Register for Mechanobiology 2016!

Foxk1 promotes cell proliferation and represses myogenic differentiation by regulating Foxo4 and Mef2
Xiaozhong Shi, Alicia M. Wallis, Robert D. Gerard, Kevin A. Voelker, Robert W. Grange, Ronald A. DePinho, Mary G. Garry, Daniel J. Garry


In response to severe injury, adult skeletal muscle exhibits a remarkable regenerative capacity due to a resident muscle stem/progenitor cell population. While a number of factors are expressed in the muscle progenitor cell (MPC) population, the molecular networks that govern this cell population remain an area of active investigation. In this study, utilizing knockdown techniques and overexpression of Foxk1 in the myogenic lineage, we observed dysregulation of Foxo and Mef2 downstream targets. Utilizing an array of technologies, we establish that Foxk1 represses the transcriptional activity of Foxo4 and Mef2 and physically interacts with Foxo4 and Mef2, thus promoting MPC proliferation and antagonizing the myogenic lineage differentiation program, respectively. Correspondingly, knockdown of Foxk1 in C2C12 myoblasts results in cell cycle arrest, and Foxk1 overexpression in C2C12CAR myoblasts retards muscle differentiation. Collectively, we have established that Foxk1 promotes MPC proliferation by repressing Foxo4 transcriptional activity and inhibits myogenic differentiation by repressing Mef2 activity. These studies enhance our understanding of the transcriptional networks that regulate the MPC population and muscle regeneration.


Adult skeletal muscle is a dynamic and highly regenerative tissue due to a resident myogenic progenitor cell (MPC) population (Mauro, 1961). In response to a severe injury that involves more than 90% of the muscle, the MPC population is capable of completely restoring the cellular architecture within a three-week period. Recent studies using genetic mouse models and transcriptome analysis have identified molecular markers for the MPC population that include Foxk1, CD29, C-met, integrin alpha7, m-cadherin, Pax3, Pax7 and Syndecan3/4 (Biressi and Rando, 2010; Shi and Garry, 2006). In addition, the C-met/Hgf, Igf, Tgfb/Myostatin/Smad3/4, Notch/Numb signaling pathways have also been shown to be essential for the MPC population (Buckingham and Vincent, 2009; Ten Broek et al., 2010). Despite these recent insights, the molecular networks that govern the MPC population remain an area of active investigation.

Forkhead/winged helix transcription factors are known to exert important regulatory functions in developmental processes including the determination of cell fate, cell cycle kinetics, cell differentiation and tissue morphogenesis (Hannenhalli and Kaestner, 2009; Myatt and Lam, 2007; Wijchers et al., 2006; Yang et al., 2009). We have previously established that Foxk1 is restricted to the MSC/MPC population in adult skeletal muscle (Garry et al., 1997). Foxk1 deficient mice have severely impaired skeletal muscle regeneration, decreased number of muscle progenitor cells, impaired progenitor cell activation, increased expression of the cyclin dependent kinase inhibitor, p21, and perturbed cell cycle kinetics of the muscle progenitor cell population (Garry et al., 2000; Hawke et al., 2003a). Transgenic, molecular biological and biochemical studies have demonstrated that Sox15 is a potent transcriptional activator of Foxk1 in the myogenic progenitor cell population, although Foxk1's downstream transcriptional program in this lineage has yet to be defined (Meeson et al., 2007). Our recent studies have demonstrated that Foxk1 recruits Sin3/Sds3 repression complex and functions to activate the myogenic progenitor although the mechanisms are incompletely defined (Shi and Garry, 2012; Shi et al., 2012).

Foxo proteins have been shown to have a broad functional role in the regulation of catabolic pathways, cell cycle kinetics, cell fate, aging and life span (Burgering, 2008; Ho et al., 2008; Partridge and Brüning, 2008). Recent studies have demonstrated that Foxo1 transgenic overexpression in skeletal muscle results in decreased body size, decreased muscle mass and increased atrogin 1 (ubiquitin ligase) expression (Kamei et al., 2004). In addition, molecular biological and biochemical studies have demonstrated that Foxo proteins directly interact with the Tgf-beta downstream effectors, Smad3/4 and transcriptionally co-activate the cyclin dependent kinase inhibitor, p21CIP and maintain the hematopoietic stem cell population in a quiescent state (Seoane et al., 2004; Tothova et al., 2007).

