Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Acute bath application to hippocampal slices of the neurotoxic metal MeHg causes a concentration- and time-dependent biphasic effect on synaptic transmission in the CA1 region. Initially MeHg increases the amplitudes of field potentials recorded extracellularly (Yuan and Atchison, 1993, 1994) and EPSPs recorded intracellularly before suppression of them to block (Yuan and Atchison, 1995a). MeHg also blocks the recurrent IPSPs (Andersen et al., 1964a,b) in the CA1 region. IPSPs appeared to be more sensitive to MeHg than EPSPs, because block of IPSPs occurred earlier than did block of EPSPs. The time to the early suppression of IPSP amplitudes appeared to correspond to the onset of the early increase in amplitudes of both population spikes and EPSPs. This suggests that the reduced IPSPs contribute to the early increase in amplitudes of population spikes and EPSPs. MeHg suppresses the GABA-induced chloride current in dorsal root ganglion cells (Arakawa et al., 1991) and modulates the muscimol-induced increases in the [³H]flunitrazepam binding to GABA_A receptors in washed cerebellar membranes (Komulainen et al., 1995). Thus, we hypothesized that block by MeHg of GABA_A receptor-mediated inhibitory synaptic transmission results in disinhibition of hippocampal excitatory synaptic transmission, and is at least partly responsible for the initial stimulatory effects of MeHg on CA1 hippocampal synaptic transmission. However, MeHg also caused biphasic changes in resting membrane potentials, i.e., initial hyperpolarization and then depolarization of pyramidal CA1 neurons in hippocampal slices (Yuan and Atchison, 1995a) and rat forebrain synaptosomes (Hare and Atchison, 1992). This effect alone could influence the observed changes in IPSP and EPSP amplitudes. Thus, nonspecific effects of MeHg on resting membrane potentials may also be involved in its early effects on synaptic transmission.

To test this hypothesis, extracellular recordings of population spikes, intracellular recordings of EPSPs and IPSPs and single-microelectrode voltage-clamp recordings of EPSCs and IPSCs from CA1 pyramidal neurons were compared with or without pretreatment of hippocampal slices with bicuculline, a GABA_A antagonist. We sought to determine: 1) whether or not MeHg and inorganic mercury (Hg⁺⁺) affect IPSPs and EPSPs in hippocampal CA1 neurons differentially, because they differentially affect GABA-mediated chloride currents in dorsal root ganglion neurons (Arakawa et al., 1991; Huang and Narahashi, 1996) and field potentials recorded in CA1 neurons of hippocampal slices (Yuan and Atchison, 1994); 2) whether or not the differential block by MeHg of IPSPs and EPSPs is due to nonspecific effects of MeHg on resting membrane potentials and 3) whether or not block of GABA_A-mediated IPSPs is primarily responsible for the early stimulation of hippocampal synaptic transmission. Because this early stimulation is a characteristic of the effects of MeHg induced on central synaptic transmission, we sought to understand how this effect occurs, and how it pertains to the overall process of MeHg-induced neurotoxicity.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Materials. Methylmercuric chloride, purchased from ICN Biomedical, Inc. (Costa, CA), was dissolved in deionized water to a final concentration of 5 mM to serve as stock solution. The applied solutions (4-500 µM) were diluted with ACSF with the following composition (in mM): NaCl 124; KCl 5; MgSO₄ 2; KH₂PO₄ 1.25; NaHCO₃ 26; CaCl₂ 2 and D-glucose 10; pH was set at 7.4 just before superfusion. MeHg and other chemicals were applied acutely to slices by bath application at a rate of 1.2 to 1.5 ml/min with a Gilson (Middleton, WI) infusion pump. Strychnine hydrochloride and muscimol were purchased from Sigma Chemical Co. (St. Louis, MO). Muscimol (25-100 µM) was applied to slices for 15 to 30 sec at an interval of 10 min to avoid desensitization of GABA_A receptors. CNQX, DNQX, AP-5 and (-)-bicuculline methbromide were obtained from Research Biochemical International (Natick, MA). CNQX or DNQX were dissolved first in DMSO and then diluted further with ACSF. The final concentration of DMSO in the applied solution was less than 0.02% (v/v), which has no significant effects on synaptic transmission.

