Several suppressive processes shape the response properties of auditory neurons, namely lateral inhibition, non-monotonic rate level function and excitation/inhibition binaural interaction. By combining intracellular recording from and staining of layers 2 and 3 pyramidal neurons (PNs) in cat primary auditory cortex, we demonstrate the temporal aspects of depolarization and hyperpolarization underlying these suppressions using pure tone stimulation. Two populations can be distinguished by the occurrence of hyperpolarization following onset depolarization (O-DEP). In layer 2 PNs there is an absence of hyperpolarization following O-DEP, while in layer 3 PNs hyperpolarization follows O-DEP. The latency of O-DEP is shortest at the neuron’s best frequency. The latency shortens as sound intensity increases. In non-monotonic PNs, hyperpolarization onset becomes shorter as sound intensity increases. This earlier onset of hyperpolarization shortens the duration of the preceding O-DEP, resulting in a decreased O-DEP amplitude. Diverse patterns in the temporal interaction of depolarization and hyperpolarization underlie the binaural suppression interaction. These results demonstrate that diverse suppressive responses result from differences in the temporal timing of excitation and inhibition. The present results also suggest the possibility of distinct connections between PNs responding in a similar manner.
Despite extensive studies on the connectivity of auditory cortex (Imig and Reale, 1980; Mitani et al., 1985; Code and Winer, 1986; Winguth and Winer, 1986; Ojima et al., 1991, 1992; Rouiller et al., 1991; Wallace et al., 1991; Huang and Winer, 2000; Smith and Populin, 2001; Winer et al., 2001), very limited information is available on how excitatory and inhibitory synaptic events create neuronal response properties (de Ribaupierre et al., 1972; Volkov and Galaziuk, 1991, 1992). This contrasts with the relatively intensive analyses of excitatory and inhibitory synaptic potential interactions in auditory subcortical areas (Wu and Oertel, 1986; Casseday et al., 1994; Covey et al., 1996; Kuwada et al., 1997; Pedemonte et al., 1997; Sanes et al., 1998) and also in the visual system (Ferster, 1986; Douglas and Martin, 1991; Douglas et al., 1991; Berman et al., 1992; Ferster, 1992; Weliky et al., 1995; Ferster and Miller, 2000).
Studies under anaesthetized or awake conditions have shown that auditory cortical neurons respond distinctly with increased or decreased spike number to different frequencies of sounds (Evans and Whitfield, 1964; Goldstein et al., 1968, 1970; Merzenich et al., 1975; Schreiner and Mendelson, 1990; Sutter et al., 1999), differences in sound intensity magnitude or amplitude envelope (Phillips and Irvine, 1981; Phillips et al., 1985; Phillips, 1988; Schreiner et al., 1992; Heil et al., 1994; Calford and Semple, 1995; Sutter and Schreiner, 1995) and different binaural cues (Hall and Goldstein, 1968; Brugge et al., 1969; Brugge and Merzenich, 1973; Imig and Adrian, 1977; Middlebrooks et al., 1980; Middlebrooks and Pettigrew, 1981; Reale and Kettner, 1986; Imig et al., 1990; Semple and Kitzes, 1993a,b; Clarey et al., 1994; Barone et al., 1996; Reser et al., 2000).
Several types of suppressions are known to shape the response properties of auditory cortical neurons. These include: (i) lateral suppression, i.e. a reduction in response to a test stimulus by the presentation of a probe stimulus of a different frequency (either prior to or simultaneously with the test stimulus) (Calford and Semple, 1995; Sutter et al., 1999); (ii) non-monotonic rate versus intensity suppression, i.e. a reduction in response to a test stimulus that results from progressive increases in the intensity of the test signal (Phillips et al., 1985; Semple and Kitzes, 1993a,b; Heil et al., 1994; Sutter and Schreiner, 1995); (iii) excitatory/inhibitory binaural interaction suppression, i.e. a reduction in response due to binaural stimulation when compared to the monaural condition (Imig and Brugge, 1978; Imig and Reale, 1981; Phillips and Irvine, 1983; Phillips, 1985; Phillips et al., 1985; Calford and Semple, 1995). In order to understand synaptic events underlying these suppressions, in this study we have intracellularly characterized membrane potential responses of pyramidal neurons (PNs) using pure tone stimulation. The results provide evidence for the temporal interactions of depolarization and hyperpolarization underlying the suppressive response properties. Following the physiological characterizations, most neurons were labeled by a tracer injected from the recording pipettes, so that the precise laminar positions of the cell bodies could be determined. In combination with previous anatomical work on local connections, these data contribute to our further understanding of network organization.
Materials and Methods
The procedures used were similar to those described previously (Ojima et al., 1991, 1992; He et al., 1997) and were approved by the Animal Care Committee of Toho University and conformed to NIH guidelines.
