Friday, August 1, 2008

Basic Anesthesia 1

Cellular and Molecular Mechanisms of Anesthesia

Alex S. Evers
C. Michael Crowder

Barash, Paul G.; Cullen, Bruce F.; Stoelting, Robert K. Title: Clinical Anesthesia, 5th Edition
Copyright ©2006 Lippincott Williams & Wilkins

The introduction of general anesthetics into clinical practice 150 years ago stands as one of the seminal innovations of medicine. This single discovery facilitated the development of modern surgery and spawned the specialty of anesthesiology. Despite the importance of general anesthetics and despite over 100 years of active research, the molecular mechanisms responsible for anesthetic action remain one of the unsolved mysteries of pharmacology.

Why have mechanisms of anesthesia been so difficult to elucidate? Anesthetics, as a class of drugs, are challenging to study for three major reasons:

1. Anesthesia, by definition, is a change in the responses of an intact animal to external stimuli. Making a definitive link between anesthetic effects observed in vitro and the anesthetic state observed and defined in vivo has proven difficult.

2. No structure-activity relationships are apparent among anesthetics; a wide variety of structurally unrelated compounds, ranging from steroids to elemental xenon, are capable of producing clinical anesthesia. This suggests that there are multiple molecular mechanisms that can produce clinical anesthesia.

3. Anesthetics work at very high concentrations in comparison to drugs, neurotransmitters, and hormones that act at specific receptors. This implies that if anesthetics do act by binding to specific receptor sites, they must bind with very low affinity and probably stay bound to the receptor for very short periods of time. Low-affinity binding is much more difficult to observe and characterize than high-affinity binding.

Despite these difficulties, molecular and genetic tools are now available that should allow for major insights into anesthetic mechanisms in the next decade. The aim of this chapter is to provide a conceptual framework for the reader to catalog current knowledge and integrate future developments about mechanisms of anesthesia. Five specific questions will be addressed in this chapter:

1. What is anesthesia and how do we measure it?

2. What is the anatomic site of anesthetic action in the central nervous system?

3. What are the cellular neurophysiologic mechanisms of anesthesia (e.g., effects on synaptic function versus effects on action potential generation) and what anesthetic effects on ion channels and other neuronal proteins underlie these mechanisms?

4. What are the molecular targets of anesthetics?

5. How are the molecular and cellular effects of anesthetics linked to the behavioral effects of anesthetics observed in vivo?


General anesthesia can broadly be defined as a drug-induced reversible depression of the central nervous system resulting in the loss of response to and perception of all external stimuli. Unfortunately, such a broad definition is inadequate for two reasons. First, the definition is not actually broad enough. Anesthesia is not simply a deafferented state; amnesia and unconsciousness are important aspects of the anesthetic state. Second, the definition is too broad, as all general anesthetics do not produce equal depression of all sensory modalities. For example, barbiturates are considered to be anesthetics, but they are not particularly effective analgesics. These conflicting problems with definition can be bypassed by a more practical description of the anesthetic state as a collection of “component” changes in behavior or perception. The components of the anesthetic state include unconsciousness, amnesia, analgesia, immobility, and attenuation of autonomic responses to noxious stimulation.

Regardless of which definition of anesthesia is used, essential to anesthesia are drug-induced changes in behavior or perception. As such, anesthesia can only be defined and measured in the intact organism. Changes in behavior such as unconsciousness or amnesia can be intuitively understood in higher organisms such as mammals, but become increasingly difficult to define as one descends the phylogenetic tree. Thus, while anesthetics have effects on organisms ranging from worms1 to man, it is difficult to map with certainty the effects of anesthetics observed in lower organisms to any of our behavioral definitions of anesthesia. This contributes to the difficulty of using simple organisms as models in which to study the molecular mechanisms of anesthesia. Similarly, any cellular or molecular effects of anesthetics observed in higher organisms can be extremely difficult to link with the constellation of behaviors that constitute the anesthetic state. The absence of a simple and concise definition of anesthesia is clearly one of the stumbling blocks to elucidating the mechanisms of anesthesia at a molecular and cellular level.


To study the pharmacology of anesthetic action, quantitative measurements of anesthetic potency are absolutely essential. To this end, Eger and colleagues have defined the concept of MAC, or minimum alveolar concentration. MAC is defined as the alveolar partial pressure of a gas at which 50% of humans do not respond to a surgical incision.2 In animals, MAC is defined as the alveolar partial pressure of a gas at which 50% of animals do not respond to a noxious stimulus, such as tail clamp,3 or at which they lose their righting reflex. The use of MAC as a measure of anesthetic potency has two major advantages. First, it is an extremely reproducible measurement that is remarkably constant over a wide range of species.2 Second, the use of end-tidal gas concentration provides an index of the “free” concentration of drug required to produce anesthesia since the end-tidal gas concentration is in equilibrium with the free concentration in plasma. The MAC concept has several important limitations, particularly when trying to relate MAC values to anesthetic potency observed in vitro. First, the end point in a MAC determination is quantal: a subject is either anesthetized or unanesthetized; it cannot be partially anesthetized. Furthermore, MAC represents the average response of a whole population of subjects rather than the response of a single subject. The quantal nature of the MAC measurement makes it very difficult to compare MAC measurements to concentration-response curves obtained in vitro, where the graded response of a single preparation is measured as a function of anesthetic concentration. The second limitation of MAC measurements is that they can only be directly applied to anesthetic gases. Parenteral anesthetics (barbiturates, neurosteroids, propofol) cannot be assigned a MAC value, making it difficult to compare the potency of parenteral and volatile anesthetics. A MAC equivalent for parenteral anesthetics is the free concentration of the drug in plasma required to prevent response to a noxious stimulus in 50% of subjects; this value has been estimated for several parenteral anesthetics.4 A third limitation of MAC is that it is highly dependent on the anesthetic end point used to define it. For example, if loss of response to a verbal command is used as an anesthetic end point, the MAC values obtained (MACawake) will be much lower than classic MAC values based on response to a noxious stimulus. Indeed, each behavioral component of the anesthetic state will likely have a different MAC value. Despite its limitations, MAC remains the most robust measurement and the standard for determining the potency of volatile anesthetics.

