The mutation in the α1A subunit gene of the P/Q-type (Cav2.1) Ca2+ channel present in tottering (tg) mice causes ataxia and motor seizures that resemble absence epilepsy in humans. P/Q-type Ca2+channels are primarily involved in acetylcholine (ACh) release at mammalian neuromuscular junctions. Unmasking of L-type (Cav1.1–1.2) Ca2+ channels occurs in cerebellar Purkinje cells of tg mice. However, whether L-type Ca2+ channels are also up-regulated at neuromuscular junctions of tg mice is unknown. We characterized thoroughly the pharmacological sensitivity of the Ca2+ channels, which control ACh release at adult tg neuromuscular junctions. Block of N- and R-type (Cav2.2–2.3), but not L-type Ca2+ channels, significantly reduced quantal content of end-plate potentials in tg preparations. Neither resting nor KCl-evoked miniature end-plate potential frequency differed significantly between tg and wild type (WT). Immunolabeling of Ca2+ channel subunits α1A, α1B, α1C, and α1E revealed an apparent increase of α1B, and α1E staining, at tg but not WT neuromuscular junctions. This presumably compensates for the deficit of P/Q-type Ca2+channels, which localized presynaptically at WT neuromuscular junctions. No α1C subunits juxtaposed with pre- or postsynaptic markers at either WT or tg neuromuscular junctions. Thus, in adult tg mice, immunocytochemical and electrophysiological data indicate that N- and R-type channels both assume control of ACh release at motor nerve terminals. Recruitment of alternate subtypes of Ca2+ channels to control transmitter release seems to represent a commonly occurring method of neuronal plasticity. However, it is unclear which conditions underlie recruitment of Cav2 as opposed to Cav1-type Ca2+ channels.
Influx of Ca2+ through high-voltage-activated Ca2+ channels triggers neurotransmitter release (Augustine and Charlton, 1986). Different Ca2+ channels are distinguished based on the genes that encode them and their pharmacological and biophysical characteristics.
Most of the subtype-specific attributes of Ca2+ channels are due to the α1 subunit, which makes up the selective pore for Ca2+, contains binding sites for various pharmacological agents, and possesses the gating regions of the channel (Zhang et al., 1993). At least five α1 subunits for neuronal Ca2+ channels are known. The α1A, α1B, and α1E subunits represent the P/Q-type (Cav2.1), N-type (Cav2.2), and R-type (Cav2.3) Ca2+ channels, respectively. The α1C or α1D represents L-type channels (Cav1.2–1.3) (Tsien et al., 1991). The anatomical location, time in development, and age of the animal all affect the expression of specific Ca2+ channel subtypes; multiple phenotypes coexist in the same cell. This redundancy modulates critical functions by allowing the various channel subtypes to act in concert; however, distinct channel subtypes may also be differentially localized and spatially separated in the same cell.
Other factors also affect the expression and localization of Ca2+ channel subtypes. In mammals, ACh release from adult somatic motor nerve terminals is mediated predominantly by P/Q-type Ca2+ channels (Uchitel et al., 1992), whereas in amphibians and birds, it is mediated by N-type Ca2+ channels (Robitaille et al., 1990; De Luca et al., 1991). P- and Q-type Ca2+ channels, originally described in cerebellar Purkinje and granule cells, respectively (Llinás et al., 1989; Randall and Tsien, 1995), are widely distributed and mediate neurotransmitter release at central and peripheral synapses. L-type Ca2+ channels can also contribute to secretory function. These participate in ACh release at neuromuscular junction in certain situations (Suguira and Ko, 1997; Urbano et al., 2003). R- and N-type Ca2+ channels also substitute for P/Q-channels in controlling ACh release (Urbano et al., 2003) at neuromuscular junction. The conditions that determine what subtypes of Ca2+ channel participate in secretory function are, as yet unclear, but they have important implications for synaptic plasticity.
Several natural mutations of P/Q-type Ca2+ channels such as tottering (tg) have been identified in mice. They have been used to study inherited neurological disorders (Burgess and Noebels, 1999; Pietrobon, 2002). The tg mutation encodes a proline-to-leucine amino acid substitution in the S5-S4 linker region of repeat domain II of the α1A subunit (Fletcher et al., 1996). This reduces whole-cell current density and voltage-dependent inactivation during prolonged depolarization of Purkinje cells, without affecting single Ca2+ channel conductance (Wakamori et al., 1998). As a result, P/Q-type Ca2+ channel function is compromised, and neurotransmitter release at hippocampal and cerebellar synapses now depends predominantly on N-type channels (Qian and Noebels, 2000; Zhou et al., 2003). Stereotypic behavior is induced by the L-type agonist BayK 8644 in tg but not wild-type (WT) animals (Campbell and Hess, 1999). L-type channel α1C subunit mRNA was also up-regulated in Purkinje cells (Campbell and Hess, 1999) and basal forebrain neurons (Etheredge et al., 2005) of tg mice, suggesting that the L-type phenotype is newly expressed, or unmasked. In hippocampal slices taken from animals lacking P/Q-type Ca2+ channels (α1A–/–), expression of functional non-P/Q-type channels is augmented. These changes included elevation of cerebellar Purkinje cell L- and N-type current density and reduction of cerebellar granule cell R-type current density (Jun et al., 1999). Hence, expression of Ca2+ channels in a given cell type is not fixed, and up- or down-regulation of other Ca2+ channel subtypes can occur following mutation or ablation of P/Q-type Ca2+ channels.
