Muscarinic Signaling in the Central Nervous System: Recent Developments and Anesthetic Implications

Until the molecular cloning of the first muscarinic receptor in 1986, investigators depended on pharmacologic tools, primarily selective antagonists, to define the several subtypes of this receptor family. Unfortunately, none of the known antagonists are completely selective, so that subtypes had to be defined by measuring the binding properties of several compounds. Thus, equilibrium binding studies with pirenzipine initially indicated the existence of two classes of cerebral muscarinic receptors, named M1 and M2. [9](Names with a capital "M" indicate pharmacologically defined subtypes, whereas those with a small "m" indicate clones.) Kinetic studies allowed differentiation of three subtypes, [10,11]and with the development of novel antagonists this number was expanded to four (M1-M4)[12]:Table 1indicates the relative selectivity of the commonly used muscarinic antagonists, and relates the pharmacologically defined types to the cloned receptor genes. An excellent recent review of this subject is available. [13].

The first muscarinic receptor (the m1 subtype) was cloned in 1986. [8]Since then, a remarkable amount of information has been gathered about the molecular biology of muscarinic signaling. Not only have the main classes of muscarinic receptor subtypes been cloned, but detailed information on their structure-activity relationship is available, which will prove useful in the development of new, highly selective agonist and antagonist drugs.

When the DNA encoding the muscarinic receptor was isolated, it was compared to already cloned sequences, and its closest relative was found to be the visual pigment rhodopsin. [8]Although at first this may appear to be an unusual relationship, the sequence similarity relates to the finding that similar intracellular systems transduce the signals generated by these molecules. In both cases, a guanosine trisphosphate (GTP)-binding protein (G protein) transduces a ligand-induced conformational change in the membrane receptor to intracellular signaling systems. Details of G protein function and their relevance to anesthesia were reviewed recently [14,15]and will not be discussed here in detail. Several hundred receptors have now been shown to belong to the G-protein-coupled receptor superfamily, of which the muscarinic receptors form a small but important cluster.

G-protein-coupled receptors all show the same molecular pattern in their amino acid sequence: most are approximately 500 amino acids in length and include seven hydrophobic domains of approximately 20 amino acids each. These domains are thought to form alpha-helices traversing the membrane, leading to the designation of these proteins as seven-transmembrane, serpentine, or heptahelical receptors (Figure 1). The presence of seven transmembrane segments so well predicts that the protein is a G-protein-coupled receptor that a number of "orphan" clones (cloned DNA to which no function has been assigned) are presumptively classified as G-protein-coupled receptors on this basis only.

The G proteins stimulated consist of a large family of related heterotrimeric proteins. They control a number of intracellular systems. Examples relevant for muscarinic signaling are Gⁱ, G^o, and G^q. Cloning studies have shown that most of these "subtypes" actually are subfamilies of closely related proteins, with the total number of members as yet undetermined. Gⁱinhibits adenylate cyclase, resulting in decreased cyclic adenosine monophosphate (cAMP) levels. G^qand G^oactivate phospholipase C, which metabolizes phosphatidylinositolbisphosphate to inositoltrisphosphate (which releases Calcium²+ from intracellular stores), and diacylglycerol (which activates protein kinase C). In addition, G proteins can activate ion channels, either through a direct interaction with the channel (e.g., activation of inwardly rectifying Potassium channels by Gⁱsubtypes), or through liberation of a diffusible messenger such as Calcium²+(e.g., I^K(M) inhibition by G^q/11 and N-type Calcium channel inhibition, discussed later).

Once the DNA sequence of one muscarinic receptor was known [8]other subtypes were isolated in rapid succession. [16–20]Thus far, five muscarinic receptors have been cloned, [21]designated m1 through m5. The clones show significant evolutionary conservation, with 89–98% amino acid sequence identity among mammalian species. Even in evolutionary distant animals like insects, muscarinic receptors are easily identified and, in fact, play important roles. [22]The apparent excess of cloned subtypes over pharmacologically defined subtypes is typical for G-protein-coupled receptors, and presumably allows finer regulation of receptor expression. The five subtypes fall into two groups, the "odd" (m1, m3, m5) and the "even" (m2, m4), based on sequence homology and second messenger signaling. The odd group signals primarily through increases in intracellular Calcium²+; the even group through decreases in cAMP production. Although the clones were numbered simply in the order they were identified, the m1 clone happens to show most of the properties of the pharmacologic M1 type, and the m2 clone those of the M2 type.

