Possible alterations in GABAA receptor signaling that underlie benzodiazepine-resistant seizures


Address correspondence to Stephen J. Moss, PhD, Department of Neuroscience, Tufts University School of Medicine, 136 Harrison Ave., Boston, MA 02111, U.S.A. E-mail: stephen.moss@tufts.edu


Benzodiazepines have been used for decades as first-line treatment for status epilepticus (SE). For reasons that are not fully understood, the efficacy of benzodiazepines decreases with increasing duration of seizure activity. This often forces clinicians to resort to more drastic second- and third-line treatments that are not always successful. The antiseizure properties of benzodiazepines are mediated by γ-aminobutyric acid type A (GABAA) receptors. Decades of research have focused on the failure of GABAergic inhibition after seizure onset as the likely cause of the development benzodiazepine resistance during SE. However, the details of the deficits in GABAA signaling are still largely unknown. Therefore, it is necessary to improve our understanding of the mechanisms of benzodiazepine resistance so that more effective strategies can be formulated. In this review we discuss evidence supporting the role of altered GABAA receptor function as the major underlying cause of benzodiazepine-resistant SE in both humans and animal models. We specifically address the prevailing hypothesis, which is based on changes in the number and subtypes of GABAA receptors, as well as the potential influence of perturbed chloride homeostasis in the mature brain.

Soon after their discovery in the late 1950s, benzodiazepines have been used as anticonvulsant therapies (for review see Goodkin & Kapur, 2009; Neligan & Shorvon, 2009). Benzodiazepines offered substantial advantages over previous medications, as noted by the early clinicians, including high efficacy, rapid onset of action, and low toxicity. Benzodiazepines such as diazepam and lorazepam are now the standard first-line treatments for status epilepticus (SE). However, mounting clinical observations soon revealed that the effectiveness of benzodiazepines decreased substantially with increasing duration of seizures (Treiman et al., 1998; Mayer et al., 2002). These diazepam-resistant seizures are defined as refractory seizures (Shorvon, 2011). This therapeutic phenomenon has persisted to this day, and the mechanistic underpinnings are somewhat unclear (Remy & Beck, 2006). The reduction of diazepam efficacy suggests that seizures incur changes in the benzodiazepine-sensitive γ-aminobutyric acid (GABA)A receptor system, and it is possible that these alterations could contribute to the maintenance of seizures during SE and the recurrence of seizures at later stages. Therefore, understanding exactly how diazepam’s efficacy is reduced could reveal important clues about the mechanisms of seizures and provide insight into potentially more effective treatments. In this review, we attempt to clarify what is actually known about refractory seizures within the context of the associated changes in benzodiazepine-sensitive GABAA receptors as well as the theoretical contribution of altered Cl homeostasis in humans and animal models (Fig. 1).

Figure 1.

Potential mechanisms of reduced GABAA receptor function and benzodiazepine resistance. The GABAA-mediated shunt current is equal to the product of the number of open GABAA receptors (conductance) and the electrochemical driving force (voltage) on the permeant ions (in this case Cl). (A) The prevailing theory of benzodiazepine resistance is based on a reduction in conductance, that is, lower number of open benzodiazepine-sensitive GABAA receptors due to increased lateral diffusion from the subsynaptic space and internalization. (B) A hypothetical mechanism based on an elevated intracellular Cl concentration, which reduces or reverses the electrochemical driving force and the GABAA-mediated current. The figure depicts a polarity reversal of the GABAA-mediated current. Red circles are GABA molecules, and black circles are benzodiazepine molecules.

