Molecular Biology, Pharmacology, and Brain Distribution of Subtypes of the Muscarinic Receptor

Frederick J. Ehlert, William R. Roeske, and Henry I. Yamamura

It has been 80 years since Dale (13) divided the actions of acetylcholine into nicotinic and muscarinic. These effects are now known to be mediated by two quite distinct classes of receptors which show little in common except their ability to bind acetylcholine. The muscarinic class of acetylcholine receptors are widely distributed throughout the body and subserve numerous vital functions in both the brain and autonomic nervous system (50). Activation of muscarinic receptors in the periphery causes a decrease in heart rate, a relaxation of blood vessels, a constriction in the airways of the lung, an increase in the secretions and motility of the various organs of the gastrointestinal tract, an increase in the secretions of the lacrimal and sweat glands, and a constriction in the iris sphincter and ciliary muscles of the eye. In the brain, muscarinic receptors participate in many important functions such as learning, memory, and the control of posture.

Until the early 1980s, muscarinic receptors seemed to represent a fairly homogeneous class of receptors, although pharmacological evidence to the contrary had actually existed since the early 1950s. Unequivocal evidence for muscarinic receptor heterogeneity came in the late 1980s when five different subtypes of the muscarinic receptor were identified using molecular biological techniques (6, 42, 43, 54, 60, 62). This result was somewhat of a surprise because, up to that time, only three subtypes of the muscarinic receptor could be reasonably identified using selective muscarinic antagonists.

The aims of this chapter are to review the pharmacology and molecular biology of the muscarinic class of acetylcholine receptors and to describe their distribution in the brain and the methods that are currently in use to study these receptors. In addition, the implications for the treatment of neuropsychiatric disease will be mentioned.

The first evidence for subtypes of the muscarinic receptor came soon after the introduction of gallamine as an adjunct in general anesthesia when it was noted that this neuromuscular blocking agent caused sinus tachycardia as a side effect (74). The results of experiments on isolated tissues showed that, while gallamine opposed the negative inotropic and chronotropic effects of acetylcholine on the heart, it was much less effective at antagonizing the actions of muscarinic agonists on intestinal smooth muscle (12, 64). Other neuromuscular blocking agents were identified which shared the cardioselective antimuscarinic effects of gallamine (64). Thus, muscarinic receptors in the heart appeared to be different from those located at other sites in smooth muscle and exocrine glands. In an elegant series of experiments, Clark and Mitchelson (12) and Stockton et al. (67) demonstrated that the blocking action of gallamine on the heart differed from competitive inhibition, but could be rationalized by an allosteric mechanism. Thus, the pharmacological specificity of muscarinic receptors is determined not only by the primary recognition site where acetylcholine binds but also by the secondary allosteric site where gallamine binds.

Evidence for additional complexity in muscarinic receptor heterogeneity came in the early 1960s when Rowskowski (65) described the properties of the novel compound, McN-A-343, which had properties of a muscarinic ganglionic stimulant. When injected into cats, McN-A-343 caused an increase in blood pressure which was antagonized by atropine but not by hexamethonium. This effect was shown to be caused by an action on sympathetic ganglia. Remarkably, McN-A-343 was almost completely inactive at other muscarinic sites, including heart and smooth muscle. Thus, as early as 1961, pharmacological evidence for three subtypes of the muscarinic receptor existed: (i) muscarinic receptors in sympathetic ganglia triggering a rise in blood pressure, (ii) muscarinic receptors on pacemaker cells mediating a slowing in heart rate, and (iii) muscarinic receptors eliciting contraction in smooth muscle. For the most part, the subsequent development of selective pharmacological agents has served to reinforce the designation of these three subtypes of the muscarinic receptor, although additional subtypes have been identified through gene cloning (see below).

Perhaps the first time that strong pharmacological evidence for the existence of subtypes of the muscarinic receptor appeared was in the early 1980s when the novel binding properties of the selective muscarinic antagonist pirenzepine were first described. This compound had been in use in Europe as an antiulcer drug. Unlike other muscarinic antagonists, it blocked gastric acid and pepsinogen secretion at doses which had little or no influence on intestinal motility, salivary secretions, and heart rate (56). It was shown subsequently that pirenzepine bound with high affinity to a subclass of the muscarinic receptor that was abundant in the cerebrum and in peripheral ganglia, whereas it displayed intermediate binding affinity for receptors in exocrine glands and low affinity for receptors in the heart (34, 35). Pirenzepine was shown to block the pressor response to McN-A-343 with high potency, whereas it only weakly antagonized the bradycardia caused by vagal stimulation (35). In contrast, more conventional muscarinic antagonists, such as N-methylscopolamine (NMS), are approximately equipotent at blocking these two effects. A variety of subsequent pharmacological experiments have demonstrated that pirenzepine can be used to divide muscarinic receptors into the same three classes mentioned above, namely, a major subtype in brain and peripheral ganglia (high affinity for pirenzepine), a cardiac subtype (low affinity for pirenzepine), and a subtype mediating responses in exocrine glands and smooth muscle (intermediate affinity for pirenzepine). These three subtypes of the muscarinic receptor are known as M1, M2, and M3, respectively. A complete definition of muscarinic receptor subtypes is given below (also see ).

