Muscle and Neural Isoforms of Agrin Increase Utrophin Expression in Cultured Myotubes via a Transcriptional Regulatory Mechanism*

Duchenne muscular dystrophy is a prevalent X-linked neuromuscular disease for which there is currently no cure. Recently, it was demonstrated in a transgenic mouse model that utrophin could functionally compensate for the lack of dystrophin and alleviate the muscle pathology (Tinsley, J. M., Potter, A. C., Phelps, S. R., Fisher, R., Trickett, J. I., and Davies, K. E. (1996) Nature 384, 349–353). In this context, it thus becomes essential to determine the cellular and molecular mechanisms presiding over utrophin expression in attempts to overexpress the endogenous gene product throughout skeletal muscle fibers. In a recent study, we showed that the nerve exerts a profound influence on utrophin gene expression and postulated that nerve-derived trophic factors mediate the local transcriptional activation of the utrophin gene within nuclei located in the postsynaptic sarcoplasm (Gramolini, A. O., Dennis, C. L., Tinsley, J. M., Robertson, G. S., Davies, K. E, Cartaud, J., and Jasmin, B. J. (1997)J. Biol. Chem. 272, 8117–8120). In the present study, we have therefore focused on the effect of agrin on utrophin expression in cultured C2 myotubes. In response to Torpedo-, muscle-, or nerve-derived agrin, we observed a significant 2-fold increase in utrophin mRNAs. By contrast, CGRP treatment failed to affect expression of utrophin transcripts. Western blotting experiments also revealed that the increase in utrophin mRNAs was accompanied by an increase in the levels of utrophin. To determine whether these changes were caused by parallel increases in the transcriptional activity of the utrophin gene, we transfected muscle cells with a 1.3-kilobase pair utrophin promoter-reporter (nlsLacZ) gene construct and treated them with agrin for 24–48 h. Under these conditions, both muscle- and nerve-derived agrin increased the activity of β-galactosidase, indicating that agrin treatment led, directly or indirectly, to the transcriptional activation of the utrophin gene. Furthermore, this increase in transcriptional activity in response to agrin resulted from a greater number of myonuclei expressing the 1.3-kilobase pair utrophin promoter-nlsLacZ construct. Deletion of 800 base pairs 5′ from this fragment decreased the basal levels of nlsLacZ expression and abolished the sensitivity of the utrophin promoter to exogenously applied agrin. In addition, site-directed mutagenesis of an N-box motif contained within this 800-base pair fragment demonstrated its essential contribution in this regulatory mechanism. Finally, direct gene transfer studies performed in vivo further revealed the importance of this DNA element for the synapse-specific expression of the utrophin gene along multinucleated muscle fibers. These data show that both muscle and neural isoforms of agrin can regulate expression of the utrophin gene and further indicate that agrin is not only involved in the mechanisms leading to the formation of clusters containing presynthesized synaptic molecules but that it can also participate in the local regulation of genes encoding synaptic proteins. Together, these observations are therefore relevant for our basic understanding of the events involved in the assembly and maintenance of the postsynaptic membrane domain of the neuromuscular junction and for the potential use of utrophin as a therapeutic strategy to counteract the effects of Duchenne muscular dystrophy.

phin gene. Furthermore, this increase in transcriptional activity in response to agrin resulted from a greater number of myonuclei expressing the 1.3-kilobase pair utrophin promoter-nlsLacZ construct. Deletion of 800 base pairs 5 from this fragment decreased the basal levels of nlsLacZ expression and abolished the sensitivity of the utrophin promoter to exogenously applied agrin. In addition, site-directed mutagenesis of an N-box motif contained within this 800-base pair fragment demonstrated its essential contribution in this regulatory mechanism. Finally, direct gene transfer studies performed in vivo further revealed the importance of this DNA element for the synapse-specific expression of the utrophin gene along multinucleated muscle fibers. These data show that both muscle and neural isoforms of agrin can regulate expression of the utrophin gene and further indicate that agrin is not only involved in the mechanisms leading to the formation of clusters containing presynthesized synaptic molecules but that it can also participate in the local regulation of genes encoding synaptic proteins. Together, these observations are therefore relevant for our basic understanding of the events involved in the assembly and maintenance of the postsynaptic membrane domain of the neuromuscular junction and for the potential use of utrophin as a therapeutic strategy to counteract the effects of Duchenne muscular dystrophy.
Duchenne muscular dystrophy (DMD) 1 is the most severe and prevalent neuromuscular disease affecting 1 in 3,500 male births (1). This disease is characterized by repeated cycles of muscle fiber degeneration and regeneration with an eventual failure to regenerate, thereby leading to a loss of muscle mass and function. The genetic defect underlying DMD, located on the short arm of the X chromosome, prevents the production of dystrophin, a large cytoskeletal protein of the spectrin superfamily (2,3). Previous studies have shown that, in muscle, dystrophin is located at the cytoplasmic face of the sarcolemma where it links the intracellular cytoskeleton network to the extracellular matrix via a complex of dystrophin-associated proteins (for reviews, see Refs. 4 -7).
