The Utrophin A 5′-Untranslated Region Confers Internal Ribosome Entry Site-mediated Translational Control during Regeneration of Skeletal Muscle Fibers*

Utrophin up-regulation in muscle fibers of Duchenne muscular dystrophy patients represents a potential therapeutic strategy. It is thus important to delineate the regulatory events presiding over utrophin in muscle in attempts to develop pharmacological interventions aimed at increasing utrophin expression. A number of studies have now shown that under several experimental conditions, the abundance of utrophin is increased without a corresponding elevation in its mRNA. Here, we examine whether utrophin expression is regulated at the translational level in regenerating muscle fibers. Treatment of mouse tibialis anterior muscles with cardiotoxin to induce muscle degeneration/regeneration led to a large (∼14-fold) increase in the levels of utrophin A with a modest change in expression of its transcript (40%). Isolation of the mouse utrophin A 5′-untranslated region (UTR) revealed that it is relatively long with a predicted high degree of secondary structure. In control muscles, the 5′-UTR of utrophin A caused an inhibition upon translation of a reporter protein. Strikingly, this inhibition was removed during regeneration, indicating that expression of utrophin A in regenerating muscles is translationally regulated via its 5′-UTR. Using bicistronic reporter vectors, we observed that this translational effect involves an internal ribosome entry site in the utrophin A 5′-UTR. Thus, internal ribosome entry site-mediated translation of utrophin A can, at least partially, account for the discordant expression of utrophin A protein and transcript in regenerating muscle. These findings provide a novel target for up-regulating levels of utrophin A in Duchenne muscular dystrophy muscle fibers via pharmacological interventions.

Utrophin up-regulation in muscle fibers of Duchenne muscular dystrophy patients represents a potential therapeutic strategy. It is thus important to delineate the regulatory events presiding over utrophin in muscle in attempts to develop pharmacological interventions aimed at increasing utrophin expression. A number of studies have now shown that under several experimental conditions, the abundance of utrophin is increased without a corresponding elevation in its mRNA. Here, we examine whether utrophin expression is regulated at the translational level in regenerating muscle fibers. Treatment of mouse tibialis anterior muscles with cardiotoxin to induce muscle degeneration/regeneration led to a large (ϳ14-fold) increase in the levels of utrophin A with a modest change in expression of its transcript (40%). Isolation of the mouse utrophin A 5-untranslated region (UTR) revealed that it is relatively long with a predicted high degree of secondary structure. In control muscles, the 5-UTR of utrophin A caused an inhibition upon translation of a reporter protein. Strikingly, this inhibition was removed during regeneration, indicating that expression of utrophin A in regenerating muscles is translationally regulated via its 5-UTR. Using bicistronic reporter vectors, we observed that this translational effect involves an internal ribosome entry site in the utrophin A 5-UTR. Thus, internal ribosome entry site-mediated translation of utrophin A can, at least partially, account for the discordant expression of utrophin A protein and transcript in regenerating muscle. These findings provide a novel target for up-regulating levels of utrophin A in Duchenne muscular dystrophy muscle fibers via pharmacological interventions.
Duchenne muscular dystrophy (DMD) 6 is the most prevalent inherited neuromuscular disorder, since it affects 1 of 3500 male births (1). Although the genetic defect underlying DMD was identified 20 years ago (2,3), there is still no effective treatment to alter the relentless progression of this debilitating neuromuscular disease (4). DMD is caused by deletions/mutations in the dystrophin gene, which prevents production of full-length dystrophin molecules in skeletal muscle fibers. One therapeutic strategy to treat DMD consists in up-regulating the endogenous levels of a protein in muscle fibers of affected patients, which, once expressed at the appropriate location, could functionally compensate for the absence of dystrophin. An ideal candidate for such a role is utrophin (5,6). Utrophin is a large cytoskeletal protein that displays a high degree of structural and functional similarity to dystrophin. In this context, overexpression of utrophin in muscle fibers of a DMD mouse model has been shown to alleviate the dystrophic pathology (7)(8)(9), thereby demonstrating that increased expression of utrophin could indeed yield important clinical benefits.
A major difference between utrophin and dystrophin relates to their pattern of expression in skeletal muscle fibers. Whereas dystrophin localizes along the cytoplasmic face of the entire muscle fiber, utrophin preferentially accumulates at the levels of the myotendinous and neuromuscular junctions (10 -12). To be valid as a therapeutic strategy for DMD, utrophin expression must therefore be extended in extrasynaptic compartments of dystrophic muscle fibers. Thus, elucidating the nature of the molecular events controlling expression of utrophin in muscle fibers becomes important, since it could lead to the rational design of pharmacological interventions aimed at up-regulating utrophin in muscle fibers of DMD patients.
Two full-length isoforms of utrophin have now been identified and named utrophin A (13) and utrophin B (14). These isoforms are encoded by mRNAs that are transcribed from two different promoters. These two mRNAs differ mostly in their 5Ј-untranslated region (UTR) as well as in their initial coding portion, resulting in protein products that slightly differ in their N termini. Utrophin A is the major full-length isoform expressed in skeletal muscle fibers, whereas expression of utrophin B is relatively higher in the vasculature (15). To date, most of the studies that have examined the mechanisms regulating utrophin A expression in muscle cells have focused on transcriptional regulation. For example, the synaptic accumulation of utrophin A has been shown to involve the local transcriptional activation of the utrophin A promoter through an N-box motif and the Ets-related transcription factor GA-binding protein, which is activated by heregulin stimulation (16,17). Recent studies have also shown that utrophin A is expressed at low levels in extrasynaptic compartments of slow muscle fibers and, in this case, is subject to regulatory events involving the calcineurin/nuclear factor of activated T cells signaling pathway (18,19).
