewal of myoblasts. In vivo, MuSCs underwent a switch from β-catenin–Lef1 to Smad3-Lef1 interactions during the postnatal switch from proliferation to quiescence, with β-catenin–Lef1 interactions recurring during damage-induced reactivation. Our findings suggest that the interplay of Wnt-Tcf/Lef and TGF-β–Smad3 signaling activates canonical Wnt target promoters in a manner that depends on β-catenin during myoblast proliferation but is independent of β-catenin during MuSC quiescence.

INTRODUCTION

Most cells in adult mammals are nondividing, with tissues largely composed of terminally differentiated cells and stem cells that persist in a reversibly arrested or quiescent state. Accumulating evidence suggests that a quiescence-induced program enhances stem cell self-renewal, which is critical for homeostasis and repair. Thus, despite low rates of transcription, translation, and metabolic activity in quiescent cells, quiescence is not a passive loss of replicative ability but rather a tightly regulated state in which protective programs are induced to prevent loss of stem cells to differentiation, apoptosis, and senescence and to prime cells for activation by external cues (1, 2). In skeletal muscle, muscle stem cells (MuSCs) called satellite cells are located in a niche formed by the plasma membrane of the multinucleated myofiber and its ensheathing basement membrane (3). MuSCs are quiescent in uninjured muscle and re-enter the cell cycle when the muscle is injured. As these cells proliferate, some of their progeny differentiate to regenerate damaged myofibers, and some repopulate the quiescent niche, enabling future repair (46).

Several signaling pathways have been implicated in MuSC function (7). Evidence suggests that growth factors such as hepatocyte growth factor and fibroblast growth factor (FGF) initiate the mitogen-activated immediate early response in quiescent MuSCs, that p38 mitogen-activated protein kinase and Notch pathways are involved in asymmetric division (8, 9), and that insulin-like growth factor, myostatin, and transforming growth factor–β (TGF-β) play roles in differentiation (1012). Notch signaling also promotes a quiescent self-renewing state (13, 14). The Wnt signaling pathway plays diverse roles in stem cell biology. For example, the Wnt pathway promotes stem cell maintenance, proliferation, and lineage determination of several adult tissues and is also important for maintaining pluripotency in embryonic stem cells (ESCs) (1517). Despite a key role for Wnt signaling in myogenesis (18), the role of Wnt signaling in adult MuSCs is still emerging.

Canonical Wnt signaling is activated when secreted Wnt ligands bind to receptors of the Frizzled (FZD) family and the co-receptors lipoprotein receptor–related protein 5 or 6 (LRP5/6), which induces the release of β-catenin from a cytoplasmic degradation complex composed of the scaffold protein Axin2, the tumor suppressor adenomatous polyposis coli, casein kinase 1, and glycogen synthase kinase-3 (GSK3). Nonphosphorylated, active β-catenin enters the nucleus, where it does not bind DNA directly but forms complexes with transcription factors of the Tcf and Lef1 (Tcf/Lef) family at target gene promoters, displacing the Groucho and Tle family co-repressors (19), to promote transcription. Unraveling the effects of this pathway is complicated by the existence of 19 Wnts, 10 FZDs, and 2 LRPs (19) in mammals, with added complexity created by cross-talk with other signaling circuits (20).

In adult muscle, activation and proliferation of MuSCs are associated with Wnt activation and nuclear localization of β-catenin (21), whereas differentiation of their progeny is accompanied by a switch from Notch to Wnt signaling (22). However, published studies offer conflicting views of the requirement for Wnt and β-catenin in MuSCs. Conditional ablation of β-catenin showed it to be dispensable for MuSC self-renewal and regeneration (23). By contrast, other studies showed that precise control of β-catenin abundance is essential for proper regeneration (24, 25). Although these studies implicate β-catenin in damage-induced muscle repair, its role in MuSC quiescence is still unclear.

We have investigated the molecular events regulating quiescence using reversibly arrested C2C12 myoblasts that model MuSCs (26, 27). Here, quiescence is induced by culturing proliferating myoblasts in suspension (G0 myoblast) and rapidly reversed by replating the cells on an adherent substrate. These conditions reproduce key molecular events associated with reversible MuSC quiescence in vivo: enhanced expression of the MuSC specification and survival factor Pax7 and the stem cell marker CD34, repression of the lineage determinants MyoD and Myf5, the absence of the key myogenic transcription factor myogenin, and activation of cell cycle inhibitors p27 and p130. Reactivation of G0 myoblasts restores MyoD and Myf5 expression, cell cycle re-entry, and differentiation potential (28). G0-associated genes identified using this in vitro model (27) have been shown to participate in quiescence and activation of MuSCs in vivo (29), validating the use of this quiescence-reactivation model for dissecting the molecular circuits involved in MuSC biology.

Previously, we showed that Wnt-dependent, Tcf/Lef-mediated transcription is induced in G0 myoblasts and suppressed upon reactivation (28), both for the Tcf reporter TOPflash (30) and endogenous Wnt-responsive target genes. Although these findings suggest a role for Wnt signaling in the entry of stem cells into quiescence, they appear to contradict reports that implicate Wnt in MuSC activation (21, 22, 31). Alternatively, we hypothesized that this paradox may be resolved if Wnt-Tcf/Lef signaling in quiescent cells is independent of β-catenin. Some evidence supporting this notion exists. For example, knockout of β-catenin specifically in hematopoietic stem cells does not affect hematopoiesis (32), but global knockout of Lef1 or Tcf1 in mice results in defects in hematopoiesis (33, 34). This suggests that Tcf/Lef-mediated transcriptional activity in these stem cells does not require β-catenin. In hematopoietic tumors, interaction of Tcf/Lef with activating transcription factor 2 leads to β-catenin–independent transcriptional activation (35). Thus, equating Wnt signaling with β-catenin transcriptional function is not always appropriate, and contradictory findings in MuSCs might reflect β-catenin–independent mechanisms.

Here, we examined the mechanisms associated with Tcf/Lef-mediated transcriptional activation in G0 myoblasts. We report that Tcf/Lef activation of canonical Wnt target genes in G0 myoblast cells was independent of β-catenin and upstream components of the canonical Wnt signaling pathway. We uncovered an alternative Lef1-dependent mechanism of transcriptional activation in G0—specifically that Lef1 cooperated with the TGF-β effector and transcriptional coactivator Smad3, instead of β-catenin, to induce and maintain MuSC quiescence. Smad3 and Lef1 interacted at Wnt target gene promoters in G0 in a manner that was independent of nuclear β-catenin. Overall, this study provides evidence for G0-specific cross-talk of the Wnt and TGF-β pathways at the level of transcription factors and shows a functional role for this cross-talk in the maintenance of reversible quiescence.

RESULTS

Tcf/Lef-dependent transcriptional activation in G0 (quiescent) myoblasts

Previously, we showed that moderate induction of Tcf/Lef luciferase reporter (TOPflash) activity occurs during entry into G0 (28). Endogenous Tcf/Lef target genes Axin2 and Lef1 were also significantly induced in quiescent cells (G0) compared to proliferating myoblasts and highest in differentiated myotubes, validating the reporter assays (Fig. 1, A and B). To determine whether Tcf/Lef-dependent transcriptional induction in G0 occurred in all cells or was restricted to a subpopulation of quiescent cells, we used fluorescence-based flow cytometry to sort cells expressing the TOPtRFP reporter, which is a variant of the TOPflash reporter in which turbo red fluorescent protein (tRFP) replaces luciferase (30). The Wnt pathway activator CHIR99021 robustly induced tRFP expression only in proliferating myoblasts transfected with TOPtRFP but not in those transfected with FOPtRFP (a reporter that contains a mutation in the Tcf/Lef target site; fig. S1A). After stimulation with CHIR99021, proliferating myoblasts carrying TOPtRFP sorted into a single population of low RFP fluorescence, whereas the entire population of TOPtRFP G0 cells shifted to a higher mean fluorescence intensity (Fig. 1C), which was completely reversed when the cells re-entered S phase. These population shifts were absent in control cells transfected with FOPtRFP (Fig. 1C). Thus, quiescence in myoblasts is associated with a specific, population-wide induction of Tcf/Lef-driven transcriptional activation.