In addition to the agents mentioned above, several other muscarinic antagonists have been useful in the classification of muscarinic receptor subtypes. Some compounds, such as 4-diphenylacetoxy-N-methylpiperidine methiodide (4-DAMP) and hexahydrosiladifenidol (HHSiD), were shown to antagonize M3-mediated contractions of intestinal smooth muscle more potently than M2mediated cardiac responses, which led to their designation as M3-selective antagonists (2, 47). However, these compounds actually exhibit high affinity for all subtypes of the muscarinic receptor except the M2, indicating that they are more appropriately defined as non-M2 antagonists. Another group of compounds, including gallamine and AF-DX 116 (11-[2-[(diethylamino)methyl]-1-piperidinyl] acetyl-5, 11-dihydro-6H-pyrido-[2, 3b] [1, 4] benzodiazepine-6-one ) , exhibit the converse selectivity -- that is, high affinity for M2 muscarinic receptors and low affinity for the other subtypes (59). It is important to note that, whereas gallamine has high affinity for the secondary allosteric site on M2 muscarinic receptors, AF-DX 116 exhibits high affinity for the primary recognition site where acetylcholine, atropine, and other competitive agents bind.

In general, none of the currently available muscarinic antagonists is highly selective. Typically, most of the compounds only exhibit about a 10-fold difference in affinity between the subtype for which they are selective and the other subtypes (see and ). Consequently, it is necessary to employ a group of antagonists to characterize a particular pharmacological response before it can be concluded what receptor subtype is involved. Unfortunately, there are no agonists available which have strong subtype-selectivity. The agonist, McN-A-343, is moderately M1-selective (37) and is very useful as a tool to activate M1 receptors in a few pharmacological assays; however, its selectivity is insufficient for widespread application as a standard M1 agonist. The problem in the use of agonists is that a high degree of selectivity is required to overcome the variation in receptor reserve (i.e., the differences in receptor density and the efficiency of coupling of a receptor to a particular response) that exists in different tissues. This variation in signal amplification can lead to large variations in agonist potency which are unrelated to differences in the ability of the agonist to discriminate among receptor subtypes.

In 1986, Numa and his colleagues (42, 43) cloned the m1 and m2 subtypes of the muscarinic receptor by screening cDNA libraries prepared from porcine cerebrum and heart, respectively. These libraries were screened with degenerate oligonucleotide probes that were synthesized to correspond with sequence information obtained from partial tryptic digests of the muscarinic receptor purified from porcine cerebrum and heart using the affinity chromatography procedure developed by Haga and Haga (33). In the next year, three more receptor subtypes (m3–m5) were cloned by screening both cDNA and genomic libraries under low-stringency conditions using oligonucleotide probes corresponding to regions of high homology between the m1 and m2 sequences (6, 54, 62). On the basis of the pharmacological properties of the expressed recombinant receptors and the distribution of their mRNA (55), it appears that the m1, m2, and m3 cloned subtypes correspond to the M1, M2, and M3 subtypes identified using pharmacological procedures (see below). The current convention is to use an uppercase "M" or a lowercase "m" to designate subtypes depending upon whether pharmacological (M) or molecular biological (m) criteria are used to identify the subtype. However, because there is correspondence between the two classification schemes with regard to the first three subtypes, we will use a single "uppercase" scheme throughout the rest of this chapter to avoid confusion.

The relationship between the pharmacological and molecular biological classification schemes for subtypes of the muscarinic receptor is summarized in . The M1 subtype exhibits high affinity for pirenzepine and is abundant in forebrain and sympathetic ganglia (21, 32, 35, 41, 82). The M2 subtype has high affinity for AF-DX 116 and gallamine and is expressed in the mammalian myocardium, where it accounts for essentially the total muscarinic receptor population (41, 70). The M2 receptor is also expressed at a relatively low, uniform density throughout the brain (21, 82), and, surprisingly, it represents the major muscarinic receptor in smooth muscle (11). It does not participate directly in the contraction of smooth muscle, but it does modulate contraction by preventing the effects of smooth muscle relaxants, such as isoproterenol and forskolin (69). The M3 muscarinic receptor has high affinity for the non-M2 antagonists, HHSiD and 4-DAMP, and represents the major muscarinic receptor in exocrine glands (41, 59). It also triggers direct contractions of smooth muscle; however, it only represents a minor fraction of total muscarinic receptor population in most smooth muscles (11). Finally, it is expressed in relatively low density throughout the brain (82). The M4 muscarinic receptor also has high affinity for 4-DAMP and HHSiD in addition to himbacine (41). It represents the major muscarinic receptor in the peripheral lung of rabbits (48), but not humans (5) and rats (27). It is also expressed abundantly in various regions of the forebrain, particularly the corpus striatum and olfactory tubercle (82). The M5 muscarinic receptor also has high affinity for the non-M2 antagonists 4-DAMP and HHSiD and is typified by its low affinity for AF-DX 116 (41). It does not appear to be expressed to any significant extent in peripheral tissues, and it represents less than 2% of the total density of muscarinic receptors in various regions of the brain (82).

Analysis of the primary sequences of the muscarinic subtypes shows that these receptors are members of a superfamily of genes including the opsins and numerous receptors which signal through G proteins (6, 42). A more complete description of G proteins is given in Cholinergic Transduction, Signal Transduction Pathways for Catecholamine Receptors, and Serotonin Receptors: Signal Transduction Pathways, in the section entitled "Signal Transduction and GTP-Binding Proteins." This family of receptors is typified by the presence of seven hydrophobic regions in their sequence which are thought to form alpha helixes which span the membrane (see ). The transmembrane (TM) segments of the muscarinic receptor represent the regions of highest homology among the different subtypes and across other members of this large family of G-protein-linked receptors. If the sequences of the five subtypes of the muscarinic receptor are aligned to achieve maximum identity, it can be seen that differences in the lengths of the sequences arise from differences in the extracellular amino terminis, the cytoplasmic carboxy terminis, and the third intracellular (i3) loop. The remaining portions of the protein—namely, the seven TM segments, the three extracellular loops, and the first two cytoplasmic loops—are all the same length. There is 63% identity among the amino acids of seven TM segments of the human M1–M5 subtypes, and most of the remaining residues in these segments are conservative replacements. The greatest divergence arises from the third cytoplasmic (i3) loop. This loop varies in length from 156 (M1) to 239 (M3) residues in the five human sequences, and it accounts for 34–45% of the total number of amino acids. A comparison of the sequences shows that the M1, M3, and M5 subtypes show maximum homology with each other, whereas the M2 and M4 subtypes constitute a separate homologous group.