In the present study, we have utilized an array of techniques to uncover the functional role of Foxk1 in the MPC population. We have knocked down Foxk1 using siRNA techniques and overexpressed Foxk1 using a transgenic technique in the myogenic lineage. Our analysis of Foxk1 knockdown and overexpression revealed dysregulation of Foxo and Mef2 downstream target genes, respectively. We demonstrate that Foxk1 directly interacts with Foxo4 and represses Foxo4 transcriptional activity, and that the repression of Foxo4 results in decreased p21 expression and increased cellular proliferation of the MPC population. We further demonstrate that Foxk1 binds to and represses Mef2c thereby restraining myogenic differentiation. Collectively, our current data concerning Foxk1 provide direct evidence for a specific role for members of this Forkhead gene family in the regulation of progenitor/stem cell function.


Foxk1 is required for the cell cycle progression

Our previous studies have defined the expression of Foxk1 in myogenic progenitor cells. Further, we have reported that loss of Foxk1 resulted in the perturbation of skeletal muscle regeneration due to impaired cell cycle regulation of the myogenic progenitor cell population. To further define the underlying mechanisms for Foxk1, we knocked down Foxk1 in C2C12 cells using siRNA oligonucleotides. From the four candidates, we identified two siRNA olgionucleotides, which efficiently knocked down Foxk1 in C2C12 cells (Fig. 1A). Using these reagents, we analyzed the effect of Foxk1 knockdown on cell cycle kinetics. As shown in Fig. 1B, the knockdown of Foxk1 resulted in cell cycle arrest using FACS analysis, which was further quantified in Fig. 1C. The gene expression studies revealed the upregulation of Foxo target genes (Fig. 1D). In addition, we observed decreased cellular proliferation with Foxk1 siRNA treatment (Fig. 1E). Taken together, these data support the notion that Foxk1 has an important functional role in the proliferation of the myogenic progenitor cell population.

Fig. 1.

Foxk1 promotes cellular proliferation. (A) Selection of Foxk1 siRNA oligonucleotides using qPCR analysis. All four siRNA oligonucleotides knocked down the endogenous Foxk1 mRNA. The No.1 and No. 2 siRNA oligonucleotides were selected for further analysis as they knocked down Foxk1 gene expression with higher efficiency. Ctrl, control. (B) Knockdown of Foxk1 using siRNA results in G0/G1 cell cycle arrest. The FACS profile is a representative sample using siRNA No. 2. (C) The quantification of the cell cycle phases from data presented in panel B (*P = 0.02, n = 4). (D) The gene expression profile was evaluated using qPCR following siRNA No. 2 treatment. The Foxo target genes are upregulated, including Gadd45a, p21, p27 and p57. Note that the expression of Foxo1, Foxo3a and Foxo4 were not affected. (E) Growth curve of C2C12 cells with Foxk1 siRNA No. 2 treatment. Ctrl, control.

Foxk1 represses transcription through a DNA-binding independent mechanism

The interaction between the winged helix domain (WHD) of Foxk1/Foxk2 and the consensus motif has been characterized using NMR spectroscopy and crystallography techniques (Chuang et al., 2002; Liu et al., 2002; Tsai et al., 2006). Using these techniques, a number of conserved amino acid residues within the WHD have been shown to contact with the DNA (Liu et al., 2002; Tsai et al., 2006). To further investigate the transcriptional repression of Foxk1, we constructed two Foxk1 WHD (winged helix domain) mutants: K333A and R340A as these conserved amino acids were important in DNA-binding (Liu et al., 2002; Tsai et al., 2006) (Fig. 2A). These mutations did not affect the protein stability in vitro (Fig. 2B). We observed that the DNA binding ability is attenuated in the K333A mutant and abolished in the R340A mutant (Fig. 2C). Transcriptional assays revealed that both mutants did not affect the Foxk1 repression activity in two distinct promoter-reporter constructs (Fig. 2D,E). Collectively, these studies suggested that Foxk1 represses transcription through a DNA-binding independent mechanism.

Fig. 2.

Transcriptional repression by Foxk1 is independent of its DNA-binding capacity. (A) Schematic illustration of the mutation in the Foxk1 DNA-binding domain (WHD). (B) In vitro translated product of Foxk1 (wild-type, WT) and its DNA-binding domain (WHD) mutants (K333A and R340A). (C) The DNA-binding ability of Foxk1 is impaired in the Foxk1 mutant K333A, and abolished in the Foxk1 mutant R340A using an EMSA. (D) The transcriptional repression by the Foxk1 mutants (K333A and R340A) is similar to the wild-type control using the Fox-binding element reporter. (E) A reporter construct which harbors a multimerized (8×) fragment from the p21 promoter is utilized in the transcriptional assays as described for panel (D). Ctrl, control.