Fig. 1 The Tcf/Lef transcriptional response is specific to G0 cells and does not require upstream components of canonical Wnt signaling.

(A and B) Quantification of Axin2 and Lef1 mRNAs by quantitative real-time polymerase chain reaction (qRT-PCR) in proliferating C2C12 myoblasts (MB), quiescent (G0) cells, and differentiated myotubes (MT). n = 3 independent experiments. (C) Flow cytometric analysis of tRFP intensity in cells stably expressing the Tcf/Lef reporter TOPtRFP (wild-type Tcf/Lef binding sites) or FOPtRFP (mutated Tcf/Lef binding sites) during proliferation (MB), quiescence (G0), and re-entry. n = 3 independent experiments. (D) Western blot showing endogenous Wnt-activated phosphorylation of LRP6 at Ser1490 (pLRP6) and total LRP6 (Tot LRP6) in MB, G0 cells, and MT. Glyceraldehyde-3-phosphate dehydrogenase (Gapdh) is a loading control. Molecular mass (in kilodaltons) markers are noted. Blot is representative of three independent experiments. (E) Densitometry analysis for Western blots of pLRP6 in MB, G0 cells, and MT normalized to the total Lrp6 in each cell type. n = 3 independent experiments. a.u., arbitrary units. (F) Western blot showing activated (not phosphorylated at Ser37 and Thr41) β-catenin (Act β-catenin) and total β-catenin (Tot β-catenin) in MB, G0 cells, and MT. Blot is representative of four independent experiments. (G) Densitometry analysis for Western blots of active β-catenin in MB, G0 cells, and MT normalized to the total β-catenin in each cell type. n = 4 independent experiments. (H) Immunofluorescence showing activated β-catenin in MB, G0 cells, and MT. The boxed regions indicate the areas magnified in the high-magnification images (Zoom) in which the nuclei are outlined. Nuclei are stained with 4′,6-diamidino-2-phenylindole (DAPI). Image is representative of three independent experiments. Scale bars, 10 μm. (I) Box plot showing image intensity analysis of active β-catenin immunostaining in the nuclei of MB, G0 cells, and MT. Quantification is based on measurements of >100 nuclei from randomly selected fields of three independent experiments. (J) Western blot showing β-catenin phosphorylated (pβ-catenin) at Tyr654 and total β-catenin in MB, G0 cells, and MT. Blot of representative of three independent experiments. (K) Densitometry analysis for Western blots of phosphorylated β-catenin (at Tyr654) in MB, G0 cells, and MT normalized to the total β-catenin in each cell type. n = 3 independent experiments. (L and M) ChIP–quantitative PCR (qPCR) analysis for β-catenin occupancy on the WRE region of the Axin2 (L) and Lef1 (M) promoters in MB, G0 cells, and MT. Immunoglobulin G (IgG) is a negative control. n = 3 independent experiments. Error bars represent mean ± SEM. Significance was determined by Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Low activation of canonical Wnt signaling in G0 myoblasts

To estimate overall Wnt pathway activation in different cell states, we analyzed phosphorylation of LRP6 at Ser1490 (36). The abundance of phosphorylated LRP6 was lower in G0 cells compared to proliferating myoblasts (Fig. 1, D and E), suggesting that signaling through this receptor is low in G0, although these cells exhibited downstream transcription factor (Tcf/Lef) activity. The abundance of both Wnt pathway–activated β-catenin (not phosphorylated at Ser37 and Thr41) and total β-catenin (including all phosphorylated forms and the nonphosphorylated active form) was significantly lower in G0 cells compared to proliferating myoblasts (Fig. 1, F and G), consistent with low abundance of activated LRP6. Active β-catenin showed strong nuclear localization in proliferating myoblasts and myotubes but was largely absent from nuclei of G0 myoblasts (Fig. 1, H and I), suggesting that Tcf/Lef activity in G0 does not depend on β-catenin. Phosphorylation of β-catenin at Tyr654, which induces its release from E-cadherin and nuclear import (37, 38), was significantly reduced in G0 cells compared to proliferating myoblasts or myotubes (Fig. 1, J and K), suggesting that Tcf/Lef activation in G0 is not caused by β-catenin that is released from the membrane. Chromatin immunoprecipitation (ChIP) analysis of the Wnt-responsive element (WRE) of Axin2 and Lef1, to which Tcf/Lef binds, showed that β-catenin occupancy was reduced in G0 cells and myotubes compared to proliferating myoblasts, consistent with lower abundance of β-catenin (Fig. 1, L and M). Three additional canonical Wnt target genes that are known inhibitors of myogenesis (3943) and repressed in myotubes (Msx1, Msx2, and Id2) showed reduced occupancy of β-catenin in their WRE regions, but their expression was also induced or maintained in G0 cells (fig. S1, B to E). Thus, β-catenin is unlikely to be responsible for the induction of Wnt-Tcf/Lef target gene expression in G0 cells.

β-catenin–independent functions of Tcf/Lef in G0 myoblasts

To investigate whether β-catenin has any role in Tcf/Lef transcriptional activation in G0 cells, we knocked down β-catenin using either a single small interfering RNA (siRNA) [siβ-cat(1)] or a pool of four siRNAs [siβ-cat(2)]. Despite a >50% reduction in both active and total β-catenin protein abundance (fig. S2, A to E), there was a little effect on Tcf/Lef reporter activity (Fig. 2A). Although the expression of endogenous targets (Axin2 and Lef1) was not altered in G0 cells transfected with siβ-cat(1) and only mildly reduced in G0 cells transfected with siβ-cat(2) (Fig. 2B), both target genes were strongly repressed in proliferating myoblasts treated with either siRNA (fig. S2, C and F). Unexpectedly, β-catenin knockdown increased colony formation by G0 cells upon plating and release from G0 (Fig. 2C), indicating increased self-renewal capacity (28). The minor effects of β-catenin knockdown in G0 myoblasts compared to proliferating myoblasts suggested state-specific regulation. These findings strongly suggest that β-catenin is neither the primary effector of Tcf/Lef-mediated transcriptional induction of G0 nor a key determinant of self-renewal in quiescent cells.

Fig. 2 Interaction with β-catenin is not required for Tcf/Lef-mediated transcriptional activation in G0 cells.

(A) Reporter activity in G0-arrested C2C12 cells stably transfected with the TOPflash Tcf/Lef reporter and transfected with siRNA-targeting β-catenin [siβ-cat(1) or siβ-cat(2)] or control siRNA (Scr). Values are presented as relative light units (RLU) normalized to the total protein (RLU per microgram protein). β-cat KD, β-catenin knockdown. (B) Quantification of Axin2 and Lef1 mRNAs by qRT-PCR in β-catenin knockdown and control siRNA in G0 cells. (C) Colony formation assay for β-catenin knockdown and control siRNA in G0 cells. (D) Tcf/Lef reporter activity in G0 cells transfected with siRNA targeting Lef1 [siLef1(1) or siLef1(2)] or control siRNA. (E) Quantification of Axin2 and Lef1 mRNAs by qRT-PCR in Lef1 knockdown and control siRNA in G0 cells. (F) Colony formation assay for Lef1 knockdown and control siRNA in G0 cells. (G) Tcf/Lef reporter activity in G0 cells expressing DNLef1, which lacks the β-catenin binding domain and untreated (UT). (H) Tcf/Lef reporter activity in G0 cells treated with iCRT3, which disrupts the interaction between Tcf and β-catenin or dimethyl sulfoxide [DMSO; control (Con)]. (I) Quantification of Axin2 and Lef1 mRNAs by qRT-PCR in G0 cells treated with iCRT3 or DMSO (control). (J) Colony formation assay in G0 cells treated with iCRT3 or DMSO (control). For all panels, n = 3 independent experiments. Error bars represent mean ± SEM. Significance was determined by Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001.

