JOURNAL HOME CME HOME THIS MONTH PAST ISSUES ETOC COLLECTIONS AUTHORS REVIEWERS EDITORIAL BOARD FEEDBACK RSS HELP
		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) An erratum has been published Alert me when this article is cited Alert me if a correction is posted Services Email this article to a colleague Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (83) Citing Articles via Google Scholar Google Scholar Articles by Sonner, J. M. Articles by Eger, E. I Search for Related Content PubMed PubMed Citation Articles by Sonner, J. M. Articles by Eger, E. I, II Related Collections Mechanisms Pharmacology

---

ANESTHETIC PHARMACOLOGY

Inhaled Anesthetics and Immobility: Mechanisms, Mysteries, and Minimum Alveolar Anesthetic Concentration

James M. Sonner, MD^*, Joseph F. Antognini, MD, Robert C. Dutton, MD^*, Pamela Flood, MD, Andrew T. Gray, MD PhD^*, R. Adron Harris, PhD, Gregg E. Homanics, PhD^||, Joan Kendig, PhD^¶, Beverley Orser, MD^#, Douglas E. Raines, MD^**, James Trudell, PhD^¶, Bryce Vissel, PhD, and Edmond I Eger, II, MD^* Section Editor

*Department of Anesthesia and Perioperative Care, University of California, San Francisco, California; Department of Anesthesiology, University of California, Davis, California; Columbia University, New York, New York; University of Texas, Austin, Texas; ||University of Pittsburgh, Pittsburgh, Pennsylvania; ¶Stanford University, Palo Alto, California; #University of Toronto, Toronto, Canada; **Department of Anaesthesia, Harvard Medical School, Cambridge, Massachusetts; and Garvan Institute of Medical Research, Darlinghurst, Australia

Address correspondence and reprint requests to James M. Sonner, MD, Department of Anesthesia, S-455, University of California, San Francisco, CA 94143-0464. Address e-mail to sonnerj{at}anesthesia ucsf.edu.

	Abstract

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
Studies using molecular modeling, genetic engineering, neurophysiology/pharmacology, and whole animals have advanced our understanding of where and how inhaled anesthetics act to produce immobility (minimum alveolar anesthetic concentration; MAC) by actions on the spinal cord. Numerous ligand- and voltage-gated channels might plausibly mediate MAC, and specific animo acid sites in certain receptors present likely candidates for mediation. However, in vivo studies to date suggest that several channels or receptors may not be mediators (e.g., -aminobutyric acid A, acetylcholine, potassium, 5-hydroxytryptamine-3, opioids, and ₂-adrenergic), whereas other receptors/channels (e.g., glycine, N-methyl-D-aspartate, and sodium) remain credible candidates.

	Introduction

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
After decades of ignorance, we now know how some anesthetics (e.g., etomidate) act (1). This knowledge resulted from information revealed by multiple approaches to studies of anesthetic mechanisms, including studies in whole animals using old-fashioned pharmacologic techniques, in vivo and in vitro neurophysiological studies, and the application of genetic engineering. Genetic engineering allows the manipulation of receptors and their study in vitro. It allows the creation of animals (usually mice) with mutated receptors that do not respond to anesthetics and that, thereby, serve as tools to test the importance of a given receptor to a particular anesthetic effect. This review discusses how these approaches have increased understanding of the mechanisms by which a particular class of anesthetics, the inhaled anesthetics, acts. The review focuses on the mechanistic bases for one characteristic of inhaled anesthetics: their capacity to cause immobility. The standard measure of anesthetic potency, MAC (the minimum alveolar concentration of an inhaled anesthetic required to suppress movement in response to noxious stimulation in 50% of subjects), reflects this capacity (2). MAC is an anesthetic 50% effective dose (ED₅₀).

At the turn of the last century, Overton (3) and Meyer (4) reported that anesthetic potency correlated with lipophilicity, suggesting the importance of a lipid-like phase to anesthetic action. For conventional and many experimental inhaled anesthetics (5–9), the correlation is remarkable (Fig. 1). Over the ensuing 80 years, a few theories of narcosis evolved from the focus on lipids, particularly the membrane bilayer, but none of these theories withstood experimental scrutiny. For example, inhaled anesthetics increase membrane disorder (increase fluidity) (10–14), and this was proposed as a basis for anesthesia. However, increases in body temperature also increase this disorder, but increases in body temperature increase rather than decrease MAC (15). The discovery of inhaled compounds with potencies that do not correlate with lipophilicity called into question the Meyer-Overton hypothesis: alcohols are more potent than predicted from their lipophilicity, whereas some compounds are less potent (transitional compounds) (5). For example, ethanol has an oil/gas partition coefficient of 108 (16), similar to the value of 98 for isoflurane (17). However, the MAC of isoflurane in rats is 1.5% (18,19), more than 10 times larger than the MAC of 0.1% for ethanol (16). At an extreme are nonimmobilizer compounds, compounds that have no anesthetic effect, alone or in combination with known anesthetics, despite a lipophilicity that predicts an anesthetizing capacity (5). For example, 1,2-dichlorohexafluorobutane (also called F6 or 2N) has an oil/gas partition coefficient of 44 (5), similar to the value of 47 found for sevoflurane (20), which has a MAC of 2.4% (21,22). However, F6 is not anesthetic at any concentration, including concentrations exceeding 4% (5).

View larger version (18K): [in this window] [in a new window]   Figure 1. The potencies (minimum alveolar anesthetic concentration; MAC) of conventional [nitrous oxide (7), halothane (6), isoflurane (6), sevoflurane (6), desflurane (6), and cyclopropane (6)] and several unconventional [xenon (9), 1A (1-chloro-1,2,2-trifluorocyclobutane) (5), benzene (8), and hexafluorobenzene (8)] inhaled anesthetics in rats correlate with their affinity to a lipid phase, as defined by the oil/gas partition coefficient. A caveat: although this relationship applies to all conventional inhaled anesthetics, several compounds (e.g., nonimmobilizers) do not conform to the relationship.

