INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

The molecular basis for myotonic dystrophy (DM) pathogenesis has been the subject of investigations for several years. Although a number of pathways have been suggested and several animal models have been generated (12, 17, 27, 31), the molecular pathway(s) by which CTG repeats cause DM is not yet known. Among several animal models generated for verification of the molecular basis for DM, CTG/CUG transgenic mice generated recently show a severe phenotype (20). Mankodi et al. presented evidence that overexpression of RNA CUG repeats in transgenic mice is sufficient to cause skeletal muscle abnormalities typical of DM disease (20). Because the authors overexpressed pure CUG repeats (20), these observations also indicated that expression of expanded CUG repeats within the mutant DMPK transcripts is a critical event in DM pathology. The RNA-based hypothesis for DM pathogenesis was previously suggested by several investigators (6, 8, 25, 29, 32-36, 39) and was further confirmed by the identification of a number of RNA binding proteins that specifically interact with RNA CUG repeats (19, 21, 34, 35). Among those, a CUG triplet repeat binding protein, CUGBP1, has been identified as a candidate whose binding activity was affected by expansion of RNA CUG triplet repeats in patients with DM and in DM cell culture models (6, 25, 29, 34, 36). Recently, several other RNA binding proteins that are highly homologous to CUGBP1 have also been described. The CUGBP family includes CUGBP1 (35), ETR-3 (19), and Brunol-1 (10). CUGBP proteins are characterized by a high level of homology, by similar structural organizations, and by similar binding activities (10, 19, 34). Members of this family are very highly conserved and are expressed in a tissue-specific manner (6, 19). Sequence comparison analysis reveals that CUGBP1 has a high level of homology to a conserved family of ELAV RNA binding proteins that play a significant role in cell differentiation and development. A deletion of the elav (embryonic lethal abnormal visual phenotype) gene in Drosophila results in embryonic death (5). It has also been shown that deletion of ETR-1 in Caenorhabditis elegans is lethal (22). The phenotype of ETR-1 mutant embryos indicates that they are defective in muscle formation and function (22). These observations show that ETR-1 is the crucial factor for muscle formation in C. elegans and suggest that CUGBP family proteins might also play a role in the development and function of skeletal muscle in humans.

The ELAV family proteins regulate cell growth and differentiation via control of posttranscriptional RNA processing such as intracellular localization, stability, and mRNA translation (2). Human ELAV proteins bind to AU-rich regions located in the 3' untranslated region (UTR) of several mRNAs such as those encoding c-myc, c-fos, GLUT1 and p21 (3, 11, 13, 18). This binding increases the stability of the corresponding mRNAs and results in increased production of the protein. Numerous studies showed that ELAV-like proteins regulate cell growth and differentiation via regulation of mRNA stability or translation. Ectopic expression of the Hel-N1 protein in mouse 3T3-L1 cells leads to alterations specific for differentiated 3T3-L1 adipocytes (11). Hel-N1 induced differentiation of 3T3-L1 cells is caused by binding of Hel-N1 proteins to GLUT1 mRNA and recruitment of the mRNA into active polysomes (11). It has been recently shown that Hel-N1 also induces differentiation of human teratocarcinoma cells via increase of translation of neurofilament M mRNA and that the recruitment of mRNA into the heavy polysomal fraction is regulated by binding of Hel-N1 to the 3'UTR of the mRNA (3).

In this paper, we present evidence showing the role of RNA CUG repeat binding protein CUGBP1 in differentiation of skeletal muscle. CUGBP1 regulates the translation of mRNA coding for an inhibitor of the cell cycle, p21, which is an important regulator of skeletal muscle differentiation. Our data demonstrate that CUGBP1 binds to GCN repeats located in the 5' region of p21 mRNA and induces the production of p21 in a cell-free translation system and in cultured cells. During differentiation of cultured skeletal muscle, CUGBP1 is induced in the cytoplasm of cells from normal patients, while DM cells do not accumulate cytoplasmic CUGBP1. The failure to accumulate CUGBP1 in the cytoplasm of DM cells leads to reduced binding of CUGBP1 to p21 mRNA and to the failure of DM cells to increase p21 protein levels during differentiation.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Muscle cells. Human myoblasts were maintained in F10 medium (Gibco-BRL) containing 15% fetal bovine serum, 5% defined supplemented calf serum (HyClone), and 1% penicillin/streptomycin. The myoblast growth medium was changed after 2 days. To initiate differentiation, the myoblasts were grown to 80% of confluency and then the myoblast growth medium was replaced with fusion medium consisting of Dulbecco modified Eagle medium from Gibco-BRL, supplemented with 2% horse serum, 0.01 M insulin, and 2.5 µmol of dexamethazone. The fusion medium was changed every day.

Immunofluorescence. MyoD expression was determined by immunofluorescence microscopy (Zeiss fluorescence microscope) using a 1:10 dilution of MyoD antibodies (Santa Cruz Biotechnology) and a 1:20 dilution of fluorescein isothiocyanate-conjugated immunoglobulin G1-specific anti-human secondary antibodies (Molecular Probe).

