II. Basic Pharmacology and Physicochemical Properties of the Halogenated Ethers

Despite the long clinical history and extensive characterization of the properties of halogenated ethers (see Fig. 1; Table 1), their precise molecular mechanisms of action remain unclear. Evidence points to pre- and postsynaptic ligand-gated ion channels and voltage-gated ion channels as highly relevant targets of halogenated ether–induced unconsciousness (Campagna et al., 2003; Covarrubias et al., 2015). Broadly speaking, general anesthetics act via depression of excitatory signaling and/or enhancement of inhibitory signaling. Under this generalization, many known protein targets exist. GABA_A, glycine, glutamate, nicotinic acetylcholine, and 5-HT₃ receptors, among others, have been implicated (Mashour et al., 2005; Forman and Chin, 2008). For further discussion of the promiscuous molecular pharmacology of general anesthetics, see the comprehensive reviews (Franks and Lieb, 1994; Campagna et al., 2003; Rudolph and Antkowiak, 2004; Hemmings et al., 2005b; Franks, 2006, 2008). In addition to neurotransmitter receptors, a growing body of evidence has emerged indicating VGSCs as mediators of some of the effects of halogenated ethers (Hemmings, 2009; Herold and Hemmings, 2012).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Molecular structures of diethyl ether and the halogenated ethers.

III. Basic Overview of Voltage-Gated Sodium Channel Biology

VGSCs are transmembrane protein complexes responsible for the initiation and propagation of action potentials in neurons as well as most excitable cells (Hille, 2001). VGSCs consist of a pore-forming α-subunit (220–260 kDa) and smaller nonpore-forming β-subunits, 30–40 kDa in size (Beneski and Catterall, 1980). VGSC α-subunits alone are sufficient for ion conduction and voltage sensing, whereas β-subunits play many roles, including the modification of channel expression, gating, localization, as well as functioning as a cell adhesion and signaling molecule (Goldin et al., 1986; Hull and Isom, 2018). VGSCs transition between several conformations through the course of an action potential, including closed, open, and inactivated states (see Fig. 2). The predominant state of the channel at resting membrane potential is the closed state, where no Na⁺ ions are conducted. Upon sufficient membrane depolarization, channels undergo a conformational change to the open state, where Na⁺ ions enter the cell and drive further depolarization. After transition to the open state, fast inactivation is enhanced, in which intracellular residues (the inactivation gate) occlude the inside face of the channel, blocking further conduction. Such fast inactivation (recovery time constants on the scale of milliseconds) is distinct from slow inactivation (recovery time constants on the scale of hundreds of milliseconds to seconds), which is discussed in more detail below. The comparatively slower rate of inactivation to activation and deactivation allows for timely termination of depolarization during an action potential and supports the refractory period between action potentials (Hille, 2001). Entry into the slow inactivated state requires repetitive or sustained depolarizations that accumulate channels in the slow inactivated state over the course of numerous action potentials (Schauf et al., 1976). Recovery from the slow inactivated state is much slower than entry, supporting longer-lasting effects on the channel than fast inactivation, which is limited to affecting fewer individual action potentials due to its fast recovery time constant. VGSC slow inactivation has the interesting property that at the depolarized potentials that support spiking (i.e., −70 mV), the recovery from slow inactivation is slow, whereas at hyperpolarized potentials (i.e., −80 mV), the recovery is much faster (Fleidervish et al., 1996; Jung et al., 1997). This voltage-dependent effect is highly relevant to neuromodulation, as discussed below.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Transition states of voltage-gated sodium channels. At rest, the activation gate is closed and the channel is in a nonconducting state. After sufficient depolarization of the membrane, the channel undergoes a conformational change to enter the activated state where both gates are open, allowing the conduction of Na⁺ ions. After transitioning to the activated state, the channel undergoes fast inactivation, where intracellular residues (the inactivation gate) occlude the inside face of the channel blocking further conduction of Na⁺ ions. After recovery from inactivation, the channel returns to the resting state.

The different conformational states rely on the activity of distinct regions of the channel, indicating that distinct channel states and their role in general anesthesia can be tested pharmacologically. VGSC α-subunits consist of four homologous domains, each composed of six transmembrane segments (see Fig. 3). Segments 1–4 (S1–S4) are primarily involved in voltage sensing, whereas S5–S6 form the pore. Positively charged lysine and arginine residues in the S4 helices act as voltage sensors. A short loop on the intracellular side of the channel linking domains III and IV forms the inactivation gate. Slow inactivation utilizes molecular mechanisms that differ from fast inactivation. Mutations that impair fast inactivation do not prevent slow inactivation (Rudy, 1978; Featherstone et al., 1996; Vedantham and Cannon, 1998; Todt et al., 1999; Mitrovic et al., 2000; Ong et al., 2000). Although numerous mutations in VGSCs have been found to affect slow inactivation, the S6 segments contain highly conserved asparagine residues that, when mutated to hydrophobic residues, completely block slow inactivation (Carr et al., 2003; Chen et al., 2006).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Schematic of the eukaryotic VGSC α-subunit. Roman numerals and numbers indicate domains and segments, respectively. The pore loop regions comprised of segments S5 and S6 are highlighted in green. Voltage-sensing S4 segments are highlighted in red. The inactivation gate and “h” particle containing the IFMT motif that is crucial for inactivation are highlighted in yellow. Small white circles of the re-entrant loops in each domain represent the amino acid residues that form the Na⁺ ion selectivity filter.

