Subtle Neuromuscular Defects in Utrophin-deficient Mice

To construct a targeting vector we isolated a 14-kb genomic fragment from a 129 mouse genomic library (Stratagene, La Jolla, CA). The fragment contained a single 75-bp exon that encoded amino acids 2851-2875 of the mouse utrophin protein and corresponded to exon 64 of the dystrophin gene (Fig. 1 a). The long arm of homology in the targeting vector was a 9.5-kb SphI genomic fragment directly upstream of the isolated exon. The short arm was a 1.1-kb BstBI–BamHI fragment that included the distal 25-bp of the exon. These fragments were inserted into cloning sites flanking a PGKneomycin resistance cassette (Tybulewicz et al., 1991). The vector included a thymidine kinase cassette distal to the short arm. The mutant gene therefore lacked the ∼2.7-kb SphI–BstBI fragment which included the first 52 bp of the isolated exon. The vector was linearized with NotI and transferred into R1 type embryonic stem (ES) cells (Nagy et al., 1993) by electroporation. Homologous recombinants were isolated by double selection using G418 and FIAU. Chimeras from two independently derived ES cells gave rise to heterozygous and homozygous mutants. The phenotypes of the two lines were indistinguishable and the results obtained with both lines are presented together.

For Southern analysis, DNA from adult liver was digested with NheI and NcoI and probed with a ³²P-labeled BstBI–BamHI fragment of the utrophin genomic clone (Fig. 1 b).

For reverse transcription-polymerase chain reaction (RT-PCR) analysis, total RNA was isolated from adult tissues and cDNA was made using AMV-RT (Promega, Madison, WI) and random primers. Four oligonucleotides, which flanked the mutated exon by 597, 197, 214, and 554 bp (primers b, a, a′, and b′, respectively, in Fig. 1 c), were used pairwise to amplify reverse transcribed RNA from utrn+/− and utrn+/− muscle. The resulting fragments were then sequenced.

For immunoblots, tissue was homogenized and sonicated in extraction buffer (phosphate-buffered saline [PBS], 1% SDS, 5 mM EDTA, and protease inhibitors). Aliquots of a low speed supernatant were analyzed for total protein concentration using a BCA protein assay (Pierce, Rockford, IL); 30 μg of total protein was loaded per lane on 6% SDS polyacrylamide gels.

For immunohistochemical analysis, tissue was frozen in liquid nitrogencooled isopentane and sectioned at 7 μm. Sections were stained with antibodies diluted in PBS with 1% BSA for 2–12 h, then rinsed with PBS. Sections were then incubated with a mixture of contrasting flurophores: fluorescein-conjugated goat anti–mouse IgG1 or goat anti–rabbit IgG and rhodamine-conjugated α-bungarotoxin (BTX) or goat anti–rabbit IgG. For thick sections, the sternomastoid was fixed in 1% paraformaldehyde in PBS, cryoprotected in sucrose, frozen, and sectioned en face at 40 μm.

Sources of antibodies were as follows: mouse monoclonal antibodies to utrophin, dystrophin, β-dystroglycan, and α-sarcoglycan (adhalin) were purchased from Novocastra Laboratories Ltd. (Newcastle upon Tyne). A rabbit polyclonal anti-utrophin antibody was made in our laboratory to a peptide corresponding to the final 10 COOH-terminal amino acids from murine utrophin. A rabbit antiserum to dystrobrevin was made to a fusion protein containing a 292–amino-acid fragment from murine dystrobrevin. Sources for antibodies to agrin, laminin-β2, laminin-1, synaptophysin, and rapsyn are described in previous publications from our laboratory (Sanes et al., 1990; Gautam et al., 1995; Gautam et al., 1996). Antibodies to MuSK (DeChiara et al., 1996) and β2-syntrophin (Peters et al., 1994) were generous gifts of G. Yancopoulos (Regeneron) and S. Froehner (University of North Carolina), respectively.

For ultrastructural studies, sternomastoid and intercostal muscles were fixed in 4% glutaraldehyde/4% paraformaldehyde in PBS, washed, refixed in 1% OsO4, dehydrated, and embedded in resin. Thin sections were stained with lead citrate and uranyl acetate. Sections were systematically scanned in the electron microscope, and all NMJs encountered were photographed at 20,000×. Lengths of nerve-muscle apposition were measured from the micrographs, and junctional folds that opened into the primary cleft were counted and expressed as folds per micron of apposition. Results did not differ significantly between sternomastoid and intercostals, so data have been pooled.

