Neuromuscular synaptic patterning requires the function of skeletal muscle dihydropyridine receptors.

Developing skeletal myofibers in vertebrates are intrinsically 'pre-patterned' for motor nerve innervation. However, the intrinsic factors that regulate muscle pre-patterning remain unknown. We found that a functional skeletal muscle dihydropyridine receptor (DHPR, the L-type Ca(2+) channel in muscle) was required for muscle pre-patterning during the development of the neuromuscular junction (NMJ). Targeted deletion of the β1 subunit of DHPR (Cacnb1) in mice led to muscle pre-patterning defects, aberrant innervation and precocious maturation of the NMJ. Reintroducing Cacnb1 into Cacnb1(-/-) muscles reversed the pre-patterning defects and restored normal development of the NMJ. The mechanism by which DHPRs govern muscle pre-patterning is independent of their role in excitation-contraction coupling, but requires Ca(2+) influx through the L-type Ca(2+) channel. Our findings indicate that the skeletal muscle DHPR retrogradely regulates the patterning and formation of the NMJ.

The establishment of specific synaptic connections lays the foundation for neural circuitry formation. At the vertebrate NMJ, one of the best-studied model synapses, synaptogenesis occurs in a stereotypic pattern along the central regions of the post-synaptic muscle cells [1]. Emerging evidence indicates that these central regions are intrinsically pre-patterned by the muscle cells. For example, genes critical for neuromuscular synaptogenesis, including the acetylcholine receptor (AChR) and the muscle-specific kinase (MuSK), are selectively activated, and their respective proteins (AChR and MuSK) are clustered in the central regions of the muscle in the absence of innervation [2-5]. In vivo imaging in zebrafish embryos demonstrates that AChR clusters are not only pre-patterned in developing muscles but also are actively incorporated into newly-formed NMJs [6,7]. Furthermore, removal of pre-patterned AChRs in mouse embryos results in excessive nerve branching and aberrant formation of the NMJ [8]. These findings challenge the long-held “neurocentric” dogma and support the opposite view that post-synaptic muscle intrinsically defines its central region for the ingrowing nerves, which subsequently fine-tune the pre-patterned AChRs so that neuromuscular synapses are generated along a narrow end-plate band of the muscle [2,9,10]. The mechanisms underlying this intrinsic pre-patterning in muscle remain unknown.

Developing myotubes generate action potentials and contract spontaneously [11], and this electrical and contractile activity is correlated with an alteration of AChR levels [12,13]. This raises the possibility that muscle electrical activity, or contractile activity, or both, are involved in pre-patterning the muscle during NMJ development. This question has been technically challenging to answer because of the difficulty in separating electrical from contractile activity, as excitation is normally coupled with contraction (E-C coupling) [14]. Most widely applied methods for manipulating activity, including tetrodotoxin (TTX), high K+ (depolarization) or electrical stimulation, all affect both the electrical and contractile activity of the muscle. Thus, the relative contributions of electrical vs. contractile activity on neuromuscular development cannot be distinguished by applying these conventional means to perturb activity.

E-C coupling is mediated by two major membrane components – the L-type Ca2+ channels (DHPRs) [15], which are at the surface of the muscle membrane and act as the voltage sensor for depolarization, and the ryanodine receptors (RyRs) at the sarcoplasmic reticulum (SR), which regulate the release of Ca2+ from the SR into the cytosol [14]. Indeed, E-C coupling is abolished in mice deficient in either DHPRs, e.g. the β1 subunit of DHPR (Cacnb1) [16] or RyRs [17-19]. Therefore, we used these mutant mice (Cacnb1 and RyR1) to investigate the role of E-C coupling in the patterning and formation of the NMJ.

Surprisingly, muscle pre-pattern was disrupted in Cacnb1 muscles, but was largely preserved in RyR1 muscles. In the absence of Cacnb1, AChR and MuSK expression was markedly increased, leading to a broad distribution of AChR and MuSK proteins and aberrant development of the NMJ. These defects are specifically due to the loss of DHPR function in the skeletal muscle because reintroducing Cacnb1 into Cacnb1 muscles genetically rescued the muscle pre-patterning defects and restored the normal patterning and formation of the NMJ. Interestingly, both electrical and synaptic activity are highly active in Cacnb1 muscles, suggesting the patterning defects were unlikely due to a lack of either electrical or synaptic activity. Taken together, our findings demonstrate that a functioning skeletal muscle DHPR is specifically required for establishing muscle pre-patterning during NMJ formation.

Results

A loss of DHPR function leads to abnormal synaptic patterning

Pre-patterned AChRs are clustered along the central region of the muscles at embryonic day 14.5 (E14.5); at this initial stage, the majority of the nascent AChR clusters are near, but not directly apposed to, the nerve terminals [2,4]. This is demonstrated by immunostaining of whole mounts of E14.5 diaphragm muscles. In the WT muscle, AChR clusters were patterned to a centrally localized end-plate band (Fig. 1b) where the nerve terminals were also confined (Fig. 1d) and extended orderly branches within the central end-plate band region (Fig. 1a, g). In contrast, in Cacnb1 muscle, AChR clusters failed to form an end-plate band but instead were broadly distributed across the muscle, including the medial and lateral edges (Fig. 1c). For example, in the dorsal quadrant of the diaphragm muscles, AChR clusters in Cacnb1 mice were distributed along a length of 569 ± 10 μm (N = 5 mice) of muscle fibers, whereas AChR clusters in the control mice were confined to a length of 133 ± 6 μm (N = 5 mice) of muscle fibers. Correspondingly, the nerves in Cacnb1 muscles were highly branched and hyper-innervated over a broad region of the muscle, especially in the dorsal quadrant, in which the nerve branches occupied the entire muscle surface (Fig. 1e, f, h).