Antognini and Schwartz (28), Antognini et al. (29), Borges and Antognini (30), Rampil et al. (31), and Rampil (32) provided another key finding: that inhaled anesthetic-induced immobility largely results from an action on the spinal cord rather than an action on higher centers. Their results informed many subsequent studies, several of which are described in this review.

Finally, the increasing capacity to study and manipulate receptors through genetic engineering provides a powerful tool for studies of anesthetic mechanisms. Results of such work complement work from traditional pharmacology. For example, as will be discussed, the nonimmobilizer F6 blocks acetylcholine receptors in vitro, and thus the acetylcholine receptor is an unlikely target for the immobilizing effect of inhaled anesthetics. Consistent with this finding, in vivo blockade of acetylcholine receptors does not affect MAC.

	The Spinal Cord as the Primary Site Mediating Immobilization

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
Several studies demonstrate that the spinal cord mediates most of the ability of inhaled anesthetics to produce immobility. The observation a decade ago that immobility during noxious stimulation does not correlate with electroencephalographic (EEG) activity prompted the hypothesis that cortical electrical activity does not control motor responses to noxious stimulation (33). In rats, precollicular decerebration (31) (Fig. 2) or complete section of the upper thoracic spinal cord (32) minimally affects isoflurane’s capacity to suppress movement. In vivo electrophysiologic examination of the spinal cord suggests that anesthetics can suppress both sensory (34) and motor (35) neuron activity. In rats (36) and humans (37), suppression of movement in response to noxious stimulation correlates with suppression of motor neuron excitability (as determined by recurrent impulses known as F waves). In goats, MAC for isoflurane delivered to the whole body is 1.2%, but delivery to only the brain increases MAC to nearly 3% (28). The separation with halothane is still greater (29). Control halothane MAC is 0.9%, but delivery confined to the head requires 3.4% to abolish movement. Some goats move during EEG silence and move at the largest cerebral halothane concentrations tested. These results indicate that inhaled anesthetics act primarily on the spinal cord to produce immobility and that only a minor component of this immobility results from cerebral effects. The common results from different species and experimental methods increase the compelling nature of these conclusions.

View larger version (15K): [in this window] [in a new window]   Figure 2. Decerebration does not change the minimum alveolar anesthetic concentration (MAC) of isoflurane in rats (31). These data and similar data gathered in Rampil’s laboratory (32) and Antognini’s laboratory (28,30) demonstrate that the spinal cord mediates much or most of the capacity of inhaled anesthetics to produce immobility. That is, effects on the spinal cord determine MAC.

	What Has Been Learned from In Vitro Spinal Cord Studies?

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
Because motor neurons integrate all input and signal muscles to move, anesthetic effects on motor neurons may be important to the production of immobility. Other spinal cord components (primary afferent terminals and interneurons) likely also play a role. The effect of anesthetics on motor neurons has been examined in the intact isolated neonatal rat spinal cord. Volatile anesthetics (38,39) and ethanol (40) depress motor neuron output evoked by dorsal root stimulation.

Volatile anesthetics enhance the currents through gamma-aminobutyric acid (GABA)_A and glycine inhibitory chloride channels that are produced by the application of GABA or glycine. This action reduces the excitability of neurons and might contribute to anesthetic depressant actions. If this is the case, then blockade of these channels should oppose the depressant effects of anesthetics. In intact spinal cord, blockade of GABA_A and glycine inhibitory chloride channels does, in fact, decrease the depressant effects of anesthetics (41). However, whether blockade is only partial reduction or is complete cannot be determined experimentally in this preparation because of uncoordinated fluctuations in the baseline brought about by the convulsant effects of these blocking drugs.

Acetylcholine receptors are also affected by volatile anesthetics and alcohols, but in opposite directions. Although acetylcholine receptors can profoundly affect spinal neuronal transmission, block of nicotinic or muscarinic receptors does not modify anesthetic potency, either in vivo (42,43) or in vitro (44). Thus, inhibitory chloride channels, but not acetylcholine receptors, may be implicated in anesthetic-induced immobility.

Spinal cord slice preparations allow isolation of postsynaptic actions on motor neurons from actions on primary afferents and interneurons. In slices, ethanol (44) and volatile anesthetics (45) depress currents evoked by glutamate application, proving that these agents can directly depress motor neuron excitability. Furthermore, ethanol (44) and volatile anesthetics (45) depress both -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) and N-methyl-D-aspartic acid (NMDA) recep-tor-mediated currents by actions independent of GABA_A and glycine receptors, suggesting that anesthetic reduction of motor output can result from depression of excitation as well as enhancement of inhibition.

Anesthetic effects on the cord are diverse and complex, and the ultimate effect on MAC is incompletely understood. Spontaneous miniature current frequency (presynaptically determined), amplitude, and kinetics (postsynaptically determined) reveal pre- and post-synaptic anesthetic actions on motor neurons. Halothane, isoflurane, and enflurane prolong glycinergic current duration but do not increase amplitude (46), in contrast to nonneuronal expression systems, in which both amplitude and duration are increased. The three anesthetics also increase the frequency of spontaneous transmitter release, but only when sodium channels are blocked (46). With intact sodium channels, halothane and isoflurane do not affect frequency, and enflurane decreases frequency. Changes in charge transfer parallel these effects: increases in presynaptically mediated frequency plus postsynaptically mediated prolongation increase the total charge transfer, but only during sodium channel blockade. That is, in normal intact preparations, anesthetics do not produce an absolute increase in inhibition. Thus, for glycinergic transmission, anesthetics affect the timing rather than the magnitude of inhibition, increasing the period during which inhibition is effective.

When glutamate AMPA currents are compared with glycinergic currents, a contrasting picture emerges. Enflurane decreases miniature current frequency regardless of sodium channel blockade (47) and depresses current amplitude without changing kinetics. Thus, volatile anesthetics tend to increase glycinergic inhibition by two effects: a presynaptic enhancement of transmitter release and a postsynaptic prolongation of current duration. A depressant action on release, probably via sodium channels, counterbalances these inhibitory actions. Enflurane’s actions on glutamate transmission are purely depressant, both pre- and postsynaptically. Immobility may result from a shift toward inhibition in the balance between excitation and inhibition, rather than from an absolute increase in inhibition.

