Fast-spiking Interneurons Contribute to Propofol-induced Facilitation of Firing Synchrony in Pyramidal Neurons of the Rat Insular Cortex

All experiments were performed in accordance with the National Institutes of Health (Bethesda, Maryland) Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee at Nihon University (Tokyo, Japan). All efforts were made to minimize the number of animals used and the suffering of these animals in experiments.

The techniques for preparing and maintaining rat cortical slices in vitro were similar to those described previously.^12,16,17  Briefly, vesicular GABA transporter–Venus line A transgenic rats¹⁸  of either sex, ranging in age from 3 to 4 weeks, were deeply anesthetized with sodium pentobarbital (100 mg/kg, intraperitoneal) and decapitated. Tissue blocks including the insular cortex were rapidly removed and stored for 3 min in ice-cold modified artificial cerebrospinal fluid (230 mM sucrose, 2.5 mM KCl, 10 mM MgSO₄, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, 0.5 mM CaCl₂, and 10 mM D-glucose). Coronal slices were cut with a thickness of 350 μm using a microslicer (Linearslicer Pro 7, Dosaka EM, Japan). Slices were incubated at 32°C for 40 min in a submersion-type holding chamber that contained 50% modified artificial cerebrospinal fluid and 50% normal artificial cerebrospinal fluid (pH 7.35 to 7.40). Normal artificial cerebrospinal fluid contained 126 mM NaCl, 3 mM KCl, 2 mM MgSO₄, 1.25 mM NaH₂PO₄, 26 mM NaHCO₃, 2 mM CaCl₂, and 10 mM D-glucose. Modified and normal artificial cerebrospinal fluid were continuously aerated with a mixture of 95% O₂/5% CO₂. The slices were then placed in normal artificial cerebrospinal fluid at 32°C for 1 h and thereafter maintained at room temperature until they were used for recording.

The slices were transferred to a recording chamber that was perfused continuously with normal artificial cerebrospinal fluid at a rate of 2.0 ml/min. Multiple whole cell patch clamp recordings were obtained from a Venus-positive fluorescent interneuron and Venus-negative pyramidal neurons identified in layer V by a fluorescence microscope equipped with Nomarski optics (×40, Olympus BX51W1, Japan) and an infrared-sensitive video camera (C3077-78, Hamamatsu Photonics, Japan). The distance between recorded neurons was less than 200 μm. Electrical signals were recorded by amplifiers (Multiclamp 700B, Molecular Devices, USA), digitized (Digidata 1440A, Molecular Devices), observed online, and stored on a computer hard disk using Clampex (pClamp 10, Molecular Devices).

The composition of the pipette solution for recordings was 135 mM potassium gluconate, 5 mM KCl, 20 mM biocytin, 0.5 mM CaCl₂, 2 mM MgCl₂, 5 mM EGTA, 5 mM HEPES, and 5 mM magnesium adenosine triphosphate. The pipette solution had a pH of 7.3 and an osmolarity of 300 mOsm. The liquid junction potentials for current clamp and voltage clamp recordings were −9 mV. In the current study, the voltage was not corrected. Thin-wall borosilicate patch electrodes (2 to 5 MΩ) were pulled on a Flaming–Brown micropipette puller (P-97, Sutter Instruments, USA).

Recordings were obtained at 30 to 31°C to avoid the rundown of GABAergic inhibitory postsynaptic currents. The seal resistance was more than 10 GΩ, and only data obtained from electrodes with access resistance of 6 to 20 MΩ and less than 20% change during recordings were included in this study. Before the unitary inhibitory postsynaptic current recordings, the voltage responses of presynaptic and postsynaptic neurons were recorded by injecting long hyperpolarizing and depolarizing current pulses (300 to 1,000 ms) to examine basic electrophysiologic properties, including the input resistance, single-spike kinetics, voltage–current relationship, and repetitive firing patterns. Among Venus-positive fluorescent interneurons, fast-spiking neurons were identified by the following characteristics: a large afterhyperpolarization amplitude with very rapid repolarization, a short spike duration, and repetitive firing at an extremely high frequency (over 100 Hz) without adaptation (fig. 1).¹⁹  All neurons were recorded under the voltage clamp condition (holding potential = −45 mV) during unitary inhibitory postsynaptic current recording. Short depolarizing voltage step pulses (1 ms, 80 mV) were applied to the presynaptic fast-spiking neurons to induce action currents (fig. 2). Postsynaptic pyramidal neurons were depolarized by direct current injection. Then 2,6-diisopropylphenol (propofol; Sigma-Aldrich, USA) was dissolved in dimethyl sulfoxide at a concentration of 10 mM and diluted to 10 μM in the perfusate. In a part of the experiment, the effect of propofol was tested at 1 μM. The membrane currents and potentials were low pass–filtered at 5 to 10 kHz and digitized at 20 kHz.