The cloned muscarinic receptor subtypes and other members of the superfamily have been used to determine the intramolecular sites involved in ligand binding and G protein coupling. As there is a high degree of similarity between subtypes, with 160 of the approximately 500 amino acids invariant, [21]specificity of ligand binding and G protein coupling must depend on relatively small changes in structure. In agreement with their functional grouping, the odd and even receptors show particularly high within-group similarities. [18,20]However, the third intracellular loop (i3, Figure 1, approximately 150 amino acids long) is highly variable, with the exception of the first and last 15–20 amino acids. Studies of bacteriorhodopsin (a related molecule for which a three-dimensional structure has been established) and adrenergic receptors have demonstrated that ligand binding takes place in a pocket, primarily consisting of the second, third, and seventh transmembrane regions (t2, t3, and t7, respectively), [23,24]whereas the i3 loop and the carboxyterminus (C) are involved in G protein binding [25–27]and regulation through phosphorylation. [28]In muscarinic receptors, the G protein binding specificity has been mapped to a small domain of approximately 20 amino acids in the i3 loop. [29–31]As in adrenergic receptors, [25,32]agonist binding to muscarinic receptors is initiated by contact with a specific aspartate residue in t3. [33–35]Exchange of (part of) t6, i3, t7, and C between the m2 and m3 subtypes resulted in a change in G protein coupling and subtype-selective ligand binding. [36]Mutation studies have identified a series of threonine and tyrosine residues in t3, t5, t6, and t7 that are of importance in agonist, but not antagonist binding, [37]again demonstrating the role of transmembrane domains for ligand binding. Thus, the functional domains of these receptors are well established.

As stated earlier, Calcium²+ and cAMP are the best-described intracellular second messengers of the "odd" and "even" receptor groups, respectively. In the brain, with its primary function of electrical signaling, and in other organs such as the heart, muscarinic systems also transduce their actions through changes in membrane potential. Several ion conductances have been shown to be affected by muscarinic stimulation, and the effects are most easily classified as depolarizing (stimulatory) or hyperpolarizing (inhibitory). [13,38]The best-known depolarizing effect is by inhibition of a noninactivating voltage-gated Potassium channel (I^K(M)) that clamps the membrane at its resting potential. [39]Stimulation of (primarily) M1 receptors inhibits this channel, resulting in a neuron more likely to fire when depolarized by other agonists. This effect has been studied in some detail, and has been shown to be mediated through the G^q/11 G protein. [40,41]A second depolarizing influence of muscarinic signaling is through inhibition of a Calcium-gated Potassium current (I^K(Ca)), which normally hyperpolarizes the cell when an action potential leads to influx of Calcium²+ through voltage-activated Calcium channels. [42]M1 receptors seem to be the primary subtype involved, which is surprising, because their stimulation leads to increases in intracellular Calcium²+ and therefore activation rather than inhibition of I^K(Ca) would be expected. Such is indeed seen in transfected cells, [43]but it has not been observed in neurons. The mediator involved has not been defined. In addition to I^K(M) and I^K(Ca), stimulation of M1 receptors can inhibit a noninactivating voltage-independent potassium current (I^K(leak)). [13,41].

Inhibitory effects of muscarinic signaling are found in many neurons, and the best-defined pathway is by muscarinic effects on voltage-activated Calcium currents (I^Ca). This appears mediated by m2 or m4 receptors activating G^oG proteins. [44]The N-type Calcium channel involved is sensitive to the Calcium-channel blocker omega-conotoxin GVIA but not to dihydropyridines. Another inhibitory effect of muscarinic signaling relevant to anesthesiologists is the activation of cardiac inwardly rectifying Potassiumm channels (K^ir) through M2 stimulation.

This is responsible for the cardiac side effects of anticholinergic drugs, and has been shown to result from direct activity of stimulated Gⁱproteins on the channel. [45]A recent review on the subject is available. [46]Much interest has been generated by findings that the G protein beta gamma subunit, traditionally considered inactive, appears to play an important role in this effect. [47–50]Although most data have been obtained in atrial cells, there is evidence that similar pathways exist in the brain. [51].

(Figure 2) summarizes the intracellular pathways involved in muscarinic signaling. This area is the subject of active investigation, and several recent, more extensive reviews are available. [13,38,52–55].

The cloning and subsequent study of the muscarinic receptors have provided a framework on which to base studies of interactions with anesthetics. It should now be possible to map the anesthetic interference with muscarinic functioning, described later, to a specific domain within the receptor or G protein, using standard molecular biology techniques such as the construction of chimeric receptors and site-directed mutagenesis. These techniques allow selective modification of the amino acid structure of proteins, so that domains thought to be relevant for anesthetic interactions can either be altered or exchanged with a corresponding domain from a related protein not sensitive to anesthetics. The role of such a domain, and even of specific amino acids within the domain, in anesthetic action can thus be defined. Although the techniques are subject to a number of caveats (the major one being that amino acid modification can change the conformation of a protein in unpredictable ways and with unpredictable functional results), they are potentially very powerful tools for determining the intramolecular site(s) of action of anesthetics. This information would not only advance our understanding of anesthetic-muscarinic interactions, but would make the muscarinic family a model for anesthetic-protein interactions in general.

The prominent role in memory and consciousness played by the CNS cholinergic system is evident from the effects of the muscarinic antagonist scopolamine.

Correlating these clinical findings with a neuronal substrate has proved more difficult. However, advances achieved using a variety of investigational techniques (intracranial drug administration, in vivo microdialysis, intracellular recording, chemical neuroanatomy, and molecular biology) have expanded our understanding of CNS muscarinic pathways and their roles considerably. Excellent summaries can be found in several recent texts. [56,57].