AChR density was assessed by fluorescence imaging, a method devised and described by Turney et al. (1996). Briefly, live mice were anesthetized and the sternomastoid was incubated with a saturating dose of rhodamineBTX for 30 min. The neuromuscular junctions were visualized using a fluorescence microscope and a video camera. Quantitative images were obtained by storing the ratio of junctional fluorescence to that of a fluorescent standard. These images were stored in a digital image processor. For analysis, 2–4 intensity profiles were taken from separate parts of each neuromuscular junction. Dedicated computer software then gave each profile a single value representing its relative illumination intensity. Two utrn+/− (70 intensity profiles from 24 endplates) and two utrn−/− mice (67 intensity profiles from 21 endplates) were studied.

For AChR quantitation by ¹²⁵I-BTX, the sternomastoid was dissected and placed in oxygenated buffer containing 50 nm ¹²⁵I-BTX (Amersham). Muscles were incubated for 2 h, washed extensively, and fixed with 4% paraformaldehyde/5% sucrose in PBS. The muscles were then teased into small muscle bundles, endplates were visualized with an anticholinesterase stain, and counted. The small bundles were then divided into two equal sections, one with and one without endplates, and counted in a gamma counter. Specific counts were the difference between the two.

Utrophin, like dystrophin, is comprised of four structural domains (Fig. 1,a; Tinsley et al., 1992; Blake et al., 1996): an amino-terminal segment that contains a putative actinbinding site, a long central rod, a highly conserved cysteine-rich region, and a COOH-terminal domain with homology to only two other known proteins, dystrobrevin (Wagner et al., 1993; Blake et al., 1996a) and DRP2 (Roberts et al., 1996). In addition, both utrophin and dystrophin exist in shorter forms, reflecting both alternative splicing and transcription from multiple promoters (Pearce et al., 1993; Ahn and Kunkel, 1993; Blake et al., 1995, 1996a). These shorter forms lack some or all of the NH₂-terminal, central rod, and COOH-terminal domains, but all known forms share the cysteine-rich region (Ahn and Kunkel, 1993; Blake et al., 1996a). This region is required for binding of dystrophin and, by implication, of utrophin to the DPC (Suzuki et al., 1994; Rafael et al., 1996); furthermore, deletion of this part of dystrophin leads to a severe dystrophic phenotype (Bies et al., 1992). To maximize our chance of disrupting the function of all known utrophin isoforms, we targeted an exon that maps to the beginning of utrophin's cysteine-rich region (Fig. 1,a). The mutation was transferred by homologous recombination to embryonic stem (ES) cells, which were then used to generate germline chimeras. Heterozygous (utrn+/−) mice appeared normal and homozygous (utrn−/−) mice were produced in expected numbers. Southern analysis confirmed disruption of the utrophin gene (Fig. 1 b).

The effects of the mutation on utrophin RNA were assessed by reverse transcription PCR (RT-PCR) (Fig. 1,c). We used primers that flanked the targeted 75-bp exon (corresponding to exon 64 of the dystrophin gene) by 197 and 214 bp. Reverse transcripted RNA from wild-type and mutant muscle was amplified by PCR using these primers. The expected 486-bp fragment was amplified from wild-type muscle, whereas a smaller (349 bp) fragment was amplified from mutant muscle. Sequencing of the PCR products revealed the mutant transcript to contain a 137-bp deletion encompassing not only the targeted exon but the preceding exon as well (corresponding to exon 63 of the dystrophin gene). These exons are located at the start of the cystein-rich region (Fig. 1 c). This deletion leads to a frameshift that introduces a stop codon 1.7-kb before the wildtype stop codon and theoretically results in a truncated utrophin protein missing both the cysteine-rich and the COOH-terminal domains. To seek transcripts with larger deletions, we repeated this analysis with primers that flanked the targeted exon by 597 and 554 bp. Only the 137bp-deleted product was recovered, suggesting that, at least in skeletal muscle, the mutated utrophin gene gives rise only to mRNAs that encode truncated proteins lacking the cysteine-rich and COOH-terminal domains.

Consistent with the RT-PCR analysis, immunoblotting with an antibody specific to the COOH terminus of utrophin failed to react with any protein in utrn−/− skeletal muscle (Fig. 1,d). However, an NH₂-terminal specific antibody also failed to detect protein, indicating that truncated utrophin is either not produced or is unstable. Similar results were obtained in adult brain and lung and in neonatal muscle (Fig. 1 d and data not shown). Thus, the utrn−/− mutant we have generated is likely a severe hypomorph for all forms of utrophin.