The tg mutation, despite altering function of the primary subtype of Ca2+channel normally involved in ACh release, does not cause neuromuscular dysfunction, aside from the obvious gait abnormality. Thus, some other Ca2+ channel subtype(s) assumes control of ACh release at tg neuromuscular junctions. Neuromuscular transmission in tg mice was investigated by Plomp et al. (2000). Two alterations were observed: 1) rundown of release during high frequency nerve stimulation was increased; and 2) MEPP frequency was increased in a Ca2+-, Mg2+-, and K+-dependent manner. Recently, Kaja et al. (2006) demonstrated a slight increase in R-type but not N-type channel contribution in 6-week-old tg mice. However, Ca2+ channel expression is age-dependent, so this may not reflect the mature pattern of Ca2+ channel dependence of ACh release in these mice.
We sought to determine the pharmacological sensitivity of neuromuscular transmission of adult tg mice, specifically, whether loss of functional P/Q-type channels unmasks L-type or another subtype of Ca2+ channels, and what Ca2+ channel subtypes sustain release in tg mice. Immunohistochemical and electrophysiological data were combined to provide an in-depth analysis of the pharmacological types of Ca2+ channels at adult tg motor nerve terminals.
Materials and Methods
Drugs and Chemicals. Nimodipine, S-(–)-BayK 8644, and HEPES were purchased from Sigma-Aldrich (St. Louis, MO). μ-Conotoxin GIIIB and ω-conotoxin GVIA were purchased from Bachem California (Torrance, CA). ω-Agatoxin IVA and ω-conotoxin MVIIC were obtained from Alomone Labs (Jerusalem, Israel). SNX 482 was obtained from Peptides International (Louisville, KY). All chemicals obtained were of the highest purity available. Toxins were prepared as stock solutions in distilled water containing 0.01% bovine serum albumin (w/v), stored frozen at –20°C, and used within a 2-week period. Before incubation with any of the toxins, 0.01% bovine serum albumin was also added to the buffered saline solution to prevent nonspecific binding of toxin to the chamber, tubing, and glassware. Nimodipine was prepared as a 10 mM stock solution in 100% ethanol and was kept in a dark bottle at 4°C until use. The final working solution with nimodipine contained only 0.1% ethanol (v/v). All experiments using dihydropyridines were done in a darkened room to avoid breakdown of the photolabile compound.
Antibodies against Ca2+ channel α1 subunits (rabbit anti-Cav2.1, P/Q-; anti-Cav2.2, N-; anti-Cav1.2, L-; and anti-Cav2.3, R-type) were obtained from Alomone Labs. Fluorescein (FITC)-conjugated AffiniPure goat anti-rabbit IgG (heavy + light chains) was purchased from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA), Pacific Blue goat anti-mouse IgG (heavy + light chains), and tetramethylrhodamine α-bungarotoxin were obtained from Invitrogen (Carlsbad, CA). Anti-mouse IgG1 monoclonal anti-syntaxin clone HPC-1 antibody was obtained from Sigma-Aldrich.
Mice. Breeding pairs of heterozygote C57BL/6J-tg mice were obtained from The Jackson Laboratory (Bar Harbor, ME), and colonies were maintained at Michigan State University Laboratory Animal Resources (East Lansing, MI). Litters were genotyped at weaning 3 weeks after birth (Plomp et al., 2000). Although the genotype has been well described, because of differences between our results and those published previously (Plomp et al., 2000; Kaja et al., 2006), we verified the genotype of all of the mice used (data not shown). Homozygote (tg/tg) mice were also identified by their characteristic phenotype consisting of a mild ataxia and occasional attacks of dyskinesia. For all the experiments, we used male mice between 3 and 9 months of age. All experiments were performed in accordance with local university (Michigan State University Laboratory Animal Resources) and national (National Institutes of Health) guidelines and were approved by the University Animal Use and Care Committee.
Electrophysiology. Animals were sacrificed by decapitation following anesthesia with 80% CO2 and 20% O2. The diaphragm muscle with its attached phrenic nerves was then removed and pinned out at resting tension in a Sylgard-coated chamber. For control recordings, the tissue was perfused continuously at a rate of 1 to 5 ml/min with oxygenated (100% O2) physiological saline solution containing 137.5 mM NaCl, 5.0 mM KCl, 1 mM MgCl2, 11 mM d-glucose, and 4 mM HEPES and remained at room temperature. pH was adjusted to 7.4 at room temperature (23–25°C) using NaOH. Muscle action potentials were inhibited by pretreating the tissue with 2.5 to 4 μM μ-conotoxin GIIIB for 15 min. This toxin preferentially blocks muscle Na+ channels (Cruz et al., 1985; Hong and Chang, 1989) and thus suppresses muscle contractility. This technique allowed recording of EPPs from intact myofibers without the complicating effects of depressing ACh release or blocking postjunctional ACh receptors as would occur in high [Mg2+] low [Ca2+]- or d-tubocurarine-treated preparations, respectively. Given that during the electrophysiological recordings the preparations were continuously perfused with physiological saline at a rate of 1 to 5 ml/min, they were retreated with 2.5 to 4 μM μ-conotoxin GIIIB for 15 min, after ∼60 to 90 min, to maintain contractile block.