The development of antibodies against the acetylcholine-synthesizing enzyme choline-O-acetyltransferase (ChAT, EC 2.3.1.6.), [58]and later the development of in situ hybridization techniques to localize its mRNA, [59]allowed charting of the organization and distribution of cholinergic neurons in detail. Figure 3shows a simplified scheme of our current understanding of the cholinergic projection systems. [60]Two groups of nuclei are involved: a brain stem group (laterodorsal tegmental nucleus and pedunculopontine tegmental nucleus), [61]and a basal forebrain group (medial septum, diagonal band of Broca, and nucleus basalis of Meynert). The brain stem group projects rostrally along a dorsal pathway to nuclei in the thalamus and the pontine reticular formation, [62]and along a ventral pathway to the basal forebrain. [63]A serotonergic pathway with a similar distribution also exists and may modulate consciousness in concert with the muscarinic system, as described later. [63]The basal forebrain projects to the neocortex and the hippocampus, [64,65]as well as to the amygdaloid complex. [66]In addition to these main pathways, cholinergic as well as noncholinergic projections exist to a number of neuronal structures in the caudal brain stem. [67,68].

Of course, techniques based on localizing sites of acetylcholine synthesis do not differentiate between muscarinic and nicotinic pathways. However, subsequent studies based on receptor expression (see later) have shown that muscarinic and nicotinic systems have different distributions. The nicotinic system, although subjected to intense investigation in view of the unusual aspects of nicotine addiction [69]and the potential use of nicotine as a cognitive enhancer in Alzheimer's disease, [70]is beyond the scope of this article. A number of recent reviews on the pharmacology of nicotine addiction are available. [71–74]Nicotinic receptor distribution has been mapped using both immunohistochemical techniques [75]and radiolabeled nicotine. [76].

The protein and mRNA distribution of the muscarinic receptors have been mapped using antibodies against individual muscarinic subtypes, as well as molecular probes recognizing their nucleotide sequences. [77,78]In both cortex and hippocampus, where projections from the basal forebrain terminate, the m1 subtype predominates, representing 35–45% of muscarinic receptor protein. It is followed, in order of decreasing protein abundance, by the m2, m4, and m3 subtypes. The m2 subtype is the most common one in thalamus (50% of muscarinic protein), and in brain stem and cerebellum (80% of muscarinic protein). Figure 4illustrates this distribution. These data are from rats, but similar findings were obtained in humans. [78,79].

The afferent inputs into the primary cholinergic nuclei include fibers from cortical areas, hippocampus, stria terminalis, preoptic area, thalamus, hypothalamus, amygdala, and several brain stem nuclei. Noncholinergic neurons link the basal forebrains of each hemisphere together. [80]Not all of these fibers have been shown to make actual contact with forebrain cholinergic neurons, although some have, e.g., hypothalamic axons. [81]Chemically, the main inputs into the basal forebrain are serotonergic, (nor)adrenergic, and dopaminergic. GABAergic input into cholinergic forebrain neurons has similarly been demonstrated. [82]The noradrenergic fibers have been shown to synapse onto cholinergic cells, [83,84]and findings that norepinephrine enhances cortical information processing [85]makes important functional interactions between the adrenergic system, with its center in the locus coeruleus, and the muscarinic system likely. Although beyond the scope of the current review, the locus coeruleus may well play an important role in anesthetic action, particularly as a site of action of alpha²-adrenergic agonists. Interested readers are referred to the literature. [85–91]A variety of brain areas have inputs into the cholinergic brain stem nuclei, [92]but the primary tracts consist of serotonergic afferents from the raphe nuclei, with a smaller contribution of noradrenergic fibers from locus coeruleus and other areas. [93]Serotonin modulates rapid eye movement (REM) sleep [94]and serotonergic fibers have been shown to contact cholinergic neurons in the mid-brain [95]and to hyperpolarize the cholinergic neurons of these regions. [96]A main function of the brain serotonin system is thought to involve facilitation of motor output and concurrent inhibition of sensory processing, [97]and part of the latter effect may be brought about by interaction with muscarinic systems. In addition, the analgesic properties of serotonin [94]and the antiemetic properties of the 5-HT sub 3 antagonist ondansetron suggest further interaction between muscarinic and serotonergic systems.

In addition to the projection pathways described here, cholinergic interneurons are found in a number of cerebral structures, including the striatum, various cranial nerve nuclei, and spinal cord. Particularly the oculomotor nucleus and the nucleus ambiguus stain intensely for choline-O-acetyltransferase. Such interneurons are considered to have modulatory functions in a variety of systems. [83]Acetylcholine is similarly an important transmitter in the vestibulocerebellar pathways, which may explain the effectiveness of muscarinic antagonists in inhibiting vestibular vertigo. [98].

