HYDROGEN SULFIDE INHIBITS GIANT DEPOLARIZING POTENTIALS AND ABOLISHES EPILEPTIFORM ACTIVITY OF NEONATAL RAT HIPPOCAMPAL SLICES

Hydrogen sulfide (H2S) is an endogenous gasotransmitter with neuroprotective properties that participates in the regulation of transmitter release and neuronal excitability in various brain structures. The role of H2S in the growth and maturation of neural networks however remains unclear. The aim of the present study is to reveal the effects of H2S on neuronal spontaneous activity relevant to neuronal maturation in hippocampal slices of neonatal rats. Sodium hydrosulfide (NaHS) (100 lM), a classical donor of H2S produced a biphasic effect with initial activation and subsequent concentration-dependent suppression of network-driven giant depolarizing potentials (GDPs) and neuronal spiking activity. Likewise, the substrate of H2S synthesis L-cysteine (1 mM) induced an initial increase followed by an inhibition of GDPs and spiking activity. Our experiments indicate that the increase in initial discharge activity by NaHS is evoked by neuronal depolarization which is partially mediated by a reduction of outward K currents. The subsequent decrease in the neuronal activity by H2S http://dx.doi.org/10.1016/j.neuroscience.2016.10.051 0306-4522/ 2016 IBRO. Published by Elsevier Ltd. All rights reserved. *Correspondence author. Address: Department of Human and Animal Physiology, Institute of Fundamental Medicine and Biology, Kazan Federal University, Kazan 420008, Kremlevskii Street, 18, Russia. E-mail addresses: alv.yakovlev@gmail.com (A. V. Yakovlev), harlequin18@gmail.com (E. D. Kurmasheva), rashid.giniatullin@uef. fi (R. Giniatullin), ilgam.khalilov@inserm.fr (I. Khalilov), guzel. sitdikova@kpfu.ru (G. F. Sitdikova). Abbreviations: 3-MST, 3-mercaptopyruvate sulfurtransferase; ACSF, artificial cerebrospinal fluid; AMPA, a-amino-3-hydroxy-5-methyl-4-iso xazolepropionic acid; CAT, cysteine aminotransferase; CBS, cystathionine beta-synthase; CNQX, 6-cyano-7-nitroquinoxaline-2,3di one; CSE, cystathionine gamma-lyase; DAO, d-amino acid oxidase; d-APV, d-2-amino-5-phosphopentanoate; EGTA, ethylene glycol-bis(2aminoethylether)-N,N,N0,N0-tetraacetic acid; GABA, c-aminobutyric acid; GDPs, giant depolarizing potentials; H2S, hydrogen sulfide; HEPES, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; MUA, multi-unit activity; NaHS, sodium hydrosulfide; NMDA, N-methyl-Daspartate; TEA, tetraethylammonium; TTX, tetrodotoxin. 153 appears to be due to the rightward shift of activation and inactivation of voltage-gated Na currents, thus preventing network activity. NaHS also reduced N-methyl-D-aspartate (NMDA)-mediated currents, without essential effect on AMPA/kainate or GABAA-mediated currents. Finally, H2S abolished the interictal-like events induced by bicuculline. In summary, our results suggest that through the inhibitory action on voltage-gated Na channels and NMDA receptors, H2S prevents the enhanced neuronal excitability typical to early hippocampal networks. 2016 IBRO. Published by Elsevier Ltd. All rights reserved.


INTRODUCTION
Hydrogen sulfide (H 2 S) has been recently identified as an important intra-and intercellular messenger, regulating various physiological and pathological processes (Kimura, 2010;Hermann et al., 2012a;Wang, 2012). In the central nervous system H 2 S induces long-term potentiation in the hippocampus (Abe and Kimura, 1996), modulates neuronal excitability of the subfornical organ, the nucleus of the solitary tract (Kuksis et al., 2014;Kuksis and Ferguson, 2015;Malik and Ferguson, 2016) and trigeminal neurons (Feng et al., 2013) and mediates central inhibition of the respiratory rhythm . In the peripheral nervous system H 2 S modulates transmitter release as well as exo-and endocytosis of synaptic vesicles in motor nerve endings (Sitdikova et al., 2011;Gerasimova et al., 2013Gerasimova et al., , 2015Mitrukhina et al., 2013). H 2 S also exerts neuro-protectant effects by preventing oxidative stress (Kimura and Kimura, 2004) and by increasing glutathione levels . Furthermore, H 2 S participates in the pathophysiology of central nervous system diseases such as epilepsy, stroke, neurodegenerative diseases and hyper-homocysteinemia (Wang, 2012;Luo et al., 2014). Neurotoxic or neuroprotective action of H 2 S is critically dependent on its concentration and cellular location (Wedmann et al., 2014).
Giant depolarizing potentials (GDPs) are spontaneous synchronized neuronal discharges generated by the hippocampal network in neonatal rats. GDPs play an important role in the development of synaptic connections between cells during intense neuronal growth and synaptogenesis (Ben-Ari et al., 1989). In the developing brain c-aminobutyric acid (GABA) is excitatory because neurons exhibit a high intracellular chloride concentration (Ben-Ari et al., 2012). The depolarizing action of GABA, together with that of glutamate underlies the initiation of GDPs (Ben-Ari et al., 1989).
