Dual Action of Hydrogen Peroxide on Synaptic Transmission at the Frog Neuromuscular Junction

Abstract

There is evidence that reactive oxygen species (ROS) are produced and released during neuromuscular activity, but their role in synaptic transmission is not known. Using a two-electrode voltage-clamp technique, at frog neuromuscular junctions, the action H₂O₂ on end-plate currents (EPC) was studied to determine the targets for this membrane-permeable ROS. In curarized or cut muscles, micromolar concentrations of H₂O₂ increased the amplitude of EPCs. Higher (> 30 μm) doses inhibited EPCs and prolonged current decay. These effects were presynaptic since H₂O₂ did not change the amplitude or duration of miniature EPCs (although it reduced the rate of spontaneous release at high concentrations). Quantal analysis and deconvolution methods showed that facilitation of EPCs was due to increased quantal release, while depression was accompanied by temporal dispersion of evoked release. Extracellular recordings revealed prolonged presynaptic Ca²⁺ entry in the presence of high H₂O₂. Both low and high H₂O₂ increased presynaptic potentiation during high-frequency stimulation. Pro-oxidant Fe²⁺ did not affect facilitation by low doses of H₂O₂ but augmented the inhibition of EPCs by high H₂O_2, indicating involvement of hydroxyl radicals. High Mg²⁺ and the ROS scavenger N-acetylcysteine eliminated both the facilitatory and depressant effects of H₂O₂. The facilitatory effect of H₂O₂ was prevented by protein kinase C (PKC) inhibitors and 4β-phorbol 12-myristate, 13-acetate (PMA), an activator of PKC. PKC inhibitors but not PMA also abolished the depressant effect of H₂O₂. Our data suggest complex presynaptic actions of H₂O₂, which could serve as a fast feedback modulator of intense neuromuscular transmission.

Reactive oxygen species (ROS) such as superoxide anion (O₂⁻), hydroxyl radical (OH•), and hydrogen peroxide (H₂O₂) are continuously produced by mitochondrial respiration, the membrane-associated NADPH oxidase complex (Vignais, 2002), or by the action of hormones, neurotransmitters and neurotrophins (Finkel, 1998; Oh et al. 2000).

H₂O₂ is of particular interest among other ROS because: (a) H₂O₂ is the most stable and long-lived ROS (Kress et al. 1995; Li & Jackson, 2002); (b) being a mild oxidant, H₂O₂ application for < 30 min does not produce non-specific peroxidation of membrane lipids (Tretter & Adam-Vizi, 1996); (c) H₂O₂ passes through cell membranes (Weiss, 1986; Waldron & Rozengurt, 2000); (d) H₂O₂ is more stable in extracellular than in intracellular space (Murrant & Reid, 2001) and it might serve, therefore, not only as an intracellular but also as an intercellular messenger (Oh et al. 2000; Servitja et al. 2000).

The mechanism of H₂O₂ action on target cells is not known. Since H₂O₂ is able to permeate membranes and can modify cysteine, methionine (Knapp & Klann, 2000) and tyrosine groups (Konishi et al. 1997) of proteins, it could modify intracellular enzymes in host or neighbouring cells. For example, PKC, a key enzyme participating in synaptic plasticity and transmitter release, is an important intracellular target for H₂O₂ (Konishi et al. 1997; Suzuki et al. 1997; Servitja et al. 2000).

The question arises of what role, if any, ROS play in synaptic transmission, since skeletal muscle fibres produce and release superoxide and H₂O₂ (O'Neill et al. 1996; Hasegawa et al. 1997; Supinski et al. 1999; Reid & Durham, 2002). Recent experiments suggest that ROS influence muscle fatigue (Hasegawa et al. 1997; Kolbeck et al. 1997; Reid & Durham, 2002). When applied to skeletal muscle in millimolar concentrations, H₂O₂ can lead to an increase or a decrease in muscle tension (Oba et al. 1996; Plant et al. 2002; Lamb & Posterino, 2003). Interestingly, Crosland (1995) has shown that exogenous ROS selectively impair nerve-mediated muscle contraction (but not that induced by direct muscle stimulation), suggesting vulnerability of synaptic transmission to ROS. In the hippocampus, ROS causes a PKC-dependent increase (Knapp & Klann, 2002) or depression of evoked excitatory postsynaptic potentials (Auerbach & Segal, 1997). In striatum, H₂O₂ induced an inhibitory action on dopamine release (Chen et al. 2001). However, no direct evidence has been reported on the action of ROS on signal transmission from motoneurons to skeletal muscle.

In the current study we analysed the effect of a wide range of H₂O₂ concentrations on pre- and postsynaptic mechanisms at the frog neuromuscular junction. We found that low (3 μM) concentrations of H₂O₂ produced facilitation while high (>30 μM) concentrations led to depression of cholinergic transmission. Our study provides direct evidence for the action of H₂O₂ on presynaptic terminals at the neuromuscular junction and suggests that ROS could serve as endogenous modulators of neurotransmission.