Foxk1 represses and interacts with Foxo4

Our above studies support the hypothesis that Foxk1 regulates gene expression via Foxo proteins. To test our hypothesis, we used conventional transcriptional assays to evaluate the role of Foxk1 on Foxo transcriptional activity. We transfected a multimerized Foxo binding element (8×FBE) fused to the luciferase reporter to evaluate Foxo transcriptional activity in the presence and absence of Foxo factors and Foxk1 in C2C12 myoblasts. As expected, we observed that Foxo4 was a potent transcriptional activator of gene expression (Fig. 3A) (Shi et al., 2010). In a dose-dependent manner, Foxk1 repressed Foxo4 transcriptional activity (Fig. 3A; supplementary material Fig. S1A). Knockdown of Foxk1 enhanced the transcriptional activity (supplementary material Fig. S1B). In addition, we utilized the Gal4 reporter system, where Foxo4 was fused to the Gal4 DNA-binding domain and Gal4 UAS-luc was the reporter (Sadowski et al., 1992). We observed that Foxk1 repressed Gal4-Foxo4 activity in a dose-dependent fashion (supplementary material Fig. S1C).

Fig. 3.

Foxk1 represses and interacts with Foxo4. (A) Transcriptional assays (using the 8×FBE-luc) reveals that Foxk1 in a dose-dependent fashion represses Foxo4 transcriptional activity. (B) Using co-immunoprecipitation (IP) assays, the protein interaction between tagged Myc-Foxk1 and Flag-Foxo4 was confirmed in the overexpression studies. WB, western blot. (C) The endogenous Foxk1 and Foxo4 could also form a complex in C2C12 myoblasts using co-IP assays. (D) Schematic summary of the Foxk1 deletion constructs (FHA, forkhead domain associated domain; WHD, winged helix domain). (E) Coomassie blue staining of the purified GST-Foxk1 deletions and the GST-control proteins. (F) The GST pulldown assay reveals that the construct containing the FHA and WHD (81–406) interacts with Foxo4. (G) Schematic summarizing of the Foxo4 deletional constructs. (H) Coomassie blue staining of the purified GST-Foxo4 deletions and the GST control protein. (I) GST pulldown assay reveals that full length Foxo4 (1–505) and the WHD (97–215) interact with Foxk1. (J) Transcriptional assays further verify that the Foxk1 truncated protein that contains the FHA and WHD (81–406) fully represses Foxo4 transcriptional activity. (K) GST pulldown assays reveal that Foxk1 could bind to all the members of the Foxo1, Fox3a and Foxo4 family with various affinities, and most avidly to Foxo4. Ctrl, control.

To uncover the regulatory mechanism of Foxo4 activity by Foxk1, we first examined the protein interaction between Foxk1 and Foxo4 using a co-immunoprecipitation assay. As shown in Fig. 3B, Foxk1 can be co-immunoprecipitated by Foxo4, and the reverse is also true. Further studies revealed the interaction of endogenous Foxk1 and Foxo4 (Fig. 3C). Using GST pulldown assays, we observed that a construct which harbors the Forkhead Associated (FHA) and WHD of Foxk1 interacts with Foxo4 (Fig. 3D–F). Similarly, we demonstrated that Foxk1 directly interacted with the winged helix (DNA binding) domain of Foxo4 (Fig. 3G–I). To complement these binding studies, we examined the Foxk1 deletional constructs for their ability to repress Foxo4 transcriptional activity. Using transcriptional assays, the truncated Foxk1 construct, which harbors the FHA and DNA-binding domains (81–406), which is capable of binding Foxo4, is sufficient to repress Foxo4 activity (Fig. 3J; supplementary material Fig. S1D). These functional studies are reinforced by biochemical studies revealing that Foxk1 most avidly interacts with Foxo4, and to a lesser extent with Foxo1 and Foxo3a, in GST pulldown assays (Fig. 3K). Using transcriptional assays, Foxk1 also represses the activity of Foxo1 and Foxo3a in a dose-dependent fashion (supplementary material Fig. S1E–F). We propose that Foxk1 governs gene expression via a DNA binding independent mechanism, which is context dependent, thereby modulating the quiescent/proliferative state of the MPC population. Collectively, these data support the hypothesis that Foxk1 directly binds to the DNA-binding domain of Foxo4 and represses Foxo4 transcriptional activity, thereby decreasing its downstream target genes including p21.