Benzodiazepine-Sensitive GABAA Receptors

GABAA receptors belong to the large family of pentameric Cysteine-loop receptors, which were formerly known as the nicotinic-acetylcholine family of receptors (Alexander et al., 2011). There are 19 GABAA receptor subunits in humans: α1-6, β1-3, γ 1-3, δ, ɛ, π, θ, and ρ1-3. GABAA receptor subunit composition dictates the subcellular localization, kinetics, and pharmacology of these receptors. Receptors containing γ subunits are sensitive to benzodiazepines, with the γ2 subunit being the major γ subunit expressed in the brain. Furthermore, the benzodiazepine-binding site is formed at the extracellular interface of a γ2 and adjacent α1-3 or α5 subunit; the α4 and α6 subunits do not support benzodiazepine binding (Puia et al., 1991; Benson et al., 1998; Rudolph & Knoflach, 2011). Most of these benzodiazepine-sensitive receptors have a low affinity for GABA and are therefore exclusively functional in the sub-synaptic space where GABA concentrations are high. The exceptions to this rule are the α5-containing receptors, which are largely extrasynaptic and mediate a portion of persistent Cl leak currents (tonic inhibition) in specific brain areas (Farrant & Nusser, 2005; Jacob et al., 2008). These specialized subunit-dependent biophysical properties suggest that benzodiazepines largely exert their inhibitory efficacy by acting on synaptic GABAergic signaling (phasic inhibition). Although the precise intramolecular mechanism that enables benzodiazepine binding to alter GABAA function is still under investigation, most evidence indicates that benzodiazepines increase the affinity of GABA (Vicini et al., 1987; Rogers et al., 1994; Goldschen-Ohm et al., 2010). The ultimate effect of benzodiazepine binding is to increase the amplitude of tonic or prolong the duration of phasic GABAA-mediated currents (Farrant & Nusser, 2005). These effects reduce excitability by decreasing the probability that excitatory postsynaptic potentials will trigger action potentials.

The Ionic Permeability of GABAA Receptors

The ionic flux through the GABAA receptor is the main determinant of its physiologic role. GABAA receptors are anion permeable receptors, in which the anion selectivity is largely determined by residues that form the ion pore lining the second transmembrane helix (Keramidas et al., 2004). Chloride is the third most abundant ion and the most abundant anion in the central nervous system. GABAA receptors are therefore predominantly Cl channels, although bicarbonate makes a significant contribution to GABAA-mediated currents under certain conditions (Bormann et al., 1987; Kaila & Voipio, 1987; Fatima-Shad & Barry, 1993; Wotring et al., 1999; Viitanen et al., 2010). For some GABAA receptors, the relative permeability of Cl to inline image is roughly 3–1. However, unlike the cation-permeable subdivision of Cysteine-loop receptors where the relative permeability of Ca2+ to Na+ is well studied (e.g., Imoto et al., 1988; Livesey et al., 2011), similar studies of the molecular determinants and GABAA subunit dependence of the Cl to inline image relative permeability are limited.

The direction that Cl and inline image flow through the GABAA receptor is entirely determined by their respective electrochemical gradients, which rely on the function of ion transporters and enzymes (Kaila, 1994; Farrant & Kaila, 2007). Active intracellular pH regulation results in a inline image reversal potential of nearly −10 to −15 mV. Therefore, under nearly all circumstances, the inline image current will depolarize the neuron. Chloride however is more complex. Typically the actions of GABA are hyperpolarizing due to the inward flux of chloride upon GABA-activated channel opening. However, under certain conditions, GABAA receptors can mediate net Cl efflux that depolarizes neurons, such as in embryonic and immature neurons (Ben-Ari et al., 2007), as well as in some subsets of mature neurons (e.g. Michelson & Wong, 1991; Verheugen et al., 1999; Song et al., 2011; Sarkar et al., 2011). The canonical hyperpolarizing Cl currents exhibited by neurons are largely due to the expression and activity of the potassium-chloride cotransporter type 2, KCC2 (Thompson & Gähwiler, 1989; Payne, 1997). KCC2 is a member of a small family of cation-chloride cotransporters (SLC12), four of which (KCC1-4) are potassium dependent and mediate K-Cl extrusion, and is the only member that exhibits constitutive transporter activity (Alexander et al., 2011). Even if a Cl load did not result in excitatory GABAergic responses, it would depolarize EGABA, thereby reducing the electrochemical driving force on GABAA-mediated Cl currents. It is therefore reasonable to propose that alterations in KCC2 function would impair GABAA receptor function, as well as the efficacy of compounds that modulate GABAA receptor activity, such as benzodiazepines.