METHODS

Preparation and solutions

Experiments were carried out on frog sartorius muscle preparations in vitro at room temperature (for details see Giniatullin et al. 1997). Briefly, muscles were dissected from tricaine-anaesthetized frogs in accordance with the European Communities Council Directive (24th November 1986; 86/609/EEC). To prevent muscle contractions we used the following preparations: (a) transverse cutting (‘cut preparation’) of muscle fibres (Barstad & Lilleheil, 1968; Glavinovic, 1979); (b) postsynaptic block of ACh receptors with 3-5 μM (+)-tubocurarine (curarized preparation); (c) addition of 6 mM Mg²⁺ with reduction of Ca²⁺ to 0.9 mM (magnesium preparation). For recording the presynaptic Ca²⁺ currents associated with evoked quantal ACh release, the extracellular solution contained 1.8 mM Ca²⁺, 30 μM (+)-tubocurarine, 100 μM 4-aminopyridine and 1 mM TEA (McArdle et al. 1981; Silinsky & Solsona, 1992).

Prior to recording, the cut muscle was rinsed for at least 40 min with physiological solution. The cutting procedure does not produce significant changes in cable properties and, in combination with the voltage clamp technique, enables long-lasting stable recording of multiquantal end-plate currents (Glavinovic, 1979). Ringer solution contained (mM): NaCl 113, KCl 2.5, CaCl₂ 1.8, NaHCO₃ 2.4, pH adjusted to 7.3 with NaOH.

Chelerythrine was purchased from Tocris. All other drugs were from Sigma.

Electrophysiology

The holding potential for cut preparations was kept at −40 mV while for all other preparations it was −70 mV. On both cut and curarized preparations the voltage dependence of the peak amplitudes of the evoked end-plate currents (EPCs) was linear in the range from 0 to −100 mV, indicating the effectiveness of the voltage clamp. Bi-exponential EPC decay was observed at some synapses (≈20 %) of cut preparations. Since the nature of this phenomenon was unclear, only junctions with mono-exponential decay phase of currents were considered. EPCs were elicited every 20 s (cut or curarized preparations) or every 2-5 s (magnesium preparation) by a single supramaximal nerve stimulation. Recording of postsynaptic end-plate currents was performed using the standard two-electrode voltage clamp technique with intracellular electrodes (resistance 3-5 MΩ). Presynaptic Ca²⁺ currents following the Na⁺ nerve action potential were recorded using focal Ringer-filled micropipettes (5-7 MΩ resistance) inserted into the perineurium (detailed description of the method and origin of the complex action potential is provided by McArdle et al. 1981 and Silinsky & Solsona, 1992). The recorded EPCs were digitized at 50 kHz, stored in a PC and analysed off-line to calculate mean values of EPC amplitudes, rise times (from 10 to 90 % of the amplitude), and decay time constants (τ).

Analysis

Usually 120-150 spontaneous miniature end-plate currents (MEPCs), 10-15 multiquantal EPCs (cut or curarized preparations), 20-50 low-quantal EPCs (magnesium preparation) or 24-48 presynaptic action potentials (Ca²⁺ current experiments) were averaged to obtain mean values. In experiments with spontaneous miniature end-plate currents (MEPCs), using the method of sequential current averaging, these uniquantal currents (digitized at 50 kHz) were repeatedly averaged (120-150 events) for at least 15 min before infusion of H₂O₂ to ensure stable baseline conditions. Spontaneous events were detected using amplitude thresholds set as a multiple (4-5 times) of the S.D. of the noise. Each event was also visually inspected to prevent noise disturbance of the analysis.

Spontaneous and elicited currents were analysed off-line using Origin software (OriginLab Corp, version 6.0). Quantal content was determined either as the ratio of averaged EPCs and averaged MEPCs or as ratio of their current areas. The time course of transmitter release was evaluated by the deconvolution method (Van der Kloot, 1988) using the average EPC and the average MEPC as detailed by Rich et al. (2002). In brief, each averaged MEPC was first fitted with the eqn (1) of Rich et al. (2002), using the least-squares method. This fitted MEPC was first adjusted to have zero latency and then deconvoluted using an approximation of the Wiener filter (eqn (3) of Rich et al. 2002). The quality of transmitter time course determination was evaluated by comparing the experimental EPC with one obtained by convolution (Van der Kloot, 1988). Measured features of the biexponential transmitter release function included the decay time constant of fast and slow components (Bukharaeva et al. 1999) and the ratio of amplitudes of two components.

The data are presented as the mean ± s.e.m. (n = number of synapses), with statistical significance assessed by Student's paired t test (for parametric data) or Wilcoxon's test (for non-parametric data). Statistical significance of data was assessed using SigmaStat (Jandel Scientific; version 2.0). A P value of < 0.05 was accepted as indicative of a statistically significant difference. All drugs were dissolved to a final concentration in the bathing solution and applied to a muscle maintained in the chamber (2.5 ml) via a superfusion system (2 ml min⁻¹). To prepare stock solutions of PMA and staurosporine, these drugs were initially dissolved in DMSO. All other compounds were water-soluble. The final concentration of DMSO in the bathing solution was 0.01 %, which did not change the amplitude or the duration of the synaptic currents and did not modify the action of H₂O₂. We therefore used these data as the additional control. Measurements were started after 20 min in the presence (or washout) of various drugs, unless otherwise stated.

