KCC2 Chloride Transport Contributes to the Termination of Ictal Epileptiform Activity

Visual Abstract


Introduction
Recurrent seizures often respond poorly to first line medications that target the inhibitory chloride-permeable GABA A receptor (GABA A -R; Painter et al., 1999;Fung et al., 2019). Low intracellular chloride concentration ([Cl -] i ) is an important determinant of inhibitory postsynaptic GABA A -R signaling. This low [Cl -] i is established by a Gibbs-Donnan system comprised of impermeant cytoplasmic and extracellular anions, and a membrane permeability for chloride salts and water provided by equilibrative cation-chloride cotransporters (CCCs; Delpire and Staley, 2014;Glykys et al., 2014). This type of system can be characterized by the reversal potential for membrane chloride currents (E Cl ) when the system is at equilibrium. GABA A -R-gated chloride currents alter the local [Cl -] i and move this system away from equilibrium. Restoration of equilibrium after synaptic signaling is achieved by the flux of chloride salts through the CCCs, such as the K 1 -Clcotransporter KCC2 (transporting 1 K 1 and 1 Clion per cycle) and Na 1 -K 1 -2Clcotransporter 1 (NKCC1; transporting 1 Na 1 , 1 K 1 , and 2 Clions per cycle; Gamba, 2005). Because GABA A -gated chloride currents are "downhill," that is, the chloride diffuses along its electrochemical gradient, the restoration of the baseline [Cl -] i requires energy; this is applied via cation cotransport. For KCC2, forward transport (canonical Clextrusion) is driven by downhill potassium extrusion that increases extracellular potassium ([K 1 ] 0 ). For NKCC1, forward transport (canonical Climport) is driven by downhill sodium import that increases intracellular sodium. The restoration of K 1 and Na 1 gradients ultimately requires energy in the form of ATP as Na 1 -K 1 -ATPase-mediated cation transport that restores the Na 1 and K 1 gradients.
High rates of synaptic Clinflux, as occurs during seizures, stresses neuronal Clhomeostasis (Staley and Proctor, 1999). To restore [Cl -] i to baseline, high rates of CCC-mediated Clefflux are required. But at these high cotransport rates, K 1 efflux is also increased, which may elevate [K 1 ] 0 (Viitanen et al., 2010). This may change the equilibrium conditions for CCC transport, leading to higher [Cl -] i . In turn, this may lead to proconvulsant depolarizing GABA conductance, while the elevated [K 1 ] 0 may depolarize neurons directly, predisposing to further seizures (González et al., 2015). Given these secondary effects of CCCs, it may not be surprising that there are conflicting results regarding the role of CCCs in seizures. When in vitro epileptiform activity was induced with 4-AP, blocking KCC2 transport activity with either VU0240551 or high doses of bumetanide abolished ictal-like epileptiform discharges (ILDs), while enhancing KCC2 activity with a high concentration of CLP257 increased the duration of ILDs (Hamidi and Avoli, 2015). On the other hand, there are both genetic and pharmacological data supporting an anticonvulsant role of KCC2 expression in chronic temporal lobe epilepsy. For example, selective inhibition of KCC2 with VU0463271 led to hyperexcitability and epileptiform discharges in hippocampal slices in vitro exposed to low magnesium, and induced seizures in vivo (Sivakumaran et al., 2015). Human loss of function mutations of KCC2 result in febrile seizures or more severe early infantile epileptic encephalopathies (Kahle et al., 2016;Duy et al., 2019). Experimental studies of gain of function mutations in KCC2 have reported resistance to 4-AP induced ILDs, lower baseline [Cl -] i , and higher Clextrusion rates after exposure to glutamate (Moore et al., 2018). However, since glutamate massively increases [K 1 ] 0 (Vargova et al., 2001) creating a proportionate increase in equilibrium [Cl -] i , increased KCC2 activity would only alter Clextrusion in recorded cells that were not in areas of high [K 1 ] 0 ; how this relates to ictal conditions (Heinemann et al., 1977) is therefore unclear (Lux and Heinemann, 1978).