Despite the absence of utrophin, utrn−/− mice developed normally, were fertile and have now lived for 10 mo without any obvious pathology. To evaluate sublethal consequences of utrophin-deficiency, we examined several tissues that normally express utrophin: brain, heart, lung, kidney, and skeletal muscle (Love et al., 1989). We first assessed the cellular distribution of utrophin in normal adult tissue. In the brain, utrophin was present in the microvasculature (Fig. 2,a), the choroid plexus and the pia mater, as previously noted by others (Khurana et al., 1992; Uchino et al., 1994). This pattern suggests that utrophin may be a part of the blood-brain barrier. In cardiac muscle, we found utrophin associated with both intercalated discs and sarcolemma (Fig. 2,c); in contrast, a previous report found utrophin only at intercalated discs (Pons et al., 1994a). In lung, utrophin localized to the membrane of most if not all the cells of the alveoli (Fig. 2,e) but was not detectable in bronchiolar or pulmonary arterial endothelium. In kidney, utrophin was localized to the cells adjacent to the glomerular basement membrane and to the basolateral membrane of a subset of tubules (Fig. 2,g). In skeletal muscle, utrophin was confined to the NMJ (Fig. 3, a and c), as detailed below.

In all tissues studied, utrophin-rich membranes abutted basal laminae, as revealed by counterstaining with antibodies to the ubiquitous basal lamina component, laminin (Fig. 2, a′, c′, e′, and g′ and data not shown). Based on this association, we used laminin counterstaining to seek residual utrophin expression in utrn−/− mice and to look for any histological abnormalities in cells that normally express utrophin. No utrophin immunoreactivity was detectable in any of the utrn−/− tissues studied (Fig. 2, b, d, f, and h, and 3, b and d). Moreover, the arrangement of basal laminae and, by implication, of the cells that abut them, were normal in all utrn−/− tissue (Fig. 2, b′, d′, f′, and h′ and data not shown). In support of this conclusion, no structural abnormalities were detected in hematoxylineosin stained sections (data not shown). Similarly, utrophin was undetectable and the tissues histologically normal in heart, brain, lung, and skeletal muscles from neonatal (P1) utrn−/− mice (data not shown).

Limited investigations have failed so far to reveal any significant functional consequences of utrophin-deficiency: (1) Serial observations detected no behavioral, motor or cardiorespiratory defects. (2) Urinalysis showed no significant proteinuria in the mutant animals. (3) Mutant mice had no evidence of a disrupted blood-brain barrier as shown by the exclusion of Evans Blue from the brain parenchyma (Chiueh et al., 1978). (4) Utrophin-deficient mothers bore normal-sized litters and reared their offspring appropriately. (5) Mutant neonates remained viable and active even when separated from their mothers for ⩾4 h.

Given the selective association of utrophin with the NMJ (Fig 3, a and c) and its putative role in synaptogenesis (see Discussion), we suspected that subtle defects might be present and therefore analyzed the NMJ in detail. Labeling the AChR with rhodamine-BTX (Fig. 4, a and b) and the nerve terminals with antibodies to synaptic vesicle proteins (not shown) revealed the utrn−/− synapse to be of normal size and geometry (Fig. 4, a and b). However, the density of AChR was ∼30% lower in utrn−/− than in utrn+/− muscle as assessed by fluorescence-imaging following incubation with rhodamine-BTX (Turney et al., 1996) (utrn−/−: 39.5 ± 0.9, n = 67; utrn+/−: 55.5 ± 1.6, n = 70; illumination intensity ± standard error of the mean, n = number of synaptic images analyzed; P < .0001 by Mann-Whitney). This difference appeared to reflect a moderate decrease in AChR density at most synapses rather than a drastic decrease in density at a specific subset of synapses (Fig. 4 c). Preliminary measurements of a total number of AChR using ¹²⁵I-BTX gave results suggesting a similar reduction (data not shown). The outwardly normal motor behavior of the utrn−/− mice despite the significant decrease in AChR is consistent with the known safety margin in the number of AChR required for normal muscle function (Patton and Waud, 1967).

