Rhabdomyosarcoma (RMS), the most common pediatric soft-tissue sarcoma, has two major histological subtypes: embryonal RMS (ERMS), which has a favorable prognosis, and alveolar RMS (ARMS), which has a poor outcome. Although both forms of RMS express muscle cell–specific markers, only ARMS cells express PAX3-FOXO1a or PAX7-FOXO1a chimeric proteins. In mice, Pax3 and Pax7 play key roles in muscle cell development and differentiation, and FoxO1a regulates myoblast differentiation and fusion; thus, the aberrant regulation of these proteins may contribute to the development of ARMS. In this paper, we report that FOXO1a is not expressed in primary ARMS tumors or ARMS-derived tumor cell lines and that restoration of FOXO1a expression in ARMS cells is sufficient to induce cell cycle arrest and apoptosis. Strikingly, the effects of FOXO1a are selective, as enforced expression of FOXO1a in ERMS-derived tumor cell lines had no effect. Furthermore, FOXO1a induced apoptosis in ARMS by directly activating the transcription of caspase-3. We conclude that FOXO1a is a potent and specific tumor suppressor in ARMS, suggesting that agents that restore or augment FOXO1a activity may be effective as ARMS therapeutics.
Rhabdomyosarcoma (RMS) is the most common soft-tissue sarcoma of childhood. The two most common histological subtypes are embryonal RMS (ERMS) and alveolar RMS (ARMS). Although recent progress in treatment has improved the outcome of patients with ERMS, those with ARMS often relapse with highly metastatic tumors (J. Anderson et al., 2001). Primary ARMS is characterized by a high apoptotic index and responsiveness to radiotherapy and chemotherapy (J. Anderson et al., 2001), yet relapsed ARMS acquires mutations in the p53 pathway, which renders these tumors resistant to therapy (Taylor et al., 2000).
ARMS cells display morphological features consistent with poorly differentiated skeletal muscle cells and typically express the muscle cell–specific transcription factor myogenin (Dias et al., 2000). ARMS is characterized by the recurrent chromosome translocations t(2;13) or t(1;13) that encode the chimeric transcription factors PAX3-FOXO1a and PAX7-FOXO1a, respectively (Barr, 2001). In mice, Pax3 is essential for primary myoblast cell migration during embryogenesis (Epstein et al., 1995; Conway et al., 1997), whereas Pax7 is required for the specification of muscle satellite cells (Seale et al., 2000). Thus, alterations of the PAX gene function by PAX3-FOXO1a and PAX7-FOXO1a are thought to contribute to muscle cell transformation. In accordance with this notion, PAX3-FOXO1 functions as an oncogene in immortal NIH-3T3 fibroblast cells (Fredericks et al., 1995; Lam et al., 1999). However, enforced expression of Pax3-FOXO1a in mice is not sufficient to cause ARMS (M. Anderson et al., 2001; Lagutina et al., 2002), and in a Myf6 promoter-driven conditional Pax3-Fkhr knock-in mouse model, ARMS-like tumors are an extremely rare event, although they occur in ∼20% of these mice when they are crossed onto an Ink4a/Arf- or p53-deficient background (Keller et al., 2004).
The FOXO family of transcription factors (FOXO1a, FOXO3, and FOXO4a) regulates diverse cellular responses, including apoptosis, cell cycle arrest, differentiation, DNA repair, and/or oxidative stress (Burgering and Kops, 2002; Birkenkamp and Coffer, 2003; Bois and Grosveld, 2003; Tran et al., 2003; Accili and Arden, 2004). How FOXO proteins direct such diverse processes is largely unknown, but some of the cellular alterations associated with skeletal muscle differentiation are similar to those typical of apoptosis. For example, remodeling of the cytoskeleton (Qu et al., 1997; Gallo et al., 1999; Hall and Nobes, 2000; Sabourin and Rudnicki, 2000; Coleman and Olson, 2002) and activation of Casp3, a key effector of apoptotic protease (Fernando et al., 2002), are common features of both processes. Furthermore, FoxO1a appears to play a role as a master regulator of muscle cell differentiation in mice (Bois and Grosveld, 2003; Nishiyama et al., 2004; Bois et al., 2005), as augmenting FoxO1a activity provokes an overfused phenotype, whereas disrupting FoxO1a function is sufficient to prevent myoblast cell fusion (Bois and Grosveld, 2003). Furthermore, the induction of myoblast differentiation triggers the nuclear translocation and accumulation of FoxO1a and leads to the expression of a cadre of target genes necessary for cytoskeleton remodeling and cell fusion (Bois and Grosveld, 2003).