Here, we studied KCC2 transport activity during recurrent ILDs in an in vitro model of epileptogenesis that does not require exogenous convulsant conditions (Berdichevsky et al., 2012) and in which both [Cl -] i and electrographic epileptiform activity could be monitored. We describe the effects of ILDs on [Cl -] i . We measured the effects of the low and high affinity KCC2 inhibitors furosemide and VU463271 (Delpire et al., 2012), as well as the putative KCC2 activator CLP257 (Gagnon et al., 2013;Cardarelli et al., 2017) on epileptiform activity as well as neuronal chloride elevation and extrusion rates during spontaneous ILDs.

Materials and Methods
All animal-use protocols were in accordance with the guidelines of the National Institutes of Health and the Massachusetts General Hospital Center for Comparative Medicine on the use of laboratory animals. All protocols were approved by the Subcommittee on Research and Animal Care (SRAC).

Culture of organotypic hippocampal slices and experimental conditions
Transverse 350-mm hippocampal slices were prepared from C57BL/6 and CLM1 (Duke University Medical Center, Durham, NC) mice at postnatal day (P)6-P7 as previously described (Dyhrfjeld-Johnsen et al., 2010;Berdichevsky et al., 2012). Acute slices were mounted on poly-L-Lysine coated glass coverslips (Electron Microscopy Sciences). Slices were incubated in 1000 ml of NeuroBasal/B27(1Â) medium (Invitrogen by Life Technologies) supplemented with 0.5 mM GlutaMAX and 30 mg/ml gentamicin (all from Invitrogen) in six-well plates with low-evaporation lid (Becton Dickinson Labware), in a humidified 37°C atmosphere that contained 5% CO 2 , placed on a rocking platform (less than one cycle per minute). Culture medium was changed bi-weekly. For acute recordings and imaging, slices were transferred to a submerged chamber and continuously superfused in oxygenated (95% O 2 and 5% CO 2 ) artificial CSF (ACSF) containing the following: 126 mM NaCl, 3.5 mM KCl, 2 mM CaCl 2, 1.3 mM MgCl 2 , 25 mM NaHCO 3 , 1.2 mM NaHPO 4 , and 11 mM glucose (pH 7.4) at 32 6 0.5°C and a flow rate of 2 ml/min. All organotypic hippocampal slices were used at days in vitro (DIV)1-DIV28.

Electrophysiological recordings and data analysis
Extracellular field potentials were recorded in the CA3 and CA1 pyramidal cell layer of organotypic hippocampal slices using custom-made tungsten-coated 50-mm wire microelectrodes. The electrical signals were digitized using an analog-to-digital converter DigiData 1322A (Molecular Devices, Inc). AxoScope 10.7 and Clampfit 10.7 (Molecular Devices), Origin 2018 (OriginLab Corporation) and SigmaPlot 11.0 (Systat Software, Inc) programs were used for data acquisition and analyses. Recordings were sampled at 10 kHz. Interictal epileptiform discharges (IEDs) were defined as synchronous network-driven bursts characterized by short (0.1-3 s) duration and large amplitude population spikes. The frequency, duration and amplitude of IEDs substantially varied between recurrent ILDs. ILDs were defined as hyper-synchronous, large-amplitude and high-frequency population spikes followed by sustained ictal-tonic and/or intermittent ictal-clonic after-discharges, with the duration of the population spikes and after-discharge complex lasting .5 s. Power spectrum analysis was performed on the electrical recordings after filtering with a Bessel high pass filter of 1 Hz and applying a Hamming window function. The power of the electrical activity was calculated by integrating the root mean square value of the signal amplitude in corresponding time windows and frequency range from 1 to 1000 Hz. For comparison between slices, power was normalized for each slice with the highest value in control conditions.

