KCC2 antagonism and gabaergic synchronization in the entorhinal cortex in the absence of ionotropic glutamatergic receptor signalling
H I G H L I G H T S
• We pharmacologically isolated GABAA receptor-mediated interictal spikes induced by 4AP in the rodent entorhinal cortex.
• Blockade of KCC2 made these interictal spikes smaller and shorter.
• KCC2 antagonism increased interneuron firing around the onset of interictal spikes.
• The recruitment of principal cells and interneurons during the interictal spike was disrupted.
Abstract
γ-Aminobutyric acid (GABA), which is released by interneurons, plays an active role in generating interictal epileptiform spikes during blockade of ionotropic glutamatergic signalling, but it remains unclear whether and how the K+-Cl− cotransporter 2 (KCC2) influences these paroxysmal events. Therefore, we employed tetrode recordings in the in vitro rat entorhinal cortex (EC) to analyze the effects of the KCC2 antagonist VU0463271 on 4-aminopyridine (4AP)-induced interictal spikes that were pharmacologically isolated by applying ionotropic glutamatergic receptor antagonists. After the addition of VU0463271, these interictal spikes continued to occur at similar rates as in control (i.e., during application of 4AP with ionotropic glutamatergic receptor antagonists) but were smaller and shorter. Despite the absence of ionotropic glutamatergic receptor signalling, both inter- neurons and principal cells increased their firing during interictal spikes. Moreover, we found that KCC2 an- tagonism increased interneuron firing but decreased principal cell firing during the interictal spike rising phase; in contrast, during the falling phase, interneuron firing decreased in the presence of VU0463271 while no change was observed in principal cell firing. Overall, our results show that KCC2 antagonism enhances interneuron excitability at the onset of interictal spikes generated by the EC neuronal networks during blockade of ionotropic glutamatergic transmission but disrupts later neuronal recruitment.
1. Introduction
While focal seizures are the hallmark of an epileptic condition, in- terictal epileptiform discharges (hereafter termed interictal spikes) are valuable to clinicians. For instance, they help clinicians to diagnose focal epilepsy and to identify epilepsy subtypes (Fisher et al., 2017; Tatum et al., 2018). Moreover, for epilepsy surgery, they have being used as markers of seizure onset zones (Jacobs et al., 2011; Tatum et al., 2018) and predictors of post-operative outcome (Coutin-Churchman et al., 2012; Dworetzky and Reinsberger, 2011; Mehvari Habibabadi et al., 2019; Rosati et al., 2003; Tatum et al., 2018). Therefore, iden- tifying the fundamental mechanisms underlying interictal spikes in experimental epilepsy models should lead to improved epilepsy diag- nosis and treatment in humans. To this end, basic science researchers have reproduced electrographic interictal spikes in the laboratory by applying drugs such as the K+ channel blocker 4-aminopyridine (4AP) to in vitro brain preparations (Avoli and de Curtis, 2011).
These studies have revealed excessive firing of interneurons, which release γ-aminobutyric acid (GABA), in coincidence with 4AP-induced interictal spikes thus pointing to the involvement of inhibitory me- chanisms in their generation (Avoli et al., 1996; González et al., 2018; Lévesque et al., 2016; Librizzi et al., 2017). In addition, these interictal spikes continue to occur in the absence of ionotropic glutamatergic receptor signalling, and they could be abolished upon blockade of GABAA signalling (Avoli et al., 2013, 1996; Hamidi and Avoli, 2015; Lévesque et al., 2016; Panuccio et al., 2010; Sudbury and Avoli, 2007). Hence, this evidence suggests that interneurons firing and subsequent GABAA signalling contribute to interictal spike generation, at least in in vitro models of epileptiform synchronization.
An important consequence of GABA released from interneurons is the activation of GABAA receptors and the resulting Cl− influx into neurons that causes hyperpolarization of the neuronal membrane po- tential (Avoli and de Curtis, 2011; Di Cristo et al., 2018; Farrant and Kaila, 2007; Kaila et al., 2014). To ensure GABAA receptor-mediated Cl− influx, neurons in the mature brain maintain a low intracellular [Cl−] using the K+-Cl− cotransporter 2 (KCC2); thus, in order to ex- trude Cl− from the neuron, KCC2 uses the [K+] gradient (Viitanen et al., 2010). Accordingly, excessive interneuron activity occurring at the onset of 4AP-induced ictal activity causes substantial elevations in extracellular [K+], presumably, through KCC2 activity (Avoli et al., 1996; González et al., 2018; Librizzi et al., 2017). Taken together, KCC2 activity and the subsequent elevations in extracellular [K+] may have important implications in epileptiform synchronization (Di Cristo et al., 2018; Kaila et al., 2014).
