Specific AIM

To identify small molecule positive allosteric modulators (PAMs) and/or allosteric agonists of the M₁ muscarinic acetylcholine receptor that are cell permeable, possess submicromolar potency and show greater than 10-fold selectivity over the other mAChRs (M₂–M₅) employing a functional HTS approach. Out of this effort aimed at M₁, which afforded a highly selective M₁ antagonist (CID 24768606), a highly selective M₁ allosteric agonist (CID 25010775) and a highly selective M1 PAM (CID 44251556), we also identified and optimized the first M₅ ligand, an M₅ PAM (CID 42633508). Starting from an unrelated lead, belonging to a novel chemotype, we have now been able to develop a second highly selective M₁ PAM with a low-micromolar EC₅₀ and a greatly improved ACh fold-shift. Another MLSCN screening effort identified a highly selective M₄ PAM (CID 864492); thus, two MLSCN/MLPCN screens have provided a toolkit of highly selective mAChR ligands available from the MLPCN to study individual mAChR function both in vitro and in vivo.

Significance

The five cloned muscarinic acetylcholine receptor subtypes (mAChR1-5 or M₁–M₅) are known to play highly important and diverse roles in many basic physiological processes.1–3 Correspondingly, muscarinic agonists and antagonists targeting one or more subtypes have been used preclinically and clinically for research and treatment of a wide range of pathologies.3,4 Based on the high sequence homology of the mAChRs across subtypes, and particularly within the orthosteric acetylcholine (ACh) binding site, discovery of truly subtype-selective compounds has proven historically difficult. Due to the scarcity of selective compounds, a detailed understanding of the precise roles of each subtype in neurobiology and in various central nervous system (CNS) disorders has thus remained elusive.3,4 In numerous Phase II and III clinical trials, pan-mAChR agonists were shown to improve cognitive performance in AD patients, but the GI-and/or cardiovascular side effects, resulting from activation of peripheral mAChRs, were deemed intolerable and the trials were discontinued.5,6 Importantly, several pan-mAChR agonists demonstrated a decline in the concentration of Aβ42 in the cerebral spinal fluid of AD patients, suggesting that mAChR activation has the potential to be disease modifying as well as providing palliative cognitive therapy.7 More recent studies in 3xTg-AD mice further support a disease modifying role for mAChR activation, and several Ph III trials demonstrated that mAChR activation lowered Aβ42 in patients.8 Interestingly, the M₁/M₄ preferring xanomeline, in addition to improving cognitive performance, had robust therapeutic effects on the psychotic symptoms and behavioral disturbances associated with AD and recently published clinical trial data indicates efficacy in schizophrenic patients.9,10 Probes developed from these efforts will greatly advance the current state of the art by aiding in the understanding of M₁’s role in cell-based physiology and may extend the clinical understanding of psychotic and cognitive symptoms associated with neurodegenerative disorders like Alzheimer’s Disease and schizophrenia.

Rationale

In recent years, major advances have been made in the discovery of highly selective agonists of other G protein-coupled receptors (GPCRs) that act at an allosteric site rather than the orthosteric binding site.11,12 By screening for compounds that act at an allosteric site on the receptor, it is anticipated that compounds that selectively activate M₁ versus the other muscarinic subtypes may be identified.13–18 While allosteric M₁ agonists have been identified, AC-42 and TBPB, they both suffer from undesirable ancillary pharmacology, poor physicochemical properties, poor pharmacokinetics and/or limited CNS exposure.19,20 Thus, to truly enable the biomedical community to dissect the relative contributions of selective M₁ activation in preclinical models of AD and schizophrenia and to understand the role of M₁ in the pronounced efficacy of the M₁/M₄ preferring xanomeline, improved M₁ probes are required. Recently, a number of novel highly subtype-selective allosteric ligands for M₁ and M₄ have emerged from functional cell-based screening efforts – several are MLPCN probes along with the prototypical M₁ PAM, BQCA.13–18,20–22 Although considerable interest was initially generated around BQCA, this level of interest appears to have waned and may point to a fatal flaw in its general chemotype (See Figure 2). Our initial report on the discovery of CID 3008304, a pan G_q M₁, M₃, M₅ PAM, also described three other series of weak M₁ PAMs, and established that different M₁ PAM chemotypes displayed different modes of activity on downstream receptor signaling/trafficking despite similar profiles in Ca²⁺ assays.13 Thus, all allosteric M₁ activation is not equivalent, and additional tool compounds representing diverse chemotypes are required to truly dissect and study M₁ function in the CNS. Subsequent to our development of an M₅ selective PAM from a pan G_q M₁, M₃, M₅ PAM,21,22 we next optimized CID 3008304 for M₁ PAM activity resulting in the first highly selective M₁ PAM MLCPN probe (CID 44251556). For the above reasons associated with downstream receptor signaling/trafficking, we were simultaneously pursuing alternative leads in an attempt to add unique chemotypes to our tool kit of selective M₁ activators.23