Two-photon imaging of Clomeleon, quantitative and morphologic analysis
Neuronal chloride concentration was determined in CA1 pyramidal neurons expressing the ratio-metric chloride indicator Clomeleon (Kuner and Augustine, 2000). High-resolution two-photon excitation laser scanning imaging of the Cl --sensitive yellow fluorescent protein (YFP) and the Cl --insensitive cyan fluorescent protein (CFP) was performed on an Olympus Fluoview 1000 MPE microscope. A mode-locked titanium-sapphire laser (MaiTai, Spectra Physics) with 860-nm two-photon excitation was used to generate fluorescence. Emitted light passed through a dichroic mirror and was bandpass filtered through 480 6 15 nm (D480/30) for CFP and a 535 6 20-nm filter (D535/40) for YFP (FV10-MRCYR/XR). Time series acquisition of 720 frames (256 Â 256 pixels for 254.46 Â 254.46 mm) with 5-to 10-s intervals was performed to measure chloride concentration as a function of time in control conditions, during a 30-to 60-min period of applications of drugs, and over a 30-to 60-min period of wash-out.
For morphologic analysis, organotypic slices were imaged through the CA1 pyramidal cell layer (z-axis dimension 0-100 mm, 1-to 2-mm step size). ImageJ 1.51 software (National Institutes of Health) was used for quantitative analysis. Regions of interest (ROIs) were selected using the chloride insensitive CFP fluorescence. The ratio of the YFP/CFP fluorescence intensity was used for [Cl -] i calculation (Kuner and Augustine, 2000;Berglund et al., 2008;Glykys et al., 2009). The CFP emission of Clomeleon was used for the high-resolution morphologic analysis .

Statistical analysis
Group measures are expressed as mean 6 SD or median (25À75%) 6 SD as indicated. The Shapiro-Wilk test was used to determine normality of the data. The Student's t test (paired or unpaired) was performed for parametric comparison of normally distributed data. The Wilcoxon signed-rank test (paired data) and Mann-Whitney test (unpaired data, two-tail) were used for non-parametric comparison of arbitrary distributed data. One-way repeated measures (RM) ANOVA was used for multiple comparison of parametric data to evaluate the differences in the mean values among the control and treatment groups. The Friedman RM ANOVA on ranks was used for non-parametric data to determine the differences in the median values among the control and treatment groups. The Tukey's test was used for all pairwise comparisons of the responses to the different treatment groups. The level of significance was set at p , 0.05.

Results
Chloride transients during ILDs and baseline chloride changes during epileptogenesis Organotypic hippocampal slice cultures from CLM-1 mice expressing the genetically encoded intracellular chloride fluorophore Clomeleon were used as a model of traumatic brain injury and epileptogenesis in vitro Hippocampal slices were incubated for three to four weeks. Non-invasive extracellular field potential recordings and two-photon fluorescence Clomeleon imaging were performed in the CA1 pyramidal cell layer to monitor neuronal network activity and [Cl -] i (Fig. 1A,B). The first week latent period was followed by spontaneous ILDs and status epilepticus (SE). Recurrent ILDs were characterized by an initial spike-and-wave bursts followed by secondary ictal tonicclonic discharges, and subsequent postictal depression. SE was defined as continuous ILDs for at least 5 min, or by sustained ictal-like tonic-clonic epileptiform discharges without recovery to baseline activity between the ILDs (Fig. 1C). The incidence of spontaneous ILDs and SE, and the mean duration and frequency of ILDs progressively increased during epileptogenesis (Fig. 1D). The averaged mean duration of ILDs increased from 15 6 2.6 s at DIV6-DIV8 (N = 12 slices) to 64.9 6 56.2 s at DIV27-DIV28 (N = 11 slices), and the mean frequency of these ILDs increased from 6 6 10.2 ILDs per hour at DIV6-DIV8, 21 6 9.7 ILDs per hour at DIV27-DIV28 (Fig. 1D).