Preparation of hippocampal slices. Hippocampal slices were prepared as described previously (Yuan and Atchison, 1993). Briefly, hippocampi isolated from brains of male Sprague-Dawley rats (125-150 g, Harlan Industries, Madison, WI) were sectioned transversely at 4°C to slices of approximately 400-µm thickness. Slices were transferred immediately to a recording chamber and incubated for at least 60 min before electrophysiological recording. A humidified gas mixture of 95% O₂/5% CO₂ was circulated over the slices continuously by bubbling in the bath water. During recording, two or three slices were kept in the chamber at a given time. The rest were maintained in a reservoir chamber for later use. All experiments were conducted at 33 to 35°C. All experiments were replicated at least four times, and no more than one slice from a given rat was used for a particular experiment.

Electrophysiological procedures. Conventional extracellular and intracellular recordings were made in the CA1 region of the hippocampal slice. Monopolar tungsten electrodes (3 M, FHC, Brunswick, ME) were used as stimulation electrodes. Borosilicated glass microelectrodes (o.d. 1.0 mm; i.d. 0.5 mm, WPI, Inc., New Haven, CT) filled with ACSF (5-15 M) or 3 M potassium acetate (80-120 M) were used for extracellular or intracellular recording, respectively. Population spikes were evoked by orthodromic-stimulation of Schaffer collaterals at an intensity level (usually 2-4 V) that gives a population spike amplitude approximately 50% of the maximum amplitude as evoked by maximum stimulation. Intracellular EPSPs were recorded at CA1 cell soma by subthreshold stimulation (0.2 Hz) of Schaffer collaterals; typically a 0.1 to 0.2 nA negative D.C. current was applied through the recording electrode to maintain the cell membrane in a somewhat hyperpolarized state to avoid evoking action potentials. The recurrent IPSPs (Andersen et al., 1964a, 1964b) were recorded by subthreshold stimulation of the alveus. IPSCs and EPSCs were recorded using single-microelectrode voltage clamp techniques (Johnston et al., 1980; Johnston and Brown, 1981, 1984). The sample frequency was set at 8 kHz or as high as possible. When measuring the current-voltage relationship, voltage step commands were generated from an internal step command generator and manually controlled by the thumbwheel switch on the front panel of an Axoclamp-2 amplifier. For each voltage step, the cell was held at that potential for 30 to 40 sec to obtain at least three to five traces of IPSCs. The membrane input resistance was monitored by D.C. current injection through the recording electrode. All stimulus pulses were generated from a Grass S88 stimulator (Grass, Inc., Quincy, MA) at 0.2 Hz and 0.1-msec duration and isolated with a Grass SIU5 stimulus isolation unit (Grass, Inc.). Recorded signals were amplified (Axoclamp-2, Axon Instruments Inc., Foster City, CA), displayed on a 2090-3 digital oscilloscope (Nicolet Instruments, Verona, WI) and recorded simultaneously to both floppy disks and magnetic tape by using a FM instrumentation recorder (model B, Vetter Instruments, Rebersburg, PA) for later analysis. All measurements in this reported paper were made based on the peak amplitude of response.

Statistical analysis. Data were collected continuously before and during application of MeHg and analyzed statistically using Student's t test or paired t test or a one-way analysis of variance; Dunnetts' procedure was used for post hoc comparisons (Steel and Torrie, 1980). Values were considered statistically significant at P < .05.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Comparative effects of MeHg and Hg²⁺ on IPSPs and EPSPs or IPSCs and EPSCs. As shown in our previous report (Yuan and Atchison, 1995a), 100 µM MeHg blocked IPSPs more rapidly than it did EPSPs; times to block were 25 ± 2 and 45 ± 3 min, respectively (fig. 1, top). In some slices, both EPSPs and IPSPs were recorded simultaneously in the same neuron. In these recordings, an early increase in EPSP amplitude or even firing of action potentials often accompanied the decrease in IPSP amplitude at the early times of application of MeHg. At the same concentration, Hg⁺⁺ blocked IPSPs with a time course similar to that of MeHg. However, Hg⁺⁺ blocked EPSPs (63 ± 10 min) even more slowly than did MeHg.

