MyoD and E-protein heterodimers switch rhabdomyosarcoma cells from an arrested myoblast phase to a differentiated state

Abstract

Rhabdomyosarcomas are characterized by expression of myogenic specification genes, such as MyoD and/or Myf5, and some muscle structural genes in a population of cells that continues to replicate. Because MyoD is sufficient to induce terminal differentiation in a variety of cell types, we have sought to determine the molecular mechanisms that prevent MyoD activity in human embryonal rhabdomyosarcoma cells. In this study, we show that a combination of inhibitory Musculin:E-protein complexes and a novel splice form of E2A compete with MyoD for the generation of active full-length E-protein:MyoD heterodimers. A forced heterodimer between MyoD and the full-length E12 robustly restores differentiation in rhabdomyosarcoma cells and broadly suppresses multiple inhibitory pathways. Our studies indicate that rhabdomyosarcomas represent an arrested progress through a normal transitional state that is regulated by the relative abundance of heterodimers between MyoD and the full-length E2A proteins. The demonstration that multiple inhibitory mechanisms can be suppressed and myogenic differentiation can be induced in the RD rhabdomyosarcomas by increasing the abundance of MyoD:E-protein heterodimers suggests a central integrating function that can be targeted to force differentiation in muscle cancer cells.

Figure 1.

MSC competes with MyoD for a limiting amount of E2A. (A) MSC copurifies with E12 but not with MyoD from RD cells. RD cells stably expressing TAP tag, or TAP-tagged MyoD or E12 were differentiated for 36 h before nuclear extraction, protein A-mediated pull-downs, and the TEV cleavage to release MyoD or E12 complexes from the IgG column (see Supplemental Fig. 1). The TEV cleavage removes the protein A from the TAP tag and leaves the calmodulin-binding domain (cbd) on MyoD or E12 as MyoDcbd or E12cbd. The TEV-released proteins were immunoblotted for E2A, MyoD, or MSC, respectively. Asterisks mark spill-over from the MyoD-cTAP pull-down. (B) The endogenous MSC in RD cells coimmunoprecipitates with E2A. Nuclear extracts from RD cells were immunoprecipitated with either MSC or E2A antibody and followed by immunoblotting of MSC, E2A, and MyoD. Normal goat serum was used as the negative control for antibody against MSC and mouse IgG for antibody against E2A in the CoIP. (C, left panel) EMSA shows that in vitro translated MyoD and E12, or E12 and MSC bind the MCK “right” E box as heterodimers. EMSA with nuclear extract from RD cells shows shifted species that are enhanced by the addition of in vitro translated E12 (cf. lanes 2 and 4), and correspond to MyoD:E12 and MSC:E12 heterodimers, as demonstrated by supershift assays (lanes 5–7), indicating that free E-protein is limiting for MyoD and MSC dimerization in RD cells. Notice the intensity of the corresponding band to either MyoD:E12 or MSC:E12 is reduced upon addition of the antibody. (D) MSC represses MyoD:E12-mediated activation on the CKM (muscle creatine kinase) reporter in RD cells. Transient transfections of RD cells with MyoD and/or E12 were done in the presence of increasing amount of MSC.

Figure 2.

