Keywords

1 Schizophrenia Symptoms, Treatment, and Challenges

Schizophrenia is a chronic, debilitating neuropsychiatric illness affecting nearly 1% of the population, often requiring lifelong treatment. Symptoms of schizophrenia are broadly categorized into three domains: positive, negative, and cognitive symptoms. The positive symptoms include hallucinations, delusions, and disorganized behavior; negative (or deficit) symptoms reflect the absence of normal social and motivational functioning, for example, avolition, alogia, anhedonia, and blunted affect; and cognitive symptoms of schizophrenia include impaired attention, executive functioning, and working memory. Early recognition that antagonism of central dopamine receptors reduces the positive symptoms of the disorder, combined with the evidence that stimulation of the dopamine system by drugs such as amphetamine could induce psychosis, established dopamine as the neurotransmitter central to the disorder and had profound effects on drug development strategies in schizophrenia. Indeed, all clinically available antipsychotic drugs antagonize central dopamine receptors. However, while dopamine’s integral role in the positive symptoms of schizophrenia is uncontroversial (see Howes and Kapur 2009), this is not the case with the other symptom domains. Increasing evidence suggests that while antipsychotic drugs can alleviate psychotic symptoms for many individuals with schizophrenia, recovery is often incomplete, leaving patients with residual positive clinical symptoms. For example, some 25% of schizophrenia patients do not respond to dopamine-targeted therapies (Hirsch and Barnes 1995).

In addition, a particularly critical barrier to improving outcomes is the elusiveness of adequate treatments for negative and cognitive symptoms. Even more than the positive clinical symptoms, cognitive deficits contribute to impaired social functioning and poor quality of life (Williams et al. 2008), and predict deficits in occupational functioning (Bellack et al. 1999; Dickinson and Coursey 2002; Gold et al. 2002; Green 1996). Moreover, cognitive deficits are associated with the onset of psychosis in individuals at risk for schizophrenia (Frommann et al. 2010). Finally, in a recent prospective longitudinal analysis, Reichenberg et al. (2010) revealed premorbid neurocognitive deficits in a wide variety of domains in individuals who went on to develop schizophrenia, including executive function, visual and verbal learning and memory, processing speed, attention, visuospatial problem solving, and working memory. These recent studies further underscore the pressing need for novel therapies to improve treatment of schizophrenia.

2 Muscarinic Acetylcholine Receptor System and Schizophrenia

Central cholinergic neurotransmission has long been known to be crucial in CNS functioning. The muscarinic acetylcholine receptor (mAChR) system plays a significant role in memory, learning, arousal, motivation, reward, and attention. Significant evidence links abnormalities in the muscarinic system to the pathophysiology of schizophrenia (for review, see Raedler et al. 2007) and to a number of other debilitating neuropsychiatric illnesses including Alzheimer’s disease (Winkler et al. 1998). Moreover, neurochemical, pharmacological, and neuropathological evidence suggests that selective targeting of mAChRs may hold therapeutic potential for schizophrenia (Friedman 2004; Raedler et al. 2007; Sellin et al. 2008; Conn et al. 2009).

2.1 Overview and Distribution of mAChRs

mAChRs are 7-transmembrane G-protein-coupled receptors (GPCRs). Five muscarinic receptor subtypes have been identified (M1–M5). The oddly numbered receptors, M1, M3, and M5, are excitatory and couple predominantly through the Gq/11 pathway to stimulate phosphoinositide hydrolysis and increase intracellular calcium. M2 and M4 receptor subtypes have inhibitory effects and couple predominantly with Gi/o proteins to inhibit cAMP. The anatomical distribution of these mAChRs places them in key dopamine pathways implicated in psychotic symptoms of schizophrenia, as well as in brain regions relevant to cognitive functioning, especially attention and memory. The M1 and M4 mAchR subtypes are concentrated heavily in the forebrain, including cerebral cortex, striatum, and hippocampus. M4 receptors are also prominently expressed in the midbrain, where their interaction with midbrain dopaminergic mechanisms in VTA and striatum suggests that they influence dopamine release into the nucleus accumbens (Langmead et al. 2008a). The M2 and M3 receptor subtypes are located in the periphery as well as in the central nervous system (CNS) and are involved in parasympathetic functions including bronchoconstriction, salivation, smooth muscle relaxation, vasorelaxation, appetite, bradycardia, akinesia, and tremor. Within the brain, M2 receptors are especially dense in forebrain and hippocampus where they regulate acetylcholine release (Billard et al. 1995; Stoll et al. 2003). Like M1 receptors, M4 receptors are also heavily concentrated in the cortex, striatum, and hippocampus. The M5 receptor subtype is expressed primarily on dopaminergic neurons in the substantia nigra pars compacta, where it modulates dopamine release to the striatum (Weiner et al. 1990). It is also prominently expressed in the ventral tegmental area, which provides dopaminergic input to limbic structures such as the nucleus accumbens (Eglen 2005). However, dissociating the relative contributions of different muscarinic receptor subtypes within the CNS has been challenging, especially in the absence of subtype-specific ligands. Major obstacles include the fact that mAChRs are often co-localized within the same brain regions, often on the same cells (Levey et al. 1991), and sometimes with opposing actions. Of the 5 mAChRs, subtypes M1 and M4 have been the target of the most widespread interest for schizophrenia.