To study the role of KCC2 in epileptic disorders, KCC2 antagonists (Delpire et al., 2009, 2012; Lebon et al., 2012; Delpire and Weaver, 2016) have been used in models of epileptiform synchronization. These experiments have established that antagonizing KCC2 enhances neu- ronal excitability by depolarizing the reversal potential of GABAA sig- nalling (Chen et al., 2019; Moore et al., 2018; Sivakumaran et al., 2015). In addition, in many cases, KCC2 antagonism blocked ictal ac- tivity while enhancing interictal spike generation (Chen et al., 2019; Hamidi and Avoli, 2015; Kelley et al., 2016; Moore et al., 2018; Sivakumaran et al., 2015). However, while previous studies have shown that GABAA signalling generates interictal spikes in the in vitro 4AP model of epileptiform synchronization (see for review Curtis and Avoli, 2016), the role of KCC2 antagonism during interictal spikes in the presence of ionotropic glutamatergic receptor antagonists was in- vestigated in only one study using VU0240551 (Hamidi and Avoli, 2015), a less potent and selective predecessor of VU0463271 (Delpire et al., 2012). Therefore, we employed here tetrode wire recordings to understand whether and how the KCC2 antagonist VU0463271 modi- fies the activity of presumptive interneurons and principal cells in the rat entorhinal cortex (EC) during interictal spikes occurring in the presence of 4AP and ionotropic glutamatergic receptor antagonists in an in vitro brain slice preparation.
2. Methods
Ethical approval. All procedures complied with the guidelines of the Canadian Council on Animal Care and were approved by the McGill University Animal Care Committee. All efforts were made to minimize the suffering and the number of animals used in the experiments.
Specimen preparation and maintenance. Brains of male Sprague- Dawley rats (250–275 g; Charles River Laboratories, Saint Constant, QC, Canada) – anaesthetized using 5% isoflurane (Fresenius Kabi Canada Limited, Toronto, ON, Canada) – were extracted and placed in ice cold artificial cerebrospinal fluid (ACSF; 124 mM NaCl, 2 mM KCl, 2 mM CaCl2, 2 mM MgSO4, 1.25 mM KH2PO4, 26 mM NaHCO3, and 10 mM ᴅ-glucose; Sigma-Aldrich, Oakville, ON, Canada) that was oxygenated with O2/CO2 (95/5%) gas mixture to maintain a pH of 7.4. Brains were sliced with a vibratome (VT1000S; Leica, Concord, ON, Canada) to obtain slices containing hippocampus and EC (thickness = 450 μm) and then transferred to an interface chamber to be maintained between warm ACSF (32 ± 1 °C) and humidified O2/ CO2 (95%/5%) gas mixture. Following approximately 1-h recovery period, epileptiform activity was induced by continuous bath applica- tion of ACSF containing 4AP (50 μM; Sigma-Aldrich, Oakville, ON, Canada) at a flow rate of approximately 2 ml/min (see also Chen et al., 2019). After the occurrence of baseline activity, the NMDA receptor antagonist (RS)-3-(2-carboxypiperazin-4-yl)-propyl-1-phosphonate (CPP; 10 μM; Tocris, Canada) and the AMPA receptor antagonist 6- Cyano-7-nitroquinoxaline-2,3-dione disodium salt (CNQX; 10 μM; To- cris, Oakville, ON, Canada) were added to ACSF containing 4AP (4AP + CPP + CNQX) to isolate GABAergic synchronous events. To establish the role of KCC2 in these ionotropic glutamatergic signalling- independent interictal spikes, N-Cyclo-propyl-N-(4-methyl-2-thiazolyl)- 2-[(6-phenyl-3-pyridazinyl)thio]acetamide (10 μM; VU0463271; To- cris, Oakville, ON, Canada; Delpire et al., 2012) was then added (4AP + CPP + CNQX + VU0463271).