Evidence in Humans of Reduced Benzodiazepine Efficacy and Benzodiazepine Binding Sites

Although clinical experience accumulated over the course of decades suggested that prolonged seizures and SE are more difficult to control than brief seizure episodes, two case studies in particular confirmed the early observations (Treiman et al., 1998; Mayer et al., 2002). Patients experiencing SE are more likely to develop resistance to first-line antiseizure drugs and require second- and third-line interventions (Kwan & Sperling, 2009). However, because of the time constraints of emergency room admittance, prior exposure to benzodiazepines, the various etiologies, and of course ethical considerations, it is difficult to determine the exact point at which the efficacy of a first-time exposure to benzodiazepines begins to wane in humans.

The simplest explanation for reduced benzodiazepine efficacy is a reduction in the number of benzodiazepine receptors. We will briefly review some of the data obtained from patients with temporal lobe epilepsies (TLEs). Patients exhibiting drug-resistant forms of TLE have a reduced number of benzodiazepine binding sites in the hippocampus that cannot be accounted for by the loss of neurons in sclerotic areas (Savic et al., 1988; Koepp et al., 1997; Hand et al., 1997; Loup et al., 2000). Further analysis indicated that the affinity of the benzodiazepine binding sites changed in some areas, suggesting a biophysical change in the types of benzodiazepine-sensitive GABAA receptors. In addition, positron emission tomography (PET) scans of benzodiazepine binding sites have suggested that deficits in the number of GABAA receptors are a useful indicator of the epileptogenic focus prior to surgery (Ryvlin et al., 1998; Bouvard et al., 2005). These data indicate that forms of chronic epilepsies exhibit a loss of benzodiazepine-sensitive receptors. However, these cases do not necessarily shed light on SE in patients with different histories or induced refractory SE in animals.

Reduced Benzodiazepine Efficacy in Rodent Models

Pilocarpine-induced seizures result in a progressive change in electroencephalography (EEG) patterns that resemble those observed in humans experiencing generalized convulsive SE, although the patients in this study had various histories including some with epilepsy (Treiman et al., 1990; Treiman, 1990). The survival rate of rats given diazepam decreases if it is administered after pilocarpine-induced seizures compared to prior administration (Morrisett et al., 1987). Walton and Treiman (1988) then correlated temporal changes in the EEG patterns after seizure induction with the efficacy of the same first-time dosage of diazepam (20 mg/kg), thereby ruling out tolerance. Indeed, SE caused a sharp decline in diazepam efficacy between 10 and 15 min after the first seizure that worsened over the next 30–180 min. Similar studies have supported these findings (Kapur & Macdonald, 1997; Rice & DeLorenzo, 1999; Gao et al., 2007) and have demonstrated a rightward shift in the therapeutic potency of diazepam by an order of magnitude between 0 and 10 min after the first stage 3 seizure (Jones et al., 2002). These experiments clearly demonstrate that pilocarpine-induced SE in rats causes a progressive reduction in diazepam’s therapeutic efficacy.

Evidence of Altered GABAA Receptor Trafficking and Benzodiazepine Efficacy within an Hour of Seizure Induction

