Hydrogen peroxide modulates synaptic transmission in ventral horn neurons of the rat spinal cord

Key points

• Excessive production of reactive oxygen species (ROS) is implicated in many central nervous system disorders; however, the physiological role of ROS in spinal ventral horn (VH) neurons remains poorly understood.

• We investigated how pathological levels of H₂O₂, an abundant ROS, regulate synaptic transmission in VH neurons of rats using a whole‐cell patch clamp approach.

• H₂O₂ increased the release of glutamate and GABA from presynaptic terminals.

• The increase in glutamate release involved N‐type voltage‐gated calcium channels (VGCCs), ryanodine receptors (RyRs), and inositol trisphosphate receptors (IP₃Rs); the increase in GABA release, which inhibited glutamatergic transmission, involved IP₃R.

• Inhibiting N‐type VGCCs and RyRs attenuates excitotoxicity resulting from increased glutamatergic activity while preserving the neuroprotective effects of GABA, and may represent a novel strategy for treating H₂O₂‐induced motor neuron disorders resulting from trauma or ischaemia–reperfusion injury.

Abstract

Excessive production of reactive oxygen species (ROS) is a critical component of the cellular and molecular pathophysiology of many central nervous system (CNS) disorders, including trauma, ischaemia–reperfusion injury, and neurodegenerative diseases. Hydrogen peroxide (H₂O₂), an abundant ROS, modulates synaptic transmission and contributes to neuronal damage in the CNS; however, the pathophysiological role of H₂O₂ in spinal cord ventral horn (VH) neurons remains poorly understood, despite reports that these neurons are highly vulnerable to oxidative stress and ischaemia. This was investigated in the present study using a whole‐cell patch clamp approach in rats. We found that exogenous application of H₂O₂ increased the release of glutamate from excitatory presynaptic terminals and γ‐aminobutyric acid (GABA) from inhibitory presynaptic terminals. The increase of glutamate release was induced in part by an increase in Ca²⁺ influx through N‐type voltage‐gated calcium channels (VGCCs) as well as by ryanodine receptor (RyR)‐ and inositol trisphosphate receptor‐mediated Ca²⁺ release from the endoplasmic reticulum (ER). In inhibitory presynaptic neurons, increased IP₃R‐mediated Ca²⁺ release from the ER increased GABAergic transmission, which served to rescue VH neurons from excessive release of glutamate from presynaptic terminals. These findings indicate that inhibiting N‐type VGCCs or RyRs may attenuate excitotoxicity resulting from increased glutamatergic activity while preserving the neuroprotective effects of GABA, and may therefore represent a novel and targeted strategy for preventing and treating H₂O₂‐induced motor neuron disorders.

Key points

• Excessive production of reactive oxygen species (ROS) is implicated in many central nervous system disorders; however, the physiological role of ROS in spinal ventral horn (VH) neurons remains poorly understood.

• We investigated how pathological levels of H₂O₂, an abundant ROS, regulate synaptic transmission in VH neurons of rats using a whole‐cell patch clamp approach.

• H₂O₂ increased the release of glutamate and GABA from presynaptic terminals.

• The increase in glutamate release involved N‐type voltage‐gated calcium channels (VGCCs), ryanodine receptors (RyRs), and inositol trisphosphate receptors (IP₃Rs); the increase in GABA release, which inhibited glutamatergic transmission, involved IP₃R.

• Inhibiting N‐type VGCCs and RyRs attenuates excitotoxicity resulting from increased glutamatergic activity while preserving the neuroprotective effects of GABA, and may represent a novel strategy for treating H₂O₂‐induced motor neuron disorders resulting from trauma or ischaemia–reperfusion injury.

Abbreviations

ADPR

adenosine diphosphate ribose

2‐APB

2‐aminoethoxydiphenil borate

Bic

bicuculline

BPB

4‐bromophenacyl bromide

CaN

calcineurin

CICR

Ca²⁺‐induced Ca²⁺‐release

CSA

cyclosporine A

Dant

dantrolene sodium

ER

endoplasmic reticulum

IP₃R

inositol trisphosphate receptor

ω‐Aga

ω‐agatoxin IVA

MPT

mitochondrial permeability transition pore

ω‐Ctx

ω‐conotoxin GVIA

ROS

reactive oxygen species

RyR

ryanodine receptor

SCI

spinal cord injury

Stry

strychnine

VH

ventral horn

VGCC

voltage‐gated calcium channel

Introduction

Reactive oxygen species (ROS) are generated naturally during cell metabolism, and their levels are tightly regulated by the intracellular antioxidant network in the central nervous system (CNS) and other organs (Giorgio et al. 2007; Rice, 2011). However, excessive ROS production is deleterious since it contributes to neuronal damage and degeneration, a hallmark of many pathological conditions of the CNS including trauma, ischaemia–reperfusion injury, and neurodegenerative diseases like amyotrophic lateral sclerosis (ALS) (Halliwell, 1992; Coyle & Puttfarcken, 1993).

Hydrogen peroxide (H₂O₂) is a diffusible and relatively stable ROS that is present in low micromolar concentrations under physiological conditions (Lei et al. 1998). H₂O₂ levels are significantly increased in ischaemia–reperfusion lesions of the brain (Hyslop et al. 1995) and in spinal cord injury (SCI) (Liu et al. 1999). Although the mechanisms of neuronal damage induced by H₂O₂ are not entirely understood, it is known that elevated H₂O₂ levels result in elevated glutamate levels (Mailly et al. 1999) and activation of the glutamate receptors including the N‐methyl‐d‐aspartate (NMDA) receptor (Mailly et al. 1999; Avshalumov & Rice, 2002). The consequent accumulation of high levels of intracellular Ca²⁺ leads to excitotoxicity, which is characterized by neuronal damage and death (Pellegrini‐Giampietro et al. 1990; Roettger & Lipton, 1996; Shaw & Ince, 1997). Attenuating excitotoxicity is therefore a potential approach for treating CNS injuries and diseases. For instance, riluzole, which reduces overall excitability and neurotransmitter release by inhibiting presynaptic voltage‐dependent sodium channels (Leach et al. 1986; Doble, 1996), is indicated for use in ALS (Shaw & Ince, 1997) and is being investigated for its therapeutic potential in SCI (Wilson & Fehlings, 2014). However, this class of drugs inhibits not only the release of glutamate but also that of other neurotransmitters including γ‐aminobutyric acid (GABA) (Leach et al. 1986; Martin et al. 1993; Jehle et al. 2000). Since GABA provides neuroprotection in cerebral ischaemia via its well‐established role in neural inhibition (Schwartz‐Bloom & Sah, 2001), selectively inhibiting glutamate release is a better strategy for neuroprotection against excitotoxicity. To date, there has been no study on the effect of H₂O₂ on synaptic transmission in spinal ventral horn (VH) neurons, despite the finding that these neurons are highly vulnerable to oxidative stress (Carri et al. 2003) and ischaemia (Sakurai et al. 1997; Nohda et al. 2007).

