Neurotransmitter release is regulated by voltage-dependent calcium channels (VDCCs) at synapses throughout the nervous system. At the neuromuscular junction (NMJ) electrophysiological and pharmacological studies have identified a major role for P- and/or Q-type VDCCs in controlling acetylcholine release from the nerve terminal. Additional studies have suggested that N-type channels may be involved in neuromuscular transmission. VDCCs consist of pore-forming α1 and regulatory β subunits. In this report, using fluorescence immunocytochemistry, we provide evidence that immunoreactivity to α1A, α1B, and α1E subunits is present at both rat and human adult NMJs. Using control and denervated rat preparations, we have been able to establish that the subunit thought to correspond to P/Q-type channels, α1A, is localized presynaptically in discrete puncta that may represent motor nerve terminals. We also demonstrate for the first time that α1A and α1B (which corresponds to N-type channels) may be localized in axon-associated Schwann cells and, further, that the α1B subunit may be present in perisynaptic Schwann cells. In addition, the α1E subunit (which may correspond to R/T-type channels) seems to be localized postsynaptically in the muscle fiber membrane and concentrated at the NMJ. The possibility that all three VDCCs at the NMJ are potential targets for circulating autoantibodies in amyotrophic lateral sclerosis is discussed.
- calcium channels
- neuromuscular junction
- motor neuron
- Schwann cells
- skeletal muscle
- amyotrophic lateral sclerosis
Voltage-dependent calcium channels (VDCCs) play a major role in neuromuscular transmission by allowing calcium ion influx into motor neuron (MN) terminals and thereby effecting acetylcholine release. VDCCs have been classified electrophysiologically into T (low-voltage-activated; LVA) and L, N, P, Q, and R (high-voltage-activated; HVA) subtypes. These channels are composed of at least three subunits: a pore-forming α1subunit and structural/regulatory α2δ and β subunits. Multiple α1 gene products have been identified, termed α1A, α1B, α1C, α1D, and α1E (Snutch et al., 1990; Williams et al., 1992a,b, 1994;Soong et al., 1993; Zahl et al., 1994). Although there has been debate as to whether α1A is the pore-forming subunit of P and/or Q channels (Zhang et al., 1993; Stea et al., 1994), α1Bis thought to encode N channels (Williams et al., 1992a,b), and α1C and α1D are thought to be components of L channels. The identity of the channel formed after expression of α1E remains unclear (Soong et al., 1993; Zhang et al., 1993; Schneider et al., 1994; Williams et al., 1994; Bourinet et al., 1996).
At the mammalian neuromuscular junction (NMJ) most evidence suggests that P and/or Q channels control acetylcholine release from MN terminals (Sano et al., 1987; De Luca et al., 1991; Uchitel et al., 1992; Protti and Uchitel, 1993; Bowersox et al., 1995; Hong and Chang, 1995; Protti et al., 1996) although there are some data suggesting that N channels also may play a role in this process (Hamilton and Smith, 1992; Hong et al., 1992; Rossoni et al., 1994). Determination of the α1 subunit(s) localized at the mammalian NMJ should help to clarify which VDCC is involved in neuromuscular transmission. In addition, identification of the α1 subunit at the human NMJ is of interest in amyotrophic lateral sclerosis (ALS), a disease characterized by degeneration of MNs, denervation atrophy of muscle (Campbell and Munsat, 1994), and defects in neuromuscular transmission (Denys and Norris, 1979; Maselli et al., 1993). In this disease it has been reported that a large proportion of patients has circulating autoantibodies directed against the α1 subunit of VDCCs (Smith et al., 1992; Kimura et al., 1994). It is thought that VDCCs in MN terminals at the NMJ may represent targets for these autoantibodies and may play a role in MN degeneration in ALS (Appel et al., 1991; Smith et al., 1994; Mosier et al., 1995; Siklos et al., 1996).
To date, there is only one previous study of α1 subunit localization at the NMJ. Ousley and Froehner (1994) demonstrated the presence of α1A-like immunoreactivity (α1A-ir) at the rat NMJ, although the pre- or postsynaptic nature of immunolabeling was not determined. In the present study we have used fluorescence immunocytochemistry with VDCC α1 subunit-specific antibodies to determine the localization of α1A, α1B, and α1E subunits at the human and rat NMJ. By comparison with markers of known distribution, the pre- or postsynaptic localization of these subunits was investigated at NMJs in control and denervated rat muscle.