Involvement of L-type channels in ACh release was determined by testing their sensitivity to the L-type antagonist nimodipine and agonist BayK 8644 (Atchison, 1989). Paired comparisons were made for each preparation between the drug-free treatment (control) BayK 8644 and following application of BayK 8644 in the presence of nimodipine. Values are expressed as the percentage of quantal content (m) from drug-treated preparations to that of preparation before addition of the drug (control). Likewise, sensitivity to ω-conotoxin GVIA (ω-CTx GVIA), SNX 482, ω-agatoxin IVA (ω-Aga IVA), and ω-conotoxin MVIIC (ω-CTx MVIIC) was used to test for the contribution of N-, R-, P-, and Q-type Ca2+ channels, respectively, to ACh release at tg motor nerve terminals. Cd2+ was used to block all Ca2+ channels nonspecifically. The P/Q-, N-, and R-type antagonists are all essentially irreversible, so only one toxin or drug was applied to any preparation. Two protocols were used to apply toxins. Because of the high concentrations used for the ω-CTx MVIIC (5 μM) study and the initial concentration-response relationship for ω-Aga IVA (up to 300 nM), for these studies the hemidiaphragm was preincubated in 5 ml of solution containing the corresponding toxin for 1 h before commencing electrophysiological recordings. The solution was constantly aerated with 100% O2. For all other experiments (final ω-Aga IVA responses, ω-CTx GVIA, SNX 482, nimodipine, BayK 8644, and Cd2+), the hemidiaphragm was superfused with the constantly oxygenated (100% O2) solution in which the specific toxin or compound was diluted. ω-Aga IVA, ω-CTx GVIA, and SNX 482 were used at 100 nM, 3 μM, and 1 μM, respectively, and diluted in 10 ml of physiological saline solution. Nimodipine, BayK 8644, and Cd2+ were used in concentrations of 10, 1, and 10 μM, respectively, and diluted in 20 ml of physiological saline solution. These concentrations were chosen based on literature determining their effectiveness at murine neuromuscular junctions (cf. Atchison, 1989; Xu et al., 1998; Santafe et al., 2000; Urbano et al., 2001, 2003; Flink and Atchison, 2002; Kaja et al., 2006).
MEPPs and EPPs were recorded using intracellular glass microelectrodes (1.0 mm o.d.; WPI, Sarasota, FL) having a resistance of 5 to 15 MΩ when filled with 3 M KCl and localized at the end-plate. The phrenic nerve was stimulated at 0.5 Hz with constant current, using a duration of 50 μs, by means of a suction electrode attached to a stimulus isolation unit (Grass SIU; Grass Instruments, Quincy, MA) and stimulator (Grass S48). Signals were amplified using a WPI 721 amplifier, digitalized using a PC-type computer and Axoscope 9.0 (Axon Instruments, Foster City, CA) software, and analyzed using MiniAnalysis 6.0 software (Synaptosoft, Decatur, GA).
Control recordings were made from neuromuscular junction preparations isolated from tg/tg, C57BL/6J-tg, and WT mice without any treatment, and following incubation with the individual chemical or toxin for 1 h. For each preparation, recordings from five to 10 end-plates were sampled and averaged to determine the mean amplitude of the EPPs and MEPPs before and after addition of any pharmacological treatment, yielding an n value of 1. Each end-plate was sampled for over 5 min, and the last end-plate recorded from, before any treatment, was the first end-plate recorded from after treatment. Amplitudes of MEPPs and EPPs were normalized to a membrane potential of –75 mV using the formula Vc = [Vo × (–75)]/E, where Vc is the corrected EPP/MEPP amplitude, Vo is the observed EPP/MEPP amplitude, and E is the resting membrane potential. Recordings were rejected if the 10 to 90% rise time was greater than 1.5 ms or if the membrane potential was more depolarized than –55 mV. The normalized EPPs were corrected for nonlinear summation (McLachlan and Martin, 1981) using the formula Vcorr = V[(1–0.8V) E–1]–1, where V is the uncorrected EPP amplitude, E is the resting membrane potential, and Vcorr is the corrected EPP amplitude. The m value was calculated using the ratio of the mean amplitude of the corrected EPPs and MEPPs. The effects of nerve terminal depolarization, induced by raising the [K+]e from 2.5 to 5, 10, and 20 mM, on MEPP frequency were also measured. After a brief period to permit the bath fluid to reach the new [K+]e (usually 2 min), at least 5 min of MEPPs were recorded at each [K+]e. When extracellular KCl concentration was raised, equimolar reductions in extracellular NaCl concentration were made.
Separate preparations were used for each unique experiment conducted. For pharmacological studies, each animal served as its own control. Experiments were replicated in at least five animals. The number of animals used in any given experiment is indicated in the respective figure legend or table. Statistical significance between the various treatment groups was analyzed using a one-way analysis of variance (Prism; GraphPad Software Inc., San Diego, CA). Post hoc differences among sample means were analyzed using Tukey's test. For all experiments, statistical significance was set at p < 0.05. Predrug values for all the animals of a given genotype (WT, C57BL/6 J-tg, tg/tg) were pooled, as the between-animal variance was homogeneous for that group (Prism statistical software; GraphPad Software Inc.).