Clampfit (pClamp 10, Molecular Devices) was used for analyses of electrophysiologic data. We defined the period between 150 ms before the first presynaptic action current up to the first presynaptic action current (0 ms) as epoch 0 (fig. 3). Similarly, the subsequent periods (150 ms each) were defined as epochs 1 to 7. To quantify the synchrony between two postsynaptic pyramidal neurons, the delay from the onset of the first action current in presynaptic fast-spiking neuron to the action potentials in postsynaptic pyramidal neurons was measured in each epoch in each trace (fig. 3C). The delay obtained from one of the postsynaptic pyramidal neurons was plotted against that of another postsynaptic pyramidal neuron (fig. 3D), and the distance from the line of identity was measured (fig. 3F). In each neuron, 10 traces were used for the measurement of line of identity, and each line of identity was obtained from each trace.

The synchronization index was calculated from the cross-correlogram obtained from 10 subsequent trials. The period for the analysis was set 50 ms before and after the first fast-spiking neuron spike. The synchronization index was obtained by dividing the number of spikes that occurred within the period of −25 to 25 ms by those within the periods of −50 ms to −25 ms and 25 to 50 ms. The synchronization index data in epochs 0, 5, 6, and 7 and in epochs 1 to 4 were summed as synchronization index in the period without and with presynaptic fast-spiking neuron stimuli, respectively. Pairs of pyramidal neurons were categorized into two classes: (1) connected pyramidal neurons, which received common inhibitory inputs from presynaptic fast-spiking neuron (e.g., fig. 3A); and (2) unconnected pyramidal neurons, which did not receive common presynaptic fast-spiking neuron inputs (e.g., fig. 4A). Therefore, the synchronization index was obtained from four categories: (1) connected pyramidal neurons without presynaptic fast-spiking neuron stimuli, (2) connected pyramidal neurons with presynaptic fast-spiking neuron stimuli, (3) unconnected pyramidal neurons without presynaptic fast-spiking neuron stimuli, and (4) unconnected pyramidal neurons with presynaptic fast-spiking neuron stimuli.

The data are mean ± SD. The sample size was based on the available data. Comparisons of the average distance from the line of identity between the control and propofol application conditions were conducted using a two-tailed unpaired t test. The synchronization index between the control and propofol application conditions was compared using a two-tailed paired t test. The normality of the data was tested by the Shapiro–Wilk test. No multiple comparisons were performed in this study. No outliers were excluded from analyses. Blinding was not used in this study. Statistical analyses were performed using SPSS (version 12.0, IBM, USA) and Prism (version 8.3.0, GraphPad Software, USA), and values of P < 0.05 were defined as statistically significant. No statistical power calculation was conducted before the study.

Among several subtypes of GABAergic interneurons, fast-spiking neurons send the most potent inhibitory inputs to pyramidal neurons,²⁰  and our previous study demonstrated that fast-spiking neuron–to–pyramidal neuron connections are the most sensitive to propofol-induced facilitation of unitary inhibitory postsynaptic currents.¹⁶  Therefore, in this study, triple or quadruple whole cell patch clamp recordings were performed from a fast-spiking neuron, which was Venus-positive and showed a short spike duration with a large afterhyperpolarization and a high frequency of repetitive spike firing without spike adaptation, and pyramidal neurons, which were Venus-negative and showed spike adaptation during repetitive spike firing (fig. 1). We focused on the effects of propofol-induced facilitation of GABAergic synaptic transmission, particularly mediated by fast-spiking neurons, on spike synchronization between pyramidal neurons. The connection rate between a fast-spiking neuron and a pyramidal neuron was approximately 50%, which was almost comparable to what we previously reported.¹⁶  Therefore, in the case of triple whole cell patch clamp recording from a fast-spiking neuron and two pyramidal neurons, the rate at which both pyramidal neurons received fast-spiking neuron inputs was approximately one quarter. In contrast, few pyramidal neuron–pyramidal neuron connections were observed in the current study.

