[1] 12w Single-unit activity in the in vitro entorhinal cortex during carbachol-induced field oscillations

[1] 250w The muscarinic receptor agonist carbachol (CCh) can induce activity in the theta range (4 -15 Hz) in the entorhinal cortex (EC), but the underlying network mechanisms remain unclear. Here, we investigated the interplay between interneurons and principal cells in the EC during CChinduced theta-like field oscillations in an in vitro brain slice preparation using tetrodes. Field oscillations at 10.1 Hz (IQR = 9.5 -10.9 Hz) occurred during bath application of CCh (100 μM; n = 32 experiments) and were associated with single-unit (n = 189) firing. Interneuron activity increased before principal cell activity at the onset of the oscillations and both interneurons and principal cells fired at specific oscillation phases with interneurons preceding principal cells, suggesting that interneurons modulate principal cell activity during such oscillations. The regularity of occurrence of CCh-induced oscillations was abolished by applying the GABA A receptor antagonist picrotoxin (100 μM; n = 13). These effects were accompanied by changes in firing with principal cells discharging action potentials before interneurons, along with a loss of preferred firing phase for principal cells in relation to the oscillation peaks. Blocking ionotropic glutamatergic transmission abolished CCh-induced field oscillations (n = 6), suggesting that ionotropic glutamatergic receptor signaling is necessary for their generation. Our results show that neuronal network interactions leading to CCh-induced theta-like field oscillations rest on the close interplay between interneurons and principal cells and that interneurons modulate principal cell activity during such oscillatory activity. Moreover, they underscore the role of ionotropic glutamatergic transmission in this type of oscillations.

[1] 122w Oscillations of neuronal networks are fundamental in brain functions (Buzsáki, 2015). During development, they are observed as giant depolarizing events that synchronize neuronal activity to assemble and to shape synapse and network formation (Griguoli and Cherubini, 2017). In the adult brain, oscillations have been most often studied in limbic networks, particularly in the hippocampus and in the entorhinal cortex (EC). These brain regions generate theta oscillations (4 -12 Hz) during sensorimotor integration, spatial navigation, memory, and learning (Düzel et al., 2010;Buzsáki and Moser, 2013;Hasselmo and Stern, 2014;Colgin, 2016), and could shape synaptic plasticity that underlie these tasks (Pavlides et al., 1988). These oscillations appear to be shaped by interneuron activity that synchronizes principal cells during these oscillatory patterns (Cobb et al., 1995).

[2] 145w The limbic system includes brain structures that have unique circuitries (Bland, 1986;Bland and Oddie, 2001). Among them, the EC plays a fundamental role in regulating the transfer of information between the hippocampal formation and other brain structures (Montoya and Sainsbury, 1985;van Strien et al., 2009;Deshmukh et al., 2010). The EC is known to receive cholinergic inputs from the medial septum, the diagonal band of Broca, and the basal forebrain (Beckstead, 1978;Alonso and Köhler, 1984;Insausti et al., 1987;Gaykema et al., 1990;Heys et al., 2012). These cholinergic inputs can influence memory, spatial navigation, and synaptic plasticity (Yun et al., 2000;McGaughy et al., 2005;Tanninen et al., 2015), as well as EC oscillatory activities (Pilly and Grossberg, 2013;Vandecasteele et al., 2014). Therefore, understanding how cholinergic signaling modifies the interplay between excitation and inhibition in the EC should help us to understand how rhythmic activities occur in this limbic area.

[3] 98w The cholinergic receptor agonist carbachol (CCh) induces oscillations in the theta frequency range when applied in vitro to different structures of the limbic system. Konopacki et al. (1992) were the first to report in vitro theta-like oscillation in the EC during bath application of CCh. These CCh-induced oscillations have been shown to depend on the activation of cholinergic M1 receptors (Cataldi et al., 2011;Kowalczyk et al., 2013). In addition, the generators of these CCh-induced oscillations are not only found in different limbic structures but also self-sustaining and independent of external inputs (D'Antuono et al., 2001;Lévesque et al., 2017).