The expression of CBS in the nervous system during the embryonic period is generally low and increases in the late embryonic to the early postnatal period (Enokido et al., 2005;Bruintjes et al., 2014). The significance of this developmental regulation is unknown. A possible role of CBS may comprise a reduction of the homocysteine level which at high concentrations contributes to pathologies of nervous system development (Rosenquist and Finnell, 2001). However, the role of H 2 S in the hippocampal network maturation remains unknown.
The aim of our study was to investigate the effects of H 2 S and L-cysteine on spontaneous network neuronal activity and membrane potential of pyramidal neurons in rat hippocampus slices during the first week after birth. We also examined the effects of H 2 S on voltagedependent Na + currents and K + currents, to highlight the contribution of these currents in the regulation of excitability. In addition we analyzed the effects of H 2 S on glutamate and GABA receptor-mediated currents of pyramidal neurons. Finally, we studied the effects of H 2 S on bicuculline-induced interictal-like events in rat hippocampus slices.

Slice preparation
Experiments were carried out on brain slices of 84 neonatal Wistar rats (postnatal days (P) 3-7). The work has been performed in accordance with EU Directive 2010/63/EU for animal experiments and all animal-use protocols were approved by the French National Institute of Health and Medical Research (INSERM, protocol N007.08.01) and Kazan Federal University on the use of laboratory animals (ethical approval by the Institutional Animal Care and Use Committee of Kazan State Medical University N9-2013). Animals were anesthetized by isofluran (4%) or were subjected to cryoanesthesia before being decapitated. After isolation, the rat brains were placed into a cooled oxygenated artificial cerebrospinal fluid (ACSF) which contained (in mM): NaCl 126; KCl 3.5; CaCl 2 2.0, MgCl 2 1.3, NaHCO 3 25, NaH 2 PO 4 1.25 and glucose 10, (pH 7.4). Horizontal slices (400 lm thick) were cut using a HM 650 V vibratome (Microm International, Germany) and kept 1 h before use in oxygenated ACSF at room temperature, then transferred to the recording chamber and superfused with oxygenated ACSF (33°C).

Chemicals
Chemicals used were: N-methyl-D-aspartate (NMDA; 50-100 lM), bicuculline (10 lM), 6-cyano-7-nitroquinoxa line-2,3dione (CNQX, 10-40 lM), d-2-amino-5phosphopentanoate (d-APV, 40 lM), CGP55845 (CGP, 2 lM), tetraethylammonium (TEA, 2 mM), tetrodotoxin (TTX, 1 lM) and L-cysteine (0.3 mM). All substances except TTX were purchased from Sigma-Aldrich (St. Louis, MO, USA). TTX was purchased from Alomone labs (Jerusalem, Israel). All drugs were dissolved to a final concentration in ASCF and applied to the preparation via a bath perfusion system. Sodium hydrosulfide (NaHS, Sigma-Aldrich, USA) was used as a source of H 2 S. In solution this compound dissociates to give HSwhich associates with H + to produce H 2 S. At 37°C 14% of total sulfide is present as H 2 S calculated from the Henderson-Hasselbalch equation (Whitfield et al., 2008). The real-time measurements of H 2 S in the chamber using amperometric sensors indicate a rapid loss of sulfide via H 2 S volatilization by bubbling with about 50% H 2 S loss within 3 min (DeLeon et al., 2012;Sitdikova et al., 2014). In our experiments NaHS was used at a concentration of 100 lM which yields about 14 lM H 2 S in the perfusion system which constantly flows to the recording chamber. Due to volatilization the H 2 S concentration further decreases by 50% of the initial level which amounts to approximately 7 lM H 2 S in the perfusate. In addition, H 2 S is quickly bound to slice preparation tissue and undergoes oxidation, as it was shown in the plasma in vivo (Whitfield et al., 2008). This will further decrease the H 2 S concentration at the target sites by an unknown amount. Stock solutions of NaHS were prepared immediately before each experiment and kept hermetically sealed in a dark place.

Electrophysiological recording
Extracellular field potentials and multi-unit activity (MUA) were recorded, from hippocampal slices from the CA3 pyramidal cell layer using tungsten wire electrodes (diameter 50 lm, California FineWire, Grover Beach, CA, USA) and a low-noise DAM-80 amplifier (WPI, GB; low-pass filter: 0.1 Hz; high-pass filter: 3 kHz; Â1000). Spontaneous GDPs were recorded using patch-clamp techniques in whole-cell configuration in voltage-clamp mode. Glass microelectrodes were located under visual control by means of an Axio Examiner A1 microscope (400Â, Carl Zeiss, Germany) using differential interference contrast optic.