Figure 2 shows a typical example of propofol-induced spike synchrony of pyramidal neurons. Triple whole cell patch clamp recordings were performed from one Venus-positive fast-spiking neuron and two Venus-negative pyramidal neurons (pyramidal neuron 1 and pyramidal neuron 2; fig. 2A). To examine whether the fast-spiking neuron projected to pyramidal neuron 1 and pyramidal neuron 2, action currents were induced in the fast-spiking neuron by applying a short depolarizing voltage pulse injection (1 ms, 80 mV; fig. 2B, upper two traces). In response to presynaptic action currents in the fast-spiking neuron, outward currents were observed in both pyramidal neuron 1 and pyramidal neuron 2, which were voltage clamped at −45 mV (fig. 2B, lower traces), indicating that both pyramidal neurons received inhibitory inputs from the fast-spiking neuron.

After confirming synaptic connections from the fast-spiking neuron to pyramidal neurons, postsynaptic pyramidal neuron 1 and pyramidal neuron 2 were recorded under a current clamp condition. To examine the effect of presynaptic inhibitory inputs from the fast-spiking neuron on spike synchrony between pyramidal neuron 1 and pyramidal neuron 2, repetitive spike firing was induced by depolarized direct currents to both pyramidal neurons. The depolarizing currents were set at the intensity (approximately 200 pA) that induced repetitive spike firing at 5 to 15 Hz.

In the first series of the experiment, four action currents were applied at the alpha range (100-ms interspike intervals, 10 Hz). Although the fast-spiking neuron sent inhibitory inputs to pyramidal neuron 1 and pyramidal neuron 2, presynaptic fast-spiking neuron activation had little effect on postsynaptic pyramidal neuron synchrony between pyramidal neuron 1 and pyramidal neuron 2 under the control conditions (fig. 2C). On the other hand, a bath application of 10 μM propofol facilitated spike synchrony between pyramidal neuron 1 and pyramidal neuron 2 just after the action current was induced in the fast-spiking neuron (fig. 2D, arrows).

In in vivo preparations, fast-spiking neurons often show burst firing in which several spikes are elicited at a high frequency, such as more than 20 Hz,^21,22  and thus, it is reasonable to estimate postsynaptic spike synchronization by applying presynaptic train pulses. Therefore, we applied four sequential voltage pulses (five train pulses at 100 Hz) with interburst intervals of 150 ms to the fast-spiking neuron (fig. 2E) during repetitive spike firing of pyramidal neuron 1 and pyramidal neuron 2. Even though the train pulses to the fast-spiking neuron induced a larger impact on postsynaptic pyramidal neurons, most pyramidal neuron pairs showed asynchronous spike firing under the control conditions. However, well-synchronized firing patterns were observed when propofol was administered (fig. 2F). Therefore, in the following experiment, we used this train pulse protocol to estimate the effect of the fast-spiking neuron → pyramidal neuron connection on synchronous spike firing among pyramidal neurons.

Figure 3, A to D, shows another example obtained from one presynaptic fast-spiking neuron and two postsynaptic pyramidal neuron connections. The spike latencies from the onset of each epoch (vertical dotted lines in fig. 3B) were measured in pyramidal neuron 1 and pyramidal neuron 2 (fig. 3C), and the latency of pyramidal neuron 2 was plotted against that of pyramidal neuron 1 (fig. 3D). The horizontal and vertical axes show the spike latency of pyramidal neuron 1 and pyramidal neuron 2, respectively. Figure 3E shows the plots of spike timing of pyramidal neuron 1 and pyramidal neuron 2 obtained from 31 pyramidal neuron pairs. In epoch 0 in the control, the plots were widely distributed. Some plots showed a cluster around the identical lines in epochs 1 and 2 in the control; however, epochs 3 to 7 again showed randomly distributed plots in the control (fig. 3E).

Similar to the results obtained for the controls, epoch 0 during the bath application of propofol showed a random distribution of plots. In contrast, the propofol group showed a cluster in the spike timing plots on the line of identity in epoch 1, suggesting spike synchronization between pyramidal neuron 1 and pyramidal neuron 2. This propofol-induced effect was also observed in epochs 2 and 3 and was weakly observed in epoch 4 (fig. 3E). However, epochs 5 to 7 did not show such deflection of plots. This result indicates that propofol facilitates the spike synchrony between pyramidal neuron 1 and pyramidal neuron 2.

