Essential roles of the acetylcholine receptor γ-subunit in neuromuscular synaptic patterning

During development, focal sites of pre- and postsynaptic cells undergo reciprocal interactions that ultimately lead to the establishment of functional synaptic connections. In developing mammalian muscles,neuromuscular synapses are generated within a narrow end-plate band that is often localized to the central region of the muscle fibers(Burden, 2002; Sanes and Lichtman, 2001). How is such a synaptic pattern established during development? Why does neuromuscular synaptogenesis take place at select sites of the muscle? It has long been thought that ingrowing nerves determine the site of synaptogenesis. For example, in vitro studies of nerve-muscle co-culture have demonstrated that innervation leads to accumulation of postsynaptic AChRs at sites of nerve-muscle contact, independent of muscle activity(Anderson et al., 1977), or the presence of pre-existing AChR clusters(Anderson and Cohen, 1977; Frank and Fischbach, 1979). Further studies have led to the discovery of agrin, a heparan sulfate proteoglycan, a potent nerve-derived factor that induces AChR clustering(McMahan, 1990; Nitkin et al., 1987; Rupp et al., 1991). Agrin activates the MuSK complex (Glass et al.,1996) and promotes neuromuscular synaptogenesis (reviewed by Bezakova and Ruegg, 2003; Bowe and Fallon, 1995). Genetic deletion of agrin (Gautam et al.,1996) or MuSK (DeChiara et al., 1996) leads to abnormal pre- and post-synaptic development at the NMJ. Agrin may also play a role in stabilizing AChRs by counteracting the dispersion effect of neurotransmitter ACh(Lin et al., 2005; Misgeld et al., 2005).

Several lines of evidence, however, challenge this long-held `neurocentric view' and suggest that post-synaptic muscle cells independently prepare pre-patterned sites at the central region of the muscle (reviewed by Arber et al., 2002; Ferns and Carbonetto, 2001; Goda and Davis, 2003). Indeed,AChR clusters are detected in aneural muscles in vitro(Bekoff and Betz, 1976), or in vivo when motor innervation was prevented either by neurotoxin injection(Braithwaite and Harris, 1979)or neuroectomy (Creazzo and Sohal,1983). However, these manipulations did not exclude the possibility that some muscle fibers were transiently contacted by motor axons,which could provide neural signals to induce postsynaptic differentiation.

Burden and his colleagues have provided the first genetic evidence for the presence of pre-patterned AChR clusters in developing muscles(Yang et al., 2000). Subsequent studies have confirmed this observation, and further demonstrated that the formation of pre-patterned AChR clusters is independent of agrin, but requires MuSK (Lin et al.,2001; Yang et al.,2001). Furthermore, ectopic expression of MuSK in muscles induces formation of ectopic synapses independently of agrin(Kim and Burden, 2008). In addition, studies of developing zebrafish NMJs in vivo elegantly demonstrate that pre-patterned AChR clusters are assembled prior to the arrival of the nerves; these AChRs are not only incorporated into the developing neuromuscular synapses but also map to the track of motor axonal growth(Flanagan-Steet et al., 2005; Panzer et al., 2005; Panzer et al., 2006). Together, these studies suggest that pre-patterned postsynaptic sites marked by nascent AChR clusters may serve as a primary determinant for subsequent innervation pattern. It is therefore essential to determine how pre-patterned AChR clusters are initiated and to what extent these clusters contribute to pre- and postsynaptic differentiation of the NMJ.

AChRs of the embryonic vertebrate muscles are pentamers composed of five membrane-spanning subunits in a stoichiometry of two α-, one β-,one δ- and one γ-subunit (γ-AChR,α ₂βδγ). The γ-subunit distinguishes the embryonic from the adult AChR (ϵ-AChR,α ₂βδϵ). Normally, the γ-subunit is expressed during embryonic and neonatal stages, and is replaced by theϵ-subunit during the first 2 weeks after birth (γ/ϵ switch)(Changeux et al., 1992; Mishina et al., 1986). The physiological role of this switch is not well understood, but evidence suggests that the γ-subunit and the ϵ-subunit each play specific roles in embryonic and adult NMJs, respectively. Deletion of theϵ-subunit gene in mice (ϵ^-/- mice) leads to progressive neuromuscular weakness and death 2-3 months postnatally(Missias et al., 1997; Witzemann et al., 1996). Expressing human AChR γ-subunit in ϵ^-/- mice rescues the lethality of the ϵ-subunit gene deletion, but the rescued mice continue to display AChR deficiency and develop phenotypes similar to human congenital myasthenic syndromes (Cossins et al.,2004). Furthermore, mutations in human AChR γ-subunit gene(CHRNG - Human Gene Nomenclature Database) cause severe prenatal myasthenia (Escobar or multiple pterygium syndrome)(Hoffmann et al., 2006; Morgan et al., 2006).