MATERIALS AND METHODS
Tissues. Human gastrocnemius muscle was obtained from patients undergoing lower limb amputation. This material was taken from patients whose primary pathology was not associated with nerve or muscle. Blocks of muscle were frozen and stored in liquid nitrogen before use. Adult female Wistar rats (180–200 gm) were used in all other experiments. For denervation, animals were anesthetized (halothane 1–3%), and a 1 cm portion of the right leg sciatic nerve was cut out. The wound was closed, and the animals were allowed to recover for 7 d, after which time the rats were stunned and killed by cervical dislocation; then the soleus muscles were removed. Contralateral leg muscles were used as controls. Blocks of muscle were either frozen for cryostat sectioning or teased into bundles of fibers to examine NMJs en face. In the case of both human and rat muscle sections, frozen blocks of tissue were cut transversely (6 μm) on a cryostat microtome, and sections were thaw-mounted onto gelatin-coated slides. NMJ-containing regions of muscle were identified via a histochemical method for demonstrating the presence of cholinesterase (Karnovsky and Roots, 1964). Sections were air-dried (45 min) before storage at −20°C. For preparation of teased fibers, the muscle was pinned out and lightly fixed in 0.5% paraformaldehyde in 0.1 m phosphate buffer, pH 7.4, for 30 min and then teased in PBS into bundles of 3–10 fibers. The fixed, teased fibers were placed in 6-well tissue culture plates in PBS before immunocytochemistry. In addition to soleus muscle, a 1.5 cm portion of the sciatic nerve was removed from the mid-thigh region of control rat hind limb. This tissue was fixed in 4% paraformaldehyde (15 min) before freezing and storage in liquid nitrogen. Fixed, frozen blocks of sciatic nerve were cut at 5 μm. Slide-mounted sections were air-dried and stored at −20°C before use.
Antibodies. Polyclonal antibodies specific for human α1A, α1B, and α1E VDCC subunits were produced and characterized as described previously (Volsen et al., 1995). Briefly, rabbits were immunized with fusion proteins comprising a region of the cytoplasmic loop between IIS6 and IIIS1 of α1A (amino acids 1048–1208), α1B (amino acids 983–1106), and α1E (amino acids 958–1058), which were attached to glutathione S-transferase (GST). Immunoglobulin fractions were isolated from the resulting antisera, followed by removal of anti-GST antibodies and immunopurification on α1-GST-linked Sepharose 4B columns. Residual cross-reacting antibodies were removed by further immunoadsorption. The specificity of each antisera was confirmed by immunofluorescent analysis of transfected HEK-293 cells and by ELISA (Volsen et al., 1995). For immunocytochemistry, the following antibody concentrations were used: 5.3 μg/ml α1A, 3 μg/ml α1B, and 1.2 μg/ml α1E. Other primary antibodies (dilutions in parentheses) used in the present study include mouse anti-synaptophysin monoclonal antibody (1:100, Dako); mouse anti-neurofilament protein RT97 monoclonal antibody (1:10; a gift from Dr. B. Anderton, Institute of Psychiatry, London, UK; Wood and Anderton, 1981), which recognizes polyphosphorylated neurofilaments; rabbit anti-cow S100 polyclonal antibody (1:100, Dako), which recognizes both subunits of the Schwann cell calcium-binding protein, S100; mouse monoclonal anti-β-spectrin RBC25C4 (used for teased fibers, 1:10; a gift from Dr. L. Anderson, University of Newcastle upon Tyne, UK) (Bewick et al., 1992); and NCLSPEC2 (used for sections, 1:50; NovoCastra Laboratories, Newcastle, UK), both of which were raised to human red blood cell ghosts but recognize β-spectrin in muscle. Labeling with primary antibodies was visualized by using rhodamine-conjugated swine anti-rabbit and rabbit anti-mouse secondary antibodies (1:100, Dako, High Wycombe, UK). The secondary antibodies were preincubated with normal rat serum (ratio 2:1) for 1 hr at room temperature before application to muscle preparations. All antibody dilutions were made in PBS containing 3% BSA and 0.1 mlysine.
