Further exploration of M₁ allosteric agonists: Subtle structural changes abolish M₁ allosteric agonism and result in pan-mAChR orthosteric antagonism

Abstract

This letter describes the further exploration of two series of M₁allosteric agonists, TBPB and VU0357017, previously reported from our lab. Within the TPBP scaffold, either electronic or steric perturbations to the central piperidine ring led to a loss of selective M₁ allosteric agonism and afforded pan-mAChR antagonism, which was demonstrated to be mediated via the orthosteric site. Additional SAR around a related M₁ allosteric agonist family (VU0357017) identified similar, subtle ‘molecular switches’ that modulated modes of pharmacology from allosteric agonism to pan-mAChR orthosteric antagonism. Therefore, all of these ligands are best classified as bi-topic ligands that possess high affinity binding at an allosteric site to engender selective M₁ activation, but all bind, at higher concentrations, to the orthosteric ACh site, leading to non-selective orthosteric site binding and mAChR antagonism.

Schizophrenia is a complex psychiatric disorder characterized by a combination of positive and negative symptoms along with significant cognitive dysfunction.^1,2 Current antipsychotic therapies can address the positive symptoms, but the negative and cognitive symptoms remain poorly managed, if at all, and are key predictors of functional disability.^3–6 A large number of anatomical, molecular, genetic, preclinical behavioral and human clinical studies have provided strong evidence that agents able to enhance cholinergic transmission or activate muscarinic acetylcholine receptors (mAChRs, M₁-M₅), notably M₁, have exciting therapeutic potential for the treatment of the positive, negative and cognitive symptoms of schizophrenia as well as cognitive dysfunction in other CNS disorders.^7–16 However, previous compounds developed to selectively activate M₁ receptors have failed in clinical development due to a lack of true specificity for this receptor subtype.^7–18 Often, many of the compounds bind to the orthosteric ACh binding site, which can result in adverse side effects as a result of M₂-M₅ activation.^7–18 Recently, multiple industrial and academic laboratories, including ours, have targeted less conserved allosteric sites on the M₁ receptor in an attempt to develop highly selective M₁ activators (both M₁ allosteric agonist and M₁ positive allosteric modulators, PAMs) and avoid activation of M₂-M₅.^17–39

For example, we have previously reported on TBPB 1, a potent, CNS penetrant and highly selective M₁ allosteric agonist that displays robust efficacy in multiple preclinical antipsychotic and cognition models, as well as significant impact on Aβ production.¹⁹ Mutagenesis and modeling efforts identified a key H-bond interaction between the central piperidine nitrogen of TBPB and Thr83 of M₁, likely contributing to TBPB’s affinity for this M₁ allosteric binding site.⁴⁰ In multiple Letters, we have also described SAR around TBPB 1, and found that halide substitutions were well tolerated on the benzimidazole core 2, as well as amide 3, sulfonamide 4 and urea 5 replacements for the benzyl amine of the distal basic piperidine nitrogen (Fig. 1); however, SAR was ‘shallow’.²¹ Moreover, these subtle structural modifications could engender D₂ antagonism, along with M₁ allosteric agonism, affording molecules with an attractive pharmacological profile for the treatment of schizophrenia.²¹ In this Letter, we describe a more detailed pharmacological profile of 1, along with the as yet unexplored SAR of the central piperidine ring of 1, and the discovery of subtle ‘molecular switches’⁴¹ that modulate modes of pharmacology from allosteric agonism to pan-mAChR orthosteric antagonism.

Figure 1.

Structures of the M₁ allosteric agonist TBPB (1), highlighting the key H-bond interaction with Thr83 for allosteric binding at M₁, and analogs 2–5 of TBPB that retain selective M₁ agonism.

Our previous characterization of TBPB revealed that this compound activates M₁ receptors.¹⁹ Here, we confirm that TBPB is a selective M₁ partial agonist and induces responses in CHO cells expressing rM₁ receptors, but not rM₂-M₅Rs (Fig. 2A). Interestingly, TBPB also inhibits ACh-induced responses in cells expressing M₂-M₅ receptors, suggesting additional activity at an orthosteric site (Fig. 2B). Based on this finding, TBPB can be described as a bi-topic ligand that binds M₁ at both an allosteric site and the orthosteric site, the former of which confers functional M₁ agonism and the latter of which confers orthosteric antagonism. These findings are similar to previously characterized M1 agonists including AC-42,^32,33 77-LH-28-1,³⁶ VU0364572,^28,30 and VU0357017^26,30, which have been characterized as bi-topic ligands.^42,43

Figure 2.