Figure 2

M₁ PAMs, CID 44251556 and BQCA.

PubChem Bioassay Name

Discovery of Novel Allosteric Modulators of the M₁ Muscarinic Receptor: Positive Allosteric Modulator (PAM)

List of PubChem bioassay identifiers generated for this screening project (AIDs)

AID-2651, AID-2425, AID-2428, AID-2430, AID-2434, AID-2433, AID-2438, AID-2626, and AID-2543.

PubChem Primary Assay Description

Chinese hamster ovary (CHO K1) cells stably expressing rat (r)M₁ were purchased from the American Type Culture Collection (ATCC, Manassas, VA) and cultured according to their recommendations. CHO cells stably expressing human (h) M₂, hM₃, and hM₅ were generously provided by A. Levey (Emory University, Atlanta, GA); rM₄ cDNA provided by T. I. Bonner (National Institutes of Health, Bethesda, MD) was used to stably transfect CHO-K1 cells purchased from the ATCC using Lipofectamine 2000. To make stable hM₂ and rM₄ cell lines for use in calcium mobilization assays, cell lines were cotransfected with a chimeric G protein (G_qi5) using Lipofectamine 2000. hM₂, hM₃, and hM₅ cells were grown in Ham’s F-12 medium containing 10% heat-inactivated fetal bovine serum, 2 mM GlutaMax I, 20 mM HEPES, and 50 μg/mL G418 sulfate. hM₂-G_qi5 cells were grown in the same medium supplemented with 500 μg/mL hygromycin B. Stable rM₄ cells were grown in Dulbecco’s modified Eagle’s medium containing 10% heat-inactivated fetal bovine serum, 2 mM GlutaMax I, 1 mM sodium pyruvate, 0.1 mM nonessential amino acids, 20 mM HEPES, and 400 μg/mL G418 sulfate; rM₄-G_qi5 cells were grown in the same medium supplemented with 500 μg/mL hygromycin B. CHO cells stably expressing rM₁, hM₃, or hM₅ were plated at a seeding density of 50,000 cells/100 μL/well. CHO cells stably coexpressing hM₂/G_qi5 and rM₄/G_qi5 were plated at a seeding density of 60,000 cells/100 μL/well. For calcium mobilization, cells were incubated in antibiotic-free medium overnight at 37 °C/5% CO₂ and assayed the next day.

Calcium Mobilization Assay

Cells were loaded with calcium indicator dye [2 μM Fluo-4 acetoxymethyl ester (50 μL/well) prepared as a stock in DMSO and mixed in a 1:1 ratio with 10% Pluronic acid F-127 in assay buffer (1xHanks’ balanced salt solution supplemented with 20 mM HEPES and 2.5 mM probenecid, pH 7.4)] for 45 min at 37 °C. Dye was removed and replaced with the appropriate volume of assay buffer. All compounds were serially diluted in assay buffer for a final 2x stock in 0.6% DMSO. This stock was then added to the assay plate for a final DMSO concentration of 0.3%. Acetylcholine (EC₂₀ concentration or full dose-response curve) was prepared at a 10x stock solution in assay buffer before addition to assay plates. Calcium mobilization was measured at 25 °C using a FLEXstation II (Molecular Devices, Sunnyvale, CA). Cells were preincubated with test compound (or vehicle) for 1.5 min before the addition of the agonist, acetylcholine. Cells were then stimulated for 50 s with a submaximal concentration (EC₂₀) or a full dose-response curve of acetylcholine. The signal amplitude was first normalized to baseline and then as a percentage of the maximal response to acetylcholine.