Immunocytochemistry. Both sections and teased fiber preparations were double-labeled with a marker for the NMJ and with antibodies to the proteins of interest. In human muscle sections, NMJs were identified by using fluorescein isothiocyanate (FITC)-conjugated dolichos biflorus lectin (DBA, 5 μg/μl, Sigma, Poole, UK), which binds to N-acetyl-d-galactosamine residues in the basal lamina that are concentrated at the NMJ (Sanes and Cheney, 1982). In rat muscle sections and teased fibers, NMJs were identified by labeling acetylcholine receptors with FITC-conjugated α-bungarotoxin (BgTX, 6 × 10−7 m; Molecular Probes, Eugene, OR). The NMJ markers were applied at the same time as the secondary antibodies.
All labeling procedures were performed at room temperature, and all washes were in PBS except where stated. Thawed cryostat sections were permeabilized with 0.1% Triton X-100 in PBS for 5 min. Then sections were washed for 15 min and incubated in primary antibody at 4°C overnight (∼15 hr). After being washed for 1 hr, sections then were incubated in secondary antibodies, together with an NMJ marker (either FITC-DBA or FITC-BgTX), for 1 hr. After several washes over 30 min, sections were fixed in 1% paraformaldehyde for 30 min. Finally, sections were washed (30 min) and mounted in anti-fading fluorescence mountant (Vectashield, Vector Laboratories, Burlingame, CA). Some sections in each experiment were incubated without primary antibodies to control for nonspecific binding or cross-reactivity of the secondary antibodies. The labeling procedure for cryosections of sciatic nerve was the same as that used for sections of muscle. However, no NMJ marker was added to the secondary antibody before it was applied to the sciatic nerve sections.
For immunolabeling of teased muscle fiber preparations, all incubations were performed in 6-well tissue culture plates. Permeabilization of teased fibers was achieved with 1% Triton X-100 (30 min) for all antibodies, except for synaptophysin, which required permeabilization with absolute alcohol (−20°C, 10 min). After permeabilization, preparations were washed (30 min) and incubated in primary antibody at 4°C overnight. Subsequently, the fibers were washed (1 hr) before incubation with secondary antibody (plus FITC-BgTX) for 2 hr. Then preparations were washed (30 min), refixed in 1% paraformaldehyde (15 min), and placed in PBS before mounting. Teased fibers were mounted onto slides in Vectashield.
Microscopy and photography. Preparations were viewed and photographed on a Nikon Optiphot-2 microscope with a Nikon 40× oil plan apo objective. A Nikon G-2A filter set was used to photograph the rhodamine fluorescence. The FITC fluorescence was photographed through a Nikon B-2A filter set, the selectivity of which was enhanced by using a Leitz 525/20 nm bandpass emission filter to avoid “bleed through” of the rhodamine signal. The exposure times and printing conditions were the same for all photographs in each figure. In some experiments double-labeled en face NMJs were viewed with a confocal laser microscope (laser λ 488 nm for FITC, laser λ 568 nm for rhodamine; Bio-Rad MRC600, Munich, Germany) with a 20× [numerical aperture (NA) 0.4] and 40× (NA 0.7) objective.
The distribution of α1A, α1B, and α1E subunits at the NMJ was studied in sections of human and rat muscle and in teased fibers from control and denervated rat muscle. To determine the localization of these subunits at the NMJ and to confirm denervation, we also labeled human and rat muscle preparations with marker antibodies to neurofilament, synaptophysin, S100, and β-spectrin, which are localized in preterminal axons, nerve terminals, Schwann cells, and the muscle fiber cytoskeleton, respectively.