Figure 1

Loss of DHPR function leads to defects in muscle pre-patterning

Whole mounts of diaphragm muscles (E14.5) were double-labeled by α-bgt for AChR and by anti-neurofilament (NF150) and anti-synaptotagmin 2 (Syt2) for the nerve. Increased nerve branching and expansion of innervation territories were detected in Cacnb1 muscle (f) compared with WT muscle (a). White dashed lines (a) delineate the myotendinous junction between the central tendon and the medial edge of the muscle fibers. The asterisk (*) (f) indicates a small branch of the intercostal nerve attached to the outer edge of the diaphragm muscle. b–e, g–h: high-magnification views of the dorsal regions of the diaphragm muscle. In the WT muscle, AChR clusters were aligned along the central region of the muscle, forming a pre-patterned end-plate band (arrow in b); the nerves extend fine branches and terminate within the central end-plate band (d, g). In contrast, in the Cacnb1 muscle, AChR clusters were distributed broadly across the entire surface of the muscle, including the medial and lateral edges of the muscle (arrowheads in c); the nerves in the Cacnb1 muscle also branched extensively and expanded their innervation territories across the entire muscle surface (e, h). At this stage, the majority of the AChR clusters were not directly apposed to the nerve terminal (g, h). i–j: AChR clusters were distributed along a central end-plate band in HB9 muscles (i), but were distributed broadly in double null Cacnb1 muscles (j). Scale bar: a, f: 200 μm; b–e, g–h: 100 μm; i–j: 400 μm.

Because the pre-patterning is muscle intrinsic, it also occurs in aneural muscles such as the diaphragm muscle in mice deficient in the motoneuron transcription factor HB9 [2,4]. To determine whether DHPRs are required for the pre-patterning in aneural muscles, we examined the diaphragm muscles from Cacnb1 and HB9 double knockout mice (Cacnb1). Interestingly, the muscle intrinsic, nerve-independent pre-patterning of AChR clusters in HB9 mice (Fig. 1i) was disrupted in Cacnb1 double knockout mice. AChR clusters were found scattered over the entire muscle (Fig. 1j). Thus, like in the innervated muscle described above, DHPRs were also required for pre-patterning the aneural muscles during development.

To determine whether the pre-patterning defects displayed in the Cacnb1 muscles simply represent developmental delay, we examined the muscles at a later stage (E18.5). AChR clusters were confined to the central end-plate band in WT muscle (Fig. 2a) but remained scattered broadly across the entire Cacnb1 muscle (Fig. 2b). To determine the number of AChR clusters in individual myofibers, we dissociated E18.5 diaphragm muscles and counted them in a total of 359 control (WT or heterozygous) and 554 Cacnb1 myofibers. We found that multiple endplates were frequently detected in dissociated Cacnb1 myofibers (Fig. 2d) but rarely in control myofibers (Fig. 2c). Similar to the distribution of AChR clusters, acetylcholinesterase (AChE) was restricted to the central regions of the WT myofibers (Fig. 2h) but was broadly distributed along the entire length of the Cacnb1 myofibers (Fig. 2i). These findings demonstrated that the patterning defects of the Cacnb1 muscles persisted throughout development in Cacnb1 muscles, and were not attributable to developmental delay.

Figure 2

Loss of DHPR function leads to multiple synaptic sites and an expansion of innervation territories

a–b: Distribution of AChR clusters (labeled by α-bgt) in whole mounts of diaphragm muscles (E18.5). AChR clusters were aligned in a central endplate band in WT (arrow in a), but were scattered over a broad region in Cacnb1 muscles (b). c-d: Dissociated myofibers stained with FITC-conjugated α-bgt and rhodamine-conjugated phalloidin. e: histogram distribution of the percentage (in log scale) of dissociated myofibers containing various numbers of endplates ranging from 1 to 7. Multiple endplates were frequently detected in dissociated Cacnb1 myofibers (d); the majority of them (72%, 399 out of 554) contained 2 or more endplates per fiber (2 patches: 42.8%; 3 patches: 16.4%; 4 patches: 8.1%; 5 patches: 3.8%; 6 patches: 0.7% and 7 patches: 0.2%). In contrast, the majority of control myofibers (99.4%, 357 out of 359) contained single endplate; less than 1% of them (0.6%, 2 out of 359) contained two endplates per fiber. f–g: The innervation territories were confined within the central region in the WT (bordered by dashed lines f), but expanded to the entire muscle in Cacnb1 (g). h–i: Distribution of AChE clusters (arrow) revealed by cholinesterase staining. AChE clusters were localized along the central region of the muscle in the WT (h), but were broadly distributed in the Cacnb1 muscle (i). Inset in i shows a high power view of individual myofibers containing multiple AChE clusters (arrowheads) in Cacnb1 muscle. Scale bars: a–b: 400 μm; c–d: 50 μm; f–g: 200 μm; h–i: 250 μm.