Ten different human VGSC α-subunit isoforms (Na_v1.1–1.9 + Na_x) encoded by the SCN(X)A genes have been characterized (see Table 2). VGSC α-subunits can be divided into two classes based on their sensitivity to tetrodotoxin (TTX). Na_v1.1–1.4, Na_v1.6, and Na_v1.7 are TTX-sensitive, as half-maximal block of Na⁺ conduction by TTX occurs at concentrations between 1 and 25 nM (Catterall et al., 2005). Na_v1.5, Na_v1.8, and Na_v1.9 are referred to as TTX-resistant, requiring >1 µM TTX for block of half-maximal conduction. All of the human VGSC subtypes share greater than 75% amino acid sequence identity (Catterall et al., 2005). When VGSC α-subunits were originally purified from rat brain, it was discovered that β-subunits were present in a heterotrimeric complex (Beneski and Catterall, 1980). Five β-subunit isoforms have been discovered (O’Malley and Isom, 2015). Heterologous expression of VGSC α-subunits alone is sufficient for functional Na⁺ conduction, but the presence of β1 or β3 and β2 or β4 confers further modularity to channel function and expression. β1 or β3 associates noncovalently, and β2 or β4 forms disulfide bonds with α-subunits. In addition to modulation of channel expression and function, β-subunits act as cell adhesion molecules, with roles in cellular aggregation, neurite extension, and axonal fasciculation (Hull and Isom, 2018). Each VGSC subtype is uniquely distributed within and across cell and tissue types, indicating that specific regional- or cell type–selective mechanisms of general anesthesia may be investigated through manipulation of specific VGSCs. Each subtype has distinctive kinetic and voltage-sensing properties. VGSC subtypes are strategically expressed based on these properties to maximize physiologic function. Na_v1.1–1.3 and Na_v1.6 are the most prominent central nervous system VGSCs, whereas the Na_v1.7–1.9 isoforms are mainly expressed in the peripheral nervous system. Na_v1.4 and Na_v1.5 are mainly expressed in skeletal and cardiac muscle, respectively (Catterall et al., 2005). The different VGSCs expressed in brain exhibit isoform enrichment in a cell type– and subcellular region–directed manner. In this study, we focus on VGSC α-subunits; for a review of β-subunit expression patterns, see Hull and Isom (2018). In the adult brain, Na_v1.1, Na_v1.2, and Na_v1.6 are the predominant VGSCs, whereas Na_v1.3 is highly expressed in the soma of neurons during development, but shows low expression in the adult brain (Yu and Catterall, 2003). In the forebrain, Na_v1.1 mRNA and protein are highly expressed in GABAergic inhibitory interneurons, although weaker expression exists in pyramidal neurons of the cortex and hippocampus (Westenbroek et al., 1989; Gong et al., 1999; Van Wart et al., 2007; Okaty et al., 2009). In interneurons, subcellular localization of Na_v1.1 protein is enriched in the cell body and to a greater degree in the axon initial segment (Ogiwara et al., 2007). Na_v 1.2 expression is enriched in the axons of somatostatin- but not parvalbumin (PV)-positive interneurons of the cortex, showing further cellular subtype enrichment in VGSC expression (Li et al., 2014). Na_v 1.2 is also expressed highly in the proximal axon initial segment of pyramidal neurons and, during development, highly at nodes of Ranvier. Na_v 1.2 expression is present, but weaker in the cell body and dendrites of pyramidal neurons (Hu et al., 2009). Na_v 1.6 protein is highly expressed in the axons of PV-positive neurons and in the distal axon initial segment of pyramidal neurons and in nodes of Ranvier (Hu et al., 2009; Li et al., 2014). Na_v 1.6 is also expressed in the cell body and dendrites of pyramidal neurons (Caldwell et al., 2000; Krzemien et al., 2000; Lorincz and Nusser, 2010). The three major adult central nervous system VGSCs show both pre- and postsynaptic localization at synaptic terminals in the hippocampus (Johnson et al., 2017). Such localization appears to be regional/cell class specific, with the central hilus of the dentate gyrus showing predominantly presynaptic Na_v 1.1 immunoreactivity, whereas in the stratum radiatum of CA1, Na_v 1.1, 1.2, and 1.6 exhibit stronger postsynaptic localization in spines and dendrites (Johnson et al., 2017). Although most of the work studying the cell type– and subregional-specific VGSC expression and function has focused on cortex and hippocampus, investigation has expanded to thalamic and subthalamic regions of the brain. Recently, it was found that targeted deletion of Scn8a (encoding Na_v1.6) in PV-positive neurons of the thalamic reticular nucleus of the thalamus was sufficient to generate absence epilepsy, a nonconvulsive seizure type that is characterized by sudden impairment of consciousness (Makinson et al., 2017). Na_v1.1 is also expressed in the thalamus, with deletion of Scn1a impacting function of thalamic reticular neurons (Kalume et al., 2015). The role of VGSCs in the mechanism(s) of the impairment of consciousness in absence epilepsy may further elucidate our understanding of the role of VGSCs in the molecular mechanism of halogenated ether-induced unconsciousness. Although the study of VGSCs in the consciousness promoting subcortical regions including the brainstem is nascent, cell type and regional effects of specific VGSCs in these regions are also likely to be highly relevant to mechanisms of general anesthesia.

