2018 Volume 66 Issue 1 Pages 37-44
Among the muscarinic acetylcholine receptor (mAChR) subtypes, the M4 receptor has been investigated as a promising drug target for the treatment of schizophrenia. These investigations have been based on findings from M4-deficient mice studies as well as on the results of a clinical trial that used xanomeline, an M1/M4 mAChRs-preferring agonist. Both orthosteric agonists and positive allosteric modulators of M4 mAChR have been reported as promising ligands that not only have antipsychotic effects, but can also improve cognitive impairment and motor dysfunction. However, challenges remain due to the high homology of the orthosteric binding site among all muscarinic receptors. In this review, we summarize our approach to the identification of M4 mAChR activators, orthosteric agonists, and positive allosteric modulators based on M4 mAChR structural information and structure–activity relationship studies. These findings indicate that selective M4 mAChR activators are promising potential therapeutic agents for several central nervous system conditions.
Neural responses to neurotransmitter signals play a crucial role in the regulation of various functions in the central nervous system (CNS). Acetylcholine (ACh) is one of the most important neurotransmitters and is synthesized by choline acetyltransferase from choline and acetyl CoAs. It is well-known that ACh system plays a crucial role on higher brain function.1,2) Aside from its role in the brain, ACh is involved in a number of regulatory mechanisms in the sympathetic and parasympathetic nerve systems. These findings promoted ACh as a pivotal target in a number of drug discovery studies and subsequently led to the identification of several CNS-acting drugs. Among them, ACh esterase inhibitors (AChEIs), such as donepezil, galantamine, and rivastigmine, are currently used for the treatment of cognitive dysfunction associated with Alzheimer’s disease (AD). However, although AChEIs can ameliorate cognitive symptoms in patients with AD, they have limited clinical utility because of unwanted side effects and their limited efficacy in treating behavioral and psychological symptoms of dementia (BPSD).3–8) Direct and subtype-selective activation of ACh receptors are therefore considered to be better approaches to treat cognitive impairment and BPSD in AD.9)
Muscarinic ACh receptors (mAChRs) belong to the rhodopsin-like G-protein-coupled receptor (GPCR) family10–13) and are classified into five subtypes (Fig. 1). M1, M4, and M5 mAChRs are predominantly expressed in the brain, while M2 and M3 mAChRs are highly distributed in peripheral tissues.12,14) Among these subtypes, M1 and M4 mAChRs are thought to be promising targets for the treatment of AD and schizophrenia. These assumptions are supported by knockout mice studies, as well as by results from clinical trials of xanomeline, an M1/M4 mAChRs-preferring agonist. It is believed that identification of M1 or M4 mAChR-selective agonists or M1/M4 mAChRs-dual activators could enhance the safety profile of mAChR ligands. In the early 2000 s, a number of M1 mAChR-selective agonists were reported and proceeded to clinical development,9) but the identification of M1/M4 mAChRs-dual or M4 mAChR-selective activators remains challenging because of high sequence similarity of the orthosteric ACh binding site among mAChR subtypes.15) In this review, we present recent findings on a novel M4 mAChR-selective activator as a promising therapeutic agent for CNS diseases.
The benefits of M4 mAChR activation for psychotic symptoms in CNS diseases are supported by a number of clinical and preclinical studies.16) Indeed, xanomeline, an M1/M4 mAChRs-preferring agonist, has been shown to improve cognition as well as behavioral and psychological symptoms in patients with AD, although it caused adverse gastrointestinal effects in a dose-dependent manner.17) In addition, clinical studies with xanomeline in patients with schizophrenia suggested that activation of M1/M4 mAChRs is effective in the treatment of positive, negative, and cognitive symptoms of this disorder.18,19) In preclinical studies, xanomeline has been shown to exhibit potent efficacies in behavioral tasks associated with psychosis (i.e., conditioned avoidance, apomorphine-induced climbing, amphetamine-induced locomotor hyperactivity, prepulse inhibition (PPI) deficit, and psychotic-like behaviors) in rodents and non-human primates.20–23) Using pharmacological magnetic resonance imaging, xanomeline suppressed the N-methyl-D-asparate (NMDA) blocker ketamine-induced activation on blood oxygen level-dependent signal across several brain regions including the association, motor, and primary sensory cortices in rats.24) Furthermore, M4 mAChR knockout mice studies have highlighted the potential of M4 mAChR activation in the treatment of psychosis. Mice lacking M4 mAChR showed increased basal locomotor activity and PPI deficits.25,26) Dopamine release, induced by amphetamine or another NMDA receptor antagonist phencyclidine, is elevated in the nucleus accumbens of M4 mAChR knockout mice, suggesting a role for M4 mAChR in preventing hyperexcitability in midbrain dopamine neurons. Woolley et al. reported that the effect of xanomeline on amphetamine-induced locomotor hyperactivity is marginally attenuated in M1 mAChR knockout mice, but is canceled in M4 mAChR knockout mice.27) Furthermore, Dencker et al. have shown that xanomeline does not have antipsychotic-like effects in mice that lack the M4 mAChR in D1 dopamine receptor-expressing cells.28)
As expected, M4 mAChR activators have also been found to possess antipsychotic effects in behavioral tests for psychosis. These compounds reversed psychostimulant- or NMDA receptor antagonist-induced locomotor hyperactivity, as well as psychostimulant-induced disruption of PPI29–38) in rodents. In a rat electroencephalogram study, the M4 mAChR positive allosteric modulator (PAM) VU0467154, like clozapine, reversed MK-801-induced elevations in high frequency gamma power, which is consistent with their antipsychotic activities.39) Interestingly, VU0467154 induced state-dependent alterations in sleep architecture and arousal. It brought about delayed rapid eye movement sleep onset, increased cumulative duration of total and non-rapid eye movement sleep, and increased arousal during waking periods. These results suggest that M4 mAChR PAMs may lack the adverse sedative effects that are commonly observed with antipsychotics.