In addition to transcriptional mechanisms, there is mounting evidence indicating that post-transcriptional mechanisms also regulate utrophin expression in muscle. This is coherent with the fact that post-transcriptional mechanisms are known to control expression of several synaptic proteins in muscle cells (20). In particular, we have previously shown that distinct elements within the utrophin 3Ј-UTR are important for controlling the stability of utrophin transcripts in muscle cells and for targeting them to specific subcellular locations (21,22).
Converging lines of evidence also indicate that utrophin may be regulated at the translational and/or post-translational level. For example, we have previously demonstrated that utrophin levels are greatly upregulated during muscle regeneration without a corresponding increase in expression of its mRNA (23). In the same study, we also noted that although utrophin levels are increased in muscle biopsy samples from DMD patients, levels of utrophin mRNA are similar to those seen in normal muscle samples (23,24). Since these initial findings, several other laboratories have also reported discordant patterns of expression of utrophin protein and transcript. Specifically, Weir et al. (15) detected an increase in the abundance of utrophin A in muscle from mdx mice with no parallel elevation in the levels of utrophin A transcript. Similarly, utrophin protein levels in muscle change significantly during the life span of mdx mice with little alteration in the expression of utrophin mRNAs (25). Finally, results from several other studies also support the idea that utrophin is regulated via translational mechanisms (26,27). With this in mind, we have therefore initiated in the present study a series of experiments to examine whether the regulation of utrophin A in vivo involves translational events.

EXPERIMENTAL PROCEDURES
Generation of Reporter Constructs-To obtain the murine utrophin A 5Ј-UTR, reverse transcription and PCR experiments using total RNA obtained from C2C12 myotubes were carried out. The 5Ј and 3Ј primers amplified a 508-base pair fragment (5Ј primer, 5Ј-GTTGTG-GAGTCGCCCT-3Ј; 3Ј primer, 5Ј-CCCCATACTTGGCCAT-3Ј) (15). This fragment was sequenced to confirm its identity. It was subsequently inserted into the multiple cloning site of pCMV SPORT-␤GAL (Invitrogen) (called p␤GAL here) to generate pUtrA/␤GAL (see Fig. 1 for all plasmids used). The utrophin A 5Ј-UTR was also inserted into the XhoI site of the bicistronic vector p␤GAL/CAT (28) to generate p␤GAL/UtrA/CAT (see Fig. 1). The p␤GAL/X-chromosome-linked inhibitor of apoptosis protein (XIAP)/CAT construct containing the ϳ1-kb XIAP 5Ј-UTR (28) was used in our experiments as a control for IRES activity. The promoterless bicistronic constructs p(-cmv)␤GAL/ CAT and p(-cmv)␤GAL/UtrA/CAT were created by removing the CMV promoter using NruI and HindIII restriction sites. Excision of the ␤-galactosidase gene was accomplished by cutting with NotI and religating to create the monocistronic pUtrA/CAT plasmid.
Cell Culture-C2C12 myoblasts were cultured in growth media in 35-mm plates as previously described in detail elsewhere (29). Cells were transfected when they reached 60 -80% confluence. Two g of p␤GAL and pUtrA/␤GAL constructs were co-transfected with 2 g of pCAT control vector (Promega, Mississauga, Canada) using 7.5 l of Lipofectamine (2 mg/ml) (Invitrogen) according to the manufacturer's instructions. Transfections using the bicistronic vectors were performed using 4 g of plasmid with 8 l of Lipofectamine (2 mg/ml). Cells were lysed 24 h after transfection using 400 l of 1ϫ reporter lysis buffer according to the manufacturer's instructions (Promega), and the reporter activity was determined as described below.
Direct Plasmid Injection and Cardiotoxin Treatment-For in vivo studies, 25 l of cardiotoxin (Latoxan, Rosans, France) at 10 Ϫ5 M were injected into the right tibialis anterior (TA) muscle of 4-week-old C57BL/10 mice (Charles River Laboratories, St. Constant, Canada) to induce muscle degeneration and regeneration as previously described (23). Seven days later, cardiotoxin-treated and control, untreated TA muscles were excised and frozen in liquid nitrogen. For immunofluorescence experiments (see below), some TA muscles were frozen in melting isopentane precooled with liquid nitrogen.
To examine the expression of the reporter constructs during muscle regeneration, 25 l of plasmid at 2 g/l in PBS was directly injected in both regenerating and control TA muscles 3 days after cardiotoxin treatment. Four days later, TA muscles were excised and frozen in liquid nitrogen. TA muscles injected with plasmids were homogenized using a Polytron homogenizer in 1 ml of 1ϫ reporter lysis buffer (Promega), freeze-thawed twice, and centrifuged for 20 min at 12,000 ϫ g. Supernatants were collected and frozen until further analysis.
Reporter Assays-Assays for ␤GAL enzymatic activity were performed on homogenized TA muscles or C2C12 myoblast lysates using the ␤-galactosidase enzyme assay system according to the manufacturer's instructions (Promega). CAT activity was measured by analyzing the conversion of chloramphenicol to butyryl chloramphenicol by incorporation of [ 14 C]butyryl coenzyme A using the CAT enzyme kit (Promega). Counting was performed in a liquid scintillation counter (Wallac 1414 liquid scintillation counter with 1414 Winspectral Windows software). Background levels for both reporter assays were determined by analyzing reporter activity in nontransduced TA muscle and nontransfected C2C12 cells.