VDCC subunit localization at the human NMJ
The distribution of VDCC α1 subunits at the human NMJ in transverse sections of gastrocnemius muscle is shown in Figure1. Sections were dual-labeled with DBA to identify the NMJ (left-hand images) and antibodies (Ab;right-hand images) directed against α1A(A), α1B (B), α1E (C), neurofilament (D), S100 (E), and β-spectrin (F). In all sections DBA labeling was concentrated at the NMJ (intense white areas) and also was evident around the outside of the muscle fiber, consistent with the labeling of sugar residues in the basal lamina surrounding the muscle fiber membrane (Sanes and Cheney, 1982). Comparison of DBA binding with antibody labeling revealed that α1A-ir (A) and α1B-ir (B) were localized only at the NMJ, similar to the labeling pattern seen with neurofilament (D) and S100 antibodies (E). The α1E antibody (C) labeled the NMJ and around the outside of the muscle fiber. This pattern of labeling is similar to that observed for the cytoskeletal muscle protein, β-spectrin (F) (Bewick et al., 1992), which is concentrated at the NMJ because of membrane folding. In addition, α1E-ir (C) was observed in blood vessels lying between muscle fibers. When primary antibody was omitted from the immunostaining procedure, virtually no labeling was observed (data not shown).
The results show that α1A-ir, α1B-ir, and α1E-ir are all localized at the human NMJ. To determine whether immunolabeling with these antibodies was localized in pre- or postsynaptic structures at the NMJ, we studied α1 subunit localization in control and denervated rat muscle.
VDCC subunit localization at the rat NMJ
Control and denervated muscle sections
VDCC subunit distribution in transverse sections from control and 7 d denervated rat soleus muscle is shown in Figure2. Sections were dual-labeled with BgTX to label postsynaptic acetylcholine receptors at the NMJ and antibodies (Ab) to the protein of interest. In control rat muscle sections (Fig. 2 A–F) comparison of BgTX binding with antibody labeling revealed that α1A-ir (A), α1B-ir (B), and α1E-ir (C) were localized at the NMJ, as were synaptophysin-ir (D), S100-ir (E), and β-spectrin-ir (F). However, some differences among the labeling patterns of the various α1 subunits were noted. At the NMJ the α1A antibody (A) produced a punctate pattern of labeling (marked by arrows). Similarly, punctate labeling of the NMJ was observed with the synaptophysin antibody (D), which is known to be present in MN terminals. In contrast, α1B-ir (B) and α1E-ir (C) were localized over the entire NMJ area demarcated by BgTX binding. In addition, α1E-ir (C) was localized around the outside of the muscle fiber, similar to β-spectrin-ir (F). Thus, in both rat and human muscle sections α1A and α1B were localized at the NMJ only, whereas α1E was localized at the NMJ and the extrajunctional muscle fiber membrane.
In denervated rat muscle sections (Fig. 2 G–L) no labeling was observed for α1A (G), synaptophysin (J), and S100 (K) antibodies at the NMJ, whereas α1B-ir (H), α1E-ir (I), and β-spectrin-ir (L) looked very similar to that in control sections. Disappearance of synaptophysin-ir in these sections is consistent with retraction and degeneration of MN terminals in response to denervation. Labeling with the Schwann cell marker S100 should persist after denervation, although the pattern of labeling may change because Schwann cells migrate to engulf nerve terminals and send processes beyond the boundaries of the NMJ after denervation (Reynolds and Woolf, 1992; Son et al., 1996). It is unclear why, in the present study, there is an apparent lack of this marker in denervated muscle sections (K). One possible explanation is that the solubility of the S100 antigen resulted in its leaching out of unfixed, denervated muscle sections during immunocytochemistry. Persistence of β-spectrin-ir in denervated muscle sections is consistent with localization of this protein in the muscle fiber membrane (Bewick et al., 1992) rather than in presynaptic structures. The finding that α1A-ir disappeared after denervation, whereas α1B-ir and α1E-ir remained, suggests that α1A is localized presynaptically, whereas α1B and α1E are localized in structures other than MN terminals. In addition, the similarity of α1E immunolabeling to β-spectrin-ir suggests that this VDCC subunit is localized in the muscle fiber membrane.