We also evaluated the role of the canonical β-catenin partner Lef1 in Tcf/Lef transcription in G0 myoblasts. Unlike knockdown of β-catenin, knockdown of Lef1 in G0 cells (fig. S2, G and H) resulted in substantial reductions in TOPflash activity, Axin2 expression, and colony formation (Fig. 2, D to F) without affecting overall cell viability (fig. S2I). Lef1 knockdown in proliferating myoblasts did not reduce Axin2 expression, suggesting that Lef1-dependent expression of Axin2 may be specific to G0 cells (fig. S2J). Together, these findings show that Lef1 promotes activation of Tcf/Lef target genes in G0 and does so potentially independently of β-catenin.

Because perturbing β-catenin or Lef1 individually had different effects on Tcf/Lef target gene expression in G0 cells, we addressed whether the interaction between Tcf/Lef and β-catenin played any role in Tcf target activation in G0. First, we perturbed the interaction between Tcf and β-catenin by expressing a dominant-negative form of Lef1 (DNLef1) that lacks the N-terminal β-catenin binding domain in G0 cells and found no significant reduction in TOPflash activity (Fig. 2G), strongly suggesting that β-catenin interaction was not required for Tcf/Lef-mediated transcriptional activation. Second, we directly perturbed β-catenin–Tcf/Lef interaction in G0 cells by treating them with the drug iCRT3 (inhibitor of β-catenin responsive transcription), which disrupts the interaction between Tcf and β-catenin (44). In myoblasts, iCRT3 suppressed Tcf/Lef-dependent TOPflash activation after stimulation with recombinant mouse Wnt3a in a dose-dependent manner (fig. S2K), confirming the importance of the interaction between Tcf and β-catenin in proliferating cells. However, in G0 myoblasts, iCRT3 treatment did not reduce TOPflash activity (Fig. 2H). Axin2 and Lef1 were not decreased significantly by low-dose (5 μM) iCRT3 treatment, but high doses (25 μM) induced Axin2 mRNA in G0 cells (Fig. 2I). As in β-catenin knockdown cells, iCRT3 treatment of G0 cells did not reduce but mildly increased colony formation (Fig. 2J and fig. S2L). We hypothesized that the unexpected increase in Axin2 expression and self-renewal upon disruption of the β-catenin–Tcf/Lef complex might reflect the interaction of Tcf/Lef with transcriptional effectors other than β-catenin. Together, these results show that the β-catenin–Tcf/Lef interaction is dispensable for Tcf/Lef-mediated transcription in quiescent cells and for the ability of quiescent cells to self-renew after release from G0. Thus, the quiescent state involves β-catenin–independent regulation of Tcf/Lef functions.

Altered balance of activating and repressive Tcf/Lef proteins in G0 myoblasts

The four Tcf/Lef proteins—Tcf 7 (also known as Tcf1), Tcf  7l1 (also known as Tcf3), Tcf   7l2 (also known as Tcf4), and Lef1—have diverse functions (45). Lef1 is always associated with target gene activation (33, 46), whereas Tcf  7l1 is generally repressive (4750) and derepressed by Wnt–β-catenin signaling (48). The other two Tcfs, Tcf   7l2 and Tcf  7, can act as activators or repressors depending on the context (51). Because we found that β-catenin is not involved in Tcf/Lef function in G0 myoblasts, we explored the possibility that an altered ratio of Tcf/Lef protein abundance may contribute to the altered WRE-dependent gene expression in G0. The abundances of Lef1 and Tcf 7 proteins were mildly increased in G0 (Fig. 3, A and B, and fig. S3, A and B), but the abundances of Tcf 7l1 and Tcf 7l2 were reduced (Fig. 3, C and D, and fig. S3, C and D). Myotubes produce two 55- to 60-kDa isoforms of Tcf   7l1 but not the ~75-kDa isoform seen in proliferating myoblasts (Fig. 3, C and D). Because Tcf  7 and Tcf  7l2 were not enriched at Axin2 and Lef1 WREs in G0 cells (fig. S3, E and F), we focused on Lef1 and Tcf   7l1. It is possible that a reduction in the amount of repressive Tcf   7l1 in G0 cells might contribute to derepression of Tcf/Lef targets. Lef1 was more abundant than Tcf  7l1 in G0 cells and myotubes, whereas similar amounts of both proteins were seen in proliferating myoblasts (Fig. 3, E and F). Lef1 and Tcf   7l1 localized to nuclear domains in myotubes that appeared to be mutually exclusive (Fig. 3E). Thus, relative reduction of repressive factor (Tcf      7l1) with persistent nuclear abundance of an activating factor (Lef1) may also contribute to Tcf/Lef-dependent transcriptional activity in G0 cells.

Fig. 3 Expression of distinct Tcf/Lef family members affects Wnt-Tcf transcriptional activation in G0.

(A) Western blot showing the transcriptional activator Lef1 in C2C12 MB, G0 cells, and MT. Gapdh is a loading control. Migration of molecular mass markers is noted (in kilodaltons). Blots are representative of four independent experiments. (B) Densitometry analysis for Western blots of Lef1 in different cellular states normalized to Gapdh in each cell type. n = 4 independent experiments. (C) Western blot showing the transcriptional repressor Tcf7l1 in MB, G0 cells, and MT. Two short isoforms of Tcf7l1 present in MT are indicated with arrows. Blots are representative of four independent experiments. (D) Densitometry analysis for Western blots of Tcf7l1, both upper (75 kDa) and lower (55 kDa) bands normalized with Gapdh in MB, G0 cells, and MT. n = 4 independent experiments. (E) Immunofluorescence showing Lef1 and Tcf7l1 in MB, G0 cells, and MT. The boxed regions indicate the areas magnified in the high-magnification (Zoom) images showing nuclear localization of Lef1 (white arrows) and Tcf7l1 (yellow arrows). Nuclei are outlined with dotted lines. Images are representative of three independent experiments. Scale bars, 10 μm. (F) Quantification of image intensity of Lef1 and Tcf7l1 immunostaining in nuclei of MB, G0 cells, and MT. Quantification is based on measurements of >100 cells from randomly selected fields of three independent experiments. (G to J) ChIP-qPCR analysis showing Lef1 (E and F) and Tcf7l1 (G and H) occupancy on the WRE region of the Axin2 and Lef1 genes in MB, G0 cells, and MT. IgG is a negative control. n = 3 independent experiments. Error bars represent mean ± SEM. Significance was determined by Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001.

Because knockdown of Lef1 reduced transcriptional activation in G0 cells (Fig. 2, D to F), we evaluated whether differential abundance of active and repressive Tcfs was reflected in altered WRE occupancy. Consistent with their abundance, Lef1 occupancy on WREs of all target genes was always higher than Tcf   7l1 occupancy on WREs in G0 cells compared to proliferating myoblasts (Fig. 3, G to J, and fig. S4, A to F). We conclude that strong WRE occupancy of Lef1 and reduced occupancy of Tcf  7l1 in G0 myoblasts may contribute to induction or sustained expression of these targets, even when strong activators such as β-catenin are absent.

Contribution of TGF-β effectors to Tcf/Lef function in G0 myoblasts

The results thus far suggest that Wnt signaling through LRP5/6 and β-catenin contributes little to Tcf/Lef-mediated transcriptional activation in quiescent myoblasts. Because Wnt–β-catenin signaling cross-talks with other pathways in a context-dependent manner (52, 53), alternate effector molecules might be involved in G0. We focused on the TGF-β pathway, which is up-regulated in G0 (28), consistent with its known role in cell cycle arrest (54). The Wnt pathway components Axin2 and GSK3β interact with Smad3, a transcriptional effector of TGF-β signaling (55). Smad2 and Smad3 can interact with Tcf/Lef proteins to induce transcriptional activation independent of β-catenin (56).

To determine whether TGF-β signaling plays a role in Tcf/Lef-mediated target gene activation, we used the compound SB431542, which blocks phosphorylation of Smad2 and Smad3 by the TGF-β type 1 receptor, thus preventing nuclear translocation of these Smads (57). SB431542 treatment of G0 myoblasts significantly reduced Tcf/Lef reporter activity, endogenous Axin2 and Lef1 expression, and colony formation (Fig. 4, A to C, and fig. S5A). These observations suggest that the TGF-β pathway contributes to Wnt effector activity in G0 cells by promoting Tcf/Lef-dependent transcription.