The foregoing discussion of MAC brings forth an important and somewhat controversial question. What drug concentration should be measured when determining anesthetic potency? When measuring potency of intravenous anesthetics, the answer to this question is relatively simple. One would like to relate the free concentration of the drug at its site of action (the biophase) to the drug's effect. It is, of course, not practical to measure the drug's concentration in the extracellular fluid of the brain, so free concentration in plasma is used as an approximation of the biophase concentration. This allows one to compare the concentration of drug required to produce anesthesia in humans to the concentrations required to produce specific effects in vitro. With the volatile anesthetics, potency is defined by MAC, which is measured in units of partial pressure. Because the partial pressure of a dissolved gas is directly proportional to the free concentration of that gas in a liquid, alveolar partial pressures are accurate reporters of the free anesthetic concentrations in plasma and in brain tissue.


In principle, general anesthesia could result from interruption of nervous system activity at myriad levels. Plausible targets include peripheral sensory receptors, spinal cord, brainstem, and cerebral cortex. Of these potential sites, only peripheral sensory receptors can be eliminated as an important site of anesthetic action. Animal studies have shown that fluorinated volatile anesthetics have no effect on cutaneous mechanosensors in cats5 and can even sensitize nociceptors in monkeys.6 Furthermore, selective perfusion studies in dogs have shown that MAC for isoflurane is unaffected by the presence or absence of isoflurane at the site of noxious stimulation, provided that the central nervous system is perfused with blood containing isoflurane.7

Spinal Cord

Clearly, anesthetic actions on the spinal cord cannot produce either amnesia or unconsciousness. However, several lines of evidence indicate that the spinal cord is probably the site at which anesthetics act to inhibit purposeful responses to noxious stimulation. This is, of course, the end point used in most measurements of anesthetic potency. Rampil and colleagues have shown that MAC values for fluorinated volatile anesthetics are unaffected in the rat by either decerebration8 or cervical spinal cord transection.9 Antognini and colleagues have used the strategy of isolating the cerebral circulation of goats to explore the contribution of brain and spinal cord to the determination of MAC. They found that when isoflurane is administered only to the brain, MAC is 2.9%, whereas when it is administered to the entire body, MAC is 1.2%.10 Surprisingly, when isoflurane was preferentially administered to the body and not to the brain, isoflurane MAC was reduced to 0.8%.11 The actions of volatile anesthetics in the spinal cord are mediated, at least in part, by direct effects on the excitability of spinal motor neurons. This conclusion has been substantiated by experiments in rats,12 goats,13 and humans,14 showing that volatile anesthetics depress the amplitude of the F wave in evoked potential measurements (F-wave amplitude correlates with motor neuron excitability). These provocative results suggest not only that anesthetic actions at the spinal cord underlie the determination of MAC, but also that anesthetic actions on the brain may actually sensitize the cord to noxious stimuli. The plausibility of the spinal cord as a locus for anesthetic immobilization is also supported by several electrophysiological studies showing inhibition of excitatory synaptic transmission in the spinal cord.15,16,17,18

Reticular Activating System

The reticular activating system, a diffuse collection of brainstem neurons involved in arousal behavior, has long been speculated to be a site of general anesthetic action on consciousness. Evidence to support this notion came from early whole animal experiments showing that electrical stimulation of the reticular activating system could induce arousal behavior in anesthetized animals.19 A role for the brainstem in anesthetic action is also supported by studies examining somatosensory evoked potentials. Generally, these studies show that anesthetics produce increased latency and decreased amplitude of cortical potentials, indicating that anesthetics inhibit information transfer through the brainstem.20 In contrast, studies using brainstem auditory evoked potentials have shown variable effects ranging from depression to enhancement of information transfer through the reticular formation.21,22 ,23 While there is evidence that the reticular formation of the brainstem is a locus for anesthetic effects, it cannot be the only anatomic site of anesthetic action for two reasons. First, as discussed earlier, the brainstem is not even required for anesthetics to inhibit responsiveness to noxious stimuli. Second, the reticular formation can be largely ablated without eliminating awareness.24

Within the reticular formation is a set of pontine noradrenergic neurons called the locus coeruleus (LC). The LC innervates a number of targets in basal forebrain and cortex including a set of GABAergic hypothalamic neurons called the ventrolateral preoptic nucleus (VLPO). The VLPO in turn innervates the tuberomammillary nucleus (TMN). The LC-VLPO-TMN pathway has been shown to be critical for non-REM sleep. Given that EEG patterns under anesthesia and non-REM sleep are quite similar, this pathway is a particularly good candidate for a site of anesthetic action. Using stereotactic techniques, Maze and colleagues tested this hypothesis by measuring whether application of a GABAergic antagonist directly onto the TMN altered the efficacy of anesthetics.25 Indeed, discrete application of the GABAergic antagonist gabazine onto the TMN markedly reduced the duration of sedation produced by systemically administered propofol or pentobarbital. This effect is unlikely to be a consequence of a nonspecific increase in arousal state because systemically administered gabazine did not antagonize the potency of ketamine whereas it did antagonize propofol and pentobarbital in a manner similar to application directly onto the TMN. This result strongly implicates the TMN as a site for the sedative action of GABAergic anesthetics like propofol and barbiturates.

