Role of GABA_A receptor subtypes in the behavioural effects of intravenous general anaesthetics

Abstract

Since the introduction of general anaesthetics into clinical practice, researchers have been mystified as to how these chemically disparate drugs act to produce their dramatic effects on central nervous system function and behaviour. Scientific advances, particularly during the last 25 years, have now begun to reveal the molecular mechanisms underpinning their behavioural effects. For certain i.v. general anaesthetics, such as etomidate and propofol, a persuasive case can now be made that the GABA_A receptor, a major inhibitory receptor in the mammalian central nervous system, is an important target. Advances in molecular pharmacology and in genetic manipulation of rodent genes reveal that different subtypes of the GABA_A receptor are responsible for mediating particular aspects of the anaesthetic behavioural repertoire. Such studies provide a better understanding of the neuronal circuitry involved in the various anaesthetic-induced behaviours and, in the future, may result in the development of novel therapeutics with a reduced propensity for side-effects.

The discovery over 170 years ago of agents that, when administered, produce a state of reversible unconsciousness revolutionized surgery and remains one of the most important medical innovations. Although during the last century the clinical use of general anaesthetics became widespread, the molecular mechanism(s) whereby they produce their remarkable repertoire of behavioural effects, which include sedation, immobility, amnesia, and unconsciousness, remained a mystery. Agents capable of inducing a state of general anaesthesia are chemically diverse (Fig. 1). This chemical diversity appeared to preclude involvement of a common molecular anaesthetic target, such as a classical drug–receptor interaction, as no obvious structure–activity relationship was evident.^1–5 Such diversity led early theories of general anaesthetic activity to focus on non-specific interactions within the central nervous system (CNS). More than a century ago, Meyer and Overton reported a striking correlation between the oil:water partition coefficients of a range of anaesthetic compounds and their ability to immobilize tadpoles, i.e. the more lipid soluble the compound, the greater is its anaesthetic potency. Based on these observations, it was subsequently proposed that general anaesthetics perturb lipid bilayers to induce ‘non-specific’ disruption of neuronal activity.⁶ As exceptions to this correlation began to emerge, the concept that particular proteins might represent viable anaesthetic targets gained traction.

An important demonstration that proteins had the potential to be relevant anaesthetic targets resulted from studies by Franks and Lieb.⁷ They demonstrated that the activity of a soluble protein, firefly luciferase, in the absence of lipid, was influenced by a variety of general anaesthetics.⁷ Subsequently, numerous candidate proteins were proposed as putative targets. However, before a protein can be considered as a plausible target certain criteria need to be met:⁸ the protein should be sensitive to clinically relevant concentrations of the anaesthetic (although the accurate determination of behaviourally appropriate concentrations in the CNS can be problematical); the protein must be expressed at appropriate anatomical sites within the CNS; and if the anaesthetic exhibits stereoselective activity, this specificity should be mirrored both by the putative target and in behavioural studies. Based on these criteria, certain transmitter-gated ion channels (TGICs) emerged as putative targets for i.v. general anaesthetics.⁹ For an account of the mechanisms of inhalation anaesthetics, see Franks.¹⁰ Here we review evidence that the GABA_A receptor (GABA_AR) is an important target for certain i.v. general anaesthetics (Fig. 1) and describe the progress being made in elucidating which GABA_AR subtypes mediate the constellation of behaviours produced by these drugs. Such studies are revealing important information on the influence of anaesthetics on the neuronal circuitry associated with sedation, unconsciousness, analgesia, and cognition, and in the future, might lead to the development of improved therapeutics.