2.2 mAChR Manipulations Influence Symptoms of Schizophrenia

One source of evidence for a role of mAChRs in schizophrenia comes from observations that muscarinic antagonists can induce an “antimuscarinic syndrome” that includes psychotic features (Bolden et al. 1991). The hallucinations induced by muscarinic antagonists are remarkably similar in character to those experienced by individuals with schizophrenia (Yeomans 1995). Atropine, scopolamine, quinuclidinyl benzilate, ditran, and other centrally acting antimuscarinic agents have been known to induce hallucinations in multiple sensory domains, as well as cognitive symptoms that bear marked resemblance to those observed in schizophrenia, including profound disturbances in attention and concentration, impaired memory, and confusion (Abood and Biel 1962; Granacher and Baldessarini 1975; Mego et al. 1988; Gershon and Olariu 1960; Neubauer et al. 1966; Clarke et al. 2004; Fisher 1991; Perry and Perry 1995; Fredrickson et al. 2008). Nonspecific muscarinic antagonists also induce learning and memory deficits in animals (Senda et al. 1997; Rasmussen et al. 2001). Finally, anticholinergic load is also associated with reduced cognitive function in schizophrenia (Minzenberg et al. 2004).

Clinical trials of mAChR agonists provide additional evidence for the putative role of mAChRs. Xanomeline, a relatively selective M1/M4 agonist, improved cognition and reduced psychotic symptoms in both schizophrenia (Shekhar et al. 2008) and Alzheimer’s disease (Bodick et al. 1997). The nonspecific muscarinic agonist arecoline has also shown cognition-enhancing effects in Alzheimer’s disease patients (Christie et al. 1981). Moreover, increasing evidence indicates that some atypical antipsychotics, namely, clozapine and olanzapine (Bymaster et al. 2003a), are partial muscarinic agonists, which has contributed to a new recognition that cholinergic facilitation may contribute to their cognition-enhancing and antipsychotic efficacy. Atypical antipsychotics drugs (i.e., ziprasidone, risperidone, clozapine, and olanzapine), but not conventional antipsychotics, increase acetylcholine release in prefrontal cortex (Ichikawa et al. 2002), which may be one mechanism by which these drugs exert their somewhat modest cognition-enhancing effects. This review will describe evidence suggesting that the muscarinic acetylcholine system is a compelling therapeutic target for schizophrenia, summarize recent progress in understanding the role specific muscarinic receptors may play in schizophrenia, and detail advances in selectively targeting receptors implicated in the pathophysiology of schizophrenia.

2.3 mAChR Abnormalities in Schizophrenia

The anatomy of the CNS cholinergic projections is consistent with a possible role in schizophrenia. The mesopontine cholinergic projection has been most associated with psychotic symptoms. It originates in the pedunculopontine and laterodorsal tegmental nuclei and projects most densely to thalamic nuclei, as well as to the substantia nigra and basal forebrain cholinergic nuclei, lateral hypothalamus, and limbic frontal cortex (Yeomans 1995). The basal forebrain cholinergic system, by virtue of its projection to hippocampal and cortical areas involved in learning and memory, has been strongly associated with cognition.

Notably, postmortem studies have suggested mAChR abnormalities in schizophrenia. Quantitative autoradiography studies measuring the binding of [(3)H]pirenzepine, a muscarinic antagonist that binds selectively to M1 and M4 receptors (Doods et al. 1987; Hulme et al. 1990), have consistently demonstrated reductions in the density of these muscarinic receptor subtypes in a number of brain regions implicated in the pathophysiology of schizophrenia (Dean et al. 1996, 2000, 2002; Crook et al. 2000, 2001; Zavitsanou et al. 2004, 2005; Deng and Huang 2005; Newell et al. 2007; Scarr et al. 2007). For example, [(3)H]pirenzepine binding has revealed low mAChR binding density in prefrontal cortex from subjects with schizophrenia (Brodmann’s areas 8, 9, 10, and 46); importantly, this decreased density is also observed in individuals with schizophrenia who had never been treated with anticholinergic drugs (i.e., benzotropine mesylate; Crook et al. 2001). Gene expression studies have also found decreased M1 (Mancama et al. 2003; Dean et al. 2002) and M4 (Dean et al. 2002) expression in prefrontal cortex in schizophrenia. In the hippocampus, while M4 receptor expression levels were significantly decreased in schizophrenia, M1 receptor levels were comparable to that in controls (Scarr et al. 2007). Reduced [(3)H]pirenzepine binding has also been reported in the anterior cingulate cortex in schizophrenia (Zavitsanou et al. 2004; Newell et al. 2007). A subsequent study by Zavitsanou et al. (2005) using the same cohort of participants tested in their 2004 study showed no differences between schizophrenia patients and other groups on [(3)H]AF-DX384 binding (Zavitsanou et al. 2005), which by inference implicates the M1 receptor in the previously observed reduction in [3H]pirenzepine binding. Schizophrenia patients also have decreased muscarinic receptor binding in the striatum (Dean et al. 1996, 2000) and throughout the hippocampal formation, including the dentate gyrus, areas CA1–CA4, subiculum, and the parahippocampal gyrus (Crook et al. 2000).