RESULTS

Bimodal effects of H₂O₂ on EPCs

On cut preparations of sartorius muscle, we tested the action of H₂O₂ on multiquantal end-plate currents (EPCs) elicited at a frequency of 0.05 Hz and miniature EPCs (MEPCs) occurring spontaneously in the same fibre. H₂O₂ at a low (3 μM) concentration increased the amplitude of EPCs from 123 ± 7 nA in control to 153 ± 11 nA (n = 6; P < 0.05 by paired t test, Fig. 1Aa and b). This reversible facilitation was stabilized after 15-20 min in the presence of H₂O₂. Enhancement of evoked currents was not accompanied by changes in the amplitude of MEPCs (1.13 ± 0.07 nA in control vs 1.12 ± 0.07 nA in H₂O₂; n = 6; P > 0.05, Fig. 1Ab). Furthermore, 3 μM H₂O₂ did not change the decay time constant of EPCs (τ_EPC = 1.21 ± 0.09 ms in control vs. 1.27 ± 0.10 ms in H₂O₂; n = 6; P > 0.05), or of MEPCs (τ_MEPC = 1.08 ± 0.13 ms in control vs. 1.02 ± 0.13 ms in H₂O₂; n = 6; P > 0.05). Estimation of the quantal content of EPCs based either on a simple ratio of peak current amplitudes, or the ratio of current areas (Fig. 1Ab; for details see Methods) revealed an enhancement of quantal release. With the first estimate (ratio of amplitudes), the quantal release was 110 ± 12 quanta in control and 139 ± 10 quanta in the presence of 3 μM H₂O₂ (n = 6; P < 0.05). The ratio of current integrals showed release of 160 ± 21 quanta in control and 224 ± 24 quanta in the presence of 3 μM H₂O₂ (n = 6; P < 0.05).

Figure 1. Opposite effects of low and high concentrations of H₂O₂ on multiquantal end-plate currents (EPCs).

A, facilitating effect of 3 μM H₂O₂ on multiquantal EPCs. Aa, representative EPCs in control, in the presence of 3 μM H₂O₂ and after washout. Ab, histograms showing the action of 3 μM H₂O₂ on peak amplitude of EPCs (A_EPC), decay time constant of EPCs (τ_EPC), amplitude of MEPCs (A_MEPC) and quantal content estimated either as the ratio of peak current amplitudes of averaged EPCs and MEPCs (QC-1) or as the ratio of current integrals (QC-2) on cut preparation (* P < 0.05; n = 6). Data in this and subsequent histograms are presented as the mean ± s.e.m. (n = number of synapses). B, depressant action of 300 μM H₂O₂ on multiquantal EPCs. Ba, EPCs in control, in the presence of 300 μM H₂O₂ and after washout. Bb, histograms showing the action of 300 μM H₂O₂ on the same parameters of EPCs and MEPCs as in Ab (* P < 0.05; n = 6). Note that depression of QC-1 was stronger (P < 0.05) than that of QC-2 (P > 0.05).

A similar enhancement of EPCs (17 ± 5 %; n = 5; P < 0.05) without any change in the time course of the EPC was obtained in curarized preparations, suggesting that the facilitatory effect of H₂O₂ was not related to the mode of muscle contraction block.

These data indicate that low, physiological doses of H₂O₂ facilitated synaptic transmission by enhancement of transmitter release from motor nerve endings.

As shown on Fig. 1B, a higher dose of H₂O₂ (300 μM) produced the opposite effect, namely reversible depression of EPC amplitude. In six fibres, the effect of 300 μM H₂O₂ on multiquantal EPCs was compared with the action on spontaneous MEPCs recorded from the same fibre. H₂O₂ significantly depressed the amplitude of EPCs from 158 ± 12 to 130 ± 11 nA (n = 6; P < 0.01 by paired t test). As with the low dose, 300 μM H₂O₂ did not change the amplitude of MEPCs (1.39 ± 0.12 nA in control vs. 1.41 ± 0.13 nA in H₂O₂; n = 6; P > 0.05; Fig. 3A). Similarly, no changes were observed in time course of MEPCs (τ_MEPC = 0.90 ± 0.03 ms in control vs. 0.93 ± 0.04 ms in H₂O₂; n = 6; P > 0.05). Whereas 3 μM H₂O₂ did not change the MEPC frequency, 300 μM H₂O₂ strongly and reversibly depressed the frequency of MEPCs by 69 ± 5 % (n = 6; P < 0.05). Unlike the action of the low dose, 300 μM H₂O₂ prolonged the decay of multiquantal EPCs (1.11 ± 0.04 ms in control vs. 1.33 ± 0.06 ms in H₂O₂; n = 6; P > 0.05). The quantal content calculated as the ratio of the peak amplitudes of EPCs and MEPCs was significantly (n = 6; P < 0.05) reduced in the presence of 300 μM H₂O₂ (Fig. 1Bb). However, the ratio of current areas did not reveal a depressant action of 300 μM H₂O₂ on evoked quantal release (n = 6; P > 0.05; Fig. 1Bb).

Figure 3. The action of H₂O₂ on short-term plasticity.

A, peak amplitudes of 20 successive EPCs during 60 Hz motor nerve stimulation of a cut preparation in control (•), and in the presence of 3 μM (○) or 300 μM (▴; n = 5-7) H₂O₂. Note the significantly enhanced potentiation of successive EPCs in the presence of both concentrations of H₂O₂.