Protein purification and Western analysis. Proteins from cultured skeletal muscle cells were isolated as described previously (19, 34, 36). A 50- or 100-µg portion of protein was loaded onto a 10 to 12% polyacrylamide gel and transferred to a nitrocellulose filter (Bio-Rad). The filter was blocked with 10% dry milk-2% bovine serum albumin (BSA) prepared in TTBS buffer (25mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.1% Tween 20) for 1 h at room temperature. Primary antibodies to CUGBP1, MyoD, p21, or p57 were added, and the filter was incubated for 1 h, washed, and then incubated with secondary antibody for 1 h. Immunoreactive proteins were detected using the enhanced chemiluminescence method (Amersham). After detection of the protein of interest, the membrane was stripped and reprobed with anti--actin. For quantitative analysis, the intensity of the signals was determined on the Alpha Imager 2000 gel documentation and analysis system. Antibodies to MyoD, p21, p57, and HuR were purchased from Santa Cruz Biotechnology. Monoclonal antibodies to -actin are from Sigma. Antibodies to CUGBP1 are described in our previous papers (19, 34). Expression of proteins was calculated as a ratio to the -actin loading control. A summary of three to five independent experiments with four differentiation protocols is presented in the paper.

RNase protection assay. p21 mRNA levels were examined using the RiboQuant Multi-Probe RNase protection assay (RPA) system (PharMingen). Total RNA (5 µg) from differentiated cells was hybridized with multiprobe hCC-2 labeled with ³²P as recommended in the instruction manual. The samples were treated with RNase and separated on a 6% denaturing acrylamide gel. The intensity of the radioactive signals was determined using PhosphorImager system.

UV cross-linking assay. Wild-type and deletion constructs of p21 (see Fig. 4A) were described previously (23). p21 mRNAs were labeled by in vitro transcription system with [³²P]UTP, purified by gel electrophoresis, and used in a UV cross-linking assay. An equal amount (200,000 cpm) of radioactive p21 RNAs was incubated with the recombinant CUGBP1. After treatment with RNase A, the RNA-protein complexes were separated by gel electrophoresis. Cold competitors were added to the reaction mixture before addition of the probe. When the RNA CUG₈ oligomer was used as the template, it was labeled with [³²P]ATP and T4 kinase. Cytoplasmic proteins were incubated with the probe, treated by UV light, and analyzed by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis. Where indicated, unlabeled RNA competitors (100 ng) were added prior to protein addition. To verify the concentration of proteins used in the UV cross-linking assay, the membrane was stained with Coomassie blue after the UV cross-linking analysis. All gels described in this paper were equally loaded.

Analysis of the effect of CUGBP1 on the translation of p21 in an RL. To examine the role of CUGBP1 in the regulation of p21, a cell-free coupled transcription-translation system in a rabbit reticulocyte lysate (RL) was used. The conditions for this assay were described in our previous papers (37, 41). Because the activity of CUGBP1 is regulated by phosphorylation, we used CUGBP1 immunopurified from HeLa cells or from the cytoplasm of cells from DM patients, where CUGBP1 has increased activity (36). p21 was translated in an RL containing [³⁵S]methionine from a wild-type construct or from mutant constructs containing several deletions in the 5' region of p21 mRNA. For diagrams of these constructs, see Fig. 4 and 5. Immunoprecipitated CUGBP1 was added to the mixture before the addition of constructs. Mock agarose was used as a negative control. After translation, p21 was immunoprecipitated with polyclonal antibodies (H164; Santa Cruz Biotechnology) and analyzed by electrophoresis. To examine the stability of p21 in RL, ³⁵S-labeled p21 was incubated in the translation mixture lacking amino acids and T7 RNA polymerase. After incubation, the mixture was separated by gel electrophoresis. Under these conditions, p21 was not degraded.

RNA electrophoretic mobility shift assay. RNA oligomers containing GCN repeats from the 5' region of p21 mRNA and GCN mutant RNA oligomers were used as probes. The sequence of these probes is as follows: wt p21, 5'- GCGGCAGCAAGGCCUGCCGCCGCCG-3' (GC islands are underlined); p21-mut, 5'-GAGAUAAUAAGUACUUACAUCUACC-3'. Analysis of binding of CUGBP1 to wild-type and GCN mutant oligomers was performed as described in our earlier papers (35, 36). Briefly, CUGBP1 was purified from HeLa cells by high-pressure liquid chromatography HPLC ion-exchange and size exclusion chromatography. The purified CUGBP1 was incubated with the wt p21 probe and separated by native gel electrophoresis. Cold competitors were incorporated in the binding reaction mixture before probe addition.