To date, the successful crystallization of a human VGSC has yet to be accomplished. However, in 2018, the structure of human Na_v1.4 coexpressed with the β1-subunit was resolved at a 3.2-Å resolution using cryogenic electron microscopy (Pan et al., 2018). This marks the highest-resolution structure of a eukaryotic VGSC currently available and may guide pharmacological investigation of VGSCs, including the action of the halogenated ethers. Selectivity of Na⁺ over other ions is an essential function for the action of VGSCs. This is achieved by the precise architecture of a VGSC region known as the selectivity filter. Approximately 4.6 Å in size, the selectivity filter is composed of an inner and outer ring of charged residues. The asymmetric charge distribution of each ring coordinates the solvated Na⁺ ions as they flow through the pore (Catterall, 2012). Through mutagenesis and subsequent electrophysiological analysis, four key residues (D: aspartate, E: glutamate, K: lysine, and A: alanine), each present in one of the domains, were identified as the inner ring of the mammalian VGSC selectivity filter (Lipkind and Fozzard, 2008). The DEKA motif found in human VGSCs is conserved across all isoforms. Mutating the KA (lysine and alanine) of the DEKA motif to glutamate (EE) abolishes Na⁺ selectivity of the channel and allows Ca²⁺ ions to pass through the pore (Heinemann et al., 1992).

IV. Molecular Interactions of Halogenated Ethers with Voltage-Gated Sodium Channels

In 1940, Heinbecker and Bartley reported a marked increase in the rate of accommodation to subthreshold depolarizations seen in the presence of 24 mM ether. It is now known that the accommodation rate is directly proportional to the rate of K⁺ channel activation and Na⁺ channel inactivation. Due to the low potassium conductance in frog nerve, this may have been the first electrophysiological indication of ether’s action on sodium channel function (Heinbecker and Bartley, 1940). In 1956, Stephen Thesleff examined the effect of numerous nonvolatile anesthetics on the action potential of frog skeletal muscle using what was, at the time, the new voltage-clamp technique (Thesleff, 1956). All agents tested increased firing threshold and reduced action potential amplitude, mostly at concentrations higher than those relevant to clinical anesthesia. At the time, the idea that voltage-controlled channels in the membrane were carrying Na⁺ ions was hypothetical. Thesleff (1956) was aware of the ionic hypothesis underlying the action potential proposed during his time (Hodgkin and Katz, 1949; Hodgkin and Huxley, 1952). Decades before the molecular characterization of VGSCs, he hypothesized that the reduction in excitability caused by anesthetic agents could be due to a reduction in the selective Na⁺ conductance across the membrane. He attributed the anesthetic reduction in action potential amplitude to what he called an “inhibition of the sodium carrying mechanism” (Thesleff, 1956).

The same sodium-dependent action potential blockade was also seen in isolated frog skeletal muscle fiber with diethyl ether and chloroform (Yamaguchi, 1961; Inoue and Frank, 1965). Combined with their finding that the local anesthetic (LA) procaine also blocked action potential conduction in isolated frog skeletal muscle fiber via a sodium-carrying mechanism, Inoue and Frank proposed that general and LAs acted via a unitary molecular mechanism in all excitable cells of the nervous system (Inoue and Frank, 1962; Frank and Sanders, 1963; Frank and Pinsky, 1966).

Motivated by the well-known positive correlation discovered between a compound’s solubility in olive oil and anesthetic potency (Meyer, 1899; Overton, 1901; Harris and Bruno, 1985), research on the mechanism of anesthetic action during the 1970s focused on a nonspecific drug perturbation of the lipid bilayer. It was proposed that volatile anesthetics acted on the lipid bilayer causing a lipid fluidization of the cell membrane, resulting in the depression of neurotransmission and eventually unconsciousness. Fluidization of the membrane was thought to cause distortions of membrane proteins (with an emphasis on ion channels such as Na⁺ channels), leading to a block of Na⁺ conductance and a breakdown of signaling in excitable cells of the central nervous system (Seeman, 1972). Phillip Seeman’s hypothesis is regarded as one of the first proposing VGSCs as specific targets of volatile anesthetic action (Seeman, 1974).

Early voltage–clamp studies of the late 1970s and early 1980s, using the voltage-clamp technique in frog, crayfish, and squid giant axon, showed that diethyl ether reduced peak Na⁺ current by accelerating VGSC inactivation, but with little to no effect at concentrations of clinical relevance (Bean, 1979; Kendig et al., 1979; Bean et al., 1981; Haydon and Urban, 1983, 1986; Urban and Haydon, 1987; Elliott et al., 1992). When investigating the effect of diethyl ether and the haloalkane halothane on Na⁺ current in the crayfish giant axon, Bean et al. (1981) observed a minor concentration-dependent enhancement of inactivation at all membrane potentials. Ether produced a hyperpolarizing shift in steady-state inactivation but did not alter the time-dependent recovery from inactivation, nor did it affect frequency-dependent stimulation (Bean et al., 1981).

Haydon and Urban (1983) reached similar conclusions when investigating the squid giant axon. No significant difference was found in membrane capacitance or Na⁺ conductance. Steady-state inactivation was shifted by a few millivolts in the hyperpolarizing direction (Haydon and Urban, 1983). The effect of ether, halothane, and enflurane inhibition on Na⁺ uptake using rodent brain synaptosomes was evaluated by Harris and Bruno (1985). All three of the anesthetics caused a 10%–20% reduction in Na⁺ uptake (Harris and Bruno, 1985).