The depression of EPCs by 300 μM H₂O₂ was reproduced on uncut curarized muscle (18 ± 6 %; n = 5; P < 0.05). The postsynaptic muscle block by curare did not allow recording of spontaneous synaptic events. In order to confirm the action of H₂O₂ on MEPCs of a different preparation, we performed experiments on uncut muscle without stimulation of the motor nerve. These experiments (including a test of the voltage dependence of τ_MEPC) confirmed the absence of any postsynaptic action of H₂O₂ obtained in the cut preparation (see Supplementary material available online at DOI: 10.1113/jphysiol. 2003. 050690).

Dose-response curve and high frequency stimulation

A systematic exploration of H₂O₂ action (0.3 μM-3 mM) on synaptic transmission of cut preparations is shown in Fig. 2. At a concentration of 0.3-1 μM, H₂O₂ had no effect, whereas at 3 μM, it elicited clear facilitatory effects on EPCs. At 30 μM, the effect of H₂O₂ was highly variable while concentrations exceeding 30 μM produced a depressant action on EPCs. The slowing of EPC decay occurred at concentrations of H₂O₂ greater than 30 μM (Fig. 2A). Unlike the dual action on the amplitude of EPCs, the effect of H₂O₂ on the time course of synaptic currents monotonically increased in a dose-dependent manner (Fig. 2B). This fact suggested that H₂O₂ changed not only the amplitude but also the time course of transmitter release (Van der Kloot 1988; Giniatullin et al. 1995). Indeed, the concentration dependence of the action of H₂O₂ on the area of EPCs (Fig. 2C) showed a pattern distinct from that of the amplitude changes (Fig. 2A). It should be noted that 3 mM H₂O₂ reduced both the amplitude (Fig. 2A) and the current area (Fig. 2C).

Figure 2. Concentration dependence of H₂O₂ action on EPCs.

A, effect of various H₂O₂ concentrations on peak EPC amplitude in a cut preparation. B, effect of H₂O₂ on the decay time constant of EPCs. C, the action of H₂O₂ on the current integrals. (* P < 0.05; n = 4-8 synapses). Note the biphasic action of H₂O₂ on the amplitude and area of EPCs and monophasic action of H₂O₂ on the time course of EPCs.

The above tests were based on the evaluation of low-frequency (0.05 Hz), evoked synaptic events. In order to explore the action of H₂O₂ on short-term presynaptic plasticity (Zucker & Regehr, 2002), we used a high-frequency (60 Hz) stimulation of the motor nerve.

As shown in Fig. 3, a 60 Hz train of 20 stimuli in control conditions elicited initial presynaptic potentiation followed by a slight depression (•). At a depressant (300 μM) concentration, H₂O₂ increased presynaptic potentiation (Fig. 3, ▴). Surprisingly, 3 μM H₂O₂ also significantly promoted potentiation during high-frequency trains (Fig. 3, ○).

Effect of H₂O₂ on the time course of transmitter release

Since high doses of H₂O₂ selectively prolonged the decay phase of multiquantal EPCs, we next tested if this action was based on changes in the time course of transmitter release. To this end we evaluated the transmitter release functions by the deconvolution method, using averaged EPCs and averaged MEPCs (Van der Kloot, 1998; Rich et al. 2002; for detail see Methods). As exemplified in Fig. 4A for signals used for deconvolution, 300 μM H₂O₂ prolonged the decay of EPCs without changing the time course of MEPCs (note that signals were scaled for comparison). Our analysis revealed that the quantal release function for multiquantal control EPCs consists of two exponential components (see also Bukharaeva et al. 1999) with decay time constants of 0.18 ± 0.04 and 1.02 ± 0.11 ms, respectively (n = 6; Fig. 4B). The ratio of release component amplitudes (A₁/A₂) was 1.27 ± 0.25 in control (n = 6). The application of 300 μM H₂O₂ did not change the fast component of release (0.18 ± 0.05 ms; n = 6; P > 0.05; Fig. 4C). The A₁/A₂ ratio tended to decrease in the presence of H₂O₂ (Fig. 4C) but this difference did not reach statistical significance (n = 6; P > 0.05). Nevertheless, the slow phase of release was significantly prolonged (1.34 ± 0.13 ms; n = 6; P < 0.05; Fig. 4), suggesting a mechanism responsible for the slowing of the time course of multiquantal EPCs by high H₂O₂.

Figure 4. Effect of H₂O₂ on the time course of transmitter release.

A, examples of averaged MEPCs (n = 120) or averaged multiquantal EPCs (n = 10) used for deconvolution in control and after 300 μM H₂O₂. All signals were normalized for comparison. Note that H₂O₂ selectively prolonged the decay of EPCs, while in the presence of 300 μM H₂O₂ the shape of MEPCs was the same as in control. B illustrates the results of deconvolution of the averaged EPC from A (transmitter release functions) with the fitted MEPCs (for detail see Methods). Deconvolution revealed two exponential components of transmitter release with decay time constants of 0.18 ± 0.04 and 1.02 ± 0.11 ms, respectively. Transmitter release functions were obtained by averaging data from 6 endplates. Note that 300 μM H₂O₂ increased the relative contribution of the slow phase of transmitter release function. C, histograms showing that 300 μM H₂O₂ significantly increased the decay time constant of the slow phase of the release function (τ₂; * P < 0.05; n = 6), while it did not affect the fast phase (τ₁; P > 0.05; n = 6) or the ratio of the amplitudes of the two components (A₁/A₂; P > 0.05; n = 6).