Perturbed skeletal muscle regeneration in Foxo4 null mice

Our previous studies have demonstrated that the skeletal muscle regeneration is delayed in p21 knockout mice (Hawke et al., 2003b). As the cell cycle inhibitor genes are downstream targets of Foxo factors, we examined the muscle regeneration capacity in Foxo4 null mice. To examine the regenerative capacity of the Foxo4 mutant skeletal muscle, cardiotoxin was injected into the gastrocnemius (GAS) muscles. In the wild-type skeletal muscle, the cellular architecture was restored within 2 weeks following cardiotoxin injury. In contrast, the Foxo4 mutant skeletal muscle had perturbed regeneration that was evident with smaller myofibers 2 weeks following cardiotoxin injury (Fig. 4A,B). To label proliferating cells two weeks following CTX injury, the injured mice of wild type and Foxo4 null were pulsed with BrdU for a 48-hour period (24 hours×2) and the respective gastrocnemius muscles were processed for BrdU and Myod immunostaining (Fig. 4C). We observed increased numbers of BrdU and Myod labeled myoblasts using morphological and quantitative assays (Fig. 4C,D). To analyze the gene expression profile, we isolated RNA from the primary myoblasts from Foxo4 wild-type (WT) and null neonatal mice. Our data demonstrated that Foxo4 target genes (Gadd45a, p21, p27 and p57) were downregulated in Foxo4 null myoblasts using qRT-PCR analysis (Fig. 4E).

Fig. 4.

Foxo4 null skeletal muscle has increased cellular proliferation. (A) The gastrocnemius muscles of Foxo4 wild-type and null mice were injured with cardiotoxin (CTX). The gastrocnemius muscles were harvested 1 week (CTX-1w) or 2 weeks (CTX-2w) following the injury. The uninjured muscle was utilized as the control (Uninjured). Note that the muscle regeneration is perturbed in Foxo4 null skeletal muscle compared with the wild-type (WT) controls (scale bar, 50 µm). (B) Quantification of the myofiber cross-sectional area (CSA) in the regenerating muscle in panel A (*P = 0.04; **P = 0.03; n = 3). (C) There are increased numbers of BrdU- and Myod-positive cells (indicated by arrowheads) in the Foxo4 null skeletal muscle compared with the WT control 2 weeks following cardiotoxin injury. DAPI staining indicates the nuclear compartment. (D) Quantification of the BrdU+ and Myod+ cells in panel C (*P = 0.01; n = 3). (E) qRT-PCR analysis reveals decreased expression (Exp) of cell cycle inhibitors (i.e. p21 expression) in the Foxo4 null myoblasts versus the WT controls (n = 3).

Overexpression of Foxk1 perturbs skeletal muscle differentiation

Our previous studies demonstrated that Foxk1 expression is downregulated during cell differentiation (Shi et al., 2010). To examine the functional role of Foxk1 in cell differentiation, we utilized the 4.8 kb MCK promoter to overexpress Foxk1 using transgenic techniques as shown in Fig. 5A (Sternberg et al., 1988). Here, western blot analysis revealed abundant HA–Foxk1 overexpression in the fast twitch extensor digitorum longus (EDL), the slow twitch soleus (SOL) and the mixed fiber muscles [tibialis anterior (TA) and gastrocnemius (GAS)] of the transgenic mice, and absence of HA–Foxk1 expression in the wild-type controls (Fig. 5B). Initial transcriptome analysis revealed the dysregulation of Mef2 target genes (data not shown) (Black and Olson, 1998). Using qRT-PCR assays, we have further verified the dysregulation of Mef2 downstream target genes in Foxk1 TG muscle (Fig. 5C). In addition, we did not observe any changes in Mef2 mRNA (Fig. 5B) or the Mef2 protein in the Foxk1 overexpression transgenic mice (supplementary material Fig. S2).

Fig. 5.

Overexpression of Foxk1 represses muscle differentiation. (A) Schematic representation of the transgenic construct using the muscle creatine kinase (MCK) promoter to direct the HA–Foxk1 fusion protein to the myogenic lineage. (B) Western blot analysis of adult skeletal muscle isolated from wild-type (WT) and transgenic (TG) mice. The anti-HA serum identifies the Foxk1 fusion protein in the extensor digitorum longus (EDL), gastrocnemius (GAS), soleus (SOL) and tibialis anterior (TA) muscles. Anti-tubulin serum was used as the loading control. (C) qPCR analysis was utilized to examine the relative gene expression (Rel Exp) of Mef2 downstream targets. (D) Representative histological analysis of skeletal muscle isolated from the 2-month-old transgenic male mice reveals normal cellular architecture (n = 3) with occasional centronucleated myofibers (indicated by arrowheads; scale bar, 50 µm). (E) Differentiation of myogenic progenitor cells is delayed in Foxk1 transgenic myogenic progenitor cells compared with the wild-type control (scale bar, 100 µm; myotubes are marked with arrowheads). (F) Quantification of the cell fusion index in panel E (*P = 0.05, n = 3). (G) Using adenoviral vectors to overexpress Foxk1 (Ad-Foxk1) or GFP as a control (Ad-GFP), we observed that overexpression of Foxk1 repressed muscle differentiation and the formation of multinucleated myotubes (scale bar, 100 µm). (H) Quantification of the fusion index in panel G (*P = 0.01, n = 3). (I) Western blot analysis of the samples in panel G reveals that overexpression of Foxk1 results in decreased expression of the myogenic differentiation program (decreased expression of MHC and myoglobin; tubulin was used as a loading control).

