PAX3-FOXO1 controls expression of the p57Kip2 cell-cycle regulator through degradation of EGR1

Abstract

The chimeric protein PAX3-FOXO1, resulting from a translocation between chromosomes 2 and 13, is the most common genetic aberration in the alveolar subtype of the human skeletal muscle tumor, rhabdomyosarcoma. To understand how PAX3-FOXO1 contributes to tumor development, we isolated and characterized muscle cells from transgenic mice expressing PAX3-FOXO1 under control of the PAX3 promoter. We demonstrate that these myoblasts are unable to complete myogenic differentiation because of an inability to up-regulate p57Kip2 transcription. This defect is caused by reduced levels of the EGR1 transcriptional activator resulting from a direct, destabilizing interaction with PAX3-FOXO1. Neither PAX3 nor FOXO1 share the ability to regulate p57Kip2 transcription. Thus, the breakage and fusion of the genes encoding these transcription factors creates a unique chimeric protein that controls a key cell-cycle and -differentiation regulator.

Rhabdomyosarcoma constitutes a group of soft tissue sarcomas of childhood and adolescence that are thought to arise from undifferentiated mesenchyme resembling various stages of early embryonic skeletal muscle development. The most aggressive pediatric subtype, alveolar rhabdomyosarcoma (ARMS), is composed of dense aggregates of poorly differentiated cells separated by a framework of fibrous septa forming “alveolar” spaces. Typical features of ARMS include: physical location of the tumor, alveolar appearance, the presence of the characteristic translocation between chromosomes 2 and 13, and immunohistochemical reactivity for the myogenic markers desmin, MYOD1, and myosin heavy chain (1). This latter feature might suggest that ARMS arises as a consequence of incomplete myogenic differentiation and abnormal proliferation coupled to transforming mutations.

The PAX3-FOXO1 fusion protein, created by the t(2;13) chromosomal translocation, is present in most cases of alveolar rhabdomyosarcoma (2). The translocation severs the transcriptional transactivation domain of PAX3 but preserves both of its two DNA-binding domains. The FOXO1 gene is disrupted in a large intron that bisects its DNA-binding domain. The chromosomal rearrangement creates a chimeric protein containing the transcriptional activation domain of FOXO1 and the DNA-binding elements of PAX3, under the control of the PAX3 promoter (3).

PAX3 is an essential myogenic regulator. Mice lacking Pax3 have multiple skeletal muscle defects, most notably delays in muscle differentiation as well as an overall decrease in muscle mass (4). In myogenesis, PAX3 functions to induce the expression of SIX1 and EYA2 (5), transcription factors that up-regulate expression of MYF5, advancing the myogenic differentiation program. FOXO1 also plays a pivotal role in mediating myogenic differentiation (6, 7).

In myoblasts, terminal differentiation and proliferation are mutually exclusive processes. p57KIP2 promotes differentiation by stabilizing MYOD1, inhibiting cyclin E-CDK2 activity and proliferating cell nuclear antigen (PCNA) function while maintaining RB1 in an active hypophosphorylated state (8–11). p57KIP2, located at 11p15, is a paternally imprinted gene whose decreased expression is a feature common to a variety of human tumors (12–17). There are several mechanisms through which this occurs, commonly involving changes in genomic imprinting or loss of the active maternal allele (18–20). It is interesting to note that the embryonal subtype of rhabdomyosarcoma is characterized by loss of the maternal 11p15 chromosomal region (21). Loss of p57KIP2 function is also implicated in Beckwith–Wiedemann syndrome, a complex overgrowth condition associated with an increased risk for developing rhabdomyosarcomas (22).