Fig. 4 The TGF-β pathway plays a role in G0-specific Tcf/Lef-mediated transcriptional activation.

(A) Tcf/Lef reporter (TOPflash) activity in G0 cells treated with the TGF-β receptor inhibitor SB431542 (SB) or DMSO (control). Values are presented as RLU normalized to total protein (RLU per microgram protein). n = 3 independent experiments. (B) Colony formation assay of G0 cells treated with the indicated concentrations of SB or DMSO (control). n = 3 independent experiments. (C) Quantification of Axin2 and Lef1 mRNAs by qRT-PCR in G0 cells treated with SB or DMSO (control). n = 3 independent experiments. (D) Western blot showing Smad3 phosphorylated at Ser423 and Ser425 (pSmad3) and total Smad3 (Tot Smad3) in G0 cells treated with SB or DMSO (control). Gapdh is a loading control. (E) Densitometry analysis for Western blots of pSmad3 normalized to total Smad3 in SB-treated G0 cells. n = 3 independent experiments. (F) Western blot showing Smad2 phosphorylated at Ser462 (pSmad2) and total Smad2 (Tot Smad2) in G0 cells treated with SB or DMSO (control). (G) Densitometry analysis for Western blots of pSmad2 normalized to total Smad2 in SB-treated G0 cells. n = 3 independent experiments. (H) Western blot showing pSmad3 and total Smad3 in MB, G0 cells, and MT. Blot is representative of three independent experiments. (I) Densitometry analysis for Western blots of pSmad3 in MB, G0 cells, and MT normalized to the total Smad3. n = 3 independent experiments. (J) Immunofluorescence showing pSmad3 in MB, G0 cells, and MT. The boxed regions are magnified in the images on the right (Zoom), with nuclei outlined. Images are representative of three independent experiments. Scale bars, 10 μm. (K) Box plot showing image intensity analysis of pSmad3 immunostaining of nuclei in MB, G0 cells, and MT. Quantification is based on measurements of >100 nuclei from randomly selected fields of three independent experiments. Error bars represent mean ± SEM. Significance was determined by Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001.

We confirmed that SB431542 treatment significantly reduced Smad3 phosphorylation in G0 myoblasts but had no effect on Smad2 phosphorylation (Fig. 4, D to G). Thus, the TGF-β pathway may function through Smad3 and not Smad2 in G0 myoblasts. Further, the abundance and nuclear localization of phosphorylated Smad3 (pSmad3) were sustained in G0 cells (Fig. 4, H to K), consistent with a transcriptional role.

Switch of Lef1 interaction from β-catenin to Smad3 in G0 myoblasts

To determine the role of Smad3 in Tcf/Lef function, we used the compound SIS3 (specific inhibitor of smad3), which inhibits TGF-β1–induced phosphorylation of Smad3 and thereby blocks Smad3 transcriptional activity (58). SIS3 treatment not only reduced Smad3 phosphorylation in G0 cells (fig. S5, B and C) but also reduced TOPflash activity, reduced endogenous Axin2 and Lef1 expression, and induced a dose-dependent suppression of colony formation (Fig. 5, A to C, and fig. S5D). Similar to Lef1 knockdown, Smad3 knockdown (fig. S5, E to G) suppressed Tcf/Lef reporter activity, colony formation, and endogenous Axin2 and Lef1 expression in G0 cells (Fig. 5, D to F, and fig. S5H). Smad3 knockdown did not affect Axin2 and Lef1 expression in proliferating myoblasts, suggesting that Smad3 is involved in Tcf/Lef-mediated transcriptional activation only in G0 cells (fig. S5I). Collectively, these observations confirm a role for the TGF-β transcriptional effector Smad3 in Tcf/Lef-dependent transcriptional activity and self-renewal in G0 myoblasts.

Fig. 5 The TGFβ effector Smad3 is a key player in Tcf/Lef-mediated transcriptional activation in G0 cells.

(A) Tcf/Lef reporter (TOPflash) activity in G0 C1C12 cells treated with the indicated concentrations of the Smad3-specific inhibitor SIS3 or DMSO (control). Values presented as RLU normalized to total protein (RLU per microgram protein). n = 3 independent experiments. (B) Colony formation assay for G0 cells treated with SIS3 or DMSO (control). n = 3 independent experiments. (C) Quantification of Axin2 and Lef1 mRNAs by qRT-PCR in G0 cells treated with SIS3 or DMSO (control). n = 3 independent experiments. (D) Tcf/Lef reporter (TOPflash) activity in G0 cells transfected with siRNA targeting Smad3 [siSmad3(1) or siSmad3(2)] or control siRNA. n = 3 independent experiments. (E) Colony formation assay for Smad3 knockdowns G0 cells. n = 3 independent experiments. (F) Quantification of Axin2, Lef1, and Smad3 mRNAs by qRT-PCR in Smad3 knockdown G0 cells. n = 3 independent experiments. (G and H) ChIP-qPCR analysis for pSmad3 occupancy on the WRE region of Axin2 (G) and Lef1 (H) in MB, G0 cells, and MT. n = 3 independent experiments. (I) Western blot (WB) showing total and active β-catenin, pSmad3, and Lef1 in pSmad3 and Lef1 immunoprecipitates (IP) from nuclear lysates of MB and G0 cells. Immunoprecipitation with IgG is a negative control. n = 3 β-catenin. (J to M) ChIP-qPCR analysis for pSmad3 (J and L) and Lef1 (K and M) occupancy on the WRE region of Axin2 and Lef1 in G0 cells treated with siRNA targeting Lef1 (siLef1) or a control siRNA. n = 3 independent experiments. (N to Q) ChIP-qPCR analysis for Lef1 (N and P) and pSmad3 (O and Q) occupancy on the WRE region of Axin2 and Lef1 in G0 cells treated with siRNA targeting Smad3 (siSmad3) or a control siRNA. n = 3 independent experiments. Error bars represent mean ± SEM. Significance was determined by Student’s t test. *P < 0.05, **P < 0.01, and ***P < 0.001.

We used ChIP analysis to evaluate whether pSmad3 is associated with Tcf/Lef factors on target promoters in G0 cells. Phosphorylated Smad3 was enriched on the WREs of all target genes tested (Axin1, Lef1, Msx1, Msx2, and Id2), both in G0 cells and myotubes compared to proliferating myoblasts (Fig. 5, G and H, and fig. S5, J to L). These same WREs were also occupied by Lef1, which activates transcription, and most showed reduced association with the repressor Tcf    7l1 (Fig. 3, G to J, and fig. S4, A to F), whereas β-catenin occupancy was minimal in G0 cells compared to proliferating myoblasts (Fig. 1, L and M, and fig. S1, C to E). To evaluate whether Lef1 and pSmad3 interacted directly, we coimmunoprecipitated Lef1 from nuclear fractions of lysates from proliferating myoblasts and G0 myoblasts. Whereas Lef1 complexes in proliferating myoblasts, as expected, contained β-catenin, Lef1 complexes in G0 cells did not (Fig. 5I). In contrast, Lef1 did not interact with pSmad3 in proliferating myoblasts but did in G0 cells, in agreement with the ChIP analysis. In the reciprocal experiment, Lef1 immunoprecipitated with Smad3 from nuclear extracts from G0 cells but not from proliferating myoblasts.

To evaluate whether Smad3 occupancy on WREs depended on interaction with Lef1, we performed ChIP analysis in Lef1 knockdown G0 cells. Knocking down Lef1 reduced both Lef1 and pSmad3 occupancy at WREs in Axin2, Lef1, Msx1, Msx2, and Id2 (Fig. 5, J to M, and fig. S6, A to F). This reduction was WRE-specific because pSmad3 occupancy of the Smad-binding element (SBE) in the TGF-β target genes Smad7 and Pai1 was not affected by Lef1 knockdown (fig. S6, G to J). Reduction of Smad3 occupancy was not due to a reduction in the abundance of pSmad3 in Lef1 knockdown G0 cells (fig. S6, K to N). Thus, the occupancy of WREs by pSmad3 depends on the presence of Lef1 at the same promoters in G0 myoblasts.