Nineteen healthy cats of both sexes, weighing 2.0–5.0 kg and having clean external acoustic meati, were pretreated with atropine sulfate (0.025 mg/kg s.c.) and dexamethasone (0.25 mg/kg i.m.) prior to general anesthesia with Nembutal (sodium pentobarbital, 35 mg/kg i.p.; Abbott Laboratories, TX). The cats underwent tracheotomy and cephalic vein catheterization. Anesthesia was maintained by infusion of diluted Nembutal (10 mg/ml saline) through a catheter connected to an infusion pump (CFV-2100; Nihon Kohden, Japan), and delivered continuously at a speed of 0.3 ml/kg/h during recording. Additional volumes of diluted Nembutal were manually administered when necessary. The cats were placed in a stereotaxic apparatus set in a single-walled, sound attenuating chamber (Aco, Japan) and secured to standard ear bars inserted into the external acoustic meati. A post was anchored to the frontal skull with screws and acrylic resin. An antero-posteriorly elongated opening was made in the skull to expose the dura mater. The dura mater was cut above the primary auditory field between the dorsalmost tips of the anterior and posterior ectosylvian sulci or slightly dorsal to this level (Ojima et al., 1991). An acrylic cylindrical chamber (inner diameter 20 mm) was attached to the skull with carbo cement (Shofu, Kyoto, Japan). The ear bars were then removed and a sound delivery system connected to a speaker enclosure (see below) was inserted into each of the external acoustic meati.
The electrocardiogram was continuously monitored, pupil size was observed with a CCD camera and areflexia was periodically checked by pinching a forepaw. The body temperature was maintained at 37.0–38.0°C using a thermostatically controlled heating pad.
Acoustic stimuli were generated digitally by a MalLab system (Kaiser Instruments, Irvine, CA) controlled by a Macintosh computer (He et al., 1997). Acoustic stimulation consisted of pure tone sets, each containing 10 bursts of pure tone of a single frequency (50 ms duration and 10 ms rise/fall time with a sigmoidal envelope, 1.2 s inter-burst interval). After briefly evaluating responses audiovisually at threshold sound intensity, approximately10 sets of tone bursts of different frequencies, from 2 to 25 kHz, were delivered at 10/20 dB above threshold (Sutter et al., 1999). For intensity functions and binaural interactions, best frequency tones of varied sound intensities with monaural and binaural stimulation (10 repeats for each parameter) were applied. Membrane potentials were monitored together with waveforms of the sound stimuli on an oscilloscope (Kikusui, Japan), and peristimulus time histograms (PSTHs) and raster displays were generated on-line to determine the best frequency (MalLab system). Sound stimulation was delivered through a calibrated silicon tube (2 cm in length and 8 mm in diameter) coupled to an enclosure containing a dynamic earphone speaker (Beyerdynamic DT-48; Beyer, Heilbronn, Germany). The silicon tubes were inserted into the right and left external meati. The gap made between the meatus wall and tube was sealed with Vaseline.
The sound delivery system was calibrated at the opening of the silicon tube for a frequency range of 0.1–25.0 kHz, using a condenser microphone (7017, 0.25 inch; Aco), with reference to a standard sound pressure level (SPL) (94 dB re 20 μPa, 1 kHz, 2126 type; Aco). The calibration data for each ear were stored in a computer file for use in controlling a digital attenuator to obtain desired SPLs.
Glass microelectrodes (CEI, UK) were pulled on a vertical puller (PE-2; Narishige, Japan) and filled with 1.5% biocytin (Sigma, MO) dissolved in 1.0 M potassium methylsulfate (ICN, OH) or 1.0 M potassium acetate (Sigma). For some neurons, the salt was replaced by 1.0 M potassium chloride (Sigma) buffered with 0.01 M Tris–HCl (pH 8.0). Microelectrode resistances ranged from 60 to 80 MΩ. A silver/silver chloride wire and plate were used as a lead from the microelectrode and a reference electrode, respectively. The microelectrode, set on a remote controlled stepping microdrive (PC-5N; Narishige, Japan) was advanced vertically into the brain through the slit in the dura. Brain pulsation was reduced by filling the chamber with warm wax (melting point 45°C). After balancing the null potential level of the recording system, the resistance and capacitance of the microelectrodes were adjusted by bridge balancing and capacity compensation, respectively (MEZ-8301; Nihon Kohden, Japan). Sound stimulation was initiated if the impalement of neurons was followed by an initial shift of membrane potential by more than –40 mV and if injury-induced action potentials disappeared within a few minutes. If these criteria were not met, the microelectrode was advanced further. This through penetration frequently led to degeneration of the neurons. Depth along an electrode track was registered for every neuron impaled. Each hemisphere had between three and five penetration tracks. At most three neurons were recorded per track. When multiple recordings were attempted along one track, the microelectrode was withdrawn by 200 μm towards the cortical surface and then advanced again to a deeper level for the next neuron. This prevented degeneration of the recorded neurons, which would otherwise be caused by the through penetration. To define laminar position accurately, recorded neurons were labeled with biocytin injected iontophoretically (1.5–2.0 nA positive current for ∼10 min). Biocytin also diffused spontaneously into neurons while recording and labeled them enough to allow identification of cell morphology if the recording lasted for >5 min. From time of surgery, experiments were completed within 18 h; typically recording sessions lasted <12 h.