Given the almost ubiquitous exposure to antipsychotic drugs in the schizophrenia population, it is possible that alterations in muscarinic receptor density could be an artifact of medication use. However, in situ hybridization studies suggest that antipsychotic exposure is unlikely to underlie these findings. For example, in rats, M1 mRNA expression increased in the substantia nigra, pars compacta, nucleus accumbens, and hippocampus following exposure to both typical and atypical antipsychotic drugs (Han et al. 2008). This finding supported an earlier study in which long-term exposure to antipsychotic drugs in rats either increased or had no effect on the density of [3H]pirenzepine binding (Crook et al. 2001). Taken together, this evidence suggests that M1 receptor alterations may be central to the pathophysiology of schizophrenia. It is also consistent with evidence that atypical antipsychotic drug actions at the M1 receptor may play a critical role in their efficacy in schizophrenia. Similarly, with respect to the M4 receptor, both typical and atypical antipsychotic drugs have either no effect or increase binding of [3H]pirenzepine and [3H]AF-DX384 (Crook et al.1999), suggesting that decreases in M4 receptor density in schizophrenia is unlikely to be attributed to antipsychotic drug exposure.

Findings from a study by Raedler et al. (2003) further supported postmortem findings of reductions in muscarinic receptor density in schizophrenia in an in vivo study of 12 unmedicated patients using [I-123]iodoquinuclidinyl benzilate ([(123)I]IQNB) single photon emission computed tomography (SPECT). In comparison to healthy controls matched for gender and age, the schizophrenia group had significant reductions (ranging from ~20 to ~33%) in muscarinic receptor availability in the cortex, basal ganglia, and thalamus. These studies provide compelling evidence that abnormalities in mAChRs, especially M1 and M4 subtypes, exist in schizophrenia independent of treatment effects from antipsychotic drugs.

3 Partially Selective mAChR Agonists

A number of relatively selective muscarinic agonists were developed in the 1990s, each of which preferentially activated either M1 or M4 (or both) subtypes. It has been suggested that compounds that selectively enhance M1 activity are effective in treating cognitive deficits in schizophrenia, while M4 agonism is effective in treating psychotic symptoms of the disorder (Felder et al. 2001; Bymaster et al. 2003a, b). Evidence in support of this hypothesis comes from studies showing that a number of these partially selective M1 receptor agonists, including xanomeline, sabcomeline, and milameline, have also demonstrated efficacy in preclinical models of cognition (Bodick et al. 1997; Harries et al. 1998; Dean et al. 2003; Weiner et al. 2004). While the M1 receptor is primarily regarded as a target for enhancing cognition, preclinical studies also implicate this receptor in psychosis. For example, M1 knockout mice show disruptions in pre-pulse inhibition and increased locomotor activity (Gerber et al. 2001; Miyakawa et al. 2001). They also exhibit increased sensitivity to amphetamine and striatal dopamine release is increased twofold compared to wild-type mice, suggesting that M1 activation inhibits dopamine release (Gerber et al. 2001). Importantly, M1 deletion does not appear to result in upregulation of other muscarinic receptor subtypes (Miyakawa et al. 2001; Hamilton et al. 1997). In addition, studies showing that mAChR agonists with partial M4 selectivity, such as BuTAC, PTAC, xanomeline, and sabcomeline, show efficacy in animal models of psychosis; specifically, they are able to inhibit dopamine agonist-induced behaviors such as conditioned avoidance responding, D1 and D2 dopamine agonist-induced rotation, and pre-pulse inhibition (Fink-Jensen et al. 1998; Jones et al. 2005; Rasmussen et al. 2001; Shannon et al. 1999; Bymaster et al. 1998). However, the particular contributions of M1 versus M4 to cognition and psychosis are still being elucidated.

Xanomeline has been of particular interest because it is a predominantly M1/M4 receptor partial agonist which has shown cognition-enhancing and antipsychotic-like properties. Xanomeline has been demonstrated to exhibit functional dopamine antagonism in vitro (Stanhope et al. 2001; Shannon et al. 2000). Xanomeline’s particular affinity for M1/M4 receptors has made it of relatively greater interest for schizophrenia due to the suggestions that agonism at the M1 receptor is relevant to cognitive deficits in the disorder, while M4 agonism may reduce psychotic symptoms (Felder et al. 2000; Bymaster et al. 2003a, b). Consistent with this hypothesis, xanomeline decreases dopamine cell firing in the ventral tegmental area (Shannon et al. 2000) and increases extracellular levels of dopamine in the prefrontal cortex (Perry et al. 2001). In primates, xanomeline inhibits unrest and stereotypies induced by dopamine agonists (Andersen et al. 2003), in spite of having no affinity for dopamine receptors (Bymaster et al. 1994; 1997).

Of the partially selective muscarinic agonists, xanomeline is the only one to progress to a clinical trial in schizophrenia patients. A small study of xanomeline’s efficacy in schizophrenia conducted by our group found statistically significant differences between xanomeline and placebo groups in several measures of learning and memory and PANSS total scores, as well as differences between groups in positive and negative symptom subscales and CGI scores in a randomized placebo-controlled, double-blind, 4-week study (Shekhar et al. 2008). Xanomeline demonstrated similar efficacy in an earlier, relatively large (n = 343) multisite clinical trial in patients with Alzheimer’s-type dementia (Bodick et al. 1997). In that study, in addition to significant differences between groups on cognitive performance measures, individuals on xanomeline fared significantly better on behavioral measures including vocal outbursts, suspiciousness, delusions, agitation, and hallucinations; moreover, these improvements were dose dependent.