Evaluation of the transgenic skeletal muscle revealed essentially normal skeletal muscle cellular architecture at 2 months of age (Fig. 5C). While we observed no evidence of tissue degeneration, there were increased number of nuclei associated with the transgenic muscle and occasional evidence of centronucleated myofibers (Fig. 5D). Furthermore, we isolated the primary myoblasts from the neonatal mice and examined their capacity for differentiation. As shown in Fig. 5E,F, the cellular differentiation (i.e. formation of multinucleated myotubes) is reduced in Foxk1 TG myoblasts.

To further explore the functional role of Foxk1, we engineered an adenoviral vector that overexpresses Foxk1. We infected C2C12CAR myoblasts with experimental (Ad-Foxk1) and control (Ad-GFP) viruses and exposed both samples to differentiation media for 48 hours. We observed a relative absence of multinucleated myotubes with Foxk1 overexpression (Fig. 5G,H). In contrast, the sample infected with the GFP expressing vector or the mock control had many multinucleated myotubes (performed in triplicate and in three separate experiments). To further support these morphological findings, we harvested the respective samples and undertook western blot analysis for Foxk1 and myogenic differentiation markers. As shown in Fig. 5I, Ad-Foxk1 overexpression was associated with decreased expression of myoglobin and MHC, which further supports the hypothesis that Foxk1 retards muscle differentiation.

Foxk1 inhibits and binds to Mef2

Our above data implicated a dual role for Foxk1 in muscle regeneration through the promotion of MPC expansion (i.e. cellular proliferation) and restraining of myogenic differentiation (through the repression of Mef2 activity). To test this hypothesis, we utilized transcriptional assays and cotransfected a multimerized Mef2-binding motif fused to the luciferase reporter in the absence and presence of increasing amounts of Foxk1. We performed the studies with Mef2c as a representative member of the Mef2 family. We observed that Foxk1, in a dose dependent fashion, repressed Mef2c transcriptional activity (Fig. 6A; supplementary material Fig. S3A). We further analyzed the Foxk1 mediated repression in the Gal4-UAS system. We observed that Foxk1 repressed Gal4-Mef2c activity in a dose-dependent fashion (supplementary material Fig. S3B). Using co-immunoprecipitation assays, we further determined that Foxk1 interacted with Mef2c using an overexpression strategy or endogenous protein in C2C12 myoblasts (Fig. 6B,C).

Fig. 6.

Foxk1 binds to Mef2 and inhibits Mef2 activity. (A) Transcriptional assays reveal that Foxk1, in a dose-dependent fashion, represses Mef2c transcriptional activity. (B) Co-immunoprecipitation (IP) assays reveal the protein interaction between Myc–Mef2c and HA–Foxk1. WB, western blot. (C) The protein interaction of endogenous Foxk1 and Mef2 is confirmed in C2C12 cells. (D) Schematic summary of the Foxk1 deletional constructs. (E) Coomassie blue staining of the purified GST-Foxk1 deletional proteins. (F) The GST pulldown assay reveals that the construct containing the FHA and WHD (81–406) interacts with Mef2c. (G) Schematic summary of the Mef2c deletional constructs. The activation domains are designated as TAD I and TAD II. (H) In vitro translated Mef2c constructs (upper panel) and the GST pulldown assay (lower panel) reveal that the MADS domain of Mef2c interacts with Foxk1. (I) Transcriptional assays further verify that the Foxk1 truncated protein that contains the FHA and WHD (81–406) represses Mef2c transcriptional activity.

To map the interacting domains between Foxk1 and Mef2c, we utilized GST pulldown assays. These assays revealed that a construct that harbors both the FHA and WHD of Foxk1 directly interacts with the MADS domain of Mef2c (Fig. 6D–H). Further, we used these deletional mutants and transcriptional assays to verify that the same Foxk1 deletional construct that interacted with Mef2c also repressed Mef2c transcriptional activity (Fig. 6I; supplementary material Fig. S3C). As the Mef2 MADS domain is involved in DNA-binding and protein–protein interactions, we hypothesized that Foxk1 prevents the interaction between Mef2c and its DNA binding motif in the target genes or alternatively Foxk1 represses the formation of the Mef2c transactivation complex and inhibits the activation of the myogenic differentiation program. To discriminate between these possibilities, we utilized electrophoretic mobility shift assays (EMSA) to examine the effect of Foxk1 on the Mef2c–DNA interaction. As shown in supplementary material Fig. S3D, the addition of Foxk1 reduced the formation of the high molecular weight Mef2c complex without affecting the low molecular complex. These studies support the notion that Foxk1 perturbs the Mef2 transcriptional complex.