Whereas some potential PAX3-FOXO1 downstream targets have been identified, the mechanism by which PAX3-FOXO1 contributes to tumor pathogenesis is unknown. The FOXO1 transactivation domain has been shown to be more robust than that of PAX3, enabling PAX3-FOXO1 to more strongly activate PAX3 consensus sequence reporter constructs (23–25). These reports have led to the hypothesis that PAX3-FOXO1 drives oncogenesis by overactivating PAX3 transcriptional targets (24, 26–29). A recent study found that PAX3-FOXO1 bypasses cellular senescence by reducing p16^INK4A levels (30), whereas others have suggested that expression of PAX3-FOXO1 results in aberrant regulation of genes involved in myogenic differentiation (reviewed in ref. 31). The present studies were undertaken to identify PAX3-FOXO1 target genes in transgenic myoblasts, to test whether these affect myogenic differentiation, and to determine how PAX3-FOXO1 expression leads to their misregulation.

Results

PAX3-FOXO1 Expression in Myoblasts Inhibits Differentiation and Decreases p57Kip2.

To study the role of PAX3-FOXO1 on myogenic differentiation, we isolated myoblasts from mice harboring the PAX3-FOXO1 fusion gene under the control of the PAX3 promoter (32). PAX3-FOXO1 transgenic myoblasts are phenotypically indistinguishable from their wild-type counterparts and express normal levels of the myoblast marker, desmin (33).

Rhabdomyosarcomas are characterized by deficiencies in myogenic differentiation and an inability to exit the cell cycle (34). Cells derived from these tumors express MYOD1 and myogenin but do not differentiate into myotubes (34). To determine if myoblasts derived from PAX3-FOXO1 transgenic animals have similar defects, wild-type and transgenic myoblasts were cultured in media that induce myogenic differentiation. As shown in Fig. 1 A and B, PAX3-FOXO1 transgenic myoblasts differentiate poorly, they fail to form myotubes, and they do not up-regulate expression of the terminal differentiation marker, myosin heavy chain.

Fig. 1.

PAX3-FOXO1 transgenic myoblasts contain low levels of p57KIP2 and are unable to differentiate. (A) (Upper) Proliferating (prolif.) myoblasts, seeded at constant density. (Lower) Myoblasts plated at a constant density and maintained in differentiation (diff.) medium for 24 h. Number of myotubes per high-power field (320×) is given below the representative photograph. A minimum of four fields were counted. (B) Immunoblot analysis of the myogenic markers myosin heavy chain (MHC) and desmin in wild-type and transgenic myoblasts maintained in either proliferation or differentiation medium. (C) Quantitative PCR analysis of p57Kip2 mRNA from early passage myoblasts. P values of the differences between wild-type and transgenic (tg.) myoblasts were calculated by using two-tailed Student's t test. Error bars represent SDs. (D) Immunoblot analysis of p57KIP2 protein levels in wild-type or transgenic myoblasts maintained in either proliferation or differentiation medium. (E) Quantitative PCR analysis of p57Kip2 mRNA accumulation during differentiation. Values shown are the mean fold up-regulation, ± SD.

The inability of transgenic myoblasts to differentiate led us to hypothesize that the PAX3-FOXO1 fusion gene might play a role in suppressing differentiation and preventing cell-cycle exit. To address the nature of the differentiation defect, expression profiling with Genechip microarrays was performed on two sets each of passage-matched primary myoblasts from transgenic and wild-type animals. For both data sets, among many genes with altered expression, the most profound effect was decreased expression of the CDK inhibitor, p57Kip2. This result was validated by using quantitative PCR on independent primary myoblasts from PAX3-FOXO1 transgenic animals, as shown in Fig. 1C. Wild-type myoblasts showed an ≈150-fold increase of p57Kip2 expression upon differentiation induction, whereas the transgenic myoblasts had <15% of that response (Fig. 1E), and this corresponded to the levels of p57KIP2 protein in the cells (Fig. 1D). These results suggest that PAX3-FOXO1 might influence the balance between myogenic differentiation and proliferation by decreasing the levels of the proliferation inhibitor, p57KIP2.