Cerebral Cortex

The cerebral cortex is the major site for integration, storage, and retrieval of information. As such, it is a likely site at which anesthetics might interfere with complex functions like memory and awareness. Anesthetics clearly alter cortical electrical activity, as evidenced by the changes in surface EEG patterns recorded during anesthesia. Anesthetic effects on patterns of cortical electrical activity vary widely among anesthetics,26 providing an initial suggestion that all anesthetics are not likely to act through identical mechanisms. More detailed in vitro electrophysiological studies examining anesthetic effects on different cortical regions support the notion that anesthetics can differentially alter neuronal function in various cortical preparations. For example, volatile anesthetics have been shown to inhibit excitatory transmission at some synapses in the olfactory cortex27 but not at others.28 Similarly, whereas volatile anesthetics inhibit excitatory transmission in the dentate gyrus of the hippocampus,29 these same drugs can actually enhance excitatory transmission at other synapses in the hippocampus.30 Anesthetics also produce a variety of effects on inhibitory transmission in the cortex. A variety of parenteral and inhalational anesthetics have been shown to enhance inhibitory transmission in olfactory cortex28 and in the hippocampus.31 Conversely, volatile anesthetics have also been reported to depress inhibitory transmission in hippocampus.32 One area of the brain that has been postulated as a potential site of anesthetic action is the thalamus. The thalamus is important in relaying sensory modalities and motor information to the cortex via thalamocortical pathways. A developing body of evidence indicates that inhalational anesthetics can depress the excitability of thalamic neurons, thus blocking thalamocortical communication potentially resulting in loss of consciousness.33


Anesthetics are able to produce effects on a variety of anatomic structures in the CNS, including spinal cord, brainstem, and cerebral cortex. Whereas certain anesthetic effects may be attributable to specific anatomic locations (e.g., purposeful response to noxious stimulation maps to the spinal cord), existing evidence provides no basis for a single anatomic site responsible for anesthesia. This difficulty in identifying a site for anesthesia might plausibly result from the various components of the anesthetic state being produced by anesthetic effects on different regions of the CNS. Nevertheless, despite the difficulty in identifying a common anatomic site for anesthesia, investigators have continued to look for other unifying principles in anesthetic action. Specifically, attention has been focused on identifying common cellular or molecular anesthetic targets that may have a wide anatomic distribution, explaining the ability of anesthetic to affect nervous system function in an anatomically diffuse manner.


In the simplest terms anesthetics inhibit or “turn off” vital central nervous system functions. They must do this by acting at specific physiologic “switches.” A great deal of investigative effort has been devoted to identifying these switches. In principle, the CNS could be switched off by several means:

1. By depressing those neurons or pattern generators that subserve a pacemaker function in the CNS,

2. By reducing overall neuronal excitability; either by changing resting membrane potential or by interfering with the processes involved in generating an action potential,

3. By reducing communication between neurons—specifically, by either inhibiting excitatory synaptic transmission or enhancing inhibitory synaptic transmission.

Pattern Generators

Information concerning the effects of anesthetics on pattern-generating neuronal circuits in the CNS is limited, but clinical concentrations of anesthetics are likely to have significant effects on these circuits. The simplest evidence for this is the observation that most anesthetics exert profound effects on respiratory rate and rhythm, strongly suggesting an effect on respiratory pattern generators in the brainstem. Invertebrate studies suggest that volatile anesthetics can selectively inhibit the spontaneous (pacemaker) firing of specific neurons. As shown in Fig. 1, halothane (1.0 MAC) completely inhibits spontaneous action potential generation by one neuron in the right parietal ganglion of the great pond snail while producing no observable effect on the firing frequency of adjacent neurons.34

FIGURE 1. Selectivity of volatile anesthetic inhibition of neuronal automaticity. Halothane (1 MAC) reversibly inhibits the spontaneous firing activity of a neuron from the parietal ganglion of Lymnaea stagnalis (A). The same concentration of halothane has no effect on the firing activity of an adjacent, and apparently identical, neuron (B). Note that in (A) halothane markedly reduces resting membrane potential in addition to inhibiting firing. (From Franks NP, Lieb WR. Mechanisms of general anesthesia. Environ Health Perspect. 1990;87:204.)

Neuronal Excitability

The ability of a neuron to generate an action potential is determined by three parameters: resting membrane potential, the threshold potential for action potential generation, and the function of voltage-gated sodium channels. Anesthetics can hyperpolarize (create a more negative resting membrane potential) both spinal motor neurons and cortical neurons,35,36 and this ability to hyperpolarize neurons correlates with anesthetic potency. In general, the increase in resting membrane potential produced by anesthetics is small in magnitude and is unlikely to have an effect on axonal propagation of an action potential. Small changes in resting potential may, however, inhibit the initiation of an action potential either at a postsynaptic site or in a spontaneously firing neuron. Indeed, hyperpolarization is responsible for the inhibition of spontaneous action potential generation shown in Fig. 1. Recent evidence also indicates that isoflurane hyperpolarizes thalamic neurons, leading to an inhibition of tonic firing of action potentials.33 There is no evidence indicating that anesthetics alter the threshold potential of a neuron for action potential generation.

However, the data is conflicting on whether the size of the action potential, once initiated, is diminished by general anesthetics. A classic paper by Larrabee and Posternak demonstrated that concentrations of ether and chloroform that completely block synaptic transmission in mammalian sympathetic ganglia have no effect on presynaptic action potential amplitude.37 Similar results have been obtained with fluorinated volatile anesthetics in several mammalian brain preparations.27,29 This dogma that the action potential is relatively resistant to general anesthetics has been challenged by more recent reports that volatile anesthetics at clinical concentrations produce a small but significant reduction in the size of the action potential in mammalian neurons.38,39 In one case, the reduction in the action potential was shown to be amplified at the presynaptic terminal resulting in a large reduction in neurotransmitter release.39 Thus, while current data still support the prevailing view that neuronal excitability is only slightly affected by general anesthetics, this small effect may nevertheless contribute significantly to the clinical actions of volatile anesthetics.