The present study investigated the mechanisms by which pathological levels of H₂O₂ affect synaptic transmission in VH neurons using a whole‐cell patch clamp approach. We demonstrate that elevated H₂O₂ levels increased glutamate and GABA release from excitatory and inhibitory presynaptic terminals, respectively. The former was partly due to increased presynaptic Ca²⁺ influx through N‐type voltage‐gated calcium channels (VGCCs) and Ca²⁺ release from the endoplasmic reticulum (ER) mediated by the ryanodine and inositol trisphosphate receptors (RyR and IP₃R, respectively), whereas increased GABAergic transmission was due to IP₃R‐mediated Ca²⁺ release from the ER, which inhibited excessive glutamate release from presynaptic terminals.

Methods

Study approval

All experimental procedures involving the use of animals were approved by the Animal Care and Use Committee at Niigata University Graduate School of Medical and Dental Sciences (Niigata, Japan).

Preparation of spinal cord slices

Neonatal Wistar rats of either sex (131 rats, 7–15 days old) were anaesthetized with urethane (1.5 g kg⁻¹ by intraperitoneal injection). A dorsal laminectomy was performed and the lumbosacral segment (L1–S3) of the spinal cord was removed (Honda et al. 2011). The rats were immediately killed by exsanguination. Spinal cord specimens were placed in pre‐oxygenated ice‐cold (2–4°C) artificial cerebrospinal fluid (ACSF) containing 117 mm NaCl, 3.6 mm KCl, 2.5 mm CaCl₂, 1.2 mm MgCl₂, 1.2 mm NaH₂PO₄, 25 mm NaHCO₃, and 11.5 mm d‐glucose. After cutting the ventral and dorsal roots, the pia‐arachnoid membrane was removed, and the spinal cord was mounted on the metal stage of a microslicer (Linear Slicer PRO 7; Dosaka, Kyoto, Japan) and cut into 500 μm transverse slices. Each of these was transferred to a recording chamber and placed on the stage of an upright microscope equipped with an infrared–differential interference contrast (IR‐DIC) system (E600FN; Nikon, Tokyo, Japan). The slice was fixed with an anchor and superfused at 5–6 ml min⁻¹ with ACSF equilibrated with a gas mixture of 95% O₂ and 5% CO₂ (pH 7.4), with the temperature maintained at 36 ± 0.5°C using a temperature controller (TC‐324B; Warner Instruments, Hamden, CT, USA). Slices were used for recording after a recovery period of ≥ 40 min.

Patch clamp recordings from spinal VH neurons

Laminar regions were identified under low magnification (5× objective) and individual neurons were identified using the 40× objective lens of an IR‐DIC microscope and monitored with a charge‐coupled device camera (C2400‐79H; Hamamatsu Photonics, Hamamatsu, Japan). Whole‐cell patch clamp recordings were made from the large VH neurons of Rexed lamina IX (size: > 25 μm), which were most frequently seen in the ventrolateral and ventromedial areas. A somal size > 20 μm was used as the cut‐off in past studies of postnatal rat lumbar motor neurons (Takahashi, 1990); however, large interneurons (> 20 μm) have been detected in the ventral half of the rat lumbar spinal cord (Thurbon et al. 1998 a,b). Therefore, putative motor neurons were identified as the largest cells in the VH (somal diameter > 25 μm). After the whole‐cell configuration was established, neurons were held in voltage‐clamp at −70 mV to record excitatory postsynaptic currents (EPSCs) or at 0 mV for recording inhibitory postsynaptic currents (IPSCs). In spinal VH neurons, the reversal potentials of EPSCs are 0 mV, and those of IPSCs are −70 mV (Aoyama et al. 2010; Honda et al. 2012; Yamamoto et al. 2012). Therefore, EPSCs in the membrane current trace were specifically recorded as downward deflections at −70 mV, while IPSCs were specifically recorded as upward deflections at 0 mV. Cell input resistance (R _in) was calculated by measuring the current response to 5 mV hyperpolarizing steps from a −70 mV holding potential (Pagnotta et al. 2005; Quitadamo et al. 2005). To investigate the effects of H₂O₂ on presynaptic terminals of fibres synaptically connected to the VH neuron being recorded, all patch clamp recordings were made in the presence of the Na⁺ channel blocker tetrodotoxin (TTX; 1μm). Since conduction of action potentials is blocked by TTX, the effects of H₂O₂ on presynaptic terminals could be isolated. In the presence of TTX, postsynaptic responses to spontaneously released neurotransmitters were detected as miniature (m) EPSCs and mIPSCs. Whole‐cell patch clamp pipettes were constructed from borosilicate glass capillaries (1.5 mm outer diameter; World Precision Instruments, Sarasota, FL, USA). The resistance of a typical pipette was 4–8 MΩ when filled with internal solution composed of 110 mm Cs₂SO₄, 5 mm tetraethylammonium, 0.5 mm CaCl₂, 2 mm MgCl₂, 5 mm EGTA, 5 mm Hepes and 5 mm ATP‐Mg.

Signals were amplified with an Axopatch 200B amplifier (Molecular Devices, Union City, CA, USA), filtered at 2 kHz, and digitized at 5 kHz. Data were stored and analysed using the pCLAMP 10.3 data acquisition program (Molecular Devices).

Drug application

The following pharmacological agents were purchased from Wako (Osaka, Japan): H₂O₂ (1 mm), TTX (1 μm), CGP35348 (20 μm), dantrolene sodium (Dant; 10 μm), 2‐aminoethoxydiphenyl borate (2‐APB; 200 μm), flufenamic acid (FFA; 100 μm), 4‐bromophenacyl bromide (BPB; 20 μm), cyclosporine A (CSA; 10 μm), nifedipine (10 μm), ω‐conotoxin GVIA (ω‐Ctx; 0.5 μm), and SNX‐482 (0.1 μm). Bicuculline (Bic; 20 μm), strychnine (Stry; 2 μm), and ω‐agatoxin IVA (ω‐Aga; 0.2 μm) were purchased from Sigma‐Aldrich (St Louis, MO, USA). H₂O₂ was freshly dissolved in ACSF immediately before the experiments. TTX, Stry, CGP35348, ω‐Aga and ω‐Ctx were first dissolved in distilled water at 1000 times the working concentration. Bic, Dant, 2‐APB, BPB, CSA, nifedipine and SNX‐482 were dissolved in dimethyl sulfoxide at a 1000 times the concentration for storage, and then diluted to the working concentration in ACSF immediately before use. The osmotic pressure of nominally Ca²⁺‐free, high‐Mg²⁺ (5 mm) ACSF was adjusted by lowering the Na⁺ concentration. Drugs were applied to the whole slice by perfusion via a three‐way stopcock without changing the perfusion rate or temperature. The volume of the recording chamber was approximately 1.0 ml. The drugs reached the recording chamber within 15 s of opening the stopcock and were completely washed out within 90 s of its being closed. Both the test inhibitor and TTX were superfused at least 4 min before starting H₂O₂ superfusion. Since the focus of the present study was on the acute toxicity of ROS, H₂O₂ was applied for 5 min at a concentration of 1 mm, which was previously shown to induce neurotoxicity in spinal neurons (Taccola et al. 2008) and hypoglossal motoneurons (Nani et al. 2010) in vitro. Slices were discarded after each H₂O₂ application experiment.