Patch-clamp amplifier Axopatch 200B (Axon Instruments, Molecular Devices, CA, USA) was employed for signal recordings using a gain of 5 mV/pA and a band path of 0-2 kHz. For whole-cell and cellattached recordings electrodes were prepared from borosilicate glass capillaries with outer and inner diameter of 1.5 and 0.86 mm, respectively (GC150F-15, Clark Electromedical Instruments, UK). Pipettes were filled with either 135 mM cesium methylsulfate or 135 mM potassium gluconate solution. Both electrode solutions also contained (mM): CaCl 2 0.1, EGTA 1, HEPES 10, NaATP 2 and NaGTP 0.4 (pH 7.25), osmolarity 290 mOsm, with the pH adjusted to 7.3 with CsOH or KOH. Patch pipettes had a resistance of 4-8 MO. The series resistance (R s ) was not compensated. The liquid junction potential for potassium gluconate-based and cesium gluconate-based internal solution were 15.6 mV and 9.9 mV, which were corrected for all data analysis. The membrane resistance was measured in current clamp whole-cell mode by the injection of negative current pulses (0.02 nA) into the neurons.
Current-clamp measurements of the resting membrane potential of CA3 pyramidal cells were performed in cell-attached configuration with null current injected after high-resistance seal formation (>1 GOm) (Mason et al., 2005;Perkins, 2006). The membrane potential (E m ) was defined as averaging data during 10-min recordings.
The activity of single NMDA channels was recorded using pipettes filled with nominally Mg 2+ free ACSF containing NMDA (10 lM) and glycine (10 lM). In each experiment the activity of NMDA channels was recorded at different voltages from +20 to À50 mV in a stepwise manner, with steps of 10-20 mV. Detection of the opening of single NMDA channels and analysis of the single-channel current amplitudes was performed by means of Axon Software after additional filtration (RC single-pole filter with cut-off frequencies below 200 Hz) of the signals, as described earlier (Tyzio et al., 2003;Yakovlev et al., 2013). Na + currents were studied using a voltage clamp protocol with depolarizing test pulses of 250 ms at 10 mV increments from À70 to +20 mV from a holding potential of À75 mV using cesium methylsulfate-based pipette solution and TEA 2 mM in the external solutions for blocking K + conductance. Current-voltage (I-V) curves were obtained from measurements of peak inward currents. To obtain a steady-state inactivation of Na + currents neurons were kept at prepulses potentials from À80 to À10 mV for 100 ms in 5 mV steps followed by test pulses at À10 mV for 500 ms (Kuksis and Ferguson, 2015). Using BoltzIV function (OriginLab Corporation, USA) the half-maximum activation (V half ), slope (k), maximal conductance (G max ), and reversal potentials (V rev ) for Na + currents were calculated (+58 mV) and the data of current-voltage curves for the peak Na + currents were fitted as follows: I = G max (V À V rev )/{1 + exp [(V À V half )/k]}, where I is the peak transient inward current. Normalized peak currents to maximal transient peak currents (I max ) were plotted versus prepulse potentials and fitted with the Boltzmann function: I/I max = 1/[1 À exp (V À V 50 )/k], where I max is the maximum current, V 50 is the membrane potential at the midpoint of the curve, and k is a slope factor. K + currents were studied using a voltage clamp protocol with depolarizing test pulses at 10 mV increments from À130 to +50 mV for 250 ms from a holding potential of À75 mV using potassium gluconatebased pipette solution and antagonists of NMDA, GABA or AMPA/kainate receptors -d-APV (40 lM), CNQX (15 lM) and bicuculline (2 lM), respectively, and TTX (1 lM for blocking voltage-gated Na + conductance) in external solutions. Current-voltage (I-V) curves were calculated from measurements of current amplitudes at the end of the 250-ms voltage pulses.

Focal puff drug application
A pneumatic picopump (PC-820, WPI, USA) was used for puff application of NMDA (50 lM + 50 lM glycine for activation of NMDA receptors), glutamate (1 mM, for AMPA/kainate receptors) or GABA (100 lM, for GABA receptors) from a glass pipette at a distance of about 50-250 lm from the neuronal soma. Pressure was applied at 5-10 psi for durations in the range 50-200 ms. To avoid desensitization of NMDA, GABA or AMPA/kainate receptors agonists substances were applied at 5-min intervals in control with NaHS (100 lM) added in the bath solution in 5, 10, 15 min of perfusion and during washout. At the end of each experiment antagonists of NMDA, GABA or AMPA/kainate receptor -d-APV (40 lM), CNQX (15 lM) and bicuculline (2 lM) were applied, respectively. The area under each trace after puff application was calculated by integration using Origin Pro 2015 (OriginLab Corp, USA).

Statistics
Group measurements are expressed as mean ± SEM, n -number of animals. All data were checked for normality using the Shapiro-Wilk test and for equal variances using F-test Origin Pro 2015 (OriginLab Corp, USA). Statistical significance between means was calculated using paired t-test. In case of heterogeneous variances Welch's correction was performed and two-sample t-test was used in Origin Pro 2015 (OriginLab Corp, USA). The level of significance was set at p 6 0.05.

RESULTS
NaHS and L-cysteine decrease neuronal activity and network-driven GDPs in the hippocampus of neonatal rats.