Until recently, it was generally believed that M4 and/or M1 mAChR activation was associated with antipsychotic efficacy, while M1, but not M4, mAChR activation was associated with enhancing cognitive function.16) However, recent preclinical work suggests that M4 mAChR also have an important role in cognitive function.40) The M4 mAChR PAM VU0152100 improves memory in a rat object recognition task. In addition, VU0467154 was shown to reverse MK-801-induced learning and memory deficits in a touch screen pairwise visual discrimination task and a fear conditioning experiment in mice.41) Importantly, the beneficial effects of VU0467154 on associative learning are absent in M4 mAChR knockout mice, which indicates that these cognitive effects are mediated by the M4 mAChR.
Currently, M4 mAChR activators are believed to have therapeutic potentials for motor dysfunction in neurological diseases. Pancani et al. reported that chronic administration of VU0467154 beginning at a pre-symptomatic age (2 months old) could prevent the appearance of deficits in both glutamatergic and dopaminergic neurotransmissions in 5 months old YAC128 mice models of Huntington’s disease.42) Motor coordination deficits and locomotor hypoactivity were also prevented by chronic treatment in YAC128 mice. Additionally, Shen et al. reported that the M4 mAChR PAM VU10010 blocks aberrant long-term potentiation in direct-pathway spiny projection neurons in an L-3,4-dihydroxyphenylalanine (L-DOPA)-induced dyskinesia (LID) mice model.43) Because giant cholinergic interneurons have dense terminal fields that overlap those of dopaminergic neurons, they promote M4 mAChR suppression of dopamine D1 receptor signaling through adenylyl cyclase,44,45) leading to the modulation of motor functions. Actually, acute treatment with two other M4 mAChR PAM, VU0467154 or VU0476406, was found to attenuate LID behaviors in mice and rhesus monkeys. It is worth mentioning that these beneficial effects of M4 mAChR PAMs on LID behaviors do not compromise the symptomatic benefits of L-DOPA treatment.43)
A number of orthosteric mAChR activators have already been reported and used clinically (Fig. 2). The representative orthosteric M4 mAChR agonist xanomeline has been shown to improve the three major symptoms (positive symptoms, negative symptoms, and cognitive impairment) of schizophrenia. Although xanomeline is reported as an M1/M4 mAChRs-preferring agonist, our calcium mobilization assay revealed that xanomeline exhibits non-selective activation of mAChRs in Chinese hamster ovary cells that express human M1–M5 mAChRs.30)
Based on the reported clinical benefits and potential risks of xanomeline, we launched an exploratory program aimed at discovering selective M1 and M4 mAChRs-dual agonists. A high-throughput screening (HTS) campaign of our chemical library led to the discovery of lead scaffolds. Using previously reported information,46,47) we devised a hypothetical pharmacophore for allosteric M4 mAChR agonistic activity and hybridized it with HTS hit compounds (Fig. 3).