Data analysis

Changes in frequency of postsynaptic currents following bath application of drugs were analysed by determining the time course of postsynaptic current frequency and amplitude before and after drug application with a time bin of 30 s using the Mini Analysis program 6.0.7 (Synaptosoft, Decatur, GA, USA). The threshold for detection of postsynaptic currents was set at 2 times the root mean square of the background noise and each automatically identified event was visually confirmed to be free of artefacts.

Statistics

All numerical data are expressed as the mean ± SEM. Student's t test (paired or unpaired) or one‐way analysis of variance (ANOVA) with a Bonferroni post hoc test was used to determine the statistical significance between means. The Kolmogorov–Smirnov test was used to compare cumulative distributions of postsynaptic current parameters in the absence and presence of the drugs. P < 0.05 was considered significant for all tests. For electrophysiological data, the number of samples n refers to the number of neurons recorded.

Results

H₂O₂ acts presynaptically to modulate spontaneous glutamate release in VH neurons

To determine whether H₂O₂ directly modulates excitatory neuronal activity, we measured mEPSCs under whole‐cell patch clamp at a holding potential of −70 mV in the presence of TTX. The mEPSCs had an average frequency of 9.5 ± 1.0 Hz and an average amplitude of 11.3 ± 0.5 pA (n = 11). H₂O₂ superfusion for 5 min produced biphasic changes in the frequency from baseline, specifically, an increase (maximum: 163.1 ± 10.9 % of the control) followed by a depression (51.7 ± 6.5% of the control) that persisted for at least 10 min following washout (Fig. 1 A–C). The maximal increase in frequency was observed 2–5.5 min after the start of H₂O₂ superfusion (mean = 3.6 ± 0.3 min) (Fig. 1 B). In contrast, mEPSC amplitude was unaffected by H₂O₂ treatment (Fig. 1 C, right panel). Figure 1 D shows that H₂O₂ increased the proportion of mEPSCs with a shorter inter‐event interval relative to the control (Kolmogorov–Smirnov test), and those with longer inter‐event intervals were observed for at least 10 min after the washout (left panel). However, H₂O₂ had no effect on the cumulative distribution of mEPSC amplitude over the recording period (Fig. 1 D, right panel).

Figure 1. H₂O₂ induces an increase followed by depression in mEPSC frequency without affecting amplitude .

A, continuous chart recording of mEPSCs before, during and after H₂O₂ (1 mm) superfusion. B, average frequency (filled circles) and amplitude (open circles) of mEPSCs in the presence of H₂O₂ relative to the control, shown as a function of time (mean ± SEM, n = 11). C, summary of mEPSC frequency (left) and amplitude (right). Values are shown as percentage change in maximal values relative to the control during H₂O₂ superfusion and after washout (mean ± SEM, n = 11 each). P values were determined by a paired t test. D, cumulative distributions of the mEPSC inter‐event interval (left) and amplitude (right) before (continuous black line), during (grey line), and after (dotted line) H₂O₂ superfusion. Values were compared to the control with the Kolmogorov–Smirnov test. H₂O₂ had no effect on amplitude (during H₂O₂ superfusion: P = 0.13; after washout: P = 0.78), but inter‐event interval was shorter during H₂O₂ superfusion (P < 0.0001) and longer after washout (P < 0.0001). A and D were obtained from the same neuron. The holding potential was −70 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

We next examined the effect of H₂O₂ treatment on R _in. Bath‐application of H₂O₂ (1 mm; 5 min) did not affect R _in values, which were 144.4 ± 40.9 MΩ before H₂O₂ superfusion, 138.9 ± 38.0 MΩ after 5 min of H₂O₂ superfusion (98.3 ± 2.1% of the control, paired t test, P = 0.47, n = 6), and 147.5 ± 42.3 MΩ after 10 min of H₂O₂ washout (101.5 ± 1.8% of the control, paired t test, P = 0.44, n = 6), when measured from the same neurons before and after application. Moreover, to examine whether the depression phase of mEPSC frequency simply reflected damage to release machinery, we investigated the influence of a prolonged (15 min) application of H₂O₂ (Fig. 2 A). At 15 min after H₂O₂ superfusion, the increase of mEPSC frequency was sustained relative to control values, which are written in square brackets at their first mention (165.3 ± 21.4% of the control [12.8 ± 3.3 Hz], paired t test, P = 0.016, n = 7). This was followed by a depression that persisted for at least 10 min following washout (42.1 ± 8.4% of the control, paired t test, P = 0.001, n = 6; Fig. 2 B, left panel). In contrast, mEPSC amplitude was not affected significantly when compared with the control (Fig. 2 B, right panel).

Figure 2. Prolonged application of H₂O₂ results in an increase followed by depression in mEPSC frequency .

A, continuous chart recording of mEPSCs before, during and after H₂O₂ (1 mm, 15 min) superfusion. B, summary of mEPSC frequency data. Values are shown as percentage change relative to the control after 15 min of H₂O₂ superfusion (n = 7) and after washout (n = 6) (mean ± SEM). P values were determined by a paired t test. The holding potential was −70 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

H₂O₂ increases mEPSC frequency by causing an elevation in presynaptic Ca²⁺ concentration

We next investigated the mechanism underlying the H₂O₂‐induced increase in mEPSC frequency. Given that the frequency of spontaneous miniature synaptic events correlates with presynaptic Ca²⁺ concentration (Augustine et al. 2003), we investigated the source of presynaptic Ca²⁺. In general, these include influx through VGCCs in the plasma membrane (Catterall & Few, 2008) and release from the ER, which stores intracellular Ca²⁺. ER release is mediated by RyR or IP₃R (Verkhratsky, 2002), whereas the mitochondria also play a role in regulating presynaptic Ca²⁺ concentration (Zenisek & Matthews, 2000; Billups & Forsythe, 2002).

Owing to the presence of exposed sulfhydryl groups, several ion channels and receptors including VGCCs (Nowicky AV & Duchen MR, 1998), RyR and IP₃R (Gerich et al. 2009) are activated by changes in cytosolic redox balance and by the level of ROS such as H₂O₂. We therefore investigated the role of VGCC‐, mitochondria‐, RyR‐, and IP₃R‐mediated Ca²⁺ signalling on H₂O₂‐induced excitatory activity by patch clamp combined with pharmacological manipulations. As a control, we recorded from cells exposed to both the inhibitor of interest and TTX but not H₂O₂. The values of the control are written in square brackets at their first mention. Given that the maximal increases in mEPSC frequency induced by H₂O₂ were observed between 2 and 5.5 min after the start of H₂O₂ superfusion, we evaluated maximal mEPSC frequency 1.5–6 min after superfusion.

We first examined whether H₂O₂‐induced increases in mEPSC frequency were dependent on extracellular Ca²⁺. In Ca²⁺‐free ACSF, H₂O₂ superfusion had little effect on mEPSC frequency (102.8 ± 6.1% of the control [8.4 ± 2.1 Hz]; paired t test, P = 0.66, n = 7; Fig. 3 A and B), suggesting that influx through VGCCs was partly responsible for the increase in intracellular Ca²⁺ level. In contrast, the subsequent depression in mEPSC frequency (48.9 ± 7.8% of the control; paired t test, P = 0.0006, n = 7) was not prevented in Ca²⁺‐free ASCF (i.e. was the same as observed in standard ACSF) (Fig. 3 A and B).

