Neuronal Transmembrane Chloride Transport Has a Time-Dependent Influence on Survival of Hippocampal Cultures to Oxygen-Glucose Deprivation

Neuronal ischemia results in chloride gradient alterations which impact the excitatory–inhibitory balance, volume regulation, and neuronal survival. Thus, the Na+/K+/Cl− co-transporter (NKCC1), the K+/ Cl− co-transporter (KCC2), and the gamma-aminobutyric acid A (GABAA) receptor may represent therapeutic targets in stroke, but a time-dependent effect on neuronal viability could influence the outcome. We, therefore, successively blocked NKCC1, KCC2, and GABAA (with bumetanide, DIOA, and gabazine, respectively) or activated GABAA (with isoguvacine) either during or after oxygen-glucose deprivation (OGD). Primary hippocampal cultures were exposed to a 2-h OGD or sham normoxia treatment, and viability was determined using the resazurin assay. Neuronal viability was significantly reduced after OGD, and was further decreased by DIOA treatment applied during OGD (p < 0.01) and by gabazine applied after OGD (p < 0.05). Bumetanide treatment during OGD increased viability (p < 0.05), while isoguvacine applied either during or after OGD did not influence viability. Our data suggests that NKCC1 and KCC2 function has an important impact on neuronal viability during the acute ischemic episode, while the GABAA receptor plays a role during the subsequent recovery period. These findings suggest that pharmacological modulation of transmembrane chloride transport could be a promising approach during stroke and highlight the importance of the timing of treatment application in relation to ischemia-reoxygenation.


Introduction
Stroke is the second leading cause of mortality worldwide, with an annual 5.5 million death toll [1]. Moreover, the incidence of stroke in young adults (18-50 years) has significantly increased in the past decade, with considerable socioeconomic impact because of high health-care costs and loss of labor productivity [2]. Despite major efforts to advance knowledge on the pathophysiology and treatment of (gabazine) [28] as well as the effect of the GABA A agonist isoguvacine, on the viability of hippocampal cultures undergoing OGD. Viability was measured using a metabolic assay which requires a concomitant reoxygenation time of 3 h. We measured the effect of drug application both during OGD (mimicking the preclinical therapeutic window), as well as post-exposure (mimicking the revascularization). Considering the role that [Cl − ] i plays in both volume regulation and excitatory-inhibitory balance, we postulated that modulating [Cl − ] i by blocking its transmembrane transport would influence cellular viability with an inversely proportional relation. Our results suggest a complex, timing-dependent effect of treatment on cellular viability.

Primary Cultures of Hippocampal Neurons
Primary hippocampal cell cultures were obtained from Wistar rat pups at postnatal day zero (P0), following a protocol previously described [29]. All animal procedures were carried out with the approval of the local ethics committee for animal research Carol Davila University of Medicine and Pharmacy (Bucharest, Romania) and in accordance with the European Communities Council Directive 86/609/EEC.
Briefly, the hippocampi were dissected, and the meninges and vascular plexuses were removed. The tissue was subjected to mechanical trituration and enzymatic digestion using papain (Worthington, 3.8 U/mL), and the cell suspension obtained was kept in culture medium (CM) containing Neurobasal-A medium (Invitrogen) supplemented with B-27 (Invitrogen), l-glutamine (HyClone, 0.5 mM), and antibiotic-antimycotic (Invitrogen). The cell suspension was plated in 24-well plates coated with poly-d-lysine (Sigma, 70,000-150,000 kDa, 100 µg/mL) at a density of 150,000 cells/well. The plates were placed in a humidified 5% CO 2 incubator at 37 • C. After 5 days in vitro (DIV), the cultures were fed by removing half of the CM volume in each well (250 µL) and replacing it with 400 µL newly prepared CM.