To confirm that diminished quantities of p57KIP2 are sufficient to inhibit differentiation, we restored p57Kip2 expression in PAX3-FOXO1 transgenic myoblasts. Early passage transgenic myoblasts were infected with either empty vector or p57Kip2 retroviruses. p57Kip2 transcript levels were increased to slightly less than the levels seen in wild-type myoblasts, ≈4-fold. As shown in Fig. 2, restoration of p57Kip2 expression profoundly improves the ability of PAX3-FOXO1 transgenic myoblasts to differentiate (P = 0.0008). This result demonstrates that p57KIP2 is a major effector of PAX3-FOXO1 in inhibiting myogenic differentiation.

Fig. 2.

Restoration of p57Kip2 levels in PAX3-FOXO1 transgenic myoblasts promotes their differentiation. (A) (Upper) Proliferating (prolif.) myoblasts, seeded at constant density. (Lower) Myoblasts plated at a constant density and maintained in differentiation (diff.) medium for 24 h. Number of myotubes per high-power field is given below the representative photograph (320×). A minimum of four fields were counted. (B) Immunoblot analysis of myosin heavy chain (MHC) in control and p57Kip2-transduced transgenic myoblasts maintained in either proliferation or differentiation medium. (C) Quantitative PCR analysis of p57Kip2 mRNA from proliferating control and p57Kip2 transduced transgenic myoblasts.

The p57Kip2 Promoter Is Responsive to PAX3-FOXO1.

To determine whether the p57Kip2 promoter is PAX3-FOXO1 responsive and which of its elements are responsible for the transcriptional repression, a series of deletions of the full-length mouse promoter linked to a luciferase transcriptional reporter were constructed (Fig. 3A). Sequence analysis showed that there are two putative FOXO1-binding sites and one putative PAX3 binding site in the full-length promoter, at positions −2,130, −2,650, and −2,900, and all were contained in the PAX3-FOXO1-responsive full-length p57Kip2 promoter construct. However, the deletion of a segment of the p57Kip2 promoter containing all of these sites (−3000 to −1800 from the transcriptional start site) did not cause the p57Kip2 promoter to become unresponsive to PAX3-FOXO1 inhibition. On the contrary, PAX3-FOXO1-responsive sites were scattered throughout the promoter, with the magnitude of the repression diminishing with decreasing promoter length. To further define the minimal sequence required for PAX3-FOXO1-dependent repression, additional deletion constructs were created. As shown in Fig. 3B, deletion of 100 nucleotides from −400 to −300, with respect to the transcriptional start site of the minimal p57Kip2 promoter, renders it PAX3-FOXO1 insensitive. These 100 nucleotides of sequence are also sufficient to mediate repression by PAX3-FOXO1 when placed upstream of a synthetic minimal promoter (Fig. 3B). Moreover, this ability is specific to the PAX3-FOXO1 fusion protein and is not evidenced by wild-type PAX3 or FOXO1 alone (Fig. 3C). These results indicate that PAX3-FOXO1 represses p57Kip2 transcription through sequences in the −400 to −300 region of the promoter that are distinct from PAX3 or FOXO1 binding sequences and that are unresponsive to PAX3 or FOXO1.

Fig. 3.

PAX3-FOXO1 represses p57Kip2 transcription through multiple GC-rich sequence elements. (A) Luciferase reporter assays of sequential deletions of the p57Kip2 promoter, cotransfected with increasing quantities of PAX3-FOXO1. The relative positions of the PAX3 (P)- and FOXO1 (F)-binding sites are indicated on the schematic. Values are expressed as mean firefly luciferase activity normalized to control Renilla luciferase activity. (B) Luciferase reporter assays of p57Kip2 sequential deletions, cotransfected with increasing quantities of PAX3-FOXO1. The −400 to −300 bp p57Kip2 promoter sequence is upstream of a minimal tk promoter. Values are expressed as mean fold activation relative to promoter cotransfected with empty vector. (C) Luciferase reporter activity of PAX3, FOXO1, and PAX3-FOXO1 on the p57Kip2 promoter, −400 to +30 bp, normalized to background activity on −300 to +30 bp. Values are expressed as mean fold activation relative to promoter cotransfected with empty vector. Results of all experiments are the mean values ± SD of triplicates and representative of at least three independent experiments.