Figure 3. H₂O₂ induces an increase in mEPSC frequency by modulating Ca²⁺ influx via N‐type VGCCs .

A, effect of H₂O₂ on mEPSCs in Ca²⁺‐free ACSF. Under this condition, the H₂O₂‐induced mEPSC increase but not its subsequent depression was completely blocked. B, summary of mEPSC frequency in Ca²⁺‐free ACSF (mean ± SEM, n = 7). Values in the charts correspond to percentage change in maximal values of mEPSC frequency relative to the control during H₂O₂ superfusion and after washout. The P value was determined by a paired t test. C, ω‐Aga (0.2 μm), a P/Q‐type VGCC blocker, partially suppressed the H₂O₂‐induced increase in mEPSC frequency. D, ω‐Ctx (0.5 μm), an N‐type VGCC blocker, completely suppressed the increase in frequency. E, summary of mEPSC frequencies observed during H₂O₂ superfusion with TTX alone (n = 11) or with nifedipine (L‐type VGCC blocker, 10 μm; n = 7), ω‐Aga (0.2 μm; n = 6), ω‐Ctx (0.5 μm; n = 6), or SNX‐482 (R‐type VGCC blocker, 0.1 μm; n = 6) relative to control (mean ± SEM). Maximal values of mEPSC frequency relative to the control 1.5–6 min after H₂O₂ superfusion are shown. P values were determined by one‐way ANOVA with a Bonferroni post hoc test. The holding potential was −70 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

Multiple types of VGCCs are expressed throughout the CNS (Frank, 2014); an immunocytochemical analysis revealed that rat spinal VH neurons express L‐, P/Q‐, N‐ and R‐type VGCCs (Westenbroek et al. 1998), although the presynaptic input to these cells presumably involves a different constellation of VGCCs. To determine the contribution of different presynaptic VGCC subtypes on the H₂O₂‐induced increase in spontaneous mEPSCs at excitatory synapses, we used VGCC inhibitors specific for each channel type during H₂O₂ superfusion of VH neurons.

Nifedipine, a dihydropyridine antagonist of L‐type VGCCs, had no effect on the H₂O₂‐induced increase in mEPSC frequency (173.3 ± 18.8% of the control [10.8 ± 1.8 Hz], n = 7). In contrast, inhibiting P/Q‐ or R‐type VGCCs with ω‐Aga and SNX‐482, respectively, partially blocked this increase (ω‐Aga, 125.5 ± 10.6% of the control [10.1 ± 1.0 Hz], n = 6, Fig. 3 C; SNX‐482, 132.6 ± 6.2% of the control [8.2 ± 1.2 Hz], n = 6), while a complete suppression was observed by blocking N‐type VGGCs with ω‐Ctx (97.7 ± 5.6% of the control [12.3 ± 2.5 Hz], n = 6; Fig. 3 D). A comparison between values obtained using VGCC inhibitors and TTX only by one‐way ANOVA revealed a significant effect of ω‐Ctx (F _5,38 = 7.73, P < 0.0001; Bonferroni post hoc test, TTX only vs. ω‐Ctx, P = 0.0001; Fig. 3 E). These VGCC blockers did not block the subsequent depression of mEPSC frequency observed after H₂O₂ washout (Nifedipine, 52.0 ± 10.6% of the control, n = 6; ω‐Aga, 52.8 ± 10.8% of the control, n = 5; ω‐Ctx, 58.5 ± 10.0% of the control, n = 6; SNX‐482, 61.3 ± 7.0% of the control, n = 6).

To determine whether intracellular Ca²⁺ stores are activated by H₂O₂, we targeted mitochondria and the ER using specific inhibitors. CSA, which blocks the opening of the mitochondrial permeability transition pore (MPT), had no effect on H₂O₂‐induced changes in mEPSC frequency (with H₂O₂ superfusion, 168.3 ± 15.3% of the control [7.4 ± 1.3 Hz]; after washout, 54.2 ± 6.2% of the control; n = 6 each). CSA also inhibits calcineurin (CaN). Therefore, these results further suggest that CaN does not mediate H₂O₂‐induced changes in mEPSC frequency in spinal VH neurons.

We then tested whether H₂O₂ affects intracellular Ca²⁺ release from the ER, which is mediated by RyR and IP₃R. Dant, an RyR inhibitor, blocked the increase in mEPSC frequency induced by H₂O₂ (92.9 ± 8.3% of the control [8.1 ± 1.4 Hz], n = 8), but did not alter the subsequent decrease in frequency after H₂O₂ washout (42.7 ± 7.6% of the control, n = 5; Fig. 4 A). In the presence of 2‐APB, an IP₃R antagonist, the H₂O₂‐induced increase in mEPSC frequency not only was blocked but was reduced after 5 min, reaching a maximum value between 1.5 and 6 min after the start of superfusion to 81.2 ± 11.0% of the control [12.7 ± 2.0 Hz] (n = 6; Fig. 4 B). These changes were accompanied by an outward current (> 5 pA) in five of six neurons (14.2 ± 1.6 pA, n = 5; Fig. 4 B). The decrease in mEPSC frequency was reversible, and subsequent depression of mEPSC frequency by H₂O₂ was completely blocked (99.5 ± 7.0% of the control, n = 6; Fig. 4 B). H₂O₂ treatment in the presence of 2‐APB had no effect in recordings of another five neurons: the R _in values were 146.9 ± 31.9 MΩ before H₂O₂ superfusion, 150.24 ± 32.5 MΩ after 5 min of H₂O₂ superfusion (102.3 ± 2.5% of the control, paired t test, P = 0.42, n = 5), and 142.9 ± 29.2 MΩ after 10 min of H₂O₂ washout (98.5 ± 3.6% of the control, paired t test, P = 0.69, n = 5).

Figure 4. H₂O₂‐induced increases in mEPSC frequency are due to intracellular Ca²⁺ release via RyR and IP3R .

A, Dant (10 μm), an RyR antagonist, completely blocked the H₂O₂‐induced increase in mEPSC frequency but not the subsequent depression. B, the IP₃R antagonist 2‐APB (200 μm) completely blocked both the increase and depression in frequency. Note that the increase not only was blocked but was reduced with respect to the control after 5 min of H₂O₂ superfusion (38.9 ± 3.2% of control, n = 6). C, summary of mEPSC frequency during (left) and after (right) H₂O₂ superfusion with TTX alone (n = 11) or with CSA (a MPT blocker/calcineurin inhibitor, 10 μm; n = 6), Dant (10 μm; n = 5–8), or 2‐APB (200 μm; n = 6) relative to the control (mean ± SEM). Values on the left indicate maximal values of mEPSC frequency relative to the control 1.5–6 min after H₂O₂ superfusion. P values were determined by one‐way ANOVA with a Bonferroni post hoc test. D, FFA (100 μm), a TRPM2 channel blocker, did not block the H₂O₂‐induced increase and subsequent depression in mEPSC frequency. E, summary of mEPSC frequency recordings in the presence of FFA (mean ± SEM, n = 7). Values in the charts correspond to percentage change in maximal values of mEPSC frequency relative to the control during H₂O₂ superfusion and after washout. The P value was determined by a paired t test. The holding potential was −70 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

A comparison between values obtained with CSA, Dant, or 2‐APB and TTX only by one‐way ANOVA showed significant effects of Dant and 2‐APB on the H₂O₂‐induced increase in mEPSC frequency (F _4,31 = 12.5, P < 0.0001; Bonferroni post hoc test, TTX only vs. Dant or 2‐APB, P < 0.0001 for each; Fig. 4 C, left panel), and a significant effect of 2‐APB on the subsequent depression (F _3,23 = 11.8, P < 0.0001; Bonferroni post hoc test, TTX only vs. 2‐APB, P < 0.0001; Fig. 4 C, right panel).