Pharmacological profile of TBPB. A) rM₁-rM₅ concentration-response curves (CRCs) of TBPB screened in agonist mode. rM₁ EC₅₀ = 79.6 nM (pEC₅₀ = 7.10±0.07), 69.3±6.4% ACh Max, rM₂-rM₅ >30 μM B) rM₂-rM₅ CRCs of TBPB screened as antagonists. rM₂ IC₅₀ = 1.29 μM (pIC₅₀ = 5.89), rM₃ IC₅₀ = 5.48 μM (pIC₅₀ = 5.26), rM₄ IC₅₀ = 493 nM (pIC₅₀ = 6.31), rM₅ IC₅₀ = 3.97 μM (pIC₅₀ = 5.40), and all reduce an ACh EC₈₀ to baseline. Human CRCs not shown. Screened as agonists, hM₁ EC₅₀ = 255 nM (pEC₅₀ = 6.59), 40% ACh Max, M₂-M₅ >30 μM. Screened as antagonists, hM₂ IC₅₀ = 3.5 μM (pIC₅₀ = 5.46), hM₃ IC₅₀ = 2.3 μM (pIC₅₀ = 5.63), hM₄ IC₅₀ = 734 nM (pIC₅₀ = 6.13), hM₅ IC₅₀ = 4.9 μM (pIC₅₀ = 5.31), and all reduce an ACh EC₈₀ to baseline.

As the central piperidine nitrogen of TBPB is thought to be critical (H-bond with Thr83) for allosteric binding,⁴⁰ we sought to disrupt this key H-bond both electronically and conformationally by installation of a β-fluorine atom (to attenuate basicity and H-bond acceptor ability) and replacement of the piperidine with a [3.3.0] system, respectively.²¹ Our expectation was that these structural modifications would abolish binding at the allosteric site, and that these analogs would bind solely at the orthosteric site and function as pan-mAChR antagonists. The synthesis was straightforward (Scheme 1), following the synthetic route previously developed.²²

Scheme 1.

Synthesis of TBPB analogs 9 (VU0465132) and 10 (VU0546403).

When screened as an agonist, neither 9 nor 10 elicited M₁ activation, suggesting that binding at the allosteric site had been abolished by disrupting the key H-bond interaction with Thr83.⁴⁰ Interestingly, both 9 and 10 proved to be weak, micromolar pan-mAChR antagonists when screened in the presence of an EC₈₀ of ACh against M₁-M₅. The [3.3.0] analog 10 was a weak M₁ antagonist (pIC₅₀ = 5.20±0.04, IC₅₀ = 6.4 μM, −1.07±1.3% ACh Max) and IC₅₀s >10 μM against M₂-M₅). The β-fluoro analog 9 (Fig. 3A) was slightly more potent at M₁ (Rat M₁ IC₅₀ = 4.9 μM (pIC₅₀ = 5.31±0.17), 13.6±1.5% ACh Max, M₂, M₃ M₄, and M₅ IC₅₀ > 10 μM (pIC₅₀s <5) (Human M₁-M₅ data not shown, but similar)). Radioligand binding experiments with [³H]-NMS (Fig. 3B) and dissociation kinetic experiments (Fig. 3C) are consistent with 9 acting as an ACh orthosteric site antagonist. Thus, a single fluorine atom serves as ‘molecular switch’ to modulate the pK_a of the central piperidine nitrogen atom (from ~11 to 9.4),⁴⁴ and likely prohibits its ability to accept a key H-bond from Thr83 in the M₁ allosteric site, an interaction that may confer M₁ functional agonist activity.⁴⁰

Figure 3.

Pharmacological profile of β-F-TBPB 9 (VU0465132). A) Rat M₁-M₅ CRCs screened as antagonists. Rat M₁ IC₅₀ = 4.9 μM (pIC₅₀ = 5.31±0.17), 13.6±1.5% ACh Max, M₂, M₃ and M₅ IC₅₀ > 10 μM (pIC₅₀s <5), M₄ IC₅₀ = 5.2 μM (pIC₅₀ = 5.29±0.09), 26.6±3.4% ACh Max (Human M₁-M₅ data not shown, but comparable). B) [³H]-NMS competition binding studies. At higher concentrations than the control atropine (Ki = 1.4 nM, pKi = 8.86±0.24), both 9 (Ki = 1.35 μM, pKi = 5.87±0.18) and TBPB (Ki = 0.60 μM, pKi = 6.22±0.14) fully displace [³H]-NMS, consistent with interactions at the orthosteric ACh site. C) Dissociation kinetics with TBPB, 9, atropine, carbachol (CCh) and Gallamine (Gall). In our assays, only the allosteric modulator Gall altered the rate of [³H]-NMS dissociation, providing further evidence that 9 acts as an orthosteric antagonist. K_off rates were as follows ± SEM: Atropine (0.0379±0.0125 min⁻¹), CCh(0.0393±0.0125 min⁻¹), Gall (0.0012±0.0006 min⁻¹), TBPB (0.0335±0.0095 min⁻¹), VU0465132 (0.0308±0.0103 min⁻¹).