Two-photon imaging of Clomeleon was performed during extracellular field potential recordings (Fig. 1B). Under control conditions, baseline [Cl -] i distribution in CA1 pyramidal cells varied from 5 to 20 mM (Fig. 1A,B). In line with our previous studies (Lillis et al., 2012;Glykys et al., 2014), [Cl -] i transiently increased in all pyramidal cells at the onset of spontaneous ILDs and was further elevated during the ILDs. Pharmacological manipulations of Cl transport could produce anticonvulsant effects that affect [Cl -] i as a consequence of reduced seizure activity, separately from the consequences of altered transport. To estimate how direct anticonvulsant effects might alter [Cl -] i , we applied the sodium channel blocker TTX (1 mM). TTX rapidly abolished recurrent ILDs and corresponding [Cl -] i transients (Fig. 1E), and significantly reduced the median baseline [Cl -] i from 8.9 (7.15-11) 6 4.7 to 7.5 (5.1-10.8) 6 4.1 mM [N = 6 slices at DIV11-DIV16, n = 32 paired cells; Wilcoxon signed-rank test, sum, of negative ranks (W) = 474, Z = 3.92, p , 0.001; Fig. 1F], suggesting activity dependent baseline neuronal chloride accumulation.
The baseline [Cl -] i in the CA1 pyramidal cell layer progressively decreased during the first week of organotypic slice incubation [N = 29 slices at DIV0-DIV28, nonlinear extreme curve fit: Y0 = 23.7 6 2.36 (median 6 SE), XC = 8.3 6 1.1, A = À16 6 2.3 M; R 2 = 0.67; ANOVA: F = 115.3, p , 0.001; Fig. 1G]. This progressive decrease in baseline [Cl -] i and corresponding negative shift in E Cl during the first week of incubation correlated with a decrease in the number of acutely damaged cells with higher [Cl -] i because of dissection (Berdichevsky et al., 2012;  Research Article: New Research al., 2012) as well as with postnatal changes in CCC expression (Stein et al., 2004;Dzhala et al., 2005;Takayama and Inoue, 2010). Starting from the second week of incubation, the median baseline [Cl -] i progressively increased with corresponding positive shifts in E Cl and the increasing incidence, duration and frequency of spontaneous ILDs (Fig.  1D,G). Positive shifts in E Cl and corresponding changes in GABA action are thought to contribute to facilitation of recurrent epileptiform discharges and resistance of ILDs to GABAergic anticonvulsants (Dzhala and Staley, 2003;Dzhala et al., 2010;Khazipov et al., 2015;Glykys et al., 2017). We therefore determined the contribution of KCC2 cotransporter activity to neuronal chloride elevation and extrusion rates, and facilitation of recurrent ILDs during epileptogenesis.
The selective KCC2 inhibitor VU0463271 reduced chloride extrusion and increased duration of ILDs VU0463271, a potent and selective inhibitor of the neuronal K 1 -Clcotransporter, KCC2 (IC 50 = 61 nM), exhibits .100-fold selectivity versus the NKCC1 (Delpire et al., 2012). Selective inhibition of KCC2 with VU0463271 led to hyperexcitability and epileptiform discharges in hippocampal slices exposed to low Mg 21 and also in vivo (Sivakumaran et al., 2015). We determined the contribution of KCC2 cotransporter activity to baseline [Cl -] i elevation and extrusion rates during recurrent ILDs, and their correlation with the frequency, duration, and power of ILDs.
At DIV14-DIV21, spontaneous neuronal network activity was characterized by short inter-ILDs and prolonged ILDs, reminiscent of seizure-like activity in vivo (Fig. 3A,B). VU0463271 at the lower concentration (0.1 mM) did not  Fig. 4B). The increased duration and power of ILDs in the presence of VU04663271 (1 mM) suggest that KCC2 transport activity contributes to recovery of baseline [Cl -] i and GABAergic inhibition during epileptiform discharges, and may contribute to termination of ILDs, although the concentration of antagonist required was substantially higher than the reported IC 50 for KCC2 (Delpire et al., 2012).