Although we used 2‐APB as an IP₃ antagonist in this experiment, 2‐APB also blocks Na⁺ and Ca²⁺‐permeable melastatin‐related transient receptor potential 2 (TRPM2) channels (Xu et al. 2005; Togashi et al. 2008). We therefore examined the effect of another TRPM2 channel blocker, flufenamic acid (FFA) (Hill et al. 2004; Naziroğlu et al. 2011; Lee et al. 2011, 2013), on the H₂O₂‐induced changes of mEPSC frequency (Fig. 4 D). FFA (100 μm) did not block the initial H₂O₂‐induced increased of mEPSC frequency (171.5 ± 10.1% of control [13.4 ± 2.5 Hz], paired t test, P = 0.0004, n = 7) or the subsequent depression after H₂O₂ washout (61.5 ± 11.0% of the control, paired t test, P = 0.025, n = 7) (Fig. 4 E).

IP₃, the natural ligand for IP₃R, arises from the enzymatic activity of phospholipase C (PLC)‐β, which is downstream of G protein‐coupled receptor activation, as detailed in Fig. 8 (Rhee & Bae, 1997). We therefore examined the effect of the PLC/A₂ inhibitor BPB on the H₂O₂‐induced increase and subsequent depression in mEPSC frequency. Unlike the IP₃R antagonist 2‐APB, BPB did not block this increase (167.8 ± 15.4% of the control [13.6 ± 3.5 Hz], paired t test, P = 0.0071, n = 6) and subsequent depression (59.5 ± 6.5% of the control, paired t test, P = 0.0034, n = 5), indicating that H₂O₂‐mediated IP₃R‐related signalling is downstream of PLC‐β.

Figure 8. Model of H₂O₂‐induced modulation of presynaptic activity in spinal VH neurons .

A, H₂O₂ stimulates glutamate release by Ca²⁺ influx via N‐type VGCC (i) and Ca²⁺ release from the ER via RyR (ii) and IP₃R (iii). Oxidation of RyR cysteine residues increases RyR sensitivity to Ca²⁺, resulting in excessive activation of RyR‐mediated CICR. The effects of H₂O₂ on IP₃R‐related signalling are downstream of PLC‐β. B, H₂O₂ increases GABAergic transmission by stimulating Ca²⁺ release from the ER via IP₃R; its site of action is downstream of PLC‐β (i). Moreover, diffusion of GABA away from the release site activates presynaptic GABA_A and to a lesser degree GABA_B receptors (ii), thereby inhibiting glutamate release from excitatory presynaptic terminals. These effects by GABA may protect postsynaptic neurons from excitotoxicity resulting from excessive glutamate. Mito, mitochondria; CaN, calcineurin; PIP₂, phosphatidylinositol 4,5‐bisphosphate; DAG, diacylglycerol.

H₂O₂‐induced enhancement of GABAergic activity reduces mEPSC frequency following the initial increase

In spinal dorsal horn neurons, ischaemia leads to an increase in the release of GABA and glycine (Kawasaki et al. 2004) and a decrease in glutamate release from nerve terminals via activation of GABA receptors (Matsumoto et al. 2003). H₂O₂‐induced increases in GABA release have also been reported in the spinal dorsal horn (Takahashi et al. 2007). Based on these findings, we investigated whether GABA and/or glycine receptors at presynaptic terminals are involved in the depression of mEPSC frequency by H₂O₂ using specific receptor antagonists.

GABA receptors are ionotropic A‐type (GABA_A) or metabotropic G protein‐coupled B‐type (GABA_B) (Kumamoto, 1997), and both receptors have been detected in nerve terminals of spinal motor neurons (Rekling et al. 2000). The addition of Bic or CGP35348 – GABA_A and GABA_B receptor antagonists, respectively – led to a robust enhancement of the H₂O₂‐mediated increase in mEPSC frequency (Bic, 241.1 ± 37.3% of the control [7.3 ± 1.8 Hz], n = 7; CGP35348, 282.3 ± 69.4% of the control [7.1 ± 0.7 Hz], n = 6; Fig. 5 A and B). The subsequent depression in mEPSC frequency was blocked by Bic (108.7 ± 10.7% of the control, n = 7; Fig. 5 A) but not by CGP35348 (48.2 ± 12.2% of the control, n = 6; Fig. 5 B). Treating neurons with the glycine receptor antagonist Stry had no effect on H₂O₂‐mediated changes in mEPSC (during H₂O₂ superfusion, 162.6 ± 19.5% of the control [8.6 ± 1.1 Hz]; after washout, 44.9 ± 7.3% of the control; n = 6 for each; Fig. 5 C). A comparison between values obtained with Bic, CGP35348, or Stry and TTX only revealed a trend of enhanced H₂O₂‐induced increases in mEPSC frequency in the presence of Bic or CGP35348; however, the differences were not statistically significant (one‐way ANOVA, F _3,26 = 2.86, P = 0.056; Fig. 5 D, left panel). In contrast, Bic had a significant effect on the subsequent depression of mEPSC frequency (one‐way ANOVA, F _3,26 = 11.05, P < 0.0001; Bonferroni post hoc test, TTX only vs. Bic, P < 0.0001; Fig. 5 D, right panel). These data suggest that GABA_A and GABA_B receptor activation in terminals that innervate VH neurons inhibits the excessive presynaptic release of glutamate induced by H₂O₂ superfusion, and that GABA_A receptor activation is involved in the subsequent depression in mEPSC frequency.

Figure 5. Effects of GABA and glycine receptor antagonists on H₂O₂‐induced changes in mEPSC frequency .

A, Bic (20 μm), a GABA_A receptor antagonist, enhanced H₂O₂‐induced increases in mEPSC frequency and blocked its subsequent depression. B, CGP35348 (20 μm), a GABA_B receptor antagonist, enhanced the increase in frequency but had no effect on the depression. C, Stry (2 μm), a glycine receptor antagonist, had no effect on H₂O₂‐induced changes in mEPSC frequency. D, summary of mEPSC frequency data during (left) and after (right) H₂O₂ superfusion with TTX alone (n = 11) or with Bic (20 μm; n = 7), CGP35348 (CGP, 20 μm; n = 6), or Stry (2 μm; n = 6) relative to the control (mean ± SEM). Values on the left indicate maximal values of mEPSC frequency relative to the control 1.5–6 min after H₂O₂ superfusion. P values were determined by one‐way ANOVA with a Bonferroni post hoc test. The holding potential was −70 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

H₂O₂ increases the frequency of GABAergic miniature inhibitory postsynaptic currents (mIPSCs)

The above findings prompted us to examine the effects of H₂O₂ on GABA release from presynaptic terminals by recording GABAergic mIPSCs in VH neurons. Since GABA and glycine are both released from axon terminals that synapse on spinal motor neurons (Ornung et al. 1994; Taal & Holstege, 1994; Jonas et al. 1998; Baer et al. 2003), GABAergic mIPSCs were isolated by adding TTX and Stry to the ACSF.