While xanomeline’s efficacy in improving cognition and reducing psychotic symptoms in schizophrenia (Shekhar et al. 2008) provided an important proof of concept with respect to mAChRs as therapeutic targets in the disorder, its clinical utility could be limited due to adverse side effects elicited by its agonism at other receptor subtypes (Bodick et al. 1997; Bymaster et al. 1998; Sur and Kinney 2005), as is the case with other multiple muscarinic receptor agonists (Schwarz et al. 1999; Wienrich et al. 2001). These adverse side effects are believed to arise due to M2 and M3 receptor activation (Bymaster et al. 2003b, c; Bodick et al. 1997). Indeed, most of the available muscarinic agonists display affinity for most of the five receptor subtypes, with varying levels of selectivity for particular subtypes (Heinrich et al. 2009; Bradley et al. 2010) in spite of early reports suggesting better subtype selectivity. Somewhat conflicting results regarding the selectivity of these agonists are believed to have arisen because they were tested in cell lines where receptor reserve was low; but in native tissue studies, selectivity declined and multiple mAChR subtypes were activated, possibly due to higher receptor reserve and systemic differences in the actions of the various compounds (Conn et al. 2009).

4 M1 and M4 Allosteric Activators

The difficulty in designing compounds with true subtype specificity at mAChR orthosteric sites, i.e., the binding site of acetylcholine, derives from their highly conserved sequence homology across the five subtypes (Wess 1996), which has inhibited drug discovery efforts (Felder et al. 2000). Recently, an alternative approach targeting allosteric receptor sites has gained momentum. Allosteric activators enhance the actions of endogenous acetylcholine but bind at a poorly conserved site (removed from the orthosteric site). This approach has proven successful for GPCRs in other neurotransmitter systems including at metabotropic glutamate receptors (Rodriguez et al. 2005; Hemstapat et al. 2007). In the muscarinic system, allosteric activators with antipsychotic-like profiles have been reported for the M1 receptor (Jones et al. 2008; Ma et al. 2009; Langmead et al. 2008b; Vanover et al. 2008; Bradley et al. 2010; Li et al. 2007, 2008) and the M4 receptor (Shirey et al. 2008; Brady et al. 2008; Chan et al. 2008; Leach et al. 2010), and are now considered highly promising targets for drug discovery efforts (Christopoulos 2002; Conn et al. 2009).

This new drug development strategy focusing on agonists and potentiators for mAChRs, especially at the M1 and M4 receptors, may provide new therapeutic compounds capable of true selectivity with fewer side effects (Conn et al. 2009). Moreover, these allosteric agents could be invaluable in dissociating contributions of different muscarinic receptor subtypes. For example, preclinical models suggest that xanomeline’s clinical efficacy is due to actions at either the M1 or the M4 receptor, or reciprocal interactions between these two receptor subtypes. However, the differential contributions of M1 versus M4 mAChRs to xanomeline’s antipsychotic and pro-cognitive effects have been an enduring question, but the lack of subtype selective agents has impeded progress in understanding their specific roles (Brady et al. 2008). Below, the recent progress in developing more selective allosteric mAChR agonists and new knowledge derived from studies using these compounds are summarized.

4.1 Selective M1 Allosteric Activators

Several M1-selective allosteric agonists and potentiators have been developed recently that have therapeutic relevance for schizophrenia. M1 is abundantly expressed in forebrain, especially striatum, hippocampus, and cortical regions (Levey et al. 1991; Wall et al. 1991; Levey 1993; Vilaro et al. 1993), all of which are implicated in the pathogenesis of schizophrenia. M1 agonism has been specifically suggested as a potential treatment for cognitive impairment in schizophrenia (Friedman 2004), and compounds with varying degrees of selectivity for this receptor have shown efficacy in preclinical animal models of cognition (Bodick et al. 1997; Harries et al. 1998; Cui et al. 2008) and in clinical populations in which cognitive deficits are prominent features of the disorder including Alzheimer’s disease (Bodick et al. 1997) and schizophrenia (Shekhar et al. 2008).

4.1.1 AC-42 and Analogs

AC-42 and its structural analogs 77-LH-28-01 and AC-260584 are potent and selective M1 allosteric agonists as determined by calcium mobilization and inositol phosphate accumulation assays (Spalding et al. 2006; Langmead et al. 2008b; Heinrich et al. 2009; Bradley et al. 2010). These compounds have shown vast improvements in subtype selectivity over orthosteric agonists including xanomeline (Heinrich et al. 2009; Bradley et al. 2010) and have some affinity for D2 and 5HT2b receptors (Vanover et al. 2008; Bradley et al. 2010; Heinrich et al. 2009), a profile that is consistent with atypical antipsychotic drugs and may confer advantages in this regard.