For this section we will discuss chemical-induced seizures in rodents at several time points after seizure induction (Löscher, 2002). An early demonstration of a reduced GABAergic efficacy was performed in freely moving rats. Systemic kainate injections ablated paired-pulse depression in the dentate gyrus, suggesting a failure of inhibition after 25 min of kainate-induced seizures (Milgram et al., 1991). In mechanically isolated CA1 neurons, a 45-min period of pilocarpine-induced SE reduced the potency of exogenous GABA applications (Kapur & Coulter, 1995). At the same time point, the modulatory efficacy of diazepam, but not pentobarbital, was also reduced in mechanically isolated dentate granule cells (Kapur & Macdonald, 1997). Recordings performed on dentate granule cells in slices obtained from seizing rats revealed several changes during the acute phase (Feng et al., 2008) (Table 1). These data were obtained from rats undergoing pilocarpine-induced seizures that were not terminated by benzodiazepines, but the slices were incubated for an hour for recovery. Immediately after the first stage 3 seizure (0 min group), the minimum inhibitory postsynaptic current (mIPSC) amplitudes were smaller and the decay times faster compared to sham controls, whereas the mIPSCs recorded from a separate group obtained 30 min after the first stage 3 seizure were of the same amplitude but had slower decay times compared to controls. The results from this study did not report any change in the baseline frequency of mIPSCs of either group compared to sham controls. mIPSCs from all three groups (sham, or 0 and 30 min after the first stage 3 seizure) were sensitive to modulation by diazepam, and the percent increase of the decay times from each group were not different compared to sham. However, the percent increase of the decay was significantly greater for the 0 min group when compared to the 30 min group. These data indicated that GABAA receptors of dentate granule cells exhibit two distinct phases that should contribute to the rapid reduction in the therapeutic efficacy of diazepam. The initial phase occurs after the first stage 3 seizure and entails a reduction of receptor number, whereas the next phase occurs over the next 30 min and involves a rebound in receptor number to control levels but with a reduction in the ability of diazepam to prolong the mIPSC, suggesting that a portion of synaptic receptors were internalized and replaced with a receptor subtype that was less sensitive to diazepam. It should be noted that measurements of mIPSC parameters do not necessarily reflect the amount of inhibition. Nevertheless, to the best of our knowledge, this is the only report that demonstrated a change in GABAA-mediated currents at these early time points in slices obtained from seizing animals.

Table 1.   Changes in GABAA receptor subunits and mIPSCs in dentate granule cells at various times after initiation of seizure activity
 0 min after Stage 330 min after Stage 360 min after SE onset
  1. NS, not statistically significant compared to controls; –, not tested.

  2. a Data were obtained from distinct populations of neurons (see text).

  3. b Data were obtained from two separate statistical comparisons from the same article (Feng et al., 2008).

GABAA R subunit
mIPSC parameter
 DecayFasterSlowerSlower, NS
 Charge transferLessGreaterLess
 FrequencyNSNSLess, NSa
 % Increase of decay by diazepamNSNS, <0 min groupbNS

Several groups then further investigated altered trafficking of GABAA subunits in slices obtained 1 h after the first stage 5 seizure, which is well into the development of diazepam resistance observed behaviorally, followed by dissection under halothane anesthesia and a 30–60 min period of slice incubation (Table 1). Immunohistochemical data indicated that the β2/3 and γ2 subunits were internalized (Naylor et al., 2005), and this was later supported by surface biotinylation assays in slices (Terunuma et al., 2008; Goodkin et al., 2008). The analysis also revealed that surface levels of α1, α2, and α4 subunits were significantly reduced, but that the α5 and δ subunits were increased (Terunuma et al., 2008). The mechanism by which the β3- and γ2-containing synaptic receptors were internalized was dependent on PKC phosphorylation of the β3 subunit and AP2-mediated clathrin-dependent endocytosis. The mIPSC amplitudes were also reduced in CA1 and dentate granule cells at the 1 h time point (Goodkin et al., 2008; Terunuma et al., 2008), but the ability of diazepam to prolong the decay of mIPSCs in dentate granule cells was not significantly different compared to controls (Naylor et al., 2005). Furthermore, SE decreased the frequency of mIPSCs in dentate granule cells (Naylor et al., 2005; Goodkin et al., 2008), but did not alter frequency in the CA1 (Terunuma et al., 2008). These data indicate that approximately 1 h of SE reduces the number of diazepam-sensitive synaptic GABAA receptors in two different populations of neurons with concurrent changes in specific extrasynaptic subtypes.