Immediately after finishing recording, animals were administered an overdose of Nembutal and perfused transcardially with saline (300–500 ml) followed by 4% paraformaldehyde (Sigma) in 0.1 M phosphate buffer (pH 7.4). The brain was then quickly removed from the skull and kept in the same fixative at 4°C overnight. Visualization of labeled neurons was identical to that described in detail previously (Ojima et al., 1991). Briefly, 50 μm sections were cut on a freezing microtome, reacted in sequence by the ABC method (Vector Laboratories, CA) and 0.06% 3′,3′-diaminobenzidine (Sigma). Sections were counterstained with 0.1% thionin (Sigma) and laminar borders were determined on the basis of the sizes and densities of cell bodies (Winguth and Winer, 1986). Layer 2 was 150–200 μm thick, corresponding to between five and six rows of cell bodies (Fig. 1). Reconstruction of microelectrode tracks and positions of labeled cell bodies were used to identify the recording points on labeled neurons.
Recording of single PNs typically lasted 10–90 min. Membrane potentials, sound stimuli, current injected and trigger signals were recorded on a 4-channel digital audio tape recorder (DC to 10 kHz, PC204Ax; Sony, Japan) together with voice commentary. Parameters were analyzed off-line on a PowerLab system (AD Instruments, Australia).
Recordings were made from a total of 193 cortical PNs. For detailed off-line analysis, the following selection criteria were used: stable resting membrane potentials greater than –50 mV, responses to a wide range of pure tone bursts sufficient to define the best frequencies, identified laminar position of the cell bodies and a recording point close to the cell body (<50 μm). Forty-seven PNs fulfilled the above criteria. Their resting membrane potentials ranged from –50.1 to –70.3 mV. An additional seven PNs could not be driven by or tuned to pure tones. The extent of diffusion of the tracer in these PNs varied from light labeling of the cell body/proximal dendrites to intense labeling of both cell body/dendrites and axons (Fig. 1).
In some traces of membrane potential, spontaneous action potentials or depolarizations comparable in amplitude to acoustically induced ones occurred during a 50 ms pre-stimulus period. These traces were excluded from the data analysis, since activation prior to the stimulus is known to affect stimulus-induced responsiveness (Ferster and Jagadeesh, 1992; Ganguly et al., 2000; Henze and Buzsáki, 2001). The remaining traces within each trial were averaged to obtain a mean trace of the membrane potentials. In some cases, for comparison of the peak amplitudes of membrane potentials induced by different stimulation variables, each trace was low-pass filtered at 100–200 Hz to eliminate action potentials and then subjected to averaging.
General Response Properties of Membrane Potential of Layer 2 and 3 PNs
Membrane potential changes of layer 2 and 3 PNs were examined in response to a variety of pure tone bursts applied binaurally at the resting membrane potential level. There were at least three common characteristics in the membrane potential response of PN populations in layers 2 and 3.
Onset depolarization (O-DEP) was induced over a frequency range which was wider than that for action potential generation. Although the response frequency range was wide, a single best frequency could be determined for most PNs on the basis of the O-DEP amplitude (see larger symbols in Fig. 2). In relation to the frequency–latency function for O-DEP, minima were almost always observed (43/44) at the best frequency (Fig. 2). Often the frequency–latency function had a second minimum at frequencies other than the best frequency (see for example filled square in layer 2 and open diamond in layer 3d in Fig. 2), consistent with recent extracellular studies (Loftus and Sutter, 2001). The intensity–latency relationship showed that as sound intensity of a pure tone increased, the latency of O-DEP concomitantly became shorter, reaching a minimum value at the highest intensity (see for example Brugge et al., 1969; Phillips and Irvine, 1981). This relationship was observed regardless of whether the O-DEP amplitude increased (monotonic pattern) or decreased (non-monotonic pattern) with increasing sound intensity. The observed latency relationships with frequency and intensity were similar in layer 2 and 3 PNs.
In contrast to the similarities in the latency relationships, distinct differences were observed in the occurrence of hyperpolarization between layer 2 and 3 PNs. Most layer 3 PNs (33/39) hyperpolarized immediately following the O-DEP. The amplitude (3.0 ± 1.77 mV, n = 33) and duration (87.1 ± 31.1 ms, n = 33) of hyperpolarization (measured from the resting membrane potential level) depended on the tone frequency and intensity and varied from neuron to neuron (Fig. 4). Unlike the layer 3 PNs, for layer 2 PNs O-DEP was not followed by hyperpolarization (7/8). This was the case even when sound intensity was raised to the highest level (e.g. 80 dB SPL).