In summary, results from studies in the isolated spinal cord suggest that actions involving both excitatory (AMPA and NMDA) and inhibitory (glycine and GABA_A) transmission might produce immobility. Actions on sodium channels and on other channels, including those that determine resting membrane potential, also may be important. Cholinergic receptors play little or no role.

	What Do In Vitro Studies of Isolated Inhibitory and Excitatory Receptors Reveal?

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
Which multiple neuronal signaling systems govern crucial components of anesthesia, such as immobility? Drugs that produce anesthesia by acting on a single signaling system offer powerful clues to the site of anesthetic action. Propofol, alfaxalone, and etomidate primarily produce anesthesia by enhancing GABA_A receptor function (i.e., by increasing the effect of a given concentration of GABA) (48). Ketamine may produce anesthesia primarily by inhibiting NMDA receptor function, although it also acts on acetylcholine receptors (49). However, blockade of NMDA alone does not appear to cause immobility (50), so ketamine probably causes immobility by more than simply blocking NMDA receptors. Emerging, detailed molecular analyses of anesthetic actions on GABA_A and glycine receptors and the potential importance of NMDA receptors prompt the following discussion focused on these three ion channels (Table 1).

Ionotropic GABA receptors include GABA_A and GABA_C classes. (A metabotropic GABA_B receptor coupled through guanosine triphosphate-binding proteins is not a ligand-gated ion channel.) GABA_A receptors mediate much of the inhibitory neurotransmission in the central nervous system (CNS) (52). GABA_A and GABA_C receptors are formed from combinations of , ß, , and subunits, each of which may have multiple types (_1–6, ß_1–3, and _1–3). Although and subunits also exist, their function remains obscure. GABA_C receptors likely form from _1–3 subunits. Homomeric (all subunits the same) ₁ receptors are functional (53,54). Most CNS GABA_A receptors have , ß, and subunits. GABA binding to extracellular sites on the receptor opens a chloride channel through the receptor. A response to GABA requires both and ß subunits; benzodiazepine enhancement of GABA function requires , ß, and subunits (55).

Two types of subunits ( and ß) form glycine receptors. Only one ß subunit has been identified, and ₁ is the major subunit in adults. Homomeric ₁ or ₂ receptors are functional (56–58). As discussed below, homomeric (e.g., ₁ or glycine ₁) receptors expressed in Xenopus oocytes have advanced studies of molecular sites of anesthetic action.

Several observations suggest the potential importance of these receptors to the immobility produced by inhaled anesthetics. Neuronal and several nonneuronal expression systems reveal that relevant (1 MAC) concentrations of volatile anesthetics and n-alcohols potentiate GABA_A and glycine receptor function (52,59–63). Most IV anesthetics also potentiate GABA_A receptor function, and some potentiate glycine receptor function (61–63). Nonimmobilizing congeners of several fluorinated anesthetics do not affect these receptors (52,59–61,64).

Comparison of the effects of anesthetics versus nonimmobilizers can be used as one test of the relevance of a particular ion channel. Although the anesthetic 1-chloro-1,2,2-trifluorocyclobutane (called 1A or F3) and the structurally similar nonimmobilizer 2N, or F6, have lipid solubilities consistent with an anesthetic capacity, only F3 affects the function of GABA_A, glycine, AMPA, kainate, and 5-hydroxytryptamine (5-HT)₃ receptors (59), and thus these receptors pass the test. The nonimmobilizer 2,3-dichlorooctafluorobutane (F8) also does not affect the function of several ligand-gated ion channels. In contrast, both F3 and F6 inhibit neuronal nicotinic receptors and several metabotropic receptors (5-HT_2C, muscarinic-1, mGluR5) (65,66), and thus these receptors do not pass the test.

How do inhaled anesthetics change the function of inhibitory neurotransmitter receptors? The differential actions of anesthetics on two homologous ligand-gated ion channels allow an exploration of the molecular sites of anesthetic action. In vitro, homomeric glycine ₁ and GABA ₁ receptors respond oppositely to volatile anesthetics, which enhance function in the former (61) and inhibit function in the latter (67). A genetic approach (one that combines various parts of homomeric glycine ₁ and GABA ₁ receptors to produce new receptors) allows identification of amino acids in transmembrane segment 2 and in transmembrane segment 3 of glycine and GABA_A receptors that permit ethanol enhancement of receptor function (68). These amino acids lie near the extracellular surface of the membrane. Results from independent studies suggest the importance of a site near the extracellular surface (69–72).

The possibility that propanethiol, an anesthetic in animals (73), could irreversibly react with cysteines engineered at those critical positions provided more direct evidence for binding of anesthetics to these amino acids in glycine and GABA_A receptors (74). Propanethiol and the related sulfhydryl-specific reagent, propyl methanethiosulfonate (PMTS), covalently bind to the mutated cysteines and permanently enhance receptor function (i.e., they produce irreversible anesthetic-like effects.) Competition experiments performed before and after irreversible binding of PMTS to the cysteine mutations in the GABA_A receptor suggested competition for a single binding site among PMTS and the inhaled anesthetics or octanol, but not between PMTS and alfaxalone. This is consistent with the binding of inhaled anesthetics and n-alcohols to the same site but the binding of alfaxalone to a different site (75). Similar conclusions apply to the glycine receptor.

As noted above, the NMDA receptor also holds promise as a site of anesthetic action. NMDA receptors constitute one glutamate receptor subtype, and they require the coagonists glycine and glutamate for activation (76). The subunits NR1, NR2 (A-D), and NR3 (A-B) form NMDA receptors (77–79). The NR1 subunit confers essential function to the NMDA receptor and is expressed throughout the CNS. In adult mouse brain, NR1, NR2A, and NR2B subunits are widely expressed, whereas NR2C occurs predominantly in the cerebellum, and small levels of NR2D exist only in the thalamus, brainstem, and spinal cord (80). Most NR3B subunits are on motor neurons (79), whereas NR3A has wide distribution (81,82).