Tetrode recordings. Collection and analysis of single-unit activity were based on previous studies from our laboratory (Chen et al., 2019, 2018; Lévesque et al., 2018, 2016). Four tungsten wires (tip dia- meter = 50 μm, California Fine Wire Company, Grover Beach, CA, USA) were twisted together to make a tetrode wire; 7 tetrode wires were inserted into a microdrive (NLX-18; Neuralynx, Bozeman, MT, USA). A ground wire was connected from the drive (EIB-36-18 Drive; Neuralynx, Bozeman, MT, USA) mounted on the microdrive to the vi- bration isolation table (Newport; Irvine, CA, USA). A channel of one tetrode was used as the reference. All tetrodes were placed in the EC to record neuronal activity at a sampling rate of 20 kHz using the Neu- roware system (2.1; Triangle Biosystems, Durham, NC, USA). Due to limitations in computer performance, neuronal activity was sampled in 10-min epochs. We performed repeated sampling to minimize the number of animals used in our study. Off-line data analysis was per- formed with Matlab (Mathworks, Natick, MA, USA; RRID:SCR_001622). Isolation of single-unit activity. Raw tetrode recordings, which were filtered between 300 and 3000 Hz (Fig. 1A), were analysed with WaveClus (Quiroga et al., 2004), an unsupervised cluster cutting al- gorithm. Peaks in the filtered signal that were 5 standard deviations (SDs) above the baseline were presumed to be action potentials and sorted based on sets of wavelet coefficients (Quiroga et al., 2004). For each identified cluster of action potentials, the experimenter visually verified that: (i) it was distinct from the noise, (ii) action potentials were visible on at least two channels of a tetrode wire, (iii) action po- tentials were of different amplitudes between channels of the tetrode, and (iv) less than 2% of the total number of action potentials occurred during the refractory period (< 3 ms) (Csicsvari et al., 1998). Verified single-unit clusters were characterised using three variables based on previously published guidelines (Chen et al., 2019; Csicsvari et al., 1998; Freund and Buzsáki, 1996; Lévesque et al., 2016; Sakata and Harris, 2009; Sirota et al., 2008): first, the amplitude from the trough to the peak; second, the asymmetry between the amplitudes of the peaks; third, the action potential width at 50% of amplitude (Fig. 1C). To cluster presumptive single units, we fitted Gaussian mix- ture models of various orders then selected the best model using Bayesian information criterion (Fig. 1D). Isolation of field potential activity. To visualize the field poten- tial activity generated by the EC, raw tetrode recordings were filtered between 1 and 500 Hz and down-sampled to 2000 Hz. Then, the four channels of the tetrodes were averaged together and normalized against the average of all signals from all tetrodes used during the recording session. Using this final processed field activity (Fig. 1A), an experi- menter manually detected interictal spikes, of which the onset was defined as the first deflection from baseline and the termination as the return to baseline activity. The duration of interictal spikes was defined as the difference between the onset and termination times. The am- plitude of interictal spikes was defined as the difference between the maximum and minimum values during each interictal spike. Patterns of single-unit activity around the onset of interictal spikes. We used raster plots, that are summarised in real-time peri- event histograms, and normalized peri-event histograms to characterize the changes in single-unit activity around the onset of interictal spikes. The raster plots were built by extracting single-unit action potentials from 1 s before to 2 s after the onset, time 0 s, of the interictal spikes. The action potentials in the raster plots were summed in 25 ms bins and normalized to the total sum to generate real-time peri-event histograms. Since interictal spikes differed in durations, normalized peri-event histograms were generated by extracting epochs of single-unit activity around interictal spikes. These single-unit activity epochs included pre- and post-event baselines with duration same to that of interictal spikes. The duration of the entire single-unit activity epoch was normalized with 0% and 100% representing the onset and termination of interictal spikes, respectively. Action potentials generated by the same single-unit type, interneurons or principal cells, in the same pharmacological condition, 4AP + CPP + CNQX or 4AP + CPP + CNQX + VU0463271, were summed in 1% bin and averaged to produce the averaged action potential densities in the normalized peri-event histograms. For visual representation, the aver- aged action potential densities were smoothed using Matlab function movmean with bin width of 11 data points. Fig. 1. Single-unit classification. A: Tetrode wire recording of single-unit and field potential activity in the rat EC. B: (a) Averaged waveform shapes of the first (solid) and last (dashed) 50 action potentials of a single unit recorded for 10 min. (b) Averaged waveform shape of the action potentials under 4AP + CPP + CNQX (solid line, n = 96) and after the addition of VU0463271 (dashed line, n = 1645). C: The waveform shape of action potentials was quantified using three variables: trough to peak, peak amplitude asymmetry, and halfwidth duration. D: Clusters of presumptive interneurons (Int, green) and principal cells (PC, blue) under 4AP + CPP + CNQX (circles) and 4AP + CPP + CNQX + VU0463271 (triangles). The centroids of the clusters are marked with x. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.) Quantification of single-unit activity around the peak of in- terictal spikes. We calculated the proportions of action potentials that occurred during the rising and the falling phases for all interictal spikes recorded for all single units. Specifically, the rising phase of interictal spikes was defined as the period between the onset and peak of inter- ictal spikes whereas the falling phase was defined as the period between the peak and end of interictal spikes (Fig. 4A). The number of action potentials that occurred in the rising and falling phases were normal- ized against the total number of action potentials generated by the single units during the interictal spikes to calculate the proportion of action potentials in the rising and falling phases, respectively. We then compared the total of the proportions of action potentials generated by the same type of single unit, either interneurons or principal cells, in the same phase, either rising or falling phases, across pharmacological conditions, between 4AP + CPP + CNQX and 4AP + CPP + CNQX + VU0463271. Fig. 2. Field potential activity recorded from the EC in the presence and absence of VU0463271. A: Interictal spikes persisted in the field poten- tial recordings upon the addition of VU0463271 to 4AP + CPP + CNQX condition. B: Quantification of the interval of occurrence, duration, and amplitude of interictal spikes re- corded under 4AP + CPP + CNQX and 4AP + CPP + CNQX + VU0463271 conditions. Wilcoxon Rank-Sum Test was used to estab- lished statistical significance, *p < 0.05. Statistical analyses. For statistical comparisons between pharma- cological conditions, we first assessed data normality using the Shapiro- Wilks Test and found that data were not normally distributed. Therefore, we expressed the data central tendencies as median (inter- quartile range) and used the Wilcoxon Rank-Sum Test for hypothesis testing. Since Wilcoxon Rank-Sum Test compares sums of rank, we also reported the mean ranks (R) of our data. We performed Monte Carlo analysis to assess for statistical sig- nificance in the normalized peri-event histograms. For each extracted single-unit activity epochs, action potential timestamps were randomly shuffled 5000 times to simulate random single-unit activity patterns. Using the simulated single-unit activity epochs, average action poten- tial densities of the normalized peri-event histograms were considered to be significant when they were 2.50 SDs above the average action potential densities of simulated activity. 3. Results 3.1. Single-unit characterisation In the EC of 25 brain slices obtained from 12 animals, we performed tetrode recordings to study the relationship between field activity and action potentials generated by single units during interictal spikes generated during application of 4AP and ionotropic glutamatergic re- ceptor antagonists (Fig. 1A). We recorded a total of 187 single units for which the averaged action potential waveforms of the first and last 50 action potentials appeared to be similar (Fig. 1Ba). Additionally, the shapes of the action potential waveform in the presence and absence of VU0463271 appeared to be similar (Fig. 1Bb). As in previous studies from our laboratory (Chen et al., 2019, 2018; Lévesque et al., 2018, 2016), we used well-established characterisations of action potential waveform (Csicsvari et al., 1998; Freund and Buzsáki, 1996; Sakata and Harris, 2009; Sirota et al., 2008) to cluster the recorded single units into presumptive interneurons (4AP + CPP + CNQX: n = 80;4AP + CPP + CNQX + VU0463271: n = 85) or principal cells (4AP + CPP + CNQX: n = 14; 4AP + CPP + CNQX + VU0463271: n = 8) (Fig. 1D). 3.2. Alterations of interictal spikes by antagonising KCC2 activity As previously reported (Avoli et al., 2013, 1996; Hamidi and Avoli, 2015; Lévesque et al., 2016), interictal spikes in the EC continued to occur during application of the ionotropic glutamatergic antagonists CPP and CNQX (n = 2480 spikes, n = 21 slices, n = 12 animals; Fig. 2A). The interictal spikes had a median interval of occurrence of 20.17 s (10.72–30.28 s), a median duration of 0.97 s (0.67–1.19 s), and a median amplitude of 22.97 μV (13.79–52.49 μV). Interictal spikes continued to occur at a similar interval of occurrence (20.65 s, 11.59–28.20 s) following the application of the KCC2 antagonist VU0463271 (n = 2560 spikes; n = 13 slices, n = 7 animals; Fig. 2A and B); however, their median duration (0.86 s, 0.59–1.07 s) along with their median amplitude (13.35 μV, 9.33–18.97 μV) were significantly decreased (duration: R4AP+CPP+CNQX = 2.74 × 103, R4AP+CPP+CNQX +VU0463271 = 2.31 × 103, p = 1.64 × 10−25, W = 5.91 × 106, z = −1.04; amplitude: R4AP+CPP+CNQX = 3.12 × 103, R4AP+CPP+CNQX +VU0463271 = 1.94 × 103, p = 4.26 × 10−184, W = 4.96 × 106, z = −2.89) (Fig. 2B). 