Layer 2 Response Properties
O-DEP induced in layer 2 PNs was not followed by hyperpolarization. Rather, the O-DEP was followed by a sustained, late depolarization. The amplitude and duration of the late depolarization varied considerably from neuron to neuron. The overall duration of the onset and sustained parts of the depolarization was, on average, 181.3 ± 61.6 ms (n = 6) at the best frequency at 10/20 dB above threshold.
In Figure 5 we show representative membrane potential responses to sets of pure tone bursts at a fixed sound intensity (50 dB) for a layer 2 PN. A frequency range from ∼6 to 16 kHz was effective in inducing various amplitudes of O-DEP (Fig. 5A). The averaged O-DEP amplitude (measured from resting membrane potential level) was maximal at a frequency of 14.3 kHz (Fig. 5B,C). The O-DEP induced at this frequency occasionally led to initiation of action potentials. This frequency corresponded to the best frequency defined by spike counting (Fig. 5D). Onset latency of the O-DEP was shortest at the best frequency (Fig. 5C,D).
(i) Response level function suppression of sound intensity.
Amplitude of O-DEP either monotonically increased with increasing loudness or increased and then decreased as sound intensity increased. Three of six layer 2 PNs examined displayed a suppressive response level relationship of O-DEP (Fig. 6A). In this study, PNs were regarded as non-monotonic if the amplitude of O-DEP was suppressed to ≤70% of the maximal value. Figure 6B shows individual traces of membrane potential of a representative layer 2 PN in response to varied sound intensities. As sound intensity increased, the O-DEP amplitude started to increase at the threshold intensity of 40 dB SPL, reached its maximal value at 50 dB SPL and then declined to 67% of the maximum at 70 dB SPL. As a consequence of the relationship between O-DEP latency and sound intensity, the latency of the O-DEP was shortest at the highest sound intensity (upper panel in Fig. 6B). This O-DEP was not followed by hyperpolarization at any magnitude of sound intensity. Even when the membrane potential was depolarized by passing a positive current (+0.2 nA) through the recording pipette, there was no hyperpolarization following the O-DEP (filled symbols in upper panel and uppermost traces in lower panel of Fig. 6B). Another example is shown in Figure 6C, in which membrane potentials are averaged for different SPLs.
(ii) Binaural interaction suppression.
Binaural interaction was examined by comparing membrane potential responses induced by monaural and binaural stimulation at a sound intensity 10/20 dB above threshold. Two of five layer 2 PNs displayed the excitation/inhibition (EI) type interaction. As shown in the averaged membrane potentials in Figure 7A,B, contralateral ear stimulation alone produced a larger amplitude of O-DEP than ipsilateral ear stimulation alone (i.e. contralateral is dominant, while ipsilateral is non-dominant). Stimulation of both ears (binaural stimulation) induced a smaller amplitude of O-DEP compared to that induced by the dominant ear stimulation alone. Although the non-dominant ear stimulation was supposed to exert a suppressive effect on the dominant ear response, hyperpolarization was not detected in the membrane potential response induced by the non-dominant ear stimulation.
Layer 3 Response Properties
As in layer 2 PNs, the frequency range over which O-DEP was induced in layer 3 PNs was wider than the frequency range over which action potentials were induced (Figs 2 and 8). In Figure 8 we show individual traces of membrane potential (Fig. 8A) in response to tone frequencies ranging from 5.1 to 11.0 kHz at 30 dB SPL (10 dB above threshold), spike rates (Fig. 8B) at the best (7.5 kHz) and nearby (8.0 kHz) frequencies and the frequency– latency relationship (Fig. 8C). Hyperpolarization followed large amplitude O-DEP in response to tone bursts whose frequency ranged from 6.1 to 11.0 kHz (Fig. 8A for individual traces and Fig. 8D for averaged traces). The largest amplitude hyperpolarization was elicited at the best frequency (i.e. 7.3–7.5 kHz; Fig. 8D). At frequencies flanking the best frequency (i.e. 5.1, 6.1 and 11.0 kHz), large amplitude O-DEPs were also elicited, but no action potentials were initiated from the O-DEP (subthreshold depolarization). Following the subthreshold O-DEP, hyperpolarization was also elicited, with an amplitude smaller than that induced at the best frequency (Fig. 8D). The sequence of subthreshold depolarization and subsequent hyperpolarization at frequencies other than the best frequency occurred for 56% of layer 3 PNs (18 of 32 layer 3 PNs that had acoustically induced hyperpolarization).
Thirty-five layer 3 PNs were tested for response level changes at their best frequencies. Non-monotonic response level suppression in membrane potential behavior was observed in 63% of the PNs (22/35). Twenty-eight layer 3 PNs were tested for binaural interactions. Suppressive binaural interaction (EI type) was found in 21% of the PNs (6/28).
(i) Response level function suppression of sound intensity.
Of the population of layer 3 PNs which showed a non-monotonic relationship between sound intensity and depolarization amplitude, nearly half (12/22) showed a strong decrease in O-DEP amplitude (≤70% of the maximal value) with increasing sound intensities. The rest displayed only a weak non-monotonic response of membrane potential.