H₂O₂ at a low (3 μM) concentration did not change the time course of transmitter release (not shown), consistent with its lack of the action on the decay of EPCs.

Effect of H₂O₂ on the presynaptic Ca²⁺currents

The best approach for understanding the mechanism of the presynaptic action of H₂O₂ would be to record presynaptic Ca²⁺ currents coupled to transmitter release. The small size of motor nerve terminals does not allow intracellular recording of presynaptic depolarization or the presynaptic Ca²⁺ currents. Relatively high-resolution, extracellular recording of these currents was obtained using the method developed by McArdle et al. (1981) in which the Na⁺ action potential is detected as downward current flow preceding the upward Ca²⁺-mediated event (see example in Fig. 5A). The Ca²⁺-mediated nature of the late upward signal was confirmed by application of the Ca²⁺ channel blocker Cd²⁺ (100 μM), which almost completely removed this signal without changing the Na⁺-mediated signal (data not shown). This main Cd²⁺-sensitive component of Ca²⁺ current (Fig. 5A) decayed with a time constant of 1.33 ± 17 ms (n = 7), which was presumably prolonged due to the presence of K⁺ channel blockers. As shown in Fig. 5A and B, the application of 3 μM H₂O₂ changed neither the Na⁺ nor the Ca²⁺ phase of the complex action potential, implying that the increase in EPCs produced by H₂O₂ was located downstream from presynaptic depolarization and Ca²⁺ influx. The action of 300 μM H₂O₂ was more effective. In this case, as for 3 μM H₂O₂, no changes in the Na⁺ phase were detected (Fig. 5A and C). However, 300 μM H₂O₂ increased the decay time constant of the Ca²⁺ current (37 ± 7 % increase; n = 6; P < 0.05), while sparing the amplitude of this current (n = 6; P > 0.05; Fig. 5A and C).

Figure 5. Effect of H₂O₂ on the presynaptic Ca²⁺ current.

A, presynaptic action potential recorded by a microelectrode inserted into the perineurium adjacent to the nerve ending. The large initial downward part of the action potential represents the initial Na⁺ current while the late upward part represents the Ca²⁺ current. Each signal was obtained by averaging 24 records evoked at 0.05 Hz. The postsynaptic response was blocked by a high dose of (+)-tubocurarine (for details, see Methods). Low (3 μM) H₂O₂ did not change the Na⁺ or Ca²⁺ phase of the presynaptic signal, while high (100 μM) H₂O₂ slowed the decay of the Ca²⁺ current without producing significant changes in the Na⁺ phase. B and C, histograms showing the action of 3 or 300 μM H₂O₂ on the amplitude (% change vs. control) or the time constant of Ca²⁺ current decay. * P < 0.05, n = 6 for both H₂O₂ concentrations.

Thus, high H₂O₂ caused prolonged Ca²⁺ entry, consistent with the slowing down of the EPC time course.

Action of H₂O₂ on magnesium preparations

The actions of H₂O₂ are Ca²⁺ dependent in many experimental models (Lee et al. 2000; Oh et al. 2000). In order to understand the role of Ca²⁺ in the modulatory action of H₂O₂ on synaptic currents, we performed experiments using a ‘magnesium preparation’ (uncut muscle, −70 mV holding potential). Adding 6 mM Mg²⁺ to the external solution, while halving the Ca²⁺ concentration, reduced the amplitude of EPCs (12 ± 10 nA; n = 5) to levels roughly 10-fold smaller than in cut preparations. This effect was presumably due to strong depression of Ca²⁺ influx into the motor nerve terminals and a reduction in the quantal content of EPCs. As shown in Fig. 6A, application of 3 μM H₂O₂ to this magnesium preparation failed to enhance the peak amplitude (4 ± 3 %) or change the area (9 ± 5 %; n = 5; P > 0.05) of EPCs. Similarly, no depressant effect was obtained after 20 min with 300 μM H₂O₂ (1 ± 1 % change in amplitude or 10 ± 6 % change in area, respectively; n = 5; P > 0.05; Fig. 6B). Thus, the level of transmitter release appeared to play a crucial role in the modulatory action of H₂O₂.

Figure 6. Modification by Mg²⁺ of H₂O₂ action on EPCs.

A, time profile for the facilitatory effect of 3 μM H₂O₂ on EPCs in control (•; [Ca²⁺]_o = 1.8 mM) and in a magnesium preparation (○; [Mg²⁺]_o = 6 mM; [Ca²⁺]_o = 0.9 mM). * P < 0.05; n = 5. B, time profile for the depressant effect of 300 μM H₂O₂ on EPCs in control (•) and in a magnesium preparation (○; * P < 0.05; n = 5).

Effects of Fe²⁺ and N-acetylcysteine

In many cases the biological action of H₂O₂ is related to the generation of hydroxyl radicals (OH•) (Winterbourn, 1995). To understand the mechanism of the action of H₂O₂, it is therefore important to know whether its effects were mediated by hydroxyl radicals. It is known that in the presence of Fe²⁺, H₂O₂ is readily converted into the hydroxyl radical and this provides a reliable tool for ascertaining the mechanism of action (Winterbourn, 1995; Kress et al. 2002).