According to the lipid-based hypotheses, anesthetics would be adsorbed by the bilayer, increasing membrane thickness, which would inhibit ion channels and subsequently axonal conduction (Franks, 2006). Controversy ensued when it was observed that the anesthetic-induced hyperpolarizing shift in sodium channel steady-state inactivation could not be due to the minor perturbations of bilayer thickness (Franks and Lieb, 1978, 1979, 1981; Elliott et al., 1984). Nicholas Franks and William Lieb demonstrated that anesthetics could act in a lipid-free environment, which brought lipid fluidization into question as the mechanism of anesthetic action (Franks and Lieb, 1981, 1982, 1984),. They proposed that direct interactions with membrane proteins were responsible for anesthetic action at a molecular level. They did not, however, believe that VGSCs were among the main proteins targeted. Before dismissing the involvement of VGSCs in their landmark Nature review, Franks and Lieb (1994) conceded the plausibility of even minor ion channel modulation disrupting communication between neurons.

A unitary hypothesis of general anesthetic action based on membrane fluidization is now widely regarded as misleading. Changes in temperature cause similar effects as general anesthetics on the properties of the cell membrane, but do not alter consciousness. Herold et al. (2014) showed that volatile anesthetics inhibit VGSCs independent of any action on the lipid bilayer of the cell membrane. Recent studies by Herold et al. (2017a,b) examined the effects of a range of structurally diverse general anesthetics on lipid bilayer properties using a gramicidin-based fluorescence assay. The authors observed no detectable effects on lipid bilayer properties at clinically relevant concentrations of most anesthetics tested, with only minimal perturbations seen in some halogenated ethers or at supratherapeutic concentrations of the agents tested (Herold et al., 2017a,b). These findings provide strong support for the notion that general anesthetics act directly on membrane proteins to produce their clinical effects, excluding indirect ion channel modulation as a pharmacologically relevant mechanism.

Molecular advances and cloning techniques permitted the analysis of isolated VGSC isoforms expressed in mammalian cells or other heterologous systems such as Xenopus oocytes. Using rat brain IIA (Na_v1.2) channels stably expressed in Chinese hamster ovary cells, Rehberg et al. (1996) were among the first to publish recordings showing halogenated ether activity with a specific VGSC isoform. All agents tested caused a reduction of peak Na⁺ current and a hyperpolarizing shift in the half-maximal voltage (V_1/2) of steady-state inactivation. The hyperpolarizing shift in the V_1/2 of inactivation ranged from −20 to −25 mV, depending on the volatile anesthetic tested and holding potential of the protocol (Rehberg et al., 1996). Duch et al. (1998) confirmed this, reporting a similar reduction in peak Na⁺ current and a hyperpolarizing shift of inactivation with whole-cell recordings testing isoflurane in rat dorsal root ganglion neurons. Stadnicka et al. (1999) tested isoflurane and halothane on human embryonic kidney cells expressing human Na_v1.5 (hH1a) and concluded both agents preferentially inhibit the inactivated state of Na_v1.5.

Why did the early electrophysiologists see no conduction block in response to halogenated ethers at similar concentrations to those used in the experiments above? The diminished effect seen with early voltage–clamp recordings of halogenated ethers may be specific to the experimental preparation (i.e., squid giant axon vs. mammalian cell line). Experimental conditions used with invertebrate preparations—including ion concentrations, recording temperature, ion channel expression patterns, and numerous other factors—can complicate the translation of results to mammalian physiology. A potential explanation is the large reserve of VGSCs present in axons, in which perturbations are likely to affect spike-timing precision and conduction velocity, but unlikely to result in overt action failure. Although axons contain the highest density of VGSCs in neurons, as discussed previously, VGSCs are expressed in a controlled manner in a variety of subcellular and cell type–specific contexts. Furthermore, as discussed above regarding neuromodulation, VGSCs exhibit differential sensitivity based on a variety of context-dependent factors, including neuromodulation, membrane potential, and isoform type.

To address this question, Ouyang et al. (2003) performed patch–clamp recordings in isolated rat neurohypophysial nerve terminals. Both isoflurane (and the intravenous anesthetic propofol) blocked Na⁺ current at clinically relevant concentrations. Isoflurane inhibited peak Na⁺ current with an IC₅₀ of 0.45 mM when holding the voltage of the membrane at −70 mV (V_h −70 mV) and 0.56 mM (V_h −90 mV). Isoflurane did not alter activation kinetics, but did cause a hyperpolarizing shift of the V_1/2 of inactivation (−14 mV with a 30-ms prepulse protocol) (Ouyang et al., 2003). The IC₅₀ values seen in the terminals were lower than previously reported (Rehberg et al., 1996; Duch et al., 1998). This indicates several possibilities, including differential inhibition of VGSC isoforms, neuromodulatory state dependence, or modulation of channel sensitivity by accessory VGSC proteins. VGSC α-subunit isoforms expressed in Chinese hamster ovary or human embryonic kidney cells do not possess the same VGSC accessory proteins present in rat neurohypophysial nerve terminals, although some cell lines exhibit background β-subunit expression (Calhoun and Isom, 2014).