Both electrical and synaptic activity are detected in Cacnb1 muscles

Despite aberrant patterning, neuromuscular synapses were established in Cacnb1 muscles. As in WT (Fig. 3a–c), every end-plate in the Cacnb1 muscle (Fig. 3d–f) was fully innervated; that is, the pre- and post-synaptic apparatus formed in close apposition (arrows in Fig. 3c, 3f). However, individual synapses appeared larger in the Cacnb1 muscle, compared to the age-matched controls. We therefore measured the sizes of individual end-plates and found a two-fold increase in size of end-plates in the Cacnb1 muscles, compared to the controls (supplementary Fig. S1). Additionally, 99.4% (n=342) of the end-plates in the control myofibers retained an oval shape (Fig. 3b) that is characteristic of embryonic end-plates [20]. In contrast, nearly half (48.6%, n=399) of the end-plates in Cacnb1 muscles appeared to be perforated (arrowheads, Fig. 3e), which was an indication that the end-plates were more matured than those in the WT muscles [20]. Thus, the post-synaptic end-plates in Cacnb1 muscles appeared precociously matured, compared with those in the age-matched WT muscles, suggesting that muscle DHPR activity may negatively regulate the maturation of the NMJ during development.

Figure 3

DHPR function is not required for synaptogenesis, but its absence leads to increased synaptic and muscle electrical activity

a–f: Confocal images of E18.5 diaphragm muscles doubly labeled by presynaptic markers (NF150 and Syt2) (a, d) and postsynaptic marker α-bgt (b, e). Every end-plate in both WT (b) and the Cacnb1 muscle (e) was fully innervated by the nerves (arrows in c, f). Endplates in the control muscle appeared predominantly as ovoid-shaped plaques (b), whereas endplates in Cacnb1 muscles were bigger, and some were perforated (arrowheads in e). g-h: Electron micrographs of the NMJ (E18.5, diaphragm muscle). In both WT (g) and Cacnb1 (h), synaptic vesicles (SV) were abundantly present at the nerve terminal (NT), and the basal lamina (white arrow in g, h) was well-defined in the synaptic cleft. i–j: Sample traces of mEPPs from control (i) and Cacnb1 (j) myofibers. The mEPP frequency was markedly increased in Cacnb1 muscles. k–l: Spontaneous action potentials (arrowheads) in control (k) and Cacnb1 (l) mice; the lower trace in k and l illustrates an expanded view of the portion of the upper trace indicated by the grey line. Arrow in k indicates the displacement of the trace resulting from muscle contraction in the control. Scale bars: a–f: 20 μm; g–h: 0.5 μm.

To further examine neuromuscular synapses, we carried out electron microscopy analyses. We found that the Cacnb1 muscle were markedly thinner and the sarcomeres were less well aligned compared with the control muscles (supplementary Fig. S2). However, the pre-synaptic nerve terminals in Cacnb1 mice contained abundant synaptic vesicles and well-defined basal lamina (arrow in Fig.3h).

To determine if the neuromuscular synapses in the Cacnb1 mice were functional we performed electrophysiological analysis. We found that neuromuscular synaptic activity in Cacnb1 mice was markedly increased compared to their littermate controls. For example, mEPP frequency (events/minute) was increased more than 60-fold in Cacnb1 muscles (57.86 ± 12.73, n = 38 cells, N = 4 mice, 8635 total events in 181 minutes), compared with controls (0.85 ± 0.16, n = 26 cells, N = 4 mice, 104 total events in 128 minutes) (P = 0.00007). Furthermore, spontaneous action potentials were more frequently detected in Cacnb1 muscle cells (58%, or 22/38 muscle cells) (Fig. 3l) compared with control muscle cells (8%, or 2/26 muscle cells) (Fig. 3k). In addition, mEPP amplitudes were also significantly increased in Cacnb1 mice (3.19 ± 0.22 mV, n=38 cells, N=4 mice), compared with the control (2.57 ± 0.19 mV, n=26 cells, N=4 mice) (P < 0.0047). Interestingly, EPP amplitudes were similar between control (13.25 ± 1.25 mV, n = 23 cells, N = 4 mice) and Cacnb1 mice (12.75 ± 1.22 mV, n = 27 cells, N = 4 mice) (P = 0.227), and action potentials were elicited in both control and Cacnb1 muscles in response to electrical stimulation of the nerve. Together, these results demonstrated that both the electrical and synaptic activity were highly active in the Cacnb1 muscles. Thus, the abnormal patterning of the NMJ that was displayed in Cacnb1 muscles cannot be attributed to a lack of electrical or synaptic activity in these muscles.

Muscle-specific expression of Cacnb1 rescues the patterning defects in Cacnb1 muscle

Since Cacnb1 is expressed in both skeletal muscles and neurons [21], we considered the possibility that the lack of Cacnb1 in neurons might have contributed to the defects of the NMJ in Cacnb1 mice. We therefore examined compound mutant mice specifically lacking Cacnb1 expression in neurons [22]. These compound mutant mice were generated based upon previous studies demonstrating that reintroducing Cacnb1 cDNA into Cacnb1 myotubes completely restores Ca2+ current, charge movements and Ca2+ transients to normal levels [23] and that human skeletal actin (HSA) promoter-driven Cacnb1 (HSA::Cacnb1) restores Cacnb1 expression specifically in Cacnb1−/− skeletal muscles. These compound mutant mice (HSA::Cacnb1; Cacnb1, hereafter referred to as the rescued mice) are normal, viable and express Cacnb1 in skeletal muscles but not in neurons [22].