Synaptic Function

Synaptic function is widely considered to be the most likely subcellular site of general anesthetic action. Neurotransmission across both excitatory and inhibitory synapses has been found to be markedly altered by general anesthetics. General anesthetics have been shown to inhibit excitatory synaptic transmission in a variety of preparations, including sympathetic ganglia,37 olfactory cortex,27 hippocampus,29 and spinal cord.17 However, not all excitatory synapses appear to be equally sensitive to anesthetics; indeed, transmission across some hippocampal excitatory synapses has been shown to be enhanced by inhalational anesthetics.30 In a similar fashion, general anesthetics have been shown both to enhance and depress inhibitory synaptic transmission in various preparations. In a classic paper in 1975, Nicoll and colleagues showed that barbiturates enhanced inhibitory synaptic transmission by prolonging the decay of the GABAergic inhibitory postsynaptic current.40 Enhancement of inhibitory transmission has also been observed with many other general anesthetics, including etomidate,41 propofol,42 inhalational anesthetics,28 and neurosteroids.43 Although anesthetic enhancement of inhibitory currents has received a great deal of attention as a potential mechanism of anesthesia,4 it is important to note that there is also a large body of experimentation showing that clinical concentrations of general anesthetics can depress inhibitory postsynaptic potentials in the hippocampus32,44,45 and in the spinal cord.18 Anesthetics do appear to have preferential effects on synapses, but there is a great deal of heterogeneity in the manner in which anesthetic agents affect different synapses. This is not surprising given the large variation in synaptic structure, function (i.e., efficacy), and chemistry (neurotransmitters, modulators) extant in the nervous system.

Presynaptic Effects

General anesthetics affect synaptic transmission both pre- and postsynaptically. However, the magnitude and even the type of effect vary according to the type of synapse and the particular anesthetic. Presynaptically, neurotransmitter release from glutamatergic synapses has consistently been found to be inhibited by clinical concentrations of volatile anesthetics. For example, a study by Perouansky and colleagues conducted in mouse hippocampal slices showed that halothane inhibited excitatory postsynaptic potentials elicited by presynaptic electrical stimulation, but not those elicited by direct application of glutamate. This indicates that halothane must be acting to prevent the release of glutamate, the major excitatory neurotransmitter in the brain.46 MacIver and colleagues extended these observations by providing evidence that the inhibition of glutamate release from hippocampal neurons is not due to effects at GABAergic synapses that could indirectly decrease transmitter release from glutamatergic neurons. Effects of intravenous anesthetics on glutamate release have also been demonstrated but the evidence is more limited and the effects potentially indirect.47,48 The data for anesthetic effects on inhibitory neurotransmitter release is mixed. Inhibition,49 stimulation,50,51 and no effect52 have been reported for volatile anesthetic and intravenous anesthetic action on GABA release. In a brain synaptosomal preparation where effects on both GABA and glutamate release could be studied simultaneously, Hemmings and coworkers found that glutamate and, to a lesser degree, GABA release were inhibited by clinical concentrations of isoflurane.53 The mechanism underlying anesthetic effects on transmitter release have not been established. The effects of anesthetics on neurotransmitter release do not appear to be mediated by reduced neurotransmitter synthesis or storage, but rather by a direct effect on the process of neurosecretion. A variety of evidence argues that at some synapses the majority of the anesthetic effect is upstream of the transmitter release machinery, perhaps on presynaptic sodium channels (see discussion later). However, genetic data in C. elegans shows that the transmitter release machinery strongly influences volatile anesthetic sensitivity;54,55 at present, it is unclear whether these findings represent species differences or different aspects of the same mechanism.

Postsynaptic Effects

Anesthetics also alter the postsynaptic response to released neurotransmitter. The effects of general anesthetics on excitatory neurotransmitter receptor function vary depending on neurotransmitter type, anesthetic agent, and preparation. Richards and Smaje examined the effects of several anesthetic agents on the response of olfactory cortical neurons to application of glutamate, the major excitatory neurotransmitter in the CNS.5657 The effects of anesthetics on neuronal responses to They found that while pentobarbital, diethyl ether, methoxyflurane, and alphaxalone depressed the electrical response to glutamate, halothane was without effect. In contrast, when acetylcholine was applied to the same olfactory cortical preparation, halothane and methoxyflurane stimulated the electrical response whereas pentobarbital had no effect; only alphaxalone depressed the electrical response to acetylcholine. inhibitory neurotransmitters are more consistent. A wide variety of anesthetics, including barbiturates, etomidate, neurosteroids, propofol, and the fluorinated volatile anesthetics, have been shown to enhance the electrical response to exogenously applied GABA (for a review, see 58). For example, Fig. 2 illustrates the ability of enflurane to increase both the amplitude and the duration of the current elicited by application of GABA to hippocampal neurons.59

FIGURE 2. Enflurane potentiates the ability of GABA to activate a chloride current in cultured rat hippocampal cells. This potentiation is rapidly reversed by removal of enflurane (wash) (Panel A). Enflurane increases both the amplitude of the current (Panel B) and the time (t1/2) it takes for the current to decay (Panel C). (Reproduced with permission from Jones MV, Brooks PA, Harrison L. Enhancement of y-aminobutyric acid-activated Cl- currents in cultured rat hippocampal neurones by three volatile anaesthetics. J Physiol. 1992;449:289.)