RNA Extraction and RT-PCR-Total RNA was isolated from C2C12 cells or TA muscle using TRIzol reagent (Invitrogen) as recommended by the manufacturer. To quantitate the levels of endogenous transcripts in cardiotoxin-treated and untreated TA muscles, RT-PCR was performed to amplify utrophin A and S12 ribosomal RNA as previously described in detail elsewhere (18). Negative controls consisted of RT mixtures in which total RNA was replaced by sterile, diethylpyrocarbonate-treated water. For quantitation of endogenous utrophin A mRNA and S12 ribosomal RNA levels, PCR products were separated on Vista Green (Amersham Biosciences)-containing agarose gels, and the intensity of the signal was quantitated using ImageQuant version 5.1 (Amersham Biosciences). The relative levels of utrophin A transcripts were standardized according to the amount of S12 ribosomal RNA levels present in the same samples.
For RT-PCR analysis of reporter expression in transfected C2C12 cells and transduced TA muscles, Trizol-extracted RNA was first DNase I-treated (Invitrogen) for 1 h to eliminate plasmid contamination according to the manufacturer's instructions. For TA muscles injected with p␤GAL and pUtrA/␤GAL and co-transduced with pCAT, RT-PCR was carried out with primers to amplify a portion of CAT (5Ј-TG-GCAATGAAAGACGGTGAG-3Ј; 5Ј-GAAAACGGGGGCGAAGA-AGT-3Ј), and a portion of ␤GAL (5Ј-GTGACGGCAGTTATCTGG-3Ј; 5Ј-TTGGCAGTGCTCGTAGTA-3Ј) mRNAs. Negative controls consisted of an RT mixture that had the reverse transcriptase replaced with diethylpyrocarbonate-treated water. Denaturation was performed at 94°C for 45 s, followed by an annealing step of 55°C of 1 min and an extension step at 72°C for 1 min for both ␤GAL and CAT PCRs. For ␤GAL amplification, 20 -27 cycles were used for quantitative analysis, whereas 23-28 cycles were used for CAT. Cycling was followed by a final extension step at 72°C for 10 min. It is important to note that for all RT-PCR experiments used to quantify the relative abundance of transcripts, cycle numbers were optimized to be within the linear range of amplification as previously described in detail (22,30,31).
To control for the presence of an intact bicistronic reporter transcript following C2C12 transfections and direct plasmid injection into TA muscles with p␤GAL/CAT and p␤GAL/UtrA/CAT plasmids, RT-PCR using total RNA was performed using a 5Ј primer spanning 142 nucleotides of the LacZ gene and a 3Ј primer spanning 459 nucleotides of the CAT gene that flanked the utrophin A 5Ј-UTR (5Ј-TTTTTCCCGATT-TGGCTACA-3Ј; 5Ј-TGAAACTCACCCAGGGATTG-3Ј). This RT-PCR strategy is summarized in Fig. 6. The RT-PCR products were subcloned using the TOPO TA cloning kit (Invitrogen) and sequenced to verify their identity.
For quantitative RT-PCR, reverse transcription was carried out using the First-Strand cDNA Synthesis kit (Amersham Biosciences) with NotI-d(T) 18 primers. The quantitative real time RT-PCR was performed using the QuantiTect SYBR green RT-PCR kit (Qiagen) and analyzed on an ABI Prism 7000 sequence detection system using the ABI Prism 7000 SDS Software. Quantitative PCRs were carried out to detect ␤GAL (5Ј-ACTATCCCGACCGCCTTACT-3Ј; 5Ј-CTGTAGC-GGCTGATGTTGAA-3Ј) and CAT (5Ј-GCGTGTTACGGTGAAAA-CCT-3Ј; 5Ј-GGGCGAAGAAGTTGTCCATA-3Ј) for muscle samples transduced with the bicistronic vectors. In order to control for DNA contamination, quantitative PCRs were performed directly on RNA samples.
Northern Blotting-For Northern blot analysis, electrophoresis of total RNA was performed using a denaturing agarose/formaldehyde gel followed by transfer onto a Hybond, positively charged nylon membrane (Amersham Biosciences) for 2.5 h using downward alkaline transfer. The membranes were preincubated for 4 h at 42°C in a shaking water bath with 10 ml of DIG easy hyb buffer (Roche Applied Science). The membrane was then incubated overnight with a linearized cDNA probe for CAT (301 bp) or ␤GAL (333 bp) labeled with Redivue [␣-32 P]dCTP (Amersham Biosciences) using the Rediprime II random prime labeling system (Amersham Biosciences). Washes were carried out for 1 ϫ 5 min and 1 ϫ 10 min at room temperature with 2ϫ SSC, 0.1% SDS, and then for 1 ϫ 2 min at 65°C with 0.1ϫ SSC, 0.1% SDS. The blot was rinsed in 2ϫ SSC and exposed to Kodak BioMax MR film (Eastman Kodak Co.).