Control and denervated teased fibers
To study VDCC subunit distribution at the NMJ in more detail, we performed immunolabeling on teased fibers from control (Fig.3 A–G) and denervated (Fig.3 H–N) rat muscle. In these preparations the NMJ could be viewed en face, together with preterminal processes and terminal boutons. In addition, some labeled, teased fiber preparations were examined with a confocal microscope where images of antibody labeling and BgTX binding could be superimposed to show regions of colocalization. Low-magnification (Fig.4 A–C) and high-magnification (Fig.4 A′–C′) confocal images are shown to illustrate clearly the labeling of preterminal structures and the extent of colocalization of antibody labeling and BgTX binding. In control teased fibers, comparison of BgTX binding with α1A-ir (Figs.3 A, 4 A,A′) showed that this subunit exhibited a punctate distribution over the surface of the NMJ. The extent of colocalization of α1A-ir with BgTX binding is illustrated in the confocal microscope images (Fig.4 A,A′), which demonstrate that α1A-labeled puncta lie within larger areas of BgTX labeling. Analysis of 11 puncta from two NMJs revealed that they were 2.7 ± 0.4 μm in length (mean ± SD), with an area of 4.7 ± 1.1 μm2 (mean ± SD). The discrete pattern of α1A-ir at the NMJ may represent localization of this subunit in clusters of active zones within individual MN terminal boutons. Labeling with the synaptophysin antibody (Fig. 3 E) was less discrete than α1A-ir and closely matched the distribution of BgTX binding. The pattern of α1B-ir was quite distinct from that of both α1A and α1E subunits. The α1B antibody labeled the entire surface area of the NMJ, with labeling extending beyond and in between the regions labeled by BgTX (Figs. 3 B, 4 B,B′). In contrast, α1E immunolabeling of the NMJ closely matched the BgTX binding pattern (Figs. 3 C, 4 C,C′). Thus, the distribution of α1E-ir was similar to that of the cytoskeletal muscle protein, β-spectrin (Fig. 3 G), indicating that α1E may be localized in the muscle fiber membrane. On the other hand, the distribution of α1B-ir suggests that this subunit probably is not localized in the muscle fiber membrane, because a protein with this localization would match closely the distribution of BgTX binding and β-spectrin-ir.
Both α1A (Figs. 3 A, 4 A) and α1B (Figs. 3 B, 4 B) antibodies labeled processes leading into each NMJ. Preterminal processes also were labeled with the axonal marker, neurofilament, in addition to neurofilament-ir in sprays of terminal branches at the NMJ (Fig. 3 D). It should be noted here that α1A-labeled (Figs. 3 A, 4 A) and α1B-labeled (Figs. 3 B,4 B) preterminal processes appeared to be slightly thicker in diameter than neurofilament-labeled axons (Fig.3 D) and exhibited node-like structures. This labeling pattern is reminiscent of that seen with the axon-associated Schwann cell marker, myelin-associated glycoprotein (N. C. Day and S. J. Wood, unpublished observations), suggesting that α1A and α1B may be present in axon-associated Schwann cells (see below). The Schwann cell marker S100 was present in ovoid-shaped nuclei and processes of perisynaptic Schwann cells at the NMJ (Fig.3 F) and also labeled axon-associated Schwann cells.
In denervated teased fiber preparations (Fig. 3 H–N), BgTX binding exhibited a similar distribution to that seen in control preparations, although the surface area of the NMJ was smaller because of atrophy of the muscle fiber after denervation. No labeling of the NMJ or preterminal processes could be detected with the neurofilament (Fig. 3 K) and synaptophysin (Fig. 3 L) antibodies, confirming that the nerve fibers and terminals had degenerated. Similarly, denervated teased fibers were devoid of α1A-ir at the NMJ and in preterminal processes (Fig.3 H). The lack of NMJ labeling in denervated preparations further supports the notion that α1A, like neurofilament and synaptophysin, is localized in presynaptic MN terminals. The lack of preterminal process labeling in denervated preparations suggests either that α1A is localized in axonal processes or that this subunit is localized in axon-associated Schwann cells, which decrease α1A expression in response to denervation. The data from teased fiber studies do not allow us to distinguish between these two possibilities.