Results from a study by Langmead et al. (2008) suggest that among the AC-42 family, 77-LH-28-01 may be a better candidate for drug development relative to AC-42 due to its in vitro and in vivo M1 receptor selectivity. In electrophysiological studies, 77-LH-28-1 showed a full agonist profile, stimulating hippocampal CA1 cell firing in single unit recordings (pEC50 = 6.3), while AC-42 did not. Carbachol initiated an almost identical response (pEC50 = 5.7) which was reversed by the M1 receptor antagonist pirenzepine, suggesting that this effect was mediated by the M1 receptor. This result also suggests higher potency and efficacy for 77-LH-28-01 relative to AC-42. 77-LH-28-01 also induced gamma oscillatory activity in hippocampus, which studies in knockout mice have demonstrated requires M1 receptors (Fisahn et al. 2002), and disruptions in gamma oscillations have been linked to schizophrenia (Spencer et al. 2003, 2004) and cognition (Kaiser and Lutzenberger 1999.

Studies of in vitro and in vivo properties of AC-260584 have demonstrated that it has a pharmacological profile similar to that of atypical antipsychotic drugs and orthosteric muscarinic agonists in several important respects. For example, it preferentially increased acetylcholine and dopamine in medial prefrontal cortex compared to that in the nucleus accumbens (Li et al. 2007, 2008). Interestingly, N-desmethylclozapine, a metabolite of clozapine, was identified as an M1 allosteric agonist (Sur et al. 2003), which may account for the pro-cognitive effects of clozapine in schizophrenia (Li et al. 2005; Spalding et al. 2006), and shared the ability of AC-260584 to induce acetylcholine release in the mPFC, an effect that was blocked by the M1 antagonist telenzepine in the mPFC, but not in the nucleus accumbens (Li et al. 2005). This mPFC finding is consistent with the actions of partial M1 agonists xanomeline and sabcomeline (Li et al. 2008) and to that of atypical antipsychotics, including clozapine and olanzapine, which increase extracellular dopamine and acetylcholine in the mPFC but not the nucleus accumbens (Kuroki et al. 1999; Ichikawa et al. 2002). Dopamine is believed to modulate critical aspects of prefrontal cortex-mediated working memory function that are compromised in schizophrenia (Braver and Cohen 2000), where dopaminergic hypofunction is believed to contribute to negative and cognitive symptoms of the disorder (Hill et al. 2004; Carter et al. 1998; Perlstein et al. 2001; Riehemann et al 2001; Weinberger et al. 1986; Wolkin et al. 1992; Andreasen et al. 1997). Taken together, these findings add to evidence that atypical antipsychotic drugs and less selective muscarinic agonists could mediate their cognitive effects through M1 receptor-mediated cholinergic and dopaminergic modulation.

Behaviorally, AC-260584 has demonstrated an antipsychotic-like profile and improved cognitive performance (Vanover et al. 2008; Bradley et al. 2010). An antipsychotic-like profile was demonstrated by AC-260584’s ability to reduce amphetamine- and MK-801-induced locomotor hyperactivity as well as reduce apomorphine-induced climbing behavior (Vanover et al. 2008). This finding, along with earlier findings that xanomeline also reduces amphetamine-induced hyperactivity in rodents (Stanhope et al. 2001) and primates (Andersen et al. 2003), suggests that M1 agonism (versus M4) may contribute to its antipsychotic effects more than previously believed. However, the activation of D2 and 5HT2b receptors (Vanover et al. 2008; Bradley et al. 2010; Heinrich et al. 2009) by AC-260584 makes it difficult to assess adequately the contribution of M1 versus M4 receptors to antipsychotic-like effects observed for compounds like xanomeline (Heinrich et al. 2009).

Preclinical studies have shown cognition-enhancing effects of AC-260584 in two animal models of learning and memory. AC-260584 also improved spatial memory on the Morris water maze (Vanover et al. 2008). Rats treated with AC-260584 demonstrated improved performance on the novel object recognition task, which was reversed by pirenzipene, a muscarinic antagonist (Bradley et al. 2010). ERK1/2 phosphorylation, which is associated with important aspects of synaptic plasticity and learning and memory processes (Giovannini 2006), was increased by AC-260584 in hippocampal cells of wild-type but not M1 knockout mice (Bradley et al. 2010). Moreover, it had low catalepsy rates (Vanover et al. 2008), which is predictive of low EPS in humans (Hoffman and Donovan 1995). Bradley et al. (2010) recently concluded that AC-260584 has high bioavailability, potency, and efficacy, and should serve as a lead compound for drug discovery efforts.

4.1.2 TPBP

Jones and colleagues (2008) recently reported that TPBP is a potent muscarinic allosteric agonist that has shown in vitro M1 selectivity. TPBP showed robust agonist activity in M1 transfected cell lines, but not in M2–M5 transfected cells. In hippocampal slices, TPBP increased NMDA receptor currents. This is consistent with findings from other studies indicating that this is an M1-mediated effect. For example, M1 receptors are co-localized with NMDA receptors in hippocampal neurons, and selective M1 antagonists block carbachol-induced potentiation of NMDA current (Marino et al. 1998). NMDA-mediated long-term potentiation is enhanced by the muscarinic agonist carbachol in wild-type and M3 knockout mice, but not in M1 knockout mice (Shinoe et al. 2005). Potentiation of NMDARs is believed to be critical to synaptic plasticity underlying learning and memory (McBain and Mayer 1994), and is consistent with the finding that the AC-260584 induced ERK1/2 phosphorylation in hippocampus (Bradley et al. 2010). Therefore, these effects support a role both for the M1 receptor in cognition and for the efficacy of TPBP in enhancing cognitive deficits in schizophrenia. Importantly, NMDARs have also been implicated in psychosis, and potentiation of NMDAR current may be a mechanism by which muscarinic agonists mediate their antipsychotic effects (Marino and Conn 2002; Jones et al. 2008). Consistent with this hypothesis, TPBP reversed amphetamine-induced hyperactivity and demonstrated a FOS expression profile similar to both xanomeline and atypical antipsychotics, and these effects were achieved at doses that did not induce catalepsy (Jones et al. 2008).

