In mammals, two combinations of muscle nicotinic acetylcholine receptors (AChRs) are used: α2βγδ (γ‐AChR) or α2βϵδ (ϵ‐AChR). After birth, γ‐AChRs are replaced by ϵ‐AChRs (γ/ϵ‐switch). The two receptors have different conductances and open times. During perinatal period, the long open time γ‐AChRs generate random myofiber action potentials from uniquantal miniature end‐plate potentials (mEPPs). ϵ‐AChRs are suitable for strong adult muscle activities. Since the effect of the γ/ϵ‐switch on neuromuscular development was unclear, despite the many differences in channel characteristics, we carried out this study to generate γ‐subunit‐deficient mice. Homozygotes born alive survived for 2 days in a stable condition, and were able to move their forelimbs. Endplate AChRs included ϵ‐subunits, and muscle fibers had multiple neuromuscular junctions. Both pre‐ and postsynapses were abnormal and spontaneous action potentials generated from mEPPs were totally absent. Results suggest a requirement for γ‐AChRs in mediating synaptically‐induced action potential activity critical for neuromuscular development.
The muscle‐type nicotinic acetylcholine receptor (AChR), a ligand gated ion channel, triggers muscle action potentials in response to the release of quantal packets of acetylcholine (ACh) from motor nerve endings (Sanes and Lichtman, 1999). In mammals, the receptor takes two different forms: α2βγδ (γ‐AChR) or α2βϵδ (ϵ‐AChR). The mouse γ‐AChR is replaced by the ϵ‐AChR (γ/ϵ‐switch) within the first 2 weeks after birth (Mishina et al., 1986; Martinou and Merlie, 1991; Bouzat et al., 1994). Primary structures of γ‐ and ϵ‐subunits are more highly conserved between species than between the two subunits (Numa, 1986). The γ‐ and ϵ‐AChRs have 39 and 59 pS single‐channel conductances and 10.4 and 5.3 ms open times, respectively (Mishina et al., 1986). The difference in conductance is ascribed mainly to the transmembrane domains M1 and M2, the channel open time to the M4 domain and the amphipathic helix (HA) between M3 and M4. Although ϵ‐AChRs are concentrated at the adult neuromuscular junction (NMJ), γ‐AChRs are also expressed outside the junctional membrane prior to innervation. In adult, γ‐subunits are re‐expressed after skeletal muscle denervation.
The accumulation of γ‐AChRs at the NMJ is nerve‐induced. The neuronal isoform of agrin induces AChR clustering through muscle‐specific kinase (MuSK) and myotube‐associated specificity component (MASC) complex in the muscle cell membrane (Gautam et al., 1996; Glass et al., 1996). Rapsyn forms a subsynaptic cytoskeletal complex with AChRs and agrin‐induced accumulation of AChRs requires rapsyn (Apel et al., 1995; Gautam et al., 1995). As the original NMJ is a preferred re‐innervation site, it is thought that the synaptic complex remains at the same site after denervation (Sanes et al., 1978).
Neuromuscular activity represses extrasynaptic γ‐AChR expression and induces the regression of polyneuronal innervation, leading ultimately to a single contact per muscle cell in mature NMJs (Witzemann et al., 1991; Gan and Lichtman, 1998). Complete blockage of the NMJ in adult muscle induces sprouting of the motor nerve endings (Brown and Ironton, 1977) and transgenic mice that overexpress neuronal GAP43 show nerve sprouting that is independent of muscle activity (Aigner et al., 1995). These data suggest that paralyzed muscle produces the nerve sprouting factor that promotes the sprouting mechanism. The situation in embryonic synaptogenesis differs slightly because descending motor impulses are rarely available in developing muscle. Instead, uniquantal miniature end‐plate potentials (mEPPs) produced by the spontaneous release of ACh dominate in generating spontaneous muscle action potentials primarily because of the long open time of the γ‐AChR (Jaramillo et al., 1988). Although the significance of mEPPs is still unknown, it is possible that spontaneous mEPPs, in conjunction with the γ/ϵ‐subunit switch and receptor‐specific differences in single channel function, may contribute directly to NMJ development.
Although the γ/ϵ‐switch is observed commonly in different mammalian species, there is no agreement as to the functions and consequences of the switch. To elucidate the systemic role of the γ/ϵ‐switch in neuromuscular development, we generated γ‐subunit gene (AChRγ) disrupted mice.