The Xenopus oocyte expression system allows study of receptors with pharmacologic properties that may mimic those of native neuronal receptors (59). NMDA receptors expressed from cloned receptors also provide evidence that inhaled anesthetics depress NMDA receptors. Clinically relevant concentrations of isoflurane, sevoflurane, and desflurane inhibit recombinant NR1/NR2A and NR1/NR2B NMDA receptors in a reversible, dose-dependent, and voltage-insensitive manner (83). Similarly, enflurane, urethane, nitrous oxide, xenon, cyclopropane, and butane inhibit NMDA-stimulated currents in oocytes expressing NMDA receptors (84–87). Thus, considerable evidence shows that a wide range of volatile anesthetics inhibit NMDA receptor function.

These results indicate that inhaled anesthetic actions on ligand-gated ion channels may explain the capacity of inhaled anesthetics to produce immobility; they point to candidate proteins that might mediate immobility. Whether the associated receptors contribute to anesthetic-induced immobility can be determined only from studies in intact organisms. A definitive test of the importance of anesthetic influence on these channels could, in principle, be achieved by designing mutant receptors that provide normal synaptic transmission but resist any anesthetic effect and expressing these receptors in animals. The molecular understanding needed to produce these mutants is emerging for the GABA_A and glycine receptors but is lacking for most other ligand-gated ion channels. As discussed elsewhere in this review, the introduction of drug-resistant receptors into animals (knockin mice) has provided remarkable insights regarding the actions of benzodiazepines (88,89) and etomidate and propofol (90) and will likely prove important for understanding inhaled anesthetic action.

	Genetic Engineering: Definitions and Implications for Studies of Anesthetic Mechanisms

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
Present technologies allow the creation of animals with genetic modifications that allow tests of specific receptor targets of anesthetics: transgenic, knockout (global or conditional), and knockin animals (Table 2). Transgenic animals overexpress or misexpress an added gene (referred to as a transgene). They are created by microinjection of DNA into the pronucleus of a recently fertilized embryo. Several disadvantages limit the usefulness of this older, fast, widely available technique. Because the added gene inserts randomly into the genome, expression can be variable, and, more importantly, the transgene must be dominant because the animals also have the endogenous gene. Mating transgenic animals to animals lacking the gene of interest (not always available) can circumvent this problem, but this is time consuming.

Conditional knockouts in which specific endogenous genes are inactivated in specific cells or tissues and/or at specific animal ages circumvent some of these limitations. This complex approach requires crossing two different lines of genetically engineered mice. Making mutations in specific cells or neuronal pathways after key developmental events have occurred minimizes compensation and simplifies the assignment of phenotype to neuroanatomic substrates. Although this approach has successfully dissected aspects of learning and memory (92), it has not been applied to studies of anesthetic mechanisms.

The gene knockin animal differs from a knockout animal (in which a gene of interest is inactivated) in that a knockin animal expresses a mutation in the gene of interest. For example, the gene might now produce a protein changed by a single amino acid. Because the normal endogenous promoter controls expression of the mutant gene, a knockin animal expresses the mutant gene in the same amount, at the same time, and in the same tissues as expressed by the normal gene. This approach allowed dissection of the contribution of individual GABA_A-receptor subunits to behavioral responses induced by benzodiazepines (93). Most importantly, it allowed the construction of mice resistant to anesthesia from propofol and etomidate (1). It also may allow the dissection of inhaled anesthetic mechanisms. Single amino acid substitutions in individual GABA_A-receptor subunit genes can change the response of the receptors in expression systems to alcohol and volatile anesthetics without changing the response to GABA (68), and mice harboring these mutations may allow for unambiguous testing of the contribution of specific receptor subunits to MAC. Such genetically engineered mice are currently being tested (94). However, the inhaled anesthetics may present a challenge not offered by anesthetics such as propofol or etomidate. Probably only one receptor (GABA_A) mediates the effects of propofol and etomidate, but multiple receptors may mediate the effects of inhaled anesthetics. Thus, results from a knockin involving a single receptor may elucidate only one aspect of inhaled anesthetic actions.

	What Do In Vivo Studies Reveal Concerning the Ion Channels Mediating MAC?

Top Abstract Introduction The Spinal Cord as... What Has Been Learned... What Do In Vitro... Genetic Engineering: Definitions... What Do In Vivo... The Place of Molecular... The 5-Angstrom Hypothesis Nonspecific Mechanisms Conclusions References

 
Pharmacologic approaches can complement the previously described genetic approaches to reveal the relevance of candidate receptors to MAC. Both assume that altering the function of a relevant receptor will change MAC. The genetic approach determines the effect of receptor mutations (e.g., elimination or modification of a specific receptor) on MAC. The pharmacologic approach determines the effect of agonists or antagonists of specific ion channels on MAC.

Genetic manipulations and pharmacologic treatments can change MAC directly or indirectly. For example, consider the action of two drugs on heart rate. Epinephrine acts directly on the sympathetic nerve terminals to increase heart rate. Atropine also increases heart rate but does so indirectly by blocking the action of acetylcholine on the parasympathetic nervous system. Similar considerations apply to MAC. Although opioids decrease MAC, inhaled anesthetics do not act via opioid receptors (see below). Opioid receptors can play an indirect (modulating) role but not a direct (mediating) role. This critical distinction will become apparent in the discussion of the role of the GABA_A receptor in MAC. A direct effect reflects a mediating role for an ion channel (the terms "direct effect" and "mediating role" are interchangeable; see Fig. 3).