Several years ago, an autosomal homologue to dystrophin was identified on chromosome 6q24 (8). This gene, now referred to as utrophin, presents a genomic organization similar to that of the dystrophin gene, indicating that both genes evolved from an ancestral duplication event (9). Cloning of a full-length cDNA and subsequent analysis of its deduced amino acid sequence revealed, in fact, that utrophin shares considerable identity with dystrophin, particularly in the actin binding domain and carboxyl terminus (10). However, in comparison to high molecular mass isoforms of dystrophin, which are predominantly expressed in brain and muscle, utrophin displays a ubiquitous pattern of expression since it can be detected in most tissues (11)(12)(13).
In normal skeletal muscle, expression of utrophin is known to be influenced by the state of differentiation and innervation of muscle fibers. In developing myotubes, for example, utrophin is first localized to the entire length of the sarcolemma (14 -17). Following the establishment of synaptic contacts, utrophin becomes highly enriched within the postsynaptic membrane domain of the neuromuscular junction (18,19). However, several studies have shown that in dystrophic muscles, utrophin expression is not restricted to postsynaptic compartments, since it extends well into extrasynaptic regions of adult muscle fibers (14, 20 -23). Such modulations in the pattern of expression indicate that distinct cellular and molecular mechanisms must exist to maintain the uneven distribution of utrophin along normal adult muscle fibers and to alter its levels and localization in developing and diseased muscles.
Despite these recent advances, however, our knowledge of the regulatory mechanisms presiding over utrophin expression in muscle is clearly lacking. A better understanding of these mechanisms appears important particularly since up-regulation of utrophin is currently envisaged as a therapeutic strategy to prevent the relentless progression of DMD (24,25). In this context, we have recently shown that the nerve exerts a profound influence on utrophin gene expression (26). Since our previous experiments also demonstrated that nerve-derived electrical activity is not a key factor regulating utrophin expression (27), we postulated in these initial studies that nervederived trophic factors likely mediate the local transcriptional activation of the utrophin gene within nuclei of the postsynaptic membrane domain (26). In the present study, we have therefore determined the effects of nerve-derived trophic factors on utrophin expression in cultured myotubes. A preliminary account of this work has previously appeared in abstract form (28).

EXPERIMENTAL PROCEDURES
Tissue Culture-C2 cells were cultured on Matrigel-coated (Collaborative Biomedical Products, Bedford, MA) 35-mm culture plates and kept at 37°C in a water-saturated atmosphere containing 5% CO 2 . Myoblasts were grown in Dulbecco's modified Eagle's medium supplemented with 20% horse serum, 10% fetal bovine serum, 100 units/ml penicillin-streptomycin, and 292 ng/ml L-glutamine until they reached confluence. At this stage, the concentration of horse serum was reduced to 5%, and fetal bovine serum was eliminated to promote myotube formation. Myoblasts were allowed to fuse into multinucleated myotubes for 3-4 days and were then used for experiments. To examine the effects of nerve-derived trophic factors, 0.1 M rat CGRP (Sigma) or 10 ng/ml purified Torpedo agrin (29) was added directly to the culture medium for 24 -48 h. Additionally, the effects of 1 nM recombinant neural (C-Ag 12,4,8 ) or muscle (C-Ag 12,0,0 ) isoforms of agrin were also examined (30).
Immunofluorescence and Quantitation of Acetylcholine Receptors (AChR) Clusters-Differentiated C2 myotubes were treated with 10 ng/ml Torpedo or recombinant agrin for 24 -48 h. Cultures were subsequently fixed for 10 min in 4% paraformaldehyde. Clusters of AChR were visualized with fluorescein isothiocyanate-conjugated ␣-bungarotoxin used at a final concentration of 4 ng/ml in phosphate-buffered saline (PBS). Following thorough washing with PBS, the myotubes were covered with a glycerol:PBS solution and a coverslip, and they were then examined by epifluorescence using a Zeiss photomicroscope.
For the determination of agrin-induced AChR clusters, the numbers of myotubes and AChR aggregates were determined in 10 fields of view per culture at a 400 ϫ magnification as described in detail in Gee et al. (31). A minimum of four cultures were quantitated for each experimental condition. Photographs were taken with Kodak T-MAX 400 black and white films.