View larger version (K): [in this window] [in a new window]   Fig. 1.   Comparison of times to block of recurrent IPSPs and EPSPs (top) or IPSCs and EPSCs (bottom) by 100 µM MeHg or Hg++. IPSPs and EPSPs were recorded in CA1 pyramidal cell soma by stimulating the alveus or Schaffer collaterals. IPSCs were recorded at their resting membrane potential levels in the presence of 20 µM AP-5 and 10 µM DNQX. EPSCs were evoked by presynaptic stimulation of Schaffer collaterals and recorded at CA1 neuronal soma at their resting potentials levels. All values are the mean ± S.E. of 5 to 12 individual experiments. The asterisk indicates a significant difference between times to block of IPSPs and EPSPs or IPSCs and EPSCs (P < .05). The black dot indicates a significant difference between times to block of EPSCs by MeHg and Hg++ [(P < .05, Student's t test)].

View larger version (K): [in this window] [in a new window]   Fig. 2.   Time course of effect of 100 µM MeHg on EPSCs recorded from CA1 neurons of hippocampal slices using single-microelectrode voltage-clamp techniques. EPSCs were evoked by electrical stimulation Schaffer collaterals and recorded at the CA1 pyramidal cell soma that was voltage-clamped at its resting membrane potential (77 mV in this case). Each trace is a representative depiction of four experiments which show the biphasic effects of MeHg on EPSCs.

Comparative effects of MeHg and Hg⁺⁺ on current-voltage relationship of IPSCs. Figures 3 compares the effects of 100 µM MeHg and Hg⁺⁺ on a family of IPSCs evoked at potentials of 40 to 90 mV. Figure 4 depicts the current-voltage relationship (I/V curve) for these IPSCs. The IPSC reversal potential is approximately 75 mV in the absence of MeHg which is close to the equilibrium potential for Cl as predicted by the Nernst equation, and similar to these values obtained by Benardo (1993) and Pitler and Alger (1994), indicating that these IPSCs are primarily GABA_A receptor-mediated chloride currents. MeHg suppressed both outward and inward currents, this effect usually started after 5 min of application of 100 µM MeHg. As shown in Figures 3 and 4 (left) exposure of slices to 100 µM MeHg for 15 min, resulted in depression of all IPSCs evoked at holding potentials of 40 to 90 mV. The I/V curve and the reversal potential were shifted to more positive potentials. In contrast, whereas 100 µM Hg⁺⁺ suppressed both outward and inward currents, Hg⁺⁺ initially caused an increase in the outward current prior to suppressing it. Moreover, Hg⁺⁺ shifted the I/V curve and the reversal potential to a more negative potential direction (fig. 4). At 20 µM the respective effects of MeHg or Hg⁺⁺ were similar but the latency to onset of action was much longer than at 100 µM (results not shown).

View larger version (K): [in this window] [in a new window]   Fig. 3.   Comparison of effects of 100 µM MeHg (A) or Hg++ (B) on IPSCs recorded at different holding potentials. IPSCs were evoked by presynaptic stimulation of Schaffer collaterals and recorded at CA1 neuronal soma of hippocampal slices in the presence of 20 µM AP-5 and 10 µM DNQX. Time 0 min represents IPSCs recorded at different holding potentials before application of MeHg or Hg++. Time 15 or 25 min indicates the time after beginning perfusion of the slice with 100 µM MeHg or Hg++. Calibration bars: vertical, 300 pA; horizontal, 50 msec.

View larger version (K): [in this window] [in a new window]   Fig. 4.   Comparison of effects of 100 µM MeHg (left) and Hg++ (right) on the current-voltage relationship (I/V curve) of hippocampal CA1 pyramidal cell IPSCs. IPSCs were normalized to their resting membrane potentials. Values are the mean ± S.E. of four to five individual experiments.

Comparative effects of MeHg and bicuculline on population spikes. Because MeHg suppresses the GABA_A-mediated chloride currents in hippocampal CA1 neurons, we sought to determine if its effects are similar to those of bicuculline, a selective GABA_A receptor antagonist. To test this, we compared the effects of 20 to 500 µM MeHg and 10 µM bicuculline on population spikes. We used these higher concentrations of MeHg because we previously showed that the higher concentrations of MeHg induced a more rapid and noticeable increase in population spikes which was often accompanied by repetitive firing (Yuan and Atchison, 1993). At 20 to 500 µM, MeHg caused a concentration- and time-dependent early increase in amplitudes of population spikes prior to blocking them (fig. 5). Higher concentrations (100 and 500 µM) of MeHg induced repetitive firing in response to single shock stimuli, suggesting that membrane excitability was increased. The early stimulatory effects of MeHg on population spikes were similar to those of bicuculline on population spikes.

