The neurotoxic effects of hydrogen peroxide and copper in Retzius nerve cells of the leech Haemopis sanguisuga

Reactive oxygen species (ROS), such as superoxide anion (O₂^•−), hydrogen peroxide (H₂O₂), and hydroxyl radical (HO^•), are produced continuously during normal cellular metabolism. There is growing evidence that H₂O₂ plays a role in normal cellular function and cell signaling (Dringen et al., 2005; Ray et al., 2012; Marinho et al., 2014), particularly in higher organisms. However, H₂O₂ shows a toxicological action, because it can produce the highly reactive HO^• and cause oxidative damage to biomolecules (Halliwell, 2006; Sabater and Martín, 2013). Compared with other ROS, H₂O₂ is relatively stable and also able of diffuse rapidly across cell membranes (Murrant and Reid, 2001). According to Halliwell et al. (2000), H₂O₂ may induce its deleterious action through direct oxidation of its substrate, or indirectly through its highly reactive byproduct HO^•. Given its low reactivity, H₂O₂ does not readily mediate oxidative injury, unless exposed to transition metal ions that can catalyze transformation of H₂O₂ to the aggressive radical, HO^• (Cohen, 1994).

The transition metal ions and their complexes in lower oxidation states were found to have the oxidative property of the Fenton reagent, and, accordingly, the mixtures of these metals with H₂O₂ were named ‘Fenton-like’ reagents (Goldstein et al., 1993). Copper (Cu) is redox-active metal which is capable of inducing oxidative injury by two different mechanisms. Firstly, in the presence of H₂O₂, it can catalyze the formation of HO^• which can induce substantial damage of biomolecules by removing hydrogen or by addition to unsaturated bonds (Simon et al., 2004). Secondly, exposition of cells to increased level of copper diminishes intracellular glutathione content. Several authors have suggested that divalent copper [Cu(II)] reacts with H₂O₂ to produce HO^•, which mediates oxidative damage. However, other researchers have disputed the formation of HO^• in reactions involving Cu(II) ions and H₂O₂, and the debate continues in the literature. Evidence of the role of copper in the production of ROS has been obtained mainly by in vitro study in which combinations of copper and a reducing agent were used. While a previous in vitro research has disclosed that Cu(II) ions are capable of reacting with H₂O₂ in a Fenton-like reaction (Gunther et al., 1995), it is still disputable whether this reaction occurs in vivo. The hydroxyl radical is small, highly mobile, water-soluble, and chemically the most reactive species of activated oxygen (Ayala et al., 2014). But, due to its very short half-life, it is effective only close to the locus of its generation. The short diffusion distance of the HO^• suggests that most metal-catalyzed oxidative modifications of proteins occur via the reaction of H₂O₂ with the sites of metals binding to the proteins (Sayre et al., 2005). HO^• is the most reactive and dangerous ROS since there are no enzymatic systems known to detoxify them. Due of its extraordinarily high reactivity, direct detection of HO^• in biological systems is very hard. For that reason, many researchers (Birinyi-Strachan et al., 2005; Huang et al., 2009a; Mokudai et al., 2012) used indirect methods for detecting and neutralizing this radical as well as antioxidants (mannitol, dimethylthiourea and dimethyl sulfoxide).

Electrophysiological analyses have shown that membrane transport proteins are susceptible to ROS. Considering neuronal network function, free radicals can attack ions channels either directly, or indirectly by causing peroxidation of membrane lipids (Carmeliet, 1999) and affecting associated signaling proteins (Hool, 2006). Oxidative modification of critical cysteine residues in redox-sensitive proteins has been proposed to constitute one of the major mechanisms that regulate protein structure and function (Zhang et al., 2006; Chung et al., 2013). In the light of this knowledge, surprisingly few experimental studies have focused on oxidative modification of K⁺ channels in nerve cells.

Having considered all the data noted above, the aim of this study was therefore to examine whether copper can enhance H₂O₂ toxicity and whether hydroxyl radical scavengers could protect leech Retzius nerve cells from toxicity induced by the H₂O₂/Cu(II) oxidizing system. The nervous system of invertebrates can be taken as a simple model for vertebrate brain studies in the aspect of synapse formation and plasticity (Schmold and Syed, 2012), and the neural basis of learning (Sahley, 1995). As multicellular organisms, the invertebrates represent a ‘simple model’, because their nervous systems are smaller and contain considerably fewer neurons than those found in brains of vertebrates. However, compared to vertebrate neurons, the structure and function of single nerve cells of invertebrates are equally complex (Burrell and Sahley, 2001; Bier and McGinnis, 2004). The main advantages of studying nerve cells in leech brain are the large sizes of Retzius nerve cells and their easy accessibility for electrophysiological recordings.