Western Blotting-A polyclonal antibody directed against utrophin A was raised as previously described (18). This utrophin A antibody was purified from production bleeds of antisera using the SulfoLink Kit (Pierce), according to the manufacturer's instructions. For Western blots, cardiotoxin-treated and untreated TA muscles from 4-week-old mice were homogenized in a previously described extraction buffer (32) using a Dounce homogenizer and then boiled and centrifuged for 10 min at 10,000 ϫ g. One hundred g of the protein extracts were separated using a 5% SDS-PAGE gel with a 4% stacking gel at 200 V for 5 h. Gels were transferred to Immobilon-P polyvinylidene fluoride membranes (Millipore Corp., Bedford, MA) overnight at 4°C. Membranes were incubated with Tris-buffered saline with 0.1% Tween 20 and 5% dried skim milk for 1 h and then incubated for 1 h with the primary antibody (1:10,000) diluted in 5% milk Tris-buffered saline buffer. The membranes were washed thoroughly in Tris-buffered saline and incubated with a horseradish peroxidase-conjugated goat anti-rabbit IgG (Jackson ImmunoResearch) secondary antibody (1:25,000) for 1 h. Protein detection was carried out using SuperSignal West Dura extended duration substrate (Pierce) and exposed to Kodak BioMAX MR film. The average intensities of the bands were quantitated using Kodak digital science 1D version 3.0.2 image analysis software. Equal loading of protein was confirmed with the SYPRO Ruby protein blot stain (Molecular Probes, Inc., Eugene, OR), as recommended by the manufacturer, and was further quantified using a monoclonal anti-actin antibody (Sigma).
Immunofluorescence-Immunofluorescence experiments were performed as previously described (18). Briefly, the polyclonal anti-utrophin A antibody was used to identify utrophin A expression on longitudinal sections of untreated or cardiotoxin-treated TA muscles. Alexa 488-conjugated ␣-bungarotoxin 594 (Jackson ImmunoResearch, West Grove, PA) was used to label acetylcholine receptors and, hence, to identify the presence of neuromuscular junctions.
Statistical Analysis-The presence of significant differences between group means was determined using Student's t test. The level of significance was set at p Ͻ 0.05. Means Ϯ S.E. are shown throughout.

Discordant Expression of Utrophin A and Its mRNA in Regenerating
Muscle-Treatment of skeletal muscle with the snake venom cardiotoxin induces severe myonecrosis and subsequent muscle regeneration while leaving the innervating nerve intact. This is a well established model to study muscle regeneration (33,34). Under these conditions, we have previously shown that the content of total utrophin (using an antibody that recognized both utrophin A and B isoforms) is significantly up-regulated several days after cardiotoxin treatment (23). Here, we examined the specific effect of cardiotoxin treatment on expression of the utrophin A isoform, which is selectively expressed in skeletal muscle (15,18). Western blot analysis of muscles treated with cardiotoxin for 3-10 days revealed that the greatest levels of utrophin A expression were obtained at 7 days after cardiotoxin treatment. At this time point, we found that utrophin A expression was increased ϳ14fold (p Ͻ 0.05) compared with untreated control muscles (Fig. 2, A and  B). Immunofluorescence experiments revealed that whereas utrophin A expression in control muscles is confined to neuromuscular junctions (Fig. 3, A and B) (10 -12), the increased expression of utrophin A in regenerating muscle fibers was observed along the entire length of the sarcolemma (Fig. 3, E and F). However, in sharp contrast to this dramatic induction of protein, utrophin A mRNA levels were modestly increased in regenerating muscle (40% increase; Fig. 2C). This relatively small increase in the abundance of utrophin A mRNA is in stark contrast to the dramatic increase in protein levels, thereby suggesting that changes in protein stability and/or in the translational efficiency of utrophin A mRNA could have an important regulatory function.
The Utrophin A 5Ј-UTR Confers Translational Regulation-On the basis of these findings, we next examined the possibility that utrophin is translationally regulated and decided to specifically assess whether mechanisms involving the utrophin A 5Ј-UTR could account for the discrepancy between the levels of protein and transcript observed during muscle regeneration. To this end, we isolated the murine utrophin A 5Ј-UTR from total RNA obtained from C2C12 myogenic cells and found that it was relatively long (15). Using the mFOLD algorithm, this sequence was predicted to have a high degree of secondary structure (35). Long 5Ј-UTRs often confer an inhibitory effect on cap-dependent translation. Therefore, we inserted this fragment into a reporter vector upstream of the ␤-galactosidase reporter gene to examine whether the utrophin A 5Ј-UTR causes such translational inhibition (see schematics of constructs in Fig. 1).
The utrophin A 5Ј-UTR reporter construct (pUtrA/␤GAL) or a parental control (p␤GAL) was directly injected into control mouse TA muscles. An indication of the role that a 5Ј-UTR of interest plays in translational regulation can be obtained by measuring protein reporter activity and standardizing to the relative levels of reporter mRNA expressed from this plasmid (36,37). Results of these experiments are thus expressed as a ratio of reporter protein activity to reporter transcript levels. As shown in Fig. 4A, we observed in control TA muscles a dramatic decrease in the ratio of protein to transcript for muscles transduced with the utrophin A 5Ј-UTR reporter construct (pUtrA/␤GAL) compared with muscles transduced with the parental vector (p␤GAL). This demonstrates that the 5Ј-UTR of utrophin A confers an overall inhibitory effect on translation in skeletal muscle fibers under control, steady-state conditions.