In contrast to the above antibodies, α1B-ir (Fig.3 I), α1E-ir (Fig.3 J), S100-ir (Fig. 3 M), and β-spectrin immunolabeling (Fig. 3 N) persisted at the NMJ in denervated fiber preparations. These findings imply that α1B and α1E are localized predominantly in structures other than MN terminals or axons. Labeling with the α1B antibody (Fig. 3 I) was similar to that seen in control preparations, with α1B-ir present at the NMJ and in preterminal processes. These observations, coupled with the finding that α1B-ir covered a more extensive surface area than BgTX binding, suggest that α1B, like S100, is not localized postsynaptically in the muscle fiber membrane but may be localized in both perisynaptic and axon-associated Schwann cells. Differences in α1B and S100 labeling at the NMJ may be attributable to intense labeling of Schwann cell nuclei by the S100 antibody, but not by the α1B antibody. The persistence of S100 labeling in denervated teased fiber preparations (Fig. 3 M), compared with the lack of staining in denervated muscle sections (Fig. 2 K, see above), could be attributable to the fact that, unlike muscle sections, teased fiber preparations were fixed before immunolabeling (possibly preventing leaching of the S100 antigen) and that the structure of the NMJ is more intact in the teased fiber preparation. The pattern of α1E-ir in denervated teased fibers (Fig.3 J) was very similar to that in control preparations, as was β-spectrin-ir (Fig. 3 N).
VDCC subunit localization in rat sciatic nerve
In an attempt to resolve whether α1A and α1B labeling of preterminal processes represented labeling of MN axons or axon-associated Schwann cells, we compared α1A-ir, α1B-ir, and α1E-ir with S100-ir in transverse sections of rat sciatic nerve (Fig.5). The S100 antibody produced a double-ring pattern of labeling (Fig. 5 A), which is consistent with labeling of both inner and outer membranes of axon-associated Schwann cells. The α1A (Fig. 5 B) and α1B (Fig.5 C) antibodies produced a similar, but fainter, pattern of labeling to S100, suggesting that both VDCC subunits are present in axon-associated Schwann cells. The complete absence of α1E-ir in sciatic nerve sections (Fig. 5 D) is consistent with other data in the present study demonstrating a lack of α1E-ir in structures other than the muscle fiber membrane.
This study is the first to present a detailed comparative study on the localization of three α1 subunits at the human and rat NMJ. We have shown that α1A, α1B, and α1E are all found at the NMJ but seem to be localized in different cells. A comparison of control and denervated rat muscle suggests that α1A is localized in MN terminals, both α1A and α1Bmay be localized in Schwann cells, and α1E is localized in the muscle fiber membrane. Because the overall distribution of VDCC subunits and marker proteins in rat muscle sections is similar to that seen in human muscle sections, it seems likely that α1A, α1B, and α1E are localized in the same cell types at the human NMJ.
VDCCs in MN terminals
Evidence that α1A is localized in MN terminals comes from several observations. First, α1A labeling of the rat NMJ, like the labeling observed with the presynaptic marker synaptophysin, completely disappeared after denervation. Second, the finding that α1A labeling at the rat NMJ appeared as discrete puncta may be consistent with localization of α1A in clusters of active zones at the presynaptic membrane of individual MN terminal boutons. Interestingly, the area of α1A-labeled puncta at the rat NMJ (4.7 ± 1.1 μm2; mean ± SD) compares favorably with MN terminal bouton area derived from electron microscopy ultrastructural studies in rat (8.6 μm2, soleus muscle; S. J. Wood, unpublished data), mouse (3.4 μm2, epitrochleoanconeus muscle; Lyons and Slater, 1991) and man (2.4 μm2, vastus lateralis; Slater et al., 1992). Although α1A immunolabeling of the NMJ in rat sections was punctate in nature, this was not the case in human sections in which the α1A antibody seemed to label the entire NMJ area demarcated by DBA. This could reflect localization of α1A in perisynaptic Schwann cells as well as in MN terminals at the human NMJ, compared with a lack of α1A-ir in rat perisynaptic Schwann cells. This observation requires further investigation, which is complicated by difficulties in obtaining suitable material.
In addition to localization at the NMJ, α1A labeling was present in preterminal processes. The disappearance of α1A-ir in preterminal processes after denervation would suggest that a proportion of this subunit may be localized in MN axons. We have shown previously that MN cell bodies in human spinal cord are immunopositive for α1A (Day et al., 1995). The presence of α1A in MN axons would be consistent with transport of this subunit from MN cell bodies to the NMJ.