Interestingly, our dissociation kinetics experiments with TBPB indicate that this compound also does not alter the off-rate of [³H]-NMS. These data are inconsistent with those of Jacobson et al., 2010, where TBPB was shown to slow the off-rate of [³H]-NMS.⁴⁰ The most likely explanation for the discrepancy between our results and those of Jacobson et al., 2010 is a difference in our assay protocols. Jacobson et al. allow 1 hour for equilibrium to occur between atropine and [³H]-NMS at room temperature for their dissociation kinetic studies whereas we allow 3 hours at room temperature for this equilibrium to occur. Traditionally, when a 1 hour equilibrium is utilized for M₁ dissociation studies, the studies are performed at 37°C.⁴³ Incubation for only 1 hour at room temperature may not allow enough time for a proper equilibrium to be established between atropine and [³H]-NMS and thus Jacobson et al. may have performed their TBPB dissociation studies under non-equilibrium conditions.

While ‘molecular switches’ that modulate modes of pharmacology and subtype selectivity are common in Family C GPCRs (eg., mGlu receptors),⁴¹ we have rarely encountered them in Family A GPCRs. Within the mAChRs, we had previously only encountered ‘molecular switches’ that modulate mAChR subtype selectivity.^27,41 We recently reported on a chemically distinct series of M₁ allosteric agonists, represented by 11 and 12 (Fig. 4), where mutagenesis work indicated a novel allosteric binding site in the third extracellular loop of M₁.²⁶ Serendipitously, when preparing additional analogs around 12, a chemical vendor sent the incorrect starting material, leading to the synthesis of the alternative regioisomer of 12, compound 13.13 proved to be a potent antagonist of M₁ (IC₅₀ = 273 nM, pIC₅₀ = 6.32±0.11), as well as M₂-M₅, and detailed molecular pharmacology showed that 13 was an orthosteric mAChR antagonist. Note, the modulation from M₁ allosteric agonist to pan-mAChR orthosteric antagonist was due to major structural modifications (ie., 3° to 2° amine and 2° to 3° amide).

Figure 4.

Structures and activities of M₁ allosteric agonists 11 and 12, and a mAChR orthosteric antagonist 13 (regioisomer of 12).

These data, coupled with the results from TBPB and 9, led us to scrutinize additional analogs of 12, in an effort to discover more subtle ‘molecular switches’ that can modulate modes of pharmacology within the VU0364572 bi-topic ligand scaffold. One of our initial efforts led to the development of VU0419757 (14) which contains a 3-bromophenyl amide (Fig. 5) and proved to be a weak partial M₁ allosteric agonist due to the addition of a 3-bromophenyl amine (EC₅₀ = 2.0 μM, pEC₅₀ = 5.69±0.08, 37.4±8.7% ACh Max). Modification with an additional bromine atom to the 5-position afforded VU0430613 (15), an equipotent M₁ antagonist (IC₅₀ = 3.4 μM, pIC₅₀ = 5.47±0.03). Furthermore, when the aryl amides in 12 were replaced with heterocyclic amides, pan-mAChR antagonists resulted. Even when fluorine atoms were employed to attenuate the basicity of the heterocycles, M₁ allosteric agonism was lost. For example, a 6-trifluomethoxy nicotinamide derivative, VU0362058 (16), proved to be another pan-mAChR antagonist (Fig. 6), favoring inhibition of M₂ (IC₅₀ = 1.1 μM, pIC₅₀ = 5.97) over M₁ (IC₅₀ = 3.2 μM, pIC₅₀ = 5.50±0.06) and M₃-M₅ (IC ₅₀s >10 μM).

Figure 5.

Structures and activities of M₁ allosteric agonist 14 and a 3,5-dibromo congener 15, an M₁ antagonist.

Figure 6.

A) Structure of VU0362058 (16) and full M₁-M₅ CRCs, highlighting the pan-mAChR antagonism via a subtle modification to the M₁ allosteric agonist 12. B) Representative additional amide analogs of 12 (17) where the nature of substitutions on the aromatic ring conferred M₁ antagonism as opposed to M₁ agonism. While some SAR trends are evident, other switches in the mode of M₁ pharmacology are random and confound chemical optimization efforts.

In summary, we have explored additional SAR around two M₁ allosteric agonists TBPB and VU0364572 (12). Within the TBPB scaffold, either electronic or steric perturbations to the benzimidazolone piperidine ring led to a loss of selective M₁ allosteric agonism and afforded pan-mAChR orthosteric antagonists. Based on mutagenesis and modeling studies with TBPB, the electronic and steric perturbations disrupted a key H-bond with Thr83 in the allosteric site on the M₁ receptor. Similar subtle ‘molecular switches’ were found within the VU0364572 series, converting potent M₁ agonists into pan-mAChR antagonists. These findings, coupled with our recent studies showing brain-region-dependent and receptor reserve-dependent pharmacology of functionally selective M₁ bi-topic agonists⁴² show the challenges associated with development of selective M₁ agonists. Therefore, we propose that positive allosteric modulation of M₁ receptors, as opposed to allosteric agonism, may be the preferred mechanism by which to develop selective M₁ activators as therapeutic agents.