We first tested the actions of NaHS and the metabolic precursor of H 2 S, L-cysteine, on the extracellular recorded neuronal activity of immature hippocampus which is characterized by spontaneous network-driven GDPs (Ben-Ari et al., 1989), consisting mainly of GABAergic postsynaptic currents and bursts of MUA (Fig. 1A). All records were obtained from the CA3 region of the hippocampus which is centrally responsible in GDP generation . In control animals, the frequency of GDPs was 0.11 ± 0.03 c À1 (n = 14 rats). Application of NaHS (100 lM) to the bath induced a drastic reduction of GDP frequency to 0.017 ± 0.01 c À1 (16 ± 8%; of control; t(8.5) = 5.68, p 6 0.001, n = 8) within 20 min (Fig. 1A, C). In about 60-70% of slices an initial short-term increase in GDP frequency to 0.18 ± 0.01 c À1 % of control (163 ± 31% of control; t(5) = 4.35, p 6 0.05, n = 6) was observed during the first 5 min of NaHS application ( Fig. 1A, C). NaHS also induced two-phase changes of MUA in CA3 pyramidal neurons ( Fig. 1A, B, D). NaHS application augmented the number of spikes up to 130 ± 7% of control relative to control (from 3.6 ± 0.5 to 4.7 ± 0.9 c À1 ; t(7) = À2.37, p 6 0.05, n = 8) during the first 5 min. By 20 min of NaHS application the number of action potentials was significantly reduced to 53 ± 6% of control (1.9 ± 0.4 c À1 ; t(7) = 4.56, p 6 0.01, n = 8; Fig. 1D). Almost complete recovery of GDPs and MUA was observed after an $30to 45-min washout ( Fig. 1A-D). The inhibitory effect of NaHS on GDPs was dosedependent from 1 to 300 lM, where 300 lM of NaHS completely abolished GDPs within 20 min. As shown in the cumulative dose-response curve of Fig. 1E, fitting the experimental points with the Hill equation gave an EC 50 value of 58 lM. In subsequent experiments we used NaHS at a concentration of 100 lM (effective concentration of 7 lM, see Experimental procedures/chemicals section).
NaHS effects on spontaneous GABA A -mediated GDPs were supported by whole-cell recordings at a holding potential of 0 mV from CA3 hippocampal pyramidal neurons using low-chloride pipette solution (Cs-based pipette solutions). Bath-applied NaHS (100 lM) reduced the amplitude and frequency of spontaneous GDPs progressively and within 20 min the currents could not be discriminated from the baseline noise ( Fig. 1F; n = 6). The effect of NaHS was reversible and spontaneous GDPs reappeared after a 20-to 30-min washout in control solution.
NaHS and L-cysteine depolarize pyramidal neurons in the CA3 hippocampal region of newborn rats In neurons and other excitable cells the value of the resting membrane potential plays a key role in electrogenesis. The effects of NaHS and L-cysteine on the membrane potential of CA3 pyramidal neurons were estimated using cell-attached current-clamp recordings with simultaneous extracellular recording of GDPs and spontaneous MUA at the CA3 layer (Perkins, 2006). Cell-attached current-clamp recordings were used in order to avoid the impact of dialysis of the cytosol by the intra-pipette solution and an artificial conductance (leakage) (Tyzio et al., 2003;Mason et al., 2005). The resting membrane potential was measured from the average of membrane potential recordings in current-clamp mode. Average membrane potential values of À72 ± 5 mV (n = 4 rats) in CA3 pyramidal cells of P4-P7 animals obtained in our study were close to the values of À77 ± 2 mV reported previously (Tyzio et al., 2003). In Fig. 2A an example of a neuron recorded after NaHS (100 lM) application is presented. Administration of NaHS or L-cysteine induced a rapid depolarization which recovered to baseline after washout. One hundred micromolar NaHS caused a decrease in the membrane potential from À72 ± 5 mV to À42 ± 10 mV (t(3) = À3.6, p 6 0.05) and L-cysteine (1 mM) to À37 ± 6 mV (t(3) = À5.1, p 6 0.05) ( Fig. 2A, B). During washout after initial hyperpolarization the membrane potential recovered within 20 min ( Fig. 2A). Synchronous field recordings from CA3 hippocampus region revealed the initial bursts of action potentials and GDPs, followed by a depression of action potential discharge.
In a next series of experiments we analyzed the effect of NaHS on the amplitude of single NMDA channels currents at different holding potentials using cell-attached recordings. This method of the cells resting membrane potential estimation is based on the fact that the reversal potential of the non-selective cationic NMDA channels is close to 0 mV (Nowak et al., 1984) and therefore NMDA currents reverse their polarity at the pipette voltage equal to the resting potential (Tyzio et al., 2003;Yakovlev et al., 2013). When the pipette voltage was equal to 0 mV single NMDA channel openings were manifested as transient, 2-to 4-ms long steps of inward current with an amplitude of 4.5 ± 1.1 pA (n = 7 rats; age of animals P4-P7). It should be noted that in the cell-attached configuration currents are directed opposite to those recorded intracellularlytherefore, inward currents in the cell-attached mode are upward. Based on the measurements of amplitudes of single currents flowing through NMDA channels at various voltages applied to the electrode, current-voltage plots were constructed of each cell (Fig. 2C). In control, NMDA currents reversed at an electrode voltage of À74 ± 2 mV (n = 7 rats). Application of 100 lM NaHS shifted the reversal potential to À44 ± 5 mV (t(6) = À6.49, p 6 0.001, n = 7) (Fig. 2C). Hence, this shift of reversal potential of NMDA currents during NaHS application indicates a depolarization of pyramidal neurons by NaHS.