The t(2;13) or t(1;13) in ARMS results in a haploid gene dose of human FOXO1a and, given the profound effects of FoxO1a on muscle cell fate, we reasoned that reductions in FOXO1a may play a role in alveolar rhabdomyosarcomagenesis. Indeed, here we report that FOXO1a expression is repressed in ARMS and that this occurs through both transcriptional and posttranslational mechanisms. In addition, FOXO1a has been shown to function as a selective tumor suppressor for ARMS but not ERMS tumors in vivo. Finally, the tumor suppressor functions of FOXO1a are also linked to its ability to regulate the transcription of Casp3.
Loss of FOXO1a expression is a hallmark of ARMS but not ERMS
To address whether the expression of FOXO family proteins (FOXO1a, FOXO3a, and FOXO4) was maintained in pediatric RMS, we immunoblotted cell lysates of frozen primary ARMS and ERMS tumor biopsies (from the tumor bank at St. Jude Children's Research Hospital, Memphis, TN) using antibodies specific for FOXO1a, FOXO3a, and FOXO4. In addition, we determined the expression of these proteins in four ERMS (Rh2, Rh6, RD, and JR1) and four ARMS (Rh3, Rh4, Rh30, and Rh41) tumor-derived cell lines, which are models for secondary RMS. Strikingly, FOXO family proteins were undetectable in primary ARMS, whereas at least one FOXO member was expressed at high levels in all primary ERMS (Fig. 1 A). Analysis of ERMS- and ARMS-derived cell lines confirmed that all ERMS cells showed robust expression of FOXO1a, FOXO3a, and/or FOXO4, whereas FOXO1a was only weakly expressed in one ARMS cell line (Fig. 1 A, Rh4 cells).
FoxO1a activity is essential for muscle cell fusion that accompanies terminal differentiation of muscle cell progenitors (Bois and Grosveld, 2003; Bois et al., 2005). To assess whether differences in the expression of FOXO proteins in ARMS versus ERMS might reflect differences in their state of differentiation, we also evaluated the expression of two muscle cell–specific markers, myogenin (intermediate differentiation) and myosin heavy chain (MyoHC; late differentiation). Interestingly, ARMS tumors and cell lines often expressed the intermediate differentiation marker myogenin, whereas most ERMS tumors and cell lines also expressed the late differentiation marker MyoHC (Fig. 1 A). Thus, ARMS tumors are more myoblast-like in their differentiation status than are ERMS tumors.
To determine whether alterations in the expression of FOXO proteins accompany muscle cell fusion and differentiation, we used C2C12 cells, which like normal myoblasts, undergo differentiation and fuse into myotubes after being transferred to a low serum medium (Rando and Blau, 1994). As expected, these cells first expressed the intermediate marker myogenin at a time when cell fusions began, and then expressed high levels of MyoHC as the cells terminally differentiated (Fig. 1 B). The FoxO family members FoxO3a and FoxO4 were not expressed at any interval during differentiation of C2C12 cells. In contrast, FoxO1a expression was detectable in proliferating muscle cell progenitors, and as expected (Bois and Grosveld, 2003), the levels of FoxO1a were induced after the receipt of differentiation cues (Fig. 1 B). However, by day 3 of culture, FoxO1a levels began to rapidly diminish such that the protein was not expressed in differentiated muscle cells (Fig. 1 B). These findings underscore the dynamic regulation and role for FoxO1a in orchestrating muscle cell fusion (Bois and Grosveld, 2003; Bois et al., 2005) and suggest that loss of FOXO1a expression in ARMS would disable the ability of these tumor cells to fuse and/or terminally differentiate.