Further abnormalities were discovered when muscles were analyzed by electron microscopy. In normal muscle, junctional folds invaginate the postsynaptic membrane; AChR and utrophin are concentrated at the tops of these folds (Sealock et al., 1991; Bewick et al., 1992). In utrn−/− sternomastoid and intercostal muscles, the number of folds was reduced by ∼50% compared to utrn+/− and utrn+/+ synapses (Fig. 5). This difference was apparent by P11 and persisted into adulthood. Despite the decrease in junctional folds, the mutant synapse displayed normal apposition of Schwann cell to nerve and of nerve to muscle as well as normally differentiated nerve terminals (Fig. 5, a–d). Thus, although utrophin is clearly not essential for neuromuscular synaptogenesis, the utrn−/− mice show morphological as well as molecular alterations in their postsynaptic apparatus.

In skeletal muscle, dystrophin and the dystrophin-associated protein (DPC) join the actin cytoskeleton to the extracellular matrix protein laminin-2 providing structural support to the sarcolemma (Ervasti and Campbell, 1993; Matsumura and Campbell, 1994). The DPC is found at the synapse as well, but in an altered form. Intracellularly, dystrophin and α1-syntrophin are present throughout the sarcolemma, whereas utrophin and β2-syntrophin are exclusively synaptic (Ohlendieck et al., 1991; thiMan et al., 1991; Khurana et al., 1991; Peters et al., 1994). Extracellularly, the β1 chain of the laminin heterotrimer is replaced by laminin-β2 in synaptic basal lamina (Sanes et al., 1990). In addition, agrin, a protein responsible for AChR clustering (Gautam et al., 1996), is concentrated in the synaptic cleft and binds to the extracellular portion of the DPC (Campanelli et al., 1994; Gee et al., 1994; Sugiyama et al., 1994; Bowe et al., 1994). Finally, studies with recombinant proteins have suggested that the synaptic DPC associates with the AChR-associated protein, rapsyn (Apel et al., 1995), which in turn associates with MuSK (Gillespie et al., 1996), a tyrosine kinase receptor thought to mediate the effects of agrin (DeChiara et al., 1996).

The utrophin-deficient mice allowed us to ask whether utrophin is needed for the DPC to acquire its specialized synaptic characteristics. We immunostained for the synapse-specific proteins rapsyn, β2-syntrophin, laminin β2, agrin, and MuSK. All of these antigens were concentrated at synaptic sites in both utrn+/− and utrn−/− mice (Fig. 3, k–n, and data not shown). In addition, we assayed the more broadly distributed DPC proteins β-dystroglycan, dystrobrevin, α-sarcoglycan, and dystrophin. These proteins were also normally distributed in utrn−/− skeletal muscle (Fig. 3, e–j, and data not shown). Thus, the synaptic DPC can exist and maintain its specialized character in the absence of utrophin.

One possible explanation for the maintenance of postsynaptic specialization in the utrn−/− mouse is that a synaptic concentration of dystrophin substitutes for utrophin. Indeed, utrophin and dystrophin are both concentrated at the adult synapse (Fig. 3, a, c, and e), although their precise distribution differs: utrophin colocalizes with AChRs at the crests of the junctional folds whereas dystrophin is primarily at the base of the folds (Byers et al., 1991; Sealock et al., 1991; Bewick et al., 1992). In newborn (P1) animals, utrophin was already enriched at the endplate, although some extrasynaptic staining was evident as well (Fig. 6,a). Dystrophin, on the other hand, was not yet enriched at the endplate at birth (Fig. 6,c), presumably reflecting the postnatal formation of junctional folds which continues for several weeks after birth. If dystrophin compensated for utrophin, then one might expect dystrophin to become enriched at the neonatal mutant synapse. In utrn−/− neonates, however, dystrophin remained evenly distributed throughout the sarcolemma with no synaptic enrichment (Fig. 6,d). Nonetheless, rapsyn and AChR remained concentrated at the neonatal utrn−/− synapse (Fig. 6, e, e′, f, and f′  ). Thus, the localization of the AChRrapsyn complex to the postsynaptic membrane can occur when neither utrophin nor dystrophin is synaptically concentrated.

Functional compensation by dystrophin may help explain the normal structure of tissues other than skeletal muscle in utrn−/− mice. We examined this hypothesis by assessing the distribution of dystrophin in several tissues that normally express utrophin. In normal heart and brain, dystrophin and utrophin had overlapping distributions; both were localized to the sarcolemma and intercalated discs in cardiac myocytes and to endothelial cells in the cerebral vasculature (Fig. 2,c, 7, a and c and data not shown). The distribution of dystrophin in heart and brain of utrn−/− mutants did not differ detectably from that in controls (Fig. 7,d, and data not shown). Moreover, the DPC component β-dystroglycan was distributed along the sarcolemma at similar levels in utrn+/− and utrn−/− cardiac myocytes (Fig. 7, e and f). Thus, in these tissues, dystrophin is appropriately placed to compensate for the loss of utrophin.