To test whether these "dual" effects of VU04663271 (1 mM; a reduction of the frequency of recurrent ILDs, but an increase in the duration and power of ILDs) were associated with changes in neuronal baseline [Cl -] i and chloride transport, we compared the baseline [Cl -] i and decay time constant of [Cl -] i extrusion during recurrent ILDs in control conditions, during application of VU04663271, and during consecutive application of the sodium channel blocker TTX in the presence of VU04663271 (Fig. 3F-H). TTX application blocks seizure activity directly, providing a means to assess the effects of the intensity of ILD activity on [Cl -] i levels and kinetics separately from effects on Cl transport. Bath application of VU04663271 (1 mM for 30 min) increased the median baseline chloride concentration from 12.1 (9.   Fig. 3H]. In contrast, VU0463271 application in the presence of TTX (Fig. 3I) did not significantly change the median baseline chloride concentration from 8.5 (5.7-11.8) 6 2.7 to 8.3 (5.7-9.9) 6 2.4 mM (N = 4 slices at DIV14-DIV17, n = 24 paired cells; Wilcoxon signed-rank test: W = 200, Z = 1.41, p = 0.16; Fig. 3I). Thus, the increased duration and power of intermittent ILDs in the presence of the KCC2 blocker VU04663271 correlates with increased magnitude and duration of [Cl -] i elevation, consistent with a reduced extrusion rate of [Cl -] i . We next used pharmacological tools to determine whether GABA A -R block prevents the epileptogenic action of the KCC2 antagonist VU04663271 (Fig. 5A,B). Under similar experimental conditions (DIV14-DIV18), bath application of the GABA A -R antagonist SR95531 (10 mM) reduced the mean frequency of spontaneous ILDs from 10 6 3 to 1.33 6 0.52 ILD/30 min and induced large amplitude inter-ILDs (IEDs). Subsequent application of VU04663271 (1 mM), in the continued presence of SR95531, did not change the mean frequency of epileptiform discharges (N = 6 slices at DIV14-DIV18; one-way RM ANOVA: DF = 17, F = 59.85, p , 0.001; Tukey's test: control compared with SR95531, df = 8.7, q = 12.3, p , 0.001; control compared with VU04663271 in the presence of SR95531, df = 10.0, q = 14.3, p , 0.001; SR95531 compared with VU04663271 in the presence of SR95531, df = 1.33, q = 1.9, p = 0.405; Fig. 5B). The net effect of the changes in ictal and interictal activity was a non-significant decrease in the mean power of electrical activity by 31%, from 852.5 6 275.3 mV 2 in control to 588.1 6 408.32 mV 2 in the presence of SR95531, and to 503.3 6 396.6 mV 2 during subsequent application of VU04663271 (1 mM), in the continued presence of SR95531 (N = 6; one-way RM ANOVA: DF = 17, F = 2.231, p = 0.16; Fig. 5B). These data suggest that the epileptogenic effects of the KCC2 blocker VU04663271 on the duration and power of ILDs require activation of GABA A receptors.
We also tested whether cation-chloride cotransport was necessary for the epileptogenic effects of VU04663271 (Fig.  5C,D). Under similar experimental conditions (DIV14-DIV19), bath application of a concentration of bumetanide (200 mM) that blocks both NKCC1 and KCC2 cotransporters significantly reduced the mean frequency of spontaneous ILDs from 12.67 6 5.6 to 6.8 6 3.2 ILD/30 min. Subsequent application of 1uM VU04663271, in the continued presence of bumetanide, did not significantly change the mean frequency of ILDs to 5.8 6 3.7 ILD/ 30 min (N = 6 slices at DIV14-DIV19; one-way RM ANOVA: DF = 17, F = 7.9, p = 0.009; Tukey's test: control compared with bumetanide, df = 5.87, q = 4.4, p = 0.026; control vs VU04663271 in the presence of bumetanide, df = 6.87, q = 5.2, p = 0.011; bumetanide vs VU04663271 in the presence of bumetanide, df = 1, q = 76, p = 0.855; Fig. 5D). In addition, both bumetanide and subsequent VU04663271 applications did not change the mean duration of ILDs (N = 5 slices; one-way RM ANOVA: DF = 14, F = 2.4, p = 0.152; Fig. 5D). The net effect of the changes in ictal and interictal activity was a significant decrease in the mean power of electrical activity by 23% from 648.8 6 172.4 mV 2 in control to 504.17 6 147.3 mV 2 in the presence of bumetanide (Fig. 5D). Subsequent application of 1 mM VU04663271, in the continued presence of bumetanide, did not significantly change the mean power of electrical activity to 547.6 6 218.1 mV 2 (N = 6; one-way RM ANOVA: DF = 17, F = 7.6, p = 0.01; Tukey's test: control compared with bumetanide, df = 144.7, q = 5.36, p = 0.009; control vs VU04663271 in the presence of bumetanide, df = 101.17, q = 3.75, p = 0.06; bumetanide vs VU04663271 in the presence of bumetanide, df = À43.5, q = 1.62, p = 0.51; Fig. 5D). These results suggest that KCC2 transporter activity is functional during recurrent ILDs, that VU04663271 antagonizes KCC2, and that this antagonism is necessary for the observed increase in the duration and power of ILDs, and induction of electrical SE (Fig. 3).