Prior to experimentation, Retzius neurons were tested in order to check whether they would keep their electrophysiological properties intact for a minimum of 20 min, to ensure that changes on their intrinsic and firing properties were strictly induced by drug administration and not by the effect of time. At this control stage, the neurons tested (n=5) were recorded during 20 min or more and no significant differences in their membrane properties were observed. Once the control process was finished, we proceeded with the recording of the study sample. Monitoring of the electrical properties was carried out on the Retzius cells which show a stable membrane potential, firing rate and spike potential duration for at least 10 min. Most Retzius nerve cells had a resting membrane potential of −45 to −55 mV. Retzius neurons with a resting membrane potential less than −40 mV in normal leech saline were considered damaged and data from such cells were not analyzed.

In the first series of experiments, we investigated the actions of H₂O₂ in the presence of Cu(II) on the resting membrane potential of Retzius nerve cells. They showed a stable membrane potential of −46.75±2.28 mV (see Table 1), and spontaneous action potential of duration, amplitude and shape typical for Retzius cells (duration 7.52±1.25 ms, amplitude 45.94±3.86 mV) were generated at a low frequency (2.43±0.26 APs/s). Combined application of 1 mM of H₂O₂ and 0.02 mM of Cu(II) induced a slow and continuous membrane depolarization. As shown in Fig. 1 and Table 1, for all observations (n=20), membrane potential was significantly depolarized 5 min after drug administration. Values slightly increased after 15 and 20 min. The average change in membrane potential was 4.58±1.4 mV in the 5 min exposure, 6.84±1.9 mV in the 15 min, and 7.41±2.1 mV after 20 min administration (Table 1). For a sample neuron, Fig. 1 illustrates the effect of 1 mM of H₂O₂ and 0.02 mM of Cu(II) on the resting membrane potential of Retzius nerve cell. After a 20 min perfusion, we observed a membrane potential depolarization of approximately 9 mV.

Exposure of leech segmental ganglia to H₂O₂ (1 mM) and Cu(II) (0.02 mM) altered the duration and shape of the action potentials of Retzius nerve cells. We found that the amplitude of action potentials decreased, while the duration augmented in a progressive way along the drug exposure time (Table 1). With prior administration of H₂O₂/Cu(II), action potential duration (APD) amounted to 7.52±1.25 ms. On average, the prolongation of action potentials amounted to 10.6±1.17 (n=16, P≤0.01) after a 20-min exposure.

In Fig. 2, we can observe the action potentials for a sample neuron that were obtained in control and experimental conditions. As seen in Fig. 2A, we found a widening effect in the spike. For a sample neuron, the duration of the action potential was 7.4 ms in control condition, 13.9 ms after 5 min, 17.5 ms after 15 min, and 19.4 ms after 20 min. Fig. 2A also demonstrates changes in the amplitude of the action potentials after H₂O₂/Cu(II) administration. The recordings show a progressive diminution of amplitude in the action potentials. In the control condition, amplitude measured 48 mV, it then decreased to 45 mV after 5 min and continued to diminish up to 42 mV after 15 min, and 37 mV after a 20-min exposure.

Our results demonstrate that most of the Retzius cells under study kept ability to discharge action potentials although their frequency decreased. Exposure of leech segmental ganglia to H₂O₂ and Cu(II) induced an initial excitation followed by depression of spontaneous electrical activity, and this effect was seen in 16 of 20 cells (Table 1). Furthermore, some of Retzius neurons lost their spontaneous activity. Thus, H₂O₂/Cu(II) led to the appearance of repetitive firing (Fig. 2B) only a few minutes after application of H₂O₂/Cu(II), which was followed by loss of excitability of the neurons. We found that 20% of Retzius cells under study showed a complete cancellation of discharge properties.

In order to examine the possibility that the broadening of action potentials of Retzius neurons were a consequence of the inhibition of the outward K⁺ current, needed for the repolarization phase of action potentials, we studied the effect of H₂O₂/Cu(II) on the outward K⁺ current. When Cu(II) (0.02 mM) was added to the leech Ringer, the concentration of the oxidant of 1 mM caused a strong inhibition of the K⁺ outward current. The typical response of a Retzius cell to H₂O₂ (1 mM) and 0.02 mM Cu(II) in an Na-free solution is depicted in Fig. 3. Outward K⁺ currents are generated by a series of depolarizing pulses, in the Tris Ringer (Fig. 3A) and after exposure of the Retzius neurons to 1 mM H₂O₂ and 0.02 mM Cu(II) for 10 min (Fig. 3B).