Results and discussion
The gene targeting strategy is shown in Figure 1A. Three independent clones were identified through Southern blot analysis of a pool of transfected embryonic stem (ES) cells (Murray et al., 1992). Selected ES clones gave rise to germline chimera. Heterozygotes (AChRγ+/−) were inter‐crossed and homozygotes (AChRγ−/−) were obtained at about half the frequency expected from Mendelian segregation. About half of these were born alive (Table 1). Northern hybridization analysis revealed no mature γ‐subunit messenger RNA (mRNA) in homozygotes (Figure 1F) (Evans et al., 1987).
Heterozygotes were obviously normal. AChRγ−/− mice had a characteristic lunar‐shaped body profile (Figure 2) and reduced body weight 18 ± 8% (mean ± SD, n = 12). They showed a shallow, irregular and paradoxical respiration (the thorax was depressed corresponding to abdominal distension) and a problem in suckling. Strong cyanosis was observed in the mutants that were dead at birth, suggesting severe respiratory failure. Mutants born alive were stable for up to 2 days, were able to move their forelimbs and reacted to stimulation. The phenotype was similar in the three lines. Because respiration was irregular but continuous, dehydration by impaired milk intake seemed to limit their survival. This idea is supported by observations of an extended survival following intraperitoneal saline infusion (data not shown).
As the mutant mice were able to move their forelimbs, but not hindlimbs, RNA was extracted separately from each muscle (Table 2). In both control and mutant mice, γ‐subunit expression was lower in the forelimb muscle than in the hindlimb muscle, whereas the opposite was true for ϵ‐subunit expression, as was suggested by the chronological order of the γ/ϵ‐switch from rostral to caudal muscle (Sakmann and Brenner, 1978). However, no compensatory increase of the ϵ‐subunit was observed in AChRγ−/− mice, in contrast to the γ‐subunit compensation seen in the ϵ‐subunit gene defect (Witzemann et al., 1996; Missias et al., 1997). In mutant mice, the expression of β and δ subunits was at the same level as that of control mice.
AChR labeling using rhodamine‐labeled α‐bungarotoxin (BTX) revealed clearly stained foci in intercostal muscle fibers in controls, but foci were barely visible in AChRγ−/− mice (Figure 3A and B). In the diaphragm muscle, the patches of labeled AChRs were distributed in a broader range along the central end‐plate band and were unequal in size and shape compared with those of controls (Figure 3C and D). These staining data show that only the diaphragm muscle was able to support respiration, and hence explain the paradoxical respiration.
To reveal the NMJ on each muscle fiber, muscle actin was simultaneously stained and the section was observed by a laser scanning confocal microscopy (MRC‐1024, Bio‐Rad). Muscle fibers of AChRγ−/− mice had unequal sizes and lower density cross‐striation than those of control mice. In addition, multiple AChR clusters were seen on a single muscle fiber, which normally has only one cluster per synaptic region (Figure 3E and F). Anti‐neuronal antibody PGP9.5 staining showed the motor neuron wandering among muscle fibers and sprouting many branches that ended at AChR clusters (Figures 3E–H and 4). Therefore, the γ‐AChR is not necessary to initiate synaptic formation (Hashemolhosseini et al., 2000).
AChRs were also stained with subunit‐specific antibodies. Since staining with anti‐δ‐subunit antibody (88B) was localized at the synaptic region, extrasynaptic AChRs were suppressed in both wild type and mutant (Figure 5A and F). Staining with anti‐ ϵ‐subunit antiserum and BTX were colocalized in the mutants, thus suggesting that the expressed AChRs included the ϵ‐subunit (Figure 5G–I). Although functional channels exist with only α, β and δ subunits experimentally (Kurosaki et al., 1987; Liu and Brehm, 1993), the single channel conductance is reduced 20‐ to 30‐fold, the binding affinity of BTX is reduced to one‐fifth or less than that of the γ‐AChR (Bouzat et al., 1994) and channel activity disappears after exposure to ACh (drop‐out phenomenon). Therefore, the viability of the mutant may depend on the level of ϵ‐subunit expression.
Communication between pre‐ and postsynapses leads to synaptic specialization. Since AChRs were clustered and the clusters were innervated in AChRγ−/− mice, the cytoskeletal complex is formed and targeted by the motor neuron in the absence of AChR γ‐subunit expression. Because the γ‐subunit can be expressed elsewhere than the NMJ, which is not the case for the ϵ‐subunit, there could be a molecule that interacts specifically with either the γ‐ or the ϵ‐subunit. However, since such molecules have not been identified, one cannot explain the synaptic abnormalities seen in AChRγ−/− mice simply by γ‐subunit interacting molecules.