Attempts to identify a physiologic “switch” at which anesthetics act have suffered from their own success. Anesthetics produce a variety of effects on many physiologic processes that might logically contribute to the anesthetic state, including neuronal automaticity, neuronal excitability, and synaptic function. The synapse is generally thought to be the most likely relevant site of anesthetic action. Existing evidence indicates that even at this one site, anesthetics produce various effects, including presynaptic inhibition of neurotransmitter release, inhibition of excitatory neurotransmitter effect, and enhancement of inhibitory neurotransmitter effect. Furthermore, the effects of anesthetics on synaptic function differ among various anesthetic agents, neurotransmitters, and neuronal preparations.


Ion channels are one likely target of anesthetic action. The advent of patch clamp techniques in the early 1980s made it possible to directly measure the currents from single ion channel proteins. It was attractive to think that anesthetic effects on a small number of ion channels might help to explain the complex physiologic effects of anesthetics that we have already described. Accordingly, during the 1980s and 1990s a major effort was directed at describing the effects of anesthetics on the various kinds of ion channels. The following section summarizes and distills this effort. For the purposes of this discussion, ion channels are cataloged according to the stimuli to which they respond by opening or closing (i.e., their mechanism of gating).

Anesthetic Effects on Voltage-Dependent Ion Channels

A variety of ion channels can sense a change in membrane potential and respond by either opening or closing their pore. These channels include voltage-dependent sodium, potassium, and calcium channels, all of which share significant structural homologies. Voltage-dependent sodium and potassium channels are largely involved in generating and shaping action potentials. The effects of anesthetics on these channels have been extensively studied by Haydon and colleagues in the squid giant axon.6061 These studies show that these invertebrate sodium channels and potassium channels are remarkably insensitive to volatile anesthetics. For example, 50% inhibition of the peak sodium channel current required halothane concentrations 8 times those required to produce anesthesia. The delayed rectifier potassium channel was even less sensitive, requiring halothane concentrations more than 20 times those required to produce anesthesia. Similar results have been obtained in a mammalian cell line (GH3 pituitary cells) where both sodium and potassium currents were inhibited by halothane only at concentrations greater than 5 times those required to produce anesthesia.62 However, a number of studies with volatile anesthetics have challenged the notion that voltage-dependent sodium channels are insensitive to anesthetics. Rehberg and colleagues expressed rat brain IIA sodium channels in a mammalian cell line and showed that clinically relevant concentrations of a variety of inhalational anesthetics suppressed voltage-elicited sodium currents.63 Hemmings and coworkers showed that sodium flux mediated by rat brain sodium channels was significantly inhibited by clinical concentrations of halothane.64 Harris and colleagues documented the effects of isoflurane on a variety of sodium channel subtypes and found that several but not all subtypes are sensitive to clinical concentrations.65 Finally as described above, in a rat brainstem neuron Wu and colleagues found that a small inhibition of sodium currents by isoflurane resulted in a large inhibition of synaptic activity.39 Thus, sodium channel activity not only appears to be inhibited by volatile anesthetics, but this inhibition results in a significant reduction in synaptic function, at least at some mammalian synapses. Intravenous anesthetics have also been shown to inhibit sodium channels, but the concentrations for this effect are supraclinical.66,67,68

Voltage-dependent calcium channels (VDCC) serve to couple electrical activity to specific cellular functions. In the nervous system, VDCCs located at presynaptic terminals respond to action potentials by opening. This allows calcium to enter the cell, activating calcium-dependent secretion of neurotransmitter into the synaptic cleft. At least six types of calcium channels (designated L, N, P, Q, R, and T) have been identified on the basis of electrophysiological properties and a larger number based on amino acid sequence similarities.69 N-, P-, Q-, and R-type channels, as well as some of the untitled channels, are preferentially expressed in the nervous system and are thought to play a major role in synaptic transmission. L-type calcium channels, although expressed in the brain, have been best studied in their role in excitation-contraction coupling in cardiac, skeletal, and smooth muscle and are thought to be less important in synaptic transmission. The effects of anesthetics on L- and T-type currents have been well characterized,62,70,71 and there are some reports concerning the effects of anesthetics on N- and P-type currents.72,73,74 As a general rule, these studies have shown that volatile anesthetics inhibit VDCCs (50% reduction in current) at concentrations 2 to 5 times those required to produce anesthesia in humans, with less than a 20% inhibition of calcium current at clinical concentrations of anesthetics (Fig. 3). However, some studies have found VDCCs that are extremely sensitive to anesthetics. Takenoshita and Steinbach reported a T-type calcium current in dorsal root ganglion neurons that was inhibited by subanesthetic concentrations of halothane.75 Additionally, ffrench-Mullen and colleagues have reported a VDCC of unspecified type in guinea pig hippocampus that is inhibited by pentobarbital at concentrations identical to those required to produce anesthesia.76 Thus, VDCCs could well mediate some actions of general anesthetics, but their general insensitivity makes them unlikely to be major targets.

FIGURE 3. Halothane inhibition of voltage-dependent Ca2+, Na + , and K+ currents. The Ca2+ channels are L-type channels from GH3 cells, and the Na+ and K+ channels are from the squid giant axon. The closed circles show the concentrations of halothane required to anesthetize humans. Note that the Ca2+ currents are inhibited about 20% by clinical concentrations of halothane whereas the Na+ and K+ currents are not inhibited at all. (Reproduced by permission from Franks NP, Lieb WR. Molecular and cellular mechanisms of anesthesia. Nature. 1994;367:607, Macmillan Magazines Ltd.)