For a quantitative analysis of the synchronized spike firing in pyramidal neuron 1 and pyramidal neuron 2, an absolute value, l, was measured and averaged (fig. 3, F and G). In the case in which spikes in pyramidal neuron 1 were completely synchronized to those in pyramidal neuron 2, the plots were aligned on the identical line, and average distance from the line of identity was 0. Figure 3G shows typical asynchronized and synchronized examples, as shown in figure 3E (epochs 0 and 1 during propofol application). The index, average distance from the line of identity, in epoch 1, in which the presynaptic fast-spiking neuron was activated, was smaller (average distance from the line of identity = 19.7 ± 19.2) than that in epoch 0 (average distance from the line of identity = 32.2 ± 25.6), supporting that average distance from the line of identity reflects the degree of synchronization.

Figure 3H shows the summary of average distance from the line of identity obtained from 31 pyramidal neuron pairs in the control and under the 10 μM propofol application. In comparison to the control, average distance from the line of identity in the propofol application was significantly smaller in epoch 1 (29.2 ± 24.1 [number of spikes: n = 510] to 19.7 ± 19.2 [n = 242], P < 0.001, difference in means = −9.4, 95% CI = −12.6 to −6.2), epoch 2 (32.0 ± 24.3 [n = 518] to 20.0 ± 19.0 [n = 255], P < 0.001, difference in means = −11.9, 95% CI = −15.1 to −8.8), epoch 3 (33.0 ± 23.5 [n = 535] to 23.9 ± 20.7 [n = 316], P < 0.001, difference in means = −9.2, 95% CI = −12.2 to −6.1), and epoch 4 (33.4 ± 24.8 [n = 613] to 30.4 ± 24.2 [n = 446], P = 0.048, difference in means = −3.0, 95% CI = −6.0 to −0.0). In contrast, propofol had little effect on the distances from the line of identity in epoch 0 (30.1 ± 23.4 [n = 430] to 32.2 ± 25.6 [n = 348], P = 0.225), epoch 5 (37.2 ± 26.3 [n = 692] to 34.3 ± 24.2 [n = 505], P = 0.056), epoch 6 (35.3 ± 23.9 [n = 609] to 36.6 ± 25.8 [n = 500], P = 0.371), and epoch 7 (35.2 ± 24.4 [n = 634] to 36.6 ± 24.8 [n = 465], P = 0.367). These results suggest that propofol facilitates spike synchronization just after GABAergic inputs are received from fast-spiking neurons but has little effect on the synchrony of spontaneous spike firing patterns.

If inhibitory inputs from fast-spiking neurons regulate the synchronization of pyramidal neuron spike firing, it is likely that uncommon inhibitory inputs do not induce synchronization in pyramidal neurons. To test this possibility, we next examined the effects of propofol on spike synchrony between two pyramidal neurons, only one of which received inhibitory input from the presynaptic fast-spiking neuron (fig. 4A).

In epochs 1 to 4, in which pyramidal neuron 1 received inhibitory inputs from the fast-spiking neuron, propofol synchronized the spikes in pyramidal neuron 1 (fig. 4B). In contrast, propofol had little effect on spike timing in pyramidal neuron 2 that did not receive inhibitory inputs from the fast-spiking neuron (fig. 4B). Consequently, propofol failed to facilitate spike synchrony between pyramidal neuron 1 and pyramidal neuron 2 even immediately after the action current was induced in the fast-spiking neuron; spike timing plots were almost randomly distributed in the control and during propofol application (fig. 4C).

In fact, average distance from the line of identity values obtained from 16 pyramidal neuron pairs was not significantly changed by propofol in each epoch: 33.7 ± 23.2 (n = 284) to 34.0 ± 24.7 (n = 229, P = 0.870) in epoch 0; 34.3 ± 23.8 (n = 271) to 36.3 ± 26.8 (n = 140, P = 0.447) in epoch 1; 32.0 ± 21.7 (n = 250) to 32.7 ± 22.9 (n = 139, P = 0.751) in epoch 2; 36.7 ± 25.0 (n = 259) to 35.5 ± 26.7 (n = 175, P = 0.644) in epoch 3; 34.1 ± 23.1 (n = 256) to 34.8 ± 24.5 (n = 166, P = 0.765) in epoch 4; 37.6 ± 26.3 (n = 346) to 34.7 ± 24.8 (n = 246, P = 0.180) in epoch 5; 34.8 ± 24.2 (n = 323) to 36.1 ± 26.6 (n = 263, P = 0.531) in epoch 6; and 35.8 ± 26.4 (n = 352) to 36.6 ± 26.6 (n = 263, P = 0.716) in epoch 7 (fig. 4, C and D). These results support the aforementioned idea that propofol-induced spike synchrony between pyramidal neurons is mediated by common inhibitory inputs from fast-spiking neurons.