GABAergic mIPSCs were measured at a holding potential of 0 mV; the frequency and amplitude were 1.9 ± 0.4 Hz and 10.0 ± 0.7 pA, respectively (n = 8). H₂O₂ superfusion resulted in a marked increase in frequency (647.8 ± 201.4% of the control) but only a minimal change in amplitude (120.0 ± 8.3% of the control; paired t test, P = 0.057; Fig. 6 A and C). These changes were accompanied by a more than 5 pA of outward current in seven of eight neurons examined (25.0 ± 10.1 pA, n = 7; Fig. 6 A). Figure 6 B shows the time course of the average change in GABAergic mIPSC frequency and amplitude in response to H₂O₂. The maximal increase in frequency was delayed, appearing 4.5–7 min (mean = 5.9 ± 0.4 min) after initiating H₂O₂ superfusion or approximately 2 min after initiating the washout (Fig. 6 C). The effect was reversible, but persisted for 10 min after the washout (Fig. 6 C); it also coincided with the decrease in mEPSC frequency observed after H₂O₂ washout (Fig. 1 B). These results suggest that changes in mEPSCs induced by H₂O₂ are mediated by GABA. Figure 6 D shows that H₂O₂ application shortened the average inter‐event interval without causing a significant change in amplitude relative to the control (during H₂O₂ superfusion: frequency, P < 0.0001; amplitude, P = 0.051; after washout: frequency, P < 0.0001; amplitude, P = 0.061; Kolmogorov–Smirnov test).

Figure 6. H₂O₂ increases the frequency of GABAergic mIPSCs .

A, continuous chart recording of GABAergic mIPSCs before during, and after H₂O₂ superfusion. H₂O₂‐induced increases in GABAergic mIPSC frequency were often accompanied by an outward current. B, average frequency (filled circles) and amplitude (open circles) of GABAergic mIPSCs after H₂O₂ superfusion relative to the control as a function of time (mean ± SEM, n = 8). C, summary of GABAergic mIPSC frequency (left) and amplitude (right) relative to the control (mean ± SEM, n = 8 each) showing an increase in frequency but not amplitude induced by H₂O₂. Middle columns of charts indicate maximal values of GABAergic mIPSC frequency and amplitude relative to the control following H₂O₂ superfusion. P values were determined by a paired t test. D, cumulative distributions of inter‐event interval (left) and amplitude (right) of GABAergic mIPSCs before (continuous black line), during (grey line), and after (dotted line) H₂O₂ superfusion. H₂O₂ did not significantly affect the distribution of amplitude (during H₂O₂ superfusion: P = 0.051; after washout: P = 0.061; Kolmogorov–Smirnov test) relative to the control, but shortened the inter‐event interval during H₂O₂ superfusion and increased the interval after washout (P < 0.0001 for each). A and D were obtained from the same neuron. The holding potential was 0 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

H₂O₂‐induced increase in GABAergic mIPSC frequency involves intracellular Ca²⁺ release via IP₃R

We next examined the role of Ca²⁺ signalling in the modulation of GABAergic mIPSC frequency by H₂O₂. In Ca²⁺‐free ACSF, H₂O₂ superfusion increased the frequency of GABAergic mIPSCs (413.0 ± 94.2% of the control [1.6 ± 0.2 Hz], n = 6; Fig. 7 A and B) to a level similar to that observed in normal ACSF (P = 0.36, unpaired t test), indicating that Ca²⁺ influx is not involved in the H₂O₂‐induced increase in GABA release. The effect was reversible, but persisted for 10 min after the washout (181.2 ± 29.4% of the control, n = 6). In addition, the H₂O₂‐induced increase of GABAergic mIPSC frequency was accompanied by an outward current (> 5pA) in three of six neurons examined (18.1 ± 10.9 pA, n = 3).

Figure 7. H₂O₂‐induced increase in GABAergic mIPSC frequency is due to intracellular Ca²⁺ release via IP₃R .

A, H₂O₂‐induced increase in GABAergic mIPSC frequency was unaffected by Ca²⁺‐free ACSF. B, summary of GABAergic mIPSC frequency relative to the control in Ca²⁺‐free ACSF (mean ± SEM, n = 6). Values in charts correspond to percentage change in maximal values of GABAergic mIPSC frequency relative to the control following H₂O₂ superfusion. P values were determined by a paired t test. C, H₂O₂‐induced increase in GABAergic mIPSC frequency and outward current were completely blocked by 2‐APB (200 μm). D, summary of GABAergic mIPSC frequencies observed during H₂O₂ superfusion with TTX combined with either Stry (2 μm; n = 8), CSA (10 μm; n = 6), Dant (10 μm; n = 6), or 2‐APB (200 μm; n = 7) relative to the control (mean ± SEM). Maximal values of GABAergic mIPSC frequency relative to the control following H₂O₂ superfusion are shown. P values were determined by one‐way ANOVA with a Bonferroni post hoc test. E, FFA (100 μm) did not block the H₂O₂‐induced increase of GABAergic mIPSC frequency. F, summary of GABAergic mIPSC frequency data during FFA treatment (mean ± SEM, n = 6). Values in the charts correspond to the percentage change in maximal values of GABAergic mIPSC frequency relative to the control following H₂O₂ superfusion. The P value was determined by a paired t test. The holding potential was 0 mV for all recordings. Washout indicates the 10 min period after initiating H₂O₂ washout.