In contrast, we detected no dystrophin in lung or kidney of either utrn+/− or utrn−/− mice (Fig. 7, i and j, and data not shown). The antibody we used was specific for the COOH terminus of dystrophin, and therefore should have detected short forms as well as full-length dystrophin. Low levels of dystrophin have been detected in these tissues by immunoblot analysis (e.g., Hoffman et al., 1988); these levels may have been too low to be detected by our immunohistochemical assay. Nonetheless, compensation by upregulation of dystrophin is unlikely in these tissues. In light of this result, we used the lung to determine whether components of the DPC persist in the apparent absence of both utrophin and dystrophin. In dystrophic muscle, loss of dystrophin leads to a marked decrease in the levels of DPC components in the sarcolemma (Matsumura et al., 1992). In lung, however, one component of the DPC, β-dystroglycan, was present in similar levels in utrn+/− and utrn−/− mice (Fig. 7, k and l). This result implies the existence of a utrophin- and dystrophin-independent mechanism for retention of the DPC.

Mice lacking a functional utrophin gene are viable and fertile, but have subtle defects in the postsynaptic apparatus of their skeletal neuromuscular junctions. In an accompanying paper, Deconinck et al. (1997) reported similar results. The allele described here removes the COOH-terminal cysteine-rich region that is shared by both forms of utrophin (Blake et al., 1995), and is likely, by analogy to dystrophin, to be important for its function (Bies et al., 1992; Suzuki et al., 1994; Rafael et al., 1996). In fact, no utrophin at all was detectable in our mutant, either because insertion of the neomycin cassette led to reduced levels of mRNA or because the mutant protein was unstable. We cannot rule out, however, the possibility that truncated utrophin is present in some tissues or at some stages of development. Deconinck et al. (1997) deleted an NH₂terminal exon, leading to complete loss of full-length utrophin. They, however, cannot exclude the possibility that shorter forms of utrophin, transcribed from a promoter COOH-terminal to their deletion (Blake et al., 1995), are present in the mutant. Thus, the similarity of the phenotype reported here to that reported by Deconinck et al. (1977) provides a strong argument that both alleles are effectively nulls.

Four sets of studies had suggested that utrophin might be critical for neuromuscular synaptogenesis and particularly for the differentiation of the postsynaptic membrane. First, agrin, a critical organizer of postsynaptic differentiation (Gautam et al., 1996), binds to dystroglycan, a component of the DPC (Campanelli et al., 1994; Gee et al., 1994; Sugiyama et al., 1994; Bowe et al., 1994). Dystroglycan, in turn, appears to associate with utrophin at the NMJ and with dystrophin extrasynaptically (Matsumura et al., 1992; Ervasti and Campbell, 1993). It seemed possible, therefore, that utrophin converted synaptic dystroglycan into a functional agrin receptor. Second, in cultured muscle cells, large but not small AChR clusters are associated with utrophin, suggesting that utrophin is important for the growth or stabilization of high density AChR aggregates (Phillips et al., 1993; Campanelli et al., 1994). Third, mice incapable of forming postsynaptic AChR clusters through targeted mutagenesis of rapsyn (Gautam et al., 1995), MuSK (DeChiara et al., 1996), or agrin (Gautam et al., 1996), lack synaptic accumulations of utrophin. Finally, forced expression of the putative dystroglycan binding domain of utrophin in cultured myotubes leads to fewer AChR clusters in response to agrin (Namba and Scheller, 1997). This presumptive dominant negative effect suggested that interfering with the utrophin-dystroglycan association attenuates the agrin-mediated AChR cluster transduction pathway.

The modest reduction in AChR density that we and Deconinck et al. (1977) find in the utrn−/− mice provides limited support for the involvement of utrophin and the DPC in postsynaptic differentiation. The nature of this involvement, however, remains unclear. One possibility is that utrophin-DPC dependent processes are required for complete AChR clustering but that other pathways play a dominant role in transmitting agrin's signals. For example, there is now evidence that a synaptically localized tyrosine kinase, MuSK, is part of or associated with a functional agrin receptor (DeChiara et al., 1996; Glass et al., 1996). Moreover, agrin fragments that are incapable of binding dystroglycan retain their AChR clustering activity in vitro (Sugiyama et al., 1994; Gesemann et al., 1995, 1996; Campanelli et al., 1996).