Most studies investigating the effects of muscarinic agonists and antagonists on consciousness have focused on sleep, particularly REM sleep. [99]Early investigations showed that intravenous administration of cholinomimetic drugs induces a REM-like state in decerebrate animals, [100]and that, conversely, anticholinergic agents decrease or abolish REM sleep in humans. [101]These effects have since been localized to brain stem cholinergic neurons: in vivo recordings show that these cells fire tonically during both wakefulness and REM, but not during non-REM sleep. [102,103]This tonic activity is associated with cortical activation, and the highest acetylcholine levels in both brain stem [104,105]and cortex [106](mediated through the basal forebrain [107]) are found during these states. Acetylcholine release in the thalamus is similarly dependent on the state of wakefulness. [108]Destruction of brain stem cholinergic neurons by toxins is associated with loss of REM sleep, and particularly with loss of the ponto-geniculo-occipital waves that characterize this state. Conversely, microinjection of cholinergic agonists into this area induces REM sleep.*.

Administration of the cholinesterase inhibitor physostigmine induces arousal in awake subjects, but REM sleep in those asleep. [109]Thus, cholinergic signaling systems enhance either wakefulness or REM sleep, depending on the prior state of the subject. Interactions between cholinergic and other monoaminergic systems, mentioned earlier, appear essential for this function. [110,111]For example, whereas cholinergic neurons increase firing before and during REM sleep, noradrenergic and serotonergic groups decrease firing rates at those times. In addition, if monoamine stores are depleted by reserpine, physostigmine no longer induces arousal, but instead causes REM sleep.

Interestingly, activation of pontine cholinergic systems can lead to prolonged enhancement of REM sleep. [112]In cats, after microinjection with the cholinergic agonist carbachol, the amount of time spent in REM sleep, as well as the incidence of ponto-geniculo-occipital waves, remained elevated for up to a week. During awake periods, increased behavioral arousal was observed, but this effect was limited to 1 day only. The distribution of muscarinic receptor subtypes in the cat brain stem has recently been determined by pharmacologic means [113]; the primary subtypes were M1, M2 and M3, found in a heterogeneous distribution with greatest level of binding in the locus coeruleus.

Nicotinic signaling may play a role in this system as well. Microinjection of nicotine into the medial pontine reticular formation induces REM sleep in cats. [114]In accordance, when smoking and nonsmoking volunteers were compared as to the amount of time spent in REM sleep, smokers had a lower REM latency (time from onset of sleep to onset of REM) and greater REM density (ratio between an eye movement activity score and total REM sleep time) than nonsmokers. [115]There is, however, evidence that such effects of nicotine are brought about by the compound acting as a presynaptic agonist that releases acetylcholine, which subsequently interacts with muscarinic receptors. [116]Nonetheless, the observation that nicotine injection into the interpeduncular nucleus prolongs recovery from halothane anesthesia [117]suggests that the nicotinic cholinergic system may sum with or facilitate the mechanisms that produce general anesthesia.

In summary, muscarinic modulation of consciousness takes place primarily in the mid-brain, and stimulation of the system induces activated behavior. In awake states this becomes evident as mild hyperactivity, characterized as "awake, non-mobile behavior," [118]whereas during sleep it becomes evident as REM sleep, which is associated with active cortical processing and dreaming. Muscarinic stimulation induces a diffuse increase in alertness, making the individual more receptive to external and internal inputs. Muscarinic inhibition, in contrast, leads to sedation or non-REM sleep, depending on the prior state of the subject. No cerebral system works in isolation, and these effects of muscarinic signaling appear to depend on the functioning of monoaminergic systems.

In contrast to muscarinic effects on consciousness, localized mainly to the brain stem, the effects on memory and learning are most easily explained by the projections from the basal forebrain to cortex and particularly hippocampus, a structure known to be of importance in memory. Many studies, using a variety of experimental systems, have demonstrated that cholinergic antagonism interferes with learning behavior, and that cholinesterase inhibitors can enhance learning. [119]The structural basis for these effects has been more difficult to determine. Evidence thus far suggests that muscarinic systems act by facilitating sensory information processing. During sensory stimulation, acetylcholine is released from cortical areas, particularly those involved in processing the stimulus. [120]In most sensory systems, acetylcholine has been shown to increase neuronal responses to presented stimuli. Similar to the long-term potentiation seen in the hippocampus, acetylcholine can induce long-term facilitation in several systems, when applied at the same time as a depolarizing pulse. These facilitating effects can last up to an hour, and could play a role in information processing and imprinting. The molecular basis for this activity is unknown, although muscarinic effects on potassium channels, discussed earlier, are likely to be important. [120].