Generation of stable clones. CUGBP1 stable clones were generated using an inducible LacSwitch mammalian system as described in our earlier papers (36, 38). The coding region of CUGBP1 was cloned into the pOP-13 vector under the Rous sarcoma virus promoter that is regulated by Lac-Repressor. Human fibroblasts were stably transfected with Lac-Repressor and pOP-13-CUGBP1 antisense plasmids. Clones resistant to hygromycin and to G418 were selected and analyzed for the CUGBP1 expression after addition of isopropyl--D-thiogalactopyranoside (IPTG). Several clones showed a two- to fivefold reduction of CUGBP1 protein by expression of antisense CUGBP1 mRNA. One clone showed a six- to eightfold reduction of CUGBP1 expression after addition of IPTG. This clone was selected for further studies. Cells were grown with 10 mM IPTG or glucose as a control. Cytoplasmic extracts were prepared 24 h after IPTG or glucose addition and analyzed by Western blotting with antibodies against CUGBP1, p21, and -actin as a control.

Analysis of cdk4 activity in the in vitro kinase assay. Kinase assays were carried out with cdk4 immunoprecipitates (IPs) from differentiated normal and DM skeletal muscle cells. Conditions for the kinase assay are follows. IPs were incubated with 1 µg of glutathione S-transferase (GST)-Rb (769-921; Santa Cruz Biotechnology) in the presence of 1 µCi of [-³²P]ATP for 30 min at 37°C. Kinase buffer contains 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 10 mM MgCl₂, 5 mM dithiothreitol, and 5% glycerol. After the kinase assay, the reaction mixture was loaded onto a 12% polyacrylamide gel, transferred to a nitrocellulose membrane (Bio-Rad), and exposed to X-ray film.

E2F gel shift assay. Nuclear proteins were isolated using high-salt extraction as described previously (38). The conditions for the E2F DNA gel shift assay were described in one of our previous papers (38). Briefly, a DNA oligomer containing an E2F consensus site from the DHFR promoter was labeled with Klenow and [-³²P]-dCTP. Then 5 to 10 µg of nuclear extracts from cultured cells was incubated with the probe in binding buffer containing 25 mM Tris-HCl, 100 mM KCl, 5 mM dithiothreitol, 10% glycerol, 2 mM MgCl₂, 0.2 µg of salmon sperm nonspecific competitor, and 50,000 cpm of the probe. To determine the identity of proteins, antibodies to the E2F family and Rb family of proteins were added before probe addition. Antibodies to E2E1-5, p107 (S9), p130 (C20), and Rb (f8) were from Santa Cruz Biotechnology. DNA-protein complexes were separated by nondenaturing polyacrylamide gel electrophoresis (5 to 6% polyacrylamide) in 0.5× Tris-borate-EDTA (TBE) buffer at 4°C. After electrophoresis, the gel was dried and exposed to X-ray film.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Differentiation of myoblasts is affected in DM patients. A number of recent publications showed that ELAV-like proteins are involved in the regulation of differentiation via control of mRNA translation or stability (2, 13, 18). Because CUGBP1 is highly expressed in skeletal muscle (6) and is able to regulate mRNA translation (37), we suggested that it might have similar activities in the regulation of skeletal muscle differentiation. To examine this hypothesis, we generated primary skeletal muscle cell culture lines derived from DM patients containing 1,200 (DM1) and 500 (DM2) CTG/CUG repeats within the DMPK gene/mRNA and from age-matched normal controls. Myoblast differentiation was induced by a shift to a medium containing a low concentration of mitogens. Normal and DM cultures grow at the same density and went through the same number of passages (5 to 10 passages).

View larger version (44K): [in this window] [in a new window]   FIG. 1.   DM cells have the lost ability to withdraw from the cell cycle. (A) Immunostaining of control and DM undifferentiated cells (day 0) and cells maintained in differentiation medium for 5 days (day 5) with antibodies to MyoD. (B) Protein levels of MyoD in control and DM cells, determined by Western blotting. Whole-cell extracts were isolated from the cells on different days after initiation of differentiation (shown on the top) and probed with antibodies against MyoD. The membrane was reprobed with antibodies to -actin. Induction of MyoD as a ratio to -actin is shown below the figure. (C) A significant portion of differentiating myoblasts from DM patients reenter the cell cycle. Differentiated myotubes (d5) from normal and DM patients were placed in growth medium. The DNA content was calculated by FACS analysis on days 2 (5-2) and 5 (5-5) after passage and growth in complete medium. The table above the diagrams shows the percentage of cells in the G1, S and G2/M phases of cell cycle.

Differentiated myoblasts from DM patients do not accumulate CUGBP1 in cytoplasm. Given the observation that RNA CUG repeats affect the RNA binding activity of CUGBP1 (25, 36), we examined whether the activity of CUGBP1 was affected in differentiated cells from normal and DM patients with expanded CUG repeats. It has been shown that CUGBP1 protein is located in both nuclei and cytoplasm (25, 29, 34); therefore, we analyzed the binding activity of CUGBP1 in these intracellular compartments during differentiation of DM cells. UV cross-linking assays show that the majority of CUGBP1 activity was observed in nuclear extracts from DM and from normal myoblasts (day 0). During differentiation, binding activity of CUGBP1 was induced in nuclear extracts of both DM and normal cells (Fig. 2A). The binding activity of nuclear CUGBP1 in differentiated cells from DM patients was two- to threefold higher than activity of CUGBP1 in normal cells, confirming previous observations that CUGBP1 activity is increased in the nuclei of DM cells (25, 29). The levels of CUGBP1 induction in DM cell nuclei during differentiation correlated with the number of CUG repeats in DM patients (1,200 repeats for DM1 and 500 repeats for DM2 [Fig. 2]). Normal muscle cells also accumulate the RNA binding activity of CUGBP1 in cytoplasm on day 3 of differentiation. In contrast, a very weak or no binding of CUGBP1 to the RNA CUG₈ probe was detected in the cytoplasm of DM cells during any stages of skeletal muscle differentiation. To verify that the induction of CUG-specific binding activity is a function of CUGBP1, proteins bound to the CUG₈ probe were precipitated with antibodies specific to CUGBP1 and separated by gel electrophoresis. Figure 2A (UV:IP) shows that the binding activity of the 51-kDa CUGBP1-immunoreactive protein was induced in the cytoplasm from normal cells but not in the cytoplasm from DM cells. The failure to induce a binding activity of CUGBP1 in the cytoplasm was observed in both DM cell lines.