In many systems the activity of FOXO1a is held in check by AKT-mediated phosphorylations (of FOXO1a-Thr24, -Ser256, and -Ser319) that target the protein for nuclear export to the cytoplasm, where it is degraded by the proteasome (Brunet et al., 1999; Matsuzaki et al., 2003). To address whether there were changes in the localization of FOXO1a in ARMS versus ERMS, we performed immunofluorescence assays in the panel of ERMS and ARMS cell lines and transduced these cells with retroviruses that express wild-type FOXO1a (FOXO1a-WT) or dominant-active FOXO1a (FOXO1a-TM) that bears alanine substitution mutations in the three AKT phosphorylation sites (Brunet et al., 1999). Although the endogenous FOXO1a protein could not be detected in any of the cells with the FOXO1a-specific antibody by immunofluorescence assays, the overexpressed FOXO1a-WT and FOXO1a-TM proteins were readily evident in all transduced cells with the FOXO1a-specific antibody (Fig. 2). Furthermore, both FOXO1a-WT and FOXO1a-TM proteins were nuclear in their localization, indicating that nuclear-to-cytoplasmic export is not necessarily augmented in ARMS. Interestingly, however, overexpression of both FOXO1a-WT and FOXO1a-TM led to rapid and distinct morphological changes in ARMS, but not ERMS, cells. Specifically, Rh3, Rh4, and RH41 FOXO1a-expressing cells showed evidence of cell shrinkage, whereas Rh30-transduced cells took on an elongated appearance (Fig. 2), indicating that FOXO1a specifically alters the biology of ARMS cells.
FOXO proteins function as potent activators of their transcription targets (Bois and Grosveld, 2003; Accili and Arden, 2004; Bois et al., 2005), suggesting that FOXO1a's function as a transcription factor might be compromised in ARMS cells. To evaluate FOXO transcriptional activity in ERMS and ARMS tumor cells, we transiently transfected them with a luciferase promoter–reporter harboring six FOXO binding sites, which is responsive to FOXO1a activity (Furuyama et al., 2000). As expected from their expression of FOXO proteins, JR1 and RD ERMS cells displayed higher levels of basal activity of this reporter than did ARMS cells. Specifically, on an absolute scale, JR1 and RD ERMS cells showed basal values of luciferase activity of 150 and 25, respectively, whereas basal luciferase activities of Rh3, Rh4, Rh30, and RH41 cells were 1.2, 2.4, 9.9, and 14.7, respectively (Fig. 3, A and B). Furthermore, when cotransfected with a FOXO1a expression construct, ERMS cells showed a marked induction of the FOXO-responsive reporter (of 28- and 5-fold for JR1 and RD cells, respectively), whereas this response was diminished in all four ARMS tumor cell lines (with a range of 1.4- to 2.8-fold; Fig. 3, A and B). Therefore, FOXO1a transcriptional activity is also diminished in ARMS cells.
To address the potential mechanism by which FOXO1a protein expression was suppressed in ARMS, we initially compared levels of FOXO1a transcripts in ERMS and ARMS cells by quantitative real-time RT-PCR. Indeed, ARMS cell lines expressed greatly reduced levels of FOXO1a transcripts compared with levels of FOXO1a mRNA present in ERMS cells (Fig. 4 A). Further, levels of FOXO1a transcripts in ARMS were much less than one would expect from the reductions of just a haploid gene dose of FOXO1a attributable to the PAX3-FOXO1a translocation present in these tumors. Therefore, reductions in FOXO1a expression in ARMS are at least in part the result of reduced levels of FOXO1a transcription and/or a reduced half-life of FOXO1a mRNA.
The reductions in the FOXO1a protein in ARMS were much more profound than the reductions in FOXO1a transcripts observed in these tumor cells. This suggested that pathways that regulate FOXO1a protein levels might be operational in ARMS. In particular, the PI3K–AKT pathway phosphorylates FOXO1a and targets the protein for destruction by the proteasome (Matsuzaki et al., 2003), and recent studies have suggested that AKT targets FOXO1a for degradation in ARMS-derived cell lines (Wan and Helman, 2003). Hallmarks of this pathway are activating phosphorylations of AKT; however, immunoblot analyses of its phosphorylation status, using antibodies specific for the phospho-Thr308 or phospho-Ser473 forms of AKT, demonstrated that AKT was not active in ARMS or ERMS cells, although active AKT was readily observed in cells lacking PTEN (phosphate and tension homologue), a negative regulator of this pathway (Fig. 4 B). Therefore, loss of FOXO1a protein in ARMS appears independent of AKT.