NaHS induces a rightward shift of activation and inactivation of voltage-gated sodium currents and a decrease of outward potassium currents in pyramidal CA3 neurons.
Depolarizing effect of NaHS may be mediated by activation of voltagegated Na + currents (I Na ) as was shown in neurons of the subfornical organ (SFO) or in the nucleus of the solitary track (Kuksis and Ferguson, 2015;Malik and Ferguson, 2016). To evaluate the effect of NaHS on I Na whole-cell patch-clamp recordings were used. I Na was induced by depolarizing voltage steps according the protocol described in 'Experimental procedures' section and was completely abolished by TTX (1 lM) (Fig. 3A). In control conditions a threshold of I Na activation was between À50 and À40 mV and a current peak was observed at À20 mV with amplitudes of 1.5 ± 0.2 nA (n = 6 rats). The voltage of halfmaximal activation (V half ) was À35 ± 1 mV and slope (k) 3.7 ± 0.6 (n = 6; Fig. 3B). After 4-to 6-min incubation NaHS induced a rightward shift of I Na activation. Threshold of I Na activation was between À40 and À30 mV and current peak was observed at 0 mV with an amplitude of 1.5 ± 0.2 nA (n = 6; Fig. 3B). The voltage of V half shifted to À11 ± 2 mV after NaHS application (t(5) = À15.9, p 6 0.001, n = 6) and k did not change significantly (4.5 ± 0.3; t(5) = À2.3, p > 0.05, n = 6). Application of an inactivation step protocol revealed a rightward shift of the inactivation curve in response to NaHS. It was shown that half-maximal inactivation (V 50 ) was shifted from À46 ± 2 mV in control to À29 ± 3 mV after NaHS application (t(5) = À4.5, p 6 0.01, n = 6) whereas the slope of the curve did not change significantly (from 5.2 ± 0.1 in control to 4.7 ± 0.2 in NaHS; t(7) = 1.33, p > 0.05, n = 6) (Fig. 3C). Thus NaHS did not affect the I Na amplitude but induced a rightward shift of the activation curve which in turn increases the threshold of action potentials generation and decreases excitability. Bath application of TTX (1 lM) did not prevent NaHS-induced depolarization of the membrane potential of neurons. It was shown that potassium channels in different tissues mediate the effects of H 2 S on membrane potential and excitability (Feng et al., 2013;Mustafina et al., 2015). NaHS induced a reduction of outward K + currents at voltages from +5 to +50 mV (t(3) = 3.56, p 6 0.05, n = 4; Fig. 3D, E) during 2 min after donor addition to the bath solution. These effects were prevented by application of TEA (5 mM), a non-specific inhibitor of voltage-gated and Ca 2+ -activated K + channels before NaHS application (Fig. 3E).
In conclusion, a reduction of K + currents by NaHS at least partially mediates the depolarization of neurons which underlies the initial increase in firing rate. A rightward shift of activation of I Na in turn decreases neuronal excitability and prevents network activity.

NaHS decreases NMDA-mediated currents without effects on GABA A -and AMPA/kainatemediated responses
In neonatal hippocampus during the first postnatal week network-driven GDPs are largely mediated by the depolarizing action of GABA and glutamate through activation of GABA A and NMDAreceptors Ben-Ari et al., 2012). Therefore, we analyzed the effects of NaHS on GABA A -, NMDA-and AMPA/kainate-mediated currents.
To isolate the effect of NaHS on GABA A -currents, GABA B receptors were inhibited by CGP (2 lM) and AMPA/kainate receptors by CNQX (15 lM), and NMDA receptors were blocked by d-APV (40 lM). GABA was focally applied (100 lM, puff duration 100 ms) and whole-cell recordings from pyramidal neurons were performed at a holding potential of À60 mV using cesiummethylsulfate-based pipette solution. Focal GABA application evoked GABA A receptor-mediated currents (Fig. 4A) which were inhibited by 10 lM bicuculline (Fig. 4A, B). Application of NaHS (100 lM) to the bath solution prior to GABA application did not significantly change the area and the amplitude of GABA A currents (mean area in NaHS was 102 ± 14% of control values; t(3) = À1.1, p > 0.05, n = 4; Fig. 4A, B). To examine the effects of NaHS on AMPA/ kainate-currents glutamate (1 mM, 50 ms) was applied at a holding potential of À60 mV in the presence of NMDA and GABA A/B receptors antagonists d-APV (40 lM), bicuculline (10 lM) and CGP (2 lM), respectively (Fig. 4C). NaHS had no effects on AMPA/ kainate currents which were blocked by CNQX (15 lM) (mean area in NaHS was 105 ± 15% of control values; t(3) = 0.32, p > 0.05, n = 4; Fig. 4C, D).

NaHS inhibits interictal-like events in rat hippocampal slices
The first two postnatal weeks are very important for brain development and for synaptogenesis, but at the same time this period is characterized by a high incidence of seizures as the result of depolarizing GABA action (Ben-Ari et al., 2012). GABA also exerts an inhibitory action in the immature brain because of a shunting mech-anism due to an increase in membrane conductance (Khalilov et al., 1999). It was shown that in neonatal rat hippocampal slices GABA A receptors antagonists block GDPs and induce glutamatergic interictal-like activity (Khalilov et al., 1999(Khalilov et al., , 2003. This activity is similar to the interictal-like epileptiform activity that is observed in adult hippocampal slices after inhibition of GABA A receptors (Khalilov et al., 1999).