Immunoblotting-C2 myotubes were treated with agrin for 48 h, washed in PBS, and then solubilized in RIPA (1% sodium deoxycholate, 0.1% SDS, 0.5% Triton X-100, 1 mM phenylmethylsulfonyl fluoride, 5 mM iodoacetamide, 2 g/ml aprotinin, 0.01 M Tris-HCl, pH 8.0, 0.14 M NaCl, and 0.025% NaN 3 ) (32). Samples were centrifuged, and the supernatant was collected and stored at Ϫ20°C until analysis. The resulting pellet was further solubilized in RIPA containing 5% SDS. Following centrifugation, the supernatant was collected and stored at Ϫ20°C. The concentration of SDS-solubilized protein was determined using the bicinchoninic acid protein assay reagent protocol (BCA; Pierce). Equivalent amounts of cell extracts (70 g) were separated on a 6% polyacrylamide gel and electroblotted onto a polyvinylidene difluoride membrane (Sigma). To ensure that equivalent amounts of proteins were loaded for each sample, membranes were also stained with Ponceau S (Sigma). Membranes were subsequently incubated with monoclonal antibodies directed against either utrophin (MANCHO-7; kindly supplied by Dr. Glen Morris, N.E. Wales Institute, UK), ␣-actinin (Sigma), or sarcomeric myosin (MF-20) (33). Bound antibodies were detected by secondary antibodies linked to horseradish peroxidase and revealed via chemiluminescence using a commercially available kit (NEN Life Science Products). Membranes were then exposed onto Bi-oMax autoradiographic films (Eastman Kodak Co.), developed, and scanned at 200 dots/inch using a Hewlett-Packard Scanjet 4C.
RNA Extraction and Reverse Transcription-Polymerase Chain Reaction (RT-PCR)-Total RNA was extracted using Trizol as recommended by the manufacturer (Life Technologies, Inc.). Briefly, cells were scraped into 1 ml of Trizol. Following addition of 200 l of chloroform, the samples were mixed vigorously and centrifuged at 12,000 ϫ g for 15 min at 4°C. The aqueous layer was then transferred to a fresh tube, and 500 l of ice-cold isopropanol were added. For RNA precipitation, the isopropanol mixture was spun, and the resultant pellets washed twice with ice-cold 75% ethanol.
For all samples, total RNA was redissolved into 20 l of RNase-free water. From each of these stocks, the RNA was further diluted to a final concentration of 50 ng/l, and only 2 l of this dilution were used for RT-PCR as described in detail in Jasmin et al. (27,34). Briefly, a RT master mix was prepared containing 5 mM MgCl 2 , 1 ϫ PCR buffer II (50 mM KCl, 10 mM Tris-HCl, pH 8.3, 1 mM dNTPs, 20 units of RNase inhibitor, 50 units of reverse transcriptase, and 2.5 mM of random hexamers (GeneAmp RNA PCR kit; Perkin-Elmer Corp.). The master mix was aliquoted, and the appropriate RNA sample was subsequently added. Negative controls consisted of RT mixtures in which the RNA sample was replaced with RNase-free water. RT was performed for 45 min at 42°C, and the mixture was heated to 99°C for 5 min to terminate the reaction.
Complementary DNAs encoding utrophin and dystrophin were specifically amplified using primers designed on the basis of available mouse cDNA sequences (27,34). Amplification of the selected cDNAs was performed in a DNA thermal cycler (Perkin-Elmer) by adding 4 l of the RT mixture to 16 l of a PCR master mix. Each cycle of amplification for utrophin cDNAs consisted of denaturation at 94°C for 1 min, primer annealing at 60°C for 1 min, and extension at 72°C for 1 min. For dystrophin amplification, each cycle consisted of denaturation at 94°C for 1 min, followed by primer annealing and extension at 72°C for 3 min. The number of cycles for utrophin and dystrophin was 26 and 44, respectively. PCR products were visualized on a 1.5% agarose gel containing ethidium bromide. The 100-bp molecular mass marker (Life Technologies, Inc.) was used to estimate the molecular mass of the PCR products. Quantitative PCR experiments were performed to determine strictly the relative abundance of transcripts following different experimental treatments. These experiments were carried out using either one of two methods. In one case, 1.5 ϫ 10 6 cpm per sample of 32 P end-labeled primers were added to the PCR master mix. PCR products were visualized and carefully excised from the agarose gel with the use of a scalpel. The level of radioactivity present in these gel bands was determined by Cerenkov counting. Alternatively, PCR products were separated in 1.5% agarose gels containing the fluorescent dye Vistra Green (Amersham Corp.), and the labeling intensity of the PCR product, which is linearly related to the amount of DNA, was quantitated using a Storm PhosphorImager (Molecular Dynamics, Sunnyvale, CA).
