As reported previously (Matsushima and Aoki 1995), field potential responses to local electrical stimulation are composed of an initial negative spike-like component (S-waves) followed by slower positive waves (P-waves); Figure 1Aa shows an averaged trace (∼30 successive responses at 10 s intervals). Histological examination confirmed that stimulation electrodes were located at ca. 500 μm dorsal to the recording sites in the IMM (Figure 1C). The S-wave contained pre-synaptic action potentials that were unmasked in Ca²⁺-free Krebs solution (Figure 1Ab), whereas the later component of the S-wave (indicated by *) was of post-synaptic origin because this and the following P-waves (indicated by #) were almost completely diminished in Ca²⁺-free conditions. In other words, the post-synaptic response was composed of the initial negative (*) and the late positive wave (#) components. Subsequent successive addition of blockers to the bath (DNQX for AMPA receptor blockade, Figure 1Bb; bicuculline for GABA-A receptor blockade, Bb; CGP52432 for GABA-B receptor blockade, 1Bc; D-AP5 for NMDA receptor blockade, 1Bd) confirmed that the P-waves represents a compound of early excitatory (glutamatergic) and late inhibitory (GABA) transmission; all traces in Ba-d are non-averaged single traces. The blue trace represents a typical single response trace in normal Krebs; inlet traces show the time expansion. DNQX blocked the initial P-waves component (blue arrow in Figure 1Ba) and the late component (brown arrow). The GABA-A blocker unmasked a lasting barrage of unsynchronized spikes (green arrow in Figure 1Bb, late P-waves). Blockade of GABA-B transmission augmented the spiking barrages (black arrow in Bc), which were abolished by D-AP5 (Bd) leaving the pre-synaptic action potentials as in Figure 1Ab.

In normal Krebs, without electrical stimulation, field potential recording did not show any spontaneous activities. Bath-applied bicuculline induced DNQX-sensitive epileptiform bursts (indicated by %) that emerged spontaneously at varying intervals of ca. 5–15 s (Figure 1Db). The activities were not localized but spread ventrally from the dorsomedial pallium. Field potentials were simultaneously recorded dorso-ventrally in medial forebrain regions, with the reference recording site set in the IMM (Figure 1F); experiments were repeated in eight slices. The intermediate medial hyperpallium apicale (IMHA) was 500 μm dorsal to the IMM, and the medial nidopallium (MN) was 500–1,000 μm ventral. Figure 1E shows three occasions, in which simultaneous IMHA (thick line traces, onset indicated by downward arrow) and IMM (thin line reference traces, onset indicated by asterisk) recordings were superimposed. The IMHA bursts led (Figure 1Ea) or followed (Eb) the IMM bursts in almost equal frequency, indicating that the bursts could emerge from epileptic focal spots scattered around IMHA-IMM regions. However, bursts in the MN almost always followed the IMM bursts. Within the frontal plane of the pallial network, IMHA-IMM could be the most excitable hotspot, from which bursts spread ventrally to the MN.

As described below, we then examined the effects of exogenously applied T₃ on local synaptic transmissions and epileptiform bursts in the IMM.

Bath-applied T₃ gradually enhanced the peak amplitude of the P-wave (Figure 2A); averaged traces (30 successive records at 10-s intervals) obtained from one control slice with vehicle (Aa) and another slice with T₃ (40 μM) (Ab) are compared. Responses recorded at 25–30 min after the onset of drug application (black traces) were superimposed on those during a 5-min period just before the application (grey traces with red upward arrows). Effects were not found on the onset slope of the initial S-wave (indicated by dashed blue lines), indicating that the neuronal excitability to the applied electrical stimulation remained unchanged.

Figure 2B shows the time course of changes in the vehicle control (upper) and the T₃ experimental (lower) slices. Recording was composed of pre-application (0–30 min), application (30–60 min) and post-application wash-out (60–90 min). Peak amplitude of the P-wave was plotted every 10 s after normalization against the responses during the last 5 min of the pre-application. Red traces indicate the population average of the superimposed recordings (13 slices for vehicle and 10 slices for T₃). The P-wave amplitude started to rise as soon as T₃ was introduced to the bath and rose continuously during the application period. After T₃ was removed from the bath, the P-wave dropped immediately, but the amplitude stayed higher than 1.0 at 90 min, at 30 min after washout.

The T₃ effects survived after AMPA receptor-mediated excitatory transmission was blocked by DNQX (Figure 2Ca) but disappeared when GABA-A receptor was also blocked (Cb). Also note that DNQX strongly suppressed the P-wave; when compared with the control traces (gray traces in Aa,b), P-wave was much smaller in DNQX (blue traces in Ca). Population data are shown in Figure 2D for the normalized slope of the S-wave (the initial drop phase; dashed blue lines) and the peak amplitude of the early P-wave (normalized against the pre-T₃ perfusion; red arrows). One-way ANOVA revealed no significant difference in the S-wave (p = 0.28), but the differences were significant in the early P-wave (p = 0.0016); further post-hoc multiple comparisons were made by Tukey’s method and revealed a significant difference between vehicle and T₃ (40 μM). The dose response relationship of T₃ effects was examined in an additional 20 slice preparations as shown in Figure 2E. Comparisons with the duplicated data of Figure 2D revealed a significant effect of dose (ANOVA, p < 0.001), although post-hoc multiple comparisons showed only a suggestive level of difference (p = 0.057) for T₃ (20 μM) over vehicle.