Exposure to Oxygen-Glucose Deprivation (OGD)
The primary hippocampal cell cultures were grown until DIV 7, a developmental time accompanied in rodents with a progressive switch of the GABA A receptor activation from excitatory to inhibitory [30]. Previous in vitro experiments in mice reported that at DIV 6 the NKCC1 expression is approximately 4 times lower than at DIV 1, while the KCC2 expression is increased by approximately 15 times compared to DIV 1, which is consistent with a mature phenotype [31].
OGD was induced on DIV 7 as previously described [24,32], by removing CM from the wells and replacing it with a deoxygenated experimental medium without glucose (EM-G), consisting of Neurobasal-A without both glucose and sodium pyruvate (Life Technologies), supplemented with HEPES (Sigma, 10 mM) and l-glutamine (HyClone, 0.5mM). The plates were then immediately placed in a humidified hypoxia chamber (Billups-Rothenberg), which was flushed with 100% N 2 for 10 min, then sealed, and placed in an incubator at 37 • C for 2 h.
Afterwards, EM-G was removed and cultures were gently washed with an assessment medium (AM) containing Neurobasal-A without phenol red (Life Technologies), supplemented with HEPES (10 mM) and l-glutamine (HyClone, 0.5 mM) and incubated for 3 h at 37 • C and 5% CO 2 in AM with resazurin (Sigma, 100 µM) to assess their metabolic viability. Control cultures were subjected to a sham exposure to normoxic conditions at 37 • C and 5% CO 2 for 2 h in an experimental medium containing Neurobasal-A with glucose 25 mM (EM + G).

Assessment of Cellular Metabolism and Viability
The viability of the hippocampal cultures was evaluated by measuring their cellular metabolism using a cell permeable redox indicator, resazurin (Aldrich) [33,34], as previously described [24]. Within cells with active metabolism, resazurin is reduced to resorufin, a fluorescent product. Resazurin requires a 3 h reoxygenation period to be metabolized, allowing for a delayed, post-exposure treatment time window. Resorufin fluorescence was read at 535 nm excitation and 595 nm emission using a multimode detector (DTX880, Beckman Coulter, Inc., Indianapolis, IN, USA).

Treatment with Chloride Membrane Transport Blockers
Treatments were applied either during the 2 h exposure to OGD/sham normoxia (T1) or postexposure, during the 3 h necessary for viability assessment (T2).
We blocked outward Cl − transport with DIOA, a KCC2 blocker (Sigma, 20 µM), and inward Cl − transport with bumetanide (Sigma, 10 µM), a NKCC1 blocker as a means for increasing, respectively decreasing [Cl − ] i either during OGD (T1) or post-exposure (T2). We also used the GABA A receptor antagonist gabazine (Sigma, 30 µM) as well as the GABA A receptor agonist isoguvacine (Sigma, 40 µM) to modulate the GABA A channel, both during OGD (T1) and during post-exposure (T2). All treatments were also applied to normoxic cultures. Experimental design and treatment times are illustrated in Figure 1.

Treatment with Chloride Membrane Transport Blockers
Treatments were applied either during the 2 h exposure to OGD/sham normoxia (T1) or postexposure, during the 3 h necessary for viability assessment (T2).
We blocked outward Cltransport with DIOA, a KCC2 blocker (Sigma, 20 µM), and inward Cltransport with bumetanide (Sigma, 10 µM), a NKCC1 blocker as a means for increasing, respectively decreasing [Cl -]i either during OGD (T1) or post-exposure (T2). We also used the GABAA receptor antagonist gabazine (Sigma, 30 µM) as well as the GABAA receptor agonist isoguvacine (Sigma, 40 µM) to modulate the GABAA channel, both during OGD (T1) and during post-exposure (T2). All treatments were also applied to normoxic cultures. Experimental design and treatment times are illustrated in Figure 1.

Data Analysis
The data was analyzed using Prism (GraphPad 7 Software). Data is presented as percentages normalized to the control values (normoxia + no treatment condition). The statistical unit was the number of wells tested per condition (n). The viability of cultures after different OGD treatments was compared to that of OGD without treatment from the same cultures. This results in slightly different OGD viabilities across treatment conditions. However, the degree of metabolic lesion was severe in all cultures, ranging from 42.5% to 59.6% (average 52.25%). All variables were tested for normality of distribution using the Shapiro-Wilk test and central tendencies are consequently reported as mean ± SD. Statistical significance was evaluated using ordinary two-way analysis of variance (ANOVA) with factors of oxygenation (normoxia, OGD) and treatment (no treatment, treatment T1, treatment T2) and Tukey's multiple comparisons test, with α set at 0.05.