The overall structure of the muscarinic receptor and other G-protein-linked receptors has not been determined, but it seems likely that it might be analogous to that of bacteriorhodopsin, the only member of this family whose structure has been determined (25). Accordingly, the seven TM segments are thought to form the staves of a barrel-like structure having a central pore (see ). Using a "helical wheel" model of the muscarinic receptor, Hulme et al. (38) have predicted that most of the conserved residues in the TM segments form the inner lining of the central pore, whereas the few nonconserved residues are on the outside. Acetylcholine and other muscarinic ligands are thought to bind at a site within this pore, and Hulme et al. (38) have pointed to the highly conserved nature of the central pore as the explanation for the present lack of highly selective muscarinic agonist and antagonists. As described below, large molecules [i.e., antibodies and toxins (63)] showing a high degree of discrimination among the subtypes have been developed; however, these agents interact with more nonconserved regions of the receptor.

It has long been known that muscarinic agonists require a positive charge to be active (10, 36). This cationic requirement is also shared by many muscarinic antagonists, and it suggests the existence of a complementary anionic site on the muscarinic receptor (4, 18). On the basis of the pH dependence of the binding of the muscarinic agonist [3H]cis-dioxolane to cerebral muscarinic receptors, it was speculated that a carboxylic acid group forms the anionic site on the muscarinic receptor (20). There is now convincing evidence that the carboxyl group from a highly conserved aspartic acid residue in the third TM segment of the muscarinic receptor (aspartic acid 105, human M1 sequence) provides the negative charge for ligand binding. This evidence includes the results of peptide mapping and sequencing studies indicating that [3H]propylbenzilylcholine mustard ([3H]PrBCM) covalently attaches itself to cognate aspartic acids 105 and 111 in the M1 and M4 sequences, respectively (39). The site of covalent attachment of [3H]PrBCM to the muscarinic receptor is likely to be the anionic site involved in ligand binding because the chemically reactive aziridinium group of [3H]PrBCM is structurally similar to the quaternary ammonium head group of many muscarinic ligands including benzilylcholine itself. Moreover, mutation of aspartate 105 to an asparaginine in the M1 sequence abolished ligand binding and the phosphoinositide response, whereas point mutations in the other conserved aspartic acids at positions 71, 99, and 122 in the M1 sequence had little or no effect on ligand binding (26).

Site-directed mutagenesis studies have also shown that conserved threonine and tyrosine residues are essential for agonist, but not antagonist, affinity. Point mutations in threonine 234 and tyrosine 506 in the M3 sequence caused 40- and 60-fold reductions, respectively, in the binding affinity of acetylcholine for M3 muscarinic receptors without affecting the binding of the antagonist, [3H]N-methylscopolamine ([3H]NMS) (78, 79). Interestingly, although these residues are on TM segments V (threonine 234) and VI (tyrosine 506), the molecular modeling studies of Brann et al. (7) suggest that both residues are adjacent to the conserved aspartic acid 105 (M1) in TM segment III. Collectively, these results provide an internally consistent picture of the residues involved with the binding of acetylcholine.

It appears that the structural properties of the individual subtypes of the muscarinic receptor determine, in part, their signaling mechanism (see Cholinergic Transduction for a more complete description of muscarinic receptor signaling mechanisms). Presumably, this specificity arises as a result of the selective coupling of the receptor subtypes to G proteins. The results of studies in which the individual receptor genes were transfected into cells previously lacking muscarinic receptors have demonstrated that the M1, M3, and M5 subtypes stimulate phosphoinositide hydrolysis, whereas activation of the M2 and M4 subtypes causes a pertussis-toxin-sensitive inhibition of adenylate cyclase (1, 44, 46, 53, 61). In most instances, the muscarinic phosphoinositide response is insensitive to pertussis toxin. However, in a few cells a pertussis-toxin-sensitive phosphoinositide response has been observed for M1 receptors (57). These observations imply that at least two types of G proteins are involved in the muscarinic phosphoinositide response. It can be seen from the foregoing observations that, at least in general terms, muscarinic receptors can be divided into two categories depending upon whether they inhibit adenylate cyclase activity (M2 and M4) or cause a robust stimulation of phosphoinositide hydrolysis (M1, M3, and M5). Interestingly, muscarinic receptor subtypes can be divided into the same two groups on the basis of sequence homology (see above).

The pattern of selective receptor coupling seen in transfected cells is also apparent in various tissues expressing a mixture of receptor subtypes. For example, the phosphoinositide response in the rat cerebral cortex is potently antagonized by the M1-selective antagonist pirenzepine (31). In smooth muscle, the phosphoinositide response is potently antagonized by HHSiD, but not by pirenzepine, which indicates the involvement of M3 receptors (11). In contrast, inhibition of adenylate cyclase in mammalian heart and intestinal smooth muscle is potently antagonized by the M2-selective antagonist AF-DX 116 (11). Interestingly, muscarinic agonists inhibit adenylate cyclase in the corpus striatum, yet the pharmacological profile of this response does not agree with the M1, M2, or M3 subclasses (19). Consequently, it is likely that the M4 receptor mediates this response in the corpus striatum, a conclusion that is supported by the great abundance of M4 receptors in this region (see below).