For the strongly non-monotonic population, the decreased O-DEP amplitude was associated with a shortened duration (Fig. 9A). The shortening of the O-DEP duration resulted mainly from the earlier onset of hyperpolarization following the O-DEP at higher sound intensities. Membrane potentials of a layer 3 PN with a representative non-monotonic response level relationship are shown in Figure 9B. As sound intensity increased from threshold (35 dB SPL), the amplitude of O-DEP increased and reached a maximal level at 45 dB SPL. The O-DEP amplitude then declined to lower levels at increasingly higher sound intensities (upper panel in Fig. 9B). In association with the decline in the O-DEP amplitude, the O-DEP duration became shortened (lower panel of Fig. 9B). In this cell, the O-DEP latency changed from 16.9 to 12.4 ms with a SPL change from 45 to 70 dB (4.5 ms shortening; arrowheads in Fig. 9B). The latency change of the hyperpolarization was greater in magnitude and ranged from 42.2 to 29.5 ms (12.7 ms shortening; short arrows in Fig. 9B). This greater degree of shortening of the hyperpolarization latency resulted in shortening of the O-DEP duration. Another example showing a similar non-monotonic relationship is shown in Figure 9C.
Thirteen of 35 layer 3 PNs displayed a monotonic response level change. O-DEP of monotonic PNs also showed a level-dependent decrease in duration as sound intensity increased. The O-DEP duration of monotonic PNs was relatively longer, however, compared with that of the non-monotonic PNs. The O-DEP duration measured at 70 dB SPL was 27.3 ± 15.9 ms (n = 11) for the monotonic population and 15.7 ± 8.2 ms (n = 9) for the strongly non-monotonic PNs (t-test, P < 0.05).
(ii) Binaural interaction suppression.
A binaural suppression response of membrane potential was also observed. Monaural stimulation of a dominant ear evoked a larger amplitude O-DEP than did binaural stimulation of both ears. Although the number of PNs displaying the binaural suppression interaction was small (six of 28 neurons tested), there were at least three variations in the pattern of membrane potential response (Fig. 10A).
The first pattern of membrane potential response (black symbol, Fig. 10A) is shown in Figure 10B. In this cell, at the best frequency (7.8 kHz) at 40 dB SPL monaural stimulation of a non-dominant (CONTRA) ear evoked a powerful hyperpolarization consisting of early short latency and late long lasting phases (single and double asterisks in upper panel of Fig. 10B). Dominant ear (IPSI) stimulation induced a large O-DEP with a relatively long duration. Binaural (BIN) stimulation (40 dB SPL to each ear) induced a significantly shortened duration of O-DEP compared with that elicited by the dominant ear stimulation. This shortened duration of the binaural O-DEP appeared to be due to a summative interaction between the two monaural membrane potential responses (upper panel of Fig. 10B). The early phase of the non-dominant hyperpolarization coincided with the peak/falling phase of the dominant depolarization. This non-dominant suppressive effect was so extensive that basically the same membrane potential response was induced at a wide range of sound intensities (middle panel of Fig. 10B). The shortened duration of O-DEP on binaural stimulation was coupled to a decreased amplitude, reflected in a decreased spike rate, i.e. 1.4, 0 and 1.0 spikes/trace in response to dominant ear (IPSI), non-dominant ear (CONTRA) and both ear (BIN) stimulation, respectively (lower panel of Fig. 10B).
The second pattern of suppression in the binaural interaction was found in two PNs (gray filled symbols, Fig. 10A). This suppression was not mediated by the strong hyperpolarization generated by non-dominant ear stimulation, but had a complicated temporal interaction of depolarization and hyperpolarization. An example (gray filled diamond, Fig. 10A) is illustrated in Figure 10C. This PN is of EI type with an ipsilateral (IPSI) ear dominance, as defined both by spike counting (upper panel) and the amplitude of O-DEP at relatively high sound intensities. As shown in the lower panel, the dominant O-DEP induced ipsilaterally was later relative to the non-dominant O-DEP induced contralaterally. This temporal sequence led to concurrence of the peak of the dominant (IPSI) depolarization with the falling phase of the preceding, non-dominant (CONTRA) depolarization. Although the ear dominance is opposite, the other neuron (gray filled triangle, Fig. 10A) showed the same temporal sequence of dominant and non-dominant ear responses (compare the latencies of largest and smallest filled symbols in Fig. 10A).
The other three PNs (open symbols, Fig. 10A) showed the third pattern of membrane potential response underlying the binaural suppression of spike rate. A representative example (open triangle, Fig. 10A) is shown in Figure 10D. Dominant CONTRA ear stimulation induced the highest spike rate (0.43 spikes/trace) and largest O-DEP amplitude. Unlike the second pattern, the dominant (CONTRA) depolarization preceded the non-dominant (IPSI) depolarization. This temporal sequence did not lead to concurrence of the peak of the dominant (CONTRA) depolarization with the falling phase of the non-dominant (IPSI) depolarization.