A novel E2A splice variant, E2A-2/5, isolated from RD cells activates less than the full-length E2A. (A) Sequencing identified a novel E2A splice variant, E2A-2/5, in RD cells missing exons 3 and 4. The top panel shows schematic domain representations of E2A (E12/E47) and E2A-2/5 proteins. The bottom panel shows the amino acid sequence at the location of exon 3 and 4 in the activation domain 1 of E2A. (B) E2A-2/5 and MSC were detected in multiple independent RMS cell lines, primary tumors, and normal human primary myoblasts. Total RNAs isolated from cell lines 1-RD (e), 2-Rh1 (e), 3-RH18 (e), 4-RHJT (a), 5-RH28 (a); primary tumors 6-RMS60 (e), 7-RMS61 (a), 8-RMS62 (a), 9-RMS63 (a), 10-RMS64 (e), 11-RMS65 (e); and human primary myoblast cell lines 12–14 were subject to RT–PCR to detect expression level of MSC and E2A. The “e” represents the embryonal and the “a” represents the alveolar subtype of RMSs. (C) EMSA using in vitro transcribed and translated (TnT) products showed that E12-2/5 dimerizes with MyoD similar to the full-length E12 (shown in lanes 4,5). The full-length E12 binds the E-box as a homodimer, whereas the E12-2/5 does not show homodimer binding but does have a faint slower migrating species that might be a higher-order multimer (cf. lanes 2 and 3). E2A-2/5 prevents the full-length E2A homodimers from binding to DNA (cf. lanes 10 ,11 and 9, cf. lanes 16,17 and 15). Numbers in percentage represent how much TnT product was used for each EMSA reaction. Lanes 19 and 20 represent reticulocyte alone and probe alone, respectively. (D) MyoD:E2A-2/5 activates less efficiently in RD cells than MyoD:E2A. Transient transfections of RD cells with MyoD and/or E2A splice variants were done in the presence of 4RTK or CKM luciferase reporter and pCS2-βLacZ as an internal control. (E) In the 3T3 cell, MyoD:E2A-2/5 activates the myogenic program less efficiently than MyoD:E12, estimated by the numbers of myosin heavy chain-positive cells. The amount of MyoD remained constant (0.75 μg) while increasing the amount of E12 or E2A-2/5 as indicated by the gradient triangles from 0.05 to 0.25 μg. (F) Western blot showing relative amounts of E2A and E2A-2/5 in RD cells. Lanes 1 and 2 are cells transfected with hE12 and hE12-2/5 to show migration. (N) Endogenous E-protein in untransfected RD cells; [ST1(si)] RD cells transfected with siRNA to E2A-2/5 (see Supplemental Fig. S5A); [N(si)] control scrambled siRNA.

Figure 3.

A dnMSC or a MyoD∼E12 forced dimer drive RD cells into differentiation. (A) RD cells were infected with retroviruses expressing either vector alone or dnMSC. Infected RD cells were selected in puromycin and differentiated for 0 h, 24 h, and 48 h in insulin transferrin medium. Myogenesis was estimated by immunostaining for myosin heavy chain (MHC) (differentiation at 0 and 48 h shown) and nuclei were stained by DAPI. (B) Corresponding expression of muscle creatine kinase (CKM) was estimated by qPCR. Values were normalized to Timm17b, an RNA that does not change with myogenic differentiation. (C) Luciferase assay was conducted on RD cells transiently transfected with MyoD and/or E12 or MyoD∼E12 forced dimer. (D) Myosin heavy chain (MHC) and DAPI staining of RD cells infected by retroviruses expressing either a control or MyoD∼E forced dimer. (E) qPCR estimated CKM expression in RD cells infected by retroviruses expressing either vector alone, MyoD, E12, or MyoD∼E12 dimer. (F) Relative folds of enrichment of MyoD binding on the Myh8 (myosin heavy chain 8) promoter and the CKM enhancer averaging three independent ChIP experiments from cells infected by retroviruses expressing MyoD, E12, or the MD∼E12 forced dimer verses the control virus at 24 h of differentiation.

Figure 4.

MyoD∼E12 forced dimer broadly inhibits myogenic repressors. (A) Reduced E2A and MSC expression in RD cells differentiated by retrovirus-mediated expression of MD∼E12 forced dimer. Correspondingly, expression of myosin heavy chain (MHC) is strongly enhanced. Whole-cell lysate (WCL) was immunoblotted for MyoD∼E12 dimer, MHC, E2A, and α-tubulin as the loading control. Nuclear extract (NE) was immunoblotted for MSC and β-actin as the loading control. (B) Duplex RT–PCR showed a preferential reduction in E2A-2/5 mRNA level but not the full-length E2A mRNA. Timm17b is a loading control. (C) Immunoblotting showed that expression of Pax7 is reduced in RD cells differentiated by MyoD∼E12 forced dimer. RT–PCR showed down-regulation of Dec1, Prominin1, and Mdfi expression. GAPDH serves as an internal control for RT–PCR.

Figure 5.

A competitive equilibrium between antagonist, partial agonist, and active bHLH dimers modulate MyoD:E2A activity to maintain a myoblast state, whereas shifting the equilibrium to favor MyoD:E2A heterodimers activates a process (factor X or signal X) that broadly suppresses the inhibitors and locks in a differentiated state.