FOXO transcription factors are degraded by the proteasome (Matsuzaki et al., 2003). Therefore, we addressed whether loss of FOXO1a in ARMS could be the result of increased degradation by the proteasome by treating these cells with the proteasome inhibitor MG132. Indeed, immunoblot analyses demonstrated that ARMS cells treated with MG132 showed a rapid and robust up-regulation of the FOXO1a protein (Fig. 4 C). Therefore, reductions in the FOXO1a protein in ARMS are at least in part attributable to high rates of proteasome-mediated turnover, and loss of FOXO1a in ARMS is the result of pathways that affect the transcription or half-life of FOXO1a mRNA and a posttranslational pathway that provokes rapid degradation of the protein.
FOXO1a induces G2/M cell cycle arrest, morphological changes, and apoptosis in ARMS
The finding that FOXO1a expression is markedly reduced in ARMS suggested that loss of the FOXO1a protein might be a crucial event in the etiology of this malignancy. To address the potential biological effects of FOXO1a, we transduced the ARMS and ERMS cell lines with retroviruses expressing FOXO1a-WT (MSCV-FOXO1a) or the AKT-unphosphorylatable, dominant-active mutant of FOXO1a-TM (MSCV-FOXO1a-TM). We reasoned that because the AKT pathway was inactive in ARMS cells (Fig. 4 B), enforced expression of FOXO1a-TM would have effects similar to those caused by overexpression of FOXO1a-WT. Strikingly, enforced expression of either FOXO1a or FOXO1a-TM induced rapid apoptosis in Rh3, Rh4, and Rh41 cells (but not in Rh30 ARMS cells; Fig. 5 B) as well as G2/M cell cycle arrest in all four ARMS cell lines (Fig. 6 A). By contrast, the deleterious effects of FOXO1a on cell growth and survival were restricted to ARMS, as ERMS cells were totally unaffected by enforced expression of FOXO1a or FOXO1a-TM (Fig. 5 B and Fig. 6, A and B). Furthermore, FOXO1a-expressing Rh4 cells, and especially Rh30 cells, displayed morphological changes reminiscent of muscle cell differentiation before undergoing apoptosis, as cells assumed elongated shapes and aligned in a manner akin to that of differentiating primary myoblasts (data shown for FOXO1a-TM–expressing cells; Fig. 6 C). However, only partial morphological changes were observed, as these cells did not contract or express the late differentiation marker MyoHC (unpublished data).
FOXO1a directly regulates the transcription of
Given FOXO1a's role as a master regulator of primary myoblast differentiation (Bois and Grosveld, 2003; Bois et al., 2005) and its ability to provoke apoptosis in ARMS (Fig. 5 B), we reasoned that the loss of FOXO1a expression in ARMS might alter the expression of components of the differentiation program of muscle cells and/or cells that are involved in apoptosis. Casp3 was an attractive candidate for such a target, especially given its role as an effector protease in apoptosis and its essential role in initiating skeletal myogenic differentiation (Fernando et al., 2002). To determine whether the biological effects of enforced FOXO1a expression in ARMS were associated with Casp3 expression and/or activity, we assessed its regulation in FOXO1a-TM–expressing ARMS and ERMS cells. Notably, all cells overexpressing FOXO1a-TM, regardless of their origin, displayed substantial increases in the level of Casp3 mRNA (Fig. 7 A). To determine if this induction was also reflected at the level of the protein, we assessed levels of pro-Casp3 and activated Casp3 (p17 and p20), by immunoblot of FOXO1a-TM–expressing ARMS and ERMS (Fig. 7 B), as well as overall Casp3 activity (Fig. 7 C). Interestingly, levels of pro-Casp3 and activated Casp3 proteins were only elevated in ARMS cells that were sensitive to FOXO1a-induced apoptosis (Fig. 7, B and C). Therefore, some level of translational control of Casp3 appears to be operational in ERMS cells and in Rh30 ARMS cells, which prevents excessive production of the Casp3 protein.