In order to study the effects of muscle regeneration on the ratio of reporter protein activity to transcript levels, we preinjected TA muscles with cardiotoxin 3 days prior to direct plasmid injection of the reporter constructs. Muscles were excised 7 days after the initial cardiotoxin injection. In these experiments, we reasoned that if the 5Ј-UTR of utrophin A plays a role in the discordant expression of endogenous utrophin A protein and transcript during muscle regeneration (see Figs. 2 and 3), then there should be a significant increase in the ratio of protein activity to transcript levels in regenerating muscles injected with pUtrA/␤GAL. In accordance with our hypothesis, the relative ratio of protein to mRNA levels was dramatically increased in regenerating muscles injected with the construct containing the utrophin A 5Ј-UTR (Fig. 4B). Together, these results demonstrate that the inhibitory effect exerted by the utrophin A 5Ј-UTR on translation in control muscles was mitigated in cardiotoxin-treated TA muscles. These experiments further indicate that the increase in utrophin A protein expression observed in regenerating muscles (Figs. 2 and 3) involves a derepression of translation inhibition conferred by the long and highly structured utrophin A 5Ј-UTR. It should be noted that plasmid DNA uptake is significantly greater in regenerating muscle than normal muscle (38), thereby making it necessary to express the data in relative ratios of protein product activity to mRNA levels. Finally, it is important to note that removal of the CMV promoter completely abolished expression of the reporter gene in control and regenerating muscles, indicating that the utrophin A 5Ј-UTR does not contain cryptic promoter activity (data not shown).
Identification of IRES Activity in the Utrophin A 5Ј-UTR in Regenerating Muscle-The above data are coherent with the idea that utrophin A is translationally regulated in regenerating muscles by mechanisms that target the 5Ј-UTR. Most commonly, translational regulation occurs at the point of translation initiation. One possible mechanism through which specific eukaryotic mRNAs are translated during periods of "cellular stress" involves translation initiation mediated by the presence of an IRES (39). We therefore speculated that the utrophin A 5Ј-UTR contains an IRES that functions during the "stress" of cardiotoxin-induced muscle regeneration. A well established approach to assess IRES activity of 5Ј-UTRs consists in using bicistronic reporter constructs (40,41). Bicistronic vectors contain two reporter proteins, which enable the identification of IRES activity in a sequence of interest inserted between the two cistrons. The downstream cistron is only expressed if the sequence of interest contains an IRES. To test this hypothesis, we thus inserted the utrophin A 5Ј-UTR into a bicistronic vector (p␤GAL/CAT; see Fig. 1) previously used to identify IRESs (28). The first cistron (␤GAL) of this vector measures cap-dependent translation, whereas the second cistron (CAT) represents cap-independent translation.
Cardiotoxin-treated and control TA muscles were injected with the bicistronic reporter constructs using the same time frame as for injection with monocistronic reporter constructs (see above). As shown in Fig. 5, IRES activity is reported as a ratio of CAT/␤GAL activity. A ratio significantly higher than that seen with the parental vector is reflective of IRES activity. In control muscles transduced with the bicistronic constructs, no IRES activity above background was detected with the utrophin A 5Ј-UTR, since similar CAT/␤GAL ratios were found between p␤GAL/CAT-and p␤GAL/UtrA/CAT-injected muscles. In cardiotoxin-treated muscles, there was a slight induction (ϳ1.5-fold) in the ratio of CAT/␤GAL for muscles injected with the parental plasmid (but see below). However, there was a striking increase of more than 9-fold (p Ͻ 0.05) in the ratio of CAT/␤GAL in regenerating muscles injected with the bicistronic vector containing the utrophin A 5Ј-UTR (Fig. 5A). The ratio of CAT/␤GAL in p␤GAL/UtrA/CAT-injected regenerating muscles was ϳ6-fold greater that that of p␤GAL/CAT-injected regenerating muscles.  Further analysis of the data expressed as individual levels of CAT and ␤GAL revealed that the increased CAT/␤GAL ratio seen for the parental vector in regenerating muscles was caused by a decrease in the level of the ␤GAL reporter and not by an increase in CAT (Fig. 5B). Thus, the increased ratio seen for the parental vector in Fig. 5A is not reflective of spurious IRES activity but rather indicates a decrease in cap-dependent translation (determined by ␤GAL). In contrast, the increased ratio of CAT/␤GAL mediated by the presence of the utrophin A 5Ј-UTR between the cistrons and seen in cardiotoxin-treated muscles was clearly due to an increase in CAT activity (Fig. 5B). Similar to what was observed with the p␤GAL/CAT constructs, we also detected a decrease in ␤GAL activity with the p␤GAL/UtrA/CAT construct to an extent similar to that seen with the parental vector (Fig. 5B). It should be noted that the levels of ␤GAL for p␤GAL/CAT and p␤GAL/UtrA/CAT in the cardiotoxin-treated muscle were not statistically different, whereas the

. The utrophin A 5-UTR regulates expression of a reporter protein during muscle regeneration.