The current-voltage (I-V) relationships for the outward K⁺ current studied were generated by standardized pulse protocols and were obtained before and at various times after the addition of H₂O₂/Cu(II). Fig. 4 demonstrates the current-voltage relationship, separately, for the peak and the entrenched steady level of the K⁺ outward current. At the test potential of +22 mV, the fast and slow steady part of the K⁺ outward current dropped from 67 to 20 nA (70.15%) and from 39 to 12 nA (69.24%), respectively.

To elucidate the mechanism underlying the effect of H₂O₂/Cu(II) oxidative activity, the actions of various antioxidants on H₂O₂/Cu(II)-induced changes were examined. The following experiments were carried out to investigate the possible contribution of HO^• production by the Fenton reaction to H₂O₂-induced electrical changes in Retzius neurons. The data presented in Fig. 5 shows that the neurotoxic effect of H₂O₂/Cu(II) on spontaneous spike electrogenesis of the Retzius neurons was reduced in the presence of dimethylthiourea (DMU; 1 mM) and dimethyl sulfoxide (1%), but not mannitol (5 mM).

In the presence of DMU (1 mM), 1 mM H₂O₂ and 0.02 mM Cu(II) produced only a weak reduction of fast and slow potassium outward current of 12.5% and 10.26%, respectively (Fig. 6A), which was significantly smaller than the reduction of 70.15% and 69.24% in the absence of DMU. Pretreatment with another scavenger, DMSO (1%), partially blocked the effect of the H₂O₂/Cu(II) oxidation system on the outward K⁺ currents. Voltage clamp experiments revealed that 1 mM H₂O₂ and 0.02 mM Cu(II) in the presence of DMSO reduced the outward potassium current by 22%, respectively (Fig. 6B). At the test potential of +24 mV, the fast and slow part of the K⁺ outward current dropped by 21.43% (compared to 70.15% in the absence of the DMSO) and 23.81% (compared to 69.24% in the absence of the DMSO). In the final experiments, Retzius neurons were pretreated with mannitol (5 mM). In contrast to DMU and DMSO, mannitol did not prevent reduction of the outward K⁺ current induced by H₂O₂ and Cu(II) (Fig. 6C). In the presence of mannitol, 1 mM H₂O₂ and 0.02 mM Cu(II) caused strong inhibition of fast and slow components of 63.34% and 60.53%, respectively (compared with the inhibition of 70.15% and 69.24% in the absence of mannitol).

The main finding of the present study is the determination of the early alteration in electrophysiological properties of Retzius neurons, i.e. membrane potential, action potential and outward repolarizing K⁺ currents. Oxidative stress induced by H₂O₂/Cu(II) produced a slow and continuous depolarization of the membrane potential. These results concur with those found by Nakaya et al. (1992) in a study with cardiac cells, in which they proposed that depolarization was partly a consequence of the inhibition of activity in the K⁺ channels. In contrast, Ostrowski et al. (2014) observed that H₂O₂ induced hyperpolarization of resting membrane potential of nucleus tractus solitarii neurons. Similar findings have been reported in rat CA1 pyramidal neurons where 3.3 mM of H₂O₂ caused a membrane hyperpolarization by inducing an increase in potassium conductance (Seutin et al., 1995). Bychkov et al. (1999) demonstrated that H₂O₂ induced both depolarization and hyperpolarization of the membrane potential via two various mechanisms. Low concentrations of H₂O₂ elicited membrane potential depolarization, mediated through the inhibition of inward-rectifying K⁺ channels, whereas higher H₂O₂ increased the amplitude of the Ca²⁺-activated K⁺ current and thus induced hyperpolarization.

We observed that the amplitude of action potentials decreased, while duration increased in a progressive way along the drug exposure time. Similar data were found in studies on rat intrinsic cardiac ganglia (Whyte et al., 2009) and cortical pyramidal neurons (Pardillo-Díaz et al., 2015) from the effects of ROS donors. The study by Du et al. (2000) suggests that down-modulation of delayed rectifier potassium currents in hippocampal neurons results in the prolongation of action potentials. Hasan et al. (2013) identified HO^• as the intermediate oxidant responsible for H₂O₂-induced inhibition of the delayed rectifier K⁺ current in the hippocampus through the oxidation of sulfhydryl groups of intracellular cysteine residues. Studies carried out on guinea pig ventricular myocytes have revealed that H₂O₂ prolongs action potentials by increasing the late sodium current (Song et al., 2006). In a study on CA1 hippocampus neurons, Angelova and Müller (2006) demonstrated oxidative inhibition of the voltage-gated transient (I_A) and delayed rectifier [I_K(V)] potassium currents by H₂O₂.