Furosemide suppressed recurrent ILDs and delayed chloride extrusion
The loop diuretic furosemide acts to inhibit both NKCC and KCC with about equal potency (K i = 25-50 mM). In contrast to the epileptogenic action of the KCC2 inhibitor VU0463271 (Figs. 2, 3), high concentrations of furosemide (1-2 mM) demonstrate strong anti-ictal effects in various in vitro models of epilepsy (Hochman et al., 1995;Gutschmidt et al., 1999;Haglund and Hochman, 2005;Blauwblomme et al., 2018). To address these discrepancies, we determined the effects of furosemide (0.1 mM and 1 mM) on spontaneous epileptiform discharges alone and in combination with GABA A receptor inhibitor SR95531, NKCC1 blocker Bumetanide and the low-affinity KCC2 blocker VU0240551 (Figs. 6, 7).

Effects of KCC2 cotransporter enhancer CLP257 on [Cl -] i and recurrent ILDs
Enhancing KCC2 transport activity may be a useful therapeutic strategy to restore and/or reduce a baseline [Cl -] i , improve GABAergic inhibition, suppress ILDs and prevent epileptogenesis (Gagnon et al., 2013;Moore et al., 2018). We therefore determined the acute anticonvulsant efficacy of the putative KCC2 activator CLP 257 (1-30 mM; Gagnon et al., 2013) and its effects on neuronal [Cl -] i in the organotypic hippocampal slices model of epileptogenesis in vitro.
In conclusion, our results demonstrated that low concentration of CLP257 (1 mM) modestly improved [Cl -] i homeostasis and reduced the duration of recurrent ILDs. However, CLP257 (1 mM) did not change the frequency of recurrent ILDs. In contrast, CLP257 at high concentration (30 mM) significantly reduced the frequency of ILDs and power of electrical activity.

Summary of findings
In an in vitro preparation exhibiting spontaneous electrographic interictal spikes and ILDs, we found two important effects that were clearly not related to CCCs: (1) GABA A receptor antagonists blocked ILDs and produced high-amplitude periodic inter-ILDs; (2) blocking ILDs reduced the baseline (steady state) [Cl -] i independently of effects on CCCs. Manipulation of cation-chloride cotransport in this preparation with currently available pharmacological agents demonstrated: (3) antagonizing KCC2 activity prolonged the ictal increase in [Cl -] i ; (4) antagonizing KCC2 activity increased ILD duration; (5) increasing KCC2 activity reduced the ictal [Cl -] i increase; (6) increasing KCC2 activity reduced ILD durations; (7) the effects of high-affinity KCC2 modulators were dependent on the activity of GABA A receptors; however, (8) the effects of the high-affinity KCC2 antagonist VU0463271 on [Cl -] i and ILDs were only apparent at concentrations 20 times higher than the IC 50 (61 nM) for KCC2 block established in dissociated cell cultures; (9) the effects of the KCC2 enhancer CLP257 on [Cl -] i and ILDs were evident at two times the EC 50 (616 nM) for KCC2 enhancement.
