Address correspondence to Howard P. Goodkin, Department of Neurology, University of Virginia, PO Box 800394, Charlottesville, VA 22908, U.S.A. E-mail: firstname.lastname@example.org
The fortuitous discovery of the benzodiazepines and the subsequent application of these agents to the treatment of status epilepticus (SE) heralds in the modern age of treating this neurologic emergency. More than 50 years after their discovery, the benzodiazepines remain the drugs of first choice in the treatment of SE. However, the benzodiazepines can be ineffective, especially in those patients whose seizures are the most prolonged. The benzodiazepines act by increasing the affinity of γ-aminobutyric acid (GABA) for GABAA receptors. A receptor’s subunit composition affects its functional and pharmacologic properties, trafficking, and cellular localization. The GABAA receptors that mediate synaptic inhibition typically contain a γ2 subunit and are diazepam-sensitive. Among the GABAA receptors that mediate tonic inhibition are the benzodiazepine-insensitive δ subunit–containing receptors. The initial studies investigating the pathogenesis of SE demonstrated that a reduction in GABA-mediated inhibition within the hippocampus was important in maintenance of SE, and this reduction correlated with a rapid modification in the postsynaptic GABAA receptor population expressed on the surface of the hippocampal principal neurons. Subsequent studies found that this rapid modification is, in part, mediated by an activity-dependent, subunit-specific trafficking of the receptors that resulted in the reduction in the surface expression of the benzodiazepine-sensitive γ2 subunit–containing receptors and the preserved surface expression of the benzodiazepine-insensitive δ subunit-containing receptors. This improved understanding of the changes in the trafficking of GABAA receptors during SE partially accounts for the development of benzodiazepine-pharmacoresistance and has implications for the current and future treatment of benzodiazepine-refractory SE.
The treatments of choice for SE during the first half of the twentieth century included paraldehyde (Whitty & Taylor, 1949; Lennox & Lennox, 1960), phenobarbital (PB), and sodium amytal (Sciarra, 1957). Although effective, all three agents, as noted by Lennox, have the potential of leaving the patient “so thoroughly drugged that vital processes and consciousness are re-established only with difficulty” (Lennox & Lennox, 1960).
The modern era of treating SE began soon after Dr. Leo Sternbach’s (Fig. 1) and Mr. Earl Reeder’s (Fig. 2) serendipitous discovery of the 1,4-benzodiazepines (BDZ) during the late 1950s followed by their synthesis of diazepam (DZP) in 1959 (Sternbach, 1978). Soon thereafter, the utility of DZP in the treatment of SE was established first in abstract form (Naquet et al., 1965) and in several publications including Gastaut et al. (1965) and Lombroso (1966). By 1969, the Medical Letter stated “Valium is the preferred drug for the treatment of recurrent prolonged seizures.” As they are effective against a variety of seizure types, have a rapid onset of action upon entering the brain, and are relatively safe, the BDZs remain the drugs of first choice in the treatment of SE (Kapur, 2002).
However, despite their efficacy in the early stages of SE, the BDZs often fail to control SE in its later stages in both adults (Treiman et al., 1998) and children (Eriksson et al., 2005; Raspall-Chaure et al., 2006; Chin et al., 2008). In the Veterans Affairs Cooperative Study (Treiman et al., 1998), the BDZs were largely ineffective in the treatment of the more prolonged subtle SE, controlling the seizure in less than 20% of these patients compared to >55% success in the treatment of the shorter overt SE. In a retrospective analysis of patients in SE of at least 10 min in duration in a large academic hospital, it was found that 31% of seizures were refractory to a combination of a BDZ and one of phenytoin (PHT), fosphenytoin, or PB (Mayer et al., 2002).