Indeed, evidence increasingly suggests that in addition to pro-cognitive effects, M1 receptor activation may also have antipsychotic effects (Vanover et al. 2008; Mirza et al. 2003; Friedman 2004), consistent with the behavioral effects of TPBP. For example, M1 knockout mice are hyperactive, most likely due to increased dopamine in the striatum (Wess et al. 2003; Gerber et al. 2001). Striatal hyperdopaminergia has been linked to acute psychotic states in schizophrenia (Schmitt Meisenzahl et al. 2009), striatal neurotransmission in schizophrenia (Gerber et al. 2001), suggesting that M1 receptor abnormalities may play a role in psychosis.

4.1.3 BQCA

Benzylquinolone carboxylate (BQCA) is a potent and highly selective M1 positive allosteric modulator that exhibits no agonist properties, but instead greatly enhances the potency of acetylcholine (Ma et al. 2009; Shirey et al. 2009). In wild-type mice but not in M −/−1 mice, oral administration of BQCA induced FOS activation in the cortex, hippocampus, and cerebellum, and significantly increased the ratio of phosophorylated ERK to total ERK (Ma et al. 2008). BQCA ALSO increased contextual fear conditioning in animals that were coadministered scopolamine, but the associative learning was blocked for animals receiving scopolamine only (Ma et al. 2009). The ability of BCQA to counteract the effects of scopolamine in this hippocampus-dependent task suggests that M1 may enhance learning by reinforcing associative learning; however, M1 receptors do not appear to be critical for contextual fear conditioning because M1 knockout mice show no acquisition deficit on this task (Anagnostaras et al. 2003; Miyikawa et al. 2001), and the allosteric selective M1 antagonist VUO255035 had no effect on contextual fear conditioning (Sheffler et al. 2009). BQCA increased the excitability of mPFC cells in slice preparations from wild-type but not M1 null mutant mice, and improved impaired PFC-dependent reversal learning in a mouse model of Alzheimer’s disease (Shirey et al. 2009).

BQCA, TPBP, and AC-42 all reduced amphetamine-induced hyperactivity; however, TBPB and AC-42 did not counteract scopolamine’s effects on fear conditioning, which may be due to a different mechanism of action on the part of BQCA, as suggested by the ability of this drug to induce β-arrestin recruitment (Ma et al. 2009). If M1 activation has both antipsychotic and pro-cognitive properties, it would be a particularly attractive target for schizophrenia-relevant therapies (Vanover et al. 2008). The finding that BQCA can mimic the antipsychotic-like profile of earlier allosteric M1 activators and also exhibited pro-cognitive effects in contextual fear conditioning, an animal model of cognition suggests that it may have considerable advantages for treatment of schizophrenia.

4.2 Selective M4 Allosteric Activators

The M4 receptor is particularly relevant to schizophrenia for several reasons. This receptor is implicated in the regulation of dopamine levels in brain regions important in the pathophysiology of schizophrenia, including the nucleus accumbens (Tzavara et al. 2004) and the striatum (Zhang et al. 2002a, b; Gomeza et al. 1999), where it is an inhibitory autoreceptor on cholinergic nerve terminals (Zhang et al. 2002a, b). The M4 receptor is also believed to play a role in the antipsychotic properties of muscarinic agonists such as xanomeline (Mirza et al. 2003) as well as the atypical antipsychotic drug clozapine (Olianas et al. 1999), whose affinity for the M4 receptor may be one source of its antipsychotic efficacy. Moreover, although the M1 receptor has been emphasized as a possible mechanism mediating cognitive improvements observed following xanomeline administration (Felder et al. 2000; Bymaster et al. 2003b) and a wealth of evidence supports a role for this receptor in mediating various aspects of cognition, the presynaptic location of M4 mAChRs excitatory neurons within the hippocampal formation (Rouse et al. 1999) suggests that they may modulate neurocognitive function as well. The finding that M4 mRNA expression is decreased in schizophrenia but M1 density being unchanged supports the argument that reductions in M4 density may play an important role in learning and memory deficits observed in schizophrenia (Scarr et al. 2007).