FOXO proteins activate transcription by binding consensus TTGTTTAC elements in the regulatory regions of their target genes (Furuyama et al., 2000). Interestingly, two regions of ∼500 bp (−3.8/−3.3 and +1.0/+1.5 kb) in the promoter-regulatory region of the mouse Casp3 gene harbor several TTGTTTAC elements. These sites are also present in the human CASP3 and Fugu Casp3 promoter-regulatory regions (unpublished data). We therefore addressed whether FOXO1a occupies these elements in mouse myoblasts undergoing differentiation using chromatin immunoprecipitation (ChIP) analyses. Because of the poor immunoprecipitation using the polyclonal FOXO1a antibody, we had to address this issue using FLAG-tagged FOXO1a. Primary myoblasts transduced with a retrovirus expressing FLAG-FOXO1a (MSCV-FOXO1a) were selected by FACS for GFP, which is expressed in cis by virtue of an IRES in this vector; enforced expression of FoxO1a-WT using these conditions has limited, if any, effects on the differentiation program of primary myoblasts (Bois and Grosveld, 2003). A ChIP analysis demonstrated that FOXO1a was recruited to these sites in the Casp3 promoter within 6 h of the induction of differentiation (Fig. 8 A). Furthermore, the activity of luciferase promoter–reporter constructs containing either of these elements in the Casp3 gene was dramatically induced in C2C12 muscle cells after the receipt of differentiation cues, and this response was augmented in myoblasts engineered to overexpress FOXO1a (Fig. 8, B and C). By contrast, luciferase promoter–reporters bearing deletions of the TTGTTTAC elements in these regions of the Casp3 promoter were not responsive to differentiation cues or to FOXO1a (Fig. 8, B and C). Therefore, FOXO1a is recruited to FoxO-binding sites at the Casp3 locus during the initial phase of myoblast differentiation, which coincides with the induction of Casp3 activity in differentiating primary myoblasts (Fernando et al., 2002). Collectively, these data support the conclusion that FOXO1a's ability to initiate apoptosis and differentiation in ARMS cells appears strongly linked to its ability to directly activate caspase-3 transcription.
FOXO1a is a tumor suppressor in ARMS
The ability of FOXO1a to selectively compromise the growth and/or survival of ARMS tumor cells suggested that loss of FOXO1a expression might be essential for tumorigenesis. To test this hypothesis, we transduced Rh30 cells with a vector expressing a tamoxifen-regulated form of FOXO1a-TM (FOXO1a-TM-ERTAM), in which FOXO1a-TM was fused in-frame with the hormone-binding domain of an estrogen receptor (ER) that has been engineered to selectively bind to the ER agonist tamoxifen (Mahfoudi et al., 1995). Attempts to establish other ARMS FOXO1a-TM-ERTAM–expressing cell lines (Rh3, Rh4, and Rh41) failed because of the leakiness of the ERTAM system and their greater sensitivity to FOXO1a-induced apoptosis (Fig. 5). As expected for the effects of enforced expression of FOXO1a-TM in Rh30 ARMS cells (Fig. 6), tamoxifen treatment of Rh30 cells expressing FOXO1a-TM-ERTAM induced G2/M cell cycle arrest and dramatic changes in cell morphology (Fig. 9, A and B). In contrast, tamoxifen treatment had no detectable effect on Rh30 cells expressing only the retroviral vector.
Tamoxifen-treated Rh30 cells expressing FOXO1a-TM-ERTAM remained elongated and arrested for up to 20 d in culture. Once elongated, this change was irreversible; i.e., the removal of tamoxifen did not release FOXO1a-TM-ERTAM–expressing cells from cell cycle arrest (unpublished data). However, after 20 d in culture, colonies that rapidly overtook the culture began to appear, and immunoblotting analysis revealed that these cells had failed to express FOXO1a-TM-ERTAM (Fig. 9 C), underscoring the essential selection for loss of FOXO1a in ARMS.