Walton and Treiman (1988) were the first to demonstrate the development of pharmacoresistance to BDZ in an experimental model of SE. DZP administered shortly after onset stopped all seizures produced by the combination of lithium and pilocarpine. However, it was effective in only 17% of animals after prolonged SE. A subsequent study demonstrated that there is a substantial reduction of BDZ potency for seizure termination with the passage of time (Kapur & Macdonald, 1997). More recently, Jones et al. (2002) demonstrated that the development of BDZ pharmacoresistance occurs rapidly after the onset of forelimb clonus and ictal spike-wave activity, and Goodkin et al. (2003) demonstrated that BDZ pharmacoresistance develops in young naive rats.
Therefore, although the BDZs have been the treatment of choice for more than 50 years, new therapies for the treatment of SE are required. The growing recognition of BDZ-refractory SE has spurred multiple investigations into the mechanism underlying BDZ-pharmacoresistance. An improved understanding of this mechanism has implications not only for the current treatment of SE but also for the development of future therapies and strategies.
Benzodiazepine Modulation of GABARs
DZP and the other BDZs act on postsynaptic γ-aminobutyric acid (GABA)A receptors (GABARs) to enhance their function (Choi et al., 1977; Macdonald & Barker, 1978). The BDZs bind to GABARs at a unique allosteric site. BDZ-site agonists (e.g., DZP) and BDZ-site inverse agonists (e.g., Ro 15-4513) are capable of positive and negative modulation of GABA’s action, respectively (Mohler et al., 2000). BDZ-site antagonists (e.g., flumazenil) prevent action of both agonists and inverse agonists. BDZ-site agonists prolong GABAR-mediated synaptic currents (Otis & Mody, 1992). In single-channel recordings, in the presence of DZP, GABA evoked more frequent channel openings without altering mean open channel times. These findings were interpreted to suggest that DZP increases the apparent association rate of GABA to GABARs (Rogers et al., 1994).
Cloning of genes for ligand-gated ion channels has resulted in the identification of numerous GABAR subunits (Schofield et al., 1987). The genes for the GABAR subunits are divided into families based on sequence homology and have been named α, β, γ, δ, ε, π, θ, and ρ subunits. Within several of the subunit families, multiple isoforms (e.g., α1-6, β1-3, and γ1-3) and splice variants (e.g., γ2s and γ2l) have been identified. Each subunit has a similar transmembrane topology, and there is a large (approximately 200 amino acid) extracellular N-terminal domain that forms the GABA binding site. Functional receptors are formed by the coassembly of five subunits around the central channel axis (Nayeem et al., 1994). The subunits form a quasi-symmetric structure around the ion channel, with each subunit contributing to the wall of the channel. The model is based on the structure of the nicotinic acetylcholine receptor, another member of the ligand-gated ion channel gene superfamily, and electron microscopic image analysis of native GABARs (Akabas, 2004). It is currently believed that the majority of GABARs are composed of 2 α subunits, 2 β subunits, and one γ2 or δ subunit. Forced assembly experiments suggest that the subunits are arranged in a specific order, with the γ2 subunit interposed between an α and a β subunit, and that these interface with the other α and β subunits (Baumann et al., 2002).
The BDZ sensitivity of GABARs is dependent on their subunit composition. The presence of the γ2 subunit is a requirement for GABAR BDZ sensitivity (Sieghart & Sperk, 2002). The BDZ binding pocket is at the α1, α2, or α5 and γ2 subunit interface in a region homologous to the GABA binding site at the interface of the α and β subunits (Sigel, 2002). A specific domain in the γ2 subunit (F-loop) appears to be involved in the transduction of BDZ binding to channel function (Hanson & Czajkowski, 2008). The γ2 subunit can be substituted by the δ subunit, and receptors containing this subunit are BDZ insensitive (Saxena & Macdonald, 1994, 1996). The δ subunit commonly assembles with the α4 or the α6 subunit, and both of these also render GABARs BDZ insensitive, even when combined with the γ2 subunit (Knoflach et al., 1996; Wafford et al., 1996).