4.2.1 VU010010 and Analogs

The first M4 allosteric potentiator was reported by Shirey et al. (2008). VU010010 selectively enhanced the affinity of acetylcholine for the M4 receptor and enhanced its efficacy. Recordings from hippocampal cells revealed that VU010010 potentiated carbachol’s depression of excitatory postsynaptic potentials at schaffer collateral-CA1 synapses in wild-type but not M4 knockout mice, suggesting a role for the M4 receptor in mediating NMDA-mediated excitatory neutrotransmission. Further optimization of VU010010 led to the development of two additional potent and selective allosteric modulators of the M4 receptor, VU0152099 and VU0152100, which have increased bioavailability and superior pharmacokinetic profiles (Brady et al. 2008; Conn et al. 2009). Both have no agonist effects at M4, but instead potentiate the effects of acetylcholine. These molecules do not bind with other G-protein-coupled receptors, muscarinic or otherwise, and both potentiated M4 response to acetylcholine as measured by enhanced calcium mobilization. Importantly, acetylcholine was more potent in the presence of these compounds as demonstrated by a dramatic increase in the ability of ACh to displace [3H] NMS. Behaviorally, both compounds reversed amphetamine-induced hyperactivity, demonstrating antipsychotic-like activity. This is consistent with evidence from M4 knockout mice that M4 receptors modulate cholinergic and dopaminergic neurotransmission and that loss of M4 function results in hyperdopaminergia (Tzavara et al. 2004). In the midbrain, cholinergic excitation activates dopamine release, and data from M4 knockout mice suggest that these mAChRs serve as inhibitory autoreceptors in the midbrain (Tzavara et al. 2004). Therefore, M4 agonism could reduce acetylcholine release and subsequent overexcitation of midbrain dopamine neurons, which would decrease dopamine release in subcortical structures. This mechanism may provide an explanation for the antipsychotic-like profile of VU0152099 and VU0152100 (Brady et al. 2008) as well as the antipsychotic properties of agents with partial M4 selectivity, including clozapine (Olianas et al. 1999), xanomeline (Stanhope et al. 2001; Andersen et al. 2003; Mirza et al. 2003; Shekhar et al. 2008), and the M2/M4 preferring partial agonist PTAC (Fink-Jensen et al. 1998).

4.2.2 LY2033298

The Eli Lilly compound LY2033298 was recently identified as a highly potent (>40-fold increase in potency) and selective allosteric potentiator of M4 receptors that acts primarily by increasing the affinity of acetylcholine for the M4 receptor as well as demonstrating agonist activity (Chan et al. 2008; Leach et al. 2010). LY2033298 has shown efficacy in two animal models of psychosis; specifically, it attenuated conditioned avoidance responding and reversed apomorphine-induced disruptions of pre-pulse inhibition (Chan et al. 2008). A reduction in conditioned avoidance responding was also observed in M4 knockout mice, but the effect was significantly smaller compared to that in wild-type mice (Leach et al. 2010). These findings are consistent with the finding that PTAC and BuTAC, which are M2/M4 partial agonists with M1/M3/M5 antagonist properties, display antipsychotic-like profiles in animal models, including inhibition of conditioned avoidance responding (PTAC; Bymaster et al. 1998), inhibition of apomorphine-induced climbing, and impaired passive avoidance responding (BuTAC; Rasmussen et al. 2001). Taken together with evidence that M4 modulates dopaminergic neurotransmission in regions implicated in positive symptoms of schizophrenia (Tzavara et al. 2004; Gomeza et al. 1999), the behavioral effects of LY2033298 provide additional evidence that M4 agonist activity may be a viable novel therapeutic approach for psychotic symptoms of schizophrenia.

5 M2 and M5 mAChRs as Potential Therapeutic Targets

The focus on muscarinic receptor-focused therapies for schizophrenia has overwhelmingly focused on M1 and M4 receptors. However, there is intriguing, but limited, evidence that the M2 and M5 receptors may also be potential therapeutic targets.

5.1 M2 Receptor

M2 receptors are found throughout the brain and CNS, including the basal forebrain, where they act primarily as inhibitory autoreceptors, regulating acetylcholine release from forebrain projections including the hippocampus (Zhang et al. 2002a, b; Kitaichi et al. 1999a, b; Rouse et al. 1999, 2000) and cortex (Zhang et al. 2002a, b). The M2 receptors have been implicated in cognitive and psychotic symptoms of schizophrenia (Eglen 2005; Fisher 2008), but are especially believed to play a significant role in learning and memory due to their prominence in the hippocampus, where they are found pre- and postsynaptically.

Numerous studies have reported that M2 receptor antagonists with various levels of selectivity have increased acetylcholine release in vitro in the hippocampus, cortex, and striatum (Billard et al. 1995; Wang et al. 2002; Carey et al. 2001; Quirion et al. 1995; Vannucchi et al. 1997), presumably through inhibition of this negative feedback mechanism. Corticostriatal recordings in rat slice preparations also showed that an antagonist of M2-like receptors, methoctramine, facilitates striatal long-term potentiation (Calabresi et al. 2000). The finding that the M2 selective antagonist SCH 55790 enhanced hippocampal, cortical, and striatal acetylcholine release (Carey et al. 2001) is consistent with reports of increased acetylcholine release in hippocampus and cortex in the presence of less selective M2 antagonists such as BIBN-99 and AF-DX 384 (Quirion et al. 1995; Vannucchi et al. 1997). Interestingly, anatomical evidence of M2 receptor localization to non-cholinergic neurons indicates that it also acts a presynaptic heteroreceptor (Rouse et al. 2000).

Behaviorally, a number of M2 antagonists have shown pro-cognitive effects in animal models. For example, bilateral infusion of methoctramine into the dorsolateral striatum of rats improved performance on a memory task (Lazaris et al. 2003). The compound (+)-14 had high oral efficacy, was highly selective for the M2 receptor, and significantly decreased passive avoidance response latency in young rats (Wang et al. 2002), a result also reported for the highly selective M2 antagonist SCH 72788 (Lachowicz et al. 2001). Both SCH 57790 and BIBN-99 induced similar improvements in preclinical models of learning and memory (Carey et al. 2001; Rowe et al. 2003).