The division of the receptor subtypes into two categories based on coupling mechanisms seems inherently accurate even though several empirical observations would appear unadaptable to this scheme. For example, M2 and M4 muscarinic receptors have been shown to stimulate phosphoinositide hydrolysis; however, the magnitude of the response is weak, and it only occurs in cells expressing high densities of these receptors (61). Moreover, it is sensitive to pertussis toxin (1, 46), unlike the phosphoinositide response to M1, M3, and M5 receptors which is usually, but not always, pertussis-toxin-insensitive. It has also been noted that M1, M3, and M5 receptors stimulate cyclic AMP accumulation in intact cells (45, 61). However, this response may be downstream from the phosphoinositide response, resulting from calcium or protein kinase C activation of adenylate cyclase. The discovery that the bg subunits of heterotrimeric G proteins activate the type II and IV adenylate cyclases (28, 68) provides another mechanism for muscarinic enhancement of adenylate cyclase activity that could be demonstrable in a broken cell preparation. This mechanism is dependent upon simultaneous activation by the a subunit of GS (stimulatory guanine-nucleotide-binding protein) and may represent the mechanism by which M4 muscarinic receptors stimulate adenylate cyclase activity in homogenates of the olfactory tubercle (58). In addition to the second messenger pathways mentioned above, muscarinic receptors can affect inwardly rectifying potassium channels by direct G-protein coupling [see Jones (40)].

An interesting difference between the two signaling groups of muscarinic receptors mentioned above is that M2- and M4-mediated inhibition of adenylate cyclase activity is practically always much more sensitive than M1, M3, and M5-stimulated phosphoinositide hydrolysis (3, 11). That is, the potency of agonists (such as carbachol and oxotremorine) for inhibiting adenylate cyclase activity is usually much greater than their respective potencies for stimulating phosphoinositide hydrolysis. This difference in sensitivity cannot be rationalized by assuming that these agonists are M2-selective, because the binding affinities of carbachol and oxotremorine for M2 and M3 receptors are practically the same when measured under physiological conditions [i.e., in the presence of GTP and physiological concentrations of salt; see Ehlert (16)]. Also, these agonists are not M2-selective on the basis of efficacy in the sense that each agonist has the same relative efficacy at M2- and M3-mediated responses, although the intrinsic efficacy of carbachol is greater than that of oxotremorine (16). There is evidence that the difference in sensitivity is related to G-protein activation. Lazareno et al. (49) have noted that nonselective muscarinic agonists are at least 10-fold more potent at stimulating [35S]GTPgS binding and GTPase activity at M2 and M4 muscarinic receptors as compared to M1 and M3 receptors. Also, several investigators have noted that guanine nucleotides cause a greater reduction in agonist affinity at M2 and M4 receptors as compared to M1 and M3 receptors [e.g., see Baumgold and Drobnick (3)]. Thus, the M2 and M4 receptors can cause a greater activation of G proteins as compared to the M1 and M3 receptors, which might explain the difference in sensitivity between the two second messenger responses. This difference could be caused by either (a) structural differences in the receptors and their complementary G proteins or (b) differences in the size of the G-protein pool with which the receptors interact. For example, a greater expression of Gi relative to Gq might cause a more sensitive M2-mediated inhibition of adenylate cyclase as compared to M3-stimulated phosphoinositide hydrolysis.

Although the mechanism for this coupling difference is unclear, it is tempting to speculate that it maintains the sensitivity of the M2 and M4 signaling pathways in the body. Many receptor signaling pathways have a series or cascade of events between the initial receptor activation and the final cellular response. This cascade provides many opportunities for amplification in the signaling process so that a relatively low level of receptor activation can elicit a near maximal response. This situation appears to reflect M3-mediated contractions of smooth muscle, where a low level of receptor occupancy by an efficacious agonist can lead to a maximum contractile response [see Candell et al. (11)]. In other words, the final response can be highly amplified without having a very sensitive initial transduction mechanism (i.e., phosphoinositide hydrolysis). In contrast, M2 and M4 signaling pathways do not typically exhibit many steps in the signaling cascade. The M2-mediated increase in potassium conductance in pacemaker cells of the heart is the result of direct G-protein coupling to potassium channels. In other words, there is only one step in the signaling pathway between G-protein activation and an increase in potassium conductance. If this hyperpolarizing response is to be as sensitive as M3-mediated contractions of smooth muscle, then the M2 activation of G proteins must be highly amplified. It is also possible that M2 and M4 activation of Gi must be highly amplified in order to achieve a sensitive and effective physiological antagonism of responses elicited by an increase in cyclic AMP by Gs-linked receptors.

A variety of evidence indicates that the i3 loop of the muscarinic receptor is important for G-protein coupling. Perhaps the most spectacular evidence of this sort comes from a study on a chimeric receptor in which the entire i3 loop of the D2 dopamine receptor was replaced with the analogous portion from the M1 muscarinic receptor (24). The chimeric receptor bound dopaminergic ligands with affinities similar to those of native D2 dopamine receptors. However, unlike the native receptor, which typically inhibits adenylate cyclase and has little influence on Ca2+, the chimeric receptor increased intracellular Ca2+ levels in response to dopamine. These data indicate that the i3 loop of the M1 muscarinic receptor is involved in coupling the receptor to G proteins which ultimately trigger calcium mobilization, presumably through phosphoinositide hydrolysis. Conversely, the M1 muscarinic receptor can be persuaded to stimulate adenylate cyclase activity via Gs if its i3 loop is replaced with that of the b-adrenergic receptor (80).