Multiple specific GABAR isoforms are expressed on the surface of hippocampal dentate granule cells (DGCs). In situ hybridization and immunohistochemistry studies have demonstrated that DGCs express mRNAs for α1, α2, α4, β1, β3, γ1, γ2, and δ GABAR subunit subtypes (Wisden et al., 1992; Sperk et al., 1997). Sun et al. (2004) used confocal laser scanning microscopy to determine the surface distribution of the DZP-sensitive (α1 and γ2) and DZP-insensitive (α4 and δ) subunits on the surface of DGCs. Brain sections containing the hippocampus were immunohistochemically double-labeled for the α1, α4, γ2, or δ subunits and glutamate decarboxylase 65 (GAD65) or gephyrin. The location, synaptic versus extrasynaptic, of these subunits was inferred by quantitative analysis of the frequency of colocalization of the various subunits with synaptic markers in high-resolution images. GAD65 immunoreactive clusters colocalized with 26% of the α1 subunit immunoreactive clusters and 32% of the γ2 subunit clusters. In contrast, only 2% of the α4 subunit immunoreactive clusters and 2% of the δ subunit clusters colocalized with the presynaptic marker GAD65. These findings were confirmed by studying colocalization with immunoreactivity of a postsynaptic marker, gephyrin, which colocalized with 28% of the α1 subunit immunoreactive clusters and 23% of the γ2 subunit immunoreactive clusters. In contrast, only 2% of the α4 subunit immunoreactive clusters and 2% of the δ subunit clusters colocalized with gephyrin. This study demonstrated that DZP-sensitive (α1 and γ2 subunit–containing) receptors are present at the synapse, whereas DZP-insensitive (α4 and δ subunit–containing) receptors are present in the extrasynaptic locations. Postembedding electron microscopy studies have also demonstrated that GABARs containing the δ subunit are expressed perisynaptically on hippocampal DGCs (Wei et al., 2003). Other postembedding electron microscopy studies demonstrate that on DGCs, the γ2 subunit is present at synapses and in the extrasynaptic membrane and the α4 subunit is present in the perisynaptic membrane (Sun et al., 2007; Zhang et al., 2007). The functional significance of these extrasynaptic receptors is discussed in subsequent text.
Impact of SE on GABAergic Synaptic Transmission. A Reduction in the Surface Expression of DZP-sensitive GABARs as a Potential Mechanism to Explain the Development of BDZ-Pharmacoresistance
The synaptically located DZP-sensitive GABARs mediate fast synaptic phasic inhibition. Synaptic GABARs containing the γ2 subunit have a low affinity for GABA and thus require high concentrations of GABA for activation. Analysis of the kinetics of synaptic currents and rapid application experiments suggests that presynaptic terminals release vesicles containing high concentrations of GABA. Synaptic GABARs open rapidly after GABA binding, followed by desensitization over a period of time, followed by closure and the dissociation of GABA (Maconochie et al., 1994; Jones & Westbrook, 1996; Bianchi & Macdonald, 2001). The entire sequence of events from activation to deactivation lasts less than 300 ms, resulting in periodic or “phasic” inhibition of the postsynaptic neuron.
Kapur and Macdonald (1997) demonstrated that the reduction in the potency of DZP during SE occurs at the level of a single principal neuron. Whole-cell GABAR currents were obtained from DGCs isolated acutely from the hippocampus of control rats and rats undergoing 45 min of SE. In DGCs from control animals, when 10 μM GABA was coapplied with 300 nM DZP, GABAR currents were markedly enhanced in all neurons. In contrast, in DGCs from animals undergoing SE, 300-nM DZP inconsistently enhanced 6- or 10-μM GABA-evoked GABAR currents. DZP concentration–response curves were obtained for enhancement of GABAR currents from neurons from both the naive and SE-treated animals. In neurons from control animals, 1 μM diazepam enhanced GABAR currents by 92%, but in neurons from animals undergoing SE, 3 μM DZP enhanced currents by only 52%. The effective concentration (EC50) for DZP enhancement of GABAR currents in neurons from control animals was 195 nM, and that in neurons from animals undergoing SE was 4.4 μM. Therefore, the prolonged seizures of SE reduced the potency and efficacy of DZP for enhancement of DGC GABAR currents. This study suggested that a rapid alteration in functional properties of the GABARs expressed on the surface of the DGCs had occurred during SE.