View larger version (49K): [in this window] [in a new window]   FIG. 2.   DM myoblasts do not accumulate CUGBP1 in the cytoplasm during differentiation. (A) UV cross-linking assay. Cytoplasm (cyto) and nuclear extracts (NE) were prepared from predifferentiated (0) and differentiated myoblasts 1 and 3 days after initiation of differentiation. CUGBP1 binding activity was determined by UV cross-linking assays with the RNA CUG8 probe. Cytoplasmic proteins were incubated with the CUG8 probe, UV cross-linked, and precipitated with antibodies against CUGBP1. (B) Protein levels of CUGBP1 are induced in the cytoplasmic compartment of differentiated cells. Nuclear (NE) and cytoplasmic (cyto) proteins were examined by Western blot assay with antibodies to CUGBP1 and HuR proteins. CRM, cross-reactive molecule which interacts with polyclonal antibodies to CUGBP1 and serves as a loading control for the CUGBP1 Western blot. The -actin control for the HuR membrane is shown below.

Protein levels of p21 are induced during differentiation of normal but not DM cells. Normal differentiation of myoblasts requires withdrawal of the cells from the cell cycle division (28, 42). Using a double-knockout animal model, Zhang et al. showed that two cdk inhibitors, p21 and p57, control the differentiation of skeletal muscle (42). Therefore, we initially examined the expression of these proteins in differentiated skeletal muscle cells from normal (control) and DM patients. In agreement with published observations (42), p21 levels were increased three- to fourfold during differentiation in normal cells (Fig. 3A and B). However, DM skeletal cells did not induce p21 protein. Expression of another cdk inhibitor, p57, was also altered in DM cells. In normal predifferentiated cells, levels of p57 were low or not detectable, but they were significantly increased during differentiation. This pattern of p57 expression is consistent with published observations (28). A quite different p57 expression pattern was observed in DM cells. Skeletal muscle cells from DM patients already contained high levels of p57 before the initiation of differentiation and retained these levels during differentiation. Taken together, these studies showed that the expression of p21 and p57, two important regulators of the cell cycle in skeletal muscle cells, is altered in DM cells. Because the major difference in differentiation of DM cells is observed at later stages of differentiation (Fig. 1) and because p57 levels in DM and control cells do not differ at that time, we focused our studies on the investigations of p21 protein in normal and DM cells.

View larger version (41K): [in this window] [in a new window]   FIG. 3.   DM muscle cells fail to induce protein levels of p21 during differentiation. (A) Nuclear extracts were analyzed by Western blotting with antibodies against p57 and p21. The same membrane was reprobed with antibodies against -actin. Bar graphs show a summary of three independent experiments with p21 protein levels calculated as a ratio to -actin. (B) Expression of p21 mRNA is similar in DM and normal cells. Total RNA was isolated from control and DM cells at different stages of differentiation. Expression of p21 mRNA was examined using RPA. Bar graphs show a summary of two independent experiments. Levels of p21 mRNA were calculated as a ratio to GAPDH. (C) Binding activity of CUGBP1 immunoreactive protein is induced during differentiation of control but not DM cells. Cytoplasmic proteins were incubated with FL p21 probe (lacking the 3'UTR), exposed with UV light, immunoprecipitated with polyclonal antibodies to CUGBP1, and separated by gel electrophoresis (UV-IP). (D) CUGBP1 immunoprecipitated from the cytoplasm of normal cells stimulates p21 translation in a cell-free translation system. CUGBP1 was immunoprecipitated from DM and normal cells at different stages of differentiation and added to a cell-free translation system programmed with p21 mRNA. 35S-labeled p21 was immunoprecipitated and loaded onto a 12% polyacrylamide gel.