A second possibility is that utrophin and the synaptic DPC might actually be crucial for postsynaptic differentiation, but that dystrophin substitutes, albeit imperfectly, for utrophin in the utrn−/− mice. In this context, it is noteworthy that mutation of the dystroglycan gene leads to a far more severe phenotype than mutation of either the utrophin or the dystrophin gene (cited in Henry and Campbell, 1996). Moreover, a reduction in the density of junctional folds is also seen at NMJs of dystrophin-deficient (mdx) mice. In this case, however, the synaptic alterations are thought to result from muscle fiber necrosis and regeneration, rather than from the absence of dystrophin per se (Torres and Duchen, 1987; Lyons and Slater, 1991).

A third possibility is that the decreased density of AChRs in utrn−/− mice may result from the decreased number of junctional folds. In normal skeletal muscle, AChRs are concentrated not only at the tops of folds but also partway down their sides (Fertuck and Salpeter, 1976). A loss of folds would, therefore, lead to a decrease in the total AChR-rich area within the NMJ. In this scenario, the decreased AChR density in utrn−/− muscle would not result from any functionally important interaction of utrophin or the DPC with AChRs. Instead, the direct effect of utrophin in muscle might be to promote the initial invagination of the postsynaptic membrane that leads to the generation of folds.

In contrast to the neuromuscular phenotype, we detected no abnormalities in other tissues that express utrophin. The early and widespread expression of utrophin led to the speculation that utrophin-deficiency might lead to embryonic lethality (Blake et al., 1996a). This hypothesis, which we have now disproven, provided an attractive explanation for the fact that no mutations of the human utrophin gene have yet been reported despite its extremely large size (Pearce et al., 1993). It is premature, however, to conclude that utrophin is unimportant for the function of nonskeletal tissues. In fact, the histological analysis so far applied to nonskeletal tissue also failed to detect defects in muscle. A more detailed analysis will be required to determine whether utrophin plays a role in such structures as the glomerular filter or the blood-brain barrier.

Utrophin levels are increased in dystrophin-deficient animals, suggesting that it may functionally compensate, in part, for the missing dystrophin (Ahn and Kunkel, 1993; Matsumura and Campbell, 1994; Blake et al., 1996a). We asked whether the reverse was true in utrn−/− mice. We obtained, however, no evidence that dystrophin was upregulated in utrophin-deficient tissue. Indeed, lung and kidney appeared histologically and functionally normal in the absence of detectable utrophin or dystrophin. Moreover, the NMJ can maintain its specialized character even when neither utrophin nor dystrophin is concentrated at synaptic sites. We still cannot exclude the possibility that an undetected isoform of dystrophin, or a dystrophin homologue such as dystrobrevin (Wagner et al., 1993; Blake et al., 1996b) or DRP2 (Roberts et al., 1996), might be compensating for the loss of utrophin. Fortunately, it will be possible to test this hypothesis since dystrophin-deficient (mdx) mice are available, and we have now generated dystrobrevin-deficient mice which are viable and fertile (Grady, R.M., Merlie, J.P., and Sanes, J.R., manuscript in preparation). We are now breeding these animals to generate doubly mutant mice deficient in utrophin and either dystrophin or dystrobrevin.

We thank M. Nichol and J. Mudd for a generation of chimeras; M. Elam, B. Klocke, and other members of the Merlie and Sanes laboratories for assistance; A. Nage for R1-Es cells; and A. Guo, S. Froehner, and G. Yancopoulos for gifts of antisera. We are especially grateful to M. Nichol for her management of the animal facility; to J. Cunningham for the electron microscopy; and to S. Culican for the fluorescence-imaging of AChR. We thank A. Vincent, K. Davies, and C. Slater for communicating results before publication, including their initial discovery that the density of AChR is decreased at utrn−/− endplates.

This work was supported by a National Research Service Award grant (R.M. Grady), and grants from the MDA of America and the National Institutes of Health (J.R. Sanes and J.P. Merlie).

AChR

acetylcholine receptors

DPC

dystrophin-associated protein complex

ES

embryonic stem

NMJ

neuromuscular junction

RT-PCR

reverse transcription-polymerase chain reaction