Evidence of Long-Term Altered GABAA Receptor Trafficking and Benzodiazepine Efficacy after Seizure Induction

In contrast to the data obtained in acute preparations, data obtained post-SE at 24 h and onward are more variable, possibly due to differences in the allowed duration of SE and whether or not spontaneous seizures were monitored just prior to tissue preparation. We will nevertheless attempt to review them in order to chronicle short- and long-term changes in GABAA receptors after chemically induced seizures. Mechanically isolated dentate granule neurons obtained 24 h after pilocarpine-induced SE exhibited reduced α1- and β1-subunit messenger RNA (mRNA) levels, whereas the α4-, δ-, and ɛ-subunit mRNA levels were increased (Brooks-Kayal et al., 1998). The change in the expression of GABAA receptor subunits was sustained during the chronic stage (1–4 months), and cells also exhibited a significant reduction in zolpidem sensitivity. In agreement, in situ hybridization analysis revealed that the mRNA level of the α1 subunit was also reduced in the CA3 and CA4 areas at the 24 h point (Friedman et al., 1994). In contrast, flumazenil autoradiographic analysis showed an increase in binding in all hippocampal areas at the 24 h point, which was followed by a decrease in some of these areas that was sustained into the chronic phase (Vivash et al., 2011). However, this study used low-dose kainate injections to induce SE that was terminated by diazepam after 4 h. Another pilocarpine study showed an increase of mIPSC amplitudes at the 24–48 h time point and 3–5 months later in dentate granule cells in slices (Leroy et al., 2004). Further analysis revealed that mIPSCs were less sensitive to diazepam modulation at the 24–48 h time point and that they eventually lost all sensitivity to diazepam 3–5 months later, but mIPSCs from both populations were sensitive to modulation by flumazenil. It should be noted that this study did not terminate SE with any antiseizure drugs.

During the latent period (6–8 days), mIPSC amplitudes in dentate granule cells were reduced, but at later stages, the mIPSCs had larger amplitudes and were insensitive to zolpidem (Cohen et al., 2003). The expression of the γ2 subunit was decreased but not significantly at 4 days before significantly increasing 60 days after SE (Peng et al., 2004). The α4 subunit was also decreased at 4 days before rising beyond control values at 30 days after SE. Further analysis revealed an apparent switch in expression of the δ subunit; it decreased in the dendrites of dentate granule cells but increased in the local interneuron population. These changes are associated with increased excitability, as assessed by medial perforant path stimulation.

In addition, in mechanically isolated neurons from chronic epileptic rats (∼6 weeks), clonazepam sensitivity increased in dentate granule cells but decreased in CA1 neurons, whereas zolpidem sensitivity decreased in dentate granule cells, suggesting a change in the relative expression of α subunits (Gibbs et al., 1997). The same study found increased current density values in dentate granule cells, but decreased values in CA1 neurons with increased GABA potency values. In situ hybridization analysis of mRNA levels during the chronic phase (2 months) indicated that the α2 and α5 transcripts were decreased throughout Ammon’s horn, but the expression of the α5 subunit increased in the granule cell layer of the dentate gyrus (Rice et al., 1996). This study also showed increased excitability in the CA1 region following Schaeffer collateral stimulation 2 months after SE induction. These studies provided evidence of synaptic and extrasynaptic GABAA receptor plasticity that lasted for prolonged periods after SE induction, which could further affect the therapeutic efficacy of benzodiazepines during SE development in patients with a history of epilepsy.

In Vitro Analysis of Rapid Alterations of GABAA Receptors

Several studies have also utilized hyperexcitable conditions in cultured neurons and organotypic slices to analyze the immediate effects of epileptiform activity. The zero-Mg2+ model produces benzodiazepine-resistant activity in cultured neurons (Sombati & Delorenzo, 1995; Deshpande et al., 2007). Exposures for longer than 10 min increased internalization rates of the β2/3 subunits utilizing an antibody feeding technique (Goodkin et al., 2005, 2007), with changes in the kinetic parameters and charge transfer values of mIPSCs that were reported for only the 2–3 h time points. Changes in mIPSC properties upon exposure to zero-Mg2+ were stated to occur between 10 and 60 min; however, no values were actually reported (Goodkin et al., 2007). It should also be noted that the antibody feeding technique does not account for changes in the insertion rate of GABAA receptors. Leaving open the possibility that zero-Mg2+ increased the overall turnover rate without altering the actual number of surface receptors, which is why the lack of reported functional values at the earliest time points is critical. In contrast, it is not known if there are changes in GABAA receptor trafficking in the zero-Mg2+ or 4-AP models of benzodiazepine resistance in the organotypic or acute-slice preparation (Albus et al., 2008; Wahab et al., 2010). However, exposure to high [K]o + N-methyl-D-aspartate (NMDA) for 1 h decreased the surface levels of the γ2 subunit but did not alter the δ subunit in organotypic slices (Goodkin et al., 2008), although it is unclear if this treatment protocol produces benzodiazepine-resistant epileptiform activity.