View larger version (23K): [in this window] [in a new window]   Figure 3. This drawing illustrates the difference between the site (or sites) that directly mediates the actions of inhaled anesthetics versus sites that may influence inhaled anesthetic potency indirectly by modulating the function of the true (directly affected) site (or sites) of action. Modulation may occur through an action on the mediating site or by affecting the input to or output from that site. For example, a tonic release of -aminobutyric acid (GABA) from an indirect site might affect the measured minimum alveolar anesthetic concentration (MAC) of all inhaled anesthetics by inhibiting activity in the anesthetic site of action. Blocking this input would increase MAC, but this increase would not mean that GABAA receptors mediate inhaled anesthetic actions.

Pharmacologic approaches can also be used to distinguish between direct and indirect effects of receptor systems on MAC. If the administration of a receptor antagonist produces no measurable effect on MAC, then that receptor neither directly nor indirectly affects MAC and is not relevant. But suppose an antagonist has an effect? Because the spinal cord mediates MAC, an intrathecal (spinal) dose of antagonist should change MAC more than the same or larger dose given IV. If it does not, then the antagonist effect is probably supraspinal and not directly relevant to MAC.

However, perhaps an antagonist affects MAC, and the systemic dose required for a given change exceeds the intrathecal dose. To be relevant, the receptor must pass one more test. To be a relevant receptor (one directly acted on to produce the anesthetic effect), the injection of an antagonist should change MAC in proportion to the in vitro capacity of each anesthetic to affect the receptor. The power of this approach is suggested in the "proof-of-principle" experiment shown in Figure 4. In rats, the noncompetitive GABA_A antagonist picrotoxin increases the immobilizing ED₅₀ of the non-GABAergic anesthetic ketamine to a ceiling of approximately 60% (reflecting an indirect antagonism of the effect of naturally occurring tonic GABA release) but increases the ED₅₀ of the GABAergic anesthetic propofol by approximately 400% and shows no ceiling effect (primarily reflecting a direct antagonism) (49,95). In this study, isoflurane acts like ketamine, the non-GABAergic anesthetic. Gabazine (a competitive antagonist at GABA_A receptors) has the same effect as picrotoxin.

View larger version (17K): [in this window] [in a new window]   Figure 4. In rats, IV infusion of the noncompetitive -aminobutyric acid (GABA)A antagonist picrotoxin increases the immobilizing 50% effective dose (ED50) of the non-GABAergic anesthetic ketamine to a ceiling of approximately 60%. Because ketamine is not known to materially influence GABAA receptors, this effect of picrotoxin probably reflects an indirect antagonism of the effect of naturally occurring tonic GABA release. Picrotoxin similarly increases the immobilizing ED50 (minimum alveolar anesthetic concentration; MAC) of isoflurane. In contrast, picrotoxin increases the ED50 of the GABAergic anesthetic propofol by approximately 400% and shows no ceiling effect (primarily reflecting a direct antagonism) (49,95). The reader will note that 0.5 MAC isoflurane was added in both the ketamine and the propofol studies. This was done because the administration of picrotoxin in the absence of isoflurane was lethal in animals anesthetized with ketamine. The administration of isoflurane prevented death. Similar results were obtained when the isoflurane concentration was 0.25 MAC.

Inhibitory Ionotropic Ligand-Gated and Voltage-Gated Channels
Glycine.
Glycine receptors are major mediators of inhibitory neurotransmission in the spinal cord. Their spinal localization and their potentiation by clinically important volatile anesthetics make them prime candidates as mediators of MAC. Evidence in animals supports this conjecture. IV and intrathecal administration of strychnine, a glycine receptor antagonist, increases MAC (96,97). The MAC increase for cyclopropane, isoflurane, and halothane produced by intrathecal strychnine administration correlates with the in vitro enhancing effect of these anesthetics on the glycine receptors (97) (Fig. 5) (98), a result expected if the enhancing action on the glycine receptor mediates part of the immobilizing effect of these anesthetics.

View larger version (28K): [in this window] [in a new window]   Figure 5. Administration of strychnine, a glycine receptor antagonist, increases minimum alveolar anesthetic concentration (MAC) for rats (96). The MAC increase for cyclopropane, isoflurane, and halothane produced by intrathecal strychnine administration correlates with the in vitro capacity of these anesthetics (at a concentration equal to MAC) to enhance (increase) the response of glycine receptors to glycine (97). Each point is the measured increase in MAC of a rat at the corresponding enhanced receptor activity calculated from pooled in vitro results (85,97). In contrast, although administration of picrotoxin, a -aminobutyric acid (GABA)A antagonist, also increases MAC for cyclopropane, xenon, and isoflurane, the increase in MAC does not correlate with the in vitro enhancing effect of these anesthetics (98) (data for xenon and halothane were not available, respectively, for glycine and GABAA receptors).

GABA_A.
Although they mediate the immobility produced by injectable anesthetics (propofol and etomidate), GABA_A receptors may not mediate the immobility produced by inhaled anesthetics. The proof-of-principle experiment illustrated in Figure 4 indicates that isoflurane resembles the non-GABAergic anesthetic ketamine, but not the GABAergic anesthetic propofol. This suggests that isoflurane’s immobilizing effect is not directly mediated by GABA_A receptors.

Similar conclusions follow from results of studies applying intrathecal picrotoxin to rats inhaling isoflurane versus xenon or cyclopropane, anesthetics with minimal effects on GABA_A receptors (Zhang Y, unpublished data) (Fig. 5). If inhaled anesthetic enhancement of the response of GABA_A receptors directly mediates the capacity of inhaled anesthetics to suppress movement in response to noxious stimuli, then blockade of GABA_A receptors should increase the MAC of isoflurane more than the MAC of xenon or cyclopropane. However, MAC increases equally for all three anesthetics, indicating that GABA_A receptors are no more important to the immobilizing action of isoflurane than for anesthetics with minimal effects on GABA_A receptors. It may be argued that the in vitro data for the effective concentration of anesthetics may be influenced by many variables (e.g., agonist concentration, use of peak current, or current decay) and that the studies of Zhang selected inappropriate in vitro concentrations. However, that argument ignores the finding that the dose-response relationship in Figure 5 is flat (i.e., it almost does not matter where the in vitro data lie on the abscissa).