Immunohistochemistry. Localization of the different Ca2+ channel α1 subunits at tg and WT mouse motor nerve terminals was compared using fluorescence immunohistochemistry in the extensor digitorum longus (EDL) and triangularis sterni (TS) muscles from animals whose diaphragm was used for pharmacological studies. The EDL is a homogeneous fast twitch type muscle; thus, concerns associated with myofiber-type-dependent differences in structure or function of the neuromuscular junctions were minimized (Gertler and Robbins, 1978; Prakash et al., 1996). The thinness of the TS muscle allowed us to label neuromuscular junction structures without using cryosectioning techniques that involved prolonged exposure of the preparation to chemicals such as sucrose, yet permitted high-quality images to be obtained. The fiber type of TS muscle is not known. Qualitatively, results obtained from EDL and TS preparation were consistent, showing the same characteristics of distribution among the different α1 subunits studied at neuromuscular junctions. No attempt was made to quantify potential differences in staining between the two muscle types. Both muscles were pinned out and lightly fixed for 30 min at room temperature in 4% (w/v) paraformaldehyde in 0.1 M phosphate-buffered saline (PBS; composition 137 mM NaCl, 2.7 mM KCl, 1.4 mM NaH2PO4, and 4.3 mM Na2HPO4, pH 7.4). The preparation was then washed in PBS for 1 min and treated with 0.1% (w/v) Triton X-100 in PBS for 30 min. EDL tissue was subsequently washed for 15 min and cryoprotected in 20 and 30% (w/v) sucrose each for 24 h. After 48 h, blocks of tissue were embedded in optimal cutting temperature compound (Tissue-Tek, Tokyo, Japan) in a plastic mold on dry ice and stored at –80°C until use. Longitudinal and cross-sections (30 μm thick) were cut on a cryostat (model 5040, Bright Microtome; Bright Instrument Company Ltd., Huntingdon, Cambridge, UK) and mounted onto gelatin-coated slides. Because the TS muscle is extremely thin, no cryosectioning was needed. Preparations were double- or triple-labeled using α-bungarotoxin as a marker for the postsynaptic ACh receptors (AChR) at the motor end-plate, commercially available monoclonal anti-syntaxin antibody, labeled subsequently with Pacific Blue-tagged secondary antibody as a presynaptic marker and antibodies to the various α1 subunits of the Ca2+ channels, labeled subsequently with FITC-tagged secondary antibody. Preliminary experiments with WT preparations demonstrated the expected juxta-position of presynaptic syntaxin with postsynaptic α-bungarotoxin (see Fig. 5). Thus, these experiments were replicated in only a limited number of preparations because the confocal microscope used for fluorescence quantitation has only two lasers and hence could not support simultaneous collection of data at three wavelengths. The preparations were washed in PBS for 15 min and then incubated at 4°C overnight (∼15 h) with the subunit-specific primary antibody and anti-syntaxin antibody. After washing for 1 h with PBS, sections were incubated for 1 h in fluorescently labeled secondary antibodies, together with rhodamine labeled α-bungarotoxin. After several washes, the preparation was mounted with anti-fading fluorescent mounting medium (Vectashield Hard Set; Vector Laboratories, Burlingame, CA). The different color fluorescent signals were collected and integrated using a Leica TCS SL laser scanning confocal microscope (Leica Microsystems, Exton, PA) to determine the spatial localization and distribution of the α1 subunits relative to the motor end-plate. This microscope allows simultaneous scanning of FITC and tetramethylrhodamine B isothiocyanate, which were excited at 510 and 580 nm, respectively, using an argon/HeNe-G laser. All pictures of the immunofluorescent signals were taken with the same confocal configuration settings (laser intensity, time exposure resolution, and magnification) for each antibody tested. Integrated signals for the different fluorophores were used to generate composite images for determining spatial localization and distribution of fluorescence. In addition, to determine the spatial localization and distribution of the α1 subunits and syntaxin relative to the motor end-plate, some preparations were also viewed on a Nikon Elipse 2000-U Diaphot-TMD microscope (Nikon, Melville, NY) with a Hamamatsu Orca 285 charge-coupled device camera (Bridgewater, NJ), and images were acquired using MetaImaging software suite (Molecular Devices, Downingtown, PA). This system allows simultaneous composite viewing of sequentially acquired images of FITC-, tetramethylrhodamine B isothiocyanate-, and Pacific Blue-labeled samples. For each TS preparation, three to five surface nerve terminals were selected for quantitation of relative fluorescence levels. Because all of the pictures were taken using the same confocal configuration settings, we used ImageJ software (National Institutes of Health, Bethesda, MD) to calculate and average the fluorescence levels from total pixels corresponding to the green dye in each picture. Averages of the mean values of fluorescence obtained from all the individual nerve terminals sampled were calculated for each specific α1 subunit studied. Averaged values of fluorescence were compared between the tg and the WT preparations using the values obtained from the WT preparation as control values. Subsequently, the percentage of juxtaposition of the green and the red dye was calculated by dividing the surface of each picture taken into an area of 5 × 5 squares for a total of 25 inner squares. Each inner square in which the green and the red dyes were juxtaposed was taken as 4% juxtaposition.
Table 1 compares the amplitudes of EPPs and MEPPs, MEPP frequency, and m for tg/tg and WT mice. When supramaximal electrical stimuli were delivered to the phrenic nerve at 0.5 Hz, the mean amplitude of EPPs and m in tg/tg preparations did not differ significantly (p > 0.05) from the value obtained in the WT preparations. Likewise, the mean amplitude and frequency of spontaneously occurring MEPPs did not differ between WT and tg/tg preparations. Thus, as suggested by their gross phenotype, neuromuscular transmission in tg/tg mice is not significantly compromised compared with that of WT. An identical set of experiments were performed in heterozygote C57BL/6J-tg preparations, and the results did not differ significantly from those obtained from WT preparations (data not shown).