We found that the levels of synaptic activity and the patterns of the neuromuscular synapses were restored to normal in the rescued mice (Fig.4): the level of neuromuscular synaptic activity was similar between the control (Fig. 4a) and the rescued mice (Fig. 4b). For example, mEPP frequency was restored to normal in the rescued mice (Cacnb1, 1.07±0.10, n=53 cells, N=7 mice) compared with the control (1.00±0.09, n=47 cells, N=5 mice). The mEPP amplitudes (control, 2.00±0.08 mV, n=51 cells, N=5 mice; rescue, 1.92±0.06 mV, n=65 cells, N=7 mice) and EPP amplitudes were also similar between control and rescued mice (control: 15.56±0.61 mV, n=61 cells, N=5 mice; rescue: 14.07±0.60 mV, n=72 cells, N=7 mice).

Figure 4

Muscle-specific expression of Cacnb1 rescues the patterning defects in Cacnb1 muscle

a–b: Sample traces of mEPP recorded from P0 diaphragm muscles in control (a) and the Cacnb1 mice that also carry a HSA::Cacnb1 transgene under the control of the muscle-specific HSA promoter, as shown in the schematic drawing [Cacnb1, referred to as the rescued mice (b)]. c–h: Wholemount diaphragm muscles at E14.5 (c–e) and E18.5 (f–h) from the rescued mice (Cacnb1 were double-labeled for AChRs (c, f) and the nerves (d, g). Similar to the WT muscles (compared to Fig. 1, Fig. 2), AChR clusters in the rescued muscles were aligned to a central endplate band (arrow in c, f) and nerve terminals were also confined to a central endplate band as shown in the merged images (e, h). i–l: AChE staining of wholemount diaphragm muscle (Dia, P0) and triangularis sterni muscle (TS, P90) from control (i, j) and the rescued mice (Cacnb1 (k, l). The patterns of AChE staining (black arrow) were similar between control and the rescued mice. Scale bars: c–e, 100 μm, f–h, 200 μm; i–l, 1000 μm.

Likewise, the patterning of the NMJs was also restored to normal in the rescued mice (Fig. 4c–h), similar to that displayed in the WT mice (Fig. 1b, d, g, Fig. 2f). Since the rescued mice were viable, we further followed their NMJ development to postnatal stages. We found that postnatal NMJ patterns and sizes of the individual end-plates were also restored to normal in the rescued mice (Supplementary Fig. S3). Furthermore, AChE expression patterns in the rescued mice (Fig. 4k–l) were indistinguishable from those of control mice (Fig. 4i–j). These results demonstrated that muscle-specific expression of Cacnb1 was sufficient to rescue the patterning defects of the NMJ in Cacnb1 mice, and that the muscle DHPR, but not the neuronal DHPR was required for the establishment of the normal pattern of the NMJ.

Distinct phenotype in pre-patterning between Cacnb1 and RyR1 muscles

DHPRs directly interact with and activate the RyRs and initiate E-C coupling. To determine whether the defects of muscle pre-patterning in the Cacnb1 mice had resulted from a disruption of E-C coupling, we examined mutant mice deficient in RyRs, which were thus lacking E-C coupling. Mice have three distinct RyR genes, RyR1, RyR2 and RyR3 [24]. Because both RyR1 and RyR3 are expressed in embryonic skeletal muscles [25], we therefore focused on double mutant mice deficient in both RyR1 and RyR3 (RyR1) [19].

Surprisingly, unlike the Cacnb1 muscles in which AChR clusters were broadly distributed (Fig. 1c), AChR clusters were confined to the central region in E14.5 RyR1 muscle (Fig. 5d). The average width of the endplate band in RyR1 muscles, however, was increased compared to E14.5 WT muscles. For example, in the dorsal quadrant of the diaphragm muscles, the average width of the end-plate band was 154 ± 13 μm in RyR1 mice (N = 5 mice) and 133 ± 6 μm (N = 5 mice) in WT mice. Importantly, a central end-plate band was present in E14.5 RyR1 muscles, indicating that the muscle pre-pattern was established in E14.5 RyR1 muscles. Interestingly, innervation in the RyR1 muscle (Fig. 5e) was markedly increased compared with the littermate control (RyR) (Fig. 5b). Nevertheless, the nerve terminals in the RyR1 muscle were confined to the central end-plate band (Fig. 5f), distinctly different from that was seen in the Cacnb1 muscles (Fig. 1e, h). As development proceeded to a later stage (E18.5), numerous ectopic AChR clusters appeared in the peripheral regions of the RyR1 muscle, but the majority of AChR clusters were still aligned to a centrally localized endplate band (Fig. 5h), in a pattern different from that seen in the Cacnb1 muscle (Fig. 5i). These data suggest that the pre-patterning defects in the Cacnb1 muscle are specifically attributed to the loss of functioning DHPRs, but neither to the lack of muscle contractile activity nor to the absence of SR Ca2+ release resulting from the loss of the RyRs.

Figure 5

RyRs and DHPRs play different roles in muscle pre-patterning

a–f: Distribution of AChR clusters (a, d), nerves (b, e) and merged images (c, f) revealed by double-immunofluroscence staining of E14.5 diaphragm muscles from RyR1 (d–f) and littermate control mice (RyR1, a–c). AChRs were clustered along central regions of the muscles in both RyR1 and RyR1 mice (arrows in a, d), in a pattern similar to that seen in the WT mice (Fig. 1b). Increased innervation was detected in RyR1 (e) compared with the control (RyR1, b). Nevertheless, nerve terminals in both RyR1 and RyR1 mice were confined to the central region of the muscle as shown in the merged images (arrowheads in c, f). g–i: AChR distribution in E18.5 diaphragm muscles. Unlike Cacnb1 muscle in which AChR clusters were broadly distributed (i), the majority of AChR clusters were aligned in a central end-plate band in E18.5 RyR1 muscle (arrow in h), similar to the end-plate band seen in the E18.5 WT muscle (arrow in g). However, some AChR clusters were ectopically localized to the peripheral regions of the E18.5 RyR1 muscle (* in h). Scale bar: a–f: 100 μm; g–i: 400 μm.