Preconditioning of muscles with 100 μM FeSO₄ depressed multiquantal EPCs by 25 ± 5 % (n = 10; P < 0.05), probably because of Fe²⁺ interference with the presynaptic voltage-activated Ca²⁺ channels. Fe²⁺ did not change the facilitatory effect of 3 μM H₂O₂, since in the presence of iron this ROS produced the same reversible enhancement of EPCs (26 ± 5 %; n = 5; P < 0.05; Fig. 7Aa) and increase in current area (27 ± 10 %; n = 5; P < 0.05; Fig. 7Ab) as in controls. However, application of 300 μM H₂O₂ in the presence of Fe²⁺ evoked a dramatic reduction in EPC amplitude of 71 ± 4 % (n = 6; P < 0.001; Fig. 7Ba) compared with a 19 % reduction in controls (P < 0.01). In the presence of pro-oxidant Fe²⁺, 300 μM H₂O₂ reduced the current area by 55 ± 5 % (n = 6; P < 0.05; Fig. 7Bb). Unlike the strong depressant effect of H₂O₂ on peak amplitude, the effect on current area was modest, presumably because the decay of EPCs was prolonged by 300 μM H₂O₂ (by 32 ± 10 %; n = 6; P < 0.05). The effect of the high concentration of H₂O₂ was slowly reversible, in contrast to the quickly reversible action of 3 μM H₂O₂. Thus, only a partial recovery to 52 ± 5 % of control EPC amplitude was achieved after 30 min.

Figure 7. Fe²⁺ selectively promoted the depressant effect of 300 μM H₂O₂.

Aa, effect of 3 μM H₂O₂ in the presence of 100 μM Fe²⁺ on multiquantal EPCs in a cut preparation. Ab, histograms showing the facilitatory action of 3 μM H₂O₂ on peak EPC amplitude or current area. Note that the facilitatory action of a low dose of H₂O₂ was not changed by Fe²⁺ (26 ± 5 % enhancement in Fe²⁺; * P < 0.05; n = 5; compare with 28 ± 7 % in control). Ba, the depressant action of 300 μM H₂O₂ on EPCs was strongly enhanced by 100 μM Fe²⁺. Bb, histograms showing the action of 300 μM H₂O₂ on peak EPC amplitude or current area in the presence of 100 μM Fe²⁺ (** P < 0.01; significantly different from the action of 300 μM H₂O₂ in control; n = 5).

These data suggest that the enhancement of EPCs by low H₂O₂ was based on the action of H₂O₂ as such, while the strong enhancement by Fe²⁺ of high H₂O₂-induced synaptic inhibition indicated the involvement of hydroxyl radicals generated via the Fenton reaction (Winterbourn, 1995).

Next, we tested the action of H₂O₂ in the presence of N-acetylcysteine, a ROS scavenger, which removes muscle fatigue created by ROS accumulation (Reid et al. 1994). The application of 100 μM N-acetylcysteine increased the amplitude of EPCs by 40 ± 6 % (n = 10; P < 0.05). Subsequent application of 3 μM H₂O₂ did not change the amplitude of EPCs (7 ± 4 %) or current areas (4 ± 4 % change; n = 5; P > 0.05). Similarly, 100 μM N-acetylcysteine prevented the depressant action of 300 μM H₂O₂ on the amplitude (3 ± 3 %; n = 5; P > 0.05) and area of EPCs (3 ± 4 % change; n = 5; P > 0.05).

Thus, a typical scavenger of ROS, N-acetylcysteine, abolished the effects of H₂O₂, indicating that the enhancing and depressant effects were indeed produced by ROS.

The role of PKC in the action of H₂O₂

To further explore the mechanisms of the action of H₂O₂ on motor nerve terminals, we tested the role of PKC, since the intracellular action of H₂O₂ is mediated by PKC in many cells (Gopalakrishna & Anderson, 1989; Konishi et al. 1997; Suzuki et al. 1997; Servitja et al. 2000).

We first blocked the activity of PKC using the non-specific inhibitor staurosporine. Staurosporine (500 nM) per se produced a slight increase in EPC amplitude (22 ± 4 %; n = 10; P < 0.05). Subsequent application of 3 μM H₂O₂ did not affect the amplitude (3 ± 5 % change; n = 5; P > 0.05) or area of EPCs (Fig. 8A). Chelerythrine (10 μM), a more specific inhibitor of PKC, also abolished the facilitatory effect of 3 μM H₂O₂ (1 ± 4 % change of amplitude and 2 ± 4 % change of area; n = 5; P > 0.05; Fig. 8A). Interestingly, pre-treatment of the muscle with 500 nM PMA, a PKC activator, also prevented the facilitatory action of H₂O₂ (Fig. 8A). PMA itself enhanced the amplitude of EPCs by 59 ± 8 % (n = 6; P < 0.05).

Figure 8. Testing the role of PKC in the action of H₂O₂.