Transmitter-gated ion channels as targets for i.v. general anaesthetics

The TGICs considered here are members of one of two genetically distinct receptor superfamilies. The ‘cys-loop’ TGICs include GABA_ARs (Fig. 1), strychnine-sensitive glycine receptors, nicotinic acetylcholine receptors (nAChRs), and 5-hydroxytryptamine 3 receptors (5-HT3Rs).¹¹ In common, these cys-loop receptors are assembled from five transmembrane-crossing subunits, arranged to form an integral ion channel pore (Fig. 1). The glutamate-gated ion channels (including alpha-amino-3-hydroxy-5-methyl-4- isoxazole propionic acid (AMPA), N-methyl-D-aspartate (NMDA), and kainate receptors), exhibit a distinct membrane topology from the cys-loop family and are composed of four transmembrane-crossing subunits, again arranged to form an integral ion channel pore.¹¹ For all TGICs, receptor activation by the neurotransmitter produces a rapid conformational change of the protein that causes the associated ion channel pore to open, permitting selective movement of anions (for GABA_AR and glycine receptors) or cations (for AMPA, NMDA, kainate, nAChR, and 5HT₃R) across the membrane.¹¹ The net flow of either cations or anions depends upon the neuronal membrane potential and the relative intracellular and extracellular concentrations of the ion species. Receptor activation of cation- or anion-associated channels usually results in neuronal depolarization or hyperpolarization, respectively.

Theoretically, general anaesthetics act to enhance the function of inhibitory receptors (GABA_ARs or glycine receptors) or inhibit excitatory receptors (nicotinic, 5HT₃, and ionotropic glutamate receptors). However, it is important to establish which members of the TGIC family are behaviourally relevant targets for i.v. general anaesthetics and to determine whether all anaesthetics exhibit a common receptor profile. In the 1980s and 1990s, a number of investigators demonstrated that the function of GABA_ARs was enhanced by anaesthetic barbiturates, alphaxalone, etomidate, and propofol, but few studies examined their relative selectivity for GABA_ARs vs other TGICs, when determined under identical recording conditions.^9,12–15 We therefore investigated the interaction of four structurally distinct i.v. general anaesthetics (etomidate, propofol, alphaxalone, and pentobarbital) with a variety of recombinant TGICs expressed in Xenopus laevis oocytes (Fig. 2). The receptors investigated included representatives of the major inhibitory anion conducting receptors in the CNS (GABA_ARs and glycine receptors), members of the major excitatory glutamate-gated cation channels (AMPA and NMDA receptors), and the cation-conducting nAChRs and the 5HT₃Rs. This profiling revealed all four anaesthetics to act as positive allosteric modulators of GABA_ARs, i.e. at behaviourally relevant concentrations they enhanced the response of the receptor to GABA (Fig. 2). At greater concentrations, all of these anaesthetics directly activated GABA_ARs in the absence of GABA (a ‘GABA-mimetic’ effect). Etomidate was of particular interest as at low micromolar concentrations it enhanced GABA-evoked responses, but had no effect on the genetically closely related anion-conducting glycine receptor or on any of the cation-conducting receptors (Fig. 2). Similarly, propofol and alphaxalone were relatively selective for the GABA_AR, although they also modestly inhibited neuronal nicotinic receptors (Fig. 2). An effect on nicotinic receptors is unlikely to be crucial to their anaesthetic activity because betaxalone, the behaviourally inert 3β-ol isomer of alphaxalone, also inhibited nicotinic receptors, but distinct from alphaxalone it had no effect on GABA_ARs.^14,15 Although pentobarbital was rather non-selective, it too enhanced GABA_AR function (Fig. 2). Therefore, GABA_ARs represent a viable target for these i.v. general anaesthetics. Note however, that not all i.v. anaesthetics enhance GABA_AR function, an important exception being the dissociative anaesthetic ketamine, for which the NMDA receptor is implicated.^10,19

Given this receptor selectivity profile, the actions of etomidate warranted further investigation. In contrast to most clinical anaesthetics, etomidate is used as the resolved R-(+)-enantiomer, rather than as a racemate.²⁰ Enantioselectivity provides a powerful tool to identify anaesthetic-relevant molecular targets.²¹ The S-(−)-enantiomer of etomidate was less effective than the R-(+)-enantiomer in causing loss of the righting reflex (a surrogate measurement that correlates with loss of response to a verbal command in humans)¹⁰ in mice and tadpoles.²² In agreement, the potency and efficacy of etomidate acting on recombinant GABA_ARs was far greater for the R-(+)-enantiomer than for the S-(−)-enantiomer.^22,23

