Discovery of VU6008677: A Structurally Distinct Tricyclic M4 Positive Allosteric Modulator with Improved CYP450 Profile

This Letter details our efforts to develop novel tricyclic muscarinic acetylcholine receptor subtype 4 (M4) positive allosteric modulator (PAM) scaffolds with improved pharmacological properties. This endeavor involved a “tie-back” strategy to replace the 3-amino-5-chloro-4,6-dimethylthieno[2,3-b]pyridine-2-carboxamide core, which led to the discovery of two novel tricyclic cores: an 8-chloro-9-methylpyrido[3′,2′:4,5]thieno[3,2-d]pyrimidin-4-amine core and 8-chloro-7,9-dimethylpyrido[3′,2′:4,5]furo[3,2-d]pyrimidin-4-amine core. Both tricyclic cores displayed low nanomolar potency against human M4 and greatly reduced cytochrome P450 inhibition when compared with parent compound ML253.

−17 This particular chemotype engendered poor solubility in earlier M 4 PAMs, varying degrees of P-gp efflux, and potency discrepancies across species.−30 For instance, xanomeline, an M 1 /M 4 -preferring agonist lacking the β-amino carboxamide moiety, has been evaluated in clinical trials.These trials have given further validation to targeting the muscarinic cholinergic system as a treatment for the psychosis and behavioral disturbances observed in both Alzheimer's and schizophrenia patients. 25,26anomeline's lack of selectivity among receptor subtypes resulted in adverse events and ultimately the discontinuation of clinical development.An effort to overcome these side effects led to the development of KarXT, which was recently submitted as a New Drug Application to the FDA.KarXT is a treatment in which trospium (a pan-selective peripheral muscarinic acetylcholine receptor antagonist) is coadministered with xanomeline to counteract the adverse events of xanomeline administration alone. 27Recently, a selective M 4 PAM was shown to not only be efficacious in preclinical assays but also exhibited fewer and less severe adverse cholinergicrelated side effects when compared with rats treated with the nonselective M 4 agonist xanomeline. 28These data suggest that receptor-subtype-selective M 4 PAMs exhibit improved safety profiles compared with agonists, and CVL-231 (a selective M 4 PAM) is currently in clinical testing. 29,30−16 Most recently, we described a novel 6,5,6-tricyclic scaffold that still afforded potent and CNS-penetrant M 4 PAMs (Figure 1, VU6007215).To identify additional novel tricyclic M 4 PAM chemotypes, we elected to further explore the "tie-back" strategy, which aids in masking the detrimental β-amino carboxamide moiety (an essential pharmacophore in earlier M 4 PAMs).Using an historical M 4 PAM (ML253) developed by our laboratory as a starting point, we employed this "tie-back" strategy to generate novel tricyclic cores. 16,31  The synthesis of tricyclic cores 5 and 6 began with a Gewald-type reaction (Scheme 1).Treatment of ethyl 2mercaptoacetate or ethyl 2-hydroxyacetate with commercially available nicotinonitriles 7a or 7b under basic conditions with irradiation in a microwave reactor afforded carboxylates 10a− c.Treatment of a heated solution of intermediates 10a−c in formamide with a formamidine acetate salt gave pyrimidone intermediates 11a−c.Pyrimidones 11a−c were then converted into chlorides 12a−c with POCl 3 , which then readily underwent nucleophilic aromatic substitutions with a variety of amines to yield desired analogues 13−15.For this work, we decided to forego exploring amino azetidines of past M 4 PAMs because of their potential metabolic instability. 15Instead, we focused on examining small aliphatic amines, as well as benzyl amines, that could aid in increasing solubility.When indicated, HCl salts of final analogues were prepared to aid in improving solubility.
Subsequently, we investigated the relevance of the methyl substituent ortho to the pyridine of core 6.Toward this end, select analogues 15 were screened against hM 4 to determine potency with results highlighted in Table 3.Interestingly, removal of the methyl at the 2-position of the tricycle gave rise to several analogues with a hM 4 EC 50 ≤ 450 nM.Most notably, several amine tails that are inactive or have a hM 4 EC 50 > 10 μM in the context of core 13 (analogues 13a, b, d) gave rise to potent compounds (15c, hM 4 EC 50 = 450 nM; 15d, hM 4 EC 50 = 320 nM; 15h, hM 4 EC 50 = 29 nM).