Deletion of the mouse γ-subunit (γ^-/- mice) leads to perinatal lethality (Takahashi et al.,2002). Genetically replacing the γ-AChRs with ϵ-AChRs in mice (γ^ϵ chimeric mice) preserves normal end-plate formation but alters the patterning of the motor nerves at postnatal stages(Koenen et al., 2005). Although these results suggest that the γ-subunit is required for normal development and is essential to survival, the contribution of theγ-subunit to neuromuscular synaptogenesis is poorly understood. This lack of understanding arises, in part, from the focus of previous studies ofγ ^-/-, or γ^ϵ chimeric mice, which were limited to neonatal tissues (Koenen et al., 2005; Takahashi et al.,2002).

In the present study, we focused on the initial stages of neuromuscular synaptogenesis in γ^-/- embryos. Specifically, we examined the timing and magnitude of AChR cluster formation and the patterning of motor nerves during development. Our results showed that the pre-patterned AChR clusters were completely absent during the initial stages of neuromuscular synaptogenesis in γ^-/- embryos. Despite the absence of AChR clustering during these initial stages, MuSK, but not rapsyn, was clustered inγ ^-/- muscles. Furthermore, we observed profound changes in patterning of the motor innervation. That is, in contrast to the wild type in which presynaptic nerves innervate a narrow region of the muscle, presynaptic nerves in the γ^-/- embryos were highly branched over a broad region of the muscle. In addition, the numbers of spinal motoneurons were markedly increased in the γ^-/- embryos. Overall, these results demonstrate that the γ-subunit is required for the assembly of pre-patterned AChR clusters, which in turn play an essential role in determining the subsequent pattern of neuromuscular synaptogenesis.

Takahashi et al. (Takahashi et al.,2002) have described the generation of mutant mice deficient in the gene encoding the AChR γ-subunit (Chrng - Mouse Genome Informatics); homozygote (γ^-/-) mice are perinatal lethal. We obtained heterozygote mice (γ^+/-) from the RIKEN BioResource Center (Tsukuba, Japan). To generate homozygous γ^-/- embryos,heterozygotes were time-mated, and the day when a vaginal plug first appeared was designated as embryonic (E) day 0.5. After selected intervals of development, embryos were collected by Cesarean section of anesthetized pregnant mice. On average, 3-12 γ^-/- embryos and a matching number of their wild-type littermate controls (+/+) were analyzed at each stage. All experimental protocols followed NIH Guidelines and were approved by the UT Southwestern Institutional Animal Care and Use Committee.

Muscle samples were fixed in 2% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.3) at 4°C overnight prior to processing as wholemounts or frozen sections. The samples were blocked in dilution buffer (500 mM NaCl,0.01 M phosphate buffer, 3% BSA and 0.01% thimerosal), and then incubated with primary antibody-neurofilament 150 (Chemicon, Temecula, CA), rapsyn (Affinity Bioreagents, Golden, CO), synaptophysin (Dako, Carpinteria, CA), synaptic vesicle protein 2 (SV2) (Developmental Studies Hybridoma Bank, University of Iowa, Iowa City, IA), MuSK (41101K) (Bowen et al., 1998), synaptotagmin 2 or syntaxin (gifts from Dr Thomas Sudhof, UT Southwestern Medical Center, Dallas, TX). Muscles were then incubated with fluorescein isothiocyanate-conjugated secondary antibodies and Texas Red-conjugated α-bungarotoxin (α-bgt) (10^-8 M,Molecular Probes). Samples were then washed with PBS and mounted in 90/10 glycerol/PBS containing 1% N-propyl gallate. Fluorescent images were acquired using a Hamamatsu ORCA-285 camera or a Zeiss LSM 510 Meta confocal microscope. Quantification of fluorescence intensity and sizes of AChR clusters was made from confocal images acquired with identical, sub-saturating gains. The mean gray value (integrated density/total pixels), area, perimeter and Feret's diameter (the length of the greatest axis) were measured using NIH ImageJ.

Detection of AChE was based on the methods previously described(Enomoto et al., 1998). Briefly, diaphragm muscles were fixed with 2% PFA, rinsed in PBS and incubated in 0.2 mM ethopropazine, 4 mM acetylthiocholine iodine, 10 mM glycine, 2 mM cupric sulfate and 65 mM sodium acetate solution at pH 5.5, for 2-4 hours at 37°C. Staining for AChE was developed by incubating the whole-mount diaphragm for 2-5 minutes in sodium sulfide (1.25%, pH 6.0), followed by extensive washing. The diaphragms were then cleared with 50% glycerol in PBS and flat mounted onto a glass slide.