To study the effect of H 2 S on epileptiform activity in the immature hippocampus (P3-P7), we induced interictal-like events by application of bicuculline (Khalilov et al., 1999). Application of 10 lM bicuculline to the bath solution completely blocked GDPs and by 6-10 min of perfusion elicited interictal-like epileptiform discharges with a frequency of 0.02 ± 0.003 c À1 (n = 5 rats; Fig. 5A). Further application of 100 lM NaHS induced a full and reversible inhibition of epileptiform discharges and reduced the frequency of action potentials (Fig. 5B).
Our results indicate that the biphasic effects of H 2 S on the physiological pattern of GDPs (transient potentiation in GDPs frequency followed by their inhibition) are due to neuronal depolarization causing network desynchronization. We hypothesize that by this mechanism H 2 S will be capable of desynchronizing and thus suppressing hyper-synchronous epileptiform discharges.

DISCUSSION
The main finding of our study is that H 2 S induces a biphasic effect on network-driven GDPs and neuronal spiking activity in rat neonatal hippocampus, where an initial increase in activity is followed by an inhibition of GDPs and MUA. We have shown that the initial phase is due to a transient depolarization mediated partially by a reduction of outward potassium currents. The second phase is due to a right shift in the activation curve of Na + current thus, decreasing neuronal excitability and preventing network activity. Furthermore, NaHS decreases NMDAmediated currents in neonatal rats without affecting AMPA and GABA responses. Finally, in this preparation H 2 S abolishes epileptiform activity induced by bicuculline.
At early stages of development in the immature central nervous system spontaneous synaptic activity regulates neuronal survival and proliferation, migration, synaptogenesis (Ben-Ari et al., 1989). Spontaneous network-driven GDPs are generated in rat hippocampal slice preparations during the first postnatal week (Ben-Ari et al., 1989). In our study we found that NaHS -a H 2 S donor -caused initial activation and subsequent suppression of GDPs and MUA in CA3 hippocampal neurons in a reversible manner during first postnatal week. The inhibitory NaHS effect was concentration-dependent from 1 lM to 300 lM with an EC 50 of 58 lM which corresponds to an effective H 2 S concentration about 4 lM taking in consideration the evaporation of H 2 S during bubbling in the perfusion system (see 'Experimental procedures' section;DeLeon et al., 2012;Sitdikova et al., 2014). NaHS at a concentration of 100 lM yields about 7 lM H 2 S which is close to endogenous H 2 S concentrations measured in mammalian tissues (from 10 nM to 3 lM, Furne et al., 2008;Ishigami et al., 2009) which probably is responsible for physiological actions and does not induce deleterious effects (Wedmann et al., 2014). Indeed in isolated cells 0.1 mM Na 2 S did not induce the changes of the mitochondrial membrane potential and decreased superoxide anion formation. Only higher doses of H 2 S (0.5 mM) caused oxidative stress in the cells (Wedmann et al., 2014). Moreover, H 2 S can even protect mitochondria especially during ischemia-reperfusion injury (Elrod et al., 2007). Concentrations used in our study may reflect the increase in local changes of H 2 S level in response to increase in endogenous release during physiological conditions. In this case the low basal concentration of H 2 S in brain would provide the ability of H 2 S to regulate cellular functions  and in experimental conditions micro-molar doses of H 2 S are able to activate cellular signaling before its level decreases .
First evidence of a neuromodulatory role of H 2 S in the rat hippocampus was obtained by Abe and Kimura (1996), where NaHS suppressed field excitatory postsynaptic potentials in the CA1 region of hippocampal slices. H 2 S-producing enzymes CBS, 3-MST/CAT or DAO are highly expressed in the hippocampus (Abe and Kimura, 1996;Enokido et al., 2005;Kimura, 2010;Renga, 2011;Bruintjes et al., 2014) and their activity can be modulated by different factors such as S-adenosyl methionine (SAM), glutathionylation, nitric oxide, carbon monoxide and Ca 2+ (Kimura, 2010). The substrate of H 2 S synthesis L-cysteine at a concentration of 1 mM (Dominy and Stipanuk, 2004) causes similar effects to NaHS on MUA and GDPs, however, the first excitatory phase was much more pronounced compared to NaHS. The initial excitatory effect of L-cysteine may result from its direct activation of NMDA receptors or by increasing GABA release from interneurons which has been observed in the presence of sulfur-containing amino acids (McBean, 2007). This suggestion is supported by the fact that a smaller concentration of 0.3 mM L-cysteine displayed only excitatory action in our experiments. At the same time NaHS at low concentrations (1-50 lM) acts only to depress network activity. It is well known that high doses of L-cysteine produced excitotoxic damage in different tissue especially in central nervous system by activation of NMDA receptors (Puka-Sundvall et al., 1995;Janaky et al., 2000), however, this neurotoxic effect occurs only during long-term exposure of tissue to high concentration of L-cysteine (more than 2-3 h) (Olney et al., 1990). At the same time L-cysteine is a precursor for glutathione synthesis in neurons and may thus be involved in neuroprotection . L-Cysteine at a concentration more than 1 mM is usually used as a substrate of H 2 S synthesis in different tissues, including brain and basal H 2 S synthesis was decreased by CBS inhibitors (Abe and Kimura, 1996;Furne et al., 2008;Linden et al., 2008;Martin et al., 2010). It was suggested that 1 mM L-cysteine in addition to its excitatory action on NMDA receptors also serves as a substrate for CBS and increases H 2 S synthesis. Neurotoxic effects of L-cysteine appear unlikely in our experiments as we apply L-cysteine only for 20 min (short-term application) and its effect was fully reversible. The effects of H 2 S on MUA and GDPs may be linked to changes in cell excitability by its direct action on ion channels and receptors. Actually a number of studies have shown that NaHS induces hyperpolarization or depolarization of excitable cells. Activation of K ATP channels mediates H 2 S-induced membrane hyperpolarization and decrease of growth hormone release in rat pituitary GH3 cells (Mustafina et al., 2015), the inhibitory action of H 2 S on the breathing cycle in medullar neurons of neonatal rats  and during hypoxic injure in the medullar respiratory center (Pan et al., 2010). On the other hand H 2 S induced depolarization of paraventricular neurons of the hypothalamus (Khademullah and Ferguson, 2013), trigeminal neurons (Feng et al., 2013) and neurons of the subfornical organ (Kuksis et al., 2014;Kuksis and Ferguson, 2015).