Structure and pharmacology of GABA_ARs

The GABA_AR structure consists of five subunits arranged around a central anion-conducting pore (Fig. 1). Mammals possess a considerable repertoire of subunits (α1–6, β1–3, γ1–3, δ, ε, π, and ρ1–3), which display a distinctive expression pattern in the CNS.^24–26 This subunit palette underpins the expression of 20–30 distinct GABA_AR isoforms, which exhibit distinct physiological and pharmacological properties, and given their distinctive distribution in the CNS, they also influence different behaviours (see next paragraph). The majority of GABA_ARs contain two α and two β subunits, together with a single γ2 subunit^24–26 (Fig. 1). Synaptic receptors often contain the γ2 subunit, although receptors incorporating this subunit can also be located outside the synapse. The synaptic receptors mediate fast phasic inhibition in response to a transient increase in neurotransmitter resulting from the vesicular release of GABA (Fig. 3). Receptors incorporating the δ subunit, in place of the γ2 subunit, are expressed extra- or perisynaptically, and are repetitively activated by ambient concentrations of GABA, thereby producing a persistent form of tonic inhibition (see Fig. 3).^29–32

The GABA_AR is an important clinical target for benzodiazepines, such as midazolam and diazepam, where they act as positive allosteric modulators to enhance the interaction of GABA with the receptor. The binding site for such drugs is at the interface of an α subunit and a γ subunit, usually the γ2 subunit.²⁶ Recombinant GABA_ARs containing the γ2 subunit, partnered with the α4 or the α6 subunit, are insensitive to benzodiazepines such as diazepam, whereas α1-, α2-, α3- and α5-βγ2 receptors are sensitive.^26,33 The molecular basis of this α-subunit-selective pharmacology was revealed by the construction of chimeric constructs of the α1 and the α6 subunit, leading to the identification of a crucial single amino acid residue located in the N-terminal portion of the α subunit.^26,33 This amino acid is a histidine (H) residue for α1–3 and α5 subunits and an arginine (R) residue for the insensitive α4 and α6 subunits. Importantly, an H to R residue exchange by site-directed mutagenesis results in diazepam insensitivity of the receptor. This finding enabled creation of ‘knock-in’ mice whereby the α1, α2, α3, or α5 subunit was replaced with a mutant subunit incorporating the H to R mutation. Such mice have been invaluable in discovering which GABA_AR subtypes mediate the behavioural repertoire produced by benzodiazepines. For example, for the α1H101R mouse the sedative action of diazepam was abolished, but the anxiolytic action was maintained.²⁶ In contrast, for the equivalent α2H101R mouse, the anxiolytic effect of diazepam was abolished, but the sedative effect remained intact.^26,34 These findings have encouraged the development of GABA_AR-isoform-selective drugs in the quest for a non-sedative anxiolytic.^26,35–37 Additionally, ligands acting as selective positive allosteric modulators of spinal cord α2βγ2 GABA_ARs show promise as analgesics^38–40 (Table 1).

Investigating the GABA_AR isoform selectivity of i.v. general anaesthetics

If GABA_ARs are an important target for general anaesthetics such as etomidate, then which GABA_ARs mediate the constellation of behaviours that constitute the anaesthetic state? Unlike the benzodiazepines, i.v. general anaesthetics such as propofol or the steroidal anaesthetics exhibit little or no selectivity for the different GABA_AR subtypes. A notable exception is etomidate. Our voltage-clamp studies of Xenopus laevis oocytes expressing human GABA_ARs revealed etomidate to selectively enhance GABA responses mediated by activation of GABA_ARs incorporating the β2 or the β3 subunit, but to have a reduced effect on equivalent GABA_ARs containing the β1 subunit.¹⁶ We therefore employed a similar strategy to that used to elucidate the molecular basis of the benzodiazepine alpha subunit selectivity. Construction of chimeric β1 and β2 subunits revealed the subunit specificity of etomidate to reside with a single amino acid [asparagine (N) for both the β2 and β3 subunits and serine (S) for the β1 subunit] located within the second transmembrane (TM2) region (a part of the protein that contributes to the lining of the associated anion-conducting ion channel).¹⁷ Site-directed mutagenesis to β2N265S reduced the GABA-modulatory and GABA-mimetic actions of etomidate, whereas the complementary mutation of the β1 residue (β1S265N) enhanced these actions of etomidate.¹⁷ A methionine (M) residue occupies the equivalent position of the Drosophila invertebrate GABA subunit.⁴¹ Mutation to β2N265M not only abolished the actions of etomidate, but additionally reduced the effects of propofol.^9,42,43