Total RNAs were isolated from de-skinned and eviscerated whole embryos at various stages using the TRI reagent (Molecular Research Center, Cincinnati,OH). The first-strand cDNA was synthesized by StrataScript reverse transcriptase (Stratagene, La Jolla, CA). The housekeeping gene glyceraldehyde phosphate dehydrogenase (GAPDH) served as an internal control. The following primers were used for PCR amplification of specific gene products: (1) AChRα-subunit primers, forward AAG CTA CTG TGA GAT CAT CGT CAC, reverse TGA CGA AGT GGT AGG TGA TGT CCA (product size: 244 bp)(Gattenlohner et al., 2002);(2) AChR ϵ-subunit primers, forward GGC AGT TTG GAG TGG CCT ACG ACA,reverse GCA GGA CGT TGA TAG AGA CCG TGC (product size: 489 bp)(Yumoto et al., 2005); GAPDH primers, forward TCA ACG GCA CAG TCA AGG CCG AGA, reverse ATG ACC TTG CCC ACA GCC TTG GCA GC (product size: 494 bp)(Cossins et al., 2004). Thirty-five cycles of PCR were performed for detection of the AChRϵ-subunit expression.

Specific TaqMan probes were used for quantitative real-time PCR [Applied Biosystems (ABI), Foster City, CA]: AChR ϵ-subunit (assay ID,Mm00437411_m1) and phosphoglycerate kinase 1 (PGK1) (assay ID, Mm00435617_m1,for internal control), according to the ABI manuals. Briefly, total RNAs were isolated from forelimb muscles of three pairs of E18.5 wild-type andγ ^-/- embryos, and from wild-type adult muscles, which later served to calibrate the assay. The first-strand cDNA were synthesized using SuperScript III cDNA synthesis kit (Invitrogen Corporation, Carlsbad, CA). Each sample was assayed in triplicate reactions using TaqMan Universal PCR master mix. The amplification difference between AChR ϵ-subunit and PGK1 in each sample was calculated and then normalized to that of adult muscle using ΔΔCt (comparative Ct) method (Applied Biosystems, Foster City, CA).

Quantification of motoneurons were carried out based on previously published methods (Buss et al.,2006; Clarke and Oppenheim,1995). E18.5 embryos were fixed with 4% PFA, equilibrated with 30%sucrose, transversely sectioned (12 μm) and stained with Cresyl Violet. Motoneurons were identified by their characteristic size and shape, and their anatomical location in the ventral horn(Clarke and Oppenheim, 1995). Motoneurons in cervical segments (C3-C8) were counted, blind to genotype, in every ninth section, and total number of motoneurons within these segments was combined and then multiplied by 9 to generate total estimates of motoneurons(Clarke and Oppenheim,1995).

Phrenic nerve-diaphragm muscle was dissected in normal rodent Ringer's solution (136.8 mM NaCl, 5 mM KCl, 12 mM NaHCO₃, 1 mM NaH₂PO₄, 1 mM MgCl₂, 2 mM CaCl₂and 11 mM d-glucose; pH 7.3) (Liley,1956) and pinned to Sylgard coated dishes. End-plate potentials(EPPs) were evoked by supra threshold stimulation (2 V, 0.1 ms) of the nerve via a suction electrode connected to an extracellular stimulator (SD9,Grass-Telefactor, West Warwick, RI). To compare EPP amplitudes recorded from muscle fibers with different resting potentials, we normalized the EPP amplitude as described (Talbot et al.,2003), except we set the correction factor to 0.8 for mouse muscles (McLachlan and Martin,1981), the reversal potential to 0 mV for the AChR gated conductance, and the common resting potential to -50 mV to reflect the low resting potentials frequently observed in embryonic muscles(Bennett and Pettigrew, 1974). EPPs that could not be separated from contaminating muscle action potentials were excluded from the quantification.

To determine the role of AChR γ-subunit in neuromuscular synaptogenesis, we followed the γ^-/- embryos during development from E13 to E18.5. These γ^-/- embryos were invariably smaller than their wild-type littermates, and displayed characteristic hunchback and wrist drop phenotypes, similar to those observed in the choline acetyltransferase (ChAT)-null mutants(Brandon et al., 2003; Misgeld et al., 2002). These external phenotypes appeared as early as E15.5, although they became more apparent at later stages (E16.5-E18.5). Mutant embryos did not show any spontaneous motor activity, nor did they respond to a mild pinch of the tail or skin.