A, p␤GAL or pUtrA/␤GAL constructs (see Fig. 1) were directly injected into normal TA muscles along with the pCAT vector, which was used to monitor the efficiency of gene transfer. Both ␤GAL reporter activity and ␤GAL mRNA levels were standardized to CAT reporter activity and CAT mRNA levels. Values are shown as a ratio of reporter protein levels, measured by ␤-galactosidase activity, to reporter mRNA levels, measured by RT-PCR, in order to obtain an indication of relative translational efficiency. The presence of the utrophin A 5Ј-UTR causes an inhibitory effect on translation of the reporter. B, muscles were treated with cardiotoxin 3 days prior to plasmid injection to induce muscle regeneration. The regenerative state of the muscle causes the inhibitory effect of the 5Ј-UTR on translation to be partially removed, causing the reporter protein to be translated more efficiently. Mean values Ϯ S.E. are shown.  p␤GAL/CAT or p␤GAL/UtrA/CAT constructs were directly injected into control (CTL) and cardiotoxin-treated (CTX) TA muscles. A shows data reported as a ratio of CAT activity, measuring cap-independent translation, to ␤GAL activity, which measures cap-independent translation. A higher ratio of CAT/␤GAL indicates IRES activity. The utrophin A 5Ј-UTR did not display IRES activity in control muscles. By contrast, we observed a significant induction (ϩ913%) of utrophin A IRES activity in regenerating muscles. B shows the individual levels of ␤GAL and CAT used to calculate the ratios shown in A. Note that the slight increase in apparent IRES activity seen with the ratios in regenerating muscles injected with the parental vector is due to a decrease in ␤GAL activity and not an increase in CAT. In regenerating muscles injected with the constructs containing the utrophin A 5Ј-UTR, note also the decrease in ␤GAL activity but, in addition, the significant increase in CAT levels. Minus and plus signs represent control or cardiotoxin-treated muscles, respectively. *, significant differences between control and cardiotoxin-treated muscles injected with p␤GAL/CAT. #, significant differences between control and cardiotoxintreated muscles injected with p␤GAL/UtrA/CAT. Mean values Ϯ S.E. are shown; n ϭ 12/group; p Ͻ 0.05. levels of CAT were 8.3-fold greater in p␤GAL/UtrA/CAT-injected regenerating muscles compared with p␤GAL/CAT-injected regenerating muscles. Therefore, the utrophin A 5Ј-UTR mediates cap-independent translation of the CAT reporter but only in cardiotoxin-treated muscle. This demonstrates that the utrophin A 5Ј-UTR contains an inducible IRES that becomes activated during muscle regeneration.
In parallel experiments, we also performed direct plasmid injection into TA muscles using a bicistronic vector containing the 5Ј-UTR of the XIAP, which is well characterized to contain an IRES element (p␤GAL/ XIAP/CAT) (28,42). Interestingly, XIAP IRES activity was dramatically lower in cardiotoxin-treated muscles compared with control muscles (data not shown), suggesting that the IRES-activating effect of muscle regeneration is gene-specific.
To determine whether the IRES activity that we observed in regenerating muscles could be linked to aberrant splicing events, several control experiments were performed to ensure that CAT expression was due to genuine IRES-mediated translation of an intact bicistronic transcript. Due to the relatively low efficiency of transduction in mouse skeletal muscle, it is necessary to use a combination of PCR experiments to control for the possibility of splicing. First, RT-PCR was used to amplify the central region of the bicistronic RNA using primers flanking the utrophin A 5Ј-UTR sequence and portions of the ␤GAL and CAT reporter mRNAs (see Fig. 6A). As shown in Fig. 6B, we amplified RT-PCR products of the expected size using RNA extracted from cardiotoxin-treated muscles injected with p␤GAL/UtrA/CAT. Sequencing of these PCR products confirmed their identity. Using this sensitive approach, we failed to detect smaller, aberrantly spliced mRNA, further demonstrating that IRES-mediated translation of utrophin A occurs during muscle regeneration.
To further control for the possibility of aberrant splicing, additional experiments were performed. Specifically, we used quantitative RT-PCR to amplify ␤GAL and CAT cistrons of the bicistronic mRNA (Fig. 6C). We reasoned that if the CAT activity from muscles transduced with the p␤GAL/UtrA/CAT plasmid was due to IRES-mediated translation from intact, full-length bicistronic transcript, then ␤GAL and CAT cDNAs would be amplified by the quantitative PCR in a ratio of 1:1. Conversely, spurious splicing of the bicistronic mRNA would result in an increase of the CAT cistron RNA relative to ␤GAL cistron. This approach has in fact been used recently by others (43). We therefore processed RNA from control and cardiotoxin-treated muscles transduced with the parental or 5Ј-UTR containing plasmids. For all samples, we found a ratio of 1:1 (Fig. 6C). This indicates that only full-length mRNAs were transcribed from both parental and 5Ј-UTR containing plasmids in both control and cardiotoxin-treated TA muscles.
The Utrophin A IRES Is Also Active in Myoblasts-In separate experiments, we also assessed the activity of the utrophin A IRES in muscle cell culture. To this end, we transfected C2C12 myoblasts with the parental or utrophin A 5Ј-UTR-containing bicistronic vectors, harvested the cells 24 h later, and assessed IRES activity by determining reporter activity. We found that the presence of the utrophin A 5Ј-UTR conferred a ϳ9-fold (p Ͻ 0.05) induction in the ratio of CAT/␤GAL compared with the ratio seen using the parental vector (Fig. 7A). This demonstrates that the utrophin A IRES element also mediates cap-independent translation in myoblasts. Interestingly, transfected cells that were induced to undergo myogenic differentiation for 4 days showed no IRES activity mediated by the utrophin A 5Ј-UTR (data not shown).