These findings indicated that T₃ enhances GABA-A-ergic inhibitory synaptic transmission without changes in the pre-synaptic action potentials, namely the neuronal excitability. The early P-waves are considered to be composed of GABA-A-ergic synaptic inhibition. At this post-hatch stage, GABA-A receptor mediated synaptic current is hyperpolarizing by Cl⁻ influx (see below and Figure 5Bb for the reversal potential of the Cl⁻ ion), giving rise to positive waves when recorded extracellularly as field potential response. However, these results do not reject the possibility that the AMPA receptor-mediated fast synaptic excitation is also modulated by T₃.

To examine if T₃ could also modulate activity-dependent synaptic plasticity, effects of low frequency tetanic stimulation were compared between two groups of slice preparations, one with T₃ bath application for 30min (with an additional interval of 30min before tetanus) and another without T₃ (control). In both groups, slice preparations were maintained in the recording chamber with test stimulation applied at 10-s intervals. As shown in Figure 3A (averaged traces for 5 min from two examples) and Figure 3B (population data), pre-treatment by T₃ yielded no significant differences; S-wave (p = 0.71; Mann-Whitney’s U-test) and early P-wave (p = 0.49), respectively. As the slice preparations were obtained from naïve inexperienced chicks (see 2.1. in Methods), we assume that non-specific saturation of synaptic plasticity did not occur as reported by Bradley et al. (1999).

The spontaneous epileptic bursts induced by bicuculline are composed of two negative components. The first peak (Figure 4Aa and Figure 4b; also see Figure 1E) represents the highly synchronized compound action potentials due to fast excitatory transmission, and the second peak coincides with the lasting strong depolarization of the NMDA receptor mediated plateau potential (Yanagihara et al., 1998). The bursts repeated for several hours, and the amplitude (both 1st and 2nd peaks) gradually became smaller (Figure 4Aa); superimposed traces recorded at 30 min (grey trace) and 60 min (black trace). The gradual decrease in amplitude was accompanied by gradual increase in frequency (number of bursts in 5 min, Figure 4Ca). Further blockade of GABA-B-ergic transmission by CGP52432 (Figure 4Ac) strongly enhanced the 2nd peak, on which barrages of epileptiform bursts were superimposed for several seconds. These barrages were suppressed by D-AP5 (data not shown), indicating that GABA-B transmission potently masked the NMDA receptor mediated long-lasting excitation of the network. However, CGP52432 failed to change the first component (1st peak in Figure 4B), which is the fast compound action potentials.

T₃ increased the amplitude and decreased the frequency of the epileptic bursts. Population data (Figure 4B) revealed that both components were enhanced by T₃; one-way ANOVA revealed significant differences among the three groups (p = 0.015 for the 1st peak, p = 0.038 for the 2nd peak), with both peaks showing significant pair-wise differences between control and T₃ (post-hoc Tukey’s multiple comparisons). In the control condition in which only bicuculline was applied, the frequency gradually increased (Figure 4Ca). When T₃ (40 μM) was added, the frequency increase was arrested (Cb), and the slope of the frequency change during the latter half of the recording session (30 min) was significantly lower (Figure 4D) (p = 0.003; Tukey’s multiple comparison). CGP52432 caused a much stronger suppression (p < 0.001) compared with the control and T₃.

These T₃ effects on epileptiform bursts can be accounted for either by 1) enhanced NMDA receptor mediated depolarization or 2) reduced GABA-B receptor mediated inhibition. Furthermore, T₃ could modulate the resting membrane potentials, spiking threshold and other factors involved in spike generation, rather than synaptic transmissions. We have therefore examined T₃ effects by intracellular recordings from IMM neurons.

NMDA receptor activation produces a bistabile property in the membrane potential by negative resistance because of depolarization-induced release of the Mg²⁺ blockade. When disinhibited by bicuculline, local electrical stimulation reliably elicited a plateau potential for a considerable period without any additional synaptic drive. Figure 5A shows an example recorded in single-electrode current clamp mode using a normal (low Cl⁻ concentration) pipette solution. After a depolarizing current injection through the recording electrode (duration = 100 msec, indicated by a rectangle above the traces), four repetitive electric shocks were applied at 40 msec intervals (dots); six consecutive responses recorded at 10 s intervals were superimposed. The depolarizing current intensity was adjusted so that one action potential was induced at the peak of the depolarization. In control conditions (normal Krebs in the bath), each shock generated quickly decaying excitatory post-synaptic potentials (EPSPs) with action potentials on the peak of the first and/or the second shocks. When bicuculline was added (Ab), a depolarizing plateau (indicated by * at the shoulder) appeared and lasted for ca. 400 msec after the electric shocks; notice that the depolarization muted the neuron and action potentials failed to occur during the plateau. Bath applied T₃ (40 μM, Ac and Ad) gradually reduced the duration of the plateau (3 and 10 min after T₃ reached the chamber) without suppressing the initial component (indicated by #); fast EPSPs were thus supposed to be insensitive to T₃. D-AP5 completely suppressed the plateau (Ae), confirming its NMDA-receptor origin. Notice also that the resting membrane potential, the depolarization by the current injection, and the threshold membrane potential for generating spikes remained unchanged.