In our experiments using NMDA-channels as membrane potential sensors together with cell-attached current-clamp recordings for analysis of membrane potential (see 'Experimental procedures' section) we found that NaHS (100 lM) and L-cysteine (1 mM) induced drastic depolarization of CA3 hippocampal pyramidal neurons of neonatal rats. Depolarization of pyramidal neurons may underlie the initial excitatory effect of NaHS. However, prolonged depolarization will produce inactivation of Na + channels and suppression of ongoing electrical and synaptic activity as previously shown in striatal spiny neurons after ischemia (Calabresi et al., 1999). Elevated Na + influx may induce a decrease in the membrane potential, however, this effect was not prevented by TTX in our experiments and similar results were obtained also in other studies (Kuksis and Ferguson, 2015;Malik and Ferguson, 2016). It was also shown the NaHS at concentrations of 1 and 10 mM increased I Na in neurons of entorhinal cortex and subfornical organ (Luo et al., 2014;Kuksis and Ferguson, 2015). In our preparation NaHS induced a rightward shift of activation and inactivation curves of I Na without affecting the current amplitude. The mechanism by which H 2 S modulates ion channel activity may relate to redox modification of channels proteins (Sitdikova et al., 2010;Tang et al., 2010). Indeed, the voltage dependence of activation and inactivation and the amplitude of I Na depend on thiol-reducing agents (Evans and Bielefeldt, 2000). The absence of NaHS effect on I Na peak can be explained by the low concentration used in our study (100 lM) and the peculiarities of expression and regulation of Na + channels in different brain regions during ontogenesis (Costa, 1996). Hence, depolarization of hippocampal neurons appears not related to Na + influx during H 2 S application. A possible mechanism which could explain depolarization may involve inhibition of different types of K + channels, including inwardly rectifying K + (Kir)channels, voltage-gated K + channels, Ca 2+ -activated K + channels or two-pore or leak K + -channels, which by decreasing K + efflux diminish their contribution to hyperpolarization and hence cause depolarization. Indeed in the paraventricular nucleus of the hypothalamus H 2 S-evoked depolarization of neurons was suggested to result from inhibition of K + currents (Khademullah and Ferguson, 2013). In neurons of rat trigeminal ganglion the increased excitability induced by H 2 S appears mediated by suppression of A-type K + currents (Feng et al., 2013). In contrast, in neurons of subfornical organ no effects of H 2 S on K + currents were observed (Kuksis and Ferguson, 2015). Moreover, a large number of studies reported an activating effect of H 2 S on different K + currents, (in particular on Ca 2+ activated BK channels) (Sitdikova et al., 2010(Sitdikova et al., , 2014Tang et al., 2010;Mustafina et al., 2015) which induces hyperpolarization of neurons Malik and Ferguson, 2016). In our experiments we observed a reduction of outward K + currents which was prevented by TEA. Such a mechanism may partially underlie the depolarization and increased neuronal excitability caused by H 2 S. On the other hand in the nucleus of the solitary tract it was shown that NaHS (5-10 mM) induced an inward current which underlies the depolarization and this current was not voltage sensitive (Malik and Ferguson, 2016). It was suggested that either ion transporters or ion channels without voltage sensitivity were responsible for the depolarization. Na + /K + -ATPase may be one of the targets of NaHS. However, the concentration of NaHS used in our study unlikely inhibits cell energy production (Erb and Althaus, 2014). Moreover, a decrease in the ATP production would activate of K(ATP) channels and induce hyperpolarization.