Considering the long-held view that general anaesthetics act in a non-specific way to disrupt the neuronal membrane, these findings, at least for etomidate, were surprising. To summarize, etomidate is a highly selective positive allosteric modulator of GABA_ARs, with little or no effect at behaviourally relevant concentrations on other transmitter-gated ion channels. The anaesthetic effects of etomidate are enantioselective, a specificity that is mirrored in their interaction with GABA_ARs, providing confidence that this receptor may be a relevant target. Furthermore, etomidate exhibits a clear selectivity for GABA_ARs that contain particular isoforms of the β subunit, a specificity that is dictated by the nature of a single amino acid of the 2000 or so that make up this pentameric receptor. Whether this residue, together with spatially related residues, contributes to an anaesthetic binding pocket is the subject of current investigation. Use of photolabelled anaesthetic analogues of etomidate and propofol, substituted cysteine modification protection techniques, and molecular modelling studies are assisting in the identification of potential binding sites within or between selective GABA_AR subunits for these i.v. anaesthetics.^44–46

Although collectively these studies identify the GABA_AR as a putative target for mediating the behavioural effects of etomidate, supporting in vivo studies are required. In this regard, an important advance was made by the development of the β2N265S and the β3N265M knock-in mice. These mutations, which in vitro studies reveal to suppress the GABA_AR actions of etomidate,^10,47–49 have been introduced into mice through genetic engineering. In the β2N265S mouse, the sedative actions of etomidate, as assessed by an activity box, were blunted, together with the effects of this agent to induce slow-wave sleep⁴⁸ (Table 1). Furthermore, the hypnotic effects of etomidate, as assessed by loss of the righting reflex, were influenced by this β2 mutation (Table 1). Complementary studies have been conducted in a β3N265M mouse. As described above, in vitro studies revealed this methionine mutation to suppress the GABA-modulatory actions of etomidate, but additionally to blunt those of propofol. In the β3N265M mouse, the sedative effects of etomidate were similar to those of the wild-type mouse.^10,18,47 Therefore, GABA_ARs containing the β2 subunit mediate the sedative effects of etomidate. However, the hypnotic effects (duration of the loss of the righting reflex) of etomidate were reduced by the β3 mutation. Collectively, these results suggest the hypnotic effects of etomidate to involve both β2- and β3-containing GABA_ARs (Table 1). The immobilizing effect of etomidate, assessed by the hindlimb withdrawal reflex, was abolished by the β3 subunit mutation (Table 1).^10,18,47

Thalamocortical pathway: a site of action for i.v. general anaesthetics

The neuroanatomical substrates of general anaesthetic action remain elusive, although electrophysiological, neuroimaging, and circuit-modelling studies consistently implicate the thalamus as an important locus for anaesthetic-induced sedation and hypnosis.^10,49–53 Indeed, the thalamus has a recognized role in controlling conscious state transitions.⁵⁴ We used the whole-cell voltage-clamp technique to record from thalamic brain slices obtained from wild-type and GABA_AR-mutant mice. Using GABA_AR subtype-selective drugs, immunohistochemistry, and a variety of GABA_AR mutant mice, we determined that the synaptic GABA_ARs of the mouse thalamocortical ventrobasal relay neurones are composed of α1, β2, and γ2 subunits⁵⁵ (Fig. 3). Upon vesicular release of GABA, these receptors are briefly activated, resulting in the near-simultaneous opening of a population of anion channels and the movement of chloride ions, usually into the neurone, to cause phasic inhibition (an inhibitory postsynaptic potential, IPSP). Under voltage-clamp conditions, activation of synaptic GABA_ARs by GABA released from a single vesicle results in a phasic miniature inhibitory postsynaptic current (mIPSC; Fig. 3). In response to a presynaptic action potential, the near-synchronous release of GABA from multiple vesicles produces an inhibitory postsynaptic current (IPSC). In wild-type mice, low micromolar concentrations of etomidate greatly prolonged the mIPSC duration, whereas this effect was blunted in the β2N265S mouse.^55,56 These ventrobasal neurones additionally express peri- or extrasynaptic GABA_ARs, composed of α4, β2, and δ subunits, that mediate a tonic form of inhibition^55,56 (Fig. 3). In wild-type mice, etomidate greatly increased tonic inhibition, but this effect was blunted in the β2N265S mouse.⁵⁶ Therefore, the effects of etomidate to enhance both phasic and tonic inhibition are compromised in the β2N265S mouse.