Other evidence supports the conclusion that the enhancing effect of volatile anesthetics on GABA_A receptors minimally influences MAC. Knockout of the ß₃ subunit of GABA_A increases enflurane MAC by 26% but increases halothane MAC by only 9% (102), and even these small increases may be attributable to compensation for lack of the subunit (41). Such results for inhaled anesthetics quantitatively agree with those found by Jurd et al. (1) for ß₃ N265M knockin mice unresponsive to the anesthetic effects of propofol: enflurane MAC increased by 15% and halothane MAC by 21%. The mutation knocked into the mice had normal in vitro sensitivity to enflurane and halothane. The MAC for a series of fluorinated alkanols (103) bears no consistent relationship to their capacity to enhance the effect of GABA on GABA_A receptors (104). Enflurane and halothane enhance GABA-mediated chloride conductance in rat hippocampal neurons (105), but a given MAC multiple of enflurane has twice the effect of halothane (i.e., the result is not quantitatively consistent across anesthetics.)

In summary, anesthetizing concentrations of many inhaled anesthetics enhance the in vitro action of GABA as much as do etomidate and propofol, anesthetics thought to produce immobility solely by enhancing the action of GABA on GABA_A receptors. Also, studies of ß₃ knockout and knockin mice indicate that GABA_A receptors may have a mediating role, albeit a minor one, in immobility. These findings suggest that enhancement of GABA actions should underlie at least part of the immobility produced by inhaled anesthetics. However, drugs that block GABA_A receptor function in vivo do not markedly increase MAC for inhaled anesthetics; the increase parallels that for ketamine, an anesthetic not thought to act by enhancement of GABA_A receptors; the increase in MAC is the same for inhaled anesthetics that do and do not enhance the effect of GABA in vitro; and, similarly, the capacity of fluorinated alcohols to enhance the effect of GABA does not predict their potency. Thus, in vivo results suggest a minor or no role for GABA_A receptors as mediators of immobility, whereas in vitro receptor studies and parallels with anesthetics such as propofol support a role. Although GABA_A receptors are unlikely mediator candidates, a conclusive determination may require the development of knockin mice with GABA_A receptors that respond normally to GABA but are not potentiated by inhaled anesthetics.

Potassium Channels.
Because potassium channels are numerous and diverse and because an increase in potassium conductance can decrease the excitability of the nervous system, potassium channels are plausible candidates as mediators of the immobility produced by inhaled anesthetics. Franks and Lieb (106) found that inhaled anesthetics potentiate potassium leak channels, and therefore reduce neuronal excitability, in the pond snail genus Lymnaea. Mammalian potassium channel subunits of the KCNK (K is potassium, CN is channel, and K is the subfamily of potassium channels) subfamily of leak channels (channels whose currents are not strongly gated by voltage) have properties similar to the properties of the Lymnaea channel, including activation by inhaled anesthetics (107–110). Although the KCNK subfamily provides a plausible anesthetic target, are there any holes in this leak theory?

No present evidence demonstrates that potassium channels directly mediate immobility. Only one (negative) study of inhaled anesthetic potency in potassium channel knockout mice has been reported (111). One study of the nonspecific KCNK activator riluzole supports the hypothesis that, although this drug has anesthetic properties, this action occurs in the brain and not the spinal cord (Gray A, Zhang Y, unpublished data). That is, IV and intrathecal infusions of riluzole that do not produce permanent injury or death do not decrease isoflurane MAC, indicating that KCNK channels do not mediate MAC. In summary, present evidence does not indicate that a specific potassium channel (or channels) mediates the immobility produced by inhaled anesthetics, but the large number and diversity of potassium channels precludes a conclusion that they play no role.

Excitatory Ionotropic Ligand-Gated and Voltage-Gated Channels
5-HT₃ Receptors.
Although most 5-HT receptors couple to second-messenger systems, potentially leading to various downstream effects, the 5-HT₃ receptor is a cationic ionophore with considerable sequence homology to GABA_A, glycine, and nicotinic acetylcholine receptors. Thus, not surprisingly, some inhaled anesthetics potentiate the ionophoric response of the 5-HT₃ receptor (112,113). This in vitro proexcitatory effect of some anesthetics suggests that the 5-HT₃ receptor cannot mediate immobility. Consistent with this conclusion, in rats, blockade of the 5-HT₃ receptor by systemic (114) (Fig. 6) or intrathecal (Flood, unpublished data) administration of ondansetron does not affect the MAC of isoflurane.

View larger version (17K): [in this window] [in a new window]   Figure 6. In rats, blockade of the 5-hydroxytryptamine-3 receptor by systemic (114) (Fig. 6) or intrathecal (Flood, unpublished data) administration of ondansetron does not affect the minimum alveolar anesthetic concentration (MAC) of isoflurane, even at doses of ondansetron modestly below the lethal dose. These findings indicate that this receptor does not directly mediate the immobility produced by inhaled anesthetics.

If antagonism of nicotinic receptors by volatile anesthetics decreases motor responses to noxious stimulation, the administration of a nicotinic antagonist (mecamylamine) to rats should, but does not (Fig. 7), decrease the volatile anesthetic concentrations required to produce immobility (42,43). Similarly, the administration of the classic agonist nicotine (at concentrations known to have behavioral affects attributed to nicotinic modulation) (117) does not alter the MAC of isoflurane (42). Finally, nonimmobilizers (inhaled compounds predicted from their hydrophobicity to have anesthetic activity that do not cause immobility alone or add to the anesthesia provided by known anesthetics) (5) inhibit nAChRs (118). Such observations suggest that nAChRs do not mediate inhaled anesthetic-induced immobility.

View larger version (52K): [in this window] [in a new window]   Figure 7. Administration of nicotinic (mecamylamine) or muscarinic (atropine or scopolamine) antagonists to rats does not decrease the isoflurane concentrations required to produce immobility (42,43). These and other data indicate that cholinergic inhibition by volatile compounds does not mediate anesthetic-induced immobility. MAC = minimum alveolar anesthetic concentration.