View larger version (K): [in this window] [in a new window]   Fig. 5.   Comparison of effects of MeHg and bicuculline on hippocampal CA1 pyramidal cell population spikes. From top to bottom are time courses of effect of 20 (A), 100 (B) and 500 µM (C) MeHg and 10 µM (D) bicuculline on population spikes. Each trace is a representative depiction of 6 to 10 experiments. Calibration bars: vertical, 5 mV; horizontal, 20 msec.

Effects of bicuculline pretreatment on MeHg-induced early stimulation of synaptic transmission. The early stimulation of excitatory synaptic transmission may be due primarily to MeHg-induced suppression of GABA_A receptor-mediated chloride currents. This in turn may lessen the inhibitory effects of interneurons on excitatory synaptic transmission. If so, then pretreatment of slices with bicuculline to block GABA_A receptor-mediated chloride currents should eliminate or suppress the MeHg-induced early increases in population spike amplitudes. To test this, we compared effects of 20 and 100 µM MeHg on population spike amplitudes in the presence or absence of bicuculline. Incubation of slices with 10 µM bicuculline for 5 to 10 min increased the amplitudes of population spikes significantly; moreover the single spike response gradually changed to multiple spike responses. After 30 to 60 min of bicuculline, population spike amplitudes typically increased to and stabilized at 150 to 200% of control. At this point, two sets of experiments were designed to examine the effects of bicuculline on the early stimulation by MeHg of excitatory synaptic transmission. In the first set of experiments, 20 or 100 µM MeHg plus 10 µM bicuculline were added to the ACSF with no change in stimulus intensity. Under these conditions, MeHg still suppressed population spike amplitude as did MeHg alone, but in four of six slices caused no further significant early increase in population spike amplitudes (fig. 6, left). The second set of experiments was performed under reduced stimulus intensity. The reason for doing this was that we were concerned that pretreatment of slices with bicuculline might increase population spike amplitudes to a ceiling amplitude, above which MeHg was unable to cause further increase, thus masking the actual effect of MeHg on population spikes. Thus, the stimulus intensity was reduced to a level that gave population spike amplitudes approximately equal to the control level before bicuculline treatment, after the bicuculline-induced increase had stabilized. MeHg (20 or 100 µM) plus 10 µM bicuculline were then applied to the slices. As seen with the results of the first set of experiments, MeHg did not cause any statistically significant early increase in population spike amplitudes but reduced or blocked completely population spikes in three of four and five of seven slices at 20 and 100 µM MeHg, respectively (fig. 6, left). In the remaining slices there was a 10 to 15% early increase in population spike amplitudes before block by MeHg. This effect was not significant, and is masked in Figure 6 due to averaging of the time courses from the individual experiments. Without pretreatment of slices with bicuculline, 20 and 100 µM MeHg caused the typical biphasic changes in amplitudes of population spikes, although the early increase in amplitude induced by 20 µM MeHg was not as prominent as that caused by 100 µM MeHg (fig. 6). Due to variations in time course of effects of MeHg among the individual experiments, figure 6 does not show any decrease in population spike amplitudes after exposure to 20 µM MeHg alone for 120 min. However, prolonging exposure of slices to 20 µM MeHg to 150 to 180 min, caused block of all population spikes (results not shown). It appears that MeHg blocked responses more rapidly in slices treated with bicuculline than in slices not pretreated with bicuculline. To test if bicuculline would prevent early increases in EPSP amplitude induced MeHg, effects of MeHg on EPSPs were examined in the presence of 10 µM bicuculline. Normally, EPSPs were evoked by subthreshold stimulation of Schaffer collaterals to avoid generation of action potentials. After application of 10 µM bicuculline for 30 to 60 min, EPSP amplitude increased dramatically and induced multiple spikes (fig. 7). Once the increase in EPSP amplitude reached a stable level, the stimulus intensity was then reduced to a level that gave a measurable EPSP but did not initiate action potentials. It was generally quite difficult to do this after pretreatment of slices with bicuculline, because either action potentials were generated or the EPSP at a given stimulus was not measurable. Thus we were only able to obtain a few successful recordings for this experiment. However, in those experiments MeHg failed to cause a significant early increase in EPSP amplitude in the presence of 10 µM bicuculline as was seen in Figure 6 for field potential recordings. Thus the early stimulatory effects of MeHg on hippocampal transmission appear to be related to its actions on GABA_A receptors.