Expression of Utrophin Promoter-Reporter Gene Constructs-Several human utrophin promoter-reporter gene constructs were used in these experiments (35). These 1.3-and 0.5-kb promoter fragments were inserted upstream of the reporter gene LacZ and a nuclear localization signal (26). Additionally, two other 1.3-kb constructs were generated with mutations of the N-box. The 1.3-kb HindIII human utrophin promoter clone (35) was digested with XhoI and PstI liberating a 300-bp fragment containing the N-box, which was then further cloned into pBSSKII(Ϫ) (Stratagene, Cambridge, UK) generating the clone pBSXP. Mutagenesis was performed using Quick Change (Stratagene) essentially following the manufacturer's instructions except for using cloned Pfu polymerase (Stratagene). Two pairs of complementary primers were generated with a single or double point mutation in the N-box (N5F, 5-U-GTG GGG CTG ATC TTC CAG AAC AAA GTT GC; N5R, 5-U-GCA ACT TTG TGG AAG ATC AGC CCC AC; N34F, 5-U-GGG GCT GAT CTT TTG GAA CAA AGT TGC TGG G; and N34R, 5-U-CCC AGC AAC TTT CTT CCA AAA GAT CAG CCC C). pBSXP was used as the template for synthesis of the mutations using these oligonucleotide primer pairs. Following 15 cycles of 95°C for 30 s, 56°C for 1 min, 68°C for 7 min, the wild-type plasmid template was destroyed using the methylation-sensitive restriction endonuclease DpnI. The mutant plasmids were cloned and sequenced to verify the addition of the mutations in the N-box and to confirm that no new mutations had been introduced into other sequences. The 300-bp XhoI/PstI was released and used to replace the equivalent nonmutated fragment at the same sites in the plasmid 1.3-kb nlsLacZ (26). The new promoter mutant/reporter constructs were then sequenced to check for no further mutations. For transfection and direct gene transfer experiments, plasmid DNA was prepared using the Qiagen mega-prep procedure (Qiagen, Chatsworth, CA).
C2 myoblasts were transfected with 3 g of the appropriate utrophin promoter-reporter gene construct using the mammalian transfection system-calcium phosphate kit (Promega, Madison, WI). Once the cultures became confluent, the medium was switched to the differentiation medium (see above) to stimulate myotube formation. Three to 4 days later, agrin was added to the medium for 48 h. Levels of ␤-galactosidase activity were then determined using either a histochemical staining procedure (26) or a biochemical assay (Promega ␤-galactosidase enzyme system). For the biochemical assays, the levels of ␤-galactosidase activity were normalized according to a cotransfected chloramphenicol acetyltransferase (CAT) plasmid (Promega) and protein content. In these experiments, the cotransfected CAT plasmid allowed for the correction of any variation due to differences in transfection efficiency between culture wells. CAT activity was determined using a CAT enzyme assay system (Promega) while protein content was determined by the BCA method (see above).
For direct gene transfer into mouse tibialis anterior muscles, experiments were performed as described previously (26). Briefly, 25 l of DNA solution (2 g/l) were injected directly into the muscles of 4-week-old mice. Muscles were excised 2 weeks following injection and they were quickly frozen in melting isopentane precooled with liquid nitrogen. Cryostat tissue sections were then processed for ␤-galactosidase and acetylcholinesterase (AChE) histochemistry (26). The position of blue myonuclei clusters, indicative of utrophin promoter activity and designated as an event, was determined and compared with the presence of neuromuscular junctions using the quantitative procedure recently established by Duclert et al. (36).
Statistical Analysis-Paired Student's t tests were performed to evaluate the effects of agrin on utrophin expression. These tests were used to strictly compare the effects of agrin-treated versus nontreated myotubes. The level of significance was set at p Ͻ 0.05. Data are expressed as mean Ϯ S.E. throughout.

Agrin Increases Expression of Utrophin in Cultured Myotubes-
In an initial series of experiments, 3-4-day-old myotubes were treated with agrin purified from Torpedo electric tissue or with recombinant agrin isoforms in attempts to identify putative extracellular cues capable of regulating utrophin gene expression. As expected, agrin treatment increased the number of AChR clusters present on the surface of these C2 myotubes (Fig. 1). Quantitative analyses revealed that the number of AChR clusters per myotube increased by approximately 15-fold (p Ͻ 0.05) following Torpedo agrin treatment (Fig. 1). Immunofluorescence experiments using the monoclonal antibody MANCHO 7 showed that utrophin was present at these AChR clusters but only at the largest ones (data not shown). As expected, treatment of myotubes with the predominant isoform of agrin expressed in muscle (C-Ag 12,0,0 ) failed to induce the formation of AChR clusters above the levels normally detected in nontreated cultures.