Muscle progenitor cells reside in adult skeletal muscle and promote tissue regeneration in response to an injury or disease. While muscle progenitor cells have a tremendous proliferative capacity, the molecular regulation of this cell population is incompletely defined (Kuang and Rudnicki, 2008). In the present study, we made three discoveries, which significantly enhance our understanding regarding the molecular mechanisms that govern the MPC population proliferation. Our first discovery demonstrates that Foxk1 represses transcription through a DNA-binding independent mechanism. This Foxk1 mechanism is via the interaction with Foxo4 resulting in the repression of Foxo4 activity, thereby promoting MPC proliferation. This protein–protein interaction confirmed that the FHA and winged helix domains of Foxk1 interacted with the winged helix domain of Foxo4 thereby repressing its transactivation of its downstream target genes including the cyclin dependent kinase inhibitor p21.

To date there are more than 300 members that belong to the forkhead/winged helix transcription factor family based on relative homology of a 110 amino acid DNA-binding domain (also referred to as winged helix domain) since the discovery of the original member Fkh in Drosophila (Clark et al., 1993; Shimeld et al., 2010; Weigel and Jäckle, 1990; Weigel et al., 1989). Many of the forkhead/winged helix factors bind directly to cognate binding motifs of genes and transactivate or repress gene expression (Wijchers et al., 2006). Typically, these Fox factor DNA binding mechanisms are mediated by interacting cofactors that result in altered transcriptional responses (i.e. transcriptional synergy through the interaction of Foxo factors and Smads) (Gomis et al., 2006; van der Vos and Coffer, 2008). Some Fox family members modulate gene expression through protein–protein interactions and have DNA-binding independent functions (Foxe1, Foxg1, Foxp3, or Foxo-mediated protein degradation) (Bettelli et al., 2005; Hanashima et al., 2002; Perrone et al., 2000; Zhao et al., 2007). In the present study, we propose that Foxk1 governs gene expression via a DNA-binding independent mechanism, which is context dependent, thereby modulating the quiescent/proliferative state of the MPC population. Our data define one pathway whereby the MPC population re-enters the cell cycle resulting in an increased number of myogenic progenitors. This expansion of the myogenic progenitors is a critical regenerative response in the repair of damaged muscle.

Our second discovery revealed that the Foxo4 knockout mouse had increased cellular proliferation following cardiotoxin injury and decreased expression of cell cycle inhibitors including p21CIP. Previous gene disruption studies have verified that Foxo1 null embryos are lethal by E10.5 due to vascular perturbations (Furuyama et al., 2004; Hosaka et al., 2004) and mice lacking Foxo3a have perturbed ovarian follicular development and are infertile (Castrillon et al., 2003; Hosaka et al., 2004). However, initial analysis of the Foxo4 null mouse revealed no overt phenotype (Hosaka et al., 2004). Due to the redundancy and overlapping expression of the Foxo factors, recent efforts were undertaken to conditionally delete Foxo1, Foxo3a and Foxo4 (Paik et al., 2007; Tothova et al., 2007). These studies revealed that lineage-specific loss of the Foxo factors resulted in decreased number and impaired cell cycle kinetics of the hematopoietic stem cell pool (Tothova et al., 2007). These hematopoietic stem cell studies are conceptually aligned with our findings in the myogenic lineage where we demonstrate a prominent role for Foxo4 as a key cell cycle regulator in the MPC population.

Transgenic technologies have been useful in uncovering the physiological role of proteins in a temporal and spatial context. Such a transgenic strategy was used to overexpress Foxo members in the muscle lineage (Kamei et al., 2004; Skurk et al., 2005). Enforced Foxo1 expression in the skeletal muscle lineage resulted in smaller body size, reduced skeletal muscle mass, perturbed fiber type diversity (i.e. a shift towards increased number of oxidative slow twitch myofibers with myogenic Foxo1 overexpression) and an altered gene expression program that enabled definition of Foxo transcriptome in this tissue (Kamei et al., 2004). In contrast to the phenotype of the Foxo1 overexpression in skeletal muscle, our data demonstrate a distinct phenotype for transgenic Foxk1 overexpression in the skeletal muscle lineage that results in normal body size, preserved cellular function but altered gene expression that includes decreased expression of Foxo and Mef2 downstream target genes (data not shown). This genetic strategy uncovered an important functional mechanism for Foxk1 and its interacting proteins in the myogenic stem/progenitor cell population.