Skeletal muscle DHPRs regulate the expression of AChR and MuSK

During normal neuromuscular development, AChR genes are selectively transcribed at myonuclei near the synaptic sites (sub-synaptic myonuclei) and suppressed at extra-synaptic myonuclei; as a result, AChR transcripts are localized at the central end-plate band [26]. This specific pattern of AChR expression is also pre-patterned in the muscle, as it is maintained in the absence of innervation [2,4]. We therefore examined the levels and patterns of AChR gene expression in Cacnb1 muscles. Using quantitative real-time RT-PCR, we found that AChRα-subunit (AChRα) expression was significantly increased in E14.5 Cacnb1 muscles compared with WT muscles (relative expression levels of AChRα: WT, 1.00 ± 0.20, N= 8 mice; Cacnb1, 2.82 ± 0.25, N= 8 mice; P = 0.007). In addition, in situ hybridization revealed that the pattern of AChRα expression in the E14.5 Cacnb1 diaphragm muscle was completely disrupted, being broadly distributed along the entire surface (Fig. 6b), rather than being localized to the central region, as in the E14.5 WT diaphragm muscle (Fig. 6a). Similar to the disruption of the AChR clustering, the disruption of AChRα expression pattern persisted throughout development. For example, AChRα transcripts were detected only in the central regions of the E18.5 WT intercostal muscles (Fig. 6c, left panel), but were uniformly distributed in the E18.5 Cacnb1 intercostal muscles (Fig. 6c, middle panel). The normal pattern of AChRα expression was restored in the intercostal muscles of the rescued mice (HSA::Cacnb1; Cacnb1) (Fig. 6c, right panel).

Figure 6

DHPRs pattern neuromuscular synapses by regulating the expression of AChR and MuSK

Wholemount in situ hybridization with DIG-labeled AChRα probes were carried out in E14.5 diaphragm muscles (a–b) or E18.5 intercostal muscles (c). AChRα transcripts were detected along the central regions in the WT (a), but were broadly distributed in the Cacnb1 muscle (b). c: AChRα transcripts were localized to the central region of the WT muscle (left panel in c), but were detected in the entire Cacnb1 muscle (middle panel in c). The normal distribution of AChRα transcripts was restored in the rescued mice (right panel in c). d: Wholemount in situ hybridization of intercostal muscles using DIG-labeled MuSK probes. MuSK transcripts were localized to the central region of the WT (left), but were broadly distributed in the Cacnb1 muscle (right). e: Relative expression levels of AChR and MuSK assayed by quantitative real-time PCR in C2C12 myotube cultures treated with L-type Ca2+ channel antagonists (verapamil, isradipine) or agonist (Bay K 8644) compared to the untreated or vehicle (DMSO) treated controls. Verapamil (10 μM) significantly increased the relative expression levels of MuSK (2.21 ± 0.13, N = 5 cultures, P = 0.0002) and AChR (2.10 ± 0.32, N = 5 cultures, P = 0.009) compared with controls (MuSK, 1.03 ± 0.12; AChR, 1.04 ± 0.16, N = 5 cultures). Similarly, isradipine (1 μM) also significantly increased the relative expression levels of MuSK (1.84 ± 0.21, N = 3 cultures, P = 0.0003) and AChR (1.68 ± 0.31, N = 3 cultures, P = 0.0189) compared with controls (MuSK, 1.01 ± 0.10; AChR, 1.00 ± 0.02, N = 3 cultures). Bay K 8644 significantly decreased the expression of MuSK (0.60 ± 0.13, N = 3 cultures, P = 0.0012) and AChR (0.83 ± 0.07, N = 3 cultures, P = 0.0337) compared with controls (MuSK, 1.02 ± 0.02; AChR, 1.04 ± 0.10, N = 6 cultures). f: Wholemount diaphragm muscles were immunostained with anti-MuSK antibodies and α-bgt. MuSK protein was clustered along the central region of the WT (left column), but broadly distributed in the Cacnb1 muscles (middle column). MuSK distribution was restored to normal in the rescued mice (right column). Data are presented as mean ± SEM. Scale bars: a–b: 500 μm; c: 500 μm; d: 400 μm; f: 50 μm.

How does DHPR function to pre-pattern the developing myofibers? Because the muscle specific kinase, MuSK, is expressed in the central regions of the muscle, independent of innervation, and overexpressing MuSK leads to a broadening of end-plate bands and innervation zones [5], we hypothesized that the patterning defects of Cacnb1 muscles may be due to altered expression of MuSK. We therefore measured MuSK expression levels in Cacnb1 muscles and found that MuSK expression was significantly increased in E14.5 Cacnb1 muscles compared with E14.5 WT muscles (relative expression levels of MuSK: WT, 1.00 ± 0.23, N= 8 mice; Cacnb1, 2.12 ± 0.21, N= 8 mice; P = 0.006). Furthermore, MuSK transcripts were broadly distributed along the entire Cacnb1 muscle (Fig. 6d, right panel) rather than being concentrated at the central end-plate band as they were in WT muscles (Fig. 6d, left panel). To examine the distribution of MuSK protein, we double-labeled diaphragm muscles with anti-MuSK antibody and α-bungarotoxin. We found that the MuSK protein was detected throughout the Cacnb1 muscle, coinciding with the broad distribution pattern of AChR clusters in the Cacnb1 muscle (Fig. 6f, middle panels), whereas MuSK protein was detected along the central region of the muscle in the WT (Fig. 6f, left panels). Like the AChR pattern, the normal distribution pattern of MuSK protein was also restored in the rescued mice (Fig. 6f, right panels). These findings demonstrated that functional DHPRs were necessary for regulating both the level and the pattern of MuSK expression.