To determine whether pSmad3 was required for Lef1 occupancy of WREs of target genes, we performed ChIP analysis in Smad3 knockdown G0 cells. Smad3 knockdown significantly reduced Lef1 occupancy on WREs (Fig. 5, N to Q, and fig. S7, A to F) but did not affect Lef1 occupancy of SBEs of Pai1 and Smad7 (fig. S7, G to J). Reduced Lef1 mRNA expression (Fig. 5F) and Lef1 protein abundance (fig. S7, K to M) in G0 are consistent with a role for Smad3 occupancy in Lef1-mediated transcriptional stimulation at WREs during the induction and maintenance of quiescence. In summary, reduced promoter occupancy by Lef1 in Smad3 knockdown cells is likely indirect through reduction of Lef1 protein abundance, but reduced promoter occupancy by Smad3 in Lef1 knockdown is likely direct due to loss of interaction with Lef1 at the promoter. Therefore, noncanonical activation of Tcf/Lef target genes in G0 myoblasts can be attributed to a direct interaction between Lef1 and pSmad3, accounting for β-catenin independence of target gene activation, exclusively in quiescent cells.

Involvement of Smad3 but not other canonical Smads in Tcf/Lef activity in G0 myoblasts

Upon TGF-β pathway activation, Smad3 complexes with Smad4 and pSmad2 and translocates into the nucleus, where the complex activates the expression of TGF-β target genes (59). However, Smad4 and pSmad2 proteins showed reduced abundance and nuclear localization in G0 cells compared to proliferating myoblasts and myotubes (fig. S8, A to H), and knockdown of either Smad2 or Smad4 did not reduce Tcf/Lef reporter activity and target gene activation in G0 cells (fig. S9, A to G). Unlike Smad3, Smad2 and Smad4 were not significantly enriched at WREs in G0 cells compared to proliferating myoblasts (fig. S9, H to K), but Smad4 and pSmad2 occupied SBEs in the canonical TGF-β promoters of Smad7 and Pai1 in G0 cells (fig. S9, L to O). As in quiescent cultured myoblasts (fig. S8, A to H), quiescent MuSCs in muscle fibers from the extensor digitorum longus (EDL) muscle of adult Pax7-nGFP mice, which express nuclear-localized green fluorescent protein (GFP) under the control of the Pax7 promoter that drives expression in MuSCs (60), did not show nuclear enrichment of either Smad2 or Smad4 (fig. S9, P and Q). Together, these data suggest that pSmad3, but not other canonical Smads (Smad2 and Smad4), plays a role in Tcf/Lef target gene activation in G0 myoblasts.

Wnt and TGF-β transcription factors in quiescent and activated MuSCs

To determine whether Wnt and TGF-β effector transcription factors were present in MuSCs, we analyzed isolated single myofibers from the EDL muscles of adult Pax7-nGFP mice. As previously reported (21), we did not observe active β-catenin in Pax7+ MuSCs (Fig. 6A). Similar to G0 myoblasts in culture (Figs. 3, A to F, and 4, H to K), MuSCs were devoid of Tcf7l1 (Fig. 6B) but contained both Lef1 and pSmad3 (Fig. 6, C and D). This correlation between quiescence in vitro and ex vivo strongly suggests that the interaction between Lef1 and pSmad3 may occur in MuSC in vivo in the absence of active β-catenin and Tcf7l1.

Fig. 6 Lef1-Smad3 may interact in quiescent MuSCs (satellite cells) in vivo.

(A to D) Immunofluorescence showing active β-catenin (A), Tcf7l1 (B), Lef1 (C), and pSmad3 (D) in MuSCs (arrow) and myonuclei (asterisks) of a single muscle fiber from the EDL of adult Pax7-nGFP mice, which express nuclear GFP in MuSCs. Images are representative of three animals. Scale bars, 20 μm. (E) Immunofluorescence showing active β-catenin, Lef1, and pSmad3 in MuSCs (white arrowheads) of single prequiescent (1-week postnatal), quiescent (8-week postnatal), and activated (8-week postnatal muscle cultured for 24 and 48 hours) EDL muscle fibers from Pax7-nGFP mice. Activated muscle fibers were cultured in the presence of EdU to label proliferating cells. Insets show the MuSC nuclei (DAPI, blue), Pax7 (green), and EdU (yellow) in the same field of view as the large image. n = 3 independent experiments, each using four mice to obtain sufficient fibers for each experiment. Scale bars, 5 μm. (F) Box plot showing fluorescence intensity analysis of active β-catenin, Lef1, and pSmad3 in nuclei of prequiescent (PreQ), quiescent (0H), and activated (24H and 48H) MuSCs from experiment in (E). Fifty to 100 cells were analyzed in each sample from three biological replicates. (G) Percentage of MuSCs on single EDL fibers that were Pax7+MyoDEdU after 3 days of culture in proliferation medium, followed by treatment with the Smad3-specific inhibitor SIS3, the TGF-β receptor inhibitor SB, the Tcf–β-catenin interaction disruptor iCRT3, or DMSO (control). n = 3 independent experiments. Three mice were used to obtain sufficient fibers for each experiment. More than 200 cells from fibers were analyzed in each treatment condition. Each experiment was normalized to the number of MuSCs scoring positive in control samples. More than 200 cells were analyzed per condition. Error bars represent mean ± SEM. Significance was determined by Student’s t test. *P < 0.05. n = 3 independent experiments. (H) Western blot showing total and active β-catenin, pSmad3, and Lef1 in Lef1 immunoprecipitates from nuclear lysates of C2C12 MB treated with iCRT3 or DMSO (control). Blot is representative of three independent experiments. (I) Model illustrates Lef1 switching partners from β-catenin to pSmad3 during the switch from the proliferative to the quiescent state. APC, adenomatous polyposis coli; CK1, casein kinase 1; TGF-βR, transforming growth factor–β receptor.

To determine the dynamics of these transcription factors during quiescence and activation, we analyzed MuSCs on isolated myofibers ex vivo at two stages of postnatal development in Pax7-nGFP mice. To identify MuSCs that had not yet entered quiescence, we labeled proliferating cells by injecting the thymidine analog 5-ethynyl-2′-deoxyuridine (EdU) into 7-day-old mice when MuSCs are still actively proliferating (61). To identify quiescent MuSCs, we isolated myofibers from adult mice (8 to 12 weeks of age) and either fixed them immediately or cultured them for 24 and 48 hours in the presence of EdU to detect activated MuSCs. Both prequiescent MuSCs and activated MuSCs were Pax7+ and incorporated EdU, whereas quiescent MuSCs were Pax7+ but did not incorporate EdU. Active β-catenin was more abundant in proliferating (both prequiescent and activated) MuSCs than in quiescent MuSCs (Fig. 6, E and F), as previously reported (21). In contrast, the abundances of both Lef1 and pSmad3 were higher in quiescent MuSCs than in either prequiescent or activated MuSCs (Fig. 6, E and F). These observations suggest that Lef1 has a higher chance of interacting with pSmad3 in quiescent MuSCs, where little active β-catenin is present, than in proliferating cells. In cycling (prequiescent and activated) MuSCs, Lef1 may preferentially interact with active β-catenin, which is abundant in these cell types. Together, these observations in MuSCs in vivo support our findings in cultured myoblasts and suggest a switch in Tcf/Lef target gene activation during quiescence, resulting from altered interactions of transcriptional coactivators.