A strong impetus to the study of muscarinic effects on memory was provided by findings that correlate cognitive and memory deficits in Alzheimer's disease patients with changes in cholinergic innervation. Decreased choline-O-acetyltransferase activity in cortex and hippocampus is one of the most significant markers for this disease, and cholinergic neurons in the basal forebrain area degenerate selectively in proportion to these changes. Cholinergic innervation of the amygdala is similarly depleted. [121]Interruption of the projection from basal forebrain to cortex in animals, and administration of scopolamine in humans, induce memory dysfunction that is similar to that seen in Alzheimer's disease. Thus, the so-called "cholinergic hypothesis," originally proposed to explain normal aging processes [122]but later applied to Alzheimer's disease as well, states that the cholinergic forebrain system is essential for cortical cognitive processing, and that the learning and memory deficits observed in the disease are caused by failure of this system. Despite the appeal of this theory, one should realize that it is equally possible that the initial dysfunction occurs in the cortex, and that neuronal degeneration in the basal forebrain results from a lack of target cells. Supporting this are findings that nerve growth factor is produced by cortical cells, binds to receptors on the cholinergic axon, and is internalized and transported to the basal forebrain. [123]Lack of nerve growth factor leads to degeneration of cholinergic neurons, as indicated by studies that show how transsection of their cortical projections leads to death of these cells, a process preventable by nerve growth factor application. [124]Despite alternative theories, the cholinergic hypothesis of Alzheimer's disease has stimulated a great amount of research. Clinical studies of the effects of cholinesterase inhibitors thus far have met with limited success, but pharmacologic investigations have resulted in a variety of novel compounds, some of which may be useful in the anesthetic setting, as described later. The realization that muscarinic systems may have limited reserve in the elderly reminds us to be careful when using muscarinic antagonists in these patients.

Systemically administered cholinergic agonists have antinociceptive actions, partially mediated by cerebral cholinergic pathways. [125]However, intrathecal administration of cholinergic agonists or anticholinesterases has a more potent antinociceptive effect, [126]as described later. In contrast to the detailed maps available of cholinergic pathways in brain, the connections of cholinergic neurons in spinal cord have not been described in detail. With the realization that spinal anticholinesterases represent a useful therapeutic method this will probably soon be corrected. Receptor binding studies with nonselective compounds have shown that most muscarinic binding sites are localized in the substantia gelatinosa in the dorsal horn, and in the motor neuron areas. [127,128]The first area is presumably the one involved in antinociceptive effects. The relatively selective M1 and M2 blocking agents pirenzepine and AFDX 116 both inhibit carbachol-induced antinociception, [129]suggesting that both types of receptors play a role. In addition, depletion of noradrenergic fibers by N-2-chloroethyl-N-ethyl-2-bromobenzylamine (DSP4) attenuated the antinociceptive effect of carbachol. Therefore, an interaction between adrenergic and cholinergic systems, as described earlier for cerebral systems, appears likely. However, the exact connections and role of muscarinic neurons in the spinal cord remain to be determined. Clinical effects are described later.

Cloning of the muscarinic receptors and subsequent understanding of their molecular biology has helped to map the locations of muscarinic neurons in detail, and this knowledge, in turn, has helped to elucidate some of the functions of this system. Many of these--effects on consciousness, memory, and pain--are of obvious relevance to anesthesia. Two questions therefore arise: are some of the effects of general anesthetics due to interference with muscarinic systems, and could manipulation of the CNS cholinergic systems be of benefit in clinical anesthesia? These issues are addressed in the next sections.

The unitary hypothesis of anesthesia, suggesting that the anesthetic state is brought about by the drug acting on a single site, has lost support in recent years as interactions of anesthetics with a variety of cellular components have been described. Most likely, interactions both with lipids and proteins (e.g., GABA^Areceptors and NMDA-activated channels) play a role in clinical anesthesia. It has become evident that muscarinic signaling is one of several systems affected significantly by anesthetics (for recent reviews on cellular effects of anesthetics, see references 130 and 131). Whereas initial electrophysiologic studies showed variable responses of cortical cholinergic neurons to volatile anesthetics, [132–134]more recent investigations have shown that most anesthetics depress muscarinic signaling.

Initial studies investigated the effects of anesthetics on binding of muscarinic agonists and antagonists. Aronstam and coworkers demonstrated that ether [135]and halothane, [136]as well as chloroform, enflurane, and isoflurane [137]increased muscarinic antagonist binding by decreasing the ligand-receptor dissociation rate. Because agonist binding was not affected, the clinical relevance of these findings is not completely evident. However, it does indicate an interaction between anesthetic and receptor. Presumably, this interaction takes place at hydrophobic domains of the receptor molecule that are relevant for antagonist, but not for agonist binding. [138]It should be noted that these studies did not study individual receptor subtypes, and that at times the anesthetic concentrations employed were quite high.

The same investigators showed that anesthetics also interfere with G protein function or receptor-G protein coupling. Nonhydrolyzable forms of GTP, such as GTP gamma S, irreversibly activate G proteins. [14]These dissociate from the receptor, and the receptor thereby shifts to a low-affinity state. Thus, addition of GTP gamma S will result in a shift in the agonist binding curve, the so-called GTP shift. A variety of volatile anesthetics eliminated this GTP shift in muscarinic systems. [135–137]Apparently, either the GTP gamma S was no longer able to activate the G protein (suggesting interference with GDP-GTP exchange) or the anesthetic stabilized the muscarinic receptor-G protein complex, so that the two could no longer dissociate.