A, histograms showing the action of 3 μM H₂O₂ on the peak amplitude of EPCs (□) and the current area (▪) in control (* P < 0.05; n = 5), in the presence of 0.5 μM staurosporine (P > 0.05; n = 5), 10 μM chelerythrine (P > 0.05; n = 5), and 0.5 μM PMA (P > 0.05; n = 5). B, the same data for the action of 300 μM H₂O₂ (n = 4-7). Note, that unlike the facilitatory action of 3 μM H₂O₂, the depressant effect of 300 μM H₂O₂ was not only preserved in the presence of PMA but even enhanced compared with control (P < 0.05).

The depressant effect of high H₂O₂ appeared to be PKC-mediated as well. Thus, staurosporine or chelerythrine completely eliminated the depressant action of 300 μM H₂O₂ on the amplitude of EPCs (2 ± 6 % and 1 ± 5 % change, respectively; for changes in current areas see Fig. 8B, ▪). In contrast to its blocking effect on the action of low doses of H₂O₂, PMA did not prevent the ability of 300 μM H₂O₂ to reduce EPCs amplitude. In fact, the depression was even stronger (30 ± 2 %; n = 5; Fig. 8B; □) than in control (19 ± 3 %; P < 0.05) and was accompanied by a significant reduction in current area (22 ± 2 %; n = 5; P < 0.05; Fig. 8B, ▪).

DISCUSSION

We have found that H₂O₂ produced opposing actions on transmission of motor signals to skeletal muscle fibres. At low physiological concentrations, H₂O₂ facilitated ACh secretion, while at high concentrations, the action of H₂O₂ resulted in synaptic depression. While the genome of the motoneuron is very susceptible to attack by ROS (Martin & Liu, 2002), current data represent another fast, presumably non-genomic, action of H₂O₂ on motoneuron terminals.

The mechanisms of H₂O₂ action on motor nerve terminals

Our data indicate that the motor nerve terminal is the preferential target for the fast action of exogenous (and potentially endogenous) H₂O₂. This was confirmed by the fact that the amplitude and decay of MEPCs were unchanged by H₂O₂. The high sensitivity of nerve terminals to H₂O₂ was consistent with earlier observations that the direct modulatory action of H₂O₂ on the muscle fibres needs much higher (mM) concentrations of this ROS (Oba et al. 1996; Lamb & Posterino, 2003).

Low H₂O₂ affected neither presynaptic Na⁺-mediated depolarization, nor subsequent Ca²⁺ current, indicating that the facilitatory action of H₂O₂ on ACh release was located downstream from Ca²⁺ influx (Hilfiker & Augustine, 1999). The potentiation of EPC amplitude by low doses of H₂O₂ during high-frequency stimulation could be explained by either enlargement of the readily releasable pool of quantal ACh (Stevens & Sullivan, 1998), or an increase in the level of residual Ca²⁺ (Zucker & Regehr, 2002).

Unlike the effect of low concentrations of H₂O₂, the actions of high doses of this ROS were more complex. First, high concentrations of H₂O₂ reduced the peak amplitude of signals but prolonged the duration of the falling phase of EPCs. Second, they prolonged Ca²⁺ influx into the motor nerve terminal. Third, high concentrations of H₂O₂, which only slightly depressed multiquantal EPCs, strongly reduced spontaneous release of ACh.

These distinct effects of high concentrations of H₂O₂ could be explained by assuming that the presynaptic action of this ROS consists of two components, one facilitatory and the other one inhibitory. Facilitation was probably related to the prolonged Ca²⁺ influx. Inhibition could be based on two mechanisms: first, on the temporal dispersion of evoked release (Van der Kloot, 1988; Giniatullin et al. 1995) consistent with the predominance of the slow component of transmitter release function found by deconvolution of EPCs and MEPCs, and second, on the impairment of releasing machinery (to explain the strong inhibition of spontaneous, Ca²⁺-independent release). Distinct factors such as Fe²⁺, PMA or an increase in H₂O₂ concentration could shift the equilibrium between facilitation and inhibition evoked by H₂O₂ and make depression the predominant phenomenon.

The effect of Fe²⁺suggests the involvement of distinct ROS

Pretreatment with Fe²⁺ dramatically increased the inhibitory potency of H₂O₂, suggesting that it was mediated by the hydroxyl radical (OH•). The OH• could be generated by the combined action of Fe²⁺ and H₂O₂ via the Fenton reaction: Fe²⁺ + H₂O₂ = Fe³⁺ + OH•+ OH⁻ (Winterbourn, 1995; Kress et al. 2002). Hydroxyl radicals, while being short-lived, are the most reactive form of ROS (Winterbourn, 1995) and participate in the depressant effects of high H₂O₂. The facilitatory action of low concentrations of H₂O₂ was unchanged by Fe²⁺, suggesting that the action was mediated by untransformed H₂O₂, with no contribution from hydroxyl radicals.

The transformation of H₂O₂ via the Fenton reaction probably occurred in the intracellular rather than the extracellular space. OH•, unlike H₂O₂, is a short-lived and cell-impermeant form of ROS that acts only at the site of its generation (Kress et al. 1995). Despite the fact that only a limited amount of free iron within the cell is in the reduced state (Kress et al. 2002), several other mechanisms could account for the transformation of H₂O₂ into OH• under physiological conditions (Liochev, 1996). If the rate of H₂O₂ breakdown was relatively low, it could explain why Fe²⁺ did not reduce the facilitatory effects of exogenous H₂O₂ when the concentration of this agent in solution was at steady state. However, breakdown could be one important factor in the action of endogenous H₂O₂ produced, for instance, during muscle activity.