Studies on chimeric receptors constructed from different subtypes of the muscarinic receptor provide further evidence that the i3 loop plays a major role in determining whether the receptor couples to stimulation of phosphoinositide hydrolysis or to inhibition of adenylate cyclase. Replacement of the i3 loop of the M2 receptor with the corresponding portion of the M3 receptor resulted in an M2/M3-i3 chimer which, like the native M3 receptor, stimulated phosphoinositide hydrolysis in a pertussistoxin-insensitive manner but did not inhibit adenylate cyclase (77). In contrast, when the i3 loop of the M3 receptor is replaced with the corresponding M2 sequence, the resulting M3/M2-i3 chimer has the coupling properties of a native M2 receptor; that is, it inhibits adenylate cyclase and causes a weak, pertussis-toxin-sensitive stimulation of phosphoinositide hydrolysis (77). Similar results have been obtained with M1/M2 chimeric receptors. Lai et al. (45) have constructed three chimeric M1/M2 receptors having splices between M1 and M2 sequence in TM segments IV and VI and have also found that the i3 loop has a major role in determining functional coupling. In these studies, the coupling behavior of the chimeric receptors could be predicted from the source of the i3 loop. The two chimeric receptors having M2 sequence in the i3 loop had no effect on phosphoinositide hydrolysis but inhibited adenylate cyclase activity, whereas the chimeric receptor having M1 sequence in the i3 loop stimulated phosphoinositide hydrolysis and increased, rather than decreased, cyclic AMP accumulation. The increase in cyclic AMP accumulation caused by this chimeric M2/M1 receptor was actually fourfold greater than that caused by the native M1 receptor, yet the maximum phosphoinositide response of the same chimeric receptor was only 30% that of the M1 response. These observations suggest that the increase in cyclic AMP is triggered by a mechanism distinct from phosphoinositide hydrolysis.

An interesting property of the i3 loop of the muscarinic receptors is that it contains an abundance of charged residues, particularly in the regions adjacent to TM segments V and VI. These residues make up approximately 30% of the total number of residues in the i3 loop of the various subtypes. Moreover, there is an excess (35–127%) of positive charge relative to negative in the i3 loop. This excess of positive charge also applies to the two other cytoplasmic loops and the cytoplasmic carboxy terminis. The positively charged nature of the i3 loop is striking when one considers that some highly negatively charged compounds have been shown to uncouple M2 muscarinic receptors as well as other G-protein-linked receptors. These compounds, including heparin, dextran sulfate, and trypan blue, have been shown to prevent M2-receptor-mediated inhibition of adenylate cyclase activity (29). These compounds also reduce high-affinity agonist binding to M2 muscarinic receptors in the heart without influencing antagonist binding (29). The high-affinity agonist receptor complex is thought to represent the receptor–G-protein complex (16). Collectively, these results are consistent with the postulate that heparin, dextran sulfate, and trypan blue uncouple M2 muscarinic receptors from Gi. Perhaps these highly negatively charged compounds bind to the abundant, positively charged residues on the i3 loop and sterically hinder the association of the receptor and G protein.

Defining the pharmacological properties of the subtypes of the muscarinic receptor is an important goal because it provides a means of establishing the function of these receptors in the body and fosters the development of useful drugs for medicine and basic research. At the present time, none of the currently available muscarinic antagonists is selective enough to block a response mediated by one subtype completely without also dampening responses mediated by the other subtypes. Consequently, when assessing the ability of an antagonist to interfere with a muscarinic response, it is essential to apply the principles of competitive antagonism (or, if appropriate, allosterism; see ref. 17) to estimate the dissociation constant (KD) of the antagonist. This functional estimate of KD can then be compared with that measured in competitive radioligand binding experiments on cells transfected with subtypes of the muscarinic receptor. A match between the functional values and the binding affinity profile of a receptor subtype provides strong evidence that the subtype mediates the response. Although the discriminatory power of hybridization and immunoprecipitation techniques greatly exceeds that of competitive muscarinic antagonists, the former methods are used primarily to identify which receptors are present in a tissue. The identification of the major receptor subtype in a given tissue does not constitute proof that this is the subtype which mediates the response of the tissue. For example, the techniques of ligand binding (11), immunoprecipitation (52), and Northern analysis (55) all demonstrate that the major muscarinic receptor in intestinal smooth muscle is the M2. However, the pharmacology of the contractile response is inconsistent with the idea that M2 receptors mediate contraction. Rather, the functional KD values of muscarinic antagonists in the intestine are in close agreement with the binding properties of recombinant m3 muscarinic receptors and a minor population of binding sites in the intestine (11). Thus, using several different techniques it is possible to conclude with certainty that M3 muscarinic receptors mediate contraction of intestinal smooth muscle.

The binding affinities of selective muscarinic antagonists for subtypes of the muscarinic receptor have been estimated in competitive binding assays on cells transfected with the M1 through M5 genes. shows a compilation of results from studies in which the competitive inhibition of [3H]N-methylQNB binding by various antagonists was measured in murine fibroblasts transfected with recombinant rat muscarinic receptors (41, 57). An important feature of these particular studies is that the binding assay was carried out at 37°C in tissue culture media so that the ionic composition of the assay buffer resembles physiological conditions. It can be seen that none of the compounds shows greater than a sixfold difference in affinity between the receptor for which it has highest affinity and the other subtypes. Pirenzepine has highest affinity for M1 receptors (pKi = 7.9), intermediate affinity for M3, M4, and M5 receptors (pKi {ewc MVIMG, MVIMAGE,!similar.bmp} 7.0), and low affinity for m2 receptors (pKi = 6.4). AF-DX 116 has highest affinity for M2 receptors (pKi = 7.0), intermediate affinity for M4 receptors (pKi = 6.5), and low affinity for M1, M3, and M5 (pKi {ewc MVIMG, MVIMAGE,!similar.bmp} 6.0). Both HHSiD and 4-DAMP have high affinities for all the subtypes except the M2. The data shown in are generally consistent with the results of similar studies by Buckley et al. (9) and Dörje et al. (15) which were carried out in 25 mM phosphate buffer containing 5 mM MgCl2. The biggest discrepancy is in the affinity values of AF-DX 116. This difference can probably be attributed to differences in the assay buffer, because Pedder et al. (59) have demonstrated that the affinity of AF-DX 116 is markedly influenced by ionic strength.