Previous studies in vertebrates support a role for Mef2c in skeletal myogenesis (Naya and Olson, 1999; Potthoff and Olson, 2007). Recent studies undertaken by Hughes and colleagues demonstrated that simultaneous morpholino knockdown of zebrafish Mef2c and Mef2d resulted in a loss of thick filament proteins and the disruption of the sarcomeric structure (Hinits and Hughes, 2007). In addition, it has been conclusively demonstrated that Mef2c is an essential upstream transcriptional activator of troponins in skeletal muscle and myofiber identity (Bassel-Duby and Olson, 2006; Blais et al., 2005). Conditional transgenic technologies have revealed a broader role for Mef2c in cellular maintenance in various lineages. The conditional deletion of Mef2c in skeletal muscle lineages using a MCK-cre transgenic line resulted in a severe decrease of type I fibers (Potthoff et al., 2007). Potthoff and colleagues clearly demonstrated that HDAC-mediated inhibition of Mef2c was the essential regulatory step in the conversion of oxidative fibers to slow-twitch, non-oxidative, fast-twitch fibers. The authors further demonstrated that the conditional loss of Mef2c in the skeletal muscle lineage, results in decreased expression of structural Mef2c target genes, including troponins I and T and myomesin2. Our third discovery demonstrated that Foxk1 interacted with the MADS domain of Mef2c and precluded its activation of the myogenic differentiation molecular program. Moreover, the overexpression of Foxk1 resulted in a delay in myogenic differentiation. These results suggest that Foxk1 has a dual function within the MPC population that includes the retardation of the differentiation process. In this fashion, the MPC population can expand to form a pool of MPCs that will respond to local cues and participate in the regenerative process.

Collectively, these studies support a model whereby Foxk1 directly interacts with Foxo4 and represses Foxo4 transcriptional activity (supplementary material Fig. S4). The repression of Foxo4 results in decreased p21 expression and increased cellular proliferation of the MPC population. We further demonstrate that Foxk1 interacts with Mef2 and inhibits Mef2 transcriptional activity thereby restraining myoblast terminal differentiation. Our current data concerning Foxk1 provide direct evidence for a specific dual role for members of this extended gene family in the regulation of progenitor/stem cell function and skeletal muscle regeneration. In this fashion, the MPC population can expand to form a pool of MPCs that will respond to local cues and participate in the regenerative process (supplementary material Fig. S4).

Materials and Methods

DNA and RNA manipulation

Mef2c expression plasmids and the Mef2 reporter were kindly provided by Dr Eric Olson (Molkentin et al., 1996; Naya et al., 1999). All of other plasmids were constructed by PCR and verified by DNA sequencing. An electrophoretic mobility-shift assay (EMSA) was done according to the protocol outlined in our previous studies. RNA extraction, cDNA synthesis, microarray, and qPCR were performed as previously described (Gallardo et al., 2003; Hawke et al., 2003b).

Western blot, co-immunoprecipitation, in vitro translation and GST pulldown

Western blot and co-immunoprecipitation were performed as described in the standard protocols with the following antibodies: anti-HA (Santa Cruz), anti-Myc (Santa Cruz), anti-Flag (Sigma), anti-Fox4 (Cell Signaling and Santa Cruz), anti-Mef2 (Santa Cruz), anti-tubulin (Sigma), anti-myoglobin (Dako), MF20 (Hyridoma Bank), and anti-Foxk1 sera as preciously described (Shi et al., 2010). In vitro protein expression was performed with TNT Quick systems (Promega) as outlined in the standard manual. GST pulldown assays utilized E. coli BL21 expressing GST fusion proteins, which were extracted with B-PER bacterial Protein Extraction Reagent (Pierce Biochemicals) and then purified with glutathione-Sepharose CL-4B (GE Healthcare). GST fusion proteins bound to Sepharose beads were incubated with 35S-labeled protein product and the BL21 cell extract. The beads pulldown complex was washed (four times) and resuspended in sample loading buffer, analyzed using a 4–20% polyacrylamide gel and imaged as previously described (Shi et al., 2010).

Tissue culture, transcriptional assays and primary myoblast preparation

C2C12 myoblasts were cultured in 35 mm dishes containing DMEM supplemented with 10% fetal bovine serum and penicillin/streptomycin. Approximately, 1.0×105 of cells were transfected with 4 µl of lipofectamine (Invitrogen) and assayed for both luciferase and β-galactosidase activity. Luciferase assays were performed using the Promega Luciferase Assay System following the manufacturer's instructions. All fold changes in luciferase activity were normalized to β-galactosidase activity, and to the vector alone as previously described (Alexander et al., 2010). All transfection experiments were performed in triplicate and replicated three times. Preparation of the primary myoblasts from Foxk1 transgenic, Foxo4 null, or wild type neonates was performed as previously described (Shi et al., 2010). Myogenic differentiation was promoted by exposing to differentiation medium (DMEM supplemented with 2% heat inactivated horse serum, antibiotics, insulin and transferrin) as previously described and was evaluated immunohistochemically by anti-MHC serum (clone MF20). The fusion index was defined as the ratio of the number of the nuclei in myotubes versus the total number of nuclei.