View larger version (K): [in this window] [in a new window]   Fig. 6.   Time courses of effects of MeHg on CA1 hippocampal pyramidal cell population spikes with or without pretreatment of slices with bicuculline (BI) or strychnine (ST). Left, Slices were pretreated with 10 µM bicuculline or 50 strychnine for 30 to 60 min (omitted in this graph), and then superfused with ACSF containing 20 (top) or 100 µM (bottom) MeHg and 10 µM bicuculline or 50 µM strychnine without changes in the stimulation intensity until 120 min or complete block of population spikes occurred. The starting time point for superfusion of MeHg, as indicated by the downward arrow, was defined as 0 min and the data collected at this point were defined as 100%. Right, After slices were pretreated with 10 µM bicuculline or 50 µM strychnine for 30 to 60 min (omitted in this graph), the stimulation intensity was reduced to a level that gave amplitudes of population spikes roughly approximate to their pretreatment values. This time point was defined as 0 min and the data collected at this point were defined as 100%. Slices were then superfused with ACSF containing 20 µM (top) or 100 µM (bottom) MeHg and 10 µM bicuculline or 50 µM strychnine, as indicated by the downward arrow, until 120 min or block of population spikes occurred. All values are the mean ± S.E. of 6 to 13 individual experiments.

View larger version (K): [in this window] [in a new window]   Fig. 7.   Time course of effects of 100 µM MeHg on hippocampal pyramidal cell EPSPs evoked in the presence of bicuculline. EPSPs were recorded from a CA1 neuron of a hippocampal slice by subthreshold stimulation of Schaffer collaterals. After perfusion with 10 µM bicuculline for 30 min (from 30 to 0 min), EPSPs were increased greatly with multi-spikes on them (note that the vertical calibration bar for the trace at 10 min is different from that for traces at other time points). At 0 min, the stimulation intensity was reduced to a level that failed to evoke any spikes. The slice was then superfused with ACSF containing 100 µM MeHg and 10 µM bicuculline until complete block of EPSPs occurred. Each trace is a representative depiction of five recordings.

View larger version (K): [in this window] [in a new window]   Fig. 8.   Effects of 20 µM bicuculline and 100 µM MeHg on muscimol-evoked responses in a hippocampal pyramidal cell. Conventional intracellular microelectrode recordings were made in CA1 pyramidal cells in the presence of 20 µM AP-5 and 10 µM DNQX in ACSF. Bath application to slices of 100 µM (top) or 50 µM (bottom) muscimol for 15 sec caused a depolarization response. Top, Effects of application of 20 µM bicuculline for 10 min (B10 min) and 20 min (B20 min) on 100 µM muscimol-evoked responses. After the muscimol-evoked response was blocked completely (B20), washing the slice with bicuculline-free ACSF for 20 min (W20) partially reversed it. Bottom, Time course of effects of 100 µM MeHg on 50 µM muscimol-evoked response. Arrows indicate the starting point of application of muscimol. Each trace is a representative depiction of three to nine experiments. Calibration bars: vertical, 20 mV; horizontal, 100 second.