PAX3-FOXO1 Repression of the p57Kip2 Promoter Is Mediated by EGR1-Binding Sequences.

The minimum sequence of the p57Kip2 promoter required for PAX3-FOXO1 repression, −400 to −300 bp, is rich in GC nucleotides, unlike the PAX3- or FOXO1-binding sites that were the anticipated PAX3-FOXO1 targets. We performed chromatin immunoprecipitation several times by using endogenous PAX3-FOXO1 from transgenic myoblasts but were unable to amplify p57Kip2 promoter sequences (data not shown). When combined with the absence of a PAX3-binding site in this region, these data suggest that PAX3-FOXO1 controls p57Kip2 expression indirectly. An alternative mechanism by which PAX3-FOXO1 might repress p57Kip2 transcription without directly binding to the p57Kip2 promoter could be by interfering with an activator of p57Kip2 transcription. Because the region of the p57Kip2 promoter from −400 to −300 bp is extremely GC-rich, we assessed three of the most-common GC box-binding transcription factors, EGR1, SP1, and SP3, for their ability to activate p57Kip2 transcription. As shown in Fig. 4A, only EGR1 was able to activate p57Kip2 transcription in a dose-dependent fashion. This result is consistent with microarray experiments that show that p57Kip2 is up-regulated by EGR1 (35, 36).

Fig. 4.

PAX3-FOXO1 interferes with Egr1-dependent transcription. (A) Luciferase reporter assays of EGR1, SP1, and SP3 on the full-length p57Kip2 promoter. Values are expressed as fold activation relative to promoter cotransfected with empty vector. (B) EMSA of in vitro transcribed/translated EGR1 on −400 to −350 bp p57Kip2 promoter. Competing DNA was either wild-type sequence from −400 to −350 bp of the p57Kip2 promoter (Left), or mutated by disruption of EGR1-binding sites (Right). (C) Luciferase reporter assay of the full-length p57Kip2 promoter cotransfected with an increasing quantity of EGR1 (E), with or without PAX3-FOXO1. (D) Luciferase reporter assay of an EGR1 synthetic reporter cotransfected with an increasing quantity of EGR1, with or without PAX3-FOXO1. For all luciferase reporter assays, results are the mean values ± SD of triplicates. Results are representative of at least three independent experiments.

To establish whether EGR1 acts directly on the p57Kip2 promoter, EMSA was performed. We used in vitro-translated EGR1 and a radiolabeled oligonucleotide representing −400 to −350 bp of the p57Kip2 promoter sequence, because this sequence contained the majority of the EGR1-binding sites in this region. Whereas EGR1 could clearly shift the mobility of the oligonucleotide, disruption of the two EGR1-binding sites in the unlabeled, competing DNA abrogates its ability to efficiently compete for EGR1 binding (Fig. 4B). Thus, EGR1 can specifically bind to and activate p57Kip2 transcription through the minimal promoter sequence important for PAX3-FOXO1 repression.

To determine whether PAX3-FOXO1 could interfere with activation of the p57Kip2 promoter by EGR1, luciferase reporter assays were performed by using the full-length p57Kip2 promoter in the presence of increasing quantities of Egr1. PAX3-FOXO1 represses EGR1 activation of the p57Kip2 promoter ≈4-fold (Fig. 4C). Because PAX3-FOXO1 has the ability to interfere with EGR1-dependent activation of the p57Kip2 promoter, we tested whether it was able to more generally suppress EGR1-mediated transcription. Luciferase assays were performed with a synthetic reporter containing three tandem copies of the EGR1 consensus sequence. Consistent with the effects observed for the p57Kip2 promoter, PAX3-FOXO1 was able to suppress EGR1 activity by ≈4-fold (Fig. 4D), suggesting that this is one of its general activities.

PAX3-FOXO1 Destabilizes EGR1.