To test the effects of enforced FOXO1a expression on the tumorigenicity of Rh30 cells, we injected male NOD/SCID mice with 5 × 106 Rh30 cells transduced with either the retroviral vector alone (Rh30 vector; n = 4) or a retrovirus harboring FOXO1a-TM-ERTAM (Rh30-FOXO1a-TM-ERTAM; n = 8). Within 5 wk, the average diameter of tumors arising from the Rh30-vector cells was 22 mm (range 20–24 mm), whereas that of tumors arising from Rh30-FOXO1a-TM-ERTAM cells was 12.5 mm (range 9–17 mm). The smaller tumors arising from Rh30-FOXO1a-TM-ERTAM cells reflected their reduced rates of proliferation (Fig. 9 A), again most likely because of the leakiness of the ERTAM system.
To assess whether loss of FOXO1a function was necessary to maintain Rh30 xenografts, mice bearing established (5-wk) tumors derived from Rh30-vector cells or Rh30-FOXO1a-TM-ERTAM cells were injected daily with 0.1 mg of tamoxifen citrate. After 1 wk of treatment the percentages of tumor cells in the S phase in vivo was determined by BrdU incorporation. Strikingly, the percentage of Rh30-FOXO1a-TM-ERTAM tumor cells in S phase (5%) was substantially lower than that of cells in Rh30-vector tumors (Fig. 9, D and E, 40 and 50%). The remaining mice were then treated with a higher daily dose of tamoxifen citrate (0.6 mg) for 9 d. Within this short interval there was a 50% reduction in overall tumor size in mice bearing Rh30-FOXO1a-TM-ERTAM–derived xenografts, and two of five tumors regressed completely (Fig. 10 A). By contrast, this treatment failed to affect the growth of xenografts in mice bearing Rh30-vector cells; indeed, these tumors grew by more than 50% during the same interval.
The tamoxifen-dependent reduction in volume of Rh30-FOXO1a-TM-ERTAM tumors suggested that FOXO1a induced apoptosis in these cells in vivo. Indeed, TUNEL analysis showed that 20% of the cells in Rh30-FOXO1a-TM-ERTAM tumors were TUNEL positive after tamoxifen treatment, whereas less than 1% of the cells in the Rh30-vector tumors were TUNEL positive (Fig. 10, B and C). To confirm that cell death was caused by apoptosis, indirect immunofluorescence assays using an antibody specific for active Casp3 was performed (Fig. 10, D and E). Again, ∼20% of the Rh30-FOXO1a-TM-ERTAM tumor cells were positive for activated Casp3, whereas active Casp3 was not detected in Rh30-vector tumor cells.
The findings reported herein establish that FOXO1a expression is suppressed in ARMS and that this transcription factor functions as a potent tumor suppressor when reintroduced in ARMS tumor cells. Interestingly, ERMS cells are insensitive to the tumor suppressor effects of FOXO1a, a finding that underscores the concept that ARMS and ERMS represent disparate forms of RMS that arise by independent means and therefore require different therapeutic strategies to improve outcome.
Restoration of FOXO1a activity blocks the in vitro growth of all ARMS tumor cell lines tested, a finding consistent with its known role as a regulator of cell proliferation (Accili and Arden, 2004). In addition, three of four ARMS cell lines engineered to express FOXO1a underwent rapid and complete apoptosis, whereas Rh30 ARMS cells displayed total cell cycle arrest as well as morphological changes reminiscent of muscle cell differentiation. Restoration of FOXO1a expression in ARMS directly induces Casp3 transcription, which is also required for muscle cell differentiation (Fernando et al., 2002), yet this response also often results in cell death, most likely because of the transformed nature of ARMS. Thus, FoxO1a regulates the differentiation of normal early mouse myoblasts (Bois and Grosveld, 2003; Nishiyama et al., 2004) and exerts the same role in transformed ARMS cells. Although different pathways might be switched on or off in t(2;13) ARMS cell lines, which model secondary tumors, the observation that all lines are sensitive to FOXO1a suggests that the loss of this protein is a pivotal step in ARMS tumorigenesis.