Importance of Smad3-Lef1 for satellite cell self-renewal

In fibers from EDL muscles that are cultured for 3 days, some MuSC re-enter quiescence by reactivating Pax7 expression and repressing both MyoD expression and the cell cycle (62). To probe the importance of co-occurrence of pSmad3 and Lef1 in satellite cells, we analyzed self-renewal ex vivo in EDL explants. Three days after isolation, myofiber cultures were treated with SB431542, SIS3, or iCRT3 for 24 hours and then pulsed with EdU for 1 hour before fixation. Perturbation of TGF-β signaling by SB431542 and SIS3 significantly reduced the frequency of self-renewing Pax7+MyoDEdU MuSCs (Fig. 6G). This finding is consistent with the C2C12 colony formation assay and suggests that reduced signaling through TGF-β and Smad3 compromises the ability of quiescent cells to self-renew. Perturbation of the interaction between Lef1 and β-catenin by iCRT3 significantly increased the number of MuSCs (Fig. 6G), in agreement with our hypothesis and the increased colony formation observed in vitro with iCRT3 treatment (Fig. 2J) or upon β-catenin knockdown (Fig. 2C) in G0 C2C12 cells. In pulldown experiments from iCRT3-treated C2C12 myoblasts, Lef1 and pSmad3 showed a stronger interaction compared to untreated cells (Fig. 6H and fig. S10, A to C). This suggests that Lef1 released from β-catenin by iCRT3 treatment is available to interact with Smad3, which may account for both the enhanced frequency of quiescent cells and the increased abundance of Lef1-Smad3 complexes, leading to higher Axin2 expression in iCRT3-treated G0 cells (Fig. 2G). Together, these experiments suggest that the Lef1-Smad3 interaction is important for quiescence and self-renewal of MuSCs.

DISCUSSION

We provide evidence of cross-talk between the Wnt and TGF-β pathways at the level of transcriptional regulation in quiescent self-renewing myoblasts in culture and in muscle satellite cells (MuSCs) ex vivo. Our study was aimed at resolving conflicting results on the role of Wnt signaling in nonproliferating, quiescent muscle cells. We demonstrated that the canonical Wnt effector β-catenin did not participate in quiescence-dependent Tcf/Lef gene activation, but its role as a cofactor of Lef1 in this context was fulfilled by the TGF-β signaling effector Smad3. The replacement of β-catenin with Smad3 in Tcf/Lef-dependent transcriptional regulation demonstrates that, even in a single cell type, signaling components from different pathways may cross-talk to provide a stage-specific outcome.

Canonical Wnt signaling and target gene activation have been implicated in the maintenance of quiescence in some adult stem cells (16, 63, 64), but little direct evidence supports a role in canonical Wnt signaling in nonproliferating MuSCs. Our previous results revealed a distinct Wnt signature with reversible quiescence-specific Tcf-Lef1 activation in G0 myoblasts (28). From our analysis, four lines of evidence argue against β-catenin in Tcf/Lef target gene activation in G0 cells. First, there was reduced activation of upstream Wnt signaling events (LRP6 phosphorylation, β-catenin phosphorylation at Tyr654, activation of β-catenin, and nuclear-localized β-catenin) in G0 cells. Second, strong reduction of β-catenin by siRNA-mediated knockdown or direct perturbation of β-catenin–Tcf/Lef interaction by iCRT3 or DNLef1 in G0 had a little effect on Tcf/Lef activity, target gene activation, and self-renewal. Third, there was negligible occupancy of β-catenin on WREs of Tcf/Lef targets with expression that is maintained or enhanced in G0. Finally, Lef1 protein interacted with active β-catenin in proliferating myoblasts but not in G0 cells. These data suggest that β-catenin has little, if any, role in self-renewal and maintenance of quiescence in myoblasts. Our findings support the observations that β-catenin knockout does not alter MuSC number or quiescence in vivo (2325). Rather, a moderate threshold of β-catenin is required to complete muscle regeneration because mice overexpressing or lacking β-catenin show premature or delayed muscle regeneration, respectively (24, 25). β-catenin–mediated transcriptional activity is also not crucial for self-renewal and pluripotency in ESCs but is required for their differentiation (6568). Therefore, we suggest that the transcriptional role of β-catenin is minor in cellular states with a high capacity for self-renewal: the ground state of ESCs and the quiescent state of adult stem cells such as MuSCs.

Because enhanced Wnt target gene expression in G0 did not depend on β-catenin, Lef1 emerges as the best candidate to mediate this transcriptional response. On the basis of abundance, nuclear localization, and WRE association, Lef1 is mainly associated with target gene activation (33, 69), whereas Tcf  7l1 is usually repressive (4750). We observed higher abundance of Lef1 than Tcf  7l1 in G0, mirrored by higher WRE occupancy. Lef1 induces higher Tcf/Lef reporter activation than does Tcf  7l1 (70, 71). Myotubes expressed smaller isoforms of Tcf  7l1 compared to proliferating and quiescent myoblasts, and different nuclei in individual myotubes displayed mutually exclusive localization with Lef1, suggesting heterogeneity in the syncytium with respect to Wnt effectors. Myotube nuclei displaying high Lef1 and low Tcf  7l1 might be responsible for the observed Tcf/Lef reporter induction. However, in G0 cells, the dominance of active Lef1 over repressive Tcf  7l1 might permit higher basal activation of Tcf/Lef promoters independently of the strong activator β-catenin.

Our study suggests a mechanism for β-catenin–independent Tcf/Lef activity in G0, implicating the TGF-β effector Smad3 and providing evidence that expression of WRE-containing target genes is not inevitably an indicator of β-catenin–mediated transcription, at least in muscle cells. The TGF-β and Wnt pathways interact in diverse physiological contexts (72) at multiple levels, including reciprocal regulation of ligand secretion, shared transcriptional targets of Smad2, Smad3, and β-catenin, and shared cytoplasmic regulators (73). TGF-β signaling participates in self-renewal and differentiation of ESCs and adult stem cells (74) and is critical for neural stem cell quiescence (75). We now report cooperation of TGF-β with Wnt pathway in muscle MuSC quiescence. Here, two major findings connect TGF-β–Smad3 signaling to Tcf/Lef function: (i) Phosphorylated Smad3 was present in G0 nuclei, and the Smad inhibitor SB431542 reduced Tcf/Lef function and self-renewal; (ii) direct Smad3 perturbation by SIS3 or siRNA reduced Tcf/Lef activity and self-renewal. We also showed that, unlike Smad3, Smad4 and Smad2 did not play any statistically significant role in β-catenin–independent Tcf/Lef activation. Similarly, Smad3 contributes more to TGF-β–induced cell cycle arrest than do Smad2 and Smad4 in keratinocytes (76), suggesting that Smad4 is not a universal effector of TGF-β signaling (7779). In addition, MuSCs had lower abundance of Smad4 and pSmad2 proteins, consistent with the finding that MuSC–specific knockout of Smad4 does not alter MuSC number (80). Thus, the G0-specific Lef1-Smad3 transcription complex may lack Smad2 and Smad4.

The phenotypes resulting from Smad3 knockdown resemble Lef1 knockdown, indicating convergence in transcriptional activation. Physical interaction of Smad3 and Lef1 has been reported in other systems (56, 81, 82), and we found that Lef1 physically interacted with pSmad3 and not with active β-catenin and that this noncanonical interaction was absent in cycling myoblasts. Thus, the interaction preference of Lef1 may switch from active β-catenin to pSmad3 when cells switch from proliferation to quiescence. In support of their interaction in G0, both Smad3 and Lef1 were enriched on Tcf/Lef target genes, whereas β-catenin occupancy of these promoters was lacking. Promoter-specific Smad3 occupancy on WREs, but not on SBEs, in G0 cells depended on Lef1, and Smad3 and Lef1 knockdowns reciprocally reduced one another’s occupancy of WREs and target gene expression specifically in G0 but not in myoblasts, suggesting a stage-specific switch in the mechanism of transcriptional activation. Collectively, this evidence suggests that, when β-catenin abundance is low, Lef1 either recruits or stabilizes Smad3 occupancy on WREs, which cooperatively induces Tcf/Lef target genes.

Our study also supports the possibility that Lef1 switches partners in MuSCs ex vivo as it does in cultured myoblasts. Active β-catenin was not detected in quiescent MuSCs, which had greater abundance of both Lef1 and Smad3 than did activated MuSCs, and Tcf7l1 was absent, increasing the likelihood that transcriptionally activating Lef1-Smad3 complexes would form instead of repressive Tcf7l1-Smad3 complexes. When MuSCs are activated by damage, β-catenin is also activated; therefore, Lef1 can interact with active β-catenin in this context. Blocking TGF-β–Smad3 signaling reduced the self-renewing MuSC population, supporting a role for this pathway in the control of quiescence. TGF-β signaling is implicated in MuSC maintenance (83, 84), and Smad3-null MuSCs show poor self-renewal (85). We found that the Smad3-Lef1 interaction was important for MuSC self-renewal ex vivo and that enhancing Smad3-Lef1 interaction by perturbing β-catenin–Lef1 interaction with iCRT3 enhanced the number of self-renewing MuSCs, consistent with results in mouse ESCs (68).