In the course of performing these experiments, we observed that cotransfection of PAX3-FOXO1 and Egr1 resulted in significantly reduced levels of EGR1. One mechanism by which PAX3-FOXO1 might interfere with EGR1-dependent transcription is by destabilizing EGR1. Cotransfection of increasing quantities of PAX3-FOXO1-HA in the presence of a fixed amount of Egr1-V5 showed that PAX3-FOXO1-HA reduced EGR1-V5 protein levels in a dose-dependent manner (Fig. 5A). Inhibition of the proteasome by the addition of MG132 abolished this effect, suggesting that the observed reduction of EGR1 protein is due to proteasomal degradation (Fig. 5A). Consistent with this, levels of Egr-V5 transcript were unaffected by PAX3-FOXO1 (data not shown). To further demonstrate that destabilization of EGR1 is specific for PAX3-FOXO1, we also performed this assay with PAX3 and FOXO1. As shown in supporting information (SI) Fig. 7, neither of these transcription factors alone has an effect on EGR1 stability.

Fig. 5.

PAX3-FOXO1 interacts with and destabilizes EGR1. (A) Immunoblot analysis of EGR1 V5 protein in the presence of increasing quantities of PAX3-FOXO1 HA, with or without the proteasome inhibitor MG132. (B) Luciferase reporter assay of the full-length p57Kip2 promoter, cotransfected with PAX3-FOXO1, in the presence of MG132. (C) Deletion mutants of PAX3-FOXO1, included residues given at left (A.A.), with their activity by luciferase reporter assay, as a mean percentage of wild type, shown at right, ± SD. The relative positions of the paired domain (P), homeodomain (H), and partial forkhead domain (F) are indicated on the schematic. (D) Coimmunoprecipitation analysis of epitope-tagged proteins, with or without MG132. (E) Coimmunoprecipitation of endogenous EGR1 and PAX3-FOXO1 from the cell line Rh28, in the presence of MG132. (F) Immunoblot analysis of Egr1 in wild-type (W.T.) and PAX3-FOXO1 transgenic (Tg.) myoblasts maintained in proliferation medium.

Because proteasome inhibition prevents PAX3-FOXO1 from destabilizing EGR1, inhibition of proteasome function would be expected to prevent the repression of p57Kip2 transcription by PAX3-FOXO1. To test this, reporter assays were performed in the presence of MG132 for 16 h; such proteasome inhibition largely reversed the repressive effect of PAX3-FOXO1 on p57Kip2 transcription (Fig. 5B). The inclusion of MG132 for the entire course of the assay, 48 h, might have completely restored p57Kip2 transcription, however this experiment could not be carried out because of the toxicity of the drug (37).

We next sought to determine which domains of PAX3-FOXO1 are required for suppression of p57Kip2 transcription by using a series of PAX3-FOXO1 deletions (Fig. 5C). Interestingly, PAX3-FOXO1, despite the loss of its primary DNA-binding domain (see construct 193-836, Fig. 5C), still retains much of its repressive activity. Additional deletions at either the N or the C terminus dramatically reduce most of the PAX3-FOXO1 transcriptional repression activity. This result suggests that the ability of PAX3-FOXO1 to suppress p57Kip2 transcription is not due to one of its particular domains and implies that the overall protein conformation might be responsible for suppression of p57Kip2 transcription.

To establish whether the destabilization of EGR1 by PAX3-FOXO1 involves their direct interaction, we performed coimmunoprecipitations of transfected epitope-tagged proteins (Fig. 5D). In the absence of proteasome inhibition, immunoprecipitation of PAX3-FOXO1-HA yields a faint smear when blots are probed for EGR1-V5. In the presence of MG132, EGR1-V5 coimmunoprecipitates with PAX3-FOXO1-HA, demonstrating that PAX3-FOXO1 and EGR1 directly associate in an unstable complex. We also performed the reciprocal experiment by immunoprecipitating EGR1-V5, resulting in a specific signal for PAX3-FOXO1 (data not shown). To demonstrate that the interaction between EGR1 and PAX3-FOXO1 was not simply a result of overexpression, we performed coimmunoprecipitation experiments on the endogenous proteins normally expressed by the ARMS cell line, Rh28, which harbors the t(2;13) translocation (38). Treatment of these cells with MG132 increases the levels of EGR1 protein (data not shown), and PAX3-FOXO1 and EGR1 form a complex in ARMS cells (Fig. 5E).