It should be noted, however, that in contrast to the findings that M2 antagonism enhanced performance in pharmacological experiments, Seeger et al. (2004) found that M2-deficient mice showed impaired learning on a hippocampus-dependent spatial learning task and impaired behavioral flexibility. In marked contrast to M4 (Gerber et al. 2001) null mutant mice, M2 knockout mice were not different from wild-type mice on locomotor activity, consistent with the hypothesis that of the two inhibitory mAChRs, M4 has a greater role in regulating dopaminergic neurotransmission (Tzavara et al. 2004).

Antagonism of presynaptic M2 receptors increases synaptic acetylcholine levels (Meyer and Otero 1985; Billard et al. 1995; Wang et al. 2002), which could lead to increased M1 receptor activation (Fisher 2008). Thus, it has been hypothesized that M2 antagonists may be a possible novel therapeutic direction for the improvement of cognitive impairment and psychotic symptoms (Eglen 2005; Clader and Wang 2005; Fisher 2008). However, to date no clinical studies of M2 selective agonists have been conducted in schizophrenia. More seriously, although they could be efficacious in treating cognitive and psychotic symptoms in schizophrenia, enthusiasm for M2-targeted therapies is limited due to their high expression in cardiac tissue (Caulfield 1993; Brodde and Michel 1999), which would likely necessitate more specific CNS targeting than is currently available (Bymaster et al. 2002; Fisher 2008).

5.2 M5 Receptor

Although the M5 receptor is found in the cerebral cortex and hippocampus, it is especially predominant in the substantia nigra and ventral tegmental brain regions, where it is localized to dopaminergic neurons (Vilaró et al. 1990; Weiner et al. 1990). Its localization to the so-called “reward pathways” has prompted speculation that it may be an important target for treatment of schizophrenia (Mirza et al. 2003) as well as drug abuse (Raffa 2009; Basile et al. 2002; Fink-Jensen et al. 2003). Dysregulation of motivational drive is a central feature of schizophrenia, implicating M5 receptors as potential targets for treatment in the disorder.

In addition to its probable role in modulating reward sensitivity, the M5 receptor has also been implicated in tonic regulation of mesolimbic and striatal dopamine levels (Blaha et al. 1996; Zhang et al. 2002a, b; Basile et al. 2002; Forster and Blaha 2003). In addition, xanomeline’s antipsychotic effects may be attributable in part to its partial agonism of the M5 receptor in striatum, although M4 is a more likely mechanism (Mirza et al. 2003). M5-deficient mice retain phasic but not sustained dopamine release into the nucleus accumbens (Forster et al. 2002). M5 receptors in VTA activate mesolimbic dopamine input to the nucleus accumbens (Yeomans et al. 2001), and M5 receptor activation may result in sustained activation of dopaminergic neurons (Forster et al 2002). A study by the same group (Wang et al. 2004) reported that compared to control animals, M5-deficient mice have improved latent inhibition and decreased amphetamine induced locomotor activity, consistent with reduced dopamine release in the nucleus accumbens. Given that earlier studies have reported that inactivation of dopamine terminals in the nucleus accumbens blocks amphetamine induced locomotion (Joyce and Koob 1981), and reduced nucleus accumbens dopamine activity results in increased latent inhibition (Joseph et al 2000; Moser et al 2000; Russig et al 2002; Gray et al 1997), it is probable that decreased dopamine release in the M5-deficient mice produced these behavioral results. Taken together, these results suggest that antagonism at the M5 receptor may reduce psychotic symptoms of schizophrenia by decreasing subcortical dopamine release.

To date, no clinical or preclinical studies of M5 selective compounds have been undertaken. However, two recent reports have described M5 allosteric modulators. The first such report characterized VU0238429, which displayed high selectivity (>30-fold) for the M5 receptor in comparison to M1 and M3, and no potentiator activity in M2 and M4 receptor transfected cells (Bridges et al. 2009). The previous study found that VU0238429 increased the potency of acetylcholine, but had poor brain penetration. The second study described the allosteric properties of the anti-arrythmia drug amiodarone, which was found to be an allosteric potentiator at the M5 receptor, but not M1 receptors; interestingly, amiodarone enhanced acetylcholine’s efficacy at the M5 receptor, but not its potency (Stahl and Ellis 2010). Discovery of these molecules provides a significant breakthrough and should lead to additional chemical modifications, electrophysiological studies, and in vivo characterizations of M5 selective modulators in order to gain additional insight into the role of this receptor in psychosis and addictive behavior.

6 Conclusion

As reviewed above, it has become increasingly evident that the muscarinic system is an attractive novel target for treating cognitive and psychotic symptoms of schizophrenia. The major obstacle to exploiting this receptor system’s therapeutic promise has been the lack of selectivity for specific receptor subtypes. Therefore, to date, the few muscarinic agonists that have been tested in humans have shown efficacy, but more selective compounds could make this approach highly fruitful in developing new therapies for schizophrenia. New generations of allosteric activators targeting M1 and M4 receptors have now demonstrated improved selectivity and some preclinical evidence of antipsychotic-like and pre-cognitive effects. These compounds may offer substantial therapeutic benefit for the treatment of cognitive and psychotic symptoms of schizophrenia and could be entering clinical trials in the next few years.