The NMDA-plateau potential was induced by bicuculline application in all 15 neurons recorded from eight slice preparations, with resting membrane potentials ranging from -50 to -67 mV (mean ± s. e.m. = -59.5 ± 1.6 mV) and stable recording maintained for 40–84 min. T₃ effects (40 μM, >15 min bath application) were successfully examined in 11 neurons, and the NMDA plateau was suppressed by T₃ in 10 out of these 11 neurons; no changes were detected in the remaining one neuron. Vehicle control was tested in other four neurons, and none of them showed changes in the plateau.

The GABA-A receptor-mediated inhibitory synaptic response was examined after glutamatergic (AMPA- and NMDA-R mediated) and GABA-B ergic transmission was blocked (Figure 5B, bridge balance mode). In normal Krebs (Ba), tonic current injections at three levels (two depolarizing and one hyperpolarizing) set the membrane potential at ca. −45 mV, −50 mV and −70 mV. Single shocks of local electrical stimulation elicited fast EPSPs, and single action potentials followed by a hyperpolarization at −45 mV (indicated by vertical dashed red line). The hyperpolarization phase turned to depolarization if held at −70 mV, indicating the reversal of the GABA-A response at approximately −50 mV. After glutamatergic and GABA-B-ergic transmission were blocked by DNQX, D-AP5 and CGP52432 (Bb), the resting membrane potential was hyperpolarized by 5 mV. Synaptic responses were examined by adjusting the tonic current injection levels. GABA-A ergic depolarization was clear at −70 mV (upward blue arrow). When T₃ was added (40 μM, Bc), the depolarizing GABA-A response increased without changes in the reversal potential, resting potential, or threshold of spike generation. The bicuculline-induced lasting depolarization was further augmented by baclofen (20 μM) (Figure 5Ca,b); note that the synaptically driven initial depolarization (indicated by #) was suppressed, thus allowing slow depolarization of varying latency and amplitude (indicated by *). The GABA-B receptor activation paradoxically facilitated the spontaneous bursting caused by NMDA receptor activation. Actually, the delayed depolarization was not accompanied by synaptic drives if measured in voltage clamp mode and disappeared with the addition of D-AP5 (data not shown). Most probably, the hyperpolarizing effect of baclofen via GABA-B receptor activation set the membrane potential within an appropriate range for the NMDA-induced plateau to spontaneously arise.

The GABA-A responses were examined in nine neurons from four slices, with resting membrane potentials ranging from −55 to −69 mV (mean ± s. e.m. = -63.9 ± 1.5 mV), and T₃ was tested in eight neurons and vehicle in 1 neuron. Of these eight neurons, enhancement by T₃ occurred in six, suppression in one and no detectable effect was found in another. The effect of baclofen on the NMDA plateau was reproduced in all of the other three neurons tested. Because the reversal potential of the GABA-A response was close to the resting potential (usually ∼5mV above than the resting level), a small accidental change in the resting level had a big influence on the GABA-A response amplitude. We therefore re-examined the effects of T₃ using high-Cl^- pipette solution.

Increase in the GABA-A response by T₃ was confirmed (Figures 5D,E). At the resting membrane potential, the GABA-A response quickly turned depolarizing after poking the neuron, and reached equilibrium within ∼10 min. In normal Krebs, two pulse electric shocks (100 msec interval) induced initial spiking followed by lasting depolarization (Da). The depolarizing response remained after DNQX, D-AP5 and CGP52432 were added (Db). The unclamped resting membrane potential was slightly hyperpolarized (from −65 to −69 mV) but the spiking threshold potential (−47mV) remained unchanged. In another neuron (Ea), T₃ enhanced the depolarizing response (Eb). The depolarization was suppressed by bicuculline (Ec) although not completely. We therefore argue that GABA-A receptor mediated responses was enhanced by T₃. T₃ application caused no changes in the spiking threshold (θ = −45 mV) in response to a rectangular depolarizing current injection through the recording electrode.

The reversed GABA-A potential was reliably recorded from six neurons from five slice preparations, with resting membrane potentials ranging from −59 to −77 mV (mean ± s. e.m. = −68.7 ± 3.1 mV), and T₃ effects (>15 min of bath application) successfully examined in all six neurons. Enhancement by T₃ occurred in five out of these six neurons, whereas suppression occurred in the other neuron.

The possibility that T₃ enhances the NMDA-receptor mediated response (see Section 3.2.3) should therefore be rejected, and an alternative hypothesis must be considered such that GABA-B mediated inhibition was suppressed. Taken together, T₃ could have opposite effects on the GABA-mediated responses, namely, enhancement of the fast GABA-A-mediated inhibition and suppression of the slow GABA-B-mediated inhibition of the IMM network.