[4] 149w While some investigators have later found that CCh can also induce oscillations in the gamma frequency range in the EC (Dickson et al., 2000(Dickson et al., , 2003;;Dickson and de Curtis, 2002), CCh-induced oscillations continue to be used as an in vitro model of the theta rhythm that occurs in vivo (Kowalczyk et al., 2013). However, it cannot be overlooked that in vivo theta oscillations occurring in awake, freely moving animals are atropine-resistant and perhaps less dependent on cholinergic inputs and on interneuron activity (Petsche et al., 1962;Kramis et al., 1975;Williams and Kauer, 1997;Buzsáki, 2002). In the present study, we employed tetrode recordings in the EC to (i) characterize the participation of putative interneurons and principal cells CCh-induced field oscillations, (ii) identify the changes induced by γ-aminobutyric acid type A (GABA A ) receptor antagonism on this oscillatory pattern, and (iii) establish the role played by ionotropic glutamatergic transmission.

[1] 164w Using tetrodes, we acquired raw data that were filtered between 300 -3000 Hz to detect action potentials of putative neurons and between 1 -500 Hz to visualise field potential activity (Figure 1A). As previously reported (Freund and Buzsáki, 1996;Csicsvari et al., 1998;Sirota et al., 2008;Sakata and Harris, 2009;Lévesque et al., 2016), action potential waveforms generated by putative interneurons (Figure 1Ba: green, Int) were narrower and more symmetric than those recorded from the putative principal cells (Figure 1Ca: blue, PC). Irrespective of the neuronal type, we could follow both types of single-units throughout the recording as revealed by the similarity between the averaged action potential waveform shapes of the first and last 50 action potentials (Figure 1Bb and Figure 1Cb). In addition, we could follow the same single-units during different pharmacological manipulations: from the baseline, CCh only condition, to a treatment condition with the application of either the GABA A receptor antagonist PTX or the ionotropic glutamatergic blockers CPP and NBQX (Figure 1Da and b).

[2] 92w When we applied k-means clustering to the 3 waveform variables extracted from all single-units, we could distinguish interneurons from principal cells (Figure 1E). Overall, from 32 slices obtained from 17 rats, 189 single-units were identified. In CCh only condition, 106 singleunits were classified as interneurons and 17 single-units were classified as principal cells. In the CCh + PTX condition (n = 13 slices), we identified 27 interneurons and 4 principal cells. Finally, in the CCh + CPP + NBQX condition (n = 6 slices), we identified 30 interneurons and 5 principal cells.

[3] 206w By filtering the field recordings between 1 -500 Hz, we identified field oscillations during application of CCh. As previously reported (Dickson and Alonso, 1997;Williams and Kauer, 1997;Cataldi et al., 2011;Kowalczyk et al., 2013;Lévesque et al., 2017), CCh induced regular oscillations (median duration = 0.6 s, IQR = 0.3 -1.1 s) in the EC (Figure 2). Visual inspection of the data revealed a diversity in single-unit discharge patterns. To quantitatively describe the diversity in activity pattern, we calculated: 1) the percentage of oscillations that contained singleunit discharges, 2) the percentage of intervals between oscillations that contained single-unit discharges, and 3) the difference in the ratios of spike discharges that occurred during CChinduced oscillations and during intervals between the oscillations. Single-units with steady and preferential firing during most of CCh-induced oscillations were categorized as type A (Figure 2A). Type B single-units were cells that fired both between and during the CCh-induced oscillations (Figure 2B). Lastly, single-unit that fired sporadically during the recording were categorized as type C (Figure 2C). As shown in Figure 2D, we found 42 type A (40%), 13 type B (12%), and 51 type C (48%) interneurons as well as 3 type A (18%), 5 type B (29%), and 9 type C (53%) principal cells.

[4] 154w To better understand the relationship between single-unit activity and CCh-induced oscillations, peri-event histograms of the average spike densities were centered around the onset of CCh-induced oscillations with the duration of the field oscillation normalized to 0 -100 %. As illustrated in Figure 3, we found increases in both type A interneuron and principal cell firings that were sustained throughout the field oscillations (Figure 3A). Upon closer inspection of the average spike densities around the onset of CCh-induced oscillation, type A interneurons appeared to increase their activity before principal cells (Figure 3A). Despite an increase in type B interneuron during CCh-induced oscillation, the increase was not statistically significant (Figure 3B). Type B principal cells firing also did not exhibit significant increases during CChinduced oscillations (Figure 3B). Type C interneurons only briefly significantly increased their firing near the end of oscillations (Figure 3C) while type C principal cell firing did not significantly increase during CCh-induced oscillations.