The binding affinities of selective muscarinic antagonists have also been measured in tissues of the rat expressing predominantly one receptor subtype. A convenient means of measuring the binding properties of M1 muscarinic receptors is to run competition experiments in the rat cerebral cortex using a low concentration of [3H]pirenzepine to label M1 receptors selectively (76). The binding properties of M2 receptors can be easily measured by running competitive binding experiments on the mammalian heart which expresses M2 receptors exclusively (70). M3 receptors can be assessed by running experiments on various exocrine glands. gives the dissociation constant of various muscarinic antagonist for M1, M2, and M3 muscarinic receptors measured in binding studies on the rat using the strategy described above (14, 59). The binding assays were carried out at 30–32°C, in a buffer having an ionic strength similar to physiological conditions. It can be seen that the results in are in excellent agreement with those shown in , indicating that the murine fibroblast transfection system provides an accurate pharmacological picture of native muscarinic receptor subtypes. None of the values in the two tables differ by more than 1.9-fold (0.28 log units; see pirenzepine, M2, and 4-DAMP, M3). The average of the absolute values of the differences between the estimates in the two tables is only 0.14 log units (i.e., 1.4-fold), and the standard deviation of this estimate is 0.09 log units.

It is informative to compare the estimates of affinity made in binding assays with those obtained from the results of antagonism studies on tissues whose functional responses are thought to be mediated by the M1, M2, and M3 subtypes of the muscarinic receptor. The functional assay system that is commonly used to screen for M1 antagonism is the rabbit vas deferens, where the M1 agonist, McN-A-343, causes an inhibition of the twitch response elicited by transmural stimulation (22, 23, 47). The standard assay for the M2 receptor is the electrically paced or spontaneously beating guinea-pig atria which exhibits a decrease in contractile force or in beating rate, respectively, when stimulated by muscarinic agonists (2, 22, 30, 47). Finally, the M3 system that is commonly used is the isolated guinea-pig ileum which contracts to muscarinic agonists (2, 30, 47). The KD values that have been measured for subtype-selective muscarinic antagonists in the functional assays described above are listed in together with the estimates of KD from binding studies on tissues of the rat. It can be seen that there is general agreement between the two estimates, which validates the use of the biological assay system.

The distribution of subtypes of the muscarinic receptor in the brain has been mapped out using a variety of methods. Radioligand binding techniques provided the first direct demonstration of the distribution of muscarinic receptors in the brain and also provided a reasonable picture of the distribution of the major subtypes. More recently, the development of subtype-selective antibodies for the M1–M5 subtypes has provided an unequivocal demonstration of the regional distribution of the subtypes in the brain. In general, there is striking agreement between the results of the two methods. Finally the results of in situ hybridization studies have demonstrated that the distribution of mRNA for the subtypes generally parallels the distribution of their protein. Convergence of the results of several types of methodologies to the same conclusion is gratifying and provides a strong validation of the results. In the remainder of this section, the distribution of muscarinic receptors in the brain will be described by reviewing the results of studies using the variety of techniques mentioned above.

Early studies on the binding properties of the muscarinic antagonist [3H](-)QNB established the distribution of muscarinic receptors in the brain (81). These studies showed that the density of muscarinic receptors was highest in various regions of the forebrain but declined in more caudal regions of the brain. In general, the distribution of ligand binding paralleled the distribution of other cholinergic markers such as acetylcholinesterase, cholineacetyltransferase, and high-affinity choline uptake. These results provided convincing evidence for the distribution of the total density of muscarinic receptors in the brain because [3H](-)QNB was highly specific for muscarinic receptors, and it labeled all subtypes of the muscarinic receptor uniformly.

By measuring the ability of subtype-selective muscarinic antagonists to inhibit the binding of a nonselective radioligand competitively, it was possible to identify subtypes of the muscarinic receptor in brain and to establish their regional distribution. The results of studies in brain where the specific binding of [3H]NMS was competitively inhibited by the M1-selective antagonist pirenzepine demonstrated that the resulting pirenzepine/[3H]NMS competition curves were inconsistent with a simple one-site model but could be rationalized by a model incorporating two sites with differential affinity for pirenzepine (21, 32). The proportion of high-affinity pirenzepine sites was highest in the forebrain and lowest in more caudal regions of the brain. When similar studies were carried out with the M2-selective antagonist AF-DX 116, analogous results were obtained except that the proportion of high-affinity AF-DX 116 sites showed exactly the converse distribution as that of the high-affinity pirenzepine sites (21, 32). That is, sites with high affinity for AF-DX 116 represented the majority of the muscarinic receptors in the cerebellum and a minor fraction of the receptors in more rostral regions of the brain. These results were also verified by direct binding studies with [3H]pirenzepine (76) and [3H]AF-DX 116 (75). Collectively, these results demonstrated that M1 receptors represented the majority of the muscarinic sites in the forebrain, whereas the M2 receptor was the most abundant site in more caudal regions of the brain.