The P/Q-type Ca2+ channel antagonist ω-Aga IVA reduced m of EPPs in WT, C57BL/6 J-tg, and tg/tg preparations; however, this reduction was statistically significant in only the C57BL/6 J-tg and WT preparations (p < 0l.05) (Fig. 1A). ω-Aga IVA (100 nM) reduced m in C57BL/6 J-tg by 60% (Fig. 1A) compared with the ∼80% reduction in the WT preparation; however, the difference between these two groups was not significant (p > 0.05). Thus, the heterozygotes (C57BL/6 J-tg) neuromuscular junctions apparently behave as a normal WT. Perhaps the expression of the normal α1A subunit protein of the P/Q-type Ca2+ channels is not sufficiently reduced in the C57BL/6 J-tg (heterozygotes) to cause alteration in the normal function of the P/Q-type channels. Consequently, there may be no need for compensation of function through other channel phenotypes.
The pharmacological sensitivity of the heterozygotes to other Ca2+ channel antagonists was also examined. These results are shown for all drugs tested in Fig. 1A. For none of these drugs did the C57BL/6J-tg heterozygote response differ significantly from that of WT (p > 0.05). Thus, for ease of comparison, the following focus is on the comparison of the tg/tg (tg) and WT groups.
L-type Ca2+Channels Do Not Contribute to Release of ACh attgNeuromuscular Junction. Nimodipine (an L-type antagonist) and BayK 8644 (an L-type agonist) were used to examine whether L-type Ca2+ channels contribute to release of ACh at tg motor nerve terminals. As shown in Fig. 1B, 10 μM nimodipine reduced m by 16.3 ± 3.8% of the nimodipine-free treatment value (control) in WT preparations and by only 4.7 ± 2.9% in tg preparations. Neither of these effects was statistically significant (p > 0.05). Subsequently the preparation was washed with a nimodipine-free solution for 5 min or until EPP amplitude returned to baseline. Subsequent addition of a solution containing BayK 8644 caused a 12.9 ± 3.8% increase in m at WT preparations, but it essentially had no effect at tg preparations. This slight increase in m induced by BayK 8644 in the WT could be prevented by pretreatment with nimodipine (Fig. 1B).
Neuromuscular Transmission intgMice Is Poorly Susceptible to the P/Q-Type Antagonists ω-Aga IVA and ω-CTx MVIIC. The effects of P/Q-type antagonist ω-Aga IVA (100 nM) on EPP amplitude of tg and WT genotypes are shown for representative examples in Figs. 2 and 3, A through D. Comparing the effects of the two toxins on m revealed that 100 nM ω-Aga IVA decreased m by ∼22% in tg and ∼82.4% in WT preparations (Fig. 2). Higher concentrations of ω-Aga IVA (300 nM) had no additional effect in m in tg preparations (Fig. 2). Likewise, ω-CTx MVIIC reduced m by 92.4% in WT and 41.3% at tg neuromuscular junctions (Fig. 3A). Thus, although the two toxins affected each of the two genotypes, 1) a greater effect was seen with the less specific ω-CTx MVIIC than with ω-Aga IVA, and 2) the effect of each toxin was greater in the WT than the tg neuromuscular junctions.
N-and R-Type Ca2+Channels Contribute to the Majority of Nerve-Evoked ACh Release attgNeuromuscular Junctions. The contribution of N- and R-type Ca2+ channels to ACh release from tg mice was tested using 3 μM ω-CTx GVIA and 1 μM SNX 482, respectively. Figure 4, A–D, shows representative tracings of the comparative effects of ω-CTx GVIA and SNX 482 applied alone, or in combination, on EPP amplitude in WT and tg mice. Neither toxin alone nor in combination was effective in WT preparations. Conversely, both ω-CTx GVIA and SNX 482 significantly reduced EPP amplitude in the tg preparations (p < 0.05). The combination of ω-CTx GVIA and SNX 482 caused a further and significant reduction in EPP amplitude of tg mice. However, even the combined presence of the two toxins did not abolish EPPs in tg mice. Although each toxin caused significant reduction in m compared with pretreatment control for tg preparations, comparing across genotypes, the effects in tg were only statistically significant from WT for SNX 482 and the combined treatment of SNX 482 and ω-CTx GVIA (Fig. 4B). Thus, Ca2+ channels sensitive to SNX 482 and ω-CTx GVIA play a dominant role in neurotransmitter release at the tg motor nerve terminals.
For comparative purposes, 10 μM Cd2+ was applied to some preparations to block all Ca2+ channels nonspecifically. This concentration of Cd2+ completely blocked ACh release in tg as well as WT neuromuscular junctions (data not shown). Thus, even though ACh release was not completely blocked by N- or R-type Ca2+ channel antagonists in the tg neuromuscular junction, it was totally sensitive to Cd2+ and in the same concentration range as was WT.
K+ Induced ACh Release Does Not Differ between tg and WT Mice. Voltage-dependent inactivation of Purkinje cell P/Q-type channels in tg mice is reduced (Wakamori et al., 1998) during prolonged depolarization. As such, we tested whether differences could be detected between tg and WT in MEPP frequency during prolonged KCl induced depolarization. Asynchronous evoked release of ACh was measured at different [K+]e as increased MEPP frequency. Increasing [K+]e incrementally from 2.5 to 20 mM increased MEPP frequency to equivalent levels in both preparations (p > 0.05; data not shown)
Differential Localization of Voltage-Dependent Ca2+Channel α1 Subunits at tg and WT Motor Nerve Terminals. The distribution of α1A, α1B, α1C, and α1E Ca2+ channel subunits at tg neuromuscular junction, was examined using fluorescence immunohistochemistry in sections of mouse EDL and TS muscles. The relative localization of these subunits with respect to the motor end-plates was assessed by comparing staining for subunit-specific antibodies to Ca2+ channels with that of fluorescent α-bungarotoxin to label postsynaptic ACh receptors and anti-syntaxin antibody to label the presynaptic membrane.