[5] 150w Next, we investigated the role of GABA A receptor signaling in CCh-induced oscillations and single-unit activity by bath applying PTX in addition to CCh. Upon PTX addition, we observed changes in the dynamics of the field oscillations (Figure 4A). Specifically, we found a nonsignificant increase in the average interval of occurrence, which changed from 2.9 s (IQR = 2.5 -3.2 s) in absence of PTX to 3.4 s (IQR = 1.9 -4.0 s) in presence of PTX (Figure 4B). The average duration of the CCh-induced oscillations significantly increased from 0.6 s (IQR = 0.3 -1.1 s) to 1.4 s (IQR = 1.0 -2.0 s) during PTX application (Figure 4B; Wilcoxon rank sum test, W = 3560, z = 5.213, p < 0.05). Coefficient of variation (CV) analyses of the duration also showed a significant increase from 0.3 A.U. (IQR = 0.2 -0.5 A.U.) in CCh only condition to 0.8 A.U.

[6] 176w (IQR = 0.4 -0.9 A.U.) in CCh + PTX condition (Figure 4B; Wilcoxon rank sum test, W = 3334, z = 4.195, p < 0.05). The average frequency of CCh-induced oscillation also changed significantly from 10.1 Hz (IQR = 9.5 -10.9 Hz) in absence of PTX to 9.1 Hz (IQR = 8.7 -9.7 Hz) in presence of PTX (Figure 4B; Wilcoxon rank sum test, W = 1344, z = -4.767, p < 0.05). Finally, CV analyses of frequency revealed a non-significant change from 0.30 A.U. (IQR = 0.27 -0.32 A.U.) in CCh only condition to 0.29 A.U. (IQR = 0.26 -0.31 A.U.) (Figure 4B). During GABA A receptor blockade, we also observed changes in single-unit discharge patterns (Figure 4C). We found that 45% (n = 12) of interneurons exhibited type A firing pattern, 11% (n = 3) exhibited type B firing pattern, and 44% (n = 12) exhibited type C firing pattern. In principal cells, we found 25% (n = 1) of type A and 75% (n = 3) of type C activity patterns (Figure 4D).

[7] 139w To better understand the relationship between single-unit activity and field oscillations occurring during application of CCh and PTX, we built peri-event histograms of the average spike densities centered around the onset of field oscillations with their duration normalized to 0 -100%. As illustrated in Figure 5A, only type A interneurons and principal cells significantly increased their firing but only transiently; thus, in contrast to what was seen in the CCh only condition, the increases in neuronal activity were not sustained. Notably, during GABA A receptor antagonism, the increase in the activity of type A principal cells was much greater in amplitude and occurred at an earlier time than the increase observed in type A interneuron firing (Figure 5A). Type B and type C interneurons and principal cells did not exhibit significant increases in firing activity (Figure 5B and 5C).

[8] 191w Peri-event histograms in figures 3 and 5 suggest that only the activity of type A cells presumably contributed to the generation CCh-induced oscillations. However, they did not reveal the single-unit discharge pattern during CCh-induced oscillations, which can be quantified as the phase of oscillation at which single-unit discharge. Therefore, we assessed the phase firing of type A single-unit discharges using the Hilbert transformation of the field potential. As shown in the inset of Figure 6, angles 0 o and 360 o corresponds to the peaks of the CCh-induced oscillations. We found that during application of CCh, type A interneuron firing phase distribution was non-uniformly distributed (Figure 6, Omnibus Test, m = 10887, p < 0.05) with a median of 258.4 o ; under these experimental conditions, Type A principal cell firing phase distribution also exhibited non-uniformity (Figure 6, upper right; Rayleigh's Test, z = 11.9, p < 0.05) with a median of 72.0 o . As illustrated in Figure 5B, during bath application of PTX, only type A interneuron firing phase distribution exhibited non-uniformity (Omnibus Test, m = 1175, p < 0.05) with a median firing phase of 314.4 o .