These analyses also revealed that ERMS represents a tumor that arises from a progenitor that can express late differentiation markers of muscle cell differentiation, such as MyoHC. The ERMS phenotype is also associated with the expression of one or more FOXO proteins, and FOXO1a is necessary for the muscle cell differentiation program (Bois and Grosveld, 2003; Nishiyama et al., 2004). In addition, FOXO1a activity appears to be regulated in ERMS tumors because overexpression of FOXO1a in ERMS cell lines has no deleterious effects. In particular, the ability of ERMS cell lines to repress Casp3 is reminiscent of the later stages of muscle differentiation (Fernando et al., 2002). Collectively, these observations underscore the differences in the oncogenic pathways involved in the establishment of ARMS and ERMS.
Our results also support the concept that a haploid FOXO1a gene dose and subsequent or coincidental loss of FOXO1a protein expression probably cooperate with oncogenic signals emanating from the PAX3-FOXO1a chimeric transcription factor to induce malignant ARMS. Loss of FOXO1a allows for the bypass of important checkpoints that induce cell cycle arrest (e.g., induction of the cell cycle inhibitor p27Kip1; Dijkers et al., 2000; Medema et al., 2000; Kops et al., 2002) or apoptosis (e.g., induction of caspase-3). Loss of FOXO1a also compromises proper remodeling of the cytoskeleton, which is necessary for muscle cell fusion and adhesion, both of which have been linked to the activity of Casp3 (Fernando et al., 2002). Thus, the differentiation block in ARMS can, at least in part, be attributed to a loss of the FOXO1a–Casp3 pathway.
Most important, the finding that secondary ARMS tumors, as modeled by the cell lines used in this study, are sensitive to FOXO1a expression suggests novel therapeutic opportunities to fight this deadly disease. Specifically, these findings suggest that FOXO1a loss of function is an essential event in ARMS tumorigenesis and that agents aimed at restoring or augmenting FOXO1a activity in ARMS cells might be of therapeutic benefit.
Materials And Methods
Early passage ARMS and ERMS cell lines (provided by S. Ragsdale and G. Germain, St. Jude Children's Research Hospital, Memphis, TN) were cultured in RPMI growth medium supplemented with 10% FBS. All experiments were performed in this medium. The C2C12 cell line was obtained from American Type Culture Collection. We used passages 1 and 2 cells and took particular care to avoid cell overgrowth, which quickly induces differentiation. C2C12 cells were cultured in DME supplemented with 10% FBS by using standard procedures. Differentiation was then induced by transferring cells to DME supplemented with 2% horse serum. Primary myoblasts were isolated from 6-d-old C57Bl/6 mice and cultured as described previously (Bois and Grosveld, 2003). Primary tumor samples were provided by the St. Jude Children's Research Hospital cell bank.
Real-time RT-PCR, protein, and ChIP analyses
Real-time RT-PCR and Western blotting were performed using standard protocols (Ausubel et al., 2001). FOXO1a (FKHR), AKT, P-AKTThr308, P-AKTSer473, and active caspase-3 antibodies were purchased from Cell Signaling Technology, Inc.; anti–procaspase-3 antibody from BD Biosciences; anti-AU1 tag from Abcam, Inc.; anti-FOXO3a (FKHRL1) antibody from Upstate Biotechnology; anti-GFP antibody from Molecular Probes, Inc.; and anti-FLAG antibody from Sigma-Aldrich. The anti-FOXO4 antibody was provided by B. Burgering (University Medical Center, Utrecht, Netherlands), and the anti-MyoHC MF20 antibody was a gift from D. Fischman (Weill Medical College of Cornell University, New York, NY). Fast Green protein dye was purchased from Sigma-Aldrich.
ChIP experiments were performed using the ChIP assay kit (Upstate Biotechnology) according to the manufacturer's protocol.
Microscopy and image acquisition
A microscope (BX51; Olympus) fitted with 20 and 40× lenses (Olympus) was used for imaging. Digital images were acquired using a SPOT RTse camera and its acquisition software (v4.0.4) for Mac OSX. Subsequent image processing (cropping, rotating, contrast, and intensity adjustments) was performed using Adobe Photoshop CS (v8.0) for Mac OSX. DNA was stained with a 10,000 dilution of a DAPI stock solution (Sigma-Aldrich). Fluorescent secondary antibodies were purchased from Molecular Probes, Inc.