As shown for the representative images in Fig. 5, for both WT and tg mice intense staining occurred with antibody against syntaxin (blue) and α-bungarotoxin (red). The two stains were highly juxtaposed. Although not evident at the level of resolution of these images (data not shown), anti-syntaxin staining frequently exhibited a punctate pattern, as described by Santafe et al. (2005).
Figure 6 demonstrates the punctate distribution of α1A subunit staining seen in WT preparations. As depicted in the representative images, the α1A subunit staining was highly juxtaposed with that of the α-bungarotoxin labeled end-plate. The panels on the right of each image depict quantitation of the extent of staining and its superimposition. The distribution and juxtaposition of the two stains is consistent with results described by Day et al. (1997). However, this pattern of distribution was not observed in the tg preparations. There was little staining of tg preparations with anti-α1A antibody, and what staining occurred did not juxtapose with the clearly demarcated end-plate.
The pattern of immunofluorescent staining for α1B, α1C, and α1E was very different from that of α1A. In the WT preparations, α1B and α1E labeling seemed to be randomly distributed, if it occurred, and it definitely did not juxtapose with α-bungarotoxin labeling (data not shown). However, in the tg group, α1B and α1E subunit labeling demonstrated expression of both subunits, with some of the staining distributed along the vicinity of the end-plates, and some degree of juxtaposition with the α-bungarotoxin label (Fig. 7). Immunofluorescence distribution of the α1C subunit seemed to run along the muscle fiber in the WT preparations (data not shown). No α1C staining was observed at any of the tg motor nerve terminals.
The relative amount of fluorescence for each subunit is compared quantitatively across all preparations for the two genotypes in Fig. 8. There was a significant increase in the amount of fluorescence corresponding to α1B (124 ± 6.4%) and especially to the α1E (2283 ± 10.4%) subunits in tg preparations compared with WT. As expected, the level of staining of α1A was significantly higher in WT than tg preparations. There was no difference quantitatively between the two genotypes in staining for α1C, which was negligible in each case.
As a further investigation of the descriptive results obtained in the initial immunohistochemical studies, we quantitated the extent of juxtaposition of α1 subunit immunofluorescence with that of α-bungarotoxin (Fig. 9). Confocal microscopic imaging demonstrated that 69% of the α1A labeling was juxtaposed with α-bungarotoxin labeling in the WT preparation, but only ∼19% in the tg preparation. Surprisingly, the percentages of juxtaposition for α1B and α1E subunit labeling in tg preparations were only 12.5 and 15% respectively; there was no juxtaposition of either of these subunits with α-bungarotoxin in the WT preparations (Fig. 9). As described above, α1C subunit seemed to run along the muscle fiber in the WT preparations with a percentage of juxtaposition of 16.66% within the end-plates. No α1C subunit staining was juxtaposed with α-bungarotoxin in the tg group.
P/Q-type channels are the normal primary regulators of nerve-evoked ACh release at mammalian neuromuscular junctions (Uchitel et al., 1992; Sugiura et al., 1995), so one might predict that α1A subunit mutations in these channels would disrupt murine junctional transmission. However, tg mice exhibit no obvious neuromuscular impairment; hence, P/Q-type channel function is somehow compensated.
This study characterized thoroughly the plasticity of ACh release at motor end-plates in these animals. Effects of the tg mutation on neuromuscular transmission have not been extensively studied. Kaja et al. (2006) suggested a possible compensation of non-Cav2.1 channels to evoked ACh release at 6-week-old tg motor nerve terminals. ω-Aga IVA reduced release by ∼75% in tg preparations, as opposed to ∼95% in WT. In addition, ∼15% sensitivity to SNX 482 and no sensitivity to ω-CTx GVIA occurred in tg preparations. Our study corroborates some of these findings, but it differs in several ways. Moreover, it extends some findings considerably.
Our results are consistent with the following conclusions: 1) P/Q-type channels contribute poorly to ACh release at adult tg neuromuscular junction, albeit heterozygote neuromuscular transmission is essentially similar to WT. 2) L-type channels do not contribute to ACh release in tg mice. 3) N- and R-type channels assume ACh release control in adult tg motor nerve terminals; however, some release remains insensitive to all toxins but equally sensitive to Cd2+. 4) At low rate (0.5 Hz) of stimulation, m does not differ between WT and tg. 5) No apparent differences occur in spontaneous release between adult tg and WT preparations.
The differential effect of ω-Aga IVA on WT and tg heterozygotes was not significant, although it displayed an interesting trend: ω-Aga IVA-sensitive channels contribute ∼82% to ACh release at WT, ∼60% at C57BL/6 J-tg, but only ∼22% at tg neuromuscular junctions. Thus, fundamentally, the heterozygote neuromuscular junction behaves similarly to WT. Adult tg mice Ca2+ channels showed higher sensitivity to the less specific blocker ω-CTx MVIIC than to ω-Aga IVA. Possible explanations for this result include 1) the Q-type splice variant is more preponderant functionally than is the P-type in tg animals. 2) The ω-CTx MVIIC binding site is more accessible in the tg mice than is the ω-Aga IVA site. 3) The selectivity of ω-CTx MVIIC, especially at concentrations as high as 5 μM, is markedly reduced and it also blocks N-type channels (McDonough et al., 2002). Greater contribution of N-type channels at tg motor axon terminals could be reflected in the greater reduction of m by ω-CTx MVIIC. This finding is consistent with the subsequent sensitivity of tg neuromuscular transmission to ω-CTx GVIA. In either case, function of P/Q-type channels is markedly reduced in adult tg mice. In this regard, our results differ significantly from those of Kaja et al. (2006), who found a higher percentage of sensitivity to ω-Aga IVA in 6-week-old tg mice. Sensitivity to ω-CTx MVIIC was not reported. As described below, difference in age of the animals used may contribute to the differences in sensitivity between these studies.