Adenoviral infection

The mouse Foxk1 cDNA was inserted under the control of the cytomegalovirus (CMV) promoter/enhancer upstream of an ires-GFP fragment to produce a bicistronic pAC shuttle plasmid (Ivanciu et al., 2007). To construct a negative control, no cDNA was inserted into this vector. Recombinant adenoviruses overexpressing Foxk1 and GFP (Ad-Foxk1) or GFP alone (Ad-GFP) were constructed using cre-loxP recombination in vitro (Aoki et al., 1999). A single cell clone of C2C12CAR myoblasts stably transformed with a human coxsackie adenovirus receptor (CAR) expression plasmid (kindly provided by Dr Susan Stevenson of Novartis) was infected with a multiplicity of 300 viruses per cell.

siRNA and cell cycle analysis

C2C12 cells were transfected with the Foxk1 siRNA oligonucleotides (Dharmacon) or with RISC (RNA-induced silencing complex)-free nontargeting duplexes as control, as previously described (Shi et al., 2010). The treated cells were fixed with cold enthanol for FACS analysis or lysed with Tripure for RNA extraction. The cell cycle profiles were analyzed on a FACScan and processed with Cell Quest software (Shi et al., 2010). For the cell growth analysis, C2C12 cells were seeded 2×104 cells/well into the 6-well plate 24 hours before siRNA treatment, and harvested 72 hours later for quantification.

Animal care, cardiotoxin-induced muscle regeneration, BrdU pulse and histology

All mice used were maintained, crossed, genotyped, injected and sacrificed in accordance with an approved Institutional Animal Care and Use Committee protocol at the University of Minnesota. Cardiotoxin (CTX, Calbiochem) induced muscle injury/regeneration model in adult mouse is an established, reliable model to study muscle regeneration (Goetsch et al., 2003). 100 µl CTX (10 µM) were delivered using an intramuscular injection into the gastrocnemius (mixed fiber type muscle group) of the adult 2-month-old male mice and the mice were sacrificed at defined time periods: control (uninjured), one weeks and two weeks (n = 3 at each time period). BrdU labeling reagent was injected into mice (Invitrogen) via the intraperitoneal route at 48 hours and 24 hours prior to sacrifice. Mice were anesthetized, perfusion fixed with 4% paraformaldehyde. The selected skeletal muscle groups were harvested and fixed in 4% paraformaldehyde, paraffin-embedded, sectioned and stained with hematoxylin and eosin (H&E) to assess skeletal muscle fiber architecture. The muscle cross-sectional area (CSA) of the regenerating gastrocnemius muscles was quantified using AxioVision 4.8. For immunostaining, the sections were incubated with a rat monoclonal anti-BrdU serum (AbD Serotec) and a polyclonal rabbit anti-Myod serum (Santa Cruz). The primary antisera were detected with species specific AlexaFluor 647 and AlexaFluor 594 fluorophore conjugated antisera (Jackson ImmunoResearch) and coverslipped with Vectashield mounting medium with DAPI and imaged using the Zeiss Axio Imager M1 microscope equipped with the AxioCam HRc camera and AxioVision 4.8 software as previously described (Meeson et al., 2007).

Generation of transgenic mice

The transgene construct (HA–Foxk1) was subcloned into the 4.8 kb MCK promoter cassette, which harbors the MCK upstream 4.8 kb to +1 base pair fragment (Sternberg et al., 1988). Transgenic mice were generated by the microinjection of the linearized constructs into fertilized F2 eggs (B6SJLF1; Jackson Labs), which were reimplanted into pseudopregnant F1 foster ICR females (Harlan) as previously described (Shi and Garry, 2010).


Student's t-tests were performed to identify significant difference (P<0.05) in data obtained from control and experimental samples. Data are presented as mean ± standard error of mean (SEM).


We are grateful to Eric Olson (UT Southwestern Medical Center) for generously providing the Mef2c expression plasmids and Mef2c reporter plasmids. We further acknowledge the support of Jennifer L. Springsteen and Kathy M. Bowlin for the assistance with the immunohistochemical analyses.


  • Funding

    Funding support was obtained from the National Institutes of Health (National Institute of Arthritis and Musculoskeletal and Skin) [grant numbers 5R01AR047850 and 5R01AR055906 to D.J.G.]. R.A.D. is supported by the Robert A. and Renee E. Belfer Institute for Applied Cancer Science. Deposited in PMC for release after 12 months.

  • Supplementary material available online at

  • Accepted August 9, 2012.


View Abstract