Membrane potential responses associated with EE (excitatory/excitatory) and EO (excitatory/null) type interactions were also observed in layer 3 PNs. Their membrane potential responses were similar to those of the third EI pattern in terms of the temporal relationship, although the binaural O-DEP amplitude was larger than or equal to the dominant, monaural O-DEP amplitude.
In this discussion we argue that tone-induced depolarization and hyperpolarization represent excitatory and inhibitory postsynaptic potentials. Following this, the differences in response pattern in layers 2 and 3 PNs will be discussed on the basis of cortical network. Subsequently, the argument continues about possible temporal interactions of depolarization and hyperpolarization to explain the suppressive responses known in auditory cortex.
Activation of Excitatory and Inhibitory Postsynaptic Potentials by Sound Stimulation
This study shows that PNs in the supragranular layer of cat primary auditory cortex respond with O-DEP with or without subsequent hyperpolarization to pure tone bursts in a population-dependent manner. It is likely that the depolarization and hyperpolarization induced by pure tone stimuli reflect excitatory and inhibitory postsynaptic potentials. This interpretation is supported by the following four points. (i) Early hyperpolarizing deflection from the resting membrane potential is believed to reflect an inhibitory postsynaptic potential if the resting membrane potential is set higher (i.e. depolarized) than the equilibrium potential for chloride ions (around –75 mV) (see McCormick, 1998). Indeed, the resting membrane potentials of the PNs examined were between –50 and –70 mV (mostly –50 to –60 mV). (ii) Inhibition is also induced by a shunting mechanism, by which local excitation can be suppressed by an increase in membrane conductance. In this situation, hyperpolarization cannot be manifested if the membrane potential is close to the equilibrium potential for chloride ions. Our trials, in which the membrane potential was set at more depolarized levels, still did not reveal hyperpolarization in layer 2 PNs. It is thus unlikely that the shunting mechanism explains the layer 2 PN suppression observed in cat primary auditory cortex. In other cortical areas shunting inhibition is known not to operate or to do so only weakly (Douglas et al., 1988; Berman et al., 1991; Ferster and Jagadeesh, 1992; however, see Borg-Graham et al., 1998). (iii) Depolarizing and hyperpolarizing shifts of the base membrane potential by constant current application also leads to increased and decreased amplitude of the hyperpolarization following the O-DEP. Since the amplitude of the hyperpolarization increases after the resting membrane potential is shifted to depolarized levels, this hyperpolarization should not be a manifestation of the transient reduction in excitatory depolarization (for more details see Berman et al., 1991). (iv) In addition, in three layer 3 PNs, chloride ions were iontophoretically injected through a recording pipette containing potassium chloride in order to shift the equilibrium potential for chloride ions toward a more positive (i.e. depolarized) value. Injection resulted in a diminished amplitude of the hyperpolarization following O-DEP (data not shown; see Krnjević and Schwartz, 1967). Together, these results strongly suggest that the hyperpolarization is a chloride ion-mediated inhibitory postsynaptic potential. Furthermore, as in the visual system, it is also likely that the in vivo inhibitory postsynaptic potential following the preceding excitatory postsynaptic potential induced by acoustic stimulation is mediated by GABAA receptors (Conners et al., 1988).
Differences in the Membrane Potential Responses of Layer 2 and 3 PNs
Intracellular recording in the present study has characterized excitatory and inhibitory postsynaptic potentials induced in PNs in layers 2 and 3. Although the absolute duration and magnitude of these membrane potentials varied considerably from neuron to neuron, there was a consistent difference in the membrane potential responses of PNs located in these two layers, as determined by dye injection.
In layer 3, most PNs displayed a sequence of excitation and hyperpolarizing inhibition. In contrast, virtually all layer 2 PNs displayed an onset depolarization which is not followed by a hyperpolarization, but rather by a sustained late depolarization. Could the different membrane potential responses be due to a different equilibrium potential for chloride ions of the two PN populations? That is, the chloride equilibrium potential level might be higher (i.e. more depolarized) for layer 2 PNs than their resting membrane potential level. This possibility can be excluded because we were able to confirm the failure in the hyperpolarization induction in layer 2 PNs by setting the membrane potential at a more depolarized level (Avoli, 1986; Ferster, 1986; Avoli and Olivier, 1989) (see Fig. 6B). This membrane potential level also excludes the possibility of involvement of a shunting inhibition mechanism (see above). Also to be considered is whether the stimulus intensity might not be strong enough to induce the hyperpolarization following O-DEP. This possibility seems unlikely, since the hyperpolarization was not induced even when the stimulus intensity was raised to the maximal value (80 dB SPL; Fig. 6C), which consistently elicited large magnitudes of the hyperpolarization in layer 3 PNs.