To gain a better understanding of the relative importance of these effects of etomidate, we first determined in more physiological conditions how phasic and tonic inhibition integrate to influence ventrobasal neurone excitability. Ventrobasal neurones are innervated by a band of GABA-ergic nucleus reticularis neurones that provide the major source of inhibition. The ventrobasal and the nucleus reticularis neurones exhibit both burst and tonic firing modes, with tonic firing dominating during waking and burst firing during periods of drowsiness and non-rapid-eye-movement (NREM) sleep.^57,58 Using paired recordings of synaptically coupled nucleus reticularis–ventrobasal neurones, we demonstrated that high-frequency burst firing of nucleus reticularis neurones produced a greatly prolonged IPSC that resulted from GABA activating synaptic α1β2γ2 receptors, but additionally from the ‘spillover’ of GABA from the synapse, which then activated the extra- or perisynaptic α4β2δ GABA_ARs²⁷ (Fig. 3). In support, the IPSC duration was reduced in equivalent recordings made from a mouse where the α4 subunit was genetically deleted (α4^−/−).²⁷ Furthermore, during burst firing, DS2, a δ-GABA_AR-selective positive allosteric modulator, greatly prolonged the IPSC duration in ventrobasal neurones derived from wild-type mice, but not those of the α4^−/− mouse.²⁷ Etomidate also greatly prolonged such IPSCs resulting from activation of synaptic, and particularly, extrasynaptic GABA_ARs.²⁸ In complementary experiments, slow IPSCs in response to nucleus reticularis burst firing were still evident in thalamocortical neurones, even though their synaptic α1-GABA_ARs had been genetically deleted.⁵⁹ Furthermore, although lacking synaptic α1-GABA_ARs, in vivo thalamic recordings from such mice still revealed slow oscillations, or sleep spindles.⁵⁹

A recent in vivo study further highlights the importance of these thalamic extrasynaptic GABA_ARs to the behavioural effects of etomidate. An increase in frontal electrocortical activity in the α–β frequency range is considered to signal an anaesthetic-induced loss of consciousness.⁶⁰ Elevated thalamic α–β activity precedes similar activity in the cortex.⁶¹ During NREM sleep, which is a state associated with burst firing, microperfusion of etomidate directly into the thalamus increased α–β activity of wild-type mice, but not that of mice where the δ subunit had been genetically deleted (δ^−/− mice).⁶⁰ Furthermore, these effects were mimicked by microperfusion of DS2, a δ-GABA_AR selective positive allosteric modulator.⁶⁰ Collectively, these studies further highlight the thalamus as a crucial locus for anaesthetic action, and in particular, emphasize the importance of the effect of agents such as etomidate to enhance spillover inhibition. The pharmacology of the thalamic synaptic and extrasynaptic receptors and their differential impact upon brief phasic, tonic, and spillover inhibition is provided in Table 2. It is noteworthy that δ-GABA_ARs are insensitive to benzodiazepines. Therefore, in these thalamocortical relay neurones, agents such as diazepam and midazolam will prolong brief phasic inhibition mediated by synaptic GABA_ARs (α1β2γ2), with no effect on tonic inhibition mediated by extrasynaptic GABA_ARs (α4β2δ). They are predicted to only exert a limited impact upon the ‘spillover’ IPSCs that occur during burst firing, which although involving synaptic GABA_ARs, are dominated by the contribution of the δ-GABA_ARs.