The current experiment revealed that fast-spiking neuron → pyramidal neuron inhibitory inputs synchronized postsynaptic pyramidal neuron spikes during the bath application of propofol. Although thalamocortical projections play a critical role in propofol-induced alpha rhythm generation in the cerebral cortex,^5,6  cortical local circuits should have the capacity to maintain alpha oscillation. However, it remains unknown whether the propofol-induced synchronization of postsynaptic spike firing in pyramidal neurons occurs even when presynaptic fast-spiking neurons are activated at a different frequency range. To answer this question, the presynaptic fast-spiking neuron was activated with interburst intervals of 1,050 ms (1 Hz), 250 ms (4 Hz), 150 ms (6.7 Hz), 100 ms (10 Hz), and 75 ms (13.3 Hz).

Figure 5 shows a typical example of postsynaptic pyramidal neuron firing responding to the presynaptic fast-spiking neuron action potentials induced at the frequencies described above. At all frequencies, the first burst of the fast-spiking neuron action currents in epoch 1 induced a period without spikes for 50 to 150 ms (arrowheads; blank period), in which GABA_A receptor–mediated inhibitory postsynaptic potentials occurred (fig. 1C), in each pyramidal neuron. However, the subsequent epochs (epochs 2 to 4) exhibited a less obvious blank period in the controls.

The blank period was more prominent, had a longer duration, and had a larger hyperpolarization during propofol application than in the controls. This blank period was induced with similar timing, and therefore, the subsequent spikes were aligned among pyramidal neurons. However, the subsequent bursts (second, third, and fourth spike trains in the fast-spiking neuron) at 1, 4, and 13.3 Hz showed less synchronized spike firing among pyramidal neurons during propofol application. In contrast, remarkable propofol-induced spike synchrony among pyramidal neurons was observed in the period when the presynaptic fast-spiking neuron was activated at 6.7 and 10 Hz.

To quantify the relationship between the synchrony of pyramidal neuron spike firing and the frequency of presynaptic fast-spiking neuron burst spike firing, we made a cross-correlogram, and the synchronization index was calculated (see Materials and Methods). Figure 6, A and B, shows examples of the cross-correlograms obtained from the results shown in figure 5. The synchronization index was compared between the control and propofol application (fig. 6C). In the cases of 1-, 4-, and 13.3-Hz stimulation, the synchronization index obtained from each category showed no statistically significant change in the propofol application (connected pyramidal neurons without presynaptic stimuli: for 1 Hz stimulation [1.2 ± 0.6 to 1.2 ± 0.5, n = 26 pyramidal neuron pairs, P = 0.869], for 4 Hz stimulation [1.2 ± 0.3 to 1.1 ± 0.3, n = 28 pyramidal neuron pairs, P = 0.328], and for 13.3 Hz stimulation [1.6 ± 0.7 to 2.1 ± 1.0, n = 18 pyramidal neuron pairs, P = 0.074]; connected pyramidal neurons with presynaptic stimuli: for 1 Hz stimulation [1.0 ± 0.2 to 1.0 ± 0.2, n = 28 pyramidal neuron pairs, P = 0.950], for 4 Hz stimulation [1.3 ± 0.5 to 1.6 ± 1.0, n = 28 pyramidal neuron pairs, P = 0.135], and for 13.3 Hz stimulation [1.2 ± 0.7 to 1.5 ± 0.7, n = 8 pyramidal neuron pairs, P = 0.532]; unconnected pyramidal neurons without presynaptic stimuli: for 1 Hz stimulation [1.2 ± 0.5 to 1.2 ± 0.5, n = 22 pyramidal neuron pairs, P = 0.969], for 4 Hz stimulation [1.3 ± 0.5 to 1.2 ± 0.4, n = 22 pyramidal neuron pairs, P = 0.605], and for 13.3 Hz stimulation [1.3 ± 0.8 to 2.4 ± 1.9, n = 14 pyramidal neuron pairs, P = 0.082]; and unconnected pyramidal neurons with presynaptic stimuli: for 1 Hz stimulation [1.1 ± 0.3 to 1.0 ± 0.2, n = 23 pyramidal neuron pairs, P = 0.388], for 4 Hz stimulation [1.3 ± 0.6 to 1.2 ± 0.3, n = 22 pyramidal neuron pairs, P = 0.601], and for 13.3 Hz stimulation [1.3 ± 0.8 to 1.3 ± 0.9, n = 11 pyramidal neuron pairs, P = 0.934]).