To date, there are limited anatomical data on the localization of VDCCs at the NMJ. Ousley and Froehner (1994) showed that α1A-ir was present at the NMJ in transverse sections of rat muscle. However, the pre- or postsynaptic nature of α1A immunolabeling was not determined in this study. Sugiura and colleagues (1995), using the Q channel ligand SNX-260, demonstrated labeling of the NMJ in teased muscle fibers from mice. A presynaptic localization for SNX-260-labeled channels was suggested by the finding that labeling was absent in denervated muscles.
Most pharmacological studies suggest that P/Q channels, but not N, L, or T channels, are involved in mammalian neuromuscular transmission. In these studies ωAga-IVA and FTX (P channel ligands; Uchitel et al., 1992; Protti and Uchitel, 1993; Bowersox et al., 1995; Hong and Chang, 1995) and ωCTX-MVIIC (Q channel ligand; Bowersox et al., 1995; Hong and Chang, 1995) blocked neuromuscular transmission, whereas ωCTX-GVIA (N channel ligand; Sano et al., 1987; De Luca et al., 1991), CGP28392 (L channel ligand; Burges and Wray, 1989), and Ni2+ (nonspecific T/R channel blocker; Wray and Porter, 1993; Porter and Wray, 1996) had no effect on neuromuscular transmission in mice. A similar study using human muscle showed that P, but not N or L, channels were involved in human neuromuscular transmission (Protti et al., 1996). The finding in the present study that α1A is localized in MN terminals at the rat NMJ is consistent with a role for P and/or Q channels in mammalian neuromuscular transmission. A few studies have reported that ωCTX-GVIA does block rodent neuromuscular transmission (Hamilton and Smith, 1992; Rossoni et al., 1994), suggesting that N channels may play a role in this process. The results of the present study suggest that a large proportion of α1B immunolabeling is localized in structures other than MN terminals. However, we cannot rule out the possibility that some α1B-ir may be presynaptic, representing low-level expression of N-type channels.
VDCCs in Schwann cells
The data in the present study suggest that both α1Aand α1B may be localized in axon-associated Schwann cells and, further, that α1B also may be found in perisynaptic Schwann cells. These conclusions are based on several observations: (1) in en face views of NMJs in teased fiber preparations, α1A-ir and α1B-ir could be seen in preterminal processes; (2) both α1A and α1B, like S100, seemed to be localized in axon-associated Schwann cells in transverse sections of sciatic nerve; (3) in contrast to α1A and α1E, α1B labeling in en face views of NMJs extended beyond and in between BgTX-labeled acetylcholine receptors on the postsynaptic membrane; and (4) like S100-ir, α1B labeling of preterminal processes and the NMJ persisted in denervated teased muscle fibers. These findings are consistent with localization of α1A and α1B in axon-associated Schwann cells. In addition, we suggest that α1B may be localized in perisynaptic Schwann cells, although the possibility that this subunit also may be localized in the muscle fiber membrane cannot be excluded. However, the contrast in labeling patterns between α1B and α1E antibodies indicates that if α1B is localized in the muscle fiber membrane, it is restricted to the NMJ region and exhibits a distribution that is quite different from that of muscle membrane proteins such as dystrophin and β-spectrin.
Evidence for the presence of VDCCs in mammalian Schwann cells comes from a study demonstrating T- and L-type currents in cultured mouse DRG Schwann cells (Amedee et al., 1991). To date, N- and P/Q-type channels, which are thought to contain α1B and α1Asubunits, respectively, have not been demonstrated in mammalian Schwann cells. VDCCs in Schwann cells may be involved in Schwann cell function and Schwann cell/neuron interaction (Verkhratsky and Kettenmann, 1996). In other types of glia, activation of VDCCs has been shown to stimulate the release of neuroactive substances (Martin, 1992), regulate glial cell activity during seizure (MacVicar et al., 1991), and control myelin oligodendrocyte formation (Kirischuk et al., 1995). Although very little is known about VDCCs in synapse-associated glial cells, it is conceivable that calcium entry into perisynaptic Schwann cells via VDCCs may regulate many cellular reactions, including the release of substances that could influence neuromuscular transmission.