Next, we examined whether agrin treatment which not only led to AChR clustering but also to the reorganization of the subsarcolemmal cytoskeleton, also influenced expression of utrophin in C2 myotubes. To this end, myotubes were treated with agrin and 48 h later, they were solubilized sequentially in RIPA containing either 0.1% or 5% SDS (see "Experimental Procedures"). Western blotting experiments showed that agrin treatment increased the levels of utrophin in 0.1% SDS-extracted proteins ( Fig. 2A). Ponceau staining of the membranes prior to immunoblotting confirmed that an equal amount of proteins had been loaded in each lane of the gel (data not shown). To further ensure that similar amounts of proteins were present in each lane, the same membranes were also processed to determine the levels of sarcomeric myosin and ␣-actinin. In these experiments, we observed that the amount of sarcomeric myosin ( Fig. 2A) and ␣-actinin (data not shown) were similar between agrin-treated versus nontreated myotubes. By contrast, the levels of utrophin extracted from the initial pellets with RIPA containing a higher concentration of SDS was not affected by agrin treatment (Fig. 2B). These results suggest that with the initial extraction buffer containing low levels of SDS, we primarily extracted utrophin not yet incorporated into the cytoskeleton which may thus reflect newly synthesized molecules. The observation that agrin increases the levels of utrophin in a readily extractable fraction indicates that agrin not only leads to a redistribution of preex- isting synaptic molecules onto the surface of myotubes but that it can also increase expression of these synaptic components.
Agrin Stimulates Transcription of the Utrophin Gene-To determine if the increase in utrophin following agrin treatment resulted from enhanced transcriptional activation of the utrophin gene, we first examined the levels of utrophin transcripts in agrin-treated versus nontreated myotube cultures by RT-PCR. Quantitative analysis revealed that utrophin mRNA levels increased significantly (p Ͻ 0.05) following Torpedo agrin treatment (Figs. 3 and 4). Recombinant neural agrin (C-Ag 12,4,8 ) had a similar effect (Fig. 4) thus ruling out the possibility that the increased expression of utrophin transcripts seen after treatment with Torpedo agrin was caused by contaminants present in this purified extract. Interestingly, treatment of myotubes with the muscle isoform of agrin (C-Ag 12,0,0 ) also increased the expression of utrophin mRNAs by approximately 2-fold (Fig. 4). Myotubes treated for 48 h with Torpedo or recombinant isoforms of agrin showed slightly higher increases in the levels of utrophin transcripts in comparison to those observed following 24 h-treatments (data not shown). In these experiments, agrin did not affect the levels of dystrophin transcripts (Fig. 3).
In separate experiments, we also determined the effects of CGRP, a neuropeptide enriched at the motor endplate and known to affect expression of AChR in cultured myotubes (for review, see Refs. 37 and 38). In contrast to the effects seen with agrin, CGRP treatment of C2 myotubes failed to induce expression of utrophin mRNA (Figs. 3 and 4). In agreement with previous reports (39), however, we nonetheless consistently observed in these experiments a small but significant 1.4-fold increase in the levels of transcripts encoding the AChR ␣-subunit following CGRP treatment (data not shown).
We next performed a series of experiments in which human utrophin promoter-reporter gene constructs were transfected into C2 myoblasts. Three-to 4-day-old myotubes were then treated with agrin, and 48 h later, the activity of ␤-galactosidase was determined and normalized to CAT activity and pro-tein content. As illustrated in Fig. 5, we observed a marked increase in the expression of the reporter gene in cultures transfected with the construct containing the 1.3-kb utrophin promoter fragment and treated with agrin. In fact, quantitative analyses showed that both muscle-(C-Ag 12,0,0 ) and nerve-derived (C-Ag 12,4,8 ) isoforms of agrin increased the expression of In the lower panel, the same membrane was subsequently striped and reprocessed for immunoblotting using the MF-20 antibody against sarcomeric myosin. Note the relative increase in utrophin following agrin treatment. B, the result of an immunoblot performed using protein extracted from the initial pellet with RIPA containing 5% SDS. Note that within this cellular fraction, utrophin levels were not affected by the agrin treatment.

FIG. 3. Representative examples of ethidium bromide-stained gels of RT-PCR products obtained from nontreated (lane 1) versus treated (lane 2) myotubes.
A and B, the effect of agrin on utrophin and dystrophin mRNA levels, respectively. Note the relative increase in utrophin mRNA levels following agrin treatment. C, the level of utrophin mRNAs in control (1) and CGRP-treated (2) myotubes. As shown, CGRP did not affect utrophin mRNA levels in these cultured myotubes. In all panels, the negative control lane is marked with a minus sign. The molecular mass of the PCR products is shown in base pairs. ␤-galactosidase by more than 2-fold (p Ͻ 0.05). In contrast, agrin treatment of myotubes transfected with the 0.5-kb utrophin promoter-reporter gene construct failed to induce expression of ␤-galactosidase above basal levels. Taken together, these results indicate therefore, that regulatory sequences contained within the deleted 800-bp fragment of the utrophin promoter are essential for transcriptional activation of the utrophin gene following agrin treatment.