Functional investigations have supported these binding studies. When cortical mRNA was expressed in Xenopus oocytes, muscarinic (primarily m1-mediated) Calcium²+ responses could be obtained from these cells. These were depressed by 2% enflurane. [139]Responses to injected GTP gamma S were similarly inhibited by the anesthetic, but responses to injected inositoltrisphosphate were not. This indicates that the signaling systems downstream of phospholipase C were unaffected (Figure 2), leaving the receptor or G protein as the most likely site of anesthetic action. Ethanol similarly depressed the function of these receptors. [140]We expressed cloned m1 muscarinic receptors in Xenopus oocytes and found their functioning inhibited by halothane, with approximately 60% depression observed in the presence of 0.34 mM halothane (approximately 0.6% in air at room temperature;Figure 5). [141]In contrast, when the AT^1Aangiotensin II receptor was expressed, no inhibitory effect of halothane was observed. The AT^1Areceptor employs the same intracellular signaling system as the m1 receptor (with the possible exception of the G protein coupled to), which indicates anesthetic interference with receptor or G protein function, or with coupling of the two. Conversely, in preliminary experiments isoflurane, at equal minimum alveolar concentrations, did not depress muscarinic signaling, [142]suggesting that muscarinic inhibition might be more relevant to side effects of anesthetic drugs than to anesthetic action per se.

As described earlier and indicated in Figure 2, the two main intracellular effects observed after muscarinic stimulation are increases in Calcium²+ and decreases in cAMP. In systems where muscarinic signaling has been studied, the pathway from inositoltrisphosphate to Calcium²+ release seems unaffected by anesthetics. [139,141],* The cAMP system has been reported to be affected variably. [143]Chloroform decreases basal adenylate cyclase activity, halothane increases it, and ether, isoflurane, and enflurane are without effect. Adenylate cyclase activation by forskolin (a direct stimulant) or isoproterenol (acting through beta-adrenergic receptors) was not affected by any of the anesthetics. Acetylcholine inhibited forskolin-stimulated adenylate cyclase, and this inhibition was blocked by ether, isoflurane, chloroform, and enflurane. However, as described earlier, this effect probably represents an action at the receptor or G protein level. Although a role for adenylate cyclase inhibition by volatile anesthetics seems doubtful, the cAMP system is nonetheless of importance for anesthesia induced by other drugs, as it has been shown clearly that inhibition of adenylate cyclase in the locus coeruleus mediates the hypnotic response to the alpha sub 2 agonist dexmedetomidine [87]and to opiates. [144].

In neuronal systems as well as in the heart, muscarinic signaling is mediated through effects on ion channels. The actions of anesthetics on muscarinic modulation of these channels have not been studied in sufficient detail to allow any definitive statements to be made.

There seems to be little doubt that many of the volatile anesthetics interfere with muscarinic signaling through an effect either on the receptor, or on receptor-G protein interaction. Although exact molecular localization of the site of action is feasible now that the muscarinic receptors and many G proteins have been cloned, it is currently undefined. One possibility would be a modulatory hydrophobic domain in the receptor molecule. Such a domain is involved in muscarinic antagonist (but not agonist) binding and has been proposed to explain the anesthetic effects on antagonist binding described earlier. [138]Another potential site of action would be the interface between receptor and G protein. Yet another alternative would be anesthetic-induced changes in membrane lipid microdomains around the receptor or G protein molecule, leading to changes in protein conformation and resulting altered function.

Whether CNS muscarinic inhibition plays a significant role in clinical anesthesia is still an unanswered question. Inhibition of muscarinic signaling by reducing acetylcholine levels or inhibiting its release certainly lowers minimum alveolar concentration of inhaled anesthetics (although intraventricular atropine does not), and physostigmine administration increases minimum alveolar concentration, [145]but such actions do not, of course, prove that muscarinic signaling has a role in the effects induced by anesthetics. Halothane administration decreases acetylcholine release in the medial pontine reticular formation. [146]This suggests disruption of cholinergic signaling by the anesthetic. In contrast, acetylcholine release in the interpeduncular nucleus is enhanced by halothane. [147]Further work is needed to define these interactions completely. Muscarinic inhibition seems, however, to be a property of many anesthetics, because ketamine also depresses signaling through m1 receptors, presumably through a direct antagonist action. [148]In contrast, propofol was shown recently to stimulate M2²receptors in heart: in an isolated preparation atropine reversed propofol-induced slowing of cardiac conduction, and the anesthetic competed with the muscarinic antagonist [sup 3 Hydrogen]-quinuclidinyl benzilate. [149]Thus, anesthetic actions on muscarinic systems may be more complex than currently considered.

Both inhibition and stimulation of CNS cholinergic pathways can be employed by the anesthesiologist. Inhibition of cerebral systems, as by scopolamine, leads to sedation and amnesia, and has antiemetic effects. Stimulation of these systems, as by physostigmine, leads to generalized arousal, which might be useful postoperatively. In addition, stimulation of spinal muscarinic pathways may soon provide us with an additional tool for pain control, and cerebroprotective effects of cholinergic stimulation are being investigated.