A major question we sought to answer was whether L-type channel function was evident at tg neuromuscular junctions. L-type channels are up-regulated in cerebellum (Campbell and Hess, 1999) and basal forebrain (Etheredge et al., 2005) of tg mice; however, we found no contribution of L-type channels to ACh release at adult tg neuromuscular junctions. Up-regulation of L-type channels is a well reported form of plasticity at murine neuromuscular junction. It occurs in adult mice treated chronically with Lambert-Eaton myasthenic syndrome plasma (Smith et al., 1995; Xu et al., 1998; Flink and Atchison, 2002), reinnervating motor endplates following acute nerve damage (Katz et al., 1996) or botulinum toxin-poisoned nerve terminals (Santafe et al., 2000). L-type channels are also involved in ACh release in neonatal rats (Sugiura and Ko, 1997). Although it is possible that L-type channel function at tg neuromuscular junctions is masked by the presence of other channel types such as Ca2+-dependent K+ channels (Flink and Atchison, 2003), immunolabeling does not reveal α1C subunit staining at tg neuromuscular junctions. Thus, L-type channel compensation is not generalized either to mutation or to ablation of P/Q-type channels.
ACh release in tg motor nerve terminals is sensitive to the N- and R-type channel antagonists ω-CTx GVIA and SNX 482, respectively. The immunolabeling demonstrates increased fluorescence for α1B, but especially for α1E in tg mice. Pharmacological studies also demonstrated a greater contribution of R-as opposed to of N-type channels to release. Hence, both types of data suggest a relocation-recruitment, or up-regulation of α1B and α1E subunits at tg neuromuscular junctions. Presumably, this occurs to compensate for the loss of function of α1A subunits. These channel subtypes apparently act synergistically in tg mice. This response is identical to that seen at neonatal mouse neuromuscular junctions of P/Q-type nullzygotes (Urbano et al., 2003). In central nervous system nerve terminals in tg mice, N-type channels apparently are solely responsible for (Qian and Noebels, 2000) or predominate in controlling glutamate release (Leenders et al., 2002; Zhou et al., 2003). Neither of these studies tested responsiveness to SNX 482, and in neither case did ω-CTx GVIA abolish release, so it is possible that an R-type component contributed to glutamate release in their experiments. We observed no significant changes at WT neuromuscular junction with either of these pharmacological treatments, and heterozygotes did not express sensitivity to ω-CTx GVIA (SNX 482 was not tested). Consequently, N- and R-type channels primarily mediate ACh release at tg neuromuscular junctions. Why multiple subtypes of Ca2+ channels are necessary to replace a process that is normally almost exclusively dependent on a single subtype of Ca2+ channel is unclear.
The finding of contributions of both N- and R-type channels to ACh release differs from those of Kaja et al. (2006). They saw no effect of ω-CTx GVIA and only a modest reduction of m (15%) with SNX 482l. The percentage reduction of ACh release caused by ω-CTx GVIA in our study was virtually identical to that in cerebellar synaptosomes (Zhou et al., 2003). The Kaja et al. (2006) study did not include immunohistochemical data, so it is unknown whether N-type channels were present at tg motor nerve endings, but did not contribute to ACh release. Differences between our two studies are most likely due to age-related factors. Several studies have demonstrated developmental changes in Ca2+ channel expression (Gray et al., 1992; Rosato Siri and Uchitel, 1999). Full expression of α1B and α1E subunits may not occur until later than 6 weeks postnatal in tg mice. Thus, at the comparatively young age of mice in which Kaja et al. (2006) examined neuromuscular function, there may have been a greater dependence on P/Q-type (hence, the larger percentage of sensitivity of their animals to ω-Aga IVA), and a lesser contribution of N- or R-type channels (slight sensitivity to SNX 482 and no sensitivity to ω-CTx GVIA) than in 3- to 9-month-old animals we used.
Although our results support up-regulation of R-type Ca2+ channels as a compensatory mechanism for the functional loss of P/Q-type channels at tg neuromuscular junctions, reports suggest that SNX 482 may not only block R-type but also P/Q-(Arroyo et al., 2003) and N-type channels (Neelands et al., 2000). However, our immunohistochemical studies correlate with the pharmacological studies in that α1B and α1E staining is clearly evident at tg neuromuscular junctions, lending credence to the notion that R-type channels are present at tg terminals and play an important role in ACh release. Similar findings are described previously (Pagani et al., 2004) at α1A knockout neuromuscular junctions. Neither our pharmacological nor immunohistochemical results suggested presence of R-type channels at WT neuromuscular junction.
Considerable phenotypic difference exists between the α1A knockout and tg mouse, both of which have impaired P/Q-type channel function, and show compensatory increases in N- and R-type channels. In α1A knockout mice, deletion of P/Q-type channels is lethal (Fletcher et al., 2001), despite compensatory plasticity changes in Ca2+ channel phenotype (Jun et al., 1999). Conversely, in tg mice, although the single amino acid substitution results in largely dysfunctional P/Q-type channels, some sensitivity to ω-Aga IVA remains even into adulthood. Nonetheless, other non-P/Q-type Ca2+ channels are recruited to control ACh release and this release is sufficient to support normal neuromuscular function to adulthood. Perhaps the presence of a small fraction of P/Q-type channels in the tg mice suffices to permit function until sufficient recruitment of non-P/Q-type channels occurs. If this is the case, the fact that in tg mice α1A protein is still expressed, albeit impaired functionally, may allow normal organization of the release apparatus and hence survival.