Using the two-electrode voltage–clamp technique, Shiraishi and Harris (2004) measured Na⁺ current in Xenopus oocytes expressing rat Na_v1.2, human Na_v1.4, rat Na_v1.6, or rat Na_v1.8. They also coexpressed rat β1-subunits, although their presence did not alter any channel’s sensitivity to the anesthetics tested. Isoflurane inhibited peak Na⁺ current in oocytes expressing rat Na_v1.2, human Na_v1.4, or rat Na_v1.6 by ∼30% (Shiraishi and Harris, 2004). The effect of isoflurane was more pronounced at depolarized holding potentials. The TTX-resistant isoform Na_v1.8 was found to be insensitive to isoflurane at concentrations up to twice the minimum alveolar concentration (defined as the concentration at which 50% of subjects do not respond with movement to noxious stimuli) (Shiraishi and Harris, 2004). A later study using the rodent neuronal expression system, ND7/23 cells, expressing rat Na_v1.8, showed that Na_v1.8 is sensitive to isoflurane in this expression system. At (V_h −70), 1.8 minimum alveolar concentration (MAC) blocked ∼55% of Na_v1.8-mediated current (IC₅₀ 0.67 ± 0.06 mM) (Herold et al., 2009). The discrepancy seen in oocytes versus the ND7/23 line indicates the importance of a native neuronal environment, suggesting a potential role for nonpore-forming β-subunit proteins, cytoskeletal environment, or post-translational modification (e.g., phosphorylation) in VGSC sensitivity to halogenated ethers. Similar results of cell line– and cell type–specific modulation of VGSC function and expression are observed with VGSC β-subunits, suggesting a general need to investigate VGSC function in a physiologic and precisely defined context. Of note is that aromatic volatile anesthetics like fluorobenzene are also observed to cause significant inhibition of Na⁺ current in Xenopus oocytes (Horishita et al., 2008).

Ouyang and Hemmings (2007) tested the selectivity of isoflurane on Chinese hamster ovary cells stably expressing rat Na_v1.2, rat Na_v1.4, or rat Na_v1.5. Na⁺ current was significantly reduced, and a hyperpolarizing shift in steady-state inactivation was seen in all three isoforms. Importantly, isoflurane displayed selective blocking properties between isoforms. At V_h −100 mV, tonic block was observed for all three isoforms. When holding at V_h −70 mV, enhanced inactivation became the dominant mechanism of isoflurane inhibition of rat Na_v1.4- and rat Na_v1.5-mediated Na⁺ current (Ouyang and Hemmings, 2007).

Isoflurane, sevoflurane, enflurane, desflurane, and the haloalkane halothane were tested via whole-cell patch clamp from Chinese hamster ovary cells stably expressing rat Na_v1.4 (Ouyang et al., 2009). Each anesthetic at ∼1 MAC significantly inhibited peak Na⁺ current. Peak current was elicited by a 25-ms step (V_h of −80 mV) to −10 or −20 mV. Reduction in peak Na⁺ current by isoflurane and sevoflurane: 14%; enflurane: 28%; and desflurane: 48% (see Fig. 4). Consistent with previous reports, each of the halogenated ethers and halothane caused a significant hyperpolarizing shift in the V_1/2 of inactivation at ∼1 MAC (Ouyang et al., 2009). The differential agent-specific inhibition displayed by the halogenated ethers is important. The higher relative ability of desflurane in blocking Na⁺ current compared with the other halogenated ethers is notable. The underlying mechanism of desflurane’s increased VGSC conduction inhibition is currently unknown.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

The halogenated ethers show differential inhibition of whole-cell Na⁺ current in Chinese hamster ovary cells expressing rat Na_v1.4 at 1 MAC. (A) Halothane; (B) isoflurane; (C) sevoflurane; (D) enflurane; (E) desflurane. Currents were evoked via 25-ms depolarizing pulses from a holding potential of −80 to −20 or −10 mV. Reproduced with permission from Ouyang et al. (2009).

Yokoyama et al. (2011) tested the effects of sevoflurane on rat Na_v1.4-, human Na_v1.7-, and rat Na_v1.8-mediated current expressed in oocytes. Sevoflurane (0.5 mM) did not cause a shift in activation or inactivation, nor did it influence peak current. Sevoflurane (1 mM) also did not change the current–voltage relationship, but did significantly inhibit the peak Na⁺ current of all three isoforms by 15%–19%. The authors hypothesized that sevoflurane inhibits VGSCs through a PKC-mediated pathway. Pretreatment with bisindolylmaleimide, a PKC inhibitor, reversed the effect of sevoflurane on all isoforms (Yokoyama et al., 2011).

V. Structural and Computational Studies Using Prokaryotic Voltage-Gated Sodium Channels

The successful crystallization of a eukaryotic VGSC has yet to be accomplished. Therefore, the exact amino acid residues of the human VGSCs that the halogenated ethers interact with can only be inferred. However, prokaryotic VGSCs such as Na_vAB from Arcobacter butzleri, Na_vMs from Magnetococcus marinus, and Na_vRh from Rickettsiales sp. HIMB114 have been crystallized (Payandeh et al., 2011; McCusker et al., 2012; Zhang et al., 2012). These bacterial sodium channels serve as powerful models to study the structure and function of eukaryotic VGSCs.