To ensure that expression of the downstream cistron was caused by genuine internal ribosome entry, we also examined the ␤GAL and CAT data individually. As shown in Fig. 7B, the increase ratio noted in cells transfected with the p␤GAL/UtrA/CAT construct was caused by a large increase (p Ͻ 0.05) in CAT activity. To further verify that this increased CAT activity was not due to the presence of a cryptic promoter in the utrophin A 5Ј-UTR, transfections of bicistronic reporter constructs in which the CMV promoter had been deleted were performed. If the utrophin A 5Ј-UTR contains a cryptic promoter, CAT expression would occur in transfected cells despite the deletion of the viral promoter. The RT lane shows the RT-PCR product amplified from RNA extracted from TA muscles injected with p␤GAL/UtrA/CAT. Note that the transcripts are of the right molecular mass. Sequencing revealed that it contains the two portions of the reporter proteins as well as the intact utrophin A 5Ј-UTR, suggesting the presence of full-length bicistronic transcripts. No smaller PCR fragments were observed, indicating that no aberrant splicing occurred. No RT, an RT-PCR amplification of RNA extracted from TA muscle injected with p␤GAL/UtrA/CAT with no reverse transcriptase in the RT reaction. This shows that amplification occurred from RNA and not from contaminating plasmid DNA, which was eliminated by DNase I treatment (see "Experimental Procedures"). C, schematic representation of the quantitative RT-PCR strategy. D, quantitative PCR using ␤GAL and CAT primers was performed on reverse transcriptase-amplified RNA from TA muscles injected with the bicistronic reporter constructs. A ratio of amplified ␤GAL and CAT cDNA of 1:1 indicates the presence of an intact bicistronic transcript. Note that the ratio of cycle thresholds (ctCAT/ct␤GAL) is basically equal to 1, and no ratios are statistically different from each other. Mean values Ϯ S.D. are shown; p Ͼ 0.05.
Removal of the CMV promoter from the parental and utrophin A 5Ј-UTR vectors (p(-cmv)␤GAL/CAT and p(-cmv)␤GAL/UtrA/CAT, respectively) resulted in the loss of both ␤GAL and CAT reporter activity (data not shown). In agreement with these findings, no expression of an mRNA containing the CAT sequence could be detected by Northern blots performed with total RNA extracted from cells transfected with the p(-cmv)␤GAL/UtrA/CAT construct (data not shown). This further shows that the IRES activity of the utrophin A 5Ј-UTR is not due to cryptic promoter activity.
To control for the possibility of splicing events causing the expression of the CAT cistron from the p␤GAL/UtrA/CAT vector in the transfected cells in culture, we performed RT-PCR and Northern blot analyses to verify the presence of an intact bicistronic transcript. RT-PCR analysis of RNA from the p␤GAL/CATand the p␤GAL/UtrA/CATtransfected C2C12 cells amplified single transcripts of the expected sizes (Fig. 8A). The larger PCR product seen in cells transfected with p␤GAL/UtrA/CAT is due to the presence of the utrophin A 5Ј-UTR between the two cistrons. Northern blotting using a 301-bp probe for CAT was used to further confirm that the downstream cistron present in the bicistronic vector was being translated from a full-length, intact bicistronic transcript. As seen in Fig. 8B (lane 1), RNA species of the expected size were detected in cells transfected with the parental bicistronic vector (p␤GAL/CAT). As expected, Northern blot using RNA from cells transfected with the bicistronic vector containing the utrophin A 5Ј-UTR revealed the presence of a slightly larger RNA with no smaller products being detected (p␤GAL/UtrA/CAT; lane 2). We also performed a Northern blot using RNA from cells transfected with a vector containing the utrophin A 5Ј-UTR and CAT, but lacking ␤GAL, to show where a potential splice product should migrate (pUtrA/CAT; lane 3, see arrow). In addition, full-length transcripts from RNA extracted from cells transfected with the p␤GAL/CAT or p␤GAL/ UtrA/CAT constructs were also detected by Northern blot using a probe for ␤GAL (data not shown). Together, these results confirm that the IRES activity conferred by the utrophin A 5Ј-UTR is caused by genuine internal ribosome entry and not by aberrant splicing events.

DISCUSSION
In recent years, several reports have implicated that utrophin is regulated via translational and/or post-translational mechanisms (see Introduction). Specifically, the discordant patterns of expression in the levels of utrophin and its mRNA in muscle cells placed under a number of experimental conditions indicate that utrophin expression is regulated by translational mechanisms and/or protein stability. The current study was designed to examine this issue by focusing on the possibility that utrophin expression, in addition to being controlled by transcriptional mechanisms, is also subjected to important translational events that ultimately modulate the overall abundance of utrophin in muscle fibers. Here, we show that the large increase in the expression of utrophin A during muscle regeneration is clearly not accompanied by a parallel increase in utrophin A transcripts. Additionally, we show that the utrophin A 5Ј-UTR has an inhibitory effect on translation of a reporter transcript in intact skeletal muscle fibers. Strikingly, a marked derepression of this inhibition is observed in muscle fibers undergoing regeneration. Finally, using several complementary approaches, we show that the utrophin A 5Ј-UTR contains an IRES whose activity becomes highly stimulated in regenerating muscle.
IRES-mediated translation is an alternative to cap-dependent trans-  lation, the main mechanism of translation initiation in eukaryotes (44). In cap-dependent translation, the eukaryotic initiation factor eukaryotic initiation factor 4E binds to the 7-methyl guanosine-capped structure at the 5Ј-end of an mRNA. Upon association of other initiation factors, Met-tRNA i Met and the 40 S ribosomal subunit, scanning is believed to occur in a 5Ј to 3Ј direction until a start codon in a favorable sequence context is encountered, and protein synthesis begins (45). By contrast, in IRES-mediated translation, the 5Ј-UTR harbors specific sequences that are thought to recruit ribosomes directly to the vicinity of initiation codon independently of the 5Ј capped structure (39). Whereas cellular mRNAs are normally translated in a cap-dependent manner, there is a growing list of cellular mRNAs that are now known to be translated via cap-independent, IRES-mediated translation. Often, these mRNAs contain long and highly structured 5Ј-UTRs.