We then pharmacologically targeted mitochondria and the ER to determine whether intracellular Ca²⁺ stores are activated by H₂O₂ at GABAergic inhibitory presynaptic terminals. CSA (an MPT blocker/calcineurin inhibitor) and Dant (an RyR antagonist) had no effect on H₂O₂‐induced increases in mIPSC frequency during H₂O₂ superfusion (CSA, 373.6 ± 72.5% of the control [1.8 ± 0.4 Hz], n = 6; Dant, 352.0 ± 63.3% of the control [2.1 ± 0.7 Hz], n = 6) and after H₂O₂ washout (CSA, 202.3 ± 34.1% of the control, n = 6; Dant, 184.6 ± 30.2% of the control, n = 6). Moreover, H₂O₂‐induced outward currents (> 5 pA) were seen in three of six neurons treated with CSA (17.4 ± 9.5 pA, n = 3) and in four of six neurons treated with Dant (28.5 ± 12.7 pA, n = 4). In contrast, 2‐APB (an IP₃R antagonist) completely blocked H₂O₂‐induced increases in mIPSC frequency during H₂O₂ superfusion (106.6 ± 11.8% of the control [2.1 ± 0.4 Hz], n = 7) and after H₂O₂ washout (101.0 ± 12.8% of the control, n = 5) as well as the outward current (Fig. 7 C). A comparison between values during H₂O₂ superfusion obtained for CSA, Dant, or 2‐APB and TTX + Stry only showed a significant effect of 2‐APB (F _3,23 = 3.31, P = 0.038; one‐way ANOVA with Bonferroni post hoc test, vs. TTX + Stry, P = 0.0046; Fig. 7 D). Moreover, the H₂O₂‐induced increase of GABAergic mIPSC frequency was not blocked by the application of the TRPM2 channel blocker FFA (during H₂O₂ superfusion, 528.1 ± 92.8% of the control [2.2 ± 0.7 Hz], paired t test, P = 0.0058, n = 6, Fig. 7 E and F; after H₂O₂ washout, 188.4 ± 20.5% of the control, paired t test, P = 0.0076, n = 6; accompanied by an outward current (> 5 pA) in four of six neurons (22.0 ± 6.5 pA, n = 4; Fig. 7 E)). These results suggest that the IP₃R‐sensitive pool in the ER is the source of the Ca²⁺ responsible for the increase in GABAergic mIPSC frequency. However, this was not blocked by the application of the PLC‐β inhibitor BPB (during H₂O₂ superfusion, 503.7 ± 95.6% of the control [1.9 ± 1.1 Hz], paired t test, P = 0.0083, n = 6; after H₂O₂ washout, 224.7 ± 40.4% of the control, paired t test, P = 0.027, n = 6), indicating that H₂O₂ acts downstream of PLC‐β, as in the case of mEPSCs.

Discussion

The present study investigated the cellular mechanisms underlying the effects of H₂O₂ on synaptic transmission in spinal VH neurons. We demonstrated that acute H₂O₂ administration induces biphasic changes in mEPSC frequency, characterized by an initial increase followed by a prolonged depression. On the other hand, prolonged H₂O₂ administration resulted in the high frequency of mEPSC that persisted before H₂O₂ washout, suggesting that the subsequent depression of mEPSC frequency observed upon acute H₂O₂ administration is associated with the removal of H₂O₂ and does not simply reflect damage to glutamate release machinery. In addition, H₂O₂ markedly increased GABA release from presynaptic terminals at a time point coinciding with the depression in mEPSC frequency, suggesting that the latter was caused by GABA‐mediated inhibitory activity. These results also indicate that H₂O₂‐induced changes in excitatory and GABAergic inhibitory activities have a presynaptic origin, given that increases in mEPSC and mIPSC frequency occurred in the absence of significant changes in amplitude.

H₂O₂ increases mEPSC frequency by Ca²⁺ influx via N‐type VGCCs

The results of VGCC inhibition suggested that the H₂O₂‐induced increase in mEPSC frequency was due to Ca²⁺ influx mainly through N‐ and, to a lesser extent, P/Q‐ and R‐type VGCCs (Fig. 8 Aa (i)). In contrast, in hippocampal neurons, H₂O₂‐induced increases in Ca²⁺ flux occur through L‐type VGCCs (Akaishi et al. 2004) whereas in Xenopus oocytes, P/Q‐type VGCCs are mainly involved (Li et al. 1998). These differences may be due to the variable expression of VGCC subtypes across cell types and/or parts of the cell. P/Q‐ and N‐type Ca²⁺ channels are predominant in synaptic nerve terminals, but their relative expression levels vary across the nervous system given the distinct physiological and pathological roles of VGCCs (Evans & Zamponi, 2006). In hippocampal neurons, the function and regulation of subtypes depends on whether they are located on excitatory or inhibitory terminals (Potier et al. 1993). In inhibitory spinal neurons, P/Q‐type VGCCs are primarily responsible for presynaptic Ca²⁺ influx (Takahashi & Momiyama, 1993), whereas our findings indicate that this function is attributable to N‐type VGCCs at excitatory presynaptic terminals of spinal VH neurons under oxidative conditions.

H₂O₂‐induced increases in mEPSC frequency are associated with intracellular Ca²⁺ release via RyR and IP₃R

Similar to the effects of several VGCC blockers, the H₂O₂‐induced increase in mEPSC frequency was suppressed by antagonists of either RyR or IP₃R, both of which mediate Ca²⁺ release from the ER. These data suggest that the mechanism underlying H₂O₂‐induced increases in mEPSC frequency involves Ca²⁺ release from the intracellular pool via RyR and IP₃R. Intracellular Ca²⁺‐induced Ca²⁺‐release (CICR) in response to activity is dependent on RyR and, to a lesser extent, on IP₃R. The oxidation of cysteine residues on RyR increases Ca²⁺ sensitivity, leading to excessive RyR‐mediated CICR (Marengo et al. 1998). Our data suggest that the increase in Ca²⁺ sensitivity of RyR is responsible for the observed H₂O₂‐induced increase in mEPSC frequency (Fig. 8 A (ii)).

On the other hand, treatment with the IP₃R antagonist 2‐APB blocked the H₂O₂‐induced increase in mEPSC frequency, which was often accompanied by an outward current. However, blocking G‐coupled protein receptor signalling, which is upstream of IP₃R, had no effect on mEPSC frequency. It was recently reported that ROS can activate IP₃ signalling and IP₃Rs by modifying cysteinyl residues, possibly in an IP₃R isoform‐dependent manner (Lencesova & Krizanova, 2012; Ivanova et al. 2014). Thus, H₂O₂ may act directly on IP₃Rs either as a ligand or by modifying the IP₃R protein in a manner that alters Ca²⁺ release in excitatory presynaptic terminals (Fig. 8 A (iii)). It is worth noting, however, that IP₃R antagonism did not simply inhibit H₂O₂‐induced increases in mEPSC frequency but instead caused a significant reduction from the baseline value for reasons that are unclear. We speculate that activation of IP₃R‐related signalling is fundamental to the increase in mEPSC frequency induced by H₂O₂, and that blocking IP₃R activated the release of neurotransmitters such as adenosine, which induced further decreases in mEPSC frequency. This is supported by previous reports that H₂O₂ activates rat hippocampal neuron adenosine A₁ receptors (Masino et al. 1999), leading to presynaptic inhibition of excitatory transmission and eliciting an outward current in spinal VH neurons (Miyazaki et al. 2008). However, it is unclear why H₂O₂ treatment in the presence of 2‐APB had no effect on R _in even though an outward current and decrease of mEPSCs frequency were observed. The 5 mV hyperpolarizing steps we applied from a −70 mV holding potential might not be sufficient to evaluate the changes of R _in induced by H₂O₂ treatment in the presence of 2‐APB. The precise mechanisms by which IP₃R regulates excitatory activity in spinal VH neurons, as well as the possible roles of adenosine signalling and the voltage dependency of the outward current by H₂O₂ treatment in the presence of 2‐APB, remain to be elucidated.