Finally, we investigated the significance of nitrogen placement within the pyrimidine ring of core 6.Select analogues 22 were screened in tandem against hM 4 to determine potency with results highlighted in Table 4. Similarly, small aliphatic amine tails provided the most potent analogues, with a hM 4 EC 50 ≤ 200 nM (22h, hM 4 EC 50 = 150 nM; 22i, hM 4 EC 50 = 65 nM; 22j, hM 4 EC 50 = 83 nM; and 22k, hM 4 EC 50 = 39 nM).It was noted that minor changes, such as a fluorosubstitution on the cyclobutanamine tail (22k vs 22j), led to a 2-fold increase in potency.Like analogues 13, reducing the size of the amine tail by replacing the piperidine ring of 22b (hM 4 EC 50 = 860 nM) with pyrrolidine (22d, hM 4 EC 50 = 430 nM) resulted in a modest increase in potency.This trend was observed as the amine tail was further reduced in size to azetidine (22h, hM 4 EC 50 = 150 nM) and the cyclopropyl amine (22i, hM 4 EC 50 = 65 nM).As there is a 2-fold difference in hM 4 potency between 22h and 22i, it can be inferred that a hydrogen bond donor is not required for activity at this position.The addition of a methylene spacer to analogue 22j (hM 4 EC 50 = 83 nM) resulted in a nearly 5-fold decrease in potency (22f, hM 4 EC 50 = 400 nM).Apart from the  cyclopropanamine analogues (13l and 22i), moving the nitrogen from the 5-position of the tricycle to the 6-position generally resulted in more potent analogues.For instance, comparing the pyrrolidine analogues (13e vs 22d), the azetidine analogues (13f vs 22h), and the 3,3-difluorocyclobutan-1-amine analogues (13i vs 22k) resulted in a 14-fold, 21-fold, and 31-fold increase in hM 4 potency, respectively.
Of these compounds, 13l, 14o−p, 15e−h, and 22i−k were advanced into a battery of in vitro drug metabolism and pharmacokinetics (DMPK) assays (Table 5).In regards to physiochemical properties, these analogues all possessed molecular weights less than 450 Da, as well as several analogues having attractive CNS xLogP values (2.79− 3.39). 32,33All compounds tested displayed high predicted hepatic clearance (CL hep ) in rat microsomes (rat CL hep s > 47 mL/min/kg).While analogue 14o displayed moderate human predicted hepatic clearance on the basis of microsomal data (human CL hep 13 mL/min/kg), all other analogues tested displayed high human-predicted hepatic clearance (human CL hep ≥ 15 mL/min/kg).
In summary, a scaffold hopping exercise utilizing a "tie-back" strategy based on M 4 PAM 4 proved to be a successful strategy in converting an early lead compound ML253 into potent, tricyclic M 4 PAM analogues devoid of the classic β-amino carboxamide moiety.Unlike an earlier observation noted with the VU6007215 tricyclic scaffolds, a thiophene as the central ring was not greatly favored over a furan in the newer tricycle  series (13l vs 14o). 22To achieve highly potent compounds within these tricyclic series (13, 14, 15, and 22), aliphatic amines were preferred over benzyl amines.Unfortunately, many of these potent analogues displayed very poor rat brain and human plasma protein fraction unbound (f u values < 0.01), as well as high human (CL hep values ≥ 15 mL/min/kg) and rat (CL hep values ≥ 46 mL/min/kg) predicted hepatic clearance.Interestingly, masking the β-amino carboxamide moiety as a tricycle was an effective strategy to drastically improve the CYP450 inhibition profile in comparison with lead ML253.
Overall, 14o (VU6008677) displayed the best pharmacokinetics (PK) profile with moderate human predicted hepatic clearance (CL hep = 13 mL/min/kg), moderate rat brain homogenate binding (f u,brain = 0.017), favorable rat plasma protein binding (f u,plasma = 0.060), and selectivity over hM 2 .Compound 14o also provided an improved CYP inhibition profile (CYP 2C9, 2D6, and 3A4 IC 50 values ≥ 30 μM) when compared with parent compound ML253.However, VU6008677 was not progressed forward because of high predicted rat microsomal clearance, high human plasma protein binding (f u,plasma < 0.01), and CYP 1A2 inhibition (IC 50 < 0.10 μM).Although this exercise did not provide M 4 PAMs with suitable DMPK profiles to warrant further advancement, it did highlight SAR insights for future scaffold designs.These refinements will be reported in due course.Calcium mobilization assays with hM 2/Gqi5 -CHO cells were performed in the presence of an EC 20 fixed concentration of acetylcholine.EC 50 values for hM 2 represent one or two independent experiments performed in triplicate.hM 2 = human M 2 b f u = fraction unbound; equilibrium dialysis assay; brain = rat brain homogenates; c Assayed in pooled human liver microsomes (HLM) in the presence of NADPH with CYP-specific probe substrates.

Table 1 .
Structures and Activities for Analogues 13 bThe HCl salt form was utilized.

Table 2 .
Structures and Activities for Analogues 14a Calcium mobilization assays with hM 4/Gqi5 -CHO cells were performed in the presence of an EC 20 fixed concentration of acetylcholine.EC 50 values for hM 4 represent at least one experiment performed in triplicate.b The HCl salt form was utilized.

Table 3 .
Structures and Activities for Analogues 15a Calcium mobilization assays with hM 4/Gqi5 -CHO cells were performed in the presence of an EC 20 fixed concentration of acetylcholine.EC 50 values for hM 4 represent at least one experiment performed in triplicate.b The HCl salt form was utilized.

Table 4 .
Structures and Activities for Analogues 22a Calcium mobilization assays with hM 4/Gqi5 -CHO cells were performed in the presence of an EC 20 fixed concentration of acetylcholine.EC 50 values for hM 4 represent at least one experiment performed in triplicate.b The HCl salt form was utilized.