Lastly, we tested whether the observed decreases in p57Kip2 mRNA were correlated with reduced EGR1 protein levels in our transgenic myoblasts. As shown in Fig. 5F, immunoblot analysis revealed that PAX3-FOXO1 significantly reduces the levels of EGR1 protein in myoblasts.

Discussion

Here, we show that the accumulation of p57Kip2 mRNA normally induced by differentiation signals is suppressed by PAX3-FOXO1, thus rendering primary PAX3-FOXO1 myoblasts refractory to such stimuli. Taken together with other findings (21, 39, 40), our results suggest that loss of functional p57KIP2 is a common feature of both subtypes of rhabdomyosarcoma. P57Kip2-null mice display a variety of developmental defects resulting from an inability of cells to differentiate, but they do not exhibit an increased susceptibility to tumors (41). Analysis of the four PAX3-FOXO1 mouse models described to date (32, 42–44) also suggests that PAX3-FOXO1 expression alone is insufficient to produce a malignant phenotype. However, disruption of the Ink4a/ARF or Tp53 pathways, targets of inactivation in human rhabdomyosarcoma, in PAX3-FOXO1 mice substantially increases the frequency of tumor formation (42). These mouse models and our present data suggest that loss of function of p57KIP2 is not a dominantly acting transforming event. Rather, our data suggest that reduction of p57KIP2 levels by PAX3-FOXO1 is more likely to predispose cells to transformation by a secondary genetic event. PAX3-FOXO1 enables cells to bypass cellular senescence checkpoints through loss of p16^INK4a (30). This function, in concert with the failure of PAX3-FOXO1-expressing cells to differentiate, would create a large pool of proliferating cells primed for transformation. This mechanism of action is different from those described for any other cancer-related fusion gene (45).

It is interesting that, in this context, PAX3-FOXO1 controls p57Kip2 transcription not by acting as a transcription factor but, rather, by interfering with the activation elicited by the intermediary transcription factor, EGR1. We had expected, based on the current model of PAX3-FOXO1 function, that previously identified PAX3 transcription targets, some of which also appear to be PAX3-FOXO1 targets, such as MITF (46), RET (47), TYRP1 (48), MET (49), NCAM (50), and BCL-xL (28), would be up-regulated in PAX3-FOXO1 transgenic myoblasts. However, transcription of these genes was unchanged in PAX3-FOXO1 transgenic myoblasts, likely reflecting the somewhat lower level of PAX3-FOXO1 expression in our transgenic myoblasts as compared with human ARMS tumor cell lines, either of which is much less than obtained by ectopic overexpression. In our system, PAX3-FOXO1 expression is driven by the endogenous PAX3 promoter, as occurs in vivo. Therefore, the expression levels in PAX3-FOXO1 transgenic myoblasts should be equivalent to those that would be seen in affected myoblasts during the early stages of ARMS. Thus, this system should accurately model disease initiation. However, in transfection experiments using other cells that are wild type for Tp53, we have routinely had difficultly in achieving even modest levels of PAX3-FOXO1 expression. These findings, taken together with our PAX3-FOXO1 mapping data, suggest that PAX3-FOXO1 may function as a misfolded protein. If misfolded proteins accumulate to the extent that they overwhelm the chaperone system, then cells will undergo programmed cell death. Apoptosis occurs through JNK, primarily through stabilization of p53 (51). Thus, overexpression of PAX3-FOXO1 would be expected to be selected against in otherwise wild-type cells.