The accumulating evidence suggests that ROS modulate neuronal excitability. It has been shown that combined application of H₂O₂ and Cu(II) disrupts both the intrinsic excitability of the neuron and the ability of generating action potentials. These findings are consistent with previous research performed in the CA1 area of the hippocampus, where H₂O₂ suppressed neuronal activity, and this effect was emphasized by co-application of H₂O₂ and Fe²⁺, proposing that the HO^• produced by the Fenton reaction mediates the action of H₂O₂ on hippocampal neuronal activity (Garcia et al., 2011). This result goes in line with the one obtained by Pardillo-Díaz et al. (2015) in which cumene hydroperoxide (10 µM; 30 min) produced a complete cancellation of discharge properties in the pyramidal neurons. In contrast, Ostrowski et al. (2014) observed, initially, a decrease of discharge of nucleus tractus solitarii neurons that is followed by sustained hyperexcitability. A study carried out on myenteric neurons (Pouokam et al., 2009) has found that H₂O₂ inhibited the Na⁺ channels, hyperpolarized the cell membranes and increased the cytosolic Ca²⁺ concentration. In our study, the HO^• seems to be involved in the mechanism of the H₂O₂/Cu(II) action. Scavenging of this radical by dimethylthiourea and dimethyl sulfoxide diminished the inhibitory effect of the H₂O₂/Cu(II) on the K⁺ channels (Fig. 5). In principle, the hydroxyl radical can induce either direct modulation of the K⁺ channels or indirect via channel-associated regulatory proteins.

Voltage-clamp experiments in the present study revealed a strong reduction of the outward K⁺ current. Outward ion currents responsible for repolarization the membrane potential use separate voltage-dependent and Ca²⁺-dependent potassium channels. Electrophysiological studies have shown that there are three classes of K⁺ channels in Retzius nerve cells of the leech: (1) the fast, (2) the slow Ca²⁺-activated K⁺ channel, (3) the late voltage regulated K⁺ channels (Beleslin et al., 1988). According to Stewart et al. (1989) in Retzius cells three principal K⁺ currents have been identified: (1) a transient, inactivating A-type current (I_A), (2) a Ca²⁺-activated K⁺ current (I_C), (3) the delayed rectifier currents (I_K1 and I_K2). Outward repolarizing K⁺ currents play a fundamental role in determining neuronal excitability and action potential duration. Normally, Retzius nerve cells are spontaneously active, firing at a quite regular rate of 0.2-3 APs/s (Beck et al., 2001). Oxidative modifications of outward K⁺ channel activity lead to changes in action potential duration and the spontaneous electrical activity.

There is accumulating evidence that regulation of ion channels by cellular redox potential may be a significant determinant of channel activity. Nevertheless, the studies have given contradictory results on whether oxidative modification increases or decreases the ion channel activity. Several studies have found that H₂O₂ enhanced the Ca²⁺-activated K⁺ channel activity (Bychkov et al., 1999; Dong et al., 2008; Huang et al., 2009b; Feng et al., 2012; Zhang et al., 2012). Other studies reached to the opposite conclusions (DiChiara and Reinhart, 1997; Zhang et al., 2006). DiChiara and Reinhart (1997), for example, reported that the oxidizing agent H₂O₂ decreases the activity of the human brain Ca²⁺-activated K⁺ channels (hslo), whereas the reducing agent DTT enhances and stabilizes the activity of K⁺ channels. Liu et al. (2009), investigated the effect of H₂O₂ on the ‘big conductance’ (BK) channels using the patch-clamp technique. They showed that BK channel activity was inhibited in inside-out patches, whereas it was increased in cell-attached configuration. Soto et al. (2002) found that the H₂O₂-induced oxidative modification of the Ca²⁺-dependent K⁺ channels was mediated by the formation HO^• and identified cysteine residues (probably located on the cytosolic side of the channel protein) as one of the targets responsible for channel inhibition. These contrary data might be due to the type channels or the type ROS generating system. However, oxidative modification of the K⁺ channels is a complex process that is not fully understood.