CUGBP1 interacts with p21 mRNA during differentiation of normal cells and is able to induce the translation of p21 mRNA in a cell-free translation system. Given the failure of DM cells to accumulate CUGBP1 in the cytoplasm and the accompanying lack of p21 induction, we suggested that CUGBP1 might regulate the translation of p21 mRNA. To test this suggestion, we first examined whether CUGBP1 interacts with p21 mRNA in cytoplasmic extracts from differentiated cells. Because a member of the ELAV family of proteins, HuR, is able to regulate p21 mRNA via binding to an AU-rich region in the 3'UTR of p21 mRNA (13, 40), we used a p21 construct that lacks the 3'UTR (FL* p21 mRNA) and is unlikely to be regulated by HuR. Incubation of FL* radioactive p21 mRNA with cytoplasmic proteins followed by UV treatment and immunoprecipitation with antibodies to CUGBP1 showed that CUGBP1 immunoreactive protein interacted with p21 mRNA (Fig. 3C). The interaction was increased during differentiation in normal cells but not in cells derived from DM patients. Because p21 mRNA levels were identically induced in both normal and DM cells (Fig. 3B) and because CUGBP1 can regulate the translation of mRNAs (37), we examined the ability of CUGBP1 to induce the translation of p21. For this goal, CUGBP1 was immunoprecipitated from differentiated cells on days 0, 3, and 5 after initiation of differentiation and added to a cell-free translation system programmed with FL* p21 mRNA. As can be seen in Fig. 3D, CUGBP1 immunoprecipitated from normal cells induced the translation of p21 mRNA, while immunoprecipitates from DM cells had little or no effect on p21 translation. Thus, these studies suggested that CUGBP1 induces the translation of p21, leading to increased levels of p21 in differentiated myoblasts from normal patients.

CUGBP1 interacts with GCN repeats within the 5' region of p21 mRNA. Given the interaction of CUGBP1 with p21 mRNA in vivo (Fig. 3C), we next carried out a detailed analysis of the interaction of CUGBP1 with the p21 mRNA in vitro. For this goal, we used p21 constructs (Fig. 4A) and bacterially expressed, purified maltose binding protein-CUGBP1 (MBP-CUGBP1). The characterization of p21 deletion constructs was described in a previous publication (23). Purification and analysis of MBP-CUGBP1 binding activity were described in our earlier papers (19, 34, 37). UV cross-linking and competition analysis with the bacterially expressed, purified MBP-CUGBP1 and FL* p21 mRNA is shown in Fig. 4B. As can be seen in Fig. 4B, CUGBP1 bound to the FL* p21 mRNA. This binding is specific because cold RNA oligomers containing high-affinity CUGBP1 binding sites (CUG, CCG, LAP, and sORF [37]) competed for the binding while a UA-rich nonspecific oligomer (GX) did not affect the binding. To further examine the specificity of the interaction of CUGBP1 with p21 mRNA and to map the region of interaction within p21 mRNA, we studied the binding of CUGBP1 to a set of p21 mRNAs that have deletions of different regions of the coding sequence of p21 mRNA (Fig. 4A). CUGBP1 bound to a nucleotide sequence located within the first 120 nucleotides of p21 mRNA, because deletion of these nucleotides (d41) abolished the interaction (Fig. 4C). Further analysis of internal deletions within the 5' region of p21 mRNA indicated that mRNA transcribed from the deletion construct d17-22 (later called dGCN-p21) showed a very weak interaction with the recombinant MBP-CUGBP1 while mRNA from an another deletion construct, d24-29, showed a strong and specific interaction with the MBP-CUGBP1 (Fig. 4C). These results demonstrate that nucleotide sequence deleted within d17-22-p21 mRNA is necessary for the interaction with CUGBP1.

View larger version (36K): [in this window] [in a new window]   FIG. 4.   CUGBP1 binds to the 5' region of p21 mRNA. (A) Diagram showing p21 constructs used in these studies. The constructs are labeled according to a deletion of amino acids (23). A sequence deleted within construct d17-22 (GCN repeats, wt and mut) is shown below. (B) Binding of the recombinant MBP-CUGBP1 to p21 mRNA. The labeled FL *p21 mRNA was incubated with MBP-CUGBP1, exposed to UV light, and analyzed by gel electrophoresis. Cold RNA competitors (shown on the top) were added to the binding-reaction mixture before the probe addition. sORF and LAP oligomers are RNA fragment from C/EBP mRNA (37); CCG oligo contains eight CCG repeats; and GX is RNA oligonucleotide of a random AU-rich sequence. (C) UV cross-linking assay with mRNAs transcribed from mutant p21 constructs shown in panel A. CUG, addition of the cold CUG8 competitor. (D) CUGBP1 binds to GCN repeats with the 5' region of p21 mRNA. Bacterially expressed, purified MBP-CUGBP1 was incubated with FL* p21 probe. Cold wt p21 and GCN mut oligomers (shown in panel A) were added to the reaction mixture before probe addition. The GCN mutant oligomer does not compete for the binding. (E) Gel shift assay. CUGBP1 was HPLC purified from HeLa cells and incubated with wt p21 probe (shown in panel A) in the presence of cold RNA competitors (shown at the top) and separated by native gel electrophoresis.