DIOA Treatment during Oxygen-Glucose Deprivation Decreases Cellular Viability of DIV 7 Hippocampal Cell Cultures
Our 2 h OGD protocol decreased neuronal viability to 57.3% ± 12.6% (n = 23, six cultures) as

Data Analysis
The data was analyzed using Prism (GraphPad 7 Software). Data is presented as percentages normalized to the control values (normoxia + no treatment condition). The statistical unit was the number of wells tested per condition (n). The viability of cultures after different OGD treatments was compared to that of OGD without treatment from the same cultures. This results in slightly different OGD viabilities across treatment conditions. However, the degree of metabolic lesion was severe in all cultures, ranging from 42.5% to 59.6% (average 52.25%). All variables were tested for normality of distribution using the Shapiro-Wilk test and central tendencies are consequently reported as mean ± SD. Statistical significance was evaluated using ordinary two-way analysis of variance (ANOVA) with factors of oxygenation (normoxia, OGD) and treatment (no treatment, treatment T1, treatment T2) and Tukey's multiple comparisons test, with α set at 0.05.

Discussion
In this study we explored the effect of pharmacologically blocking chloride membrane transport via the NKCC1 and KCC2 co-transporters and the effect of GABAA receptor inhibition or activation on post-ischemic neuronal viability, when applied either during OGD or post-exposure.

Treatment with Bumetanide Is Associated with Increased Cellular Viability of DIV 7 Hippocampal Cells after Oxygen-Glucose Deprivation
Our results show that bumetanide treatment exhibits a neuroprotective effect when added during the OGD insult, but has no benefit on neuronal viability when added post-exposure. These findings extend previous studies on cultured cortical neurons suggesting that NKCC1 contributes to glutamate mediated excitotoxicity, its blocking being ineffective in preventing cell death when added after OGD exposure [35]. The underlying mechanism is thought to be the prevention of cell swelling secondary to Na + and Clentry via NKCC1, which is prevalent during the rapidly triggered glutamate-mediated excitotoxicity in early OGD, thus rendering NKCC1 a valuable therapeutic target during, but not after, OGD episodes [35,38]. Moreover, N-methyl-D-aspartate receptor activation and high extra-cellular K + , factors that are known to participate in ischemic damage and excitotoxicity, have been reported to stimulate NKCC1 activity in neurons [40][41][42]. There is also compelling evidence suggesting that the activity of NKCC1 increases through phosphorylation of threonine residues of its structure during ischemia, its expression remaining unchanged [19,22,39,43]. This increased NKCC1 activity creates a functional phenotype resembling the immature state where

Discussion
In this study we explored the effect of pharmacologically blocking chloride membrane transport via the NKCC1 and KCC2 co-transporters and the effect of GABA A receptor inhibition or activation on post-ischemic neuronal viability, when applied either during OGD or post-exposure.

Treatment with Bumetanide Is Associated with Increased Cellular Viability of DIV 7 Hippocampal Cells after Oxygen-Glucose Deprivation
Our results show that bumetanide treatment exhibits a neuroprotective effect when added during the OGD insult, but has no benefit on neuronal viability when added post-exposure. These findings extend previous studies on cultured cortical neurons suggesting that NKCC1 contributes to glutamate mediated excitotoxicity, its blocking being ineffective in preventing cell death when added after OGD exposure [35]. The underlying mechanism is thought to be the prevention of cell swelling secondary to Na + and Cl − entry via NKCC1, which is prevalent during the rapidly triggered glutamate-mediated excitotoxicity in early OGD, thus rendering NKCC1 a valuable therapeutic target during, but not after, OGD episodes [35,38]. Moreover, N-methyl-D-aspartate receptor activation and high extra-cellular K + , factors that are known to participate in ischemic damage and excitotoxicity, have been reported to stimulate NKCC1 activity in neurons [40][41][42]. There is also compelling evidence suggesting that the activity of NKCC1 increases through phosphorylation of threonine residues of its structure during ischemia, its expression remaining unchanged [19,22,39,43]. This increased NKCC1 activity creates a functional phenotype resembling the immature state where bumetanide has been initially reported to have a neuroprotective effect during OGD, in both neonatal rat hippocampal slices [44] and immature primary hippocampal cultures [24].