To further elucidate the mechanisms by which DHPRs regulate the expression of AChR and MuSK, we tested whehter Ca2+ influx through the L-type channels was involved. We treated WT myotube cultures (C2C12) [27] with L-type Ca2+ channel antagonists, including verapamil and isradipine [28]. We found that both verapamil and isradipine significantly increased the relative expression levels of MuSK and AChR compared with controls. Conversely, the L-type Ca2+ channel agonist (Bay K 8644) significantly decreased the expression of AChR and MuSK compared with controls (Fig. 6e). Together, these results demonstrated that muscle DHPRs regulate the expression of AChR and MuSK through Ca2+ influx.

Discussion

Using genetic approaches in mice, we have revealed an unexpected role of the DHPRs in the establishment of muscle pre-patterning during neuromuscular synaptogenesis. We found that DHPR function is necessary for regulating proper levels and patterns of AChR and MuSK expression. In the absence of muscle DHPR function, AChR and MuSK expression is up regulated and AChR and MuSK protein is broadly distributed, resulting in pre-patterning defects. We further demonstrated that DHPRs regulate the expression of AChR and MuSK through Ca2+ influx, because application of L-type Ca2+ channel antagonists to WT myotubes increased the expression of AChR and MuSK, whereas application of L-type Ca2+ channel agonists decreased their expression (summarized in Supplementary Fig. S4).

The mechanism by which DHPRs govern muscle pre-patterning is likely independent of their roles in E-C coupling, as distinct phenotype in AChR clustering is displayed between Cacnb1 and RyR1 muscles, which both lack E-C coupling. These findings indicate that DHPRs are the major contributors for regulating the establishment of muscle pre-patterning. This is further supported by our genetic rescue experiments that restore the normal muscle pre-pattern when Cacnb1 is specifically reintroduced into Cacnb1 muscles.

Interestingly, at a later stage (E18.5) of development, numerous AChR clusters appear ectopically outside of the central end-plate band in RyR1 muscles. These results indicate that RyRs, although not being required for the establishment of muscle pre-patterning at the initial stage of neuromuscular synaptogenesis (E14.5), are important for the subsequent development of the NMJ. In addition, like Cacnb1 muscles, innervation is markedly increased in RyR1 muscles. One possible contributing factor for the emergence of ectopic AChR clusters in E18.5 RyR1 muscles and the resemblance of increased innervation between Cacnb1 and RyR null muscles is that the L-type Ca2+ current is significantly reduced, although not completely eliminated, in RyR null muscles [18,29]. Alternatively, increased innervation in both Cacnb1 and RyR null muscles could be attributed to an enhanced motoneuron survival resulted from a blockade of muscle contractile activity in these mutant muscles; the underlying mechanism remains to be further elucidated.

One of the most intriguing findings from previous studies is that aneural muscles in HB9 mice are capable of pre-patterning the expression of AChR [2,4] and MuSK [5] in the absence of the nerves. Our present study demonstrates that the pre-patterning of aneural muscle in HB9 mice requires DHPR function, because HB9 double knockout mice fail to establish muscle pre-patterning. This finding provides an important step for elucidating mechanisms underlying muscle intrinsic pre-patterning. Muscle fibers depolarize spontaneously during development [11]. Membrane depolarization is detected by the DHPR, which functions both as a voltage sensor and a L-type Ca2+ channel and initiates Ca2+ influx [14]. This depolarization-initiated Ca2+ influx through the DHPR negatively regulates AChR and MuSK expression. In the absence of Cacnb1, L-type Ca2+ currents are absent [16,30], this negative regulation is lost, allowing increased AChR and MuSK expression and a disruption of muscle pre-patterning.

In addition to their predominant localization in the transverse tubular membrane, DHPRs are also found in discrete foci in the sub-sarcolemmal region of the muscle [31], raising the possibility that DHPRs may directly regulate gene expression independent of their roles in regulating Ca2+ channel activity. Indeed, previous studies have shown that the C terminus of the L-type Ca2+ channels in cardiac muscles functions as a transcription factor and directly regulates gene transcription [32,33]. Additionally, the Ca2+ channel β3 subunit, a homolog of Cacnb1, regulates gene transcription by interacting with a novel isoform of Pax6 [34]. These studies suggest a possibility that Cacnb1 (DHPR β1) may also exert its effect on gene expression, like DHPR β3 does, by functioning directly as a transcription factor or by binding to one or more novel transcription factor. However, our data demonstrate that the underlying mechanism that leads to an increase in AChR and MuSK expression in Cacnb1 muscles is attributed to the loss of L-type Ca2+ channel activity, because application of L-type Ca2+ channel agonist and antagonists in WT muscles lead to a significant decrease and increase, respectively, of both AChR and MuSK expression.