Potassium channels are the most diverse of the ion channel types and include voltage-gated, second messenger and ligand-activated, and so-called inward rectifying channels; some channels fall into more than one category. High concentrations of both volatile anesthetics and intravenous anesthetics are required to significantly affect the function of voltage-gated K+ channels.617778 Similarly, classic inward rectifying K+ channels are relatively insensitive to sevoflurane and barbiturates.79,80,81 However, some ligand-gated K+ channels are reasonably sensitive to volatile anesthetics as discussed below.


Existing evidence suggests that most VDCCs are modestly sensitive or insensitive to anesthetics, but some reports argue for significant heterogeneity in the anesthetic sensitivity of specific channel types and subtypes. In particular, some sodium channel subtypes are inhibited by volatile anesthetics and this effect may be responsible in part for a reduction in neurotransmitter release at some synapses. Additional experimental data will be required to establish whether anesthetic-sensitive VDCCs are localized to specific synapses at which anesthetics have been shown to inhibit neurotransmitter release.

Anesthetic Effects on Ligand-Gated Ion Channels

Fast excitatory and inhibitory neurotransmission is mediated by the actions of ligand-gated ion channels. Synaptically released glutamate or GABA diffuse across the synaptic cleft and bind to channel proteins that open as a consequence of neurotransmitter release. The channel proteins that bind GABA (GABAA receptors) are members of a superfamily of structurally related ligand-gated ion channel proteins that include nicotinic acetylcholine receptors, glycine receptors, and 5-HT3 receptors.82 Based on the structure of the nicotinic acetylcholine receptor, each ligand-gated channel is thought to be composed of five nonidentical subunits. The glutamate receptors also comprise a family, each receptor thought to be a tetrameric protein composed of structurally related subunits.83 The ligand-gated ion channels provide a logical target for anesthetic action because selective effects on these channels could inhibit fast excitatory synaptic transmission and/or facilitate fast inhibitory synaptic transmission. The effects of anesthetic agents on ligand-gated ion channels are thoroughly cataloged in a review by Krasowski and Harrison.58 The following section provides a brief summary of this large body of work.

Glutamate-Activated Ion Channels

Glutamate-activated ion channels have been classified, based on selective agonists, into three categories: AMPA receptors, kainate receptors, and NMDA receptors. Molecular biologic studies indicate that a large number of structurally distinct glutamate receptor subunits can be used to form each of the three categories of glutamate receptors.84 This structural heterogeneity is reflected in functional heterogeneity within each category of glutamate receptor. AMPA and kainate receptors are relatively nonselective monovalent cation channels involved in fast excitatory synaptic transmission, whereas NMDA channels conduct not only Na+ and K+ but also Ca++ and are involved in long-term modulation of synaptic responses (long-term potentiation). Studies from the early 1980s in mouse and rat brain preparations showed that AMPA- and kainate-activated currents are insensitive to clinical concentrations of halothane,85 enflurane,86 and the neurosteroid allopregnanolone.87 In contrast, kainate- and AMPA-activated currents were shown to be sensitive to barbiturates; in rat hippocampal neurons, 50 µM pentobarbital (pentobarbital produces anesthesia at approximately 50 µM) inhibited kainate and AMPA responses by 50%.87 More recent studies using cloned and expressed glutamate receptor subunits show that submaximal agonist responses of GluR3 (AMPA-type) receptors are inhibited by fluorinated volatile anesthetics whereas agonist responses of GluR6 (kainate-type) receptors are enhanced.88 In contrast both GluR3 and GluR6 receptors are inhibited by pentobarbital. The directionally opposite effects of the volatile anesthetics on different glutamate receptor subtypes may explain the earlier inconclusive effects observed in tissue, where multiple subunit types are expressed. These opposite effects have also been used as a strategy to identify critical sites on the molecules involved in anesthetic effect. By producing GluR3/GluR6 receptor chimeras (receptors made up of various combinations of sections of the GluR3 and GluR6 receptors) and screening for volatile anesthetic effect, specific areas of the protein required for volatile anesthetic potentiation of GluR6 have been identified. Subsequent site-directed mutagenesis studies have identified a specific glycine residue (Gly-819) as critical for volatile anesthetic action on GluR6-containing receptors.89

NMDA-activated currents also appear to be sensitive to a subset of anesthetics. Electrophysiological studies show virtually no effects of clinical concentrations of volatile anesthetics,85,86 neurosteroids, or barbiturates87 on NMDA-activated currents. It should be noted that there is some evidence from flux studies that volatile anesthetics may inhibit NMDA-activated channels. A study in rat brain microvesicles showed that anesthetic concentrations (0.2-0.3 mM) of halothane and enflurane inhibited NMDA-activated calcium flux by 50%.90 In contrast, ketamine is a potent and selective inhibitor of NMDA-activated currents. Ketamine stereoselectively inhibits NMDA currents by binding to the phencyclidine site on the NMDA receptor protein.91,92,93 The anesthetic effects of ketamine in intact animals show the same stereoselectivity as that is observed in vitro,94 suggesting that the NMDA receptor may be the principal molecular target for the anesthetic actions of ketamine. Two other recent findings suggest that NMDA receptors may be an important target for nitrous oxide and xenon. These studies show that N2O95,96 and xenon97 are potent and selective inhibitors of NMDA-activated currents. This is illustrated in Fig. 4, showing that N2O inhibits NMDA-elicited, but not GABA-elicited, currents in hippocampal neurons.