Effects of strychnine on MeHg-induced early stimulation of CA1 synaptic transmission. GABA is generally believed to be the major inhibitory transmitter in the mammalian central nervous system. However, glycine also serves as an inhibitory transmitter in the central nervous system, especially in the spinal cord and brain stem (Aprison and Daly, 1978; Pycock and Kerwin, 1981; McCormick, 1990). Additionally, glycine can potentiate the action of glutamate at NMDA receptors (Johnson and Ascher, 1987), although this response is generally assumed to be strychnine insensitive (Kishimoto et al., 1981). To test whether or not a putative glycine receptor also plays a role in MeHg-induced early stimulatory effects on hippocampal synaptic transmission, slices were perfused with 50 µM strychnine, a glycine receptor antagonist, before and during exposure to 20 or 100 µM MeHg. In a similar manner to that of bicuculline, strychnine also caused a significant increase in population spike amplitudes and induced repetitive firing, although not as prominently as did bicuculline. However, unlike the effects of MeHg on population spikes in the presence of bicuculline, MeHg caused a further significant increase of population spike amplitude above that already elevated by strychnine. This effect occurred irrespective of whether or not stimulus intensity was reduced (fig. 6). Thus glycine receptors do not play a major role involved in the MeHg-induced early stimulation of hippocampal synaptic transmission. Figure 9 summarizes the effects of MeHg, bicuculline and strychnine alone and in combination with MeHg on population spike amplitude. Clearly, MeHg, bicuculline and strychnine all increase population spike amplitudes significantly. However, pretreatment of slices with bicuculline prevented the MeHg-induced early increase in population spike amplitudes, whereas pretreatment of slices with strychnine failed to do so.

View larger version (K): [in this window] [in a new window]   Fig. 9.   Comparison of the maximum stimulation of CA1 hippocampal pyramidal cell population spikes by 50 µM strychnine (ST), 10 µM bicuculline (BI) or MeHg alone, with the combinations of MeHg with bicuculline (BI + MeHg) or strychnine (ST + MeHg). In the BI + MeHg- or ST + MeHg- groups, slices were superfused first with strychnine or bicuculline for 30 to 60 min, and then with ACSF containing both MeHg and strychnine or bicuculline. The asterisk indicates a significant difference between values of control and slices treated with ST, BI, MeHg, or BI + MeHg (P < .05, analysis of variance). The dagger indicates a significant difference between values of ST and ST + MeHg (P < .05, paired t test).

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Previously we showed that acute bath application of MeHg caused an initial stimulation of hippocampal synaptic transmission prior to suppression to block (Yuan and Atchison, 1993, 1995a). Under similar conditions, Hg⁺⁺ blocked synaptic transmission in the CA1 region of hippocampal slices but did not induce the early stimulatory effects (Yuan and Atchison, 1994). The primary objective of the present study was to identify the potential factor(s) responsible for the early stimulatory effects of MeHg on hippocampal synaptic transmission. Previous results of microelectrode current-clamp recordings suggested that effects of MeHg on inhibitory synaptic transmission and on resting membrane potentials may be involved in the MeHg-induced early stimulation of hippocampal synaptic transmission, because MeHg blocked IPSPs more rapidly than it did EPSPs, and caused biphasic changes in resting membrane potentials of CA1 pyramidal neurons (Yuan and Atchison, 1995a). In our study, we reconfirmed that IPSPs are more sensitive to MeHg than are EPSPs, and demonstrated that this effect is not related to MeHg-induced changes in resting membrane potentials of CA1 neurons, because times to block by MeHg of IPSCs and EPSCs recorded under voltage-clamp conditions were similar to those for block of IPSPs and EPSPs recorded under current-clamp conditions. Moreover, voltage-clamp of neuronal membranes at their resting potential levels failed to prevent the MeHg-induced early increase in EPSC amplitude. In contrast, Hg⁺⁺ also blocked IPSPs more rapidly than it did EPSPs. However, it blocked IPSCs and EPSCs similarly when CA1 neuronal membranes were voltage-clamped at their resting potentials, suggesting that the early block by Hg⁺⁺ of IPSPs compared with that for EPSPs may be due simply to changes in resting membrane potential. Thus, MeHg blocked inhibitory synaptic transmission more preferentially, although it also blocked excitatory synaptic transmission, whereas Hg⁺⁺ blocked both inhibitory and excitatory transmission to the same extent and relatively slowly. This is consistent with our previous observations that MeHg caused early stimulatory effects on hippocampal synaptic transmission, although Hg⁺⁺ did not (Yuan and Atchison, 1994).