We also found evidence of effects of CCC antagonists that were not mediated by KCC2 or NKCC1. The largest of these effects was the profound GABA A receptor-dependent anti-ictal effect of high concentrations of furosemide (Hochman et al., 1995;Gutschmidt et al., 1999;Haglund and Hochman, 2005;Blauwblomme et al., 2018) that was enhanced in the presence of NKCC1 and KCC2 antagonists (Fig. 7). This effect was correlated with a prolongation of the postictal [Cl -] i transient that greatly exceeded the maximum effect of specific CCC blockers. These findings raise the possibility that blockade of an unidentified CCC, perhaps KCC3, mediates these remarkable anticonvulsant effects, but it is equally possible that the anticonvulsant effects are not related to the reduced Cltransport. Rather, other nonspecific effects (Andreasen and Nedergaard, 2017) may be responsible for this profound anticonvulsant effect, for example, interference with vesicular glutamate uptake (Roseth et al., 1995), or other membrane Clexchangers or channels, particularly GABA A receptor-operated channels in light of our first finding (Korpi and Lüddens, 1997;Mtchedlishvili and Kapur, 2006).

The role of KCC2 transport activity in termination of ILDs
Activity-dependent neuronal Claccumulation and a consequent transient depolarizing shift in the reversal potential of the GABA A -R (E GABA ) have been suggested to contribute to generation of ictal-clonic afterdischarges (Fujiwara-Tsukamoto et al., 2006Ellender et al., 2014). We determined the role of KCC2 transport activity in extrusion of neuronal Cland termination of ictal-like tonic-clonic activity using pharmacological inhibition of KCC2 transport activity. Our data demonstrated that the high-affinity KCC2 inhibitor VU0463271 increased ILD-induced [Cl -] i elevation and the duration of ILDs (Fig. 3). Neuronal baseline Clremained elevated during prolonged ictal-like activity and the mean decay time of chloride extrusion increased from the mean 21.4 6 7.6 s in control to 188.1 6 111 s in the presence of VU04663271. The sodium channel blocker TTX in the presence of VU0463271 abolished recurrent ILDs and recovered baseline [Cl -] i within the mean 185.5 6 67.2 s. In addition, VU0463271 application in the presence of TTX did not significantly change the mean baseline [Cl -] i . Our results demonstrate that KCC2 transport activity efficiently recovers E Cl during ILDs, restores GABAergic inhibition and these effects are correlated with more rapid termination of ILDs. Inhibition of KCC2 transport activity delays chloride extrusion rate suggesting that impaired KCC2 function may contribute to the transition from the short ictal-like events to sustained SE.

Modulation of CCCs for control of drug-resistant ILDs
Neuronal chloride concentration ([Cl -] i ) is an important determinant of both postsynaptic GABA A -receptor-mediated signaling and cell volume regulation. Clequilibrium is mediated by a Donnan system that includes intra and extracellular impermeant anions, and in which the CCCs comprise the requisite cation and chloride membrane permeability (Glykys et al., 2017). Increasing this permeability by increasing the maximum velocity of cation-chloride cotransport should not change the baseline [Cl -] i . However, increasing the maximum velocity of transport could increase the ability of neurons to buffer synaptically-mediated Clinflux. Importantly, this increased ability is predicated on the extracellular potassium; increases in extracellular potassium directly alter the steady state The maximum velocity of CCCs is regulated by a system of the WNK-SPAK/OSR1 kinase complex pathways [serine-threonine kinase WNK (with no lysine) and SPS1related proline/alanine-rich kinase (SPAK) or the SPAK homolog oxidative stress-responsive kinase 1 (OSR1)] that result in the compensatory phosphorylation and dephosphorylation processes (Kahle et al., 2006;Melo et al., 2013;Alessi et al., 2014;de Los et al., 2014). In immature neurons and isotonic conditions KCC2 is phosphorylated at two C-terminal threonine (Thr906 and Thr1007) and inactive, whereas CNS development and hypotonic conditions promote their dephosphorylation and activation Pisella et al., 2019;Watanabe et al., 2019).