We first asked whether the γ-subunit is required for clustering of AChRs during the initial stages of neuromuscular synaptogenesis. To address this issue, we labeled whole-mount diaphragm muscles (E13-E15.5) with Texas Red-conjugated α-bgt. In wild-type diaphragm muscles, AChR clusters were first seen at E13 (Fig. 1A) -at this stage, they appeared as tiny and dim speckles concentrated along the central region of the muscle (arrowheads in Fig. 1A). As development proceeded, AChR clusters in wild-type muscles increased in size and fluorescence-labeling intensity (E14.5, Fig. 1B; E15.5, Fig. 1C). By contrast, AChR clusters were completely absent from the γ^-/- diaphragm muscles; instead, diffused fluorescence was observed across the muscle surface(Fig. 1F-H), although the diffused fluorescence did appeared slightly more intense along the central regions, as illustrated by a line-scan analysis (see Fig. S1 in the supplementary material).

As different muscles vary in their onset of AChR clustering(Pun et al., 2002), we further examined a broad range of muscles that represent various segments of the anterior-posterior axis, including sternomastoid, triangularis sterni,intercostals, extensor digitorum longus (EDL) and soleus (Sol) muscles. Consistent with the diaphragm muscles, AChR clusters were absent from the other γ^-/- muscles during E13-E15.5(Fig. 1I,J and data not shown). These results demonstrated that the γ-subunit was required for the formation of pre-patterned AChR clusters.

Interestingly, AChR clusters appeared on the surface ofγ ^-/- muscles at E16.5(Fig. 2D), and remained throughout the subsequent stages (Fig. 2E,F). However, AChR clusters in γ^-/- muscles were distributed across a much broader region, compared with those observed in wild-type muscles. For example, within the ventral quadrant of the diaphragm muscle, the average width of AChR cluster band in γ^-/- was about twice that observed for wild types: 161.8±8.2 μm(n=3, number of embryos) in E16.5 γ^-/- muscles,versus 77.7±15.6 μm (n=3) in E16.5 wild-type muscles;152.6±21.2 μm (n=3) in E18.5 γ^-/- muscles,versus 81.7±5.0 μm (n=3) in E18.5 wild-type muscles.

Furthermore, the individual AChR clusters in the γ^-/-muscles were less intensely labeled by α-bgt and appeared larger in size than those in the wild-type muscles. For example, the mean gray value of the fluorescence intensity of individual AChR clusters in E18.5γ ^-/- muscles was 68±8 (n=176 clusters, n=3 embryos), only about 55% of the mean gray value measured in wild-type muscles (116±22, n=213, n=3). However, the average area, perimeter and Feret's diameter of AChR clusters inγ ^-/- muscles were 108±17 μm²,41±3 μm and 15±1 μm (n=176, n=3),respectively; by contrast, the same parameters measured in the wild-type muscles were 67±5 μm², 33±2 μm and 12±1μm (n=213, n=3), respectively. In other words, the average area, perimeter and Feret's diameter of AChR clusters inγ ^-/- muscles were about 160±25%, 127±10% and 130±8%, respectively, of those observed in the wild-type muscles(Fig. 2G).

The delayed emergence of AChR clusters in the γ^-/- embryos raises the following possibilities. First, in the absence of theγ-subunit, various combinations of the other AChR subunits, such asαβ, αβδ orα ₂βδ₂, may assemble into functional receptors, although with reduced ligand-binding affinity(Kurosaki et al., 1987; Liu and Brehm, 1993; Sine and Claudio, 1991). Alternatively, the delayed appearance of AChR clusters may be due to expression of the AChR ϵ-subunit in embryonic muscles(Yumoto et al., 2005). Therefore, we performed RT-PCR to determine the expression of theϵ-subunit. In parallel with ϵ-subunit expression, we also examined the expression of the AChR α-subunit and a `housekeeping gene',glyceraldehyde phosphate dehydrogenase (GAPDH), as controls. Although the expression of the α-subunit and GAPDH was readily detectable in both wild-type and γ^-/- muscles from E13.5 to E18.5, noϵ-subunit expression was detected prior to E16.5. A low level of theϵ-subunit expression first became detectable at E16.5 in both wild-type and γ^-/- muscles (asterisk in Fig. 3), and the level was sharply increased at E18.5 in both wild-type and γ^-/- muscles(^** in Fig. 3). To further determine the level of ϵ-subunit expression, we carried out quantitative real-time PCR analysis. We detected a similar expression level of the ϵ-subunit gene in both wild-type and γ^-/- muscles,which was ∼3% of the expression observed for wild-type adult muscle(Fig. 3B). These results demonstrated that the emergence of AChR clusters at later stages in theγ ^-/- muscles was probably due to the expression of theϵ-subunit.