The effect of H₂O₂/Cu(II) on the electrical properties of Retzius nerve cells in our study, seemed to be mediated by the HO^•, as Cu(II), which leads to the production of this radical from H₂O₂, and strongly potentiates the influence of an ineffective concentration (1 mM) of the H₂O₂. Previous results showed that the oxidizing agent H₂O₂ did not significantly change the electrical properties of Retzius neurons (Jovanovic and Jovanovic, 2013). To confirm whether the mechanism action of the H₂O₂/Cu(II) system on action potential duration and K⁺ channels is in any way attributable to HO^•, the action of the specific scavengers were investigated. It was indirectly shown that an H₂O₂/Cu(II) system elicited oxidative stress, through the examinations with antioxidants where we showed the partial reversal of H₂O₂/Cu(II)-induced outward K⁺ channels' inhibition. Also, that the HO^• is involved in modification of the electrical properties of Retzius neurons with H₂O₂/Cu(II) was demonstrated in the experiments where DMU almost completely prevented and DMSO partially inhibited the effects of H₂O₂/Cu(II) on Retzius cells. Mannitol was less effective in prevention of prolongation of action potential duration and inhibition of outward K⁺ channels. Complete protection was not seen with any of the antioxidants used in this study; however, pretreatment with DMU and DMSO significantly reduced the prolongation of action potential duration and the inhibition of the outward K⁺ current, while mannitol did not prevent the inhibition of the outward K⁺ current induced by H₂O₂/Cu(II).

There are at least two possible explanations for the incomplete recovery of the action potential duration and the outward K⁺ current in the presence of HO^• scavengers. One possibility is that H₂O₂ in reaction with Cu(II) ions forms two types of oxidizing species, namely HO^•, and also some kind of copper-oxygen complex that antioxidants do not scavenge. This agrees with the findings of Kawanishi et al. (2002), who found that the oxidative DNA damage by the H₂O₂/Cu(II) was induced by the generation of ROS, such as HO^• and a copper-oxygen complex with similar reactivity to HO^•. It is possible that the HO^• as well as the equally reactive alkoxy radicals produced by the Cu(II)-catalyzed Fenton reaction may be partly responsible for the observed prolongation of action potential duration and for the inhibition of the outward K⁺ channels in our study. The other explanation for the incomplete protection of the action potentials and outward K⁺ channel in the presence of HO^• scavengers is that a large portion of the oxidants formed are site-specific, i.e. the metal ion producing the radical is attached to ion channel proteins, in such a way that a scavenger has no possible chance of interfering before the radical has hit the target. Several previous studies have reported that DMSO exhibits dual behavior: as an antioxidant and proxidant. DMSO is an antioxidant able to decrease both protein oxidation and lipid peroxidation in rat brain. It also has the ability to trap hydroxyl and hydroperoxyl radicals (Sanmartín-Suárez et al., 2011). However, DMSO also revealed pro-oxidant characteristics, which is a result of the specific activity of DMSO with thiol groups in proteins (Liu et al., 2009). In this study mannitol did not show protective effects, which is probably related to its low permeability and its limited distribution to sites where HO^• is generated. The most likely explanation is that some ROS produced from H₂O₂/Cu(II) are not neutralized by this antioxidant. The study by Winter et al. (2005) suggests that mannitol is the relatively weak HO^• scavenger, which does not have adequate access to the reactive oxygen metabolites.

According to the obtained results, we concluded that the prolongation of the action potentials of Retzius neurons is the result of the effect of the H₂O₂/Cu(II) system on outward K⁺ channels. Outward repolarizing K⁺ currents are critical determinants of membrane excitability and action potential firing in neuronal cell (Duprat et al., 1995). Accordingly, oxidative modifications of K⁺ channels results in brain hyperexcitability and eventually cell death. Studies have shown that ROS-mediated oxidation of K⁺ channels is a cause of reduced cognitive function during normal aging (Alshuaib et al., 2001; Cotella et al., 2012; Wu et al., 2013) and neurodegenerative diseases including Alzheimer's and Parkinson's (Cai and Sesti, 2009).

The oxidative mechanism proposed in the present study might have a wider importance not only to simple invertebrate systems such as leeches, but also, similarly, to the brains of mammals. A better understanding of oxidative modifications of ion channels may allow the development of new and specific ion channel therapies in the treatment of brain disorders such as neurodegeneration, epilepsy, and pain.