3.3. Effects of VU0463271 on single-unit activity pattern around the onset of interictal spikes Next, we characterised the pattern of single-unit activity around the onset of interictal spikes generated during application of 4AP and io- notropic glutamatergic receptor antagonists with and without VU0463271 in the medium. Examples of interictal spikes are illustrated in Fig. 3Aa. As illustrated in Fig. 3Ab, both interneurons and principal cells started firing around the onset of interictal spikes in both absence and presence of VU0463271 (4AP + CPP + CNQX: n = 1588 interictal spikes for interneurons, n = 185 interictal spikes for principal cells; 4AP + CPP + CNQX + VU0463271: n = 1764 interictal spikes for interneurons, n = 134 interictal spikes for principal cells). When we examined the action potential density in the real-time peri-event his- tograms, we noted that the peak of interneuron action potential density was higher in the 4AP + CPP + CNQX + VU0463271 condition than in the 4AP + CPP + CNQX condition (Fig. 3Ac). Moreover, in the normalized peri-event histograms, we found that single-unit action potential densities of both interneurons and principal cells significantly increased after the onset of interictal spikes under both the 4AP + CPP + CNQX and the 4AP + CPP + CNQX + VU0463271 conditions (Fig. 3B). Moreover, by performing this type of analysis, we noted that the peak of interneuron action potential density in the 4AP + CPP + CNQX + VU0463271 condition was higher than that in the 4AP + CPP + CNQX condition. Fig. 3. Single-unit activity around the onset of interictal spikes. (A) Single-unit activity around the onset of interictal spikes in the 4AP + CPP + CNQX and 4AP + CPP + CNQX + VU0463271 conditions. Interictal spikes (a) along with corresponding raster plots (b) and real-time peri-event histograms (25 ms bins) (c) are illustrated. In the real-time peri-event histogram, peaks of interneuron and principal cell action potential densities in the 4AP + CPP + CNQX condition were both identified at 0.06 s after the onset of interictal spikes. In the 4AP + CPP + CNQX + VU0463271 condition, peaks of interneuron and principal cell action potential densities were identified at 0.06 s and 0.11 s, respectively, after the onset of interictal spikes. (B) Normalized peri-event histograms with time 0 representing the onset of interictal spikes. Note that in the 4AP + CPP + CNQX condition, the peak of interneuron action potential density during interictal spikes (n = 1588) occurred at 8.50% of the normalized duration while the peak of principal cell action potential density during interictal spikes (n = 185) occurred at 9.50%. In the 4AP + CPP + CNQX + VU0463271 condition, the peak of interneuron action potential density during interictal spikes (n = 1764) occurred at 8.50% whereas the peak of principal cell action potential density during interictal spikes (n = 139) occurred at 15.50%. Dashed lines indicate significance thresholds. 3.4. Effects of VU0463271 on the occurrence of single-unit activity during interictal spikes As shown in the real-time and normalized peri-event histograms in Fig. 3, interneuron action potentials after the onset of interictal spikes were more concentrated after the addition of VU0463271, suggesting that the pattern of interneuron activity was altered upon the antag- onism of KCC2. Therefore, we adapted the methods outlined by Spoljaric et al. (2019) to quantify single-unit activity around the peak of interictal spikes recorded with and without VU0463271 in the medium. Specifically, we calculated the proportion of single-unit action poten- tials that occurred in the rising and falling phase of interictal spikes (Fig. 4A). As shown in Fig. 4B, the proportion of interneuron action potentials in the rising phase was significantly higher in the presence than in the absence of VU0463271 (4AP + CPP + CNQX: 0.00%, 0.00–50.00%; 4AP + CPP + CNQX + VU0463271: 16.67%, 0.00–66.67%; R4AP+CPP+CNQX = 2.01 × 103, R4AP+CPP+CNQX +VU0463271 = 2.30 × 103, p = 8.54 × 10−17, W = 4.17 × 106, z = −8.32). As for principal cells, we observed that the proportion of action potentials in the rising phase was significantly lower after the addition of VU0463271 than before (4AP + CPP + CNQX: 0.00%, 0.00–38.46%; 4AP + CPP + CNQX + VU0463271: 0.00%,0.00–20.00%; R4AP+CPP+CNQX = 2.84 × 102, R4AP+CPP+CNQX +VU0463271 = 2.40 × 102, p = 2.71 × 10−4, W = 5.38 × 104, z = −3.64) (Fig. 4B). Fig. 4. Quantification of single-unit activity during the rising and falling phases of interictal spikes. (A) Examples of interictal spikes recorded in the 4AP + CPP + CNQX and 4AP + CPP + CNQX + VU0463271 conditions. The grey box outlines the rising phase of the interictal spike while the white box outlines the falling phase of the interictal spike. (B) Quantification of the pro- portion of interneuron and principal cell action potentials in the rising and falling phases across pharmacological conditions. Wilcoxon Rank-Sum Test was used to established statistical significance, *p < 0.05. As shown in Fig. 4B, during the falling phase of the interictal spikes, the proportion of interneuron action potentials was significantly lower after the addition of VU0463271 than before (4AP + CPP + CNQX: 33.33%, 0.00–66.67%; 4AP + CPP + CNQX + VU0463271: 25.00%, 0.00–60.00%; R4AP+CPP+CNQX = 2.23 × 103, R4AP+CPP+CNQX +VU0463271 = 2.11 × 103, p = 6.62 × 10−4, W = 4.62 × 106, z = 3.40). In contrast, the proportion of principal cell action potentials occurring during the falling phase of the interictal spikes was not sig- nificantly different between the two pharmacological conditions (4AP + CPP + CNQX: 20.94%, 0.00–57.14%; 4AP + CPP + CNQX + VU0463271: 10.00%, 0.00–66.67%) (Fig. 4B). 4. Discussion The main findings of our study can be summarised as follows: 1) we confirmed that KCC2 antagonism does not alter the interval of occur- rence of interictal spikes generated during application of 4AP and io- notropic glutamatergic receptor antagonists but decreased their am- plitude and duration; 2) despite the changes in field activity, interictal spikes were still associated to significant increases in interneuron and principal cell action potential densities under KCC2 antagonism; 3) KCC2 antagonism led to an increase in the proportion of interneuron action potentials in the rising phase concomitant to a decrease in the proportion of principal cell action potentials; and 4) during the falling phase, the proportion of interneuron action potentials decreased when VU0463271 was present in the medium while no change was observed in the proportion of principal cell action potentials. 4.1. KCC2 antagonism alters the dynamics of interictal spikes In line with findings reported by Hamidi and Avoli (2015), we found that the application of VU0463271 to medium containing 4AP and io- notropic glutamatergic receptor antagonists does not alter the interval of occurrence of interictal spikes. The similarity between intervals of occurrence in our pharmacological conditions suggests that inter- neurons can periodically and synchronously fire action potentials in the absence of ionotropic excitatory synaptic transmission, which was ori- ginally reported by Perreault and Avoli (1992, 1991), as well as during reduced KCC2 function. Moreover, our data suggest that these periodic and synchronous interneuron action potentials are capable of gen- erating postsynaptic responses since they were mirrored by field po- tential events. Interestingly, the increase in firing associated to inter- ictal spikes in absence of ionotropic glutamatergic signalling could also be identified in principal cells. These increases in firing could result from elevations in extracellular [K+] that accompany “excessive” ac- tivation of postsynaptic GABAA receptors (Avoli et al., 1996; Morris et al., 1996; Smirnov et al., 1999; Viitanen et al., 2010) and HCO −- mediated depolarizing currents (Lamsa and Kaila, 1997). In addition, depolarizing GABAA signalling has been reported in hippocampal (Michelson and Wong, 1991; Perreault and Avoli, 1992, 1991) and EC neurons (Lopantsev and Avoli, 1998). Finally, action potential firing may result from gap junction coupling (Fujiwara-Tsukamoto et al., 2010; Gigout et al., 2006; Yang and Michelson, 2001). Confirming previous data from our laboratory (Hamidi and Avoli,2015), we found that the duration and amplitude of ionotropic gluta- matergic signalling-independent 4AP-induced interictal spikes became lower in the presence of VU0463271. In light of previous experiments showing that extracellular [K+] can influence the duration and the amplitude of synchronous field potentials such as epileptiform interictal spikes (Chizhov et al., 2015; Hablitz and Lundervold, 1981; Ogata et al., 1976), we interpret such changes in duration and amplitude as the result of decreased elevations in extracellular [K+] that are known to accompany 4AP-induced interictal spikes (Avoli et al., 1996) and that should become smaller than in control during pharmacological KCC2 antagonist (Viitanen et al., 2010). Therefore, our findings are in agreement with previous studies that highlighted the role of transient elevations in extracellular [K+] in the generation of field interictal spikes (Avoli et al., 1996; Morris et al., 1996; Smirnov et al., 1999).