To determine whether the ϵ-subunit was assembled into functional AChRs during embryonic stages, we measured EPPs in response to distinct toxins that specifically recognize the γ-AChRs [αA-conotoxin OIVB orαA-OIVB (Teichert et al.,2005)] and ϵ-AChRs [waglerin 1(McArdle et al., 1999)]. As shown in Fig. 4, in normal Ringer's solution, the EPP amplitude in the γ^-/- muscle were∼37% smaller than that in the wild-type muscle; the average EPP amplitude was 14.2±0.8 mV (n=35, number of muscle fibers; n=8,number of embryos) in the wild-type and 8.9±0.5 mV (n=37, n=8) in the γ^-/- muscles. The reduced EPP amplitude is consistent with the reduced fluorescence of the endplates labeled byα-bungarotoxin in the γ^-/- muscles(Fig. 2). Application of theγ-AChR blocker αA-OIVB (10 μM) resulted in ∼79% reduction of EPP amplitude in wild-type muscles: the average EPP amplitude declined to 3.0±0.4 mV (n=22, n=5) from the pre-toxin control value of 14.1±1.3 mV (n=16, n=5). By contrast,application of αA-OIVB had little effect on EPP amplitude in theγ ^-/- muscles; that is, mean EPP amplitude was 8.7±1.4 mV (n=17, n=4) and 8.0±1.1 mV (n=18, n=4) before and after treatment, respectively(Fig. 4A,B). Thus,γ ^-/- muscles were insensitive to the γ-AChR blocker.

By contrast, application of the ϵ-AChR blocker, waglerin 1 (1 μM),completely blocked the EPPs in the γ^-/- muscles, but had much less effect on the wild-type muscles. That is, in the presence of waglerin 1(1 μM), the average EPP amplitude was reduced to zero (n=24, n=5) in the γ^-/- muscles, compared with the pre-waglerin 1 value of 8.9±0.6 mV (n=20, n=4)(P<0.001). Similar exposure of the wild-type muscle to waglerin 1 reduced the average EPP amplitude to 10.7±1.3 mV (n=13, n=3) from the pre-waglerin 1 value of 14.2±0.4 mV(n=19, n=3) (Fig. 4C,D). These electrophysiological results demonstrate thatϵ-subunits were assembled into functional AChRs in both wild-type andγ ^-/- muscles at E18.5 and that the postsynaptic response at the NMJ of the γ^-/- muscles was mediated by ϵ-AChRs. By contrast, the postsynaptic response in E18.5 wild-type muscles was mediated by both the γ- and ϵ-AChRs, suggesting co-existence of both subtypes of AChRs at the same synaptic site at this stage (E18.5).

We next tested whether the development of presynaptic nerves was affected in γ^-/- muscles. We used neurofilament or syntaxin antibodies to label pre-terminal nerves, and synaptophysin, synaptotagmin 2 or SV2 antibodies to label synaptic terminals. We found striking presynaptic defects in γ^-/- muscles. As shown in Fig. 5, the nerves in theγ ^-/- embryos branched extensively and projected over a broad region of the muscle (Fig. 5B,D,F,H), whereas the nerves in wild-type embryos were nicely confined to the central region of the muscle(Fig. 5A,C,E,G). Increased nerve branching and extended projections in γ^-/- embryos were also observed in limb muscles, for example, in tibialis anterior, EDL, medial and lateral gastrocnemius and Sol (see Fig. S2 in the supplementary material).

Interestingly, synaptic vesicle proteins, such as synaptophysin,synaptotagmin 2 or SV2, were seen accumulated in nerve terminals of both wild-type and γ^-/- muscles. Preferential accumulation of synaptic vesicle proteins at nerve terminals was detectable as early as E15.5,when they appeared as numerous `puncta', immunolabeled by either synaptophysin(Fig. 5E,F), synaptotagmin 2(Fig. 5G,H, see also insets) or SV2 (data not shown). As development proceeded to subsequent stages(E16.5-E18.5), presynaptic nerve terminals became more intensely labeled by synaptic vesicle proteins (see Figs S3 and S4 in the supplementary material,and Fig. 6). To determine the spatial relationship between AChR clusters and nerve terminals, we double-labeled diaphragm muscles with α-bgt and Syt2, and analyzed the number of AChR clusters and nerve terminals at both E16.5 and E18.5 stages. In E16.5 wild-type muscles, we identified 780 AChR clusters and 763 nerve terminals within the ventral quadrant diaphragm muscles from three embryos;there were slightly more AChR clusters than nerve terminals within the same region of the muscle. However, all nerve terminals were colocalized with AChR clusters: among 763 nerve terminals we observed, all of them were closely apposed by AChR clusters (see Fig. S3C, arrowheads in the supplementary material); a small fraction of AChR clusters remained aneural (around 2%, 17 out of 780 clusters were aneural) (see Fig. S3C, arrows in the supplementary material). In striking contrast, for E16.5 γ^-/- muscles,there were more than twice as many nerve terminals as AChR clusters in the corresponding region. The majority of nerve terminals were not directly apposed by AChR clusters (supplementary material Fig. S3, arrows in F); among a total of 1183 nerve terminals examined, only 510 of them were apposed by AChR clusters (43%) (see Fig. S3F, arrowheads in the supplementary material). At E18.5, all nerve terminals in both wild-type and γ^-/-muscles were directly apposed by AChR clusters(Fig. 6). We counted 267 neuromuscular synapses from the γ^-/- and 505 neuromuscular synapses from the wild-type muscles; there was 100% co-localization of nerve terminals and AChR clusters in both genotypes.