Examination of fine structure revealed a rather well‐developed presynapse with active zones, mitochondria and accumulations of synaptic vesicles, although the polarity of the vesicles was not clear (Figure 6B). Despite movements of the forelimb and diaphragm, indicating good conduction of motor nerve impulses to these muscles, the presence of well‐developed but sprouted nerve terminals indicates a history of inactivity. This could be explained by the loss of a fraction of action potentials generated by other than descending neural impulses.
To examine this possibility, intracellular recordings were obtained from proximal forelimb muscle (triceps) and the diaphragm, which indicated the presence of functional neuromuscular contacts. There was no significant difference (p > 0.05) in the resting potentials (RP) between the normal littermate, which had an average of −52 ± 4.6 mV (mean ± SD, n = 5 fibers), and the mutant mouse, which had an average of −45 ± 4.2 mV (mean ± SD, n = 7 fibers). Recordings from normal P0 littermates indicated that the large amplitude uniquantal mEPPs characteristic of this stage of development (Muniak et al., 1982) often elicited spontaneous action potentials (Figure 7A and C). These recordings also showed clusters of irregularly spaced action potentials (Figure 7A, traces 5–7) with inflections indicative of synaptic activity (Figure 7A, trace 5). This suggests that the spontaneous clusters of large synaptic uniquantal activity seen in tetrodotoxin‐treated fibers at this stage elicit irregular bursts of action potentials in untreated fibers (Carlson, 1992). In other instances, isolated action potentials were produced by single large‐amplitude mEPPs, which occurred at an overall frequency of 1–2 events/min (Figure 7C). Approximately 45% (5/11 using five fibers) of the mEPPs recorded from normal endplates produced action potentials in these P0 fibers (Jaramillo et al., 1988). In AChRγ−/− fibers, mEPPs were recorded at a frequency similar to those of normal littermates, but the mEPPs were subthreshold and never (0/12 using seven fibers) resulted in action potentials (Figure 7B). The absence of action potentials elicited by mEPPs in AChRγ−/− fibers was not secondary to a reduction in membrane excitability, since AChRγ−/− fibers were actually hyperexcitable and occasionally exhibited fibrillatory action potentials immediately upon impalement (data not shown).
Although it is impossible to measure the average mEPP amplitude in the control muscle without blocking muscle action potentials, mEPPs in AChRγ−/− fibers had an average value of 0.87 ± 0.47 mV (mean ± SD, n = 12 mEPPs). The average time constant of decay for the subthreshold mEPPs was 9.1 ± 1.6 ms (mean ± SD, n = 4 mEPPs) and 6.2 ± 0.9 ms (mean ± SD, n = 7 mEPPs) for control and mutant muscle fibers, respectively. This is consistent with shorter open times of the mutant AChR.
Since the first description of spontaneous quantal release (Fatt and Katz, 1952), the physiological role of the mEPP itself has remained unclear (Jaramillo et al., 1988; Kidokoro and Saito, 1988). Synapses are formed and maintained without action potentials (Cohen, 1972), but action potentials suppress γ‐subunit expression in non‐endplate regions. There is also evidence that mEPP‐triggered spontaneous muscle action potentials are related to cross‐striation formation of the myocytes (Kidokoro and Saito, 1988). The γ‐subunit is highly conserved in mammalian species, even though it is normally expressed only during the fetal to neonatal period. We have shown here that spontaneous mEPPs in conjunction with endplate γ‐AChRs may play an important role in synapse elimination and muscle and synaptic maturation. Expression of γ‐AChR can therefore be critical for early neuromuscular development, while high conductance, short open time ϵ‐AChRs are optimized for adult skeletal muscle fibers.
The mouse AChR γ‐subunit gene was isolated from a 129 genome library (a gift from T. Doetschman) using the human γ‐subunit cDNA as a probe (Shibahara et al., 1985). The 11 kb EcoRI–SmaI fragment was subcloned (Figure 1A). PGKneo (Stratagene) was inserted at the AatI site of exon 4. HSV‐tk was inserted at the 3′‐end of the insert. Feeder‐dependent RW4 ES cells (6.6 × 107; Genome Systems) were transfected with 25 μg of the linearized (NotI) targeting vector by electroporation (625 V/cm, 25 μF; EasyJect, EquiBio), divided into 64 wells and cultured against G418 (200 μg/ml) for 1 week. Twenty micrograms of genomic DNA extracted from each pool (Blin and Stafford, 1976) were digested with BamHI, electrophoresed and transferred to nitrocellulose filters. The filters were hybridized with the riboprobe, which was prepared as described previously (Melton, 1985) using the 1 kb EcoRI fragment subcloned from the 5′ flanking region as a template. The pools that showed a faint recombinant band (Figure 1B) were subcloned. Southern blot analysis was carried out for each subclone with an acrg5 probe, which is the 0.4 kb EcoRI–PstI fragment 5 kb upstream of the γ‐subunit gene (Figure 1C), as described previously (Honjo et al., 1979). Recombination was confirmed with acrg5 and pCRγ probes (Figure 1D and E).