Hybridization of the hit compound 1 and the M4 mAChR pharmacophore led to the preferable second lead compound 2, which showed higher M1 and M4 mAChR agonistic activity relative to 1. However, compound 2 also exhibited agonistic activity for M2 and M3 mAChRs. Replacement from a benzene to a pyridine ring attenuated M3 mAChR agonistic activity. In addition, optimization of the substituents at the nitrogen atom gave the M1 and M4 mAChRs-selective agonist 3. Compound 3 displayed strong M1 (91% at 1 µM) and M4 (134% at 1 µM) mAChRs activation with weak or negligible activation of M2 (27% at 10 µM), M3 (3% at 10 µM), and M5 (3% at 10 µM) mAChRs in a calcium mobilization assay. In addition, compound 3 showed moderate human ether-a-go-go related gene (hERG) inhibition (IC50=2.9 µM)48) (Fig. 4).
However, pharmacokinetic (PK) studies of this compound revealed low bioavailability (BA) (2.6% at 2.5 mg/kg, per os (p.o.)) and rapid clearance (CL) (49.3 mL/min/kg) in rats. An analysis of compound 2 metabolites in urine revealed dealkylation. We then turned our attention to the newly fused ring moiety. Hybridization of compound 3 and the HTS hit compound 4 led to the 7-azaindoline derivative 5. Although compound 5 maintained high mAChR subtype selectivity (M1: 55%, M2: 13%, M3: 5%, M4: 77% at 1 µM, M5: 11% at 10 µM), it also showed strong hERG inhibition (IC50=0.23 µM). Eventually, conversion to the N,N′-dimethyl urea moiety provided compound 6 with good M1 and M4 subtype selectivity (M1: 76%, M2: 10%, M3: 2%, M4: 118% at 0.3 µM, M5: 11% at 10 µM) and ameliorated hERG inhibition (IC50>10 µM). In addition, compound 6 showed improved BA (51.1% at 2.5 mg/kg, p.o. and CL (28.3 mL/min/kg), and good blood–brain barrier penetration (brain/blood ratio=0.7)29) (Fig. 5).
Our search also led to the identification of M1 and M4 mAChRs partial agonists; i.e., N-substituted oxindole derivatives. In this case, the high versatility M4 mAChR agonist pharmacophore was first incorporated into the HTS hit compound 7, and then the ring size of the amine linker was modified to give the lead compound 8, which had good selectivity for M1 and M4 over the M2, M3, and M5 mAChRs. Replacement of the piperidine moiety with tropane, a bulky amine ring, led to the discovery of compound 9, which partially but selectively activated M1 (EC50=12 nM, IA=60%) and M4 (EC50=29 nM, IA=42%) mAChRs over M2, M3, and M5 mAChRs. In addition, compound 9 (3 µM) showed negligible binding to 68 off-target proteins, including a broad range of GPCRs, ion channels, enzymes, and transporters. Finally, compound 9 exhibited potent CNS penetration (brain/blood ratio=2.0) in rats.30)
In addition to the discovery of an M4 mAChR partial agonist, dihydroquinazolinone derivatives were identified as selective M1 full and M4 mAChRs partial agonists (compound 10; M1: 81%, M2: 1%, M3: 2%, M4: 49% at 0.3 µM, M5: 3% at 30 µM)31) (Fig. 6).
Finally, we succeeded in identifying a selective M4 mAChR agonist.32) The introduction of a methanesulfonyl group into the 7-azaindoline scaffold resulted in a decrease in M1 mAChR agonistic activity, while M4 mAChR agonistic activity was maintained. Further optimization around the N-carbethoxypiperidine, by introducing a methyl group at the 4-position, led to compound 11, which displayed high M4 mAChR selectivity (M1: 5%, M2: 3%, M3: 4%, M4: 94% at 0.3 µM, M5: 3% at 10 µM) and moderate hERG inhibition (IC50=3.63 µM). In addition, compound 11 showed good BA (49% at 1 mg/kg, p.o.) and acceptable brain penetration (brain/blood ratio=0.9) in rats. To the best of our knowledge, this is the first example of such a selective M4 mAChR agonist. Compound 11 is therefore expected to be both a promising candidate for the treatment of CNS diseases and a suitable tool for the elucidation of M4 mAChR function (Fig. 7).
Using our promising pharmacophore for M4 mAChR activation, we could identify various M1/M4 and M4 mAChR agonists. Recently, researchers at Heptares Therapeutics reported a unique pharmacophore that helped to identify selective M1/M4 mAChRs and M4 mAChR activators using the Alphascreen Surefire phospho-ERK1/2 assay.49–54) Other novel selective mAChR agonists may be identified by optimizing the hetero ring moiety in compound 11, or even by modifying the pharmacophore for M4 mAChR activation.