Although the nature of the pirenzepine/[3H]NMS and AF-DX 116/[3H]NMS competition curves mentioned above were both consistent with two-site models, it was impossible to rationalize the competition data in terms of only two types of binding sites. The inadequacy of the two-site model was particularly apparent in the corpus striatum, where the sum of the densities of the high-affinity binding sites for pirenzepine (M1, 31%) and AF-DX 116 (M2, 7%) only amounted to 38% of the total density of muscarinic receptors in this brain region (19). Clearly there must be a large population of sites lacking high affinity for both AF-DX 116 and pirenzepine. The simplest hypothesis to account for the data is a three-site model incorporating a high-affinity pirenzepine site (M1), a high-affinity AF-DX 116 site (M2), and third site (non-M1, non-M2) lacking high affinity for both AF-DX 116 and pirenzepine. This model predicts that AF-DX 116 has similar, but perhaps not identical, affinities for the M1 and the non-M1, non-M2 site and that pirenzepine has about the same affinities for the M2 site and the non-M1, non-M2 site.

The question then arises, What is the relationship between these three binding sites and the five cloned subtypes (M1–M5) of the muscarinic receptor? It seems clear that the high-affinity pirenzepine and AF-DX 116 sites represent the M1 and M2 subtypes, respectively, because of the reasons described above (see ). It follows that the non-M1, non-M2 site represents the remainder of the subtypes (i.e., M3, M4, and M5). However, because the results of immunoprecipitation studies with M5selective antibodies indicate very little M5 subtype in brain (less than 2% of the total, see below), it seems appropriate to designate the non-M1, non-M2 binding site as the sum of the densities of the M3 and M4 subtypes. The consequences of this hypothesis indicate that, whereas the high-affinity component of the pirenzepine/[3H]NMS competition curve in brain represents the M1 receptor, the low-affinity component is composed of the M2, M3, and M4 receptors. A comparison of the negative log of the KD of the high-affinity pirenzepine site in rat brain (pKH = 7.82; see ref. 21) shows good agreement with that (7.89) determined for recombinant rat M1 receptors (see ). Moreover, the estimate of pKH was essentially the same in various regions of the brain, indicating that the high-affinity component of pirenzepine binding represented a single type of site (21). In contrast, the low-affinity KD (pKL) of pirenzepine varied from a low value of 6.15 in the cerebellum to a high value of 6.65 in the corpus striatum. This variation is consistent with the postulate that the low-affinity component actually represents a mixture of sites whose affinities are similar, but not identical, and that the measured pKL of pirenzepine in a given brain region represents the weighted average of pKD values for the M2, M3, and M4 receptors (6.38, 6.72, 7.07, respectively; see ). Accordingly, the low pKL value of pirenzepine in the cerebellum (6.15) is consistent with the great abundance of M2 receptors in this tissue which have the lowest affinity for pirenzepine (pKD = 6.38; see ). Also, the high pKL value in the corpus striatum (6.65) is consistent with the great abundance of M4 receptors in this tissue which have relatively high affinity (i.e., compared to M2 and M3) for pirenzepine (pKD = 7.07; see ). An analogous type of argument can be made for the behavior of the AF-DX 116/[3H]NMS competition curves in various regions of the brain. Thus, it can be seen that the method of fitting the pirenzepine/[3H]NMS and AF-DX 116/[3H]NMS competition curves to a two-site model is an approximation; nevertheless, the error associated with this technique is low as indicated by the good agreement between this method and the immunoprecipitation method described below.

The results of the studies described above on the regional binding properties of muscarinic receptors in the brain are summarized in . It can be seen in that the proportion of M1 and non-M1, non-M2 sites (M3 + M4) was greatest in various telencephalic structures and decreased proportionately the more caudal the brain region. In contrast, the proportion of M2 sites was greatest in the cerebellum and decreased proportionately the more rostral the brain region. The actual densities of the M1, M2, and non-M1, non-M2 sites (M3 + M4) can be calculated by multiplying their respective proportions by the total density of muscarinic receptors in various regions of the brain. shows the distribution of the M1, M2, and non-M1, non-M2 sites (M3 + M4) in various regions of the brain. It can be seen that in terms of absolute density, the M2 site has a relatively low uniform density throughout the brain, whereas the M1 and non-M1, non-M2 sites (M3 + M4) are most abundant in the forebrain and decrease in more caudal regions of the brain.

By taking advantage of the slow rate at which [3H]NMS dissociates from M3 and M4 receptors relative to the M1 and M2 receptors, Waelbroeck et al. (71) have actually estimated the proportion of the M3 and M4 receptors in some regions of the brain. These investigators have obtained himbacine/[3H]NMS and methoctramine/[3H]NMS competition curves after [3H]NMS has been washed off the M1 and M2 sites. Because himbacine and methoctramine have much higher affinity for M4 receptors relative to M3, it was possible to divide the non-M1, non-M2 class of sites into M3 and M4 receptors. The estimates of the proportion of M3 and M4 receptors made by Waelbroek et al. (71) in the cerebral cortex, hippocampus, and corpus striatum are also given in .

Quantitative autoradiography has been used to map out the distribution of M1, M2, and non-M1, non-M2 receptors in the brain. The distributions of the M1 and M2 sites have been determined by directly labeling these receptors with [3H]pirenzepine and [3H]AF-DX 116, respectively (66). The distribution of the non-M1, non-M2 site was determined by labeling these sites with [3H]NMS in the presence of pirenzepine and AF-DX 116 so that radioligand binding to the M1 and M2 sites was blocked (66). shows the results of a quantitative autoradiographic analysis by Smith et al. (66) in which the strategy described above was used to map out the distribution of the M1, M2, and non-M1, non-M2 subtypes in the brain. It can be seen that the M1 and the non-M1, non-M2 sites represent the most abundant receptor classes in the brain, with high densities occurring in various regions of the forebrain. In contrast, the M2 receptor has a very low, uniform distribution throughout the brain. These results are in general agreement with those shown in and . The highest densities of M1 receptors occur in the hippocampus, various subcortical telencephalic nuclei, and various layers of the cerebral cortex. The highest densities of the non-M1, non-M2 site occur in the islands of Calleja, olfactory tubercle, caudate putamen, and nucleus accumbens. High densities also occur in other subcortical telencephalic structures, the hippocampus, and various layers of the cerebral cortex (see also Colocalization in Dopamine Neurons).