[9] 148w To investigate the role of ionotropic glutamatergic signaling in CCh-induced field oscillations, we applied CPP and NBQX while recording single-unit activity generated by interneurons and principal cells. As previously reported (Dickson and Alonso, 1997;Lévesque et al., 2017), CChinduced oscillations were abolished by this pharmacological procedure (Figure 7A). However, both interneurons and principal cells continued to fire action potentials under these experimental conditions (Figure 7B). As illustrated in figure 7C, there was a non-significant increase in firing frequency of interneurons from 1.1 Hz (IQR = 0.3 -4.4 Hz) in the CCh only condition to 1.2 Hz (IQR = 0.5 -1.9 Hz) in the CCh + CPP + NBQX condition. Similarly, the principal cell firing frequency in the CCh only condition (1.1 Hz, IQR = 0.3 -4.4 Hz) was also not significantly different when compared with the CCh + CPP + NBQX condition (1.2 Hz, IQR = 0.5 -1.9 Hz).

[10] 101w Despite the non-significant changes in firing frequency, we observed a change in the rhythmicity of firing as revealed by the autocorrelograms of action potential discharge patterns (Figure 7D). Namely, interneurons in the CCh only condition exhibited rhythmic firing at approximately 17 Hz, or a peak delay of approximately 0.06 s. In contrast, during application of CPP and NBQX, no rhythmicity in interneuron firing was observed. While principal cells did not exhibit a clear rhythmic firing pattern under both CCh only and CCh + CPP + NBQX conditions, the autocorrelograms suggested that principal cell firing was more irregular during the latter condition.

[1] 29w In this study, we performed tetrode recordings to examine single-unit activity during field oscillations that were induced by the cholinergic agent CCh in the rat EC maintained in vitro.

[2] 94w The main findings obtained from our experiments are as follows: (i) during bath application of CCh, approximately half of the recorded single-units increased their firing rate in coincidence with CCh-induced oscillations; (ii) around the onset of CCh-induced oscillations, interneuron activity increased before principal cell activity; (iii) blockade of GABA A -receptor signaling led to a transient increase in neuronal firing during the field oscillations that was not sustained to the end of CCh-induced oscillations; (iv) antagonizing ionotropic glutamatergic signaling abolished the field oscillations but did not change the firing frequency while affecting their rhythmicity.

[1] 243w Brain slice preparation, maintenance, and treatment -All procedures were conducted in compliance with the guidelines of the Canadian Council on Animal Care and the McGill Animal Care Committee. Brains were extracted from male Sprague-Dawley rats (250-275 g; n = 17, Charles River Laboratories, Saint Constant, Quebec, Canada) under isoflurane-induced anesthesia then chilled for 3 minutes in ice-cold artificial cerebrospinal fluid (ACSF) continuously bubbled with O 2 /CO 2 (95/5%) gas. The ACSF had the following composition (mM): 124 NaCl, 2 KCl, 2 CaCl 2 , 2 MgSO 4 ,1.25 KH 2 PO 4 , 26 NaHCO 3 , and 10 ᴅ-glucose. Horizontal brain slices (thickness = 450 µm, n = 32) were obtained with a Vibratome (VT1000S; Leica, Concord, Ontario, Canada) and were placed in an interface chamber between warm ACSF (31-33 o C; pH 7.4; 305 mOSM/kg) and humidified gas O 2 /CO 2 (95/5%). Brain slices were allowed to recover for 1 h before starting the continuous bath application of CCh (100 µM) at a flow rate of 2 ml/min To investigate the contributions of ionotropic GABAergic and glutamatergic signaling, additional pharmacological agents were applied after stable CCh-induced oscillations were observed. In 13 slices, picrotoxin (PTX; 100 μM; Sigma-Aldrich, Canada) was applied to block ionotropic GABA A receptors (CCh + PTX condition). In 6 slices, 3-(2-carboxypiperazin-4-yl)propyl-1-phosphonate (CPP; 10 μM; Sigma-Aldrich, Canada) and 2,3-dihydroxy-6-nitro-7sulfamoyl-benzo[f]quinoxaline-2,3-dione (NBQX; 10 μM; Sigma-Aldrich, Canada) were applied to block ionotropic glutamatergic signaling (CCh + CPP + NBQX condition).