Despite the importance of SNX 482-sensitive ACh release to tg neuromuscular transmission, an aspect of the immunohistochemical data is puzzling. Although staining for α1E was greater than that of α1B in tg mice (Fig. 8), the percentage of juxtaposition of α1E staining with α-bungarotoxin was low, in fact, no higher than that of α1A or α1B (Fig. 9). This implies that the R-type channels may not be closely localized to active zones. Urbano et al. (2003) suggest that there is a “preferred order” of insertion of high-voltage-activated Ca2+ channels into nerve terminal membrane in the active zones. R-type channels are suggested to be “preferred” over N-type. Certainly our data indicate more control by R- than N-type channels of release in tg mice, but one would have expected a greater level of juxtaposition of the α1E staining with the α-bungarotoxin than was observed. More extensive analyses of this will be needed to resolve this conundrum.
Previous reports (Plomp et al., 2000; Kaja et al., 2006) indicated that resting MEPP frequency and response to depolarization with 10 mM KCl was increased in tg mice. However, we found no increase in resting MEPP frequency in either tg or C57BL6/tg mice, and the response to sustained depolarizations at [KCl] up to 20 mM was equivalent in tg and WT mice. The basis for this difference in result may be methodological. Although no details of the MEPP recordings were provided in Plomp et al. (2000), Kaja et al. (2006) report measuring at least 30 MEPPs at each neuromuscular junction. At the reported frequencies in their article (2.04 MEPPs/s in tg; 0.96 MEPPs/s in WT), this implies recordings of as little as 15 to 30 s. We sampled over a much longer interval. At a frequency of 1 Hz, for 5 min (our minimal sampling duration), we would sample ∼300 MEPPs. Short time intervals could easily magnify apparent differences in MEPP frequency, which would “average out” over longer intervals. This is especially true if “clusters” of MEPPs occurred (Fatt and Katz, 1952; Kriebel and Stolper, 1975; Vautrin and Kriebel, 1991) for whatever reason during a brief recording episode. Alternatively, perhaps younger mice exhibit a greater difference in MEPP frequency between the two genotypes. This would need to be examined more rigorously, however, to substantiate.
In conclusion, the type of channel that can control ACh release at mammalian neuromuscular junction is not fixed. R- and N-type Ca2+ channels contribute to ACh release at mammalian neuromuscular junctions under specific conditions such as following P/Q-type channel ablation in α1A knockout animals (Jun et al., 1999; Urbano et al., 2003) or P/Q-type channel mutation. Therefore, recruitment of alternate subtypes of Ca2+ channels to overcome deficiency in the normal complement seems to represent a commonly occurring method of neuronal plasticity. However, the identity of the compensatory type of Ca2+ channel(s) is not constant. Furthermore, even in the same genotype, plasticity varies from central nervous system to peripheral synapses. In each case, the apparent goal is to preserve synaptic function. However, different α1 subunits have distinct biophysical as well as pharmacological properties, so substitution of one phenotype of Ca2+ channel with another is unlikely to be “seamless”. The “rules” by which this up-regulation occurs are not yet clear, but they could play an important role in determining compensation at synapses in which Ca2+ channel function is altered.
The excellent word processing assistance of Erin E. Koglin and Tara S. Oeschger is appreciated.
This work was supported by National Institutes of Health Grant R01 NS051833 (to W.D.A.), in part by an Osserman/Sosin/McClure Fellowship (to N.E.P.) from the Myasthenia Gravis Association, and by a grant from the Muscular Dystrophy Association of America. The Leica confocal microscope was provided by funding from the Life Sciences Corridor Program of the Michigan Economic Development Corporation. Preliminary results from this research were presented as abstracts. Pardo NE and Atchison WD (2003) Pharmacological analysis of acetylcholine release from motor nerve terminals of tottering mice. Program No. 686.8, in 2003 Abstract Viewer/Itinerary Planner, Society for Neuroscience, Washington, DC; and Pardo NE and Atchison WD (2004) Acetylcholine release from motor nerve terminals in tottering mice is sensitive to SNX 482 and ω-conotoxin GVIA, Program No. 736.1, in 2004 Abstract Viewer/Itinerary Planner, Society for Neuroscience, Washington, DC.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
ABBREVIATIONS: ACh, acetylcholine; tg, tottering; WT, wild-type; MEPP, miniature end-plate potential; EPP, end-plate potential; BayK 8644, S-(–)-1,4-dihydro-2,6-dimethyl-5-nitro-4-(2-[trifluoromethyl]phenyl)-3-pyridine carboxylic acid methyl ester; ω-Aga IVA, ω-agatoxin IVA; ω-CTx MVIIC, ω-conotoxin MVIIC; ω-CTx GVIA, ω-conotoxin GVIA; FITC, fluorescein isothiocyanate; m, quantal content; [K+]e, extracellular K+ concentration; EDL, extensor digitoris longus; TS, triangularis sterni; PBS, phosphate-buffered saline; AChR, acetylcholine receptor; 3D, three-dimensional; NMJ, neuromuscular junction; OCT, optimal cutting temperature.
- Received June 1, 2006.
- Accepted September 13, 2006.
- The American Society for Pharmacology and Experimental Therapeutics