4.2. KCC2 antagonism alters the pattern of single-unit activity during interictal spikes
Using tetrodes, we found that interneuron and principal cell averaged action potential densities significantly increased around the onset of interictal spikes generated in the absence of ionotropic gluta- matergic receptor signalling, showing that GABAA signalling can recruit single units through the mechanisms discussed in the previous section (e.g. depolarizing GABAA signalling, gap junctions, etc.). Upon closer examination, we found changes in the proportions of single-unit action potentials during interictal spikes recorded under KCC2 antagonism, suggesting changes to the underlying interplay between single units. Namely, we found that the proportion of interneuron action potentials in the rising phase was higher in the condition of reduced KCC2 func- tion than in the condition of intact KCC2 function; this increase in the proportion of interneuron action potentials in the rising phase may mirror an increase in interneuron excitability, supporting our previous demonstrations of neuronal hyperexcitability during KCC2 blockade in the in vitro 4AP model (Chen et al., 2019). Furthermore, it has been demonstrated in cultured hippocampal neurons that downregulating KCC2 expression increases neuronal excitability by inducing a depo- larizing shift in the GABAA reversal potential (Kelley et al., 2018; Moore et al., 2018; Sivakumaran et al., 2015). Supporting the studies on cul- tured hippocampal neurons, both in vitro cell cultures expressing mu- tated SLC12A5 genes and the hippocampal formation of brain slices taken from pilocarpine-treated animal showed reduced KCC2 cell sur- face expression with depolarized GABAA reversal potential (Pathak et al., 2007; Puskarjov et al., 2014). Together, these studies indicate that KCC2 downregulation can promote neuronal hyperexcitability by depolarizing the GABAA reversal potential.
Despite the known neuronal hyperexcitability induced by KCC2 antagonism, we found significant decreases in the proportion of prin- cipal cell action potentials in the rising phase – in contrast with con- ditions of intact ionotropic glutamatergic receptor signalling (Chen et al., 2019) – and in the proportion of interneuron action potentials in the falling phase. Since interictal spikes were recorded in the absence of ionotropic glutamatergic signalling, the proportion of principal cell action potentials throughout interictal spikes and the proportion of interneuron action potentials in the falling phase likely reflects a dis- ruption in single-unit recruitment during interictal spikes. Our results are therefore in line with the view that increases in extracellular [K+] promotes neuronal recruitment (Avoli and de Curtis, 2011; Viitanen et al., 2010). These findings in single-unit activity also support the observations in the field potential activity – namely, the significant decreases in the duration and the amplitude of interictal spikes that could be attributed to extracellular [K+]. Taken together, our data emphasize the important role played by extracellular [K+] in the dy- namics of interictal spikes.
5. Conclusions
Our findings highlight the interaction between KCC2 antagonism and GABAA signalling during interictal spikes generated by EC neuronal networks in vitro during blockade of ionotropic glutamatergic trans- mission. Specifically, our data show that KCC2 antagonism is not only capable of inducing neuronal hyperexcitability but that it can also disrupt neuronal recruitment during interictal spikes. The dual influ- ence of KCC2 antagonism on neuronal activity altogether could lead to smaller and shorter spikes suggesting that pathological neuronal ac- tivity depends on neuronal excitability as well as synchrony.