[2] 102w Tetrode recordings -Neuronal activity was acquired with tetrode wires according to Lévesque et al. (2016). Briefly, 7 tetrodes, each made from twisting 4 tungsten wires together, were inserted into individual microdrives (NLX-18; Neuralynx, Bozeman, MT, USA). A ground wire was connected to the recording table. A channel of a tetrode was used as a reference for online data visualization. Tetrodes were slowly lowered into the EC to record 10 minutes of neuronal activity at a sampling rate of 20 kHz using the Neuroware (2.1) software from Triangle Biosystems (Durham, NC, USA). Off-line data analysis was performed with Matlab (Mathworks, Natick, MA, USA).

[3] 139w Raw data processing for single-unit activity -Single-unit analyzes were adapted from Lévesque et al. (2016). Briefly, raw data filtered between 300 -3000 Hz were analyzed with WaveClus (Quiroga et al., 2004), which considered action potentials that were 5 standard deviations (SDs) above the threshold as putative discharges from single-units and clustered action potentials according to selected sets of wavelet coefficients. Then the experimenter verified that (i) the putative single-unit clusters were distinct from the noise clusters, (ii) the action potentials had to be visible on at least 2 out of 4 channels on a tetrode (iii) the action potentials on different channels had to differ in amplitude, and (iv) less than 2% of the total number of action potentials of a single-unit cluster could occur during the refractory period (< 3 ms) (Csicsvari et al., 1998;Lévesque et al., 2016).

[4] 83w The classification of putative single-units as interneurons or principal cells were based on previously published studies (Freund and Buzsáki, 1996;Csicsvari et al., 1998;Sirota et al., 2008;Sakata and Harris, 2009;Lévesque et al., 2016). Briefly, 3 variables were calculated from the average waveform of action potentials taken from the channel that recorded the largest action potential amplitude: (i) the amplitude from the trough to the peak, (ii) the peak amplitude asymmetry, and (iii) the action potential width at 50% of amplitude (Figure 1Ba and 1Ca).

[5] 64w Average waveforms of the first and last 50 action potentials recorded were plotted together to demonstrate the stability of our recording (Figure 1Bb and 1Cb); similarly, pharmacological manipulations with GABA A and ionotropic glutamatergic receptor antagonists did not alter action potential waveform shape (Figure 1D). Therefore, a k-means clustering analysis was applied to all single-units to identify putative interneurons and principal cells (Figure 1E).

[6] 67w Raw data processing for field potential activity -Custom Matlab scripts were used to identify CCh-induced oscillations in the field potential recordings. Raw data were filtered between 1 -500 Hz and down-sampled to 2000 Hz. The 4 channels of the tetrode were averaged together and normalized against an average of all field potentials recorded from all tetrodes used in the recording session, creating the final processed field potential.

[7] 186w To identify the onset and end times of each CCh-induced oscillation, the processed field potential was high-pass filtered above 3 Hz and band-stop filtered between 55 -65 Hz. The resulting signal was enhanced by raising it to the 5 th power. The noise fluctuations of the processed field potential were estimated by calculating the moving standard deviations (MVSDs) with 1 s windows. The signal fluctuations were estimated by calculating the MVSDs with 0.0625 s window. The CCh-induced oscillation candidates were identified as signal epochs with (i) signal fluctuations greater than the 75 th percentile of the noise fluctuation and (ii) duration greater than 0.0625 s; CCh-induced oscillation candidates that were less than 0.25 s apart were merged together. The end times of CCh-induced oscillation candidates were adjusted by searching after the original end times where signal fluctuations became less than 25 th percentile of the noise fluctuations. Finally, the algorithm searched before the onset and after the end of CCh-induced oscillation candidates to identify the closest time points when the signal was 0 μV, which were respectively labeled as the onset and end of CCh-induced oscillations.