One bacterial sodium channel known as NaChBac, which has not yet been crystallized, from Bacillus halodurans is considered an exemplar that has been used as a model on which to base hypotheses of halogenated ether binding to a eukaryotic channel. Like all other bacterial VGSCs studied to date, NaChBac contains a single domain with six-transmembrane α-helices (see Fig. 5) (Catterall, 2001). Despite being one-tenth the amino acid size of a mammalian VGSC, NaChBac still conserves many of its integral functions. The channel similarly activates in response to voltage changes. Both activation and inactivation kinetics are slower for NaChBac compared with mammalian channels. The Na⁺ selectivity mechanism consists of a symmetric glutamate (EEEE) motif. This selectivity mechanism is used by voltage-gated calcium channels and diverges from the mammalian DEKA motif (Lipkind and Fozzard, 2008). Interestingly, NaChBac is not blocked by TTX, but is sensitive to the L-type calcium channel blocker class of dihydropyridines (Charalambous and Wallace, 2011). Its small size and homotetrameric organization make it a powerful tool to study VGSC structure, function, and pharmacology.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Comparing the eukaryotic and prokaryotic VGSC structures. (Left) Two-dimensional schematic of the eukaryotic VGSC α-subunit. (Right) Two-dimensional schematic of NaChBac, the prokaryotic VGSC α-subunit found in Bacillus halodurans.

Ouyang et al. (2007) performed whole-cell patch–clamp recordings with isoflurane on human embryonic kidney cells expressing NaChBac. Isoflurane caused a concentration-dependent inhibition of peak Na⁺ current (IC₅₀ = 0.35 mM at V_h= −80 mV). Both activation and inactivation kinetics were altered. At clinically relevant concentrations, isoflurane positively shifted the voltage dependence of activation. Consistent with earlier reports, isoflurane also shifted the V_1/2 of NaChBac inactivation in the hyperpolarizing direction. The investigators found NaChBac 8-fold more susceptible to isoflurane-induced enhanced inactivation compared with tonic block of the activated state of the channel (Ouyang et al., 2007). This NaChBac study provides important evidence of phylogenetic conservation for halogenated ether binding to VGSCs.

Raju et al. (2013) performed molecular dynamics simulations of isoflurane binding to NaChBac and proposed three distinct binding sites that emerge at clinically relevant concentrations. The first is an extracellular site at “the intersubunit interface between the P-loops” (Raju et al., 2013). The second is a linker site between the N and C termini. The authors hypothesized that multiple isoflurane molecules occupy this site simultaneously, with mainly nonpolar drug-channel interactions. The third site was deemed the pore site. Even though the molecular dynamics simulations were carried out with the activation gate of NaChBac closed, isoflurane was still able to access the ionic pore to form hydrophobic interactions with the channel. Two of these residues (T220 and F227) align with residues from mammalian VGSCs that are known to mediate LA binding (F1764 and Y1771 rat Na_v1.2 numbering) (Raju et al., 2013).

Using a combination of electrophysiology and molecular dynamics simulations, Barber et al. (2014) proposed that sevoflurane caused open channel blockade while also altering the activation and inactivation kinetics of NaChBac. Sevoflurane (0.2 mM) caused a −10 mV shift in the V_1/2 of activation and a −14 mV shift in the V_1/2 of inactivation. This resulted in a ∼15% increase in peak Na⁺ current at 0.2 mM sevoflurane and a ∼7% decrease at 2 mM (Barber et al., 2014). Sevoflurane also accelerated the time-dependent recovery from inactivation, whereas previous studies with halogenated ethers and mammalian VGSCs showed a delay in recovery from inactivation. Molecular dynamics simulations indicated sevoflurane binding to multiple sites. The authors noted a state-dependent interaction with the central cavity of the pore, leading them to emphasize an open channel block. Consistent with the findings of Raju et al. (2013), residues involved in LA binding (F1764 and Y1771 rat Na_v1.2 numbering) were highlighted by Barber et al. (2014) as sevoflurane binding sites.

Kinde et al. (2016) sought to experimentally validate binding of isoflurane to NaChBac by nuclear magnetic resonance. ¹⁹F probes were inserted into critical residues of the three putative binding sites (extracellular site, pore site, and linker site) suggested by the computational studies (Raju et al., 2013; Barber et al., 2014). ¹⁹F NMR saturation transfer spectroscopy confirmed multiple major sites of isoflurane interaction with NaChBac. The study revealed two primary mechanisms by which isoflurane inhibits the channel. One mechanism is at the inner ring of the selectivity filter, occluding the pore. The other is at a hinge site, in which isoflurane binds to the S4–S5 linker that controls inactivation gating via interactions with S6 (Kinde et al., 2016).

The effects of isoflurane of NaChBac gating via site-directed mutagenesis were recently explored (Sand et al., 2017). After expressing NaChBac in human embryonic kidney cells, two point mutations were introduced to residues involved in activation: G229A, which inhibits activation, and G219A, which enhances activation. Their analysis suggested isoflurane primarily inhibits NaChBac via enhanced activation and inactivation, not through an open channel block of NaChBac (Sand et al., 2017).