We cannot exclude the inhibitory effect of NaHS on Na + /K + -ATPase although the data about the direct inhibitory effect of NaHS on this pump is rare. An inhibitory effect of NaHS (1 mM) on Na + /K + -ATPase was shown in Xenopus lung epithelia (Erb and Althaus, 2014) and an indirect action of NaHS on Na + / K + -ATPase activity was shown in renal tubular epithelial cells where H 2 S activates epidermal growth factor receptors and causes subsequent endocytosis of Na + /K + -ATPase (Ge et al., 2014). In neurons of frog sympathetic ganglion where NaHS induces reversible depolarization Na + /K + ATPase activity in the presence of sulfide was not different to control (Reiffenstein et al., 1992).
In our experiments we revealed the decrease in membrane resistance during H 2 S application indicates activation of an unknown conductance which may be involved in the depolarizing action of H 2 S as was also shown in paraventricular nucleus of the hypothalamus (Khademullah and Ferguson, 2013) and the nucleus of the solitary tract (Malik and Ferguson, 2016).
We tested the hypothesis whether the depression of GDPs in response to H 2 S may result from its direct action on GABA and/or glutamate receptors underlying GDPs generation. Focal application of GABA, AMPA/ kainate or NMDA receptors agonists revealed that only NMDA receptor-mediated currents were affected by H 2 S. We found a decrease in the amplitude and area of NMDA receptor-mediated currents. This result suggests at least some contribution to the suppression of GDPs induced by H 2 S. Our results are controversial to the data obtained in the studies of Abe and Kimura (1996) where H 2 S in a concentration-dependent manner stimulated NMDA receptor-mediated currents in CA1 region of hippocampal neurons of 2-to 3-week-old rats by enhancing the sensitivity of NMDA receptors to glutamate. However, for inducing inward currents Abe and Kimura used a different protocol -bath application of 20 lM NMDA for 90 s without blocking GABA A and GABA B receptors. Furthermore, studies using NR1 and NR2A NMDA receptors subunits expressed in Xenopus oocytes showed that NaHS decreased the onset time of NMDA responses without changing of their amplitude and this effect was mediated by the production of cAMP (Kimura, 2000). However, it should be noted that at early postnatal development of the hippocampus in rodent brain principal neurons are abundant of NR2B subunits compared to mature neurons in the adult brain (Ko¨hr, 2006). During the first postnatal week in hippocampus NMDA receptors are composed of NR1-NR2B subunits and of NR1, NR2A, and NR2B in the adult brain (Chang et al., 2009). Thus, in our conditions NMDA receptors probably mainly contain NR2B subunits which can explain the differences to previous studies.
At early stages of the brain development hyperexcitability of neurons may increase the probability of seizure-like events (Ben-Ari et al., 2012). In this respect H 2 S may play a protecting role supporting the inhibitory/ excitatory balance in the developing brain. This suggestion is supported by the fact that the H 2 S-producing enzyme levels in the nervous system are developmentally regulated and the expression level of CBS and 3-MST increases from the late embryonic to the early postnatal period (Enokido et al., 2005). This assumes that CBS and 3MST are involved in the growth and maturation of neural networks.
Deficiency of CBS leads to hyper-homocysteinemia, a condition associated with elevated plasma homocysteine which is known as a risk factor for several neurological diseases, such as cognitive impairment, dementia or epilepsy (Mattson and Shea, 2003). High homocysteine level results in neural tube defects during CNS development in most animal models (Rosenquist and Finnell, 2001). Therefore, CBS might exert protection of neurons by reducing homocysteine concentrations and by producing H 2 S, which was shown to protect neurons from glutamate toxicity (Kimura and Kimura, 2004) by increasing the glutathione level  and inhibition of hypoxia-induced apoptosis in cultured hippocampal neurons (Luo et al., 2012).
A protective role of H 2 S was shown in rat models of recurrent febrile seizure (Han et al., 2005) which is the most common seizure type in children, often causing hippocampal damage. The expression of CBS in the hip-pocampus and the plasma level of H 2 S were dramatically increased by recurrent febrile seizures which may be a compensatory response to suppress neuronal hyperexcitability and thus diminish the neuronal damage in the hippocampus (Han et al., 2005). Our experiments support this notion by the finding of a complete depression of bicuculline-induced interictal-like activities in the neonatal rat hippocampus by NaHS which can be explained by network desynchronization due to depolarization of neurons. Our results are not in accordance with the results of Luo et al. (2014) where H 2 S aggravated both pentylenetetrazole -as well as pilocarpine-induced seizure-like events in rats (postnatal 15-25 days), whereas both low Mg 2+ /high-K + -and 4-aminopyridineinduced seizure-like events had shorter latency after preliminary H 2 S treatment of brain slices. However, in uˆin vitroy´experiments these authors recorded the activity of single cells without analysis of the network activity (Luo et al., 2014). Hence, the differences may be explained by the different protocols used and also may reflect agedependent effects of H 2 S.

CONCLUSION
Our experiments reveal a potential neuromodulatory and neuroprotective role of H 2 S in the developing nervous system by regulation of hippocampal neuronal excitability, which is supported by our finding that H 2 S completely depressed epileptiform activity in the hippocampus of neonatal rats. On the other hand excessive H 2 S production induces inhibition of neuronal activity which may have deleterious effects on neonatal brain development. Our results extend the knowledge on the physiological and pathophysiological functions of H 2 S in the immature brain and indicate that H 2 Sassociated signaling shall be targeted for the treatment of developmental disorders.