Numerous results suggest the potential importance of NMDA receptors as mediators of the immobilizing capacity of inhaled anesthetics. Diethyl ether, chloroform, methoxyflurane, halothane, enflurane, and isoflurane decrease glutamate-stimulated binding of MK-801 to the NMDA receptor (MK-801 [also called dizocilpine] is a standard blocker of NMDA receptors). Application of glutamate normally increases binding of MK-801, which thereby serves as an indication of NMDA receptor activity. Thus, a decrease of binding produced by the volatile anesthetics indicates decreased NMDA receptor activity (121). Inhaled anesthetics inhibit NMDA receptors in receptor expression systems (87). Nitrous oxide and xenon have greater in vitro effects than potent inhaled anesthetics on NMDA receptors (122), and their actions on NMDA receptors and/or sodium channels may contribute more to their immobilizing capacity than is the case with anesthetics such as isoflurane. As discussed above, both enflurane (45) and ethanol (44) can decrease NMDA-mediated glutamate currents in motor neurons in the spinal cord slice, an effect independent of effects on GABA_A and glycine receptors.

Masaki et al. (123) applied the NMDA antagonist D (-)-2-amino-5-phosphonopentanoic acid (D-AP5) intracerebroventricularly and noted a sevoflurane MAC-sparing effect, suggesting a supraspinal effect. Ishizaki et al. (124,125) found a maximal decrease of 30% in isoflurane MAC in rats given intrathecal bolus doses of the NMDA antagonists AP5, MK-801, CPP, and 7CKA. McFarlane et al. (126) reported an 80% decrease in MAC of halothane from IV application of the competitive NMDA receptor antagonist CGS 19755.

MK-801 administration to rats decreases isoflurane MAC, and the decrease primarily correlates with MK-801 concentrations in the spinal cord, rather than whole brain or cerebral cortex concentrations, a finding consistent with mediation of the decreased MAC by the cord (50). Do inhaled anesthetics produce effects similar to MK-801: do they block NMDA receptors so as to produce immobility (i.e., do NMDA receptors in the spinal cord directly mediate MAC)?

Temporal summation is the cumulative effect of a succession of repeated stimuli presented at sufficiently close intervals (127). Phrased another way, a shortening of the intervals between stimuli increases the collective stimulation and is more likely to provoke a response. In vitro, blocking NMDA receptors blocks temporal summation, indicating that NMDA receptors lie in the pathway that mediates temporal summation (128). These observations suggest that persistence of temporal summation during anesthetic administration reflects intact NMDA receptor function. Studies with isoflurane suggest that approximately 40% of the generation of movement evoked by noxious stimulation (MAC) depends on the interstimulus interval, suggesting the persistence of temporal summation and transmission via NMDA pathways (Fig. 8). Strengthening this interpretation, if temporal summation persists, then the administration of the NMDA blocker MK-801 should abolish summation, and that is what appears to occur (Fig. 8). Also consistent with the interpretation that isoflurane does not block temporal summation, electrophysiologic studies of neuronal windup show that temporal summation can occur during anesthesia (129). However, another interpretation is possible: perhaps isoflurane causes suppression of NMDA receptor transmission, and blockade with MK-801 simply substitutes for the blockade produced by isoflurane.

View larger version (19K): [in this window] [in a new window]   Figure 8. Isoflurane concentrations (mean ± SD) required to produce immobility in response to noxious stimulation (intermittent 50-V, 0.5-ms square-wave pulses) increase with decreasing interstimulus interval (ISI) (127). The concentration required to suppress movement at an ISI of 0.1 s does not differ from the minimum alveolar anesthetic concentration (MAC). MK-801 (an N-methyl-D-aspartate antagonist that impairs temporal summation) administration added to isoflurane administration prevents shorter ISIs from provoking movement at larger isoflurane concentrations.

In summary, evidence suggesting the importance of NMDA receptors to MAC include the observations that (1) in vitro, inhaled anesthetics can block NMDA receptors; (2) blockade of NMDA receptors (e.g., with MK-801) decreases MAC; and (3) this decrease in MAC correlates with the concentration of MK-801 in the lower part of the spinal cord. However, studies indicating that temporal summation persists during isoflurane anesthesia and that MK-801 blocks such summation argue against a role for NMDA receptors as mediators of isoflurane MAC. Conversely, the contrasting data for xenon and isoflurane suggest the possibility that NMDA receptors might mediate a portion of the capacity of some anesthetics (xenon), but not others (isoflurane), to produce immobility. None of these results excludes a role for NMDA receptors in the mediation of other aspects of anesthesia, including the capacity of inhaled anesthetics to impair learning and memory (i.e., produce amnesia).

AMPA Receptors.
AMPA receptors mediate the fast (initial) component of excitatory postsynaptic transmission and provide plausible targets for volatile anesthetics. Such fast transmission contrasts with the slower (later) excitatory transmission mediated by NMDA receptors. Competitive antagonists for AMPA receptors, administered IV (130) or intrathecally (124), markedly decrease the concentration of halothane that causes immobility, findings consistent with the mediation of anesthetic-induced immobility by AMPA receptors. However, although showing that movement requires AMPA receptor function, these results do not prove or disprove that anesthetics impair AMPA receptor function. Indeed, one study found that intrathecal antagonists fail to affect isoflurane MAC (131), but this study used smaller doses of antagonist than in the study that did show a decrease.

In vitro electrophysiologic studies demonstrate that volatile anesthetics inhibit AMPA-mediated excitatory postsynaptic currents, that enflurane inhibits the postsynaptic action of glutamate on AMPA receptors in mouse spinal cord (45), and that halothane similarly affects the hippocampus at larger MAC values (25). Also, clinically relevant concentrations of volatile anesthetics partially inhibit native and recombinant AMPA receptors activated by exogenous agonists (45,87,132–134). Whether minor inhibition of AMPA receptors materially influences neuronal network activity remains unclear.