On the basis of these findings, it became important to determine whether the increase in the activity of ␤-galactosidase was due to an increase in the number of myonuclei expressing detectable levels of the reporter gene or, alternatively, to an enhanced level of expression in myonuclei already expressing ␤-galactosidase. To address this issue, we histochemically stained transfected cultures for ␤-galactosidase and counted the number of positive myonuclei in control versus agrintreated myotube cultures. This analysis was justified and statistically valid for two reasons. First, our biochemical experiments (see above) showed that transfection efficiency did not vary markedly from one culture dish to another as evidenced by the relatively constant levels of CAT used to normalize ␤-galactosidase activity. In fact, we noted in these experiments that CAT levels varied by less than 15% between transfected culture dishes. Second, quantitative analysis showed that the number of ␤-galactosidase-positive myonuclei increased significantly (p Ͻ 0.05) following agrin treatment (Fig. 6), thereby eliminating the contribution of a random experimental event, such as transfection efficiency, to the overall results. Similar to our data obtained by determining biochemically the activity of ␤-galactosidase and normalizing it to CAT activity and protein content (Fig. 5), this effect was observed with both muscle (C-Ag 12,0,0 ) and neural (C-Ag 12,4,8 ) isoforms of agrin (Fig. 6). Taken together, these data show therefore that the increase in ␤-galactosidase activity observed in our biochemical assays resulted primarily from a greater number of nuclei expressing the 1.3-kb utrophin promoter-reporter gene construct. In agreement with our biochemical data, we also observed that agrin treatment of myotubes transfected with the construct containing the 0.5-kb utrophin promoter fragment failed to increase the number of ␤-galactosidase-positive nuclei, thereby further highlighting the importance of regulatory elements contained within the deleted 800-bp promoter fragment for the transcriptional activation of the utrophin gene in response to agrin.
Role of the N-box Motif in Regulating Utrophin Gene Expression-Based on recent studies, which have shown that the N-box motif plays a crucial role in regulating the expression of genes encoding the ␦and ⑀-subunits of the AChR (36,40), we examined the contribution of this DNA element in the transcriptional regulation of the utrophin gene by agrin. For these studies, site-directed mutagenesis was used to introduce single or double-base pair mutations into the N-box motif contained within the utrophin promoter (26,35). Two different mutants were generated and differed from the wild-type N-box (TTC-CGG) by one (N5 ϭ TTCCAG) or two bases (N34 ϭ TTTTGG). The mutant utrophin promoter fragments were inserted upstream of the nlsLacZ reporter gene.
In contrast to the 2-3-fold induction in the activity of ␤-galactosidase driven by the wild-type 1.3-kb utrophin promoter fragment seen following agrin treatment (Figs. 5 and 7), both N-box mutant constructs failed to display a similar responsiveness to agrin (Fig. 7). Quantitative analyses revealed that expression of ␤-galactosidase driven by either one of the two N-box mutant promoter fragments was not significantly (p Ͼ 0.05) different between agrin-treated versus non-treated myotube cultures. These results strongly indicate therefore that the N-box motif is involved in the regulatory mechanism governing expression of the utrophin gene in response to agrin.
The N-box Motif Regulates the Synaptic Expression of the Utrophin Gene in Vivo-To determine whether the N-box motif participates also in the regulation of the utrophin gene in vivo Asterisks denote significant differences from levels seen with the 1.3-kb fragment (p Ͻ 0.05). (26,36,40), we injected directly into mouse tibialis anterior muscles constructs containing either the 1.3-kb wild-type utrophin promoter fragment or the N-box mutants. In agreement with our previous findings (26), we observed that ϳ55% of all blue myonuclei clusters seen in muscles injected with constructs containing the wild-type 1.3-kb promoter fragment coincided with the presence of neuromuscular junctions (Fig. 8).
Mutations of the N-box, however, led to a marked reduction in the percentage of synaptic events. In fact, quantitative analysis revealed that, in muscles injected with either one of the mutant constructs, less than 20% of all blue myonuclei clusters were located in the vicinity of neuromuscular junctions (Fig. 8).
These results indicate, therefore, that the N-box motif regulates also in vivo expression of the utrophin gene since it modulates its pattern of synaptic expression.
Finally, to gain insight into the mechanisms contributing to the local transcriptional regulation of the utrophin gene along muscle fibers in vivo, we determined the total number of synaptic versus extrasynaptic events per muscle following injection of constructs containing the wild-type 1.3-kb utrophin promoter fragment or the N5 mutant. In these experiments, we focused our analysis on the N5 mutant since the total number of ␤-galactosidase-positive fibers seen after injection with this construct was similar to that observed following injection with the construct containing the wild-type promoter fragment (Fig.  9A). Interestingly, we observed a significantly lower number of synaptic events per muscle following injection of the N5 mutant construct as compared with the wild-type 1.3-kb utrophin promoter fragment (Fig. 9B). In contrast, the number of events in extrasynaptic regions of muscle fibers was similar between these two constructs (Fig. 9C). Therefore, these results suggest that the N-box motif contributes to the local transcriptional activation of the utrophin gene within myonuclei of the postsynaptic sarcoplasm by increasing its expression in this specialized region of muscle fibers as opposed to repressing its activity in extrasynaptic compartments (see Duclert et al. (36) for further discussion).