Our findings lead to the hypothesis that PAX3-FOXO1 contributes to rhabdomyosarcoma development by repressing the transcription of p57Kip2. Reduction of EGR1 protein levels by PAX3-FOXO1 results in inadequate quantities of p57KIP2 protein, preventing PAX3-FOXO1 transgenic myoblasts from completing the differentiation program (Fig. 6). Thus, translocation-positive myoblasts that are unable to exit the cell cycle, could establish a pool of proliferating precursor cells susceptible to a secondary transforming event, thus illustrating that tumor-specific genetic alterations could be central to the interplay between development and cancer. Finally, these results may have therapeutic applications, because they suggest that ARMS is, at least in part, a protein degradation-driven disease. Proteasome inhibitors, such as bortezomib, might be a new and effective approach for treating ARMS.

Fig. 6.

Model for PAX3-FOXO1-driven suppression of myogenic differentiation (adapted from ref. 54).

Materials and Methods

Plasmids.

Mammalian expression vectors encoding EGR1, PAX3-FOXO1, PAX3, FOXO1, and PAX3-FOXO1 mutants, both wild-type and epitope-tagged, were derived from pcDNA3.1 (Invitrogen, Carlsbad, CA). SP1 and SP3 plasmids were derived from pCMV. Epitope tags are as follows: the V5 tag (Invitrogen) was fused to the N terminus of EGR1 and PAX3-FOXO1; the 3xHA tag was fused the C terminus of PAX3, FOXO1, PAX3-FOXO1, and all of the PAX3-FOXO1 deletion mutants.

Mouse p57kip2 promoter deletions were constructed by using the firefly luciferase vector, pGL3 basic (Promega, Madison, WI). The Egr1 reporter, containing three tandem EGR1 consensus sites (GCGGGGGCG; see ref. 52) separated by spacers, was derived from the pLucMCS vector (Stratagene, La Jolla, CA).

Cell Culture.

All transfections were performed with human 293T cells, in DMEM with 10% FBS. Cells were transfected by using 1 μg of DNA and 8 μl of Lipofectamine 2000 (Invitrogen) per 3 × 10⁶ cells. For the reporter assays, 1 × 10⁶ cells were transfected with 0.5 μg of DNA and 1 μl of Lipofectamine 2000.

Western Blots.

Ten micrograms of protein were harvested in RIPA buffer [150 mM NaCl/1.0% Triton X-100/0.5% sodium deoxycholate/0.1% SDS/50 mM Tris, pH 8.0/complete protease inhibitor (Roche, Indianapolis, IN)], added to lithium dodecyl-sulfate (LDS) sample buffer (Invitrogen) separated on either 4–12% or 3–8% NuPAGE gels (Invitrogen), transferred to nitrocellulose, blocked (PBS containing 0.1% Tween 20 and 5% nonfat milk), and incubated for 1 h at room temperature with primary antibodies. Antibodies and sources are in SI Table 1.

Isolation of Mouse Myoblasts.

Isolation and culture of mouse myoblasts was performed as described in ref. 53. Briefly, hindlimb muscle tissue was dissected from 3-day-old mice, minced, and digested with Liberase Blendzyme 3 (Roche) followed by trypsin. After washing, this mixture was plated and cultured in myoblast growth media [20% FBS in a 50/50 mix of DMEM/Ham's F10 (Invitrogen) and 25 ng/ml bFGF (Dako, Glostrup, Denmark)]. Fibroblasts were removed by selective plating at each subsequent passage. For expression analysis, RNA was harvested from early passage cells (before or at passage 7) from cultures grown in triplicate with the RNeasy Plus Kit (Qiagen, Valencia, CA).

For differentiation experiments, cells were cultured in DMEM with 5% horse serum (Invitrogen). After 24 h (48 h for p57KIP2) in differentiation medium, protein was harvested in RIPA buffer [150 mM NaCl/1.0% Triton X-100/0.5% sodium deoxycholate/0.1% SDS/50 mM Tris, pH 8.0/complete protease inhibitor (Roche, Indianapolis, IN)], and RNA was extracted as above.