Thus, the current study has revealed a synergistic interaction between Fe²⁺ and H₂O₂ in depression of synaptic transmission, indicating participation of hydroxyl radicals in this effect.

Second messenger cascades

PKC, a key enzyme for the expression of synaptic plasticity and control of transmitter release, is an important intracellular target for H₂O₂ (Gopalakrishna & Anderson, 1989; Konishi et al. 1997; Waldron & Rozengurt, 2000; Shibukawa et al. 2003).

Consistent with this notion, the facilitatory action of low H₂O₂ on ACh release was abolished by the PKC inhibitors staurosporine and chelerythrine. Furthermore, the facilitatory effect of H₂O₂ was prevented when PKC activity was up-regulated by PMA.

Unexpectedly, PKC was also involved in the depressant action of H₂O₂, as two distinct PKC inhibitors abolished the H₂O₂-mediated inhibition. However, unlike the facilitatory effect, the inhibitory action of H₂O₂ was preserved after PMA treatment.

There are several explanations for the mechanisms by which PKC might produce opposite effects. First, we cannot exclude the possibility that distinct isoforms of PKC participate in facilitation or depression, because H₂O₂ can activate either conventional, novel or atypical PKCs (Konishi et al. 1997). Remarkably, phorbol esters activate only conventional and novel, but not atypical, PKCs (Ron & Kazanietz, 1999). Second, H₂O₂ might activate PKC at low concentrations and inactivate it at high concentrations (Gopalakrishna & Anderson, 1989). The ability of PMA to counteract the facilitatory but not the inhibitory action of H₂O₂ supports this suggestion. It has been shown that the active form of PKC is more susceptible to oxidative inactivation (Gopalakrishna & Anderson, 1989), consistent with our observation that PMA pre-treatment increased the depressant action of high H₂O₂. In the nerve terminals of Calyx of Held, Ca²⁺-independent εPKC (a novel isoform) determines PMA-induced synaptic facilitation (Saitoh et al. 2001). Another isoform, γPKC, widely expressed in synapses, does not appear to be involved in the action of phorbol esters on transmitter release (Goda et al. 1996). It remains to be established what isoforms of PKC are present at motor nerve endings and are activated (or inhibited) by low or high doses of H₂O₂.

In summary, our results indicate a principal role of PKC in the action of H₂O₂ on transmitter release from the motor nerve endings.

Functional implications

Since the changes in synaptic currents were observed with very low concentrations of H₂O₂, our data suggest that ROS-mediated modulation of synaptic transmission might take place in physiological conditions. The production (O'Neill et al. 1996; Hasegawa et al. 1997; Supinski et al. 1999; Reid & Durham, 2002) and extracellular release (Reid et al. 1992) of endogenous ROS during muscle contraction provide the conditions for the feedback action of these compounds on the terminal axonal branches in skeletal muscle. At low doses, H₂O₂ could enhance ACh release, thus leading to better performance of neuromuscular transmission. In contrast, during prolonged muscle activity, a high level of H₂O₂ could participate in fatigue mechanisms. The functional output presumably depends on the concentration of H₂O₂ in the synaptic cleft. Mitochondrial respiration, one of the major sources of ROS production in skeletal muscle (Murrant & Reid, 2001) produces superoxide (the parent of H₂O₂) at a rate of approximately 1.2 nmol min⁻¹ mg⁻¹ (Boveris et al. 1972). An at least several-fold increase in ROS production during exercise has been well documented by many research groups (reviewed by Murrant & Reid, 2001). However, the exact value of the H₂O₂ concentration in the vicinity of nerve terminals remains unknown. For central synapses, high (up to millimolar) local concentrations of H₂O₂ have been suggested (Chen et al. 2001). The direct determination of ROS during reperfusion of the brain after short periods of ischaemia revealed about 200 μM H₂O₂ (Hyslop et al. 1995). Currently, it is unknown whether the presynaptic action of H₂O₂ is direct or involves other partners. For instance, the possibility exists that the action of H₂O₂ is mediated by glial cells in analogy with the well-established action of reactive nitrogen species (Castonguay et al. 2001).

Another interesting but still poorly understood phenomenon is the hypoxia- and re-oxygenation-related overproduction of ROS such as the superoxide anion (O₂⁻) and H₂O₂ (Li & Jackson, 2002). The modulatory effect of H₂O₂ on synaptic transmission should become even more important under these conditions and potentially damaging to motor nerve endings.

Conclusions

The production of cell-permeable H₂O₂ by skeletal muscle during intense contractions could generate contrasting actions on motor nerve endings. Low doses of H₂O₂ in the synaptic cleft would be expected to enhance sustained ACh release from motor nerve endings, while high doses of H₂O₂ inhibit synaptic currents. Both processes are sensitive to Ca²⁺ entry and PKC activity and might be significantly modified by ROS transformation. We hypothesize that the opposite actions of H₂O₂ represent a fast feedback mechanism for fine-tuning transmitter release during muscle activity.

Supplementary material

The online version of this paper can be found at:

DOI: 10.1113/jphysiol.2003.050690

and contains material entitled:

Effect of H₂O₂ on miniature end-plate currents (MEPCs)