Several studies have utilized single particle tracking methods to analyze how GABAA receptors behave under hyperexcitable conditions in cultured neurons. Treatment with 4-AP increased the lateral mobility of endogenous GABAA receptors after just 3 min and decreased the immunolabeled size of GABAA clusters (Bannai et al., 2009). These data were supported by an electrophysiologic analysis that utilized direct current injection protocols to generate hyperexcitability, which decreased the amplitudes of mIPSCs. It is unclear if these conditions produce diazepam-resistant activity, although one could predict that such effects would reduce diazepam’s inhibitory efficacy. In contrast, 4-AP did not increase the mobility of transfected γ2-tagged subunits at a lower temperature of 29°C (Bouthour et al., 2012). It is likely that both the transfection and lower temperature could account for this small discrepancy. Indeed, another group found that 4-AP treatment increased the lateral diffusion of γ2-containing receptors within 2.5 min, which correlated with a reduction in immunolabeled GABAA clusters (Niwa et al., 2012). Tracking experiments utilizing glutamate or NMDA applications also demonstrated that GABAA receptors become more mobile within 10 min (Muir et al., 2010; Niwa et al., 2012). Again, it is not evident that these conditions cause a reduction in the inhibitory efficacy of diazepam. Nevertheless, these single-particle tracking studies suggest that GABAA receptors exhibit increased migratory behavior during hyperexcitable states, thereby raising the likelihood that receptors reach an endocytic zone and are internalized (Smith et al., 2012). Future experiments should reveal the precise temporal correlations between the onset of diazepam-resistant epileptiform activity with reductions in the function of GABAergic synapses and GABAA receptor trafficking in cultured neurons, and acute and organotypic slice preparations. Such investigations would support the therapeutic value of targeting the mechanisms that regulate the molecular determinants of the formation, size, and stability of synaptic GABAA receptor clusters (Saiepour et al., 2010; Papadopoulos & Soykan, 2011; Mukherjee et al., 2011).

Chloride Homeostasis and Refractory Seizures

Antiseizure compounds such as benzodiazepines, barbiturates, and propofol potentiate GABAA receptor activity. However, GABAA receptors have a dual role in neurons. As discussed above, GABAA receptors can either depolarize or hyperpolarize a neuron, which is well documented during the course of neuronal maturation. As mentioned earlier, the canonical inhibitory hyperpolarizing effect of GABAA-mediated currents is largely determined by Cl homeostasis and KCC2 function. If KCC2 activity is reduced and Cl homeostasis perturbed, the efficacy of these antiseizure drugs would also theoretically be altered. In fact, it has been demonstrated that the Cl gradient has a significant impact on the efficacy of GABAA modulators (Staley, 1992). To the best of our knowledge, this is the only article to date that has explored this relationship. However, this investigation utilized the whole-cell patch-clamp technique, which dictates the reversal potential of GABAA-mediated currents (EGABA) and therefore did not investigate any changes in endogenous Cl gradients under pathophysiologic conditions. It should also be noted that in this study high intracellular Cl concentrations reduced the inhibitory efficacy of the benzodiazepine flunitrazepam and the barbiturate pentobarbital. Although refractory SE is by definition poorly responsive to benzodiazepines, it is sensitive to the anesthetics pentobarbital and propofol (Shorvon, 2011). Nevertheless, this discovery could be vitally important to the mechanisms underlying changes in the efficacy of benzodiazepines during SE. Curiously, no single investigation has yet attempted to demonstrate that altered Cl homeostasis contributes to the reduced inhibitory or therapeutic efficacy of benzodiazepines under pathophysiologic conditions, although our research groups are currently undertaking this task.