Traumatic injury to the brain alters the equilibrium value of chloride, resulting in an elevated baseline [Cl -] i and depolarizing shift in E GABA (van den Pol et al., 1996;Pond et al., 2006;Kahle et al., 2008;Blaesse et al., 2009;Dzhala et al., 2010Dzhala et al., , 2012. Acute brain injury results in cytotoxic edema, in which there is a net neuronal uptake of chloride salts and water. Suppressing CCC activity may be a useful therapeutic strategy in this condition to prevent [Cl -] i accumulation, reduce [Cl -] i and swelling in injured neurons, restore GABAergic inhibition and suppress acute ILDs (Kahle et al., 2008;Gagnon et al., 2013;Glykys et al., 2017). Previous studies in in vitro and in vivo models of hypoxia-ischemia, recurrent seizures and neuronal brain injury demonstrated that inhibition of NKCC1 reduced [Cl -] i in injured neurons, enhanced GABAergic inhibition and enhanced efficacy of GABAergic anticonvulsants (Dzhala et al., 2005(Dzhala et al., , 2008Pond et al., 2006;Nardou et al., 2009Nardou et al., , 2011Cleary et al., 2013;Dzhala and Staley, 2015;Sivakumaran and Maguire, 2016).
In the current study, we investigated the role of CCC inhibitors in chronic epilepsy in vitro rather than seizures in the setting of acute brain injury in vivo. Studies in acute brain slice preparations from patients with chronic epilepsy have supported the acute efficacy of bumetanide (Pallud et al., 2014). However, in acute brain slices, there is also a significant degree of acute brain injury , making it difficult to separate effects because of acute versus chronic changes. We found only GABA A receptor-dependent pro-ictal effects of KCC2 antagonists at concentrations that also prolonged ictal [Cl -] i transients. This is consistent with the idea that reducing KCC2 activity degrades Clhomeostasis and the ability to restore [Cl -] i and the proper polarity of GABA A signaling after synaptic activity. KCC2 antagonists have a potential anti-ictal effect in their reduction of [K 1 ] 0 transients. We did not find evidence for an anti-ictal effect, but our slice cultures were perfused via the bath rather than via the vasculature, so it is possible that bulk ACSF flow reduced this potential anti-ictal effect on KCC2-dependent [K 1 ] 0 accumulation (Viitanen et al., 2010).
There is substantial interest in overexpression and enhancement of KCC2 transport activity as a novel therapeutic strategy to improve synaptic GABAergic inhibition in neurologic disorders (Kahle et al., 2008;Gagnon et al., 2013;Moore et al., 2018;Magloire et al., 2019). Highthroughput screening identified a new high-affinity compound CLP257 (EC 50 = 616 nM) and its carbamate prodrug derivative CLP290 that selectively activates KCC2 over other KCC family members, NKCC1 and GABA A receptors (Gagnon et al., 2013). CLP257 restored impaired Cltransport in adult spinal cord slices with impaired KCC2 function, restored Clextrusion and renormalized stimulus-evoked responses in adult neurons in an experimental model of neuropathic pain. However, the pharmacodynamic profile of CLP257, including KCC2 specificity and mode of action, has not been determined. High concentrations of CLP257 (30-50 mM) was reported to potentiate GABA A receptors (Cardarelli et al., 2017). Even higher concentrations of CLP257 (100 mM) increased the duration of ILDs without affecting their frequency (Hamidi and Avoli, 2015).
We investigated whether enhancing KCC2 transport activity with a low concentration of CLP257 affects baseline [Cl -] i and [Cl -] i extrusion rates in injured neurons during ictal-like events, and the downstream effects on frequency and duration of recurrent ILDs in the organotypic hippocampal slice in vitro model of epileptogenesis (Fig.  8). We found that low concentrations of CLP257 (1 mM) improved [Cl -] i homeostasis, increased postictal [Cl -] i extrusion rates, and modestly reduced ictal duration without effecting ictal frequency. Our data suggest that the modest anticonvulsant actions of low dose CLP257 are likely partially mediated by enhanced KCC2 transport activity that more efficiently restores neuronal baseline Clduring termination of ictal-like events. In addition, our data validate CLP257 as a promising target of investigation for antiepileptic therapy and highlight the ongoing need to develop more specific activators of KCC2 cotransport.