In the cases of 6.7- and 10-Hz stimulation, connected pyramidal neurons with presynaptic stimuli showed a statistically significant increase in the synchronization index (6.7 Hz: 1.4 ± 0.5 to 2.3 ± 1.5, n = 27 pyramidal neuron pairs, P = 0.004, difference in means = 0.9, 95% CI = 0.3 to 1.5; 10 Hz: 1.5 ± 0.5 to 2.2 ± 1.0, n = 19 pyramidal neuron pairs, P = 0.030, difference in means = 0.7, 95% CI = 0.1 to 1.4), although the other stimulation categories did not show a statistically significant change in the synchronization index (connected pyramidal neurons without presynaptic stimuli: for 6.7 Hz stimulation [1.3 ± 0.5 to 1.3 ± 0.5, n = 28 pyramidal neuron pairs, P = 0.591] and for 10 Hz stimulation [1.4 ± 0.5 to 1.8 ± 0.8, n = 23 pyramidal neuron pairs, P = 0.056]; unconnected pyramidal neurons without presynaptic stimuli: for 6.7 Hz stimulation [1.2 ± 0.6 to 1.4 ± 0.7, n = 21 pyramidal neuron pairs, P = 0.267] and for 10 Hz stimulation [1.4 ± 0.5 to 1.5 ± 0.7, n = 19 pyramidal neuron pairs, P = 0.576]; unconnected pyramidal neurons with presynaptic stimuli: for 6.7 Hz stimulation [1.2 ± 0.5 to 1.3 ± 0.6, n = 21 pyramidal neuron pairs, P = 0.358] and for 10 Hz stimulation [1.5 ± 0.6 to 2.1 ± 1.7, n = 19 pyramidal neuron pairs, P = 0.141]).

In agreement with the result of the synchronization index increase by 10 μM propofol in the category of connected pyramidal neurons with presynaptic stimuli, 1 μM propofol significantly increased the synchronization index at 10 Hz (1.9 ± 1.3 to 3.7 ± 2.6, n = 12 pyramidal neuron pairs, P = 0.042, difference in means = 1.8, 95% CI = 0.1 to 3.5) without changing the synchronization index at 4 Hz (1.3 ± 0.5 to 1.5 ± 1.2, n = 14 pyramidal neuron pairs, P = 0.577). On the other hand, 1 μM propofol induced a slight but insignificant increase in the synchronization index at 6.7 Hz (1.6 ± 0.7 to 2.0 ± 3.4, n = 21 pyramidal neuron pairs, P = 0.587), which contradicts the result obtained for 10 μM propofol. These results indicate that pyramidal neurons show synchronous firing in response to fast-spiking neuron activation with an interburst interval of 100 to 150 ms.

Using a triple or quadruple whole cell patch clamp technique, we found that the intravenous anesthetic propofol facilitates spike synchrony among pyramidal neurons by enhancing inhibitory input from fast-spiking neurons in rat cortical slices. The bath application of 1 and 10 μM propofol promoted postsynaptic pyramidal neuron firing synchrony when a presynaptic fast-spiking neuron was activated with 100- and 100/150-ms interburst intervals, respectively. On the other hand, propofol failed to synchronize pyramidal neuron firing when the presynaptic fast-spiking neuron was activated with 1,050-, 250-, or 75-ms interburst intervals. These findings suggest that fast-spiking neurons play a critical role in inducing the synchronization of pyramidal neuron firing during propofol application.

Propofol prolongs the decay time of GABA_A receptor–mediated inhibitory currents, and among the connections from various interneurons to pyramidal neurons, the fast-spiking neuron → pyramidal neuron connection is the most sensitive to propofol.¹⁶  Propofol-induced increases in GABA_A receptor–mediated Cl⁻ conductance reduce the spike firing frequency.¹²  Several studies have demonstrated that the enhancement of inhibitory input results in facilitating neural synchronization. For example, spike-timing precision and neuronal synchrony are enhanced by synaptic inhibition in the amygdala.²³  In the somatosensory cortex, synchrony of spike firing between interneurons correlates with inhibitory synaptic transmission.²⁴  Therefore, the propofol-induced enhancement of inhibitory input from fast-spiking neurons is likely to increase spike synchrony in the insular cortex, part of which may be mediated by a propofol-induced reduction of asynchronous spike firing.