Three muscarinic antagonists are used routinely by anesthesiologists: atropine, glycopyrrolate, and scopolamine. Of these, scopolamine has the most pronounced CNS effects, and is used primarily for its CNS actions. Atropine can, in high doses, similarly lead to CNS effects, whereas glycopyrrolate does not cross the blood-brain barrier. Although the CNS actions of atropine were originally thought to be excitatory, it is now evident that they are similar to those of scopolamine. This is not surprising, because the compounds are structurally almost identical, and both nonselectively inhibit multiple muscarinic receptor subtypes.

In addition to its sedative and memory effects, scopolamine is a potent antiemetic, presumably through an effect on vestibular pathways. [96]It is most useful when applied transdermally, because after oral or parenteral administration it has too short a duration of action to be useful. Transdermally obtained blood concentrations are very low, but remain constant for up to 72 h and are effective in the prevention of both nausea and motion sickness. When transdermal scopolamine, droperidol, and placebo were compared in 96 patients undergoing superficial surgery, both scopolamine and droperidol were effective in reducing nausea. [150]However, the incidence of vomiting was unchanged. In addition, scopolamine induced more sedation than did droperidol.

An advance to be expected in the coming years is the development of more selective antagonists. Compounds have been developed that act as agonists on one receptor subtype but as antagonists on others, indicating the feasibility of obtaining subtype-selective activity. For example, L-689,660 is a potent partial agonist at M1 and M3 receptors, but a competitive antagonist at M2 receptors. [151]Another novel compound has agonist activity at M1 sites, but acts as antagonist at both M2 and M3 receptors. [152]Whereas classical antagonists like pirenzepine discriminate at most 50-fold among subtypes, novel compounds have selectivities greater than 100-fold. [153]Although we still lack a full understanding of the receptor subtypes involved in the various functions of the muscarinic system, the distinct tissue distribution of the subtypes suggests a different role for each. This might allow separation of the antiemetic from the sedative effects, for example, or the sometimes bothersome peripheral actions, such as antisialagogue effects, from sedation.

Exogenously administered acetylcholine has a very short duration of action because it is inactivated rapidly by butyrylcholinesterase (in serum) and acetylcholinesterase (at the synapse). Compounds such as acetyl-beta-methylcholine that are not a substrate for cholinesterase have longer half-lives. Nonetheless, the most efficient method for prolonged stimulation of the CNS muscarinic system is by the administration of long-acting anticholinesterases. This has the additional advantage of generating high concentrations of acetylcholine only at the synapse, where the agonist serves a useful function. Physostigmine is the classic centrally acting cholinesterase inhibitor. It has been used primarily to treat atropine or scopolamine toxicity, [154]but it is evident that depressant effects of a variety of drugs, including antihistamines, tricyclic antidepressants, benzodiazepines, and fentanyl-droperidol combinations, can be antagonized by the compound as well. [155–157]Recognizing that CNS muscarinic stimulation induces a relatively diffuse arousal, this is not surprising. A review on cholinesterases is outside the scope of this article, and the interested reader is referred to a recent review. [158]Nonetheless, it should be understood that cholinesterases come in a large variety of forms, which potentially could be inhibited selectively. Physostigmine is relatively nonselective, which makes it a less than optimal compound for targeted pharmacologic applications. More specific actions could be expected from inhibitors acting only on certain spatially localized cholinesterases. Cholinesterase inhibitor development has received a major impetus from Alzheimer's research. New compounds show increased blood-brain barrier permeability and prolonged duration of action, and clinical trials should soon provide an understanding of their clinical effects.

Most exciting have been recent studies indicating the feasibility of using spinal cholinesterase inhibitors as antinociceptive drugs. Rats with chronically implanted intrathecal catheters have been shown to respond with increased tail flick latencies and decreased writhing response to several cholinomimetics. [126,129,159]Paw withdrawal in response to heat in the rat [160]and in response to pressure in sheep** was similarly delayed after intrathecal cholinergics. A baseline level of muscarinic activity in the spinal cord appears to exist, as spinal atropine or scopolamine lower the pain threshold. [161]Although interactions with substance P [159]and noradrenergic systems [129]have been proposed, a recent thorough study [160]employing isobolographic analysis [162]showed that the analgesia after cholinesterase inhibition is mediated through muscarinic, but not through nicotinic, opioid or alpha²-adrenergic receptors. Importantly, there is synergistic interaction with the antinociceptive effects of intrathecal micro and alpha²agonists. [160,163]The most noticeable side effect in animals is an increase in blood pressure and heart rate through preganglionic sympathetic stimulation. This suggests that combining clonidine and an anticholinesterase such as neostigmine might be useful, and indeed it has been shown that coadministration of these drugs not only has synergistic antinociceptive effects [160]but minimizes hemodynamic changes. [164]Recently, the first report of intrathecal neostigmine use in humans was published. [165,166]This Phase I study administered 50–750 micro gram neostigmine to healthy volunteers, either through a spinal catheter or a small-gauge needle. The compound was effective in reducing visual analog pain scores to foot immersion in ice water in a dose-dependent manner. However, the side effects were significant: nausea and vomiting were the most common, but leg weakness, decreased deep tendon reflexes, sedation and, at high concentrations, anxiety and increased blood pressure and heart rate were observed as well. Nonetheless, the absence of dangerous side effects warrants further examination of the compound as a clinical analgesic.