The abnormal innervation defects displayed in Cacnb1 muscles resemble those reported in mice with muscular dysgenesis (mdg) [35-37], a lethal autosomal recessive mutation leading to an absence of the α1s subunit of DHPR [38]. It was proposed that the defects seen in mdg mice are “neurogenic” rather than “myogenic” in nature, based on experiments in which mdg muscles that were co-cultured with normal spinal cord neurons were thought to regain normal muscle activity [39,40]. Our transgenic rescue experiment unequivocally demonstrates that the defects of abnormal NMJ development in Cacnb1 mice are myogenic, i.e., the result of a lack of muscle DHPR function, with no effect exerted by neuronal DHPRs.

Previous studies have demonstrated that electrical activity plays crucial roles in regulating neuromuscular synapse formation during development and regeneration. For example, blocking electrical activity increases AChR [26,41,42] and MuSK [43] expression in the skeletal muscles. Conversely, enhancing electrical activity suppresses AChR expression in skeletal muscles [44]. Recent studies have further indicated that the activity-dependent gene expression is regulated by histone deacetylases (HDACs), such as HDAC9 [45], or HDAC4 [46]. Because muscle electrical activity is not only present but also significantly enhanced in Cacnb1 muscles, one would expect AChR and MuSK expression in Cacnb1 muscles to be suppressed. To the contrary, we found that AChR and MuSK expression are significantly increased and that AChR and MuSK transcripts are broadly distributed along the Cacnb1 muscle. These results suggest that the mechanism by which electrical activity suppresses AChR and MuSK expression is that it must activate the muscle DHPRs, the L-type Ca2+ channel in skeletal muscles. Interestingly, a previous study has shown that L-type Ca2+ channel activity regulates the metabolic stabilization of end-plate AChRs in chronically denervated soleus muscles in adult rats [47]. Furthermore, L-type channels are also involved in regulating neuronal gene transcription [48,49]. Therefore, L-type Ca2+ channel activity is likely one of the common mechanisms underlying activity-dependent gene expression in various types of excitable cells including nerve and muscle cells.

Online Methods

Mice

Mutant mice deficient in the DHPR β1 subunit (MGI: 2181804, Cacnb1tm1Rgg, or Cacnb1) [16], RyR1RyR3 [19] and HAS::Cacnb1 transgenic mice [22] have been described previously. HB9 mutant mice [50] were obtained from the Jackson Lab (stock number 006600, strain name: B6.129S1-Mnx1tm4(cre)Tmj/J). Homozygous null (Cacnb1, HB9 and RyR1) or rescued Cacnb1 mutants (Cacnb1; HAS::Cacnb1) were obtained by dated pregnancies of heterozygous mating; the day when a vaginal plug first appeared was designated as embryonic (E) day 0.5. After selected intervals of development, null embryos and their littermate controls [including wild-type (WT, +/+) and heterozygous (+/−)] were collected by Cesarean section of anesthetized pregnant mice. All experimental protocols followed NIH Guidelines and were approved bythe UT Southwestern Institutional Animal Care and Use Committee.

Light and electron microscopic analysis of the NMJ

Immunofluorescence labeling of the embryonic NMJs was performed as described [8]. Whole mounts of diaphragm and intercostal muscles were fixed in 2% paraformaldehyde (PFA) in 0.1 M phosphate buffer (pH 7.3) overnight at 4°C. The samples were blocked in dilution buffer (500 mM NaCl, 0.01 M phosphate buffer, 3% BSA and 0.01% thimerosal), and then incubated overnight at 4°C with anti-neurofilament 150 (NF150) plus anti-synaptotagmin-2 (Syt 2) antibodies, or with anti-MuSK antibodies. After extensive washes, muscle whole mounts were then incubated with fluorescein isothiocyanate (FITC)-conjugated secondary antibody and Texas Red-conjugated α-bungarotoxin (α-bgt) (10−8 M, Invitrogen Corporation, Carlsbad, CA). Samples were then washed with PBS and mounted in VECTASHIELD mounting medium (Vector Laboratories, Inc. Burlingame, CA). Images were acquired using a Zeiss LSM 510 Meta confocal microscope. Acetylcholinesterase (AChE) staining was carried out in whole mounts of diaphragm or triangularis sterni muscles. Muscles were fixed with 2% PFA in 0.1 M phosphate buffer (pH 7.3) overnight at 4°C. After extensive wash in PBS, the muscles were pre-incubated at 37 °C for 1 hr in a solution (pH 5.5) containing ethopropazine (0.2 mM), acetylthiocholine iodine (4 mM), glycine (10 mM), cupric sulfate (2 mM) and sodium acetate (65 mM). The muscle samples were then incubated at room temperature and under a fume hood for 2–5 minutes in sodium sulfide (1.25%, pH 6.0), followed by extensive wash in water (under a fume hood). The muscles were then flat mounted onto a glass slide and imaged under a stereomicroscope (SteREO Discovery V8, Carl Zeiss).

Quantification of endplates (AChR patches) in dissociated single myofibers was carried out in E18.5 diaphragm muscles isolated from 3 pairs of control (WT or heterozygous) and Cacnb1 mice. The diaphragms were digested with 0.2% of collagenase in Dulbecco’s Modified Eagle’s Medium (DMEM) at 37°C for 30 minutes, fixed in 1% PFA for 15 min on ice, washed in PBS and stained with FITC-conjugated α-bungarotoxin (10−8 M, Invitrogen Corporation, Carlsbad, CA) to label endplates and Rhodamine conjugated phalloidin (1:200, Invitrogen Corporation, Carlsbad, CA) to label myofibers. Images were acquired using a Zeiss LSM 510 Meta confocal microscope.