To test this hypothesis, the effects of propofol on pyramidal neuron synchrony were examined in minimum neuronal circuits consisting of two or three postsynaptic pyramidal neurons and one presynaptic fast-spiking neuron using multiple whole cell recordings. We found that propofol effectively synchronized spike firing among the pyramidal neurons. It is worth noting that fast-spiking neurons are electrically coupled via gap junctions because these electrical synapses contribute to the synchronization of spike firing among fast-spiking neurons.^25,26  Therefore, the synchronized activity in the minimum circuits may propagate to the adjacent local circuits, resulting in spike firing synchrony of pyramidal neurons distributed in a large cortical area. The weaker synchronization in epoch 4 compared with that in epochs 1 to 3 may be due to the subsequent depression of inhibitory postsynaptic currents in fast-spiking neuron → pyramidal neuron connections, as we previously reported.¹⁶ 

We chose five train bursts to activate fast-spiking neurons because of the lower number of the inhibitory inputs from fast-spiking neuron to pyramidal neurons in the slice preparation compared with those in in vivo preparations; this lower number of inputs in the slice preparation means that some chemical and electrical connections are lost. In the minimum local circuit, 1 and 10 μM propofol consistently promoted fast-spiking neuron-activated synchronization when the fast-spiking neuron was activated with 100 ms of the interburst interval. However, there was a partial discrepancy in propofol-induced fast-spiking neuron-activated synchronization between 10 and 1 μM propofol: synchronization induced by fast-spiking neuron activation at 150 ms of the interburst interval was potentiated by 10 μM but not 1 μM propofol. The larger effect on GABAergic currents induced by 10 μM propofol may expand the effective range of synchronization compared with that induced by 1 μM propofol. We consider that propofol-induced alpha oscillation is not always induced in the cortical circuit. As previously reported, thalamocortical and corticothalamic circuits play an essential role in the induction of alpha oscillations (8 to 13 Hz) in the cerebral cortex, and alpha oscillations cannot be induced without thalamic inputs.^5,6  In sensory cortices such as the visual and barrel cortices, layer IV neurons are the principal target of thalamocortical inputs and project to layers II/III. Layer II/III glutamatergic neurons then project to layer V/VI neurons, some of which project their axons to the thalamus.²⁷  These glutamatergic circuits also receive GABAergic inputs. Therefore, in addition to the thalamocortical and corticothalamic circuits, the cortical local circuits between them possibly contribute to alpha oscillations by mediating alpha oscillation activity from thalamocortical inputs to corticothalamic outputs. Our finding of propofol-induced synchrony close to the alpha range in the minimum cortical circuits shows the capacity of the cortical circuit for maintaining alpha synchrony, at least when cortical fast-spiking neurons are activated.

The current study demonstrated that 10 μM propofol increased synchronized activity among pyramidal neurons by fast-spiking neuron activation not only at 10 Hz (alpha rhythm) but also at 6.7 Hz, which is slightly lower than the frequency of the alpha rhythm. This finding suggests that other mechanisms, such as non–fast-spiking neuron → pyramidal neuron connections and inhibitory connections among fast-spiking neurons and non–fast-spiking neurons, are involved in alpha rhythm generation in addition to fast-spiking neuron → pyramidal neuron connections. In the insular cortex, there are several other types of GABAergic interneurons: late-spiking, regular-spiking, and low-threshold spike neurons.¹⁹  Several studies have demonstrated the critical role of non–fast-spiking neurons in propofol-induced alpha rhythms. For example, computational models showed that low-threshold spike neurons, a subtype of non–fast-spiking neurons, are crucial for propofol-induced alpha rhythms,^1,28  although low-threshold spike neurons can be replaced with fast-spiking neurons.¹ 

On the other hand, somatostatin-expressing non–fast-spiking neurons show preferential inhibition of other interneurons, including fast-spiking neurons, rather than pyramidal neurons.²⁰  In addition, somatostatin-expressing non–fast-spiking neurons are principal targets of vasoactive intestinal peptide–expressing non–fast-spiking neurons, whereas parvalbumin-expressing fast-spiking neurons preferentially inhibit other fast-spiking neurons and pyramidal neurons.²⁰  With this information, along with our previous study demonstrating the preferential facilitation of inhibitory input in fast-spiking neuron → pyramidal neuron rather than non–fast-spiking neuron → pyramidal neuron/interneurons and fast-spiking neuron → interneurons by propofol,¹⁶  it is reasonable to postulate that non–fast-spiking neurons play a minor role in the spike synchrony of pyramidal neurons. Further studies are required to identify the modulatory effect of a non–fast-spiking neuron network on fast-spiking neuron firing during anesthesia.