However, there were substantial differences in the developing presynaptic nerve terminals between the γ^-/- and wild-type muscles. One of the striking differences was that the presynaptic nerves in theγ ^-/- muscles extended beyond the central region of the muscle so that numerous nerve sprouts projected towards the edge of the muscle (see Fig. S4B, arrows). By contrast, the nerves in the wild-type muscle were largely confined to the central region (see Fig. S4A, arrowheads in the supplementary material) and only a few nerve sprouts extended beyond the central region (see Fig. S4A, arrows in the supplementary material). The nerve sprouts in γ^-/- muscles appeared to be extrasynaptic(Fig. 6F, arrows), as they extended beyond the synaptic sites (arrowheads in Fig. 6F). By contrast, in the wild-type muscle, nerve sprouting occurred to a much lesser extent (arrows in Fig. 6C). We quantified the difference between the wild-type and γ^-/- muscles by measuring the width of AChR clusters and the width of innervation band within the ventral quadrant of the diaphragm muscle. The mean width of the innervation band in γ^-/- muscles was 355.0±21.4 μm,more than twice that of the end-plate band (152.6±21.2 μm, n=3 embryos), whereas the mean width of the innervation band in the wild-type muscle was 84.6±8.5 μm, close to that of the end-plate band (81.7±5.0, n=3). Thus, both the innervation band and end-plate band were greatly expanded in the γ^-/- muscles,compared with the wild-type muscles; however, the AChR clusters were always localized within the innervation band in both wild-type andγ ^-/- muscles.

AChE is a reliable marker for differentiated postsynaptic membrane (for a review, see Rotundo, 2003). We have previously observed that clustering of AChE occurs in aneural muscles independently of innervation, suggesting that AChE may also be prepatterned by mechanisms intrinsic to muscles (Lin et al., 2001). To determine whether AChE distribution depends on the presence of the pre-patterned AChR clusters, we carried out AChE staining inγ ^-/- muscles at E15.5, a stage when AChR clusters were absent(Fig. 1H). Our results showed that AChE clusters were present in γ^-/- muscles(Fig. 7C), despite the total absence of AChR clustering this stage (Fig. 1H). However, AChE clusters were distributed over a broader surface of the γ^-/- muscle(Fig. 7C) when compared with the wild-type muscle (Fig. 7A). The abnormally broad distribution of AChE clusters in theγ ^-/- muscles was also detected at E18.5(Fig. 7D).

We next investigated mechanisms that might lead to the absence of pre-patterned AChRs in γ^-/- muscles. The receptor tyrosine kinase MuSK is specifically expressed at postsynaptic sites(Glass et al., 1996; Jennings et al., 1993; Valenzuela et al., 1995). Therefore, we sought to determine whether the localization of MuSK was altered in γ^-/- muscles. We examined whole-mount diaphragm muscles at E15.5, a stage when clustering of AChRs was absent from γ^-/-muscles (Fig. 1). As shown in Fig. 8, MuSK was localized to the central region of the γ^-/- muscle (arrows in Fig. 8D,H), despite the absence of AChR clusters (Fig. 8C,G). The localization of MuSK clusters in the γ^-/- muscle was similar to that observed in the wild-type muscle (arrowheads in Fig. 8B,F), and consistent with the previously reported pattern of MuSK antibody staining in wild-type embryonic muscles (Bowen et al.,1998). Therefore, the phenotype developed in theγ ^-/- muscles is unlikely to be due to alternation of MuSK expression; instead, it probably resulted from the lack of the γ-subunit and consequently the absence of prepatterned AChRs(Fig. 1).