Increases in H₂O₂‐induced GABA release from presynaptic terminals is attributable to Ca²⁺ release from the IP₃R‐sensitive intracellular pool

We demonstrated that presynaptic IP₃Rs have an essential role in modulating H₂O₂‐induced increases in GABA release from presynaptic terminals of spinal VH neurons (Fig. 8 B (i)). In the rat hippocampus, H₂O₂ induces Ca²⁺ release from the ER via RyR and IP₃R by acting on the redox‐sensing elements of their hyper‐reactive sulfhydryl groups (Gerich et al. 2009). In contrast, in mouse spinal dorsal horn neurons this process depends on IP₃R but not RyR (Takahashi et al. 2007). Thus, the two receptors have distinct roles depending on the CNS region, likely to be due to differences in expression and neuronal architecture. In addition, as observed for excitatory presynaptic terminals, blocking PLA₂/C had no effect on the H₂O₂‐induced increase in GABA release, suggesting that H₂O₂ acts downstream of PLC‐β (Fig. 8). However, an alternative IP₃ generating pathway may also be involved.

The H₂O₂‐induced increase in GABAergic mIPSC frequency was often accompanied by an outward current. Although the underlying mechanism is unknown, this may involve hyperpolarization of the resting membrane potential by GABA_A activation, which would lead to Cl⁻ influx into the cell body (Sivilotti & Nistri, 1991; Laube et al. 2002). Another possibility is that H₂O₂ increases the GABA_A receptor‐mediated tonic currents in hippocampal neurons (Penna et al. 2014). Moreover, as we used neonatal rats, an exploration of the possibility of a postsynaptic uncovering of ‘silent synapses’ is needed. It has been reported that silent synapses reflect the functional presence of NMDA receptors but not α‐amino‐3‐hydroxy‐5‐methyl‐4‐isoxazole‐4‐propionic acid (AMPA) receptors, and that they are involved in a postsynaptic mechanism of long‐term potentiation (Kerchner & Nicoll, 2008). Recently, it has been reported that there is a similar form of GABA synapse plasticity (Inoue et al. 2013) and that mitochondrial ROS regulates the strength of GABA‐mediated synaptic transmission through a mechanism that recruits α3‐containig GABA_A receptors into the synapse (Accardi et al. 2014). Additional insight may be provided by future studies examining the postsynaptic effects of H₂O₂ including the possible involvement of silent synapses.

H₂O₂‐induced increases in the frequency of mEPSC and GABAergic mIPSCs do not involve Ca²⁺ influx via TRPM2 channels

TRPM2 channels are widely expressed in the brain and have been implicated in neuronal damage induced by oxidative stress (Halliwell, 2006; Hara et al. 2002). Moreover, it has been reported that TRPM2 channels are also expressed in spinal VH of neonatal rats (Bianchetti et al. 2013) and can be activated by H₂O₂ in neuronal cells (Naziroğlu, 2011; Lee et al. 2011, 2013). In our acute H₂O₂ administration experiments, the TRPM2 channel blocker FFA did not block H₂O₂‐induced changes in mEPSCs and GABAergic mIPSCs, suggesting that the effects of H₂O₂ do not involve Ca²⁺ influx via TRPM2 channels in spinal VH neurons.

Although the mechanism of TRPM2 channel activation by H₂O₂ is not entirely resolved, it appears that the gating mechanism relies primarily on its ability to release adenosine diphosphate ribose (ADPR) from mitochondria (Ayub & Hallett, 2004) and the nucleus (Caiafa et al. 2009; Esposito & Cuzzocrea, 2009; Fauzee et al. 2010). This suggests that exogenous H₂O₂ may be needed to cross the plasma membrane and generate ADPR from mitochondria and/or the nucleus to activate TRPM2 channels. TRPM2 activation by exogenous H₂O₂ shows slow kinetics (Sumoza‐Toledo & Penner, 2011) and may therefore occur at a time point beyond our experimental range. Moreover, applying FFA for only 4 min before starting H₂O₂ superfusion might not have been sufficient to antagonize TRPM2 channels completely. Further study is needed to clarify the time dependence of FFA and H₂O₂ effects on TRPM2 channels in rat VH neurons.

H₂O₂‐induced increases in GABA release inhibit glutamatergic transmission

When H₂O₂ superfusion was combined with GABA_A or GABA_B receptor antagonist, a robust increase in mEPSC frequency was observed. In addition, GABA_A receptor antagonism blocked the depression in mEPSC frequency observed after H₂O₂ washout, indicating that this was caused by GABA_A receptor activation at presynaptic terminals. These results suggest that excess GABA released from terminals may activate presynaptic GABA_A and GABA_B receptors of glutamatergic interneurons, thereby suppressing glutamate release and excitatory activity (Fig. 8 B (ii)). In support of this possibility, a similar form of synaptic modulation in the cerebellar glomerulus is thought to result from GABA in single inhibitory axons (Mitchell & Silver, 2000). Nonetheless, it is unclear why subsequent depression of mEPSC frequency was blocked only by activation of GABA_A receptors. One explanation is the differential distribution and physiological roles of GABA receptor subtypes in presynaptic terminals of spinal VH neurons. For example, in group Ia dorsal root afferent axons that make monosynaptic connections to motor neurons in the spinal cord, the GABA_A but not the GABA_B receptor, which is also present at Ia terminals, plays a predominant role in presynaptic inhibition (Stuart & Redman, 1992). Finally, as predicted by the absence of presynaptic glycine receptors in the spinal VH (Rekling et al. 2000), application of a glycine receptor antagonist had no effect on glutamate release induced by H₂O₂. However, as glycine is a key inhibitory transmitters in the spinal cord, the effect of H₂O₂ in glycinergic synaptic transmission should be studied.

Physiological significance of H₂O₂‐induced modulation of synaptic transmission

Our results demonstrate that both glutamatergic excitatory and GABAergic inhibitory synaptic transmission were enhanced by H₂O₂‐activated increases in Ca²⁺‐dependent transmitter release, albeit by different mechanisms. We do note that the apparent lack of an effect of some inhibitors might reflect the limited time (4 min) of exposure to them prior to H₂O₂ application. Notably, N‐type VGCC, RyR, and IP₃R were fundamental to H₂O₂‐induced increases in glutamate release, while only IP₃R was implicated in the increase in GABA release induced by H₂O₂. These results suggest a novel strategy for preventing and/or treating H₂O₂‐induced motor neuron disorders resulting from trauma or ischaemia–reperfusion injury by selectively blocking N‐type VGCCs or RyR but not IP₃R, which would specifically suppress excitotoxicity caused by the H₂O₂‐induced increase in glutamate release without blocking the neuroprotective effects of GABA.

Additional information

Competing interests

The authors have declared that no conflicts of interest exist.

Author contributions

M.O. and N.O. conducted experiments and analysed data. M.O., T.H., K.W., K.K, N.E., H.B. and T.K conceived and designed the project. M.O. and T.K. wrote the manuscript. T.H., K.W., K.K., N.O., H.B. and N.O. revised the manuscript critically for important intellectual content. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.

Funding

This work was supported by a Grant for Promotion of Niigata University Research Projects (25C032), a Tsukada Grant for Niigata University Medical Research, and a Grant‐in‐Aid for Exploratory Research (grant number 25670668 and 15K19989) from the Ministry of Education, Culture, Sports, Science and Technology of Japan (Tokyo, Japan).