GABA release from fast-spiking neurons to pyramidal neurons induces inhibitory postsynaptic currents that are principally mediated by GABA_A receptors.¹⁹  Our previous study demonstrated that the 80% to 20% decay time of unitary inhibitory postsynaptic currents in fast-spiking neuron → pyramidal neuron connections in the insular cortex was 12 ms,¹⁶  and as shown in figure 1B, the duration of inhibitory postsynaptic currents at baseline was approximately 50 ms. The time course of inhibitory postsynaptic currents often overlies the afterhyperpolarization evoked by an action potential. Therefore, the summative effect of inhibitory postsynaptic potentials and afterhyperpolarization may be potentiated by propofol (10 μM), which prolongs the decay kinetics approximately twofold (approximately 25 ms) with a slight increase in the peak amplitude.¹⁶  This propofol-induced modulation of the temporal kinetics of inhibitory postsynaptic currents may play a critical role in regulating the spike timing of postsynaptic pyramidal neurons. Under current clamp mode, repetitive spike firing is reset by unitary inhibitory postsynaptic potentials. This role of GABA_A receptors was potentiated by propofol, and therefore, spikes rarely occurred during this inhibitory postsynaptic potential period (fig. 2D). As a result, the next spike was temporally aligned among pyramidal neurons that received inhibitory postsynaptic potentials from a common fast-spiking neuron.

Fast-spiking neuron spike firing at the alpha range is necessary for pyramidal neuron spike firing around the alpha range because fast-spiking neuron spike firing at other bands does not induce synchronized activities in pyramidal neurons. Autapses are a crucial factor in regulating fast-spiking neuron firing. Autaptic transmission mediated by GABA_A receptors is frequently observed in cortical fast-spiking neurons and improves the precision of spike timing rigidity.^29,30  Propofol is likely to prolong decay times of autaptic inhibitory postsynaptic current, which may result in slowing fast-spiking neuron firing so that it is within the alpha range. In fact, when single action currents were generated in the presynaptic fast-spiking neuron at the alpha range, postsynaptic pyramidal neurons tended to show synchronous firing (fig. 2D), although the modulatory effect of the fast-spiking neuron on pyramidal neuron firing was lower for five train pulses.

The current study demonstrated that cortical pyramidal neurons tend to be synchronized at a certain frequency that includes the alpha range by the bath application of propofol using slice preparations. Such synchronization preferences among pyramidal neurons may be dependent on the modulation of the excitatory and inhibitory balance of inputs to pyramidal neurons, and the other anesthetics may establish synchronous firing at different ranges. This modulation might be a mechanism for various states of consciousness, e.g., hypersynchrony during seizure and alpha coma.³¹  Our findings may elucidate some of the mechanisms at the cellular level for how anesthetics and pathophysiological status cause a certain EEG pattern in humans.

The authors thank Drs. Keisuke Kaneko, D.D.S., Ph.D., and Kiyofumi Yamamoto, Ph.D., Department of Pharmacology, Nihon University School of Dentistry, Tokyo, Japan, for their excellent technical support with the experiments. Vesicular GABA transporter–Venus transgenic rats were generated by Yuchio Yanagawa, M.D., Ph.D., Department of Genetic and Behavioral Neuroscience, Gunma University Graduate School of Medicine, Maebashi, Gunma, Japan; Masumi Hirabayashi, Ph.D., Section of Mammalian Transgenesis, Center for Genetic Analysis of Behavior, National Institute for Physiological Sciences, Okazaki, Japan; and Yasuo Kawaguchi, M.D., Ph.D., Division of Cerebral Circuitry, National Institute for Physiological Sciences, Okazaki, Japan, using pCS2-Venus provided by Atsushi Miyawaki, M.D., Ph.D., Laboratory for Cell Function and Dynamics, Brain Science Institute, RIKEN, Wako, Japan.

This work was supported by Grant-in-Aid for Scientific Research Nos. 15K20561 and 18K09731 (to Dr. Koyanagi), 16K11765 and 19K10345 (to Dr. Oi), and 16H05507 and 19H03821 (to Dr. Kobayashi) from the Japan Society for the Promotion of Science, Tokyo, Japan, and by the Sato Fund and research grants from the Dental Research Center, Nihon University School of Dentistry, Tokyo, Japan.

The authors declare no competing interests.