The development of subtype-selective antibodies has provided a powerful means of measuring the distribution of subtypes of the muscarinic receptor in different regions of the brain. Two groups of investigators (51, 82) have raised antisera to fusion proteins containing the i3 loop of the different subtypes of the muscarinic receptor. The high selectivity of these antibodies rests on the divergent nature of the i3 loop of the different subtypes of the muscarinic receptor. Wolfe and co-workers (52, 72, 73, 82) were able to immunoprecipitate all five subtypes of the muscarinic receptor from brain. Moreover, the total amount of receptors immunoprecipitated in various regions of the brain varied from 86% to 99% of the total amount of [3H](-)QNB binding sites. These data indicate the highly quantitative recovery of protein that is achievable using the antibodies. These data also indicate that it is unlikely that there are significant amounts of additional subtypes of the muscarinic receptor in brain because the total amount of receptors that can be immunoprecipitated from CHO cells transfected with the M1–M5 subtypes ranges from 83% to 95%.

The data in show the relative distribution of the M1–M5 subtypes of the muscarinic receptor in various regions of the rat brain. The data have been estimated from Fig. 8 in the publication by Yasuda et al. (82) and have been normalized with respect to the total amount of receptors immunoprecipitated. It can be seen that there is very little M5 muscarinic receptor expressed in the brain, and that it accounts for less than 2% of the total receptor population. The M1 and M4 muscarinic receptors represent the most abundant receptors in various forebrain regions, and their relative proportion declines in more caudal regions of the brain. In contrast, the relative abundance of the M2 receptor is low in various forebrain regions; however, it represents the major receptor in more caudal regions of the brain. When the absolute densities of the subtypes of the muscarinic receptor are considered, it can be seen that the M1 and M4 subtypes are most abundant in the forebrain, and their density declines in more caudal regions of the brain. In contrast, the M2 receptor has a low, relatively uniform distribution throughout the brain. The M3 receptor is not expressed to a great extent in the brain; however, its density is greatest in the forebrain, and it decreases in more caudal regions of the brain. It can be seen by comparison of the binding data in with the immunoprecipitation data in that there is a striking agreement between the two sets of data.

The results of a study (8) using in situ hybridization histochemistry has demonstrated that the relative abundances of mRNA for the M1 and M4 subtypes of the muscarinic receptor are greatest in various forebrain structures such as the cerebral cortex, hippocampus, and caudate putamen and that the message for these receptors declines in more caudal regions of the brain. The abundance of M3 mRNA was greatest in the cerebral cortex and hippocampus, but low in the caudate putamen and in caudal regions of the brain. In contrast, M2 mRNA had a relatively low, uniform distribution throughout the brain. In summary, it can be seen that the distribution of mRNA for subtypes of the muscarinic receptor generally agrees with the distribution of these proteins as determined by radioligand binding and immunoprecipitation (see ).

A variety of different techniques can be used to study the pharmacology, signaling mechanisms, regional distribution, and molecular biology of subtypes of the muscarinic receptor. Molecular biological and immunological techniques have provided the most powerful methods for the identification of subtypes of the muscarinic receptor. Moderately selective muscarinic antagonists have been developed which discriminate among the different subtypes of the muscarinic receptor. By using a panel of antagonists including pirenzepine, AF-DX 116, and HHSiD, it is possible to determine whether a given response is mediated by an M1, M2, or M3 subtype of the muscarinic receptor. If methoctramine or himbacine is used in combination with the other antagonists, it should be possible to discriminate among the M1–M4 subtypes. To date, few agonists have been developed which exhibit high selectivity for the different subtypes of the muscarinic receptor. The great abundance of M1 receptors in the cortex and hippocampus suggests that perhaps M1-selective agonists which penetrate the brain might be efficacious in the treatment of neurological conditions associated with hypocholinergic function in the forebrain (e.g., Alzheimer's disease). Moreover, the existence of a secondary allosteric site on the muscarinic receptor suggests that it might be possible to develop novel allosteric muscarinic agonists that potentiate the effects of endogenous acetylcholine much in the same way that benzodiazepines potentiate GABA. Presumably, allosteric muscarinic agonists would only activate muscarinic receptors that are occupied by acetylcholine. Consequently, they offer an advantage to muscarinic agonists which act at the primary recognition site in that they should preserve the temporal pattern of receptor activation. Moreover, overdose from an allosteric muscarinic agonist should have less disastrous consequences than a directly acting agonist, because the effect of an allosteric agonist would have a ceiling depending upon how much it potentiated endogenous acetylcholine. Although no such allosteric muscarinic agonists have been identified to date, they could be very efficacious in the treatment of Alzheimer's disease. Finally, the great abundance of M4 receptors in the caudate putamen suggests that centrally active M4-selective muscarinic antagonists might be useful in the treatment of parkinsonism. Such agents would lack the peripheral side effects of classic muscarinic antagonists which block M2- and M3-mediated responses throughout the parasympathetic nervous system. Centrally active M4-selective muscarinic antagonists might be useful in parkinsonism by themselves or as adjuncts to levodopa.

Portions of the authors' work cited in this chapter were supported by National Institutes of Health grants NS-26511, NS-30882, HL-20984, and MH-45051 and by the American Heart Association. Frederick J. Ehlert is a recipient of a United States Public Health Service, Research Career Development Award from the National Institute of Neurological Disorders and Stroke.

published 2000