For electron microscopy, E18.5 diaphragm and intercostal muscles were immersion-fixed in freshly prepared 1% glutaraldehyde plus 4% PFA in 0.1-M phosphate buffer (pH 7.4). Tissues were then dissected and remained in the same fixative overnight. The tissues were post-fixed with 1% osmium tetroxide and embedded in Epon 812. Ultrathin-sections (70–80 nm) were stained with uranyl acetate and lead citrate, and observed using a Tecnai transmission electron microscope.

Electrophysiology

Electrophysiological analyses were carried out as previously described [8]. Briefly, phrenic nerve-diaphragm muscles (E18.5 or P0) were dissected in oxygenated (95% O2 and 5% CO2) rodent normal Ringer’s solution (NaCl 136.8mM, KCl 5mM, MgCl2 1mM, NaH2PO4 1mM, NaHCO3 12 mM, D-Glucose 11mM, CaCl2 2mM). The preparation was left in the oxygenated Ringer’s solution for at least 30 minutes before recording. All the recordings were performed at room temperature (22°C). The end-plate region was impaled by a glass micropipette (30–40 MΩ) filled with solution containing 2 M potassium citrate and 10 mM KCl. A square-pulse stimulation (0.1ms, 2V) was delivered to the nerve via a suction electrode connected to an external stimulator (Model: SD9, Grass, USA). The signal was amplified by an AxonClamp-2B amplifier (Molecular Devices, USA) and digitized at 10kHz by Digidata 1322A (Molecular Devices, USA). Data were recorded by Clampex software (Molecular Devices, USA) and stored in a PC for further off-line analysis.

Myotube culture

Mouse C2C12 cells [27] were obtained from American Type Culture Collection (ATCC, Manassas, VA). Cells were plated in 60 mm culture dishes coated with 1% gelatin overnight. Cells (myoblasts) were initially cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, GlutaMAX) with 20% fetal bovine serum and 0.5% chick embryo extract, without antibiotics, at 37 °C, 8% CO2. Upon reaching 90% confluence, the culture medium was replaced with DMEM plus 2% horse serum to induce myotube fusion. After an additional 72 hours, myotube cultures were treated with either the L-type Ca2+ channel antagonists including isradipine (1 μM) for 48 hours or verapamil (10 μM) for 24 hours, or treated with the L-type Ca2+ channel agonist (S)-(−)- Bay K 8644 (10 μM) for 24 hours. Vehicle controls include DMSO or medium only. To maximize the uniformity in drug treatment, all reagents were freshly prepared and were mixed gently with culture medium prior to application to the cultures. Both L-type Ca2+ channel antagonists (isradipine and verapamil) and agonist (S)-(−)-Bay K 8644 were purchased from Tocris Bioscience (Ellisville, MO).

RT-PCR and quantitative real-time PCR

Total RNAs were isolated from E14.5 hindlimbs of mouse embryos, or from C2C12 myotube cultures using TRI reagent (Molecular Research Center, Cincinnati, OH). First-strand cDNAs were synthesized using SuperScript III cDNA synthesis kit (Invitrogen Corporation, Carlsbad, CA). Quantitative real-time PCR was performed by using Fast SYBR Green Master Mix on a 7500 Fast Real-Time PCR System (Applied Biosystems, Foster City, CA). Each sample was assayed in triplicate reactions. The amplificationdifference between AChR α-subunit and muscle creatine kinase (Mck), or between MuSK and Mck in each sample was calculated and then normalized to that of wild-type muscles or control myotube cultures using the comparative CT method (Applied Biosystems, Foster City, CA). The following primers were used for PCR amplification of specific gene products: (1) AChR α-subunit - forward AAG CTA CTG TGA GAT CAT CGT CAC, reverse TGA CGA AGT GGT AGG TGA TGT CCA; (2). MuSK -forward CCC TGC AAG TGA AGA TGA AA, reverse TTC AAG AAC TGC GAT TCT GG; (3). Mck – forward CGT GTC ACC TCT GCT GCT, reverse CCT TCA TAT TGC CTC CCT TCT [5].

Statistical Analysis

The Student’s t-test was applied to assess statistical differences. Differences were considered statistically significant if the p-value (P) was less than 0.05. All quantitative data were presented as means ± standard error of the mean (SEM).

In situ hybridization

Whole mount in situ hybridization was carried out as previously described [2]. Briefly, ribcages including diaphragm and intercostal muscles were fixed in freshly prepared 4% PFA in 0.1 M phosphate buffer at 4°C overnight and then dehydrated through a series of methanol solutions (25, 50, 75 and 100%). Hybridization was carried out at 70°C overnight in hybridization buffer containing 50% formamide, 1.3X SSC, 50 μg/ml yeast tRNA, 0.2% Tween-20, and 1% Triton X-100, using digoxigenin-labeled α-AChR [26] or MuSK [5] riboprobes. After hybridization, the samples were washed with TBS containing 1% Tween-20, blocked with 5% goat serum in dilution buffer, and incubated with alkaline phosphatase conjugated anti-digoxigenin (1:1000, Boehringer Mannheim) overnight at 4°C, and hybridization signals were detected in a staining solution containing 100 mM Tris (pH 9.5), 0.4 mg/ml nitro blue tetrazolium chloride (NBT), 0.19 mg/ml 5-Bromo-4-chloro-3-indolyl phosphate (BCIP), 100mM NaCl and 50 mM MgCl2.