In dorsal root ganglion neurons, MeHg suppressed GABA-mediated chloride currents, although Hg⁺⁺ greatly enhanced these currents in a concentration-dependent manner (Arakawa et al., 1991; Huang and Narahashi, 1996). In our study, the IPSCs recorded at CA1 neurons appear to be primarily GABA_A-mediated chloride currents, because their reversal potentials are close to the equilibrium potential of Cl and these currents can be blocked by bicuculline. At 20 and 100 µM, MeHg suppressed all inward and outward currents generated at different holding potentials and shifted the I/V curve to more positive potentials, suggesting that MeHg may block the GABA_A-mediated chloride channels. MeHg has also been shown to inhibit muscimol-stimulated agonist binding in cerebellar P₂ membrane fractions (Komulainen et al., 1995). In contrast, whereas Hg⁺⁺ also suppressed to block all inward and outward Cl currents, it took longer to do so than did MeHg. Unlike the effects of MeHg on GABA_A-activated Cl currents, Hg⁺⁺ initially caused an increase in GABA_A-mediated outward Cl currents before suppressing them, indicating that Hg⁺⁺ may, as it did to the GABA_A-mediated chloride channels in dorsal root ganglion neurons (Arakawa et al., 1991; Huang and Narahashi, 1996), initially increase the open probability of GABA_A-activated chloride channels. Moreover, similar to its effects on the tetrodotoxin-, bicuculline- and picrotoxin-insensitive slow inward currents induced in dorsal root ganglion neurons (Arakawa et al., 1991), Hg⁺⁺ shifted the I/V curve and the reversal potential to more negative potentials, indicating that ions other than Cl may be also involved. These differential effects of MeHg and Hg⁺⁺ on GABA_A receptors may explain why MeHg causes the early stimulatory effects on hippocampal synaptic transmission, although Hg⁺⁺ does not.

If effects of MeHg on GABA_A receptors are indeed responsible for the MeHg-induced early increase in population spike or EPSP amplitude, then pretreatment of slices with the GABA_A antagonist bicuculline should eliminate the early increased phase in either population spikes or EPSPs. After pretreatment of slices with bicuculline, MeHg no longer caused an initial increase in population spike and EPSP amplitudes but still decreased them to block. The failure to induce the early increase in amplitude of population spikes was not due to a ceiling effect caused by bicuculline, although bicuculline significantly increased population spike amplitude to 180 to 200% of control. At the time bicuculline-stimulated amplitudes of population spikes reached maximal levels, increasing stimulation intensity still caused a further increase in population spike amplitude. Moreover, MeHg failed to cause the early stimulatory effects even under conditions in which the stimulation intensity was reduced to prebicuculline control level after bicuculline had increased population spike amplitude to a stable level. The most likely explanation for these results is that MeHg may directly act at GABA_A receptors to cause disinhibition in a similar manner to the effects of bicuculline on GABA_A receptors. This explanation was further supported by the data that MeHg directly blocks responses evoked by bath application of muscimol with a similar time course to that of block of IPSPs or IPSCs.

In the hippocampal CA1 region, at least two subtypes of GABA_A receptors coexist in pyramidal neurons (Pearce, 1993; Gordey et al., 1995). One type is located at the soma or initial segment of the axon. When activated, they hyperpolarize the CA1 pyramidal cell membrane. The other type is located in the dendrites. When activated, they depolarize the CA1 pyramidal cell membrane (Gordey et al., 1995). The responses evoked by bath application of muscimol in the present study are likely to represent a net response of both types of GABA_A receptor to muscimol. Thus, block of responses evoked by bath application of muscimol indicated that MeHg affects both types of GABA_A-mediated responses. We cannot exclude the possibility that presynaptic effects of MeHg on the interneurons or some factors other than GABA_A receptors contribute to the increased effect in hippocampal excitability, since in some slices pretreated with bicuculline, MeHg still caused a delayed increase of about 10-15% in population spike amplitude, although this was not statistically significant. In hippocampus, in addition to GABA_A receptors, GABA_B receptors are located both pre- and postsynaptically in the CA1 region and regulate synaptic transmission (Dutar and Nicoll, 1988a, 1988b; Thompson et al., 1992; Otis et al., 1993; Isaacson et al., 1993; Pitler and Alger, 1994; Wu and Saggau, 1995). At postsynaptic CA1 neurons, GABA_B receptors are coupled to K⁺ channels via a G-protein to cause hyperpolarization of cells. This is expressed as the slow IPSP (Dutar and Nicoll, 1988b; Thompson and Gähwiler, 1992; Otis et al., 1993; Pitler and Alger, 1994). Perhaps the delayed increase in population spike amplitude by MeHg-induced in the presence of bicuculline was due to an effect on GABA_B receptors. Alternatively, effects of MeHg on intracellular Ca⁺⁺ homeostasis may also be involved in the early stimulatory effects of MeHg on hippocampal synaptic transmission, because MeHg increases intracellular Ca⁺⁺ concentrations in several types of neurons (Denny et al., 1993; Hare et al., 1993, 1995). In fact, in hippocampal slices after block of voltage-dependent Na⁺ channels using the local anesthetic QX-314, MeHg also caused an initial increase in Ca²⁺ spike amplitudes prior to decreasing them to block (Y. Yuan and W. D. Atchison, unpublished observation).