DISCUSSION
In a recent study, we demonstrated that utrophin transcripts accumulate preferentially within the postsynaptic sarcoplasm of muscle fibers and that this accumulation resulted from the local transcriptional activation of the utrophin gene in myonuclei concentrated beneath the neuromuscular junction (26). Induction of ectopic synapses at sites distant from the original neuromuscular junctions further revealed that nuclei located in extrasynaptic regions were capable of expressing utrophin upon receiving appropriate neuronal cues. Together with the demonstration that levels of utrophin in muscle are largely insensitive to elimination of nerve-evoked electrical activity (19,27), these experiments led us to postulate that nervederived trophic factors regulate locally the expression of the utrophin gene (25,26). Among the molecules known to regulate the expression or localization of AChR (for review, see Refs. 37 and 38), agrin appeared as a plausible candidate for several reasons. For example, detailed analysis of agrin (41)-and muscle-specific kinase (42)-deficient mice has led to the suggestion that, in vivo, agrin may ultimately affect transcription of genes encoding synaptic proteins such as AChR. Moreover, in response to exogenously applied agrin, cultured myotubes show increase numbers of AChR clusters with only large ones containing utrophin (43,44). Although agrin treatment leads to a redistribution of normally diffusing AChR molecules, it is unlikely that it causes a similar clustering of presynthesized, membrane-attached utrophin. The presence of utrophin in large AChR clusters may thus result from compartmentalized de novo expression of utrophin by nuclei located in the vicinity of the growing clusters. In the present study, we have therefore focused on the effect of agrin on utrophin expression.
In attempts to determine whether agrin treatment induced utrophin expression, we initially measured levels of utrophin and its mRNA in cultures of treated versus nontreated myotubes. In addition to causing the clustering of AChR, agrin treatment also increased the levels of utrophin. In these exper- FIG. 7. The N-box motif is critical for mediating the response to agrin. Human utrophin promoter constructs (wild-type 1.3-kb or N-box mutants N5 and N34) were inserted upstream of the reporter gene nlsLacZ and transfected in myoblasts. Myotubes were then incubated with agrin and 48 h later, the levels of ␤-galactosidase activity were determined and normalized to CAT activity and protein content. Note that the increase in the activity of the reporter gene driven by the 1.3-kb utrophin promoter fragment following treatment with both muscle (C-Ag 12,0,0 ) and neural (C-Ag 12,4,8 ) isoforms of agrin is abolished in myotubes transfected with constructs containing the N-box mutants (N5 and N34). Shown are the results of a minimum of five independent experiments. Asterisks denote significant differences from control levels (p Ͻ 0.05). iments, we observed that utrophin levels increased within an easily dissociated cellular fraction, thereby suggesting that this increase resulted from a newly synthesized pool of utrophin not yet intertwined within the existing cytoskeleton. Similarly, we also noted that agrin treatment induced a significant 2-fold increase in the levels of utrophin transcripts. Interestingly, both nerve-and muscle-derived isoforms of agrin had a comparable stimulatory effect on utrophin expression. These increases are in fact of similar magnitude to those reported recently by Jones et al. (45) who examined the impact of both muscle and neural isoforms of agrin on expression of transcripts encoding the AChR ⑀-subunit. However, a major difference between the two studies is that we were able to observe an effect on utrophin gene expression without the necessity of agrin being substrate-bound (45). Although the reason for this difference remains currently obscure, it appears reasonable to assume that it likely arises from differences in culture conditions. In particular, recent experiments have revealed that Matrigel TM is capable of binding agrin (46,47). Since, in our experiments, myotube cultures are plated on Matrigel-coated plates, it appears likely that Torpedo agrin as well as recombinant agrin fragments may become bound to this substrate via an unknown mechanism (see Denzer et al. (46,47) for further discussion) and therefore do not remain in a "soluble" form (45). Nonetheless, since the pattern of expression of the utrophin gene along muscle fibers resembles that of the ⑀-subunit gene (26,36,48), these results are coherent with the notion that expression of genes encoding membrane and cytoskeletal proteins of the postsynaptic membrane are co-regulated and therefore involve a common signal transduction pathway.
Transfection experiments with utrophin promoter-reporter gene constructs indicated that the increase in utrophin mRNA levels following agrin treatment resulted from the transcriptional activation of the utrophin gene. In agreement with our previous in vivo studies (26), deletion of 800 bp from the 3Ј region of the 1.3-kb promoter fragment significantly reduced the activity of the reporter gene in transfected cells. More importantly, it also abolished the response to agrin treatment. Together, these results indicate that DNA elements contained within the deleted 800 bp are not only regulating the basal level of utrophin gene expression in muscle cells in vivo (26) and in vitro (this study), but they also confer to the utrophin promoter its sensitivity to neuronal cues including agrin. Among the putative elements that may play a crucial role in this regulatory mechanism is the N-box motif (26,35,40), which was shown recently to be essential for the synapsespecific expression of AChR ␦and ⑀-subunit genes (36,40). In the present study, site-directed mutagenesis confirmed that the N-box motif is indeed essential in this regulatory mechanism. These results further suggest that the N-box motif may in fact represent the ultimate target within the utrophin promoter that mediates the agrin effect in cultured myotubes. In addition, it appears that this DNA element also plays an essential role in vivo in the regulation of the utrophin gene, since direct injection of constructs containing mutant utrophin promoter fragments into tibialis anterior muscles failed to induce synapse-specific expression of the reporter gene as observed with the wild-type 1.3-kb utrophin promoter fragment (26).