Although relevant data are not available for the auditory system, a similar membrane potential behavior has been reported for PNs in in vitro cat motor cortex. Electrical stimulation of superficial layer 3 elicited a large amplitude hyperpolarization in layer 2 PNs, whereas that of deep layer 3 evoked no hyperpolarization (Kang et al., 1994). This raises the possibility that polysynaptic excitation may result in a depolarization without a following hyperpolarization. Anatomical findings have shown that the medial geniculate-cortical afferent projection terminates mainly in layers 3 and 4 (Hashikawa et al., 1995; Smith and Populin, 2001) and that PNs in the lower half of layer 3 (and also in layer 4) are the major population activated monosynaptically by thalamic afferents (Smith and Populin, 2001). Thus layer 2 PNs may not be monosynaptic targets of the thalamic afferent (for the visulal system see Ferster and Lindstrom, 1983). Since other anatomical findings in both visual (Gilbert and Wiesel, 1979; Martin and Whitteridge, 1984) and auditory cortices (Ojima et al., 1991) have shown that a rich axon plexus extends from layer 3 PNs into layer 2, it is likely that a tonally driven excitation mediated by the thalamic axons can reach layer 2 neurons polysynaptically via excitatory synapses of the layer 3 PNs (Mitani and Shimokouchi, 1985). This polysynaptic excitatory connection may not activate inhibitory inputs, via interneurons, to layer 2 PNs in the present stimulus paradigm.
The present findings show that onset time of a sequence of depolarization and hyperpolarization varies depending on tonal frequency, intensity and ear of stimulation (i.e. monaural or binaural). Membrane potential responses of layer 2/3 PNs display a non-monotonic as well as monotonic pattern to increasing SPLs.
One explanation of the decreased amplitude of depolarization in non-monotonic layer 3 PNs follows from the fact that depolarization and hyperpolarization counteract each other in a summative fashion when they temporally coincide (Eccles, 1964, 1968). Our results show that at higher sound intensities the latency of the hyperpolarization following depolarization becomes shorter. Because of this earlier onset of hyperpolarization, it is likely that the counteraction between the preceding depolarization and subsequent hyperpolarization begins earlier, leading to a shortened duration of the depolarization (Fig. 9A and B). This shortened duration can result in a decrease in the peak amplitude of the depolarization for nonmonotonic PNs.
The O-DEP induced in layer 2 PNs is not followed by hyperpolarization. Because of the lack of hyperpolarization, it is unlikely that inhibitory inputs to layer 2 PNs are engaged in reducing the amplitude of their O-DEPs. There are at least two explanations for the decreased magnitude of the O-DEP in non-monotonic layer 2 PNs. First, the decreased amplitude of the O-DEP in non-monotonic layer 2 PNs can be thought to result from the decreased number of excitatory inputs activated at higher sound intensities. This situation is equivalent to a situation where stimulation is given at lower sound intensities. Lower sound intensities should induce lower O-DEP amplitudes but result in longer onset latencies of O-DEP. However, this is not the case, since the onset latency of the O-DEP of non-monotonic layer 2 PNs gradually shortens as sound intensity increases (Fig. 3). Second, there may be a preference for layer 2 PNs to receive afferents from layer 3 PNs with similar response features (nonmonotonic). As discussed above, layer 2 PNs are likely to receive excitatory afferents from layer 3 PNs (Ojima et al., 1991; Kang et al., 1994) and these connections can be activated by the present stimulation paradigm. If this layer 3 PN population responds in a non-monotonic manner and converges on subsets of layer 2 PNs, the depolarization of these layer 2 subsets are also likely to have a non-monotonic response pattern. In this case, the hyperpolarization cannot necessarily be generated after the depolarization. Furthermore, since the O-DEP latency shortens in non-monotonic layer 3 PNs with increasing sound intensity, this temporal shift of the O-DEP latency should also be reflected in the temporal pattern of O-DEP of layer 2 PNs. This is indeed the case (Figs 3 and 6B). It can be hypothesized that non-monotonic layer 3 PNs are preferentially connected to non-monotonic layer 2 PNs.
Monaural and Binaural Interactions
In layer 3 PNs of EI type, ipsilaterally and contralaterally induced synaptic potentials interact with each other to suppress the magnitude of the binaurally induced membrane potential. From a sample of six neurons, three patterns of suppressive interaction in membrane potential response are suggested.