Overall, these data support a model for signal-dependent transcription factor partner switching wherein Lef1 changes partners from β-catenin to Smad3 to control MuSC quiescence (Fig. 6I). This interaction is likely to activate specific target genes including inhibitors of differentiation, cell cycle repressors, and inhibitors of apoptosis, all of which are key components of the quiescence program (28). Transcription factors function in a spatiotemporal context, permitting cell type– or cell state–specific gene activation. In myoblasts, pro–B cells, and ESCs, co-occupancy of Smad3 with lineage-specific transcription factors determines the distinct effects of TGF-β in each cell type (86). Transcription factors may also mark genes for anticipatory activation in a signal-dependent fashion (87). We speculate that in the absence of β-catenin, Lef1 might mark the promoters of Wnt pathway genes in muscle cells for Smad3 binding, thereby integrating Wnt and TGF-β signaling.

Strong Wnt–β-catenin activation in MuSCs leads to precocious differentiation and death (24, 31), suggesting that a threshold amount of β-catenin is required for target activation (24, 25). Our data support a role for β-catenin in proliferating, but not G0, cells. The low abundance of active β-catenin in MuSCs may reflect distinct scenarios. First, because Tcf7l1 is not abundant, β-catenin may not be required to release this repressor from target promoters (88, 89). Second, when β-catenin is low, Lef1-Smad3 complexes might activate a unique set of Tcf/Lef targets necessary for maintenance of the G0 state (Fig. 6I). We speculate that quiescence may be maintained by a distinct threshold of Tcf/Lef target activation that is stimulated by TGF-β–Smad3 signaling independently of β-catenin. In summary, our study reveals a previously unrecognized stage-specific mechanism of Tcf/Lef target gene activation by Lef1-Smad3 interaction in muscle, which is independent of the Wnt–β-catenin pathway and implicates the TGF-β pathway in activating Tcf/Lef-dependent transcription in quiescence.

MATERIALS AND METHODS

Inhibitors and antibodies

Drugs SB431542 and SIS3 were all from Sigma-Aldrich. CHIR99021 drug was from StemRD. iCRT3 was prepared in-house (44). Recombinant mouse Wnt3A was from R&D Systems. Antibodies recognizing β-catenin phosphorylated at Tyr654 (ab24295), Lef1 (ab85052), Tcf7 (ab71251), Tcf7l1 (ab86175), Tcf7l2 (Tcf4, ab32873), pSmad3 (ab51451), Smad4 (ab208804), total Smad3 (ab28379), and GAPDH (ab9484) were from Abcam. Antibodies recognizing phosphorylated LRP6 (2568), total LRP6 (3395), pSmad2 (3108), and total Smad2 (5339) were from Cell Signaling Technology. Antibodies recognizing active β-catenin (clone e7, 05-665) was from Upstate-Millipore, total β-catenin (610154) was from BD Transduction Laboratories, total β-catenin (C2206) was from Sigma-Aldrich, Tcf7l1 (Tcf3, M-20, Sc-8634) was from Santa Cruz Biotechnology, Pax7 monoclonal antibody supernatant was from Developmental Studies Hybridoma Bank (AB_528428), and polyclonal Pax7 (ARP32742) from Aviva Systems Biology. Mouse and rabbit IgG were obtained from Calbiochem.

Plasmids

The Tcf/Lef reporter plasmid Super 8x TOPflash was a gift from R. T. Moon (30). TOPtRFP (multimerized Tcf/Lef binding site) and FOPtRFP (mutated Tcf/Lef binding site) plasmids were generated from TOPflash and FOPflash constructs by removing luciferase using restriction digestion with Hind III and Hpa I and ligating the plasmid with PCR-amplified tRFP fragment flanked by the same restriction sites. 67DNLEF1 was a gift from M. L. Waterman (University of California, Irvine).

Cell culture

C2C12A2 (27), a subclone of the C2C12 skeletal muscle myoblast cell line (90, 91), was used in this study. Cells were maintained as proliferating myoblast cultures in growth medium [high-glucose Dulbecco’s modified Eagle’s medium (DMEM) and 20% fetal bovine serum (FBS); Invitrogen]. For differentiation to myotubes, near-confluent cultures were transferred into low-serum conditions (2% horse serum) for 3 to 5 days. For synchronizing cells in the G0 state, suspension culture (26) was used with some modifications. Briefly, C2C12 cells were trypsinized, resuspended in DMEM containing 1.3% methyl cellulose, 20% FBS, 4 mM glutamine, antibiotics, and 10 mM Hepes (pH 7.4) to a final density of 105 cells/ml, and mixed thoroughly. All cells arrested in G0 by 48 hours in suspension culture, and >99% of cells re-enter the cell cycle upon replating using this method (27).

Knockdown analysis

All knockdown studies were performed in proliferating myoblasts because G0 cells in methylcellulose cannot be efficiently transfected. Lipid-mediated transfection of myoblasts with 50 pmol of siRNAs targeting β-catenin (Silencer, Ambion, and siGENOME SMARTpool, Dharmacon), Smad2 and Smad4 (siGENOME SMARTpool, Dharmacon), or Lef1, Smad3 (ON-TARGETplus siRNA, Dharmacon), or control siRNA (Ambion, Dharmacon) was performed using Lipofectamine RNAiMAX (Thermo Fisher Scientific), as described (92). Twenty-four hours after transfection, cells were cultured in suspension for a further 48 hours to induce G0 arrest.

TOPflash reporter assay

Stable cell lines derived from single clones of TOPflash (28) in C2C12 cells were used for testing Wnt signaling induction in different cellular states. For reporter analysis in knockdown cells, TOPflash cells were first transfected with specific siRNA in myoblasts using Lipofectamine RNAiMAX. After confirming knockdown 12 to 24 hours later (by analyzing RNA and protein using real-time PCR and Western blotting, respectively), cells were cultured in suspension for 48 hours to induce G0 arrest. For iCRT3 validation, TOPflash stable cells were treated with rmWnt3a (5 ng/ml) and iCRT3 (5, 50, or 100 μM) for 24 hours. Luciferase assay was performed as per the manufacturer’s protocols (Roche) on cells recovered from suspension culture. Protein quantification was performed using BCA (bicinchoninic acid) assay kit (Thermo Fisher Scientific), and RLU measured in a luminometer (Sirius) were normalized to the total amount of protein in the lysate (RLU per microgram protein).

RNA isolation and real-time PCR

RNA was isolated using TRIzol (Invitrogen), and complementary DNA (cDNA) was prepared from equal amounts of RNA using SuperScript III cDNA synthesis kit (Invitrogen). qRT-PCR was performed on an ABI 7700HT thermal cycler, as previously described (93). All mRNA cycle threshold (Ct) values were normalized to GAPDH in the same sample for ΔCt and then normalized to the control sample for ΔΔCt, and fold change (2−ΔΔCt) was calculated. Primer sequences are provided in table S2.

Flow cytometry

Cell clones stably transfected with TOPtRFP or FOPtRFP were allowed to proliferate and either induced to enter G0 or reactivated into the cell cycle by replating for 24 hours before RFP intensity was analyzed by fluorescence-activated cell sorting using FlowJo software (FlowJo LLC). Drug [iCRT3 (5 and 25 μM), SB431542 (1 and 5 μM), and SIS3 (5 and 10 μM)]– or siRNA (50 pmol)–treated cells in G0 were checked for viability using the Annexin V/Propidium Iodide (PI) Kit (Thermo Fisher Scientific), as per the manufacturer’s instructions, and flow cytometry. Cells positive for annexin V were considered apoptotic, and annexin V– and PI-positive cells were considered necrotic. Unstained cells were scored as live cells.