The molecular mechanism by which nerve-and muscle-derived isoforms of agrin lead to the transcriptional activation of the utrophin gene remains to be established. In this context, however, there are several pathways that may be currently envisaged. One signaling pathway involves binding of agrin to a complex that includes the tyrosine kinase receptor musclespecific kinase and a myotube-specific accessory component (49). This binding is known to trigger a series of biochemical events that culminate in the clustering of AChR on the surface of myotubes and in a reorganization of the underlying cytoskeleton. However, this pathway is probably not directly involved since only neural agrin activates muscle-specific kinase and induces AChR clustering (49).
A more likely mechanism responsible for the agrin-induced effects on utrophin gene expression involves not only clustering of AChR but also of other postsynaptic membrane proteins that, in turn, may directly participate in the regulation of utrophin. For example, it has been recently demonstrated that intramuscular injections of plasmid DNA-encoding agrin into extrasynaptic regions of denervated soleus muscle fibers induced, in addition to AChR clustering, the aggregation of muscle-derived ARIA along with its receptors, erbB2 and erbB3 (50). Since these molecules are known to regulate expression of AChR subunit genes (51-53), agrin treatment may thus ultimately stimulate ARIA-dependent gene expression via an autocrine mechanism involving muscle ARIA and its receptors FIG. 9. The N-box motif increases expression of the utrophin gene in synaptic regions of muscle fibers. A, the absolute number of ␤-galactosidase-positive fibers per muscle injected with constructs containing either the wild-type 1.3-kb utrophin promoter fragment or the N-box mutant N5. B, the number of synaptic events per muscle for each construct. Note that the amount of synaptic events was significantly decreased in the N5 mutant-injected muscles. Conversely, there was no difference in the number of extrasynaptic events per muscle between muscles injected with the wild-type 1.3-kb promoter fragment and the N5 mutant (C). Shown are the results obtained with a minimum of 17 injected muscles per construct. The asterisk denotes a significant difference between the two constructs (p Ͻ 0.05). (45,50). Accordingly, agrin may be sufficient for: (i) the initial events underlying AChR clustering and (ii) the positioning of other molecules involved in regulating expression of synaptic proteins. Such a role for agrin would thereby ensure the proper growth of developing postsynaptic membrane domains as well as their long term maintenance. Furthermore, it could also explain the presence of utrophin only in large AChR clusters, since recruitment of all necessary components would parallel the growth of the clusters. In fact, this mechanism is consistent with our statistical analysis demonstrating that the agrin effect on the activity of the reporter gene was caused by a significantly greater number of nuclei expressing the 1.3-kb construct as opposed to a similar number of nuclei increasing their level of expression. These results indicate therefore, that the effect of agrin is to stimulate transcription of the utrophin gene in normally quiescent nuclei; an expected effect given that agrin increases the number of clusters containing AChR and other synaptic proteins on the surface of these myotubes. In the case of muscle-derived agrin however, the effect on utrophin gene expression likely occurs via a mechanism altogether distinct from that involving the muscle-specific kinase-dependent pathway (45). Finally, it is also conceivable that the effects of both muscle and neural isoforms of agrin occurs via a distinct and unique pathway involving therefore a MuSK-independent mechanism. For example, as a protein of the extracellular matrix, agrin may activate transcription of synaptic genes by first binding to other receptors such as the integrins or ␣-dystroglycan which are known to accumulate at developing postsynaptic membrane domains (54,55). We are currently examining these possibilities using several experimental approaches.
In a recent study, Tinsley et al. (24) showed that expression of utrophin in extrasynaptic regions of muscle fibers from mdx mice functionally compensated for the lack of dystrophin and alleviated the dystrophic pathology. These findings demonstrate that up-regulation of utrophin may indeed represent an effective treatment for DMD. In this context, the next logical step is naturally to identify molecules capable of increasing utrophin gene expression in skeletal muscle fibers. Our observation that agrin increases levels of utrophin protein and mRNA via a transcriptional regulatory mechanism is therefore not only relevant for our basic understanding of the events involved in the assembly and maintenance of the postsynaptic membrane domain of the neuromuscular junction but also, for the potential use of utrophin as a therapeutic strategy for DMD.