[8] 68w To quantify the frequencies of CCh-induced oscillations, discrete short-time Fourier transform with a 1 s hamming window and 80% overlap was applied to the processed field potential to calculate the power spectral densities (PSDs). The mean of frequencies between 4 -15 Hz, which is the reported CCh-induced oscillation frequency (Buzsáki, 2002), with PSD above the 95 th percentile during CCh-induced oscillations was used to estimate the oscillation frequency.

[9] 109w Classification of single-unit activity types -To classify single-unit activity, we calculated the percentage of oscillations and intervals between oscillations that contained single-unit discharges in order to evaluate the consistency of single-unit firing during recurring CChinduced oscillations. Furthermore, the difference in the proportions of spikes that occurred during CCh-induced oscillations and intervals between field oscillations was also calculated to assess single-unit firing preference. Single-units were classified in three groups: type A comprised of single-units with steady firing during CCh-induced oscillations (Figure 2A), type B comprised of single-units that also fired in between field oscillations (Figure 2B), and type C comprised of single-units with sporadic firing during CCh-induced oscillations (Figure 2C).

[1] 97w between single-unit activity and CCh-induced oscillations, single-unit discharges were extracted around CCh-induced oscillations. Each single-unit activity epoch contained segments before, during, and after the CCh-induced oscillation with the same duration as the oscillation. The durations of the single-unit activity epochs were normalized with 0% and 100% as, respectively, the onsets and ends of CCh-induced oscillations. The spike density of each single-unit was calculated by averaging the number of action potentials that occurred in each 1% bin. Average spike densities of single-units of the same neuronal type in the same pharmacological conditions were averaged to produce peri-event histograms.

[2] 97w To assess for significant relationship between single-unit activities and CCh-induced oscillations, the Monte Carlo methods were applied. For each subset of data of the same neuronal type and pharmacological condition, 10,000 single-units were randomly sampled with replacement. Their action potential timestamps were randomly shuffled to simulate random single-unit discharge patterns. Average spike densities for simulated random single-unit activities were calculated using the same procedure mentioned above. Average spike densities of the acquired data that were 2.5 SDs above the average spike densities of simulated random activities were considered to be statistically significant changes in single-unit discharge activity.

[3] 76w Single-unit phase firing during CCh-induced oscillations -The Hilbert transform of processed field potential filtered between 6 -14 Hz was used to define the phase of the CChinduced oscillations with 0 o and 360 o representing the peaks of oscillations. The probability that single-unit discharges would fire at distinct phases was plotted in polar plots to examine the phase firing of single-units during CCh-induced oscillations. The autocorrelations of single-unit discharge patterns were computed using MATLAB function xcorr.

[4] 119w Statistics -The Shapiro-Wilks Test was applied to assess the normality of the data distributions. Normally distributed data were compared using the two-sample t-tests; otherwise, data were compared using the Mann-Whitney U-Test. Throughout the text, results are expressed as median (interquartile range). Circular statistics were performed with CircStat Matlab toolbox (Berens, 2009). The Kuiper's Test was performed to compare single-unit discharge firing phase data to von Mises distributions. The Rayleigh Test for uniformity was performed on circularly normal data; otherwise, the Omnibus Test was performed. If nonuniformity was found in the data, the median of the phase data was calculated and plotted in black arrows and expressed as median throughout the text. Statistical significance was established at p < 0.05.

[5] 196w In line with previous in vitro studies perfomed in the hippocampus (Bland et al., 1988;Williams and Kauer, 1997;Konopacki et al., 2006), our findings show that CCh-induced field oscillations are associated with different single-unit activity patterns. Specifically, we found that approximately half of the principal cells and interneurons increased their activity during CChinduced oscillations. Upon closer examination, we also observed that the increases in interneuron activity preceded that in principal cells at the onset of CCh-induced oscillations. Furthermore, interneurons fired at a preferred phase of 258.4 o , which was earlier than the principal cell preferred phase of 72.0 o . Together, our data suggest that interneurons could be modulating principal cell activity during CCh-induced oscillations in the EC. Previously, Konopacki et al. (2006) have also conducted phase firing analysis and found that unclassified cells firing during CCh-induced oscillations exhibit a mean preferred firing phase of 234.6 o ± 47.29 o in the hippocampus, which is similar to what was observed here in EC interneurons. In our study, we further categorized our cells into interneurons and principal cells, and we show a presumptive role that is played by the interneurons in modulating principal cells during CChinduced oscillations.