The first and second patterns of binaural suppression can be explained by the temporal coincidence of depolarization and hyperpolarization induced by dominant and non-dominant ear stimulation. In the first pattern (see Fig. 10B), stimulation of the non-dominant, inhibitory ear alone induces a strong hyperpolarization inhibition with no preceding depolarization. This inhibitory response induced by non-dominant ear stimulation is likely to be responsible for the binaural suppression. Our findings show that the hyperpolarization initiated by the inhibitory ear stimulation temporally coincides with the depolarization peak induced by the excitatory ear stimulation. This enables the hyperpolarization to counteract the depolarization induced by the dominant ear stimulation to suppress its peak/falling phase. In the second pattern of binaural suppression (see Fig. 10C), the non-dominant ear does not induce a strong hyperpolarization inhibition, unlike the first pattern. Dominant ear stimulation alone rather induces a depolarization and hyperpolarization sequence. The onset/peak of the depolarization induced by dominant ear stimulation is delayed relative to that of the depolarization induced by non-dominant ear stimulation alone. Because of this delay, the dominant depolarization can coincide temporally with the hyperpolarization inhibition following the non-dominant depolarization (Calford and Semple, 1995). On binaural stimulation, this temporal interaction can result in a partial or full suppression of the dominant depolarization component, leading to a smaller amplitude depolarization on binaural stimulation than on dominant ear stimulation.
In the third pattern of binaural suppression (Fig. 10D), the O-DEP peak induced by dominant ear stimulation does not coincide with the hyperpolarization inhibition following the O-DEP induced by non-dominant ear stimulation. So the reduced magnitude of the depolarization induced binaurally cannot be explained by temporal summation of the membrane potentials. Such a response pattern would result from convergence of presynaptic neurons, located in binaural bands of the same features in the cortex (Imig and Adrian, 1977; Imig and Brugge, 1978; Middlebrooks et al., 1980) and/or medial geniculate body (Calford and Webster, 1981; Middlebrooks and Zook, 1983; Rodrigeus-Dagaeff et al., 1989).
Subthreshold Responses and their Relationships to Lateral Inhibition
The present study shows that layer 3 PNs respond to pure tone bursts with a sequence of depolarization and hyperpolarization at their best and flanking frequencies. This temporal sequence can underlie a phenomenon named ‘forward masking inhibition’ that has been described in primary auditory cortex (Calford and Semple, 1995). In forward masking, two pure tone stimuli are applied sequentially. A first pure tone (masker tone) with varied frequencies precedes a second pure tone (probe tone) with a fixed frequency (usually the best frequency). It is known that following masker tone stimulation at the best frequency, spike rate on probe tone stimulation is reduced. Such suppression can be explained by a temporal interaction of the hyperpolarization following the masker tone depolarization with the subsequent probe tone depolarization, i.e. if the probe tone depolarization coincides temporally with the hyperpolarizing inhibition that follows the preceding masker tone depolarization, the probe tone depolarization can be suppressed.
A spike rate suppression effect by tones with frequencies outside the response field of a neuron (lateral suppression) can also be explained by a similar temporal interaction. The occurrence of a sequence of subthreshold depolarization and following hyperpolarization suggests that probe tone depolarization can be suppressed if the masker tone stimulation induces a sequence of subthreshold depolarization followed by hyperpolarizing inhibition at frequencies flanking the best frequency (see Fig. 8A and D).
Relationship to Two Tone Simultaneous Masking
Cortical suppression has also been observed when the masker and probe tones are presented simultaneously (Sutter et al., 1999). The probe tone has a fixed intensity (10/20 dB above threshold) at the best frequency and the masker tone has various combinations of frequency and intensity. Suppression of the probe tone responses by the masker tone is observed when the masker tone is applied at frequencies neighboring the best frequency and at higher intensities.
Some of the response properties in this simultaneous suppression (see fig. 11 in Sutter et al., 1999) may be explained by latency dissociation of the masker and probe tone responses. The present data show that the onset latency of depolarization is shorter at the best frequency than at any other frequencies (see Fig. 2). They also show that the O-DEP latency becomes shorter as sound intensity increases (see Fig. 3). These results indicate that, depending on the combinations of frequency and intensity, the onset latency of the masker tone depolarization can become shorter relative to that of the probe tone depolarization. Because of this dissociation in latency, it is possible that the probe tone depolarization can coincide temporally with the hyperpolarizing inhibition following the masker tone depolarization.
Using intracellular recording from primary auditory cortical PNs, we have demonstrated membrane potential responses that can explain forward masking, lateral inhibition, non-monotonic rate level suppression and binaural suppression. These suppressive responses can be determined by the temporal interactions of excitatory and inhibitory membrane potentials.
Another mechanism underlying the suppressive responses could also be suggested. Membrane potential responses of a postsynaptic neuron may reflect the convergence of similarly responding presynaptic neurons. The suppressive membrane potential responses of layer 2 PNs and some binaural suppression in layer 3 PNs are likely to be explained by this mechanism.
The present results suggest that PNs in primary auditory cortex act as band-pass filters whose specific characteristics are defined by the temporal interactions of frequency-, level- and monaural/binaural-dependent excitation and inhibition. These interactions also show a laminar-specific distinction.
The authors thank Drs J. Brugge, M. Sutter and K.S. Rockland for their critical reading of the manuscript and for editing the English. We are also grateful to Dr K. Kishi for his continuing encouragement and to Mr M. Sakai for his technical assistance. The study was supported by a grant-in-aid (C) from the Japanese Ministry of Education, Science and Culture (no. 08680827).