CUGBP1 increases the translation of p21 in a cell-free translation system via the interaction with GCN repeats. We previously showed that CUGBP1 is associated with polysomes and is able to regulate the translation of mRNAs (37). In addition, we showed that CUGBP1 from differentiated cells is able to induce p21 translation in a cell-free translation system (Fig. 3). Therefore, we performed a detailed analysis of the effect of CUGBP1 on translation of p21 by using an RL cell-free translation system. Since phosphorylation of CUGBP1 increases the ability of CUGBP1 to regulate translation (41), we immunopurified CUGBP1 from HeLa cells or from normal myoblasts on day 3 of differentiation (Fig. 3D). In agreement with data in Fig. 3, the addition of CUGBP1 immunoprecipitated from HeLa cells to the translation mixture programmed with FL* p21 mRNA led to increased production of p21 protein (Fig. 5B, translation). This effect was not due to stabilization of p21 protein, because under the conditions of our experiments, p21 protein was not degraded in the RL and therefore was not stabilized by CUGBP1 in this system (Fig. 5B, incubation). To examine whether GCN repeats (CUGBP1 binding site) within the 5' region of p21 mRNA are responsible for CUGBP1-dependent induction of p21 in the cell-free translation system, p21 was translated from the p21 deletion constructs, dGCN-p21 (previously referred to as d17-22) and d24-29-p21 (control), in the absence and presence of CUGBP1. The nucleotide sequence of FL* p21 and the dGCN-p21 construct is shown in Fig. 5A. Analysis of these p21 mutants in a cell-free translation system showed that deletion of the CUGBP1 binding site abolished the CUGBP1-dependent induction of p21 in the cell-free translation system (Fig. 5C). These data demonstrate that CUGBP1 induces the translation of p21 in the cell-free translation system via an interaction with GCN repeats located in the 5' region of p21 mRNA.

View larger version (30K): [in this window] [in a new window]   FIG. 5.   CUGBP1 induces the translation of p21 in an in vitro cell-free translation system and in cultured cells. (A) Nucleotide sequence of the dGCN deletion construct. Four GCN repeats are deleted within the dGCN-p21 mRNA. (B) Immunopurified CUGBP1 induces the translation of p21. The plasmid coding for FL *p21 mRNA was transcribed and translated in the presence of [35S] Met (translation). Addition of increasing amounts of CUGBP1 to the translation reaction mixture results in increased translation of p21. Incubation, CUGBP1 was incubated with 35S-p21 under conditions of translation involving a lack of amino acids and p21 mRNA. p21 is not degraded, and CUGBP1 does not affect the stability of p21 protein (C) The deletion of GCN repeats abolishes CUGBP1-dependent induction of p21 translation. p21 was translated from mRNAs transcribed from dGCN-p21 and d24-29-p21 constructs in the presence of CUGBP1. Arrows show the positions of corresponding p21 products. Agarose precipitates serve as a negative control. (D) CUGBP1 regulates p21 protein in cultured cells. HT1080 cells were transfected with an empty vector (V) or with vector expressing CUGBP1, and the levels of CUGBP1 and p21 were examined by Western blotting. (E) Levels of p21 are reduced in a stable clone expressing antisense CUGBP1 mRNA. Expression of antisense CUGBP1 mRNA was induced by IPTG (I), and proteins were isolated from Glucose- (G) and IPTG-treated cells and analyzed by Western blotting with antibodies specific to CUGBP1 and p21. The p21 membrane was reprobed with antibodies to -actin. Expression of mRNA was examined by RPA. Levels of p21 and GAPDH mRNAs are shown.

CUGBP1 induces the translation of p21 in cultured cells. To examine whether CUGBP1 regulates p21 protein in cultured cells, two approaches were used; transient-transfection studies and use of a stable clone expressing antisense CUGBP1 mRNA. CUGBP1-expressing vector (pOP-CUGBP1) (36) was transfected into HT1080 cells, and total-protein lysates were analyzed by Western blotting with antibodies to p21 and CUGBP1. In these experiments, the efficiency of transfection was 40 to 50%. Figure 5D shows that increased expression of CUGBP1 led to elevation of p21 protein levels. The induction of p21 was proportional to CUGBP1 expression. These data show that CUGBP1 induces p21 levels in cultured cells. To confirm this observation, stable clones expressing antisense CUGBP1 mRNA were generated using a LacSwitch inducible system as described in our earlier publications (36, 38). The antisense strategy was chosen because levels of endogenous CUGBP1 are relatively high in cultured cells (34, 35). Expression of antisense CUGBP1 mRNA (induced by IPTG) leads to a significant reduction of endogenous CUGBP1 protein levels. In agreement with the results obtained in a cell-free translation system and in transient-transfection studies, the levels of p21 were reduced in cells expressing low levels of CUGBP1 (Fig. 5E). Reprobing the membrane with antibodies to -actin showed equal protein loading. Examination of p21 mRNA levels by RPA showed no significant difference in mRNA expression (Fig. 5E, mRNA). Taken together, these studies demonstrate that CUGBP1 regulates the translation of p21 protein in vitro and in cultured cells.