We briefly discuss some of the articles on altered cation-chloride cotransporters in patients with chronic epilepsy, although several comprehensive reviews have already summarized this literature (Payne et al., 2003; Galanopoulou, 2007; Kahle et al., 2008; Blaesse et al., 2009; Löscher et al., 2012). KCC2 function was found to be decreased in patients with temporal lobe epilepsy (Palma et al., 2006; Munoz et al., 2007; Eichler et al., 2008). Subsets of neurons in the subiculum of resected human epileptic brain tissue exhibit depolarizing GABA responses (Cohen et al., 2002), which is attributable to decreased KCC2 expression (Huberfeld et al., 2007). Further analysis revealed that the type 1 Na+/K+/Cl cotransporter, NKCC1, inhibitor bumetanide blocked the interictal epileptiform activity. This suggests that perturbed Cl homeostasis directly contributed to interictal events, although the pathophysiologic role of these events was unclear in these experiments (Staley et al., 2005; Avoli et al., 2006). Although not investigated directly on these particular tissue samples, these forms of epilepsy are generally refractory to benzodiazepines. So it is possible that the altered Cl homeostasis contributes to the reduced therapeutic efficacy of these compounds in the chronic phase. Even though patients with a history of epilepsy can develop refractory SE (Treiman et al., 1990), changes in Cl homeostasis might occur at later stages, and therefore these data do not shed light on the initial loss of diazepam efficacy observed during chemically induced refractory seizures in animals or humans.

There is evidence in rodents of altered cation-chloride cotransporter expression and/or perturbed Cl homeostasis after chemically induced seizures in adults (Löscher et al., 2012). To the best of our knowledge, the first study that demonstrated a positive shift in EGABA after SE induction was performed by Kapur and Coulter (1996). This study used the whole-cell patch configuration and was performed on mechanically isolated CA1 neurons after 45 min of SE. To date, this is the earliest time point after SE induction that EGABA values have been characterized. Pilocarpine-induced SE that proceeded for an hour after the first stage 5 seizure, caused a marked reduction in KCC2 surface and total levels in the hippocampus after 1 h of seizures, with a concurrent enhancement of tyrosine phosphorylation of KCC2 residues Y903 and Y1087 (Lee et al., 2010). Blocking of all tyrosine-dependent phosphorylation in cultured neurons decreased KCC2 surface activity and clustering without altering the surface or total protein levels (Watanabe et al., 2009), indicating that dynamic regulation of functional activity at the cell surface is important. Indeed, KCC2 and NKCC1 are dynamically regulated by phosphorylation. These processes can have rapid effects on Cl homeostasis, which can be either transient or sustained and are potentially more relevant than total or even surface protein levels, thereby adding an extra degree of complexity for this system (for review see Kahle et al., 2010; Chamma et al., 2012).