A recent study using whole-cell patch clamp examining the effects of propofol on NaChBac showed that this anesthetic also inhibits peak Na⁺ current of the prokaryotic channel at clinically relevant concentrations (Yang et al., 2018). The authors propose that propofol acts via a hyperpolarizing shift in steady-state inactivation. Molecular dynamics simulations with propofol suggest it directly interacts with NaChBac at multiple distinct sites on the channel (Wang et al., 2018). These results indicate that VGSC inhibition may be involved in the mechanism of action of anesthetic agents beyond the halogenated ethers.

VI. Presynaptic Voltage-Gated Sodium Channel Inhibition Leads to Depressed Neurotransmission

What are the neurochemical consequences of activity-dependent VGSC inhibition by halogenated ethers? Hemmings (2009) has proposed a mechanism by which presynaptic VGSC inhibition leads to reduced depolarization of the presynaptic terminal. The VGSC inhibition reduces Ca²⁺ influx into the terminal via voltage-gated calcium channels (Hemmings, 2009). Reduced Ca²⁺ influx ultimately leads to reduced exocytosis and neurotransmission (see Fig. 6). Previous studies have revealed that isoflurane and other volatile anesthetics suppress synaptic vesicle exocytosis and neurotransmission via presynaptic actions (Schlame and Hemmings, 1995; Maclver et al., 1996; Hemmings et al., 2005a; Ouyang and Hemmings, 2005). Due to most neurons in the human brain utilizing glutamatergic neurotransmission, this mechanism would have important consequences for neural networks if occurring during volatile anesthesia in vivo.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 6.

Basic overview of voltage-gated Na⁺ channels in synaptic transmission. (1) At rest, presynaptic voltage-gated Na⁺ and Ca²⁺ channels are closed. (2) Voltage-gated Na⁺ channels open, driving membrane depolarization and action potential propagation. (3) Voltage-gated Ca²⁺ channels open, raising intracellular levels of Ca²⁺. (4) Elevated intracellular Ca²⁺ triggers exocytosis. (5) Neurotransmitters are released into the synaptic cleft, activating postsynaptic receptors leading to a depolarization of the postsynaptic membrane. Due to the instrumental role voltage-gated Na⁺ channels play in this process, their inhibition by the halogenated ethers is likely to lead to dysfunction in neurotransmitter release.

Studies conducted by Berg-Johnsen and Langmoen (1986, 1990, 1992) revealed a depression of neuronal excitability and excitatory synaptic transmission by isoflurane in the rat hippocampus and human neocortex. Using isolated rat cerebrocortical nerve terminals, Westphalen and Hemmings (2006a,b) showed that isoflurane inhibition of presynaptic VGSCs was more highly coupled to glutamate release than GABA. By measuring the release of radiolabeled glutamate and GABA evoked by either 4-aminopyridine (4-AP; VGSC-dependent release) or elevated levels of extracellular K⁺ (VGSC-independent release), the investigators found the VGSC-dependent release using 4-AP more sensitive to inhibition by isoflurane than VGSC-independent release using elevated extracellular K⁺. Moreover, the IC₅₀ values for 4-AP–evoked release support the preferential inhibition of isoflurane for glutamate over GABA release. Isoflurane IC₅₀ values of 4-AP–evoked release were 0.44 and 0.58 mM for glutamate and GABA, respectively (Westphalen and Hemmings, 2003; Hemmings, 2009).

Additionally, even a small inhibition of the action amplitude in the presynaptic terminal induced by isoflurane inhibits excitatory postsynaptic current amplitude and exocytosis in the rat calyx of Held synapse with no significant effect on Ca²⁺ currents. Wu et al. (2004) reported that a majority (∼70%) of the isoflurane-induced inhibition of the excitatory postsynaptic current resulted from inhibition of the presynaptic action potential. In cultured rat hippocampal neurons, isoflurane significantly inhibits synaptic vesicle exocytosis through a reduction in presynaptic Ca²⁺ influx without affecting the Ca²⁺ sensitivity of exocytosis. This effect was more pronounced when targeting glutamatergic over GABAergic neurons (Baumgart et al., 2015).

It has long been observed that ether and some halogenated derivatives display a depression of excitatory synaptic transmission that varies in different brain regions (Richards and White, 1975; Richards et al., 1975). The suppression of neurotransmitter release by isoflurane was shown to have regional specificity (Westphalen et al., 2011). This regional specificity could, in part, result from the differential localization of VGSC subtypes within nerve terminals of different cell types and their expression patterns within various regions of the brain (Westphalen et al., 2010). Isoflurane was recently shown to inhibit persistent and transient Na⁺ current in hippocampal excitatory pyramidal neurons from mice at clinically relevant concentrations (Zhao et al., 2019). Moreover, the differential localization of VGSC subtypes in pre- and postsynaptic compartments most likely contributes to localized variations in neurotransmitter suppression observed (Johnson et al., 2017).