Rapsyn is a cytoplasmic protein colocalized with AChRs in the postsynaptic muscle membrane (Froehner et al.,1981). Synaptic localization of rapsyn occurs at the initial stages of neuromuscular synaptogenesis(Noakes et al., 1993), and deletion of rapsyn leads to a total absence of AChR clustering(Gautam et al., 1995). Furthermore, rapsyn is absent from synaptic sites in AChR mutants of zebrafish(Ono et al., 2004; Ono et al., 2002). Because of its essential role in AChR clustering, we asked whether the localization of rapsyn was affected in γ^-/- muscles. We carried out triple-immunostaining assay on muscle sections with antibodies against Syt2(to label the nerve terminal) and rapsyn, as well as α-bgt (to label AChRs). In E15.5 wild-type muscles, as expected, rapsyn(Fig. 9C) was localized at synaptic sites marked by Syt2 antibodies(Fig. 9A) and α-bgt(Fig. 9B). However, in E15.5γ ^-/- muscle, rapsyn staining was diffusely distributed across the entire sarcoplasm (Fig. 9F), with no specific localization at the synaptic sites marked by Syt2 antibodies (Fig. 9D). These results demonstrated that although rapsyn was expressed inγ ^-/- muscles, it failed to cluster at synaptic sites in the absence of the γ-subunit. Interestingly, synaptic localization of rapsyn was observed in E18.5 γ^-/- muscle(Fig. 9L). This is probably due to the emergence of the ϵ-subunit in E18.5 γ^-/-muscles.

Increased nerve branching in the γ^-/- muscles(Fig. 5) raised a possibility there were increased motoneurons in the γ^-/- embryos. Alternately, motoneuron number may remain unchanged, but motor axons may branch more extensively in the γ^-/- embryos. We therefore compared motoneuron numbers between the wild type and γ^-/-embryos. As shown in Fig. 10,motoneuron numbers in the γ^-/- embryos were significantly increased. For example, within the cervical spinal segments (C3-C8), there was an ∼65% increase of motoneuron number in the γ^-/-(7473±283, n=3 embryos) compared with wild-type embryos(4539±247, n=3).

In the present study, we focused our studies on the initial stages of neuromuscular synaptogenesis (E13-E15.5) and showed that pre-patterned AChR clusters were completely absent in γ^-/- muscles. AChR clusters emerged at later embryonic stages (E16.5-E18.5) owing to the expression of the ϵ-subunit. Thus, genetic ablation of the AChRγ-subunit specifically affected the formation of pre-patterned AChR clusters during the initial stages of neuromuscular synaptogenesis. The absence of pre-patterned AChR clusters in γ^-/- muscles was associated with excessive nerve branching, increased motoneuron survival, as well as aberrant distribution of AChE and rapsyn. However, clustering of MuSK proceeded normally in γ^-/- muscles.

The postsynaptic apparatus of the NMJ is assembled from large protein complexes that include AChRs, MuSK, rapsyn, utrophin, laminin, AChE,dystrobrevin and dystrophin-glycoprotein complex (DGC)(Bewick et al., 1996; Campbell, 1995; Ervasti and Campbell, 1991; Grady et al., 2000; Jacobson et al., 2001; Ohlendieck et al., 1991). These protein complexes interact intricately with each other, and altering one component may consequently affect the entire complex. For example, AChR clusters are closely associated with DGC, which in turn plays a crucial role in anchoring and stabilizing AChR clusters at the NMJ(Banks et al., 2003b; Henry and Campbell, 1996; Sunada and Campbell, 1995). Thus, it is possible that aberrant development of the NMJ in theγ ^-/- muscles may result from a broad disruption of postsynaptic proteins.

Our results demonstrated that MuSK, but not rapsyn, was clustered in the E15.5 γ^-/- muscles, suggesting the clustering of MuSK may precede the clustering of AChRs. Interestingly, recent studies by Burden's group have shown that MuSK is prepatterned in developing muscles and dictates the formation of the NMJ (Kim and Burden,2008). Thus, pre-patterning of MuSK may be one of the earliest events occurred during neuromuscular synaptogenesis. Activating MuSK at pre-patterned postsynaptic sites may then lead to the formation of pre-patterned AChR clusters, which may in turn sense the release of neurotransmitter from ingrowing axons(Chow and Poo, 1985; Hume et al., 1983; Young and Poo, 1983) and therefore establish communication with the nerve.

Clustering of rapsyn in the γ^-/- muscles at E18.5, but not at E15.5, also suggests that synaptic targeting of rapsyn does appear to require the full complement of AChR subunits. Indeed, in vitro studies have shown that rapsyn is closely associated with various AChR subunits, including the β-subunit (Burden et al.,1983), the α-subunit(Maimone and Merlie, 1993),and the β-, γ- and δ-subunits(Huebsch and Maimone, 2003). In addition, rapsyn appears to associate with AChRs in an intracellular compartment prior to its insertion into the plasma membrane(Marchand et al., 2002; Moransard et al., 2003),suggesting that rapsyn and AChRs may arrive at the plasma membrane together. Furthermore, rapsyn clusters are absent from the NMJ in AChR δ-subunit mutants of zebrafish, demonstrating that AChRs are required for synaptic localization of rapsyn (Ono et al.,2004; Ono et al.,2002). Together, these observations support the hypothesis that the full complement of AChR subunits is required for targeting rapsyn to synaptic sites.