FIGURE 4. Nitrous oxide inhibits NMDA-elicited, but not GABA-elicited, currents in rat hippocampal neurons. (Panel A) 80% N2O has no effect on holding current (upper trace), but inhibits the current elicited by NMDA. (Panel B) N2O causes a rightward and downward shift of the NMDA concentration-response curve, indicating a mixed competitive/noncompetitive antagonism. (Panel C) 80% N2O has little effect on GABA-elicited currents. In contrast, an equipotent anesthetic concentration of pentobarbital markedly enhances the GABA-elicited current. (Reproduced with permission from Jevtovic-Todorovic V, Todorovic SM, Mennerick S, et al. Nitrous oxide (laughing gas) is an NMDA antagonist, neuroprotectant, and neurotoxin. Nature Medicine. 1998;4:460)

GABA-Activated Ion Channels

GABA is the most important inhibitory neurotransmitter in the mammalian central nervous system. GABA-activated ion channels (GABAA receptors) mediate the postsynaptic response to synaptically released GABA by selectively allowing chloride ions to enter and thereby hyperpolarizing neurons. GABAA receptors are multisubunit proteins consisting of various combinations of α, β, γ, δ, and ε subunits, and there are many subtypes of each of these subunits. The function of GABAA receptors is modulated by a wide variety of pharmacological agents including convulsants, anticonvulsants, sedatives, anxiolytics, and anesthetics.98 The effects of these various drugs on GABAA receptor function varies across brain regions and cell types. The following section briefly reviews the effects of anesthetics on GABAA receptor function.

Barbiturates, anesthetic steroids, benzodiazepines, propofol, etomidate, and the volatile anesthetics all modulate GABAA receptor function.59, 98,99,100,101 These drugs produce three kinds of effects on the electrophysiological behavior of the GABAA receptor channels: potentiation, direct gating, and inhibition. Potentiation refers to the ability of anesthetics to increase markedly the current elicited by low concentrations of GABA, but to produce no increase in the current elicited by a maximally effective concentration of GABA.85,102 Potentiation is illustrated in Fig. 5, showing the effects of halothane on currents elicited by a range of GABA concentrations in dissociated cortical neurons. Anesthetic potentiation of GABAA currents generally occurs at concentrations of anesthetics within the clinical range. Direct gating refers to the ability of anesthetics to activate GABAA channels in the absence of GABA. Generally, direct gating of GABAA currents occurs at anesthetic concentrations higher than those used clinically, but the concentration-response curves for potentiation and for direct gating can overlap. It is not known whether direct gating of GABAA channels is either required for or contributes to the effects of anesthetics on GABA-mediated inhibitory synaptic transmission in vivo. In the case of anesthetic steroids, strong evidence indicates that potentiation, rather than direct gating of GABAA currents, is required for producing anesthesia.103 Anesthetics can also inhibit GABA-activated currents. Inhibition refers to the ability of anesthetics to prevent GABA from initiating current flow through GABAA channels and has generally been observed at high concentrations of both GABA and anesthetic.104,105 Inhibition of GABAA channels may help to explain why volatile anesthetics have, in some cases, been observed to inhibit rather than facilitate inhibitory synaptic transmission.32

FIGURE 5. The effects of halothane (Hal), enflurane (Enf), and fluorothyl (HFE) on GABA-activated chloride currents in dissociated rat CNS neurons. (Panel A) Clinical concentrations of halothane and enflurane potentiate the ability of GABA to elicit a chloride current. The convulsant fluorothyl antagonizes the effects of GABA. (Panel B) GABA causes a concentration-dependent activation of a chloride current. Halothane shifts the GABA concentration-response curve to the left (increases the apparent affinity of the channel for GABA) whereas fluorothyl shifts the curve to the right (decreases the apparent affinity of the channel for GABA). (Reproduced with permission from Wakamori M, Ikemoto Y, Akaike N. Effects of two volatile anesthetics and a volatile convulsant on the excitatory and inhibitory amino acid responses in dissociated CNS neurons of the rat. J Neurophysiol. 1991;66:2014.)

Effects of anesthetics have also been observed on the function of single GABAA channels. These studies show that barbiturates,99 propofol,101 and volatile anesthetics106 do not alter the conductance (rate at which ions traverse the open channel) of the channel, but that they increase the frequency with which the channel opens and/or the average length of time that the channel remains open. Collectively, the whole cell and single channel data are most consistent with the idea that clinical concentrations of anesthetics produce a change in the conformation of GABAA receptors that increases the affinity of the receptor for GABA. This is consistent with the ability of anesthetics to increase the duration of inhibitory postsynaptic potentials (IPSPs), because higher affinity binding of GABA would slow the dissociation of GABA from postsynaptic GABAA channels. It would not be expected that anesthetics would increase the peak amplitude of a GABAergic IPSP because synaptically released GABA probably reaches very high concentrations in the synapse. Higher concentrations of anesthetics can produce additional effects, either directly activating or inhibiting GABAA channels. Consistent with these ideas, a study by Banks and Pearce showed that isoflurane and enflurane simultaneously increased the duration and decreased the amplitude of GABAergic inhibitory postsynaptic currents in hippocampal slices.107

Despite the similar effects of many anesthetics on GABAA receptor function, there is significant evidence that the various anesthetics do not act by binding to a single common binding site on the channel protein. First, even anesthetics that directly activate the channel probably do not bind to the GABA binding site. This is most clearly demonstrated by molecular biologic studies in which the GABA binding site is eliminated from the channel protein but pentobarbital can still activate the channel.108 Direct radioligand binding studies have demonstrated that benzodiazepines bind to the GABAA receptor at nanomolar concentrations and that other anesthetics can modulate binding but do not bind directly to the benzodiazepine site.98,109 A series of more complex studies examining the interactions between barbiturates, anesthetic steroids, and benzodiazepines indicates that these three classes of drugs cannot be acting at the same sites.98 The actions of anesthetics on GABAA receptors are further complicated by the observation that steroid anesthetics can produce different effects on GABAA receptors in different brain regions.110 This suggests the possibility that the specific subunit composition of a GABAA receptor may encode pharmacological selectivity. This is well illustrated by benzodiazepine sensitivity, which requires the presence of the γ2 subunit subtype.111 Similarly, sensitivity to etomidate has been shown to require the presence of a β2 or β3 subunit.112 More recently, it has been shown that the presence of a δ or ε subunit in a GABAA receptor confers insensitivity to the potentiating effects of some anesthetics.113,114