Earlier findings from ligand binding studies (Young and Snyder, 1973) and autoradiography (Zarbin et al., 1981; Frostholm and Rotter, 1985; Probst et al., 1986) using [³H]strychnine indicated that glycine receptors are predominately confined to the spinal cord, brain stem and other areas of the lower neuraxis. However, recent studies using immunocytochemistry with monoclonal antibodies (Van den Pol and Gorcs, 1988; Becker et al., 1988), autoradiography with [³H]glycine (Bristow et al., 1986), Northern blot hybridization (Grenningloh et al., 1990; Kuhse et al., 1990a; Malosio et al., 1991) and polymerase chain reaction (Kuhse et al., 1990a, 1990b, 1991) demonstrated a wide distribution of glycine receptors in the higher regions of the central nervous system including cerebral cortex and hippocampus. These glycine receptors, unlike those in the spinal cord and brain stem that primarily express the 1 subunit, a component of the "classical" strychnine-sensitive glycine receptor (Bristow et al., 1986; Becker et al., 1988; Belz, 1990), express a different ligand binding subunit (2), which displays only low affinity for binding of strychnine (Bristow et al., 1986; Becker et al., 1988) or low sensitivity to strychnine upon heterologous expression in Xenopus oocytes (Kuhse et al., 1990a). However, to date we are unaware of any direct report of the existence and the physiological role of functional glycine receptors in hippocampal CA1 neurons, although the above evidence suggests their presence in the hippocampus. In our study, pretreatment of slices with strychnine caused a dramatic increase in population spike amplitude and induced multiple spike responses, although it was not as effective in this regard as was bicuculline. This suggests that there may be a small population of strychnine-sensitive subtype of glycine receptors located in the CA1 hippocampal region, or that strychnine cross-reacts with GABA_A receptors, because they both belong to a superfamily of ligand-gated ion channels and share significant sequence similarity in primary structure and transmembrane topology (Grenningloh et al., 1987; Schofield et al., 1987; Langosch et al., 1988; Schmieden et al., 1993). The latter possibility seems less likely, inasmuch as strychnine did not prevent or suppress the MeHg-induced early stimulation, as did bicuculline pretreatment. Another possible explanation for the failure of strychnine pretreatment to block MeHg-induced early stimulation of synaptic transmission is that these heterologous glycine receptors in hippocampal neurons may be not blocked completely by strychnine due to their low sensitivity to strychnine as suggested by previous studies (Young and Snyder, 1973; Frostholm and Rotter, 1985; Probst et al., 1986; Bristow et al., 1986; Kuhse et al., 1990a). This may be one of the reasons that strychnine was less potent in increasing population spike amplitude than was bicuculline. This possibility also seems less likely, because pretreatment of slices with bicuculline alone completely suppressed MeHg-induced early increase in field potentials. Thus, if there are glycine receptors located on postsynaptic membranes, they do not play a primary role in MeHg-induced early stimulation of hippocampal synaptic CA1 cell transmission.

In conclusion, the preferential block by MeHg of inhibitory synaptic transmission, mediated primarily by GABA_A receptors, appears to be primarily responsible for the MeHg-induced early stimulatory effects on hippocampal synaptic transmission. The importance of this disinhibition caused by MeHg to its overall neurotoxicity also remains unknown.