Generation of null mutant mice.
Each clone was injected into C57BL/6 blastocysts. Chimeric male mice were crossed with C57BL/6 females to obtain agouti offspring. Targeted alleles were determined by Southern blot hybridization or PCR of tail DNAs (data not shown).
Northern blot hybridization.
Total RNA was extracted from the fore‐ and hindlimbs using guanidine HCl and organic solvents (Chomczynski and Sacchi, 1987). Ten micrograms each was electrophoresed after glyoxal treatment and transferred to nylon membranes (McMaster and Carmichael, 1977). Membranes were hybridized with 32P‐labeled DNA probes containing the 3′‐part of β, γ, δ or ϵ‐subunit cDNA, which were obtained by cloning PCR amplification products (Table 3), and were confirmed by sequencing (ABI373S, Perkin Elmer). Stripping off the probe, each filter was re‐hybridized with β‐actin probe. For denervation, the sciatic nerve of AChRγ+/− mice was cut on P12. After 4 days (P16), RNA was extracted from the hindlimb muscle.
Immunohistochemical and ultrastructural analysis.
Neonates were killed following the institutional guidelines and fixed with 1% formaldehyde for 10 min. Muscles were stained with 1 μg/ml of rhodamine‐labeled BTX (Molecular Probes) overnight. 88B (Affinity BioReagents), PGP9.5 (Ultraclone) and AChR ϵ‐subunit‐specific antiserum (a gift from J.R. Sanes) were used for staining as described by Mizoguchi et al. (1994). Actin was stained with phaloidin. For ultrastructural studies, neonates were fixed by transcardial perfusion with 2% paraformaldehyde and 2.5% glutaraldehyde in 100 mM phosphate buffer pH 7.4 for 10 min. Diaphragms were dissected out, soaked with the same fixative for 4 h, washed, post‐fixed with 1% OsO4 for 1 h, dehydrated and embedded in epoxy resin. Ultrathin sections were stained with uranyl acetate and lead citrate, and observed under an electron microscope (JEM‐1200EX, JEOL).
Intracellular recordings were obtained from P0 muscle fibers by using fiber‐filled glass micropipettes (Warner Instruments, 3 M KCl; resistances, 30–45 MΩ), and the recordings were amplified using a Warner Model IE201 Intracellular Electrometer and a Grass P15 amplifier. Both the diaphragm and triceps muscles were placed in a normal Ringer saline solution (125 mM NaCl, 5 mM KCl, 24 mM NaHCO3, 2 mM CaCl2, 1 mM NaH2PO4, 11 mM glucose pH 7.3; T = 23–25°C), continually bubbled with 95% O2 and 5% CO2. To obtain recordings from the triceps muscle, the entire forelimb and scapula were removed and pinned to a Sylgard base. Intracellular recordings were obtained in the usual way from the synaptic region of visually identified fibers (Carlson, 1992) and were identified by resting potential measurements while the electrode was advanced in order to sample spontaneous quantal activity. Because of fibrillatory activity after the muscle preparation, it was necessary to incubate 1–2 h until mEPP recordings started, a procedure that which would not affect the mEPP activity (Purves and Sakmann, 1974). Subsequently, fibrillatory activity was seen in some fibers within the first few minutes after impalement. Only fibers in which impalements could be maintained beyond this period were used for measuring mEPPs. The data were stored on a PCM recorder (Vetter Model 200T modified Toshiba recorder) and examined using PCLAMP software (v 4.0) and a Labmaster A‐D convertor. Records were presented using Sigmaplot v 5. The decay time constant (τ) was determined by negative exponential curve fits (V = A e−t/τ; V, voltage; t, time; A, constant) to individual mEPP decay phases.
We thank S. Kumagai, A. Takagi and H. Tango for technical assistance, T. Doetschman for the mouse genomic library, J.R. Sanes for the antibody, and C. Morrison and E. May for critical reading of the manuscript. This work was supported by a donation from the Shimadzu Co., Ltd. The Numa Project is a memorial project of the late Dr S. Numa. The authors dedicate this paper to his memory.
- Copyright © 2002 European Molecular Biology Organization