[6] 110w Interneurons have been proposed to control the activity of principal cells and to modulate rhythmic activities of the brain (Cobb et al., 1995;Colgin, 2016). In vivo optogenetics studies have also shown that interneurons could modulate the frequencies of in vivo hippocampal theta oscillations, gamma rhythm, and ripples (Stark et al., 2013;Lasztóczi and Klausberger, 2014;Amilhon et al., 2015). In addition to physiological brain rhythms, interneurons also play a role in pathological brain oscillations, such as the initiation of epileptiform activities (Sessolo et al., 2015;Yekhlef et al., 2015;Shiri et al., 2016). Therefore, our findings support the view that interneurons can modulate the activity of principal cells during CCh-induced oscillations in the EC.

[7] 45w By antagonizing GABA A -receptor signaling with PTX, we found an increase in the duration of CCh-induced oscillations, which may be attributed to a loss of regularity in the pattern of CChinduced oscillations, as demonstrated by a significant increase in the CV after PTX application.

[8] 212w These results are in keeping with previous findings (Williams and Kauer, 1997;D'Antuono et al., 2001;Lévesque et al., 2017). Therefore, in light of a previous investigation showing that CCh could induce membrane potential oscillations in the theta frequency range in a subtype of neocortical interneurons (Blatow et al., 2003), we propose that GABA A -receptor signaling forms a network that modulates CCh-induced oscillations. When we examined single-unit activity, we found that both interneurons and principal cells only fired transiently around the onset of the field oscillations recorded during GABA A -receptor antagonism, in contrast to their sustained firing when GABA A -receptor signaling was preserved. In addition, during application of medium containing CCh and PTX, principal cells increased their firing activity before interneurons at the onset of the field oscillations. Finally, preferred firing phase analyses showed that, while interneurons still exhibited a preference in their firing at 314.4 o , principal cells did not fire at a preferred phase in absence of GABA A -receptor signaling. This evidence therefore supports the hypothesis that interneurons are involved in synchronizing the activity of principal cells and in modulating the structure and dynamics of CCh-induced oscillations. However, in the absence of GABAergic signaling, CCh-induced oscillations could still be present but with less regularity in their pattern.

[9] 86w We also found that CCh-induced oscillations were abolished by antagonizing ionotropic glutamatergic receptors. These data are consistent with what was reported in previous studies (Dickson and Alonso, 1997;Lévesque et al., 2017), and suggest that excitatory ionotropic glutamatergic receptor signaling is necessary for the generation of CCh-induced oscillations. By examining single-unit activity, we could identify a change in the rhythmicity of both interneuron and principal cell firing patterns. Together, our data suggest that ionotropic glutamatergic receptor signaling plays a crucial role in the generation of CCh-induced oscillations.

[10] 131w Several authors have emphasized that in vitro CCh-induced oscillations are similar to the atropine-sensitive in vivo theta rhythm that occur under anesthesia (Bland et al., 1988;Kowalczyk et al., 2013). These in vivo theta rhythms have been widely studied because they appear to facilitate sensorimotor integration, spatial navigation, memory, and learning (Düzel et al., 2010;Buzsáki and Moser, 2013;Hasselmo and Stern, 2014;Colgin, 2016). These physiological processes are believed to support long-term potentiation (LTP) (Bliss and Collingridge, 1993). In line with this view, in vivo hippocampal theta rhythm has been shown to facilitate LTP in anesthetized animals (Pavlides et al., 1988). Similarly, CCh-induced oscillations not only exhibit frequency ranges that overlap with in vivo theta rhythm but also have been shown to facilitate synaptic plasticity when applied to hippocampal slices (Huerta and Lisman, 1993).