Using bicistronic reporter vectors, we showed that the utrophin A 5Ј-UTR displayed, as expected, no IRES activity in intact muscles. However, in muscle induced to degenerate and regenerate by cardiotoxin treatment, the utrophin A 5Ј-UTR mediated a 9-fold increase in the expression of the downstream reporter. Several control experiments revealed that the bicistronic mRNA did not undergo aberrant splicing, while also showing that the utrophin A 5Ј-UTR does not contain cryptic promoter activity. Based on these findings, it appears therefore that the induction of the muscle regenerative program serves as an appropriate signal to create an environment causing activation of the utrophin A IRES. Thus, cap-independent translation via the utrophin A 5Ј-UTR IRES can at least partially account for the large discrepancy in protein and transcript levels observed in regenerating skeletal muscle.
Skeletal muscle regeneration that follows after injury or chemical insult involves the activation of quiescent satellite cells and promotion of their entry into the mitotic phase. These cells proliferate, terminally differentiate, and fuse to form newly formed mutlinucleated muscle fibers (33). Interestingly, and in agreement with our in vivo findings, we observed in the present studies that the utrophin A 5Ј-UTR also displayed IRES activity in myoblasts growing in culture at 9.1-fold over parental levels (Fig. 7). In contrast, after transfected myoblasts were induced to differentiate to form multinucleated myotubes, we observed no significant IRES activity (data not shown). This is unlikely to be coincidental, since before they fuse to form myotubes, C2C12 myoblasts are similar to satellite cells that are activated during muscle regeneration. Our results obtained with cultured cells as well as with muscle in vivo are strongly supported by the fact that cap-dependent translation is often down-regulated during cell cycle activation and differentiation, whereas IRES-mediated translation of specific genes persists under these conditions (46,47). Indeed, ␤GAL reporter levels, which in our experimental system are indicative of cap-dependent translation, were reduced in cardiotoxin-treated samples.
Many of the mammalian IRES elements found to date mediate capindependent translation of proteins that are expressed during periods of "cellular stress" when cap-dependent translation may become less efficient (39,44,48). An interesting possibility therefore is that the "stress" of muscle regeneration causes the induction of IRES activity for a variety of mRNAs. In this context, several other proteins that are known to be involved in muscle regeneration have also been shown to be regulated via IRES-mediated translation. Specifically, it is known that injection of fibroblast growth factor II in muscles of mdx mice improves regeneration (49). Additionally, expression of insulin growth factor I receptor is up-regulated during muscle regeneration (50). It is therefore highly relevant to note that the fibroblast growth factor II 5Ј-UTR contains an IRES (51) and that the receptor for insulin growth factor I is also subject to IRES-mediated translational control (52). In addition, both the leader 2 5Ј-UTR of insulin growth factor II (53) and the A and C leaders of fibroblast growth factor I (54) can cause translation initiation via internal ribosome entry. Thus, muscle regeneration may indeed represent a type of "cellular stress" that promotes IRES-mediated translation of a subclass of mRNAs whose expression may be crucial to ensure the success of the regenerative response.
In recent years, there has been an emergence of studies identifying translational regulation at specific postsynaptic sites. The majority of such work has focused on translational events that occur in dendrites and that impact on synaptic plasticity (55)(56)(57). Such local regulation provides a mechanism for protein synthesis to occur rapidly in a distinct subcellular compartment under the influence of incoming inputs from presynaptic nerve terminals. Map 2 and Arc are two mRNAs that are localized to dendrites and whose translation can be regulated by the presence of IRES elements in their 5Ј-UTRs (58). These findings are important and supportive of our current results, since (i) utrophin A is an important component of the postsynaptic apparatus in muscle fibers, and (ii) both Map 2 and Arc are also cytoskeletal proteins.
There is solid evidence now indicating that translational regulation does occur at the Drosophila neuromuscular junction (59,60). Furthermore, there are indirect observations implicating translational regulation as an important control step in the synthesis of synaptic proteins in mammalian muscle. For instance, it has been shown that expression of the acetylcholine receptor ␣-subunit is regulated at the translational level in rat primary muscle cells (61). Moreover, a decrease in translation has been shown to account for the reduction in acetylcholinesterase expression in skeletal muscle following glucocorticoid treatment (62). Finally, we have recently demonstrated that staufen, a double-stranded RNA-binding protein involved in mRNA targeting and translation, accumulates within the postsynaptic membrane domain of the neuromuscular junction (63).
In summary, we have demonstrated that the utrophin A 5Ј-UTR can drive IRES-mediated translation during muscle regeneration. This finding provides an explanation for the discrepancy between utrophin A protein and transcript levels observed in various studies. Although these findings do not exclude the possibility that post-translational mechanisms, such as alterations in protein stability, may also partially account for these discrepancies between protein and mRNA levels (25,27,64), our data add considerably to the growing literature highlighting the importance of translational control for the regulation of synaptic proteins in muscle and nerve. Additionally, our results provide an additional target from which pharmacological interventions may be rationally designed to stimulate utrophin A expression in muscle fibers from DMD patients in attempts to alter the progression of this neuromuscular disorder and provide functional benefits.