VII. In Vivo Modulation of Halogenated Ether Anesthesia

B. Modulation of Anesthetic Sensitivity Mediated by Genetic Manipulation of Voltage-Gated Sodium Channels

Na_v1.6 (encoded by SCN8A) is the most abundantly expressed VGSC subtype in the adult central nervous system (Goldin, 2001). The subtype is highly localized to the neuronal axon initial segment and nodes of Ranvier (Lorincz and Nusser, 2010). As a major contributor to neuronal excitability, Na_v1.6 is a subtype of interest in halogenated ether anesthesia. Pal et al. (2015) showed that Na_v1.6 was important for mediating in vivo sensitivity to isoflurane and sevoflurane anesthesia induction. Using mice with reduced Na_v1.6 activity (SCN8A^medJ/medJ and SCN8A^9J/9J mice), the authors found the mutant mice to be significantly more sensitive to isoflurane and sevoflurane than their wild-type littermate controls (see Fig. 7). The SCN8A^medJ/medJ mutation does not change the amino acid sequence, but reduces expression levels of Na_v1.6 by 90%–95% of that of wild-type levels by disrupting splicing efficiency. The SCN8A^9J/9J mutation is an in-frame deletion in segment 6 of domain IV, resulting in a dramatic decrease in channel activity (Pal et al., 2015). The concentration required for 50% of tested mice to lose their righting reflex (IC₅₀) was significantly lower for the SCN8A^medJ/medJ and SCN8A^9J/9J mice compared with their littermate controls.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 7.

Increased anesthetic sensitivity in mice with loss-of-function mutations in Scn8a (the gene that encodes Na_v1.6). Best-fit dose–response curves along with 95% confidence band (dashed lines) show the behavioral response (loss of righting reflex) of Scn8a^medJ/medJ (A and B) and Scn8a^9J/9J (C and D) mice and their respective littermate control groups to increasing concentrations of isoflurane and sevoflurane anesthesia. EC₅₀ shows a statistically significant leftward shift in the mutant mice, indicating hypersensitivity to anesthesia mediated by Na_v1.6. Reproduced with permission from Pal et al. (2015).

Notably, the time to emerge from anesthesia (return of righting reflex) was not different in the mutant mice versus control under isoflurane. There was a significant difference in the return of righting reflex after sevoflurane for SCN8A^9J/9J mice versus wild-type. Intracranial electroencephalogram was performed on the mice. Significant reductions in θ power for both SCN8A^medJ/medJ and SCN8A^9J/9J mice in the waking state compared with controls excluded sleep deprivation as an explanation for the increased anesthetic sensitivity. Increased θ power is a marker for sleep deprivation in the waking state of mice. The differential effect on induction sensitivity supports a potential role for VGSC involvement in the maintenance of consciousness. The authors noted that differential expression patterns of Na_v1.6 could lead to nonspecific physiologic effects (Pal et al., 2015).

An accumulating body of evidence indicates halogenated ethers inhibit sodium channels via stabilization of the fast-inactivated state. Purtell et al. (2015) used the prototypical halogenated ether isoflurane to further investigate this effect on rat Na_v1.2-mediated sodium current in ND7/23 cells. To eliminate the endogenous VGSC current in the neuronal rodent line, a phenylalanine to serine mutation (F385S) was introduced to rat Na_v1.2 to render it TTX resistant, and whole-cell recordings were performed in the presence of 250 nM TTX. A modest reduction of peak Na⁺ current (∼10%) was observed at 1.5 MAC isoflurane. A ∼−4 mV shift in the V_1/2 of steady-state inactivation was seen with no alterations in activation. The time-dependent recovery from inactivation was delayed following treatment with 1.5 MAC (Purtell et al., 2015). After observing that the delayed recovery from inactivation was more pronounced at depolarized holding potentials, the authors tested the activity-dependent effects of isoflurane. Following 10 depolarizing 5-ms pulses at 50 Hz (V_h of −70 or −90 stepping to 0 mV), 1.5 MAC isoflurane significantly reduced the fraction of Na⁺ current at pulse 10 to pulse 1. The reduction in activity-dependent Na⁺ current was increased when starting from a V_h of −70 mV instead of V_h of −90 mV. Because V_h of −70 mV is closer to the V_1/2 of inactivation, more VGSCs will be inactivated during these protocols than when holding at V_h of −90 mV. Therefore, the larger reduction in activity-dependent Na⁺ current further confirms isoflurane preferentially inhibits channels in the inactivated state. Increasing the frequency from 50 to 100 Hz further increased the block by isoflurane, although only a subset of neurons is capable of firing at this rate in vivo, including fast-spiking GABAergic interneurons. Using the same protocol, the activity-dependent inhibition by isoflurane was also observed in rat neurohypophysial nerve terminals. The authors then constructed a three-state Markov model with the data obtained from their electrophysiology experiments. The model provided quantitative support of the hypothesis that the activity-dependent block of isoflurane is achieved via stabilization of the inactivated state (Purtell, 2015; Purtell et al., 2015). The novel activity-dependent action of isoflurane on Na_v1.2 is nonetheless important as it is one of the major VGSC isoforms expressed in the central nervous system. The authors claim the additional ∼20% increase in Na⁺ current block at 50 Hz stimulation from V_h of −70 mV has major implications for the in vivo systems level mechanism of isoflurane. If isoflurane increasingly blocks VGSCs as neurons fire more frequently, information-rich networks of fast-firing neurons will be differentially suppressed. Recent experiments have provided supporting evidence that isoflurane is a subtype-selective inhibitor of the neuronal VGSCs. In ND7/23 cells, clinically relevant concentrations of isoflurane showed a voltage-dependent reduction (between 15% and 20%) of peak Na⁺ current mediated by Na_v1.3 and Na_v1.6, but no significant effect was observed in cells expressing Na_v1.1 (Zhou et al., 2019). Analysis of the activity other halogenated ethers have on the neuronal subtypes Na_v1.1, 1.3, and 1.6 will be important to determine whether this activity-dependent and subtype-selective effect is agent-specific.