A subsequent study performed a more detailed analysis of the effect of SE on drug modulation of synaptic GABARs on DGCs (Feng et al., 2008). Animals were studied immediately after a grade 3 seizure of the Racine scale or 30 min after the first grade 3 seizure. These time-points were chosen because a previous study demonstrated that resistance to DZP termination of SE occurs within minutes of the first grade 3 seizure (Jones et al., 2002). Compared to controls, the amplitude and charge transfer of action potential–independent inhibitory postsynaptic currents (mIPSC) were reduced in animals studied immediately after the first stage 3 seizure. However, these reductions in synaptic currents were restored when animals were studied 30 min after the first stage 3 seizure. The mIPSCs recorded from animals immediately after the first grade 3 seizure were less sensitive to DZP and zinc modulation than those from sham controls and animals studied 30 min after first stage 3 seizures. The authors concluded that there were substantial plastic changes in synaptic GABAR function and their allosteric modulation during SE.
Other studies (see subsequent text) investigating the impact of SE on GABAergic synaptic transmission reached a similar conclusion that the population of synaptic GABARs is modified during SE, and extended the findings reviewed in the preceding text by demonstrating that the rapid modification in synaptic GABARs is partially the result of activity-dependent, subunit-specific trafficking of GABARs during SE.
Goodkin et al. (2005) studied the acute effects of prolonged epileptiform bursting on GABA-mediated synaptic transmission in a network of cultured hippocampal neurons. In this in vitro model of SE, the period of prolonged epileptiform bursting resulted in an approximately 35% reduction in the mIPSC amplitude.
Using an antibody-feeding technique, it was demonstrated that constitutive internalization of β2/3 subunit–containing GABARs was rapid, with 50% of the trafficked receptors internalized in less than approximately 11 min. Under the condition of prolonged epileptiform bursting, the percentage of receptors in traffic increased, and 50% of the receptors in traffic were internalized within approximately 7 min. Given that the number of receptors at the synapse is an important determinant of mIPSC amplitude (Edwards et al., 1990; Nusser et al., 1998) and the finding that the rate of GABAR intracellular accumulation correlated with neuronal activity, these authors posited that activity-dependent trafficking was a potential mechanism to explain the reduction in GABA-mediated inhibition and the development of BDZ-pharmacoresistance that occur during SE.
Naylor et al. (2005) confirmed these findings using an in vivo model of SE. They recorded mIPSCs from the DGCs of animals in lithium/pilocarpine–induced SE of 1 h in duration. The amplitude of mIPSCs recorded from DGCs of animals in SE was approximately 30% smaller than that recorded from controls. DZP enhances mIPSCs in DGCs by prolonging their decays (Mody et al., 1994). In this study DZP enhancement of mIPSCs was evident in DGCs from SE-treated and control animals. However, DZP did not restore total charge transferred during mIPSC (a measure of inhibition) in SE animals to the level observed in controls. The mIPSC data were fitted to a multistate model of synaptic GABARs to derive the number of GABARs expressed per DGC inhibitory synapse in the dentate gyrus of the control and SE-treated animals. The authors suggested that during SE the number of receptors expressed at the synapse had declined to 18 receptors per synapse compared to the 36 receptors per synapse in control animals. Using immunocytochemical methods, they were also able to demonstrate a redistribution of GABARs, which contained either a β2/3 or γ2 subunit to the intracellular compartment.