[11] 90w Therefore, it could be argued that CCh-induced oscillations represent a model of in vivo theta rhythm. However, it has been reported that in vivo theta oscillations in freely moving animals can be atropine-resistant as well as that they also do not rest upon principal cell involvement as much as CCh-induced oscillations (Buzsáki et al., 1983;Williams and Kauer, 1997;Buzsáki, 2002). We have found here that principal cell activity is indeed necessary for the generation of CCh-induced oscillations in vitro, suggesting that they may not fully model the in vivo theta rhythm.

[12] 672w The mechanisms underlying CCh-induced oscillations could be related to oscillatory activities that occur during epileptiform discharges. In human depth electrode recordings, it has been documented that some seizure onset patterns are marked by rhythmic activity in the theta frequency range (Wennberg et al., 2002;Perucca et al., 2014). In the tetanus toxin animal model, spontaneous hippocampal seizures have been shown to occur more frequently during periods of prominent theta rhythm, namely the rapid eye movement sleep and exploratory wake behaviors (Sedigh-Sarvestani et al., 2014). Furthermore, recent studies in the hippocampus of the pilocarpine animal model of epilepsy have noted the appearance of oscillatory activities in the theta frequency range that preceded ictal events (Grasse et al., 2013;Fujita et al., 2014;Toyoda et al., 2015;Karunakaran et al., 2016). Upon closer examination, interneuron firing rate and coherence with field oscillations have been shown to increase during the appearance of this pre-ictal theta rhythm in the CA3 (Grasse et al., 2013;Karunakaran et al., 2016). Since several studies have found that CCh-induced oscillations are generated locally (Dickson and Alonso, 1997;Konopacki et al., 2000;Lévesque et al., 2017), we propose that the pre-ictal theta rhythm, which is observed in the hippocampus of epileptic animals, could be acting through the same network underlying CCh-induced oscillations described here. Williams and Kauer (1997) have reported that CCh-induced oscillations were longer in duration and smaller in amplitude when extracellular K + was elevated from 2.5 mM to 5 mM. These findings could be further explored using tetrode recordings to uncover the contributions of interneurons and principal cells. A: An example of raw data filtered between 300 -3000 Hz to detect action potential discharges (bars) and between 1 -500 Hz to visualize field potential activity. In this example, the single-unit is classified as a putative interneuron. B: Representative single-unit shown in A classified as a putative interneuron (Int) based on the average action potential waveform shape (a). Single-unit activity was stable throughout the recording as shown by the similarities between the average action potential waveform shape of the first 50 (solid line) and last 50 (dash line) action potentials during each tetrode recording (b). C: Representative single-unit classified as a putative principal cell (PC). D: Average waveform shape of 2 single-units that were followed under different pharmacological conditions: first under bath application of only CCh (a and b, solid line) then under bath application of CCh with the GABAergic antagonist PTX (a, dash line) or the glutamatergic antagonists CPP and NBQX (b, dash line). E: K-means clustering to classify putative single-units into interneurons (green) and principal cells (blue). A: Representative recordings of type A interneuron (a, green) and a putative principal cell (b, blue). The field potential was high-pass filtered with cutoff frequency of 3 Hz for visual clarity. A selected oscillation was expanded in the inset showing the oscillation without additional highpass filtering with black arrowheads marking the computer-identified onset and termination of the oscillation. B: Same as A for type B interneurons (a, green) and principal cells (b, blue). C: Same as A for type C interneurons (a, green) and principal cells (b, blue). D: Pie charts summarizing the proportion of type A, B, and C interneurons and principal cells identified in CCh only condition. induced field oscillations. A: A peri-event histogram of the average spike densities of type A interneuron (green) and principal cell (blue) around the onset of CCh-induced oscillations. The dashed line shows the significance threshold (2.5 SDs above the simulated random neuronal activities). B: Same as A for type B interneuron (green) and principal cell (blue) activity around the onset of CCh-induced oscillations. C: Same as A for type C interneuron (green) and principal cell (blue) activity around the onset of CCh-induced oscillations. upon the addition of PTX. B: Boxplots summarizing the distribution of the average interval of occurrence (left), average duration (center, left), coefficient of variation (CV) of duration (center), average oscillation frequency (center, right), and CV of oscillation frequency (right) of CChinduced field oscillations of the single-unit in CCh only and CCh + PTX conditions. The Shapiro-