Allosteric modulators bind to the allosteric site of a GPCR instead of the orthosteric site, which has theoretical advantages. Generally, PAMs display high target selectivity due to higher sequence divergence in the allosteric sites compared with the orthosteric site of the receptor. In addition, PAMs do not induce independent agonistic activity, indicating that they generally possess a good safety profile.55)
In nature, thiochrome, an oxidation product and metabolite of thiamine (vitamin B1), acts as an M4 mAChR PAM. Thiochrome enhances the affinity of ACh by 5-fold at the M4 mAChR56) (Fig. 8).
The pioneers of M4 mAChR PAMs are Eli Lily, with LY2033298,33) and Vanderbilt University, with VU10010.57) Chan et al. described LY2033298 as a potent M4 mAChR PAM that selectively enhances the ACh response at the human M4 mAChR. Szabo et al., on the other hand, reported on the structure–activity relationship (SAR) of the thieno[2,3-b]pyridine core of LY2033298.58) They found that the allosteric properties of LY2033298 progressively decrease when the O- and N-alkyl chains are substituted by groups greater than two carbon atoms. Although a chlorine atom was tolerated, it significantly decreased M4 mAChR positive modulation. Conversely, acylation of the primary amine completely abolished activity (Fig. 9).
Eli Lily researchers also reported LY2119620, an M2 and M4 mAChRs-dual PAM built from a thieno[2,3-b]pyridine core and a cyclopropyl amide scaffold.59) The introduction of an N-methyl acetylpiperazine moiety into LY2033298 led to LY2119620, with a potentiated muscarinic agonist iperoxo at both M2 and M4 mAChRs (Fig. 10).
Conversely, VU10010 was reported by Shirey et al., who based their research on LY2033298 and used cheminformatics and medicinal chemistry as an initial approach.57) They performed a search of the chemical database of ChemBridge Corporation and picked up 232 compounds containing a core structure similar to that of LY2033298. According to the SAR of the selected compounds, a dimethyl substitution on both the pyridine ring and the primary amine was preferable. After optimization of the substituents on the amide, the para-chlorobenzyl VU10010 was identified as a compound that binds to the allosteric site of M4 mAChR and selectively potentiates (47-fold) M4 mAChR-ACh concentration with an EC50 value in the 400 nM range. However, VU010010 showed unfavorable physiochemical properties, including high lipophilicity (Log P ca. 4.5), poor solubility, and P-glycoprotein efflux, which prevented its use in in vivo studies. As a result of a SAR study, functionalized benzyl amides, as well as pyridyl methyl congeners, were found to be well tolerated, providing selective M4 mAChR PAMs with EC50 values ranging from 380 nM to 3.7 µM, and shifts in the ACh dose-response curve ranging from 8.6- to 70-fold. The identified compounds; i.e., VU0152099 (rat M4 EC50=0.4 µM, 29.7-fold shift in ACh) and VU0152100 (rat M4 EC50=0.38 µM, 70.1-fold shift in ACh), had improved Log P values (3.65 and 3.6, respectively).34) Following administration of these compounds (56.6 mg/kg intraperitoneally (i.p.)) to rats, the area under the curve (AUC) brain/AUC plasma ratio for VU0152099 was 0.39±0.01, whereas that for VU0152100 was 0.86±0.08. Further optimization of VU0152099 and VU0152100 produced ML173 (human M4 EC50=500 nM, rat M4 EC50=900 nM)60) and VU0448088 (human M4 EC50=77 nM, rat M4 EC50=176 nM)35) as selective human and rat M4 mAChR PAMs (Fig. 11).
Wood et al. described 5-amino-thieno[2,3-c]pyridazine as a new chemotype that expands the 3-amino-thieno[2,3-b]pyridine core chemical diversity and improves on some of the remaining issues, including a steep SAR, species differences, and insufficient pharmacokinetic properties.36) The identified VU0467154 showed strong and selective M4 mAChR positive allosteric modulation with a good drug metabolism and pharmacokinetics (DMPK) profile. However, this compound displayed different M4 mAChR modulating activity between humans and rats (35 times less potent activity for the human M4 mAChR). Further optimization on the benzene ring led to the identification of the preclinical candidate, VU0467485/AZ13713945.37) This compound showed improved M4 mAChR modulating activity between humans and rats (human M4 EC50=78.8 nM, rat M4 EC50=26.6 nM) and an attractive DMPK profile in rats (CL=29 mL/min/kg, F=79%, brain/blood ratio=0.31 at 1 mg/kg iv and 3 mg/kg p.o.). VU0467485/AZ13713945 also showed clean CYP450 inhibition (3A4, 2D6, 2C9, 1A2 IC50s >30 µM in human hepatic microsomes), appropriate induction (3A4, 1A2, 2B6 EC50s >50 µM, Emaxs ≤1.0 in cryopreserved human hepatocytes), and weak hERG inhibition (IC50=11 µM) or significant off-target activity (200 targets in an internal AZ/Cerep panel). Moreover, Tarr et al. have recently reported a substitution of the benzyl linker with a 3-amino azetidine moiety, leading to VU6000918.61) This compound also showed an excellent M4 mAChR potency (human M4 EC50=19 nM) and DMPK profile in rats (CL=16 mL/min/kg, F=38%, brain/blood ratio=0.77). VU6000918 also demonstrated robust efficacy in a rat amphetamine-induced hyperlocomotion reversal model (minimum efficacious dose=0.3 mg/kg) (Fig. 12).