The activity of cdk4 is altered in DM muscle cells. Given the failure of DM cells to increase p21 during differentiation, we examined the expression and activity of proteins that are regulated by p21 and are important regulators of skeletal muscle differentiation. An in vitro kinase assay utilizing cdk4 precipitated from differentiated cells showed that cdk4 activity was reduced during differentiation of normal skeletal muscle, while in DM cells the cdk4 activity was significantly induced during differentiation (Fig. 6A). Western analysis indicated that protein levels of cdk4 were similar in DM and normal cells and that cyclin D1 levels were slightly induced at later stages of differentiation in DM cells. Because the induction of cdk4 activity in DM cells correlated with the failure of these cells to increase p21 protein (Fig 3), we examined whether the difference in cdk4 activity is due to association with p21. To do this, cdk4 was immunoprecipitated from differentiated cells and p21 was examined in cdk4 IPs by Western blotting with monoclonal antibodies. Figure 6A shows that the amounts of p21 in cdk4 IPs were significantly induced in normal cells at later stages of differentiation while p21 was not detectable in cdk4 IPs in DM cells. These data are consistent with the hypothesis that the failure of DM cells to induce p21 leads to failure to reduce the activity of cdk4 via p21-dependent pathway.

View larger version (41K): [in this window] [in a new window]   FIG. 6.   Expression and activity of cdk4 and E2F are altered in DM myoblasts. (A) cdk4 activity in control and DM skeletal muscle during differentiation. cdk4 was immunoprecipitated from nuclear extracts and analyzed in an in vitro kinase assay with a GST-Rb substrate (kinase assay). Levels of cdk4 and cyclin D1 were examined by Western blotting. -Actin loading control is shown for the cdk4 membrane. For the IP-cdk4-Western, p21 was examined in cdk4 immunoprecipitates from differentiated cells. (B) Composition of E2F complexes is altered in DM myoblasts. Nuclear extracts from control and DM cells on different days of differentiation (indicated at the top) were incubated with an E2F probe (E2F consensus from the DHFR promoter) and separated by gel electrophoresis. E2F complexes in predifferentiated cells are numbered. E2F complexes 3 and 4 are present only in predifferentiated DM cells.

DM cells do not form Rb/E2F growth repressor complexes. Given the observations that DM cells are able to reenter cell cycle (Fig. 1C), we examined additional proteins that are crucial for cell cycle withdrawal. It has been shown that formation of Rb-E2F complexes is necessary for the inhibition of cell cycle progression and for cell cycle withdrawal (7, 9, 14, 24). Therefore, we examined E2F-Rb complexes in control and DM cells. Gel shift analysis with an E2F probe showed that the pattern of E2F binding during the differentiation of normal myoblasts differed from the E2F binding pattern observed in DM cells (Fig. 6B). The difference in E2F complexes was detectable in predifferentiated cells and continued to differ at later stages of differentiation. In predifferentiated normal cells, two E2F complexes (complexes 1 and 2) were detected (Fig. 6B, control, 0). A very different pattern of E2F binding was observed in predifferentiated DM cells. E2F complex 1 was not detectable in DM cells, and two additional low-mobility E2F complexes (complexes 3 and 4, Fig. 6B, DM, 0) were abundant. During differentiation, E2F complexes 1 and 4 were induced in normal cells while one of these complexes (complex 4) was not detectable after the onset of differentiation in DM cells (Fig. 6B).

View larger version (42K): [in this window] [in a new window]   FIG. 7.   Identity of proteins involved in the formation of E2F complexes in DM and normal cells. (A) Gel shift/supershift analysis with nuclear extracts from predifferentiated cells. Antibodies to E2F4, p107, p130, and Rb (indicated at the top) were incorporated into the E2F reaction mixture with nuclear extracts from predifferentiated (day 0) control and DM cells. (B) Gel shift/supershift assay with nuclear extracts isolated on day 3 after the initiation of differentiation. Antibodies to E2F1 to E2F-4 and Rb family proteins were incorporated into the E2F binding reaction mixture.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Studies of transgenic mice expressing multiple CUG repeats indicate that RNA CUG repeats play a crucial role in the DM pathogenesis (20). A number of observations suggested that the expansion of CUG repeats within the 3' region of the mutant DMPK mRNA might affect the differentiation of skeletal muscle in DM. Data from Korneluk's laboratory showed that the overexpression of CUG repeats within the mutant DMPK mRNA inhibits the differentiation of skeletal muscle (30). In agreement with these observations, two other studies demonstrated that the mutant DMPK mRNA with 200 CUG repeats selectively inhibits myoblast differentiation in a cell culture model (1, 4). Although the RNA-based hypothesis for DM pathogenesis has been confirmed by the above-mentioned studies, the molecular mechanisms of how RNA CUG repeats affect biochemical pathways are not well understood. A recent study shows that overexpression of RNA CUG repeats in DM patients and in cultured cells leads to the sequestration of proteins that specifically interact with CUG repeats (36). Similar to the effect on CUGBP1, overexpression of RNA CUG repeats might affect other RNA binding proteins that are involved in the regulation of cell differentiation.