Treatment with DIOA Is Associated with Decreased Cellular Viability of DIV 7 Hippocampal Cell Cultures after Oxygen-Glucose Deprivation
DIOA treatment was detrimental when added during OGD. This effect could be explained by the supplementary increase in intracellular Cl − concentration following the blockage of its outward transport, since KCC2 is one of the primary Cl − extruders in mature neurons [45,46]. However, no viability changes were noted when DIOA was added during reoxygenation. These results are supported by literature data showing KCC2 protein levels significantly dropping by 30% at 1 h post OGD and further dropping by 70% at 2 h post OGD, in hippocampal slices [19]. Thus, KCC2 availability during reoxygenation might be too low for DIOA treatment to give rise to any significant changes in neuronal viability. All this data point to a narrow window of opportunity for DIOA action at the beginning of OGD, in accordance with our results. Furthermore, KCC2 has been shown to serve two main functions in mature neurons: maintaining a low [Cl − ] i to allow Cl − influx via ligand-gated Cl − channels, and buffering of external K + concentration [47,48]. Concerning the latter function, it has been shown that, under conditions that mimic ischemia/OGD, KCC2-mediated transport will reverse in response to small increases in extracellular K + , generating a Cl − influx [39,47,49].

4.3.
Blocking of the GABA A Receptor Using Gabazine during Reoxygenation, but Not during Oxygen-Glucose Deprivation, Is Associated with Decreased Cellular Viability of DIV 7 Hippocampal Cells GABA release during OGD has been hypothesized to be either neuroprotective, because of its hyperpolarizing properties counteracting the glutamate-mediated excitotoxicity, or neurotoxic, due to GABA A receptors' activation facilitating Cl − entry into neurons and consequent cell swelling [19,50]. We found that blocking GABA A with gabazine during OGD in primary hippocampal cultures elicits no changes in cellular viability. In hippocampal slices, OGD triggers an increased glutamate and GABA release through two sequential mechanisms: exocytosis followed by a brisk anoxic depolarization and reversal of the glutamate/GABA transporters [38,51]. The anoxic depolarization is accompanied by a significant disruption of ionic homeostasis. There is a sharp increase in [Ca 2+ ] i , which causes the inactivation of GABA A receptors and subsequently prevents Cl − influx during OGD [38,52,53]. This is a long-lasting inactivation, reported to persist over 1 h for high-calcium loads [53]. Other ischemia-related mechanisms involved in GABA A receptor activity reduction and desensitization, such as generation of eicosanoids and reactive oxygen species or ATP level reduction, have also been described [54].
Our data is in accordance with these previous findings, as it shows that application of GABA A antagonist gabazine during OGD does not significantly affect cellular viability, most likely due to the already diminished GABA A activity which naturally occurs during ischemic episodes [38].
Additionally, our findings indicate that blocking GABA A during reoxygenation is detrimental. Beyond the 3 h reoxygenation period, GABA A inactivation gradually diminishes as ionic homeostasis is presumably restored in surviving neurons, allowing for GABA A chemical manipulation. The high [Cl − ] i following OGD [19,22,35,36] is expected to cause Cl − outflow, as the Cl − equilibrium potential (E Cl ) is markedly shifted towards positive values [55][56][57][58]. However, literature data suggests that the inhibitory to excitatory shift would require that E Cl exceeds the membrane potential for spiking, which only occurs in severe ischemic insults [54]. Thus, most studies in animal and in vitro models of ischemia report neuroprotective effects of GABAergic drugs [59][60][61]. Our results support these effects, by showing that the blocking of these neuroprotective mechanisms has detrimental effects.

Enhancing GABA A Activation Using Isoguvacine Does Not Influence Cellular Viability Either during Oxygen-Glucose Deprivation or Post-Exposure
Inducing sedation by enhancing the activity of the GABA A receptor is quite common in stroke patients and it is only sensible to ask whether this pharmacological manipulation could potentially have any detrimental effects on neuronal viability in this setting, especially since literature data are controversial regarding this matter [62][63][64]. It has also been suggested that benzodiazepines may be neuroprotective when given soon after stroke, but harmful when administered during late recovery [63]. We did not find a significant influence of GABA A receptor activation with isoguvacine on cellular viability of neurons during either OGD or post-exposure.

Conclusions
In conclusion, we showed that the timing of treatment application in relation to the moment of the ischemia-reoxygenation is of paramount importance: the function of chloride co-transporters NKCC1 and KCC2 has a decisive impact on neuronal viability during the acute ischemic episode, while the GABA A receptor plays a key role in the recovery period that follows. Our findings highlight a temporal widow for pharmacological modulation of chloride membrane transport as a promising neuroprotective strategy.

Conflicts of Interest:
The authors declare no conflicts of interest.