The delayed emergence of AChR clusters at E16.5 in γ^-/-muscles was consistent with the onset of ϵ-subunit expression. Our RT-PCR and electrophysiology results demonstrate that the ϵ-subunit is expressed at late embryonic stages in both wild-type and γ^-/- muscles. Indeed, a recent study shows that the expression of ϵ-subunit is detectable at E18 in leg muscles (Yumoto et al., 2005). However, immunostaining with anti-ϵ antibody shows that the epsilon subunit is only expressed after P0(Missias et al., 1996). This discrepancy could be due to limited sensitivity of antibody immunostaining or lack of translation of the ϵ-subunit transcripts. The delayed AChR clusters appeared significantly larger in γ^-/- muscles,compared with those observed in the wild-type muscles. Such increases in end-plate sizes are probably due to decreases in muscle activities in theγ ^-/- muscles. Consistent with this idea, similar increases in AChR cluster size were also observed in mutant embryos deficient in ChAT,which leads to a total blockade of ACh-mediated synaptic transmission(Brandon et al., 2003; Misgeld et al., 2002).

The delayed AChR clusters in γ^-/- muscles could be nerve-or/and agrin-induced. Indeed, we observed 100% colocalization between the nerve terminals and AChR clusters at E18.5 in both wild-type andγ ^-/- muscles; there were no aneural AChR clusters at E18.5. The absence of aneural clusters at E18.5 could be due to dispersal effect of the ACh released from the nerve (Lin et al., 2005; Misgeld et al.,2005), or establishment of contacts by the newly arrived nerve terminals, or both. We cannot distinguish between these possibilities without using time-lapsed video-microscopy to follow the same nerve terminals/AChR clusters over time, as elegantly demonstrated in the zebrafish NMJ(Flanagan-Steet et al., 2005; Panzer et al., 2005; Panzer et al., 2006).

One of the hallmarks of presynaptic differentiation is progressive accumulation of synaptic vesicle proteins at the nerve terminal(Dahm and Landmesser, 1991; Lupa and Hall, 1989). Mechanisms underlying this process remain unclear. Our data suggest that the presence of pre-patterned AChR clusters is not required for initiating pre-synaptic differentiation. These results are consistent with studies in zebrafish NMJ previously reported: presynaptic nerve terminals develop normally in mutants lacking functional AChRs in zebrafish NMJ(Li et al., 2003; Ono et al., 2001; Panzer et al., 2006; Westerfield et al., 1990).

Previous studies have shown that motoneuron survival is markedly enhanced when ACh-mediated transmission is blocked either pharmacologically(Dahm and Landmesser, 1988; Hory-Lee and Frank, 1995; Oppenheim et al., 1989; Oppenheim et al., 2000; Pittman and Oppenheim, 1979; Pittman and Oppenheim, 1978)or genetically (Brandon et al.,2003; Misgeld et al.,2002). However, these approaches may unavoidably affect both the neuronal nicotinic AChRs in presynaptic cells and the muscle nicotinic AChRs in the postsynaptic cells. As the γ-subunit is expressed only in postsynaptic muscle cells, our results demonstrate blockade of ACh-mediated transmission specifically in postsynaptic muscle cells promotes motoneuron survival. Our results are consistent with previous reports demonstrating enhanced motoneuron survival in mutant mice lacking rapsyn, MuSK or agrin(Banks et al., 2001; Banks et al., 2003a; Terrado et al., 2001). Together, these studies suggest a postsynaptic mechanism regulates motoneuron survival during development.

We thank the RIKEN BioResource Center (Tsukuba, Japan) for maintaining and providing the AChR γ-subunit knockout mice, and Dr Atsushi Yoshiki at the RIKEN BioResource for coordinating the transferring of the mice. We are indebted to Drs Thomas Sudhof, Jane Johnson, Helmut Kramer, Paul Blount and Jonathan Terman for their critical comments on manuscript drafts, and to Drs William Betz, Mark Rich and Yoshie Sugiura for their valuable suggestions on electrophysiology. This work was supported by grants (to W.L.) from NIH/NINDS(NS055028), from Robert Packard Center for ALS Research at Johns Hopkins, from the Edward Mallinckrodt, Jr Foundation and from the Cain Foundation in Medical Research.