Novel potential uses of cholinesterase inhibitors are being investigated. For example, these compounds appear to have a cerebroprotective effect during hypoxia [167]and ischemia, [168,169]although not all studies have found this. [170].

With scopolamine and physostigmine, anesthesiologists have access to two prototypical drugs for modifying CNS muscarinic functioning. However, much greater specificity is needed before this class of drugs can find a broader use. If, as discussed earlier, muscarinic inhibition is indeed a part of general anesthesia, postoperative antagonism of muscarinic blockade by a selective centrally acting cholinesterase inhibitor might be appropriate. Similarly, a muscarinic antagonist with selective sedating, amnestic, or antiemetic effects would be useful. It appears that with the rapid advances in molecular biology described earlier such specificity may be attained.

During the last decade, major advances have been made in our understanding of the physiology and pharmacology of CNS muscarinic signaling. It is time to emphasize that the well-known peripheral parasympathetic and cardiovascular actions represent only one component of muscarinic signaling. Interestingly, many new findings have the potential to influence the practice of anesthesiology. Inhibition of muscarinic signaling may explain some of the anesthetic state, and subtype-selective drugs may allow wider perioperative manipulation of CNS muscarinic systems. The next years will doubtlessly see progress in this area, and our specialty may well reap the benefits.

Acetylcholinesterase = the type of cholinesterase present in the synaptic cleft.

Aminoterminus = the end of a polypeptide chain carrying a free amino group.

Butyrylcholinesterase = the official name of the enzyme usually called plasmacholinesterase or pseudocholinesterase.

Carboxyterminus = the end of a polypeptide chain carrying a free carboxy group.

Chimeras = proteins recombinantly formed from segments of two other molecules. Chimeras often retain the properties of the parent molecules, and can be used to localize functional domains within a protein.

GTP shift = shift in ligand binding affinity induced in a receptor by addition to the binding assay of a nonhydrolyzable GTP analog such as GTP gamma S. GTP gamma S irreversibly binds G proteins, preventing their reassociation with the receptor, thereby keeping the receptor in a low-affinity state. Presence of a GTP shift is presumptive evidence that the receptor under study signals through a G protein.

Homology = a measure of the similarity between different DNA sequences, which takes into account the number and location of both identical amino acids and conservative substitutions, where a change in amino acid has resulted from a single nucleotide mutation.

Immunohistochemistry = method where agonists or antagonists labeled with immunoreactive groups are allowed to bind to their receptors in a thin slice of tissue. The tissue, mounted on a microscope slide, is then exposed to fluorescently labeled antibodies against the immunoreactive group, and the location of the receptor can be determined at the cellular level by fluorescence microscopy.

In situ hybridization = method whereby a radioactively labeled segment of RNA is allowed to bind to its messenger RNA counterpart in a thin slice of tissue. The tissue, mounted on a microscope slide, is subsequently covered with photographic emulsion, and silver grains will develop over the sites where the RNA bound. This indicates the location of the messenger RNA of interest at the cellular or subcellular level. Alternative techniques use fluorescent labels.

Ionotropic receptor = membrane receptor where ligand binding leads to opening of an ion channel present within the receptor molecule. Examples are the nicotinic acetylcholine receptors and the GABA^Areceptors.

Metabotropic receptor = membrane receptor where ligand binding leads to changes in the concentration of an intracellular second messenger such as Calcium²+ or cAMP. Examples are the muscarinic acetylcholine receptors and opiate receptors.

Nerve growth factor = a 13 kd, 3-subunit protein with pronounced ability to induce neurite outgrowth in some neurons.

Orphan clones = cloned genes without an assigned function. Often a presumptive function can be deduced from the amino acid sequence. For example, presence of seven hydrophobic domains suggest that the clone encodes a G-protein-coupled receptor.

Site-directed mutagenesis = method whereby selected amino acids within a protein molecule can be changed to other amino acids, thus allowing assessment of their role in the functional molecule.

Transfection = method whereby a foreign gene is introduced into a cell, where it is subsequently expressed as protein. Such cells can then be used for functional or binding studies.

*Okamura A, Ohkubo K, Yamamura T, Kemmotsu O, Tohda M, Nomura Y: Inhalational anesthetics depress second messenger-mediated response of mRNA expressed Xenopus oocyte [sic](abstract). ANESTHESIOLOGY 1993; 79:A735.

**Detweiler DJ, Eisenach JC: Intrathecal neostigmine potentiates the analgesic effects of intrathecal clonidine (abstract). ANESTHESIOLOGY 1992; 77:A840.