For viral transduction, viruses in the murine stem cell virus (MSCV) vector were obtained by cotransfecting 293T cells with an equal mass of MSCV plasmid and the envelope encoding vector, pCL-Eco (Imgenex, San Diego, CA). Viral supernatant was prepared 48 h posttransfection by passing the media through 0.45-μm filters. Myoblasts (3 × 10⁶ cells) were infected with fresh virus diluted 1:2 with myoblast growth media and 3.2 μg/ml fresh polybrene (Sigma, St. Louis, MO) for 8 h. Forty-eight hours postinfection, transduction efficiency was measured by maintaining a separate pool of myoblasts infected with GFP. The entire procedure was repeated four times to obtain a cell population >90% GFP-positive.

Quantitative Real-Time PCR.

Total RNA was harvested as above and reverse transcribed (SuperScript III First Strand Kit; Invitrogen). Quantitative PCR was performed on the iCycler IQ using IQ Syber Green (Bio-Rad, Hercules, CA) according to the manufacturer's instructions and default machine settings with an annealing temperature of 60°C (primer sequences can be found in SI Table 2).

EMSA.

EGR1 was transcribed and translated in vitro (TNT T7 Quick; Promega) in a reaction with 10 μM ZnSO₄ (Sigma). The EMSA was performed with 0.2 pmol of ³²P- labeled p57Kip2 promoter sequence (−400 to −350 bp) at 30°C for 30 min in a reaction buffer containing 20% glycerol, 5 mM MgCl2, 2.5 mM EDTA, 2.5 mM DTT, 250 mM NaCl, 50 mM Tris·HCl (pH 7.5), 0.25 mg/ml poly(dI-dC).poly(dI-dC), and 10 μM ZnSO₄. Reactions were separated on 6% DNA retardation gels (Invitrogen) that were dried and subjected to autoradiography.

Reporter Assays.

293T cells were cotransfected with the firefly luciferase reporter plasmid and the indicated combinations of expression plasmid. The Renilla luciferase plasmid pRL (Promega), driven by a minimal tk promoter, was included as an internal control. Forty-eight hours posttransfection, the Dual-Luciferase Reporter Assay (Promega) was performed according to the manufacturer's instructions, and values were read on the GENios Pro (Tecan, Durham, NC).

EGR1 Destruction Assay.

293T cells were transfected with a fixed mass of epitope-tagged EGR1 plasmid with increasing quantities of epitope-tagged PAX3-FOXO1 plasmid. Total DNA mass was kept constant by the addition of pcDNA3.1 plasmid. Sixteen hours before harvest, cells were treated with either 5 μM MG132 dissolved in DMSO or DMSO alone.

Coimmunoprecipitation.

293T cells were transfected with equal quantities of epitope-tagged PAX3-FOXO1 and EGR1 by using Lipofectamine 2000 according the manufacturer's instructions. Cells were lysed in Nonidet P-40 buffer and either 1 mg of lysate was incubated overnight with 2 μg of anti-V5 antibody or 2 mg of lysate was used with 2 μg of an anti-HA antibody in the presence of Dynabeads Protein G (Invitrogen). Immune complexes were washed [150 mM NaCl/50 mM Tris·HCl, pH 8/1% Nonidet P-40/0.5% sodium deoxycholate/0.05% SDS/complete protease inhibitor (Roche)]. Complexes were reduced, denatured, and eluted in LDS sample buffer.

For the coimmunoprecipitation of endogenously expressed proteins, the ARMS-derived cell line Rh28 (38) was used. Cells were treated with 20 μM MG132 (EMD Biosciences, San Diego, CA) for 2 h before harvesting in Nonidet P-40 buffer. For the EGR1 IP, 10 mg of protein was incubated overnight in the presence of 2 μg of the anti-EGR1 antibody (C19) with Dynabeads Protein A (Invitrogen).

Abbreviation

ARMS

alveolar rhabdomyosarcoma.