Interestingly, GABAA receptors composed of ρ-type subunits (referred to as GABAC receptors) have been shown to be inhibited rather than potentiated by volatile anesthetics.115 This property has been exploited, using molecular biologic techniques, by constructing chimeric receptors composed of part of the ρ receptor coupled to part of an α, β, or glycine receptor subunit. By screening these chimeras for anesthetic sensitivity, regions of the α, β, and glycine subunits responsible for anesthetic sensitivity have been identified. Based on the results of these chimeric studies, site-directed mutagenesis studies were performed to identify the specific amino acids responsible for conferring anesthetic sensitivity. These studies revealed two critical amino acids, near the extracellular regions of transmembrane domains 2 and 3 (TM2, TM3) of the glycine and GABAA receptors that are required for volatile anesthetic potentiation of agonist effect.116 It is not yet clear if these amino acids represent a volatile anesthetic binding site, or whether they are sites critical to transducing anesthetic-induced conformational changes in the receptor molecule. Interestingly, one of the amino acids shown to be critical to volatile anesthetic effect (TM3 site) has also been shown to be required (in the β2/β3 subunit) for the potentiating effects of etomidate.117 In contrast, the TM2 and TM3 sites do not appear to be required for the actions of propofol, barbiturates, or neurosteroids.118 Interestingly, a distinct amino acid in the TM3 region of the β1 subunit of the GABAA receptor has been shown to selectively modulate the ability of propofol to potentiate GABA agonist effects.118 Collectively, these molecular biologic data provide strong evidence that there are multiple unique binding sites for anesthetics on the GABAA receptor protein.

Other Ligand-Activated Ion Channels

Other members of the ligand-gated receptor superfamily include the nicotinic acetylcholine receptors (muscle and neuronal types), glycine receptors, and 5-HT3 receptors. A large body of work has gone into examining the effects of anesthetics on nicotinic acetylcholine receptors. The muscle type of nicotinic receptor has been shown to be inhibited by anesthetic concentrations in the clinical range119 and to be desensitized by higher concentrations of anesthetics.120 The muscle nicotinic receptor is an informative model to study because of its abundance and the wealth of knowledge about its structure. It is, however, not expressed in the central nervous system and hence not involved in the mechanism of anesthesia. However, a neuronal type of nicotinic receptor, which is widely expressed in the nervous system, might plausibly be involved in anesthetic mechanisms. Older studies looking at neuronal nicotinic receptors in molluscan neurons121 and in bovine chromaffin cells122 indicate that these channels are inhibited by clinical concentrations of volatile anesthetics. More recent studies using cloned and expressed neuronal nicotinic receptor subunits have shown a high degree of subunit and anesthetic selectivity. Acetylcholine-elicited currents are inhibited, in receptors composed of various combinations of α2, α4, β2, and β4 subunits, by subanesthetic concentrations of halothane123 or isoflurane.124 In contrast, these receptors are relatively insensitive to propofol. Most interestingly, receptors composed of α7 subunits are completely insensitive to both isoflurane and propofol.124,125

Subsequent pharmacological experiments using selective inhibitors of neuronal nicotinic receptors led to the conclusion that these receptors are unlikely to have a major role in immobilization by volatile anesthetics.126,127 However, they might play a role in the amnestic or hypnotic effects of volatile anesthetics.128

Glycine is an important inhibitory neurotransmitter, particularly in the spinal cord and brainstem. The glycine receptor is a member of the ligand-activated channel superfamily that, like the GABAA receptor, is a chloride-selective ion channel. A large number of studies have shown that clinical concentrations of volatile anesthetics potentiate glycine-activated currents in intact neurons85 and in cloned glycine receptors expressed in oocytes.129,130 The volatile anesthetics appear to produce their potentiating effect by increasing the affinity of the receptor for glycine.130 Propofol,101 alphaxalone, and pentobarbital also potentiate glycine-activated currents, whereas etomidate and ketamine do not.129 Potentiation of glycine receptor function may contribute to the anesthetic action of volatile anesthetics and some parenteral anesthetics. 5-HT3 receptors are also members of the genetically related superfamily of ligand-gated receptor channels. Clinical concentrations of volatile anesthetics potentiate currents activated by 5-hydroxytryptamine in intact cells131 and in cloned receptors expressed in oocytes.132 In contrast, thiopental inhibits 5-HT3 receptor currents131 and propofol is without effect on these receptor channels.132 5-HT3 receptors may play some role in the anesthetic state produced by volatile anesthetics and may also contribute to some unpleasant anesthetic side effects such as nausea and vomiting.


Several ligand-gated ion channels are modulated by clinical concentrations of anesthetics. Ketamine, N2O, and xenon inhibit NMDA-type glutamate receptors, and this effect may play a major role in their mechanism of action. A large body of evidence shows that clinical concentrations of many anesthetics potentiate GABA-activated currents in the central nervous system. This suggests that GABAA receptors are a probable molecular target of anesthetics. Other members of the ligand-activated ion channel family, including glycine receptors, neuronal nicotinic receptors, and 5-HT3 receptors, are also affected by clinical concentrations of anesthetics and remain plausible anesthetic targets.

(to be continued in the next Part )