VDCCs in the muscle fiber
Our data suggest that α1E is localized in the postsynaptic muscle fiber membrane because of its similar distribution to the cytoskeletal muscle protein, β-spectrin, and the finding that α1E-ir persisted after denervation. Expression studies have produced conflicting data on the nature of the channel formed by α1E. Some studies suggest that α1E forms a channel that shares properties with LVA currents (Soong et al., 1993;Schneider et al., 1994; Bourinet et al., 1996), which groups it in the same category as the T channel. Other studies suggest that expressed α1E resembles HVA R-type currents (Zhang et al., 1993;Schneider et al., 1994; Williams et al., 1994). Neither T nor R currents have been described in adult mammalian muscle, although the T current is expressed in immature muscle during the first 2 weeks of postnatal development (Beam and Knudson, 1988; Garcia and Beam, 1994). At present, the only VDCC described in adult muscle is the L-type channel that is involved in excitation–contraction coupling and is localized in the T-tubular network within the muscle fiber (Tanabe et al., 1988). The physiological role of α1E-containing VDCCs in the muscle fiber membrane remains to be determined.
α1 subunits at the NMJ and amyotrophic lateral sclerosis
The type of α1 subunit at the NMJ is of interest in ALS, because it has been reported that a large proportion of patients with this disease has circulating autoantibodies directed against VDCCs (Smith et al., 1992), and these autoantibodies bind to the α1 subunit (Kimura et al., 1994). Several lines of evidence suggest that ALS autoantibodies may interact with VDCCs in MN terminals to affect calcium flux and MN function. First, ALS IgG has been shown to increase calcium flux through P-type channels (Llinás et al., 1993) and to stimulate the spontaneous release of acetylcholine at the mouse NMJ (Appel et al., 1991). Second, mice injected with ALS IgG exhibited an increase in synaptic vesicle number and intracellular calcium levels in MN terminals (Engelhardt et al., 1995). Finally, application of ALS IgG to a differentiated hybrid MN cell line caused an increase in calcium flux (Mosier et al., 1995) and was cytotoxic to these cells (Smith et al., 1994). It has been suggested that interaction of ALS antibodies with VDCCs in MN terminals may stimulate processes leading to cell death as well as stimulating acetylcholine release (Appel et al., 1991). Interestingly, studies of muscle biopsies from ALS patients have demonstrated an increase in the probability of quantal store release (Maselli et al., 1993) and increased calcium in MN terminals (Siklos et al., 1996). The precise mechanism by which interaction of ALS autoantibodies with VDCCs at the NMJ may cause MN death is unclear, although it may involve calcium overload in MN terminals, resulting in enhanced glutamate release (Maselli et al., 1993) or altered efficacy of oxidative phosphorylation in the cell body (Siklos et al., 1996). In addition to interacting with P-type channels, ALS IgG will bind to (Smith et al., 1992) and affect calcium flux through L-type channels (Delbono et al., 1991), suggesting that VDCC autoantibodies interact with an epitope or epitopes common to several calcium channels subtypes. Thus, it seems likely that VDCCs in MN terminals, Schwann cells, and muscle will all be targets for circulating ALS autoantibodies.
In conclusion, the present study has shown that α1A, α1B, and α1E are localized differentially at the mammalian NMJ. The major findings of our study are that α1A is localized in MN terminals and that α1A and α1B may be present in Schwann cells, whereas α1E is localized in the muscle fiber membrane. Localization of α1A in MN terminals predicts that the channel incorporating this subunit may be involved in neuromuscular transmission. The role of α1Aand α1B in Schwann cells and α1E in muscle fibers is unclear, although they may play a role in regulating calcium-dependent processes within these cell types. Finally, interaction of ALS autoantibodies with α1A, α1B, and α1E subunits at the NMJ may have important consequences for MN function and the pathogenesis of ALS.
We are grateful to Lilly Research, the Muscular Dystrophy Group of Great Britain, and the Wellcome Trust for financial support. We thank Trevor Booth for technical assistance with confocal microscopy and Carol Young for producing the figures for this paper. We thank Dr. Louise Anderson for the gift of spectrin antibodies.
Correspondence should be addressed to Dr. Nicola Caroline Day, Pharmagene Laboratories, 2A Orchard Road, Royston, Hertfordshire SG8 5HD, UK.