Luciferase reporters carrying the mouse −3.8/−3.3 or +1.0/+1.5 kb caspase-3 promoter regions containing the FoxO1a binding sites as well as constructs deleted for these sites were generated by inserting PCR-amplified fragments upstream of the minimal promoter of the pGL3-basic firefly luciferase vector (Promega). C2C12 myoblasts were transiently transfected with the promoter–reporter constructs using FuGene6 (Roche Molecular Systems, Inc.) following the supplier's protocol. Luciferase detection was performed using the Luciferase Assay System (Promega) as per the manufacturer's recommendations. A β-actin promoter–driven secreted alkaline phosphatase (SEAP) reporter was cotransfected (100 ng per transfection) to normalize transfection efficiency; SEAP activity was determined as described previously (Berger et al., 1988). To avoid vector squelching, we transfected empty MSCV with the reporter construct when exogenous expression of FoxO1a was not required.
Apoptosis and caspase-3 activity assays
Apoptosis was measured by staining with phycoerythrin-conjugated annexin V according to the manufacturer's instructions (BD Biosciences). TUNEL assays were performed using the ApopTag kit (InterGen) as per the manufacturer's protocol. To analyze Casp3 activity, we collected floating cells, combined them with cells growing on the dish, and washed them twice with PBS. The cells were lysed in caspase lysis buffer (25 mM Hepes-NaOH, 0.1% sucrose, 1% CHAPS, 2 mM EDTA, and 10 mM dithiothreitol; pH 7.4). Cell lysates were mixed with caspase assay buffer (25 mM Hepes-NaOH; pH 7.4), 10 mM dithiothreitol, and the fluorogenic substrate Ac-DEVD-AMC (50 μm; caspase-3). After incubation at 37°C for 1 h, the fluorometric detection of released AMC product was performed on a Fluorescence Multi-well Plate Reader (CytoFluor Series 2350; Millipore) using a 400-nm excitation filter and a 530-nm emission filter.
Virus production, cell transduction, and cell sorting
D. Persons (St. Jude Children's Research Hospital, Memphis, TN) provided the MSCV-IRES-GFP vector. M. McMahon (University of California, San Francisco, San Francisco, CA) and K. Helin (European Institute of Oncology, Milan, Italy) provided the pBabe vectors. Dominant-negative FADD cDNA and AU1 epitope-tagged dominant-negative FADD (NFD-4; a gift from J.M. Lahti and V.J. Kidd, St. Jude Children's Research Hospital, Memphis, TN) were cloned into pMSCV-IRES-GFP. All constructs were generated using standard molecular biology protocols (Ausubel et al., 2001). Cells were transduced using standard methods (Ausubel et al., 2001), and after 2–3 d, transduced cells were FACS sorted for GFP expression or were selected in puromycin-containing medium.
NOD/SCID mice were purchased from The Jackson Laboratory. Mice were maintained at St. Jude Children's Research Hospital under the Institutional Animal Care and Use Committee guidelines. Generation of xenografted animals and tamoxifen-sensitive animals was performed as previously described (Houghton et al., 1995; Hayashi and McMahon, 2002).
We thank the staff of the tumor bank at St. Jude Children's Research Hospital for providing the primary RMS samples and Glenn Germain and Susan Ragsdale for providing RMS cell lines and helpful advice.
This research was supported in part by grants from the National Cancer Institute (CA-71907, CA-87952, CA-96696, and CA-23099), the Cancer Center Support (CA-21765), the Van Vleet Foundation of Memphis, and the American Lebanese Syrian Associated Charities. P.R.J. Bois is a fellow of the Van Vleet foundation of Memphis.
Abbreviations used in this paper: ARMS, alveolar RMS; Casp, caspase-3; ChIP, chromatin immunoprecipitation; ER, estrogen receptor; ERMS, embryonal RMS; FOXO1a-TM, dominant-active FOXO1a; FOXO1a-TM-ERTAM, tamoxifen-regulated form of FOXO1a-TM; FOXO1a-WT, wild-type FOXO1a; MyoHC, myosin heavy chain; RMS, rhabdomyosarcoma; SEAP, secreted alkaline phosphatase.