In the chemically induced models of SE, there is some disagreement in the literature at time points greater than 1 h after SE induction. Kainate-induced SE in mice caused a marked increase in KCC2 total protein levels at 2, 6, and 24 h prior to returning to basal levels at the 48-h time point in the hippocampus (Shin et al., 2012). This study also found a small decrease in the expression of NKCC1 at 6 h. Similarly, pilocarpine-induced SE caused a transient up-regulation of KCC2 mRNA levels at the 6-h time point (Zhu et al., 2012). In sharp contrast, several studies indicated a shift in the Cl homeostatic mechanism that would favor a depolarizing shift in EGABA values, although some measurements were obtained at different time points. Pilocarpine-induced SE resulted in an increase in NKCC1 protein expression in the subiculum and dentate gyrus at the 24-h time point (Brandt et al., 2010). Pilocarpine-induced SE depolarized EGABA values in dentate granule cells at the 24-h and 1-week time points, with a concurrent reduction in KCC2 total protein (Pathak et al., 2007). Similarly, pilocarpine produced a positive shift in EGABA values in dentate granule cells, subiculum neurons, and CA1 neurons between 7 and 14 days after SE induction, with a concurrent reduction in the ratio of mRNA expression of KCC2 relative to NKCC1 (Barmashenko et al., 2011). Pilocarpine-induced SE caused a marked increase in NKCC1 and a decrease in KCC2 protein and mRNA levels at the 1-, 14-, and 45-day time points in the CA1 region, although the reduction in KCC2 protein levels were only significant at 14 and 45 days (Li et al., 2008). Similarly, pilocarpine-induced SE increased NKCC1 and decreased KCC2 protein and mRNA levels at the 2- and 3-week time points in the entorhinal cortex (Bragin et al., 2009). Pilocarpine also caused a reduction of KCC2 mRNA levels and depolarized EGABA values in the subiculum at the 4-month time point (De Guzman et al., 2006). In summary, the bulk of data supports a reduction in the Cl extrusion capacity in multiple subregions of the hippocampus and associated areas beginning at 1 h after SE induction that lasted weeks. Future studies of Cl homeostasis will need to entail a more detailed analysis of the biochemical properties of the cation-chloride cotransporters at time points during the earliest observable seizures in order to test if Cl homeostasis has an impact, if any at all, on the progressive reduction of diazepam’s efficacy during SE.

From a technical standpoint, protocols usually have a 1-h incubation period to allow the brain slices to recover from the acute slice preparation; it is therefore possible that short term, virtually instantaneous changes in EGABA are missed. Chloride plasticity, or fluctuations in EGABA, following intense neuronal activity can occur even in the presence of functional KCC2, and these alterations can recover quickly within minutes to baseline values if the stimulation or insult is terminated (Kaila, 1994; Rivera et al., 2005; Buzsáki et al., 2007). Therefore, in vitro studies on neurons and brain tissue have the potential to yield far greater detail in terms of the temporal correlation between the inhibitory efficacy of benzodiazepines and changes in EGABA or the biochemistry of the Cl homeostatic mechanism. To date, several models of pharmacoresistant epileptiform activity have been established (Wahab et al., 2010). For example, prolonged exposure to Mg2+-free solution caused the development of late recurrent discharges that were resistant to a number of antiseizure drugs including benzodiazepines in acute hippocampal-entorhinal brain slices (Zhang et al., 1995; Drier et al., 1998) and organotypic slices (Albus et al., 2008). Of interest, zero-Mg2+ protocols caused a reduction in KCC2 protein expression and Cl extrusion capacity in acute slice preparations (Rivera et al., 2004) and a depolarizing shift of EGABA in organotypic slice preparations (Ilie et al., 2012). The latest evidence suggests that the molecular underpinnings of these zero-Mg2+–induced events involve calpain-dependent degradation of KCC2 with a concurrent reduction in Cl extrusion capacity (Puskarjov et al., 2012). In terms of the hypothesized role of Cl homeostasis, this molecular mechanism could also explain the refractoriness of resected human brain tissue, which exhibited increased levels of calpain activity (Feng et al., 2011; Das et al., 2012). Future in vitro analyses must perform a more detailed analysis of the potential influence of Cl homeostasis on diazepam-resistant activity.

Concluding Remarks

The bulk of the research on the development of benzodiazepine-resistant seizures has focused on changes in GABAA receptors in both humans and animals. However, we believe that Cl homeostasis may play a significant role in the development of diazepam-resistant seizures. Further research on this possibility is imperative to at least rule it out. Furthermore, changes in GABAA receptor trafficking and perturbed Cl homeostasis are not mutually exclusive. It is possible that both aspects contribute to reduce diazepam’s efficacy, and experimentally, both are certainly capable. Future work will undoubtedly shed more light on the role of GABAA receptors and Cl homeostasis in this widespread and costly debilitating disorder.


This work was supported in part by U.S. National Institutes of Health/National Institute of Neurological Disorders and Stroke grant NS036296 awarded to SJM; JM is funded by NS073574.


None of the authors has any conflict of interest to disclose.

We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.