A growing number of recent publications describe an important role of RNA binding proteins in cell growth and differentiation (2). Interaction of RNA binding proteins with mRNAs coding for cell cycle proteins leads to alterations in the stability or translation of these mRNA. CUGBP1 is similar to ELAV-like proteins but regulates the translation of mRNAs via interaction with GCN repeats within mRNAs (37, 41). Given the observation that the activity of CUGBP1 protein is altered in DM patients (6, 25, 29, 34, 36), we examined whether CUGBP1 is involved in the control of differentiation in normal skeletal muscle cells and whether CUGBP1 function is affected during differentiation of skeletal muscle cells from DM patients. We found that the intracellular localization of CUGBP1 protein is altered in DM cells during differentiation and that this alteration leads to changes in the expression of p21 protein, which is known to be involved in the growth arrest of skeletal muscle cells (42). Our previous studies showed that the regulation of CUGBP1 activity occurs primarily at the posttranslational level (36, 37, 41). In this paper, we present evidence that the activity of CUGBP1 can also be controlled by regulation of its intracellular localization. It is interesting that a similar pathway of regulation was described for the ELAV-like HuR protein. Under basal conditions, HuR is located in nuclei, while in response to UV treatment, HuR translocates to the cytoplasm and stabilizes p21 mRNA (40). Similar to HuR, the binding activity of CUGBP1 is located in nuclei of skeletal muscle cells from normal patients and is detectable in cytoplasm only at later stages of differentiation (Fig. 2). However, in DM cells, CUGBP1 is observed only in nuclei during all stages of differentiation.

CUGBP1 is a multifunctional protein and is involved in the regulation of splicing and translation of mRNAs (25, 37, 41). In this study, we identified a new target of CUGBP1: p21 mRNA. Our data show that CUGBP1 binds to the 5' region of p21 mRNA and enhances the production of p21 in a cell-free translation system and in cultured cells. The ability of CUGBP1 to control p21 translation suggests that DM cells do not increase the p21 level during differentiation due to lack of CUGBP1 in cytoplasm. This suggestion is consistent with increased binding of CUGBP1 to the p21 mRNA observed during differentiation of normal cells (Fig. 3). In addition to alterations in p21 expression, DM cells contain a high level of p57, which does not change during differentiation. The difference in p57 levels may also contribute to alterations in cdk4 and Rb activities. In this study, we investigated in detail the regulation of p21 by CUGBP1 and the expression and activities of downstream targets of p21 in DM cells. Although there are many pathways for the regulation of p21 in cells, investigations of p21 expression in transient-transfections studies showed that CUGBP1 induced protein levels of p21 in cultured cells. In agreement with these observations, results with a stable clone expressing antisense CUGBP1 mRNA show that reduction of CUGBP1 levels leads to a decrease of the p21 level. In vitro studies suggest that CUGBP1 up-regulates p21 translation via interaction with GCN repeats located in the 5' region of p21 mRNA. Further studies are necessary to understand the mechanism of this control. It is interesting that a recent paper from the Ashizawa laboratory demonstrated that p21 expression is reduced in cultured cells from DM patients with expanded CTG/CUG repeats and that DM cells possess increased proliferative capacities (15). These observations are consistent with data presented in our paper and suggest that RNA CUG repeats play a key role in the alterations of cell cycle withdrawal and proliferation observed in DM patients.

Analysis of the ability of cells to reenter the cell cycle showed that DM differentiating myoblasts are able to proliferate when they are cultured in growth medium while myotubes from normal patients do not proliferate. The analysis of cell cycle proteins during differentiation of normal and DM cells indicates that a number of other activities that are necessary for cell cycle withdrawal do not occur in DM cells. Our studies show that the activity of cdk4 is not reduced in DM cells during differentiation. E2F-Rb complexes, which are important for cell cycle withdrawal, are induced in myotubes from normal controls but not in myotubes from DM patients. In addition to alterations in E2F-Rb complexes, we detected changes in the E2F complexes with other Rb family proteins in DM cells. In particular, E2F-p107 complexes were detected in predifferentiated DM cells but not in normal cells. It has been shown that the formation of E2F-p107 complexes takes place primarily during S phase (9). The increased amounts of the S-phase-specific E2F-p107 complexes in DM cells provide an additional demonstration that the proliferative capacities of DM cells differ from those of cells from normal patients. Alterations of E2F-Rb family complexes in predifferentiated cells suggest that in addition to p21, CUGBP1 might regulate other mRNAs required for cell cycle control. There is also the possibility that in addition to CUGBP1, perhaps some other factors are affected in DM cells. Khajavi et al. have recently described that DM cells with large CTG expansion had increased proliferative capacities (15). The authors showed that the increased proliferation of DM cells correlated with down-regulation of the p21 protein. Data in our paper are consistent with these observations and provide more mechanistic insight into the molecular basis for alterations of cell cycle in DM cells.

Taken together, data in this paper show the role of an RNA binding protein, CUGBP1, in the regulation of p21 expression and in skeletal muscle differentiation. Similar to ELAV-like proteins, CUGBP1 is involved in the regulation of genes that are necessary for cell differentiation and growth arrest. Multiple changes of cell cycle proteins in DM cells indicate that cell cycle withdrawal is impaired in patients with expanded CUG repeats and with abnormal expression of CUGBP1.