In two subsequent studies (Goodkin et al., 2008; Terunuma et al., 2008), a biotinylation pull-down assay was used to demonstrate that the increase in the intracellular accumulation of these GABARs had the net effect of reducing the surface expression of BDZ-sensitive GABARs containing the γ2 subunit. Modulation of the surface expression of the BDZ-sensitive γ2 subunit–containing GABARs could occur at single or multiple steps in the GABAR trafficking pathway (Michels & Moss, 2007). Synaptic GABARs are endocytosed rapidly via a clathrin-dependent mechanism (Tehrani & Barnes, 1993;Tehrani et al., 1997). In order to be endocytosed, the receptors associate with the adaptin AP2 protein complex (Kittler et al., 2000). As reviewed previously in this article, synaptic GABARs endocytose at a rapid pace (Goodkin et al., 2005). In their study, Terunuma et al. (2008) were able to demonstrate that the reduction in the surface expression of GABARs containing the β3 subunits that occurs during SE was due to a deficit in phosphorylation of this GABAR subunit at serine residues 408/9, which resulted in an unmasking of a patch-binding motif for the clathrin adaptor AP2 and the endocytosis of the receptors containing this subunit.
During SE, the signaling cascade that results in a reduction in the surface expression of the BDZ-sensitive γ2 subunit–containing GABARs could be initiated by a ligand-independent process such as an increase in neuronal excitability as the result of stimulation of excitatory amino acid receptors (Stelzer et al., 1987; Kapur & Lothman, 1990) or initiated by excessive extracellular GABA (ligand-dependent). In one study (Goodkin et al., 2008), a combination of high external potassium and N-methyl-d-aspartate (NMDA) receptor activation was sufficient to reduce the surface expression of γ2 subunit–containing GABARs in an organotypical hippocampal culture model, consistent with a ligand-independent process. Although NMDA-receptor activation was necessary to induce the reduction, NMDA receptor activation was not sufficient to induce a statistically significant decrease in the surface expression of the γ2 subunits. This finding suggests that other changes in neuronal excitability induced by the high external concentration of potassium were required for the reduction in the surface expression of this subunit.
In contrast, when dissociated cultures and organotypic cultures were incubated in excessive extracellular GABA, the surface expression of the γ2 subunit was unchanged. Studies in which the organotypic cultures were exposed to either excessive extracellular GABA concentration with GABA uptake blocked or to the high affinity GABAR agonist muscimol also demonstrated that modulation of the surface expression of the γ2 subunit was not dependent on direct ligand binding of the receptor. Although previous studies have demonstrated a ligand-dependent mechanism for the regulation of the surface expression of GABARs, these studies used longer incubation periods and higher concentrations of GABA (Friedman et al., 1996; Tehrani et al., 1997).
Taken together these studies performed in different laboratories using multiple models of SE reached remarkably similar conclusions: GABAergic synaptic inhibition is diminished during SE, and this reduction relates to a reduction in the surface expression of BDZ-sensitive γ2 subunit–containing GABARs.
Effect of SE on DZP-insensitive GABARs
In addition to the γ2 subunit, DGCs express the α4 and δ subunit, and as explained previously, these subunits are exclusively expressed in the extrasynaptic space. Studies of recombinant GABARs constituted of these subunits have demonstrated that receptors composed of these subunits have a high affinity for GABA, desensitize slowly, and are DZP-insensitive (Saxena & Macdonald, 1994, 1996; Haas & Macdonald, 1999; Brown et al., 2002). This unique set of properties enables α4 and δ subunit–containing receptors to mediate a novel nonsynaptic form of DZP-insensitive GABAergic inhibition, commonly referred to as tonic inhibition (Stell & Mody, 2002; Mtchedlishvili & Kapur, 2006).
The impact of SE on the tonic inhibition of DGCs has also been investigated. In one study (Naylor et al., 2005), the amplitude of tonic inhibition recorded from DGCs of animals that had undergone SE was larger in amplitude than that recorded from controls. The authors modeled their results and suggested that this increase in tonic inhibition was due to increased GABA in the extracellular space. Another study (Goodkin et al., 2008) further confirmed this finding that the tonic inhibition of DGCs measured in the presence of GABA uptake blockers is enhanced following SE.