Genetically modified mice provide insights into the importance of inhibition of AMPA receptors to volatile anesthetic actions. GluR1–4 subunit combinations form AMPA receptors. Mice lacking the GluR2 subunit have inconsistent phenotypic responses to inhaled anesthetics. They show a greater sensitivity to halothane, isoflurane, and sevoflurane than wild-type littermates for loss of the righting reflex and antinociception but no difference in MAC (133). In electrophysiologic studies, clinical concentrations of isoflurane and halothane minimally inhibit AMPA receptors (both GluR2-containing and -deficient) (133). Therefore, blockade of AMPA receptors cannot underlie enhanced sensitivity (for righting reflex and antinociception) in GluR2 knockout mice. Other evidence suggests that decreased excitatory neurotransmission due to altered AMPA receptor kinetics renders GluR2-deficient neurons more sensitive to anesthetics acting on other target receptors (135).

Thus, studies using GluR2 knockout mice support the postulate that different neuronal circuits mediate MAC versus antinociception and loss of the righting reflex (133). Motor neurons in the ventral spinal cord (i.e., that might mediate MAC) lack GluR2 subunits (136–138), but neurons that mediate the righting reflex and nociception do express GluR2-containing AMPA receptors. Thus, knockout of the GluR2-containing receptors should not change MAC but might change the righting reflex and antinociception. Furthermore, neuronal circuits in the spinal cord that mediate MAC do not appear to require the GluR2 subunit. In summary, although in vivo data suggest that ablation of the GluR2 subunit does not affect MAC, this cannot exclude the possibility that other subunits of AMPA receptors contribute to MAC.

Kainate Receptors.
Combinations of GluR5–7, KA1, and KA2 subunits form the kainate subtype of ionotropic glutamate receptors (139). In vitro, inhaled anesthetics enhance currents mediated by kainate receptors containing GluR6 (51). However, GluR6 knockout mice have a normal isoflurane MAC (Sonner, unpublished data). These findings are difficult to interpret because kainate receptors can be assembled from other kainate subunits, even in the absence of the GluR6 subunit. Perhaps other forms of GluR6 mutation, such as GluR6 editing mutant mice (140), will provide insights that a knockout may not. Presently, the importance of kainate receptors to MAC remains speculative.

Sodium Channels.
A lingering dogma asserts that clinical concentrations of inhaled anesthetics do not block axonal conduction and, thus, voltage-dependent sodium channels. However, more recent data show that anesthetics can inhibit presynaptic terminal release of neurotransmitters, particularly glutamate (141). Inhibition of these presynaptic sodium channels could produce this action. A tantalizing finding is that systemic injection of lidocaine decreases MAC in animals (142,143) and humans (144). The effect has a floor (a maximum decrease of 40%–50%) (143) and thus cannot simply result from blockade of all nerve conduction.

Molecular cloning provides a basis for reconciling old ideas about axon conduction with anesthetic inhibition of sodium channel function. Cloning reveals nine different sodium channel subunits and three ß subunits differentially localized in the brain and the periphery. Different members of this family may vary in their anesthetic sensitivity. Several volatile anesthetics, including 1A, or F3, may inhibit synaptosomal sodium channels (141,145–156), but the nonimmobilizer 2N, or F6, does not inhibit (157). Thus, sodium channels pass this test of anesthetic relevance. Studies of volatile anesthetic action on cloned sodium channels show somewhat divergent results that, in part, may be due to differences in the origin (neural versus nonneural origin of the channel) (148,158,159).

Thus, consistent with the notion that sodium channels directly mediate anesthesia, several inhaled anesthetics, but not the nonimmobilizer F6, inhibit sodium channels. Key questions remain: which sodium channel subunits exhibit sensitivity or resistance to anesthetics, and what is the neuroanatomical localization of the sensitive channels? Finally, what is the mechanism of channel inhibition—do anesthetics act directly on the protein or act through protein kinase C (PKC) or other modulators?

Metabotropic Receptors
Much of the preceding discussion focused on ion channels because so much is known about inhaled anesthetic effects on these receptors. In contrast to ion channels, metabotropic receptors act through second messengers released by an intracellular component such as the G-protein coupled to the receptor (160). Often, these effects are prolonged, lasting seconds to minutes. They influence the electrophysiologic properties of a neuron, including membrane resting potentials (160). As noted below, changes in some of these receptors (e.g., 5-HT₂, opioid, and ₂ adrenoreceptors) can significantly decrease MAC. However, of these examples, only the 5-HT₂ receptor might directly mediate the capacity of inhaled anesthetics to produce immobility.

Neuronal Muscarinic Acetylcholine Receptors.
Isoflurane inhibits muscarinic receptors through the activation of PKC (117). If muscarinic antagonism by volatile anesthetics decreases motor responses to noxious stimulation, muscarinic antagonists (e.g., atropine or scopolamine) given to rats should, but do not (Fig. 7), decrease the volatile anesthetic concentrations required to produce immobility (43), and intrathecal administration of atropine does not alter the MAC of isoflurane (43). Even the combination of muscarinic and nicotinic blocking drugs does not change MAC (Fig. 7). Thus, the hypothesis that cholinergic inhibition by volatile compounds mediates anesthetic-induced immobility cannot be correct.

Opioid Receptors.
Analgesia constitutes a nominal, albeit sometimes contested, feature of the anesthesia provided by inhaled anesthetics. Thus, it would be reasonable to suppose that inhaled anesthetics enhance the release of endogenous opioids and that this release contributes to the anesthetic state. However, inhaled anesthetics do not appear to increase endogenous opioids in cerebrospinal fluid (161), and they do not suppress autonomic or ventilatory responses to surgical stimulation at concentrations that suppress movement (162). Also, small doses of opioids markedly decrease inhaled anesthetic concentrations that prevent movement (163). That is, the opioids supply something (analgesia) that inhaled anesthetics do not. Finally, the administration of enormous doses of naloxone, a drug that blocks the effect of opioids, has no effect on the MAC of halothane (Fig. 9) (164) or nitrous oxide (165) in rodents.

View larger version (20K): [in this window]