The experiments were performed on Retzius nerve cells in isolated leech segmental ganglia at room temperature (20-25°C). Leeches of the species Haemopis sanguisuga were obtained from local commercial suppliers. All experimental protocols were approved by the Animal Research Ethics Committee (School of Medicine, University of Belgrade, Serbia). The experimental procedure complies with institutional research council guidelines. The leeches were first anaesthetized in 10% ethanol. Then, the ventral nerve cord was removed from the animal in short segments of four ganglia via a ventral longitudinal incision. Dissected segments were immediately transferred to a 2.5 ml plastic chamber containing a leech Ringer solution (for the composition, see Solutions) and fixed by means of fine steel clips. The plastic chamber was then placed in a grounded Faraday cage mounted on a fixed table in a manner that prevents vibrations. Each segmental ganglion contains approximately 200 pairs of neurons. The largest neurons in the leech central nervous system are Retzius cells (40-60 µm in diameter) which exhibit stable resting membrane potential and which are nonbursting neurons with a low spontaneous firing rate. It is well known that the resting potential of Retzius nerve cells of medical and horse leeches is lower then in other neurons. The resting potential of the Retzius cells ranges from −40 to −60 mV (Lent, 1977; Beleslin et al., 1988; Angstadt, 1999) and the action potentials were generally between 20 and 50 mV and did not overshoot. Because Retzius cells are large and easily identifiable they must be among the most thoroughly investigated single nerve cells (Lent, 1977).

Transmembrane action potentials were recorded with conventional microelectrode techniques. Isolated cells were impaled with glass microelectrodes pulled from 1.5 mm borosilicate glass (1.5 mm outside diameter, 0.6 mm inside diameter, Clark Electromedical Instruments, Edenbridge, UK) and filled with a 3 M KCl to give final resistances of 15-20 MΩ. A microelectrode was dipped into the solution and 20-30 min were allowed for equilibration. The recordings were amplified using a Bioelectric Instrument DS2C high input resistant amplifier. Microelectrodes were connected to the amplifier via an Ag-AgCl junction. The ground electrode was an Ag-AgCl wire in a separate chamber filled with Ringer solution connected to the experimental chamber by a 3 M KCl 3% agar bridge.

Activity K⁺ channels were studied in the Retzius nerve cells by using the voltage-clamp technique. Long-lasting depolarizing pulses (to 300 ms) in sodium free Tris Ringer in the neurons where the holding potential was more negative than −40 mV induced a progressive decay of the outward current. The data were leak corrected by using hyperpolarizing pulses of equal magnitude and by assuming a constant leak conductance. Command pulses were derived from a Tektronix 161 pulse generator. Voltage and current records were displaced on a Tektronix 564 oscilloscope. Data were acquired by a Digidata 1200 analog-to-digital board (Axon Instruments, Jakarta, Indonesia), and stored for analysis in a computer. Duration of an action potential was determined at the 90% level of repolarization. Amplitude was the voltage increment between the resting level and spike voltage peak.

Leech Ringer solution composed of (mM): NaCl, 115; KCl, 4; CaCl₂, 2; Na₂HPO₄, 1.2; NaH₂PO₄, 0.3 (pH 7.2). In the Na⁺-free Ringer, 115 mM NaCl was completely replaced with an equal amount of Tris (hydroxymethyl) aminomethane-Cl (Tris Ringer), and Na₂HPO₄ and NaH₂PO₄ were omitted. Pharmacological agents were prepared and dissolved immediately before application in the physiological salt solution at the concentrations stated. H₂O₂-containing solutions were prepared fresh, just before each experiment by dilution of a 30% H₂O₂ stock solution (Zorka Pharma, Sabac, Serbia) and added to the Ringer solution (or Tris-Ringer solution) at a final concentration of 1 mM. The CuCl₂ (Sigma, St. Louis, MO, USA) concentration was 0.02 mM. The mannitol, dimethylthiourea and dimethyl sulfoxide were obtained from Sigma, and added to the Ringer solution at final concentrations of 5 mM (mannitol), 1 mM (dimethylthiourea) and 1% (dimethyl sulfoxide). The Retzius nerve cells were treated for 20 min with H₂O₂/Cu(II) in the presence or absence of mannitol, dimethylthiourea and dimethyl sulfoxide. To change solutions the chamber was flushed continuously with a volume of fluid at least 10 times that of the chamber volume.

Results are given as mean±standard deviation (s.d.) with the number (n) of investigated neurons. Statistical analysis was made using Student's t-test. P values <0.05 were considered significant.