Using the biotinylation pull-down assay, Terunuma et al. (2008) and Goodkin et al. (2008) were able to demonstrate that the surface expression of the GABAR δ subunit was not decreased in hippocampal slices acutely obtained from animals in SE compared to hippocampal slices obtained from naive animals. The findings from both of these studies demonstrated that the activity-dependent trafficking of the GABARs is subunit-specific, resulting in a decrease in the surface expression of the BDZ-sensitive γ2 subunit–containing receptors and preservation of the extrasynaptic BDZ-insensitive δ subunit–containing receptors.
The Treatment Implications of Activity-Dependent Subunit-Specific Trafficking of GABARs During SE
There has been increasing recognition that the treatment of SE should commence quickly (Lowenstein et al., 1999; Treiman, 2008). However, treatment is often delayed. Pellock et al. (2004) found that only 41% of prolonged seizures in their database had been treated within 30 min onset. In the recently published description of the first 119 subjects enrolled in the FEBSTAT trial (Shinnar et al., 2008), the prospective, multicenter study of the consequences of febrile SE in children, the median duration of febrile SE was 68.0 min, with greater than 20% of the episode lasting more than 2 h. Surprisingly, the physicians caring for these children frequently underestimated the SE duration or, in more than one-third of the cases, failed to recognize the seizure as SE.
When treatment is delayed, the reduction in the surface expression of the BDZ-sensitive γ2 subunit–containing GABARs is likely one factor that contributes to the reduced efficacy of the BDZs in terminating the more prolonged episodes of SE. In addition to providing an explanation for the development of BDZ-pharmacoresistance, activity-dependent, subunit-specific trafficking of GABARs during SE also provides a rational for newer treatment strategies and the development of new therapies. The BDZ-insensitive δ subunit–containing receptors are sensitive to general anesthetics such as propofol and pentobarbital (Lees & Edwards, 1998; Feng & Macdonald, 2004; Feng et al., 2004) as well as the natural (e.g., allopregnalone) and the artificial (e.g., ganaxolone) neurosteroids. Therefore, the finding that the surface expression of these GABARs is preserved during SE suggests that once BDZs have failed that prompt treatment with a general anesthetic is required or that the neurosteroids represent a future potential treatment of BDZ-refractory SE (Kokate et al., 1994, 1996). Other potential future targets for therapy of BDZ-refractory SE would be blocking the trafficking of the BDZ-sensitive GABARs, through a mechanism that increases the fraction of phosphorylated GABARs or that inhibits the interaction of dephosphorylated GABARS with the endocytic machinery.
However, until such agents can be developed, other potential targets that contribute to the self-sustaining nature of SE need to be considered including the reduction in the surface expression of Kv4.2 (Lugo et al., 2008) and alterations in neuropeptide expression (Liu et al., 1999). In addition, the role of the excitatory neurotransmitter system and trafficking of these receptors during SE needs to be more thoroughly investigated. There is likely to be increased activation of these receptors during SE. As reviewed previously, activation of these receptors and repetitive neuronal activation contributes to the reduction in the surface expression of the BDZ-sensitive GABARS via a ligand (e.g., GABA) independent mechanism. Martin and Kapur (2008) tested whether the combination of an NMDA antagonist and DZP would be useful in treating lithium/pilocarpine–induced SE at a time-point that BDZ-refractoriness was well-established. They found that when used in combination, these two agents had a synergistic effect in the treatment of prolonged SE. What remains to be determined is whether this synergistic effect is partially the result of an increase in the surface expression of the BDZ-sensitive receptors or occurs via another mechanism.
The discovery of the BDZs was fortuitous. However, it has resulted in a dramatic improvement in how SE is treated and in our understanding of the pathogenesis of SE. This improved understanding of the changes in the surface expression of GABARs during SE can have an immediate impact on how we treat SE today and serves as a foundation for developing novel therapies and strategies in the treatment of SE.
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.
Disclosures: HPG and JK receive support from the National Institutes of Health (NIH). HPG serves as a consultant to MedImmune, Inc.