Role of GABA_ARs in the impairment of cognition associated with i.v. general anaesthetics

Short-term impairment of memory is an important property of general anaesthetics; however, clinically, their use and the accompanying surgery can be associated with a postoperative cognitive impairment, lasting for days to months after administration.⁶² The hippocampus is known to play an important role in the processes associated with both learning and memory. Hippocampal CA1 pyramidal neurones express extrasynaptic receptors composed of α5, β, and γ2 subunits that mediate tonic inhibition.⁶³ This tonic conductance is not evident in CA1 neurones derived from α5^−/− mice.⁶³ Such tonic currents are enhanced by general anaesthetics such as etomidate⁶⁴ and are reduced by subsequent co-application of L-655,708, a selective negative allosteric modulator of α5βγ2 GABA_ARs.⁶⁴ Hippocampal long-term potentiation, a form of synaptic plasticity associated with learning and memory, is suppressed by etomidate in CA1 neurones obtained from wild-type but not from α5^−/− mice.⁶⁵ Furthermore, suppression of long-term potentiation by etomidate was prevented by co-application of L-655,708.^64,65 Behaviourally, spatial and non-spatial hippocampus-dependent learning tasks were impaired by prior etomidate treatment, a deficit also prevented by treatment with L-655,708.^64,65

Intriguingly, administration of sedative and anaesthetic doses of etomidate impaired memory performance in the novel object recognition (NOR) test for more than 3 days after sedation and 7 days after anaesthesia, i.e. long after the anaesthetic was eliminated. Implicating α5-GABA_ARs, this impairment did not occur in the α5^−/− mouse and was prevented by treatment with the α5-GABA_AR negative allosteric modulator L-655,708.⁶⁶ How does etomidate produce such a prolonged effect on cognition, maintained long after elimination of the anaesthetic? In tandem with these behavioural impairments, ex vivo recordings from CA1 neurones of mice previously exposed to etomidate, revealed a persistent increase of the CA1 neurone tonic current mediated by α5-GABA_ARs and an associated decrease in the magnitude of long-term potentiation.⁶⁶ Complementary biochemical experiments revealed increased cell surface expression of α5-GABA_ARs in the hippocampus. The mechanism of this anaesthetic-induced plasticity is not known. However, this effect of etomidate could be replicated in hippocampal cell cultures and required the presence of astrocytes, which presumably release some as yet unknown factor.⁶⁶ A possible candidate is the inflammatory cytokine interleukin-1β (IL-1β). Inflammation triggered by surgical trauma, by the anaesthetic per se, or both may increase circulating concentrations of IL-1β.⁶⁷ In the hippocampus, this cytokine has been shown to increase cell surface expression of α5-GABA_ARs and, consequently, to increase the CA1 tonic current.⁶⁷ Indeed, etomidate and IL-1β act synergistically to enhance greatly the CA1 tonic current mediated by α5-GABA_ARs.⁶⁸

Conclusions

The GABA_AR is an important target for the behavioural actions of general anaesthetics such as etomidate and propofol. Mammals express 20–30 different GABA_AR isoforms. These receptor subtypes are not uniformly expressed throughout the CNS, but exhibit specific expression profiles, not only within different regions of the CNS, but even within individual neurones. Given this very specific neuroanatomical expression profile, it is perhaps not surprising that specific GABA_AR subtypes mediate distinct components of the behavioural repertoire of drugs such as benzodiazepines, etomidate, and propofol. The use of genetic techniques that permit the manipulation of wild-type or mutant GABA_ARs in a neurone-specific manner, coupled with advances in optogenetic manipulations, should further develop our understanding of the neuronal circuitry implicated in these drug-induced behaviours. The use of mice engineered to express anaesthetic-insensitive GABA_ARs has identified the molecular targets for developing novel therapeutics with a reduced propensity for side-effects. The challenge now is for the medicinal chemists to exploit the molecular differences between the various GABA_AR isoforms to develop a new generation of medicines.

Authors’ contributions

All authors contributed equally to the preparation of the manuscript, tables, and figures.

Declaration of interests

C.J.W. was a member of the editoral board of BJA Education (2007–2017). S.J.M. and J.J.L. have no conflicts to declare.

References