Coimmunoprecipitation

Cells were lysed on ice in hypotonic buffer containing 0.2% NP-40 and 10% glycerol for 10 min on ice. Nuclei were pelleted by centrifugation at 6000g for 10 min at 4°C and lysed at 4°C in nuclear lysis buffer [50 mM tris-Cl (pH 7.5), 150 to 350 mM NaCl, 0.7% NP-40, 0.1 mM EDTA, and 10% glycerol]. Four micrograms of antibody was added to 500 μg of nuclear protein, and immune complexes were allowed to form overnight at 4°C. Immune complexes were recovered using protein A–agarose beads (for 4 hours). Bead-protein complexes were washed four times with washing buffer [50 mM tris (pH 7.5) and 150 to 350 mM NaCl] before the protein complexes were eluted using 2× Laemmli sample buffer and analyzed by SDS–polyacrylamide gel electrophoresis (PAGE) gel using 10% of input as control.

Protein isolation and immunoblotting

Protein was isolated from C2C12 cells using 2× Laemmli sample buffer after two phosphate-buffered saline (PBS) washes. Lysates were centrifuged for 3 min at 15,000g to remove debris. Supernatants representing equal amounts of protein were loaded onto 10% SDS-PAGE gels, transferred to polyvinylidene difluoride membrane (Bio-Rad), blocked with 5% Blotto (Santa Cruz Biotechnology) for 1 hour at room temperature, and probed with primary antibody for 1 hour at room temperature or overnight at 4°C, followed by horseradish peroxidase–conjugated secondary antibody 1 hour at room temperature. Blots were developed using ECL reagent (Amersham) and captured using ImageQuant (Amersham). Gapdh protein was used as the loading control. Intensity of bands in Western blots was analyzed by densitometry using Fiji (ImageJ) software.

Colony formation assays

G0 cells treated with drug [iCRT3 (5 and 25 μM), SB431542 (1 and 5 μM), and SIS3 (5 and 10 μM)] or siRNA (50 pmol) were as previously described (28). Two hundred fifty or 500 viable cells were seeded per 150-mm dish. Colonies were stained by methylene blue after 7 days.

Immunocytochemistry

Cells were cultured on glass coverslips, fixed with 2% paraformaldehyde for 20 min, permeabilized for 15 min with 0.5% Triton X-100 in PBS, blocked with 2% horse serum in 0.5% Triton X-100–PBS for 1 hour, and incubated with primary antibody diluted in blocking buffer at room temperature for 1 hour or overnight at 4°C. The cells were then incubated in the appropriate secondary antibody for 45 min at room temperature. Nuclei were counterstained with DAPI (1 μg/ml) in PBS for 15 min. Samples were imaged on a LSM510 Meta (Zeiss) or Leica SP8 confocal microscope. Image intensity was calculated using Fiji (ImageJ) software, and corrected mean intensity (CMI = total intensity of signal − area of signal × mean background signal) was determined for more than 100 nuclei per sample.

Muscle fiber analysis

Animal experiments were conducted with prior approval of Institute for Stem Cell Biology and Regenerative Medicine (InStem) and Center for Cellular and Molecular Biology (CCMB) Institutional Animal Ethics Committees and in accordance with Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA) guidelines. Single muscle fibers were isolated from EDL muscles of adult Pax7-nGFP mice (60) and cultured using methods described, with some modification (94). Briefly, the EDL muscles from both hindlimbs were dissected from 8-week-old mice. Isolated muscles were digested with type I collagenase (400 U/ml; Worthington) in DMEM at 37°C, until single fibers dissociated. All dissociated fibers were transferred into fresh DMEM. Fibers were cleaned by transfer through media washes (three times) and then to fiber culture media [DMEM (Gibco) + 20% FBS (Gibco), 10% horse serum (Gibco), 2% chick embryo extract (Gentaur), 1% penicillin-streptomycin (Gibco), and basic FGF (5 ng/ml; Sigma-Aldrich)]. EdU (10 μM) was added to media to mark MuSC activated during culture. For the drug treatment experiments, EDL fibers were cultured for 3 days and treated with DMSO (control), SB431542 (5 μM), SIS3 (10 μM), or iCRT3 (25 μM) for 24 hours. EdU (10 μM) was added to the media for the final 1 hour of culture. Cultured fibers were fixed in 4% paraformaldehyde for 5 min at room temperature. Freshly isolated or cultured fibers were analyzed by immunostaining.

For prequiescent MuSC (satellite cell) isolation, 1-week-old Pax7-nGFP mice were injected intraperitoneally with two doses of 200 μg of EdU, 8 hours apart. Sixteen hours after the last injection, mice were euthanized, and the entire hindlimb muscles were isolated. Muscles were digested in type I collagenase (200 U/ml; Worthington) in DMEM for 45 min at 37°C, and digestion was quenched using serum-containing media. Muscles were triturated with wide-bore fire-polished glass pipettes. Dispersed single fibers were fixed in 4% paraformaldehyde for 5 min, washed 3× with PBS, and placed on charged slides (Thermo Fisher Scientific) for immunostaining.

Chromatin immunoprecipitation

ChIP was performed as previously described (28, 93). Two micrograms of primary antibody was used for each experiment (antibody details are in table S1). Enriched genes were analyzed by real-time PCR. Fold change was calculated over the Ct values of input and IgG control using 2−ΔΔCt method. ChIP analysis for knockdown G0 samples was performed using LowCell# ChIP kit (C01010072, Diagenode), as per the manufacturer’s protocol. Primer details are in table S2.

Statistical analysis

Two-tailed t tests were performed with Microsoft Excel. Error bars represent SEM, and P values were indicated by *P < 0.05, **P < 0.01, and ***P < 0.001. All molecular, cellular, and animal experiments were performed using three or more biological replicates, each comprising multiple technical replicates.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/11/540/eaan3000/DC1

Fig. S1. Canonical Wnt pathway components in different cellular states.

Fig. S2. β-Catenin and Lef1 knockdown and iCRT3 treatment.

Fig. S3. Expression and activity of Tcf/Lef proteins in different cellular states.

Fig. S4. Occupancies of Lef1 and Tcf7l1 on Tcf/Lef target genes.

Fig. S5. Role of TGF-β and Smad3 in G0.

Fig. S6. Effect of Lef1 knockdown on Smad3 occupancy at target gene promoters in G0.

Fig. S7. Effect of Smad3 knockdown on Lef1 occupancy at target gene promoters in G0.

Fig. S8. Smad2 and Smad4 in different cellular states.

Fig. S9. Role of Smad2 and Smad4 in G0-specific Tcf/Lef-mediated transcriptional activation.

Fig. S10. Effect of iCRT3 treatment on Smad3-Lef1 interaction in myoblasts.

Table S1. Antibodies used in this study.

Table S2. Primers used in this study.

REFERENCES AND NOTES

Acknowledgments: We thank R. Sambasivan and V. Radha for the comments on the manuscript, R. Moon and M. Waterman for the plasmids, and S. Tajbakhsh for the Pax7-nGFP mice. We are grateful for the access to Confocal Imaging and Flow Cytometry Facility (CIFF) at National Centre for Biological Sciences (NCBS) and Advanced Imaging Facility at CCMB. Funding: Animal work was partially supported by National Mouse Research Resource grant (BT/PR5981/MED/31/181/2012; 2013–2016) from Department of Biotechnology (DBT). This work was supported by a doctoral fellowship from Tata Institute of Fundamental Research–NCBS to A.A., a California Institute of Regenerative Medicine–InStem grant to R.D., Indo-Denmark and Indo-Australia grants from DBT to J.D., and core funds from InStem and CSIR-CCMB. Author contributions: A.A., R.D., and J.D. designed the experiments. A.A. performed the experiments. A.A. and J.D. wrote the manuscript with input from R.D. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or the Supplementary Materials. The Pax7-nGFP mice used in this study were obtained from the Pasteur Institute under a material transfer agreement (MTA). These mice can only be obtained under an MTA executed directly with the Pasteur Institute.
View Abstract

Navigate This Article