Other chemotypes have also been reported in the literature; VU0409524, with a benzothiazole scaffold and human M4 EC50 of 13 µM, was reported by Salovich et al.62) In addition, compounds with a thieno[2,3-d]pyrimidine core or a 6-fluoroquinazoline core, such as VU6002703 (human M4 EC50=0.60 µM) and VU6003130 (human M4 EC50=0.6 µM) were described by Wood et al.63) (Fig. 13).
Despite being useful for SAR studies, these compounds have limited design for titration curves because they yield apparently “flat” SAR profiles when compound activity relies on a single parameter. Huynh et al. reported useful information on parameters for the investigation of allosteric modulator activity.64) These parameters are the affinity of the allosteric ligand for the free receptor (KB); the cooperativity factors that define the magnitude and direction of the allosteric ligand’s effect on the orthosteric ligand’s affinity (α) and/or downstream efficacy (β); and the intrinsic agonist efficacy of the allosteric ligand (τB). It is worth mentioning that all synthesized compounds that demonstrate PAM activity of the M4 mAChR also exhibit intrinsic activation (τB) at the allosteric site in their own right.
Well-designed M4 mAChR PAMs have been reported by researchers at Vanderbilt University. As mentioned above, their effort led to an improvement in the benefits of M4 mAChR allosteric modulators. It is therefore expected that further work will validate the efficacy and safety of M4 mAChR PAMs in humans.
Until recently, only the crystal structures of M2 and M3 mAChRs have been reported.65,66) However, Thal et al. recently reported the crystal structures of M1 and M4 mAChRs bound to the inverse agonist tiotropium.67) In their work, the intracellular loop 3 of M4 mAChR was replaced with a minimal T4 lysozyme fusion to aid crystallization. It was also necessary to remove the first 21 residues of the amino terminus from the M4 mAChR to improve diffraction. This led to the determination of the M4 mAChR structure with a resolution of 2.6 Å. Overall, the structures of the M1 and M4 mAChRs are similar to the previously identified structures of inactive M2 and M3 mAChRs, with similar positioning of the seven-transmembrane bundle and root mean squared deviations of 0.6–0.9 Å. In M4 mAChR, the rotameric change of D1123.32 resulted in pointing D1123.32 away from the tiotropium and is accompanied by slight movements of Y4397.39 and Y4437.43, allowing the formation of a hydrogen bond network between D1123.32 and S852.57, W1083.28, Y4397.39, and Y4437.43, which is distinct from the M1, M2, and M3 mAChR structures.
Furthermore, Thal et al. revealed the similarity of the orthosteric binding site and the difference of the allosteric biding site between M1–M4 mAChRs subtypes owing to differences in amino acid composition. The alanine mutations of M4 mAChR are believed to be fundamental for interaction with the selective M4 PAM LY2033298. Specifically, W4357.35 at the top of TM7, Y1133.33, Y4166.51, and Y4397.39, which form the roof of the orthosteric site, as well as Y892.61, W1083.28, and L1093.29, are all considered to contribute to the PAM binding pocket. These findings are helpful in the search for novel M4 mAChR activators.
Based on the interaction between M4 mAChRs and dopamine D1 receptors or other targets, an M4 mAChR activator has the potential to be a therapeutic agent for psychotic symptoms and cognitive dysfunction in schizophrenia and AD, as well as for motor dysfunction in Parkinson’s disease and Huntington’s disease. Potent, selective, and brain penetrant M4 mAChR orthosteric agonists and M4 mAChR positive allosteric modulators have already been reported as promising ligands that display antipsychotic effects and are beneficial for cognitive and motor dysfunctions. Finally, revelation of the structure of the M4 mAChR will help us to understand the SAR of ligands and thus identify novel M4 mAChR activators for clinical use.
We are grateful to Dr. Takaaki Sumiyoshi for his support and encouragement.
The authors are employees of Sumitomo Dainippon Pharma Co., Ltd..