Summary of Screen

This screen was performed in the pilot phase, the MLSCN, when the MLSMR compound collection at Vanderbilt only contained ~65,000 compounds. Results from the primary M₁ screen of these compounds identified ~12 putative M₁ PAMs with an average Z’ score of 0.70±0.09. The confirmation screen (singles at 10 μM) produced two lead compounds, one of which was optimized into our first M₁ PAM probe (CID 44251556). The other, CID 2157678, represented a viable lead structure already endowed with good receptor subtype selectivity across the M₃/M₅ receptors (Figure 1). However, its potency upon reconfirmation from dry solid (M₁ EC₅₀ = 12.9 μM) was not particularly attractive and so a program was initiated to develop this lead into a pharmacologically more useful tool by improving its potency while maintaining its M₂–M₅ receptor subtype selectivity. Ultimately, it was envisioned that an M₁ PAM not structurally related to the known M₁ PAMs (CID 44251556 and BQCA, Figure 2) may possess a distinct pharmacology at the M₁ receptor, thereby enabling an expanded understanding of the M₁ muscarinic receptor’s downstream signaling/trafficking.

Figure 1

CRCs at M₁–M₅ for HTS lead CID 2157678.

Probe Chemical Lead Optimization Strategy

Our initial optimization strategy is outlined in Figure 3, and as SAR with allosteric ligands is often shallow, we employed an iterative parallel synthesis approach, along with targeted syntheses for structures encompassing more speculative modifications. Attempted modifications of the Eastern oxazole-amide, although not extensive, met with no success, returning only compounds with undetectable activity (Such as CID 44129586, 44634499, 44634501 and 44634503). In a straightforward attempt to reduce molecular weight the benzyl group attached to the indole nitrogen was omitted, but met with a similar lack of success (CID 751482, EC₅₀ > 10 μM).

Figure 3

Initial optimization strategy for CID 2157678.

Accordingly, libraries were concomitantly prepared as shown in Scheme 1 and predominately surveyed diversity on the Southern benzyl moiety. Methyl thioglycolate (1) was arylated with indole using iodine and potassium iodide, followed by saponification of the ester using lithium hydroxide to give acid 2. This acid was then peptide coupled with 3-amino-5-methyl-isoxazole (3) employing PyClU (chlorodipyrrolidinocarbenium hexafluorophosphate), in DCE, with microwave heating to yield thioether 4. The sulfur was then oxidized to the sulfone with Oxone, thereby allowing the benzylation of the indole nitrogen to be performed as the final step in the preparation of the initial library (Analogs 5, Table 1). As would be expected for this benzylation, bis-alkylation was observed to a small extent, and occasionally allowed for the isolation of analogs like CID 44129599 (Figure 4). While it was not terribly surprising that these analogs were devoid of M₁ PAM activity, it would eventually be ascertained that even the introduction of a solitary methyl group α to the amide abolished any measureable M₁ PAM activity (CID 44634500, M₁ EC₅₀ > 10 μM, is the μ-methylated analog of the probe compound). Alternatively, if the Oxone oxidation step is omitted or replaced with a milder oxidant (e.g. FeCl₃/H₅IO₆) then thioethers (CID 3305286) or sulfoxides (CID 44247543) analogous to the lead structure were produced. However only the sulfoxide (CID 44247543) retained measureable activity (M₁ EC₅₀ = 5.23 μM), but did not represent a significant improvement over its sulfone analog (5b, CID 44129591, Table 1).

Scheme 1

Synthesis of CID 2157678 and the route to its analogs.

Table 1

Structures and activities of analogs 5.

Figure 4

A representative over-alkylation product from Scheme 1.

Focusing on the library of analogs appearing in Table 1, we quickly discovered that substitution in the 3-position of the phenyl ring was preferred. Moving the chlorine from the 2-position in the lead structure (5a) to the 3-position (5b) improved the M₁ EC₅₀ by about 50%. While, either a methoxy (5e) or a fluorine (5f) at this position afforded a small improvement in their EC₅₀ values, a 2-fold increase in potency could be obtained by introducing a bromine at the 3-position providing 5h, which possessed an EC₅₀ = 3.79 μM. The remainder of Table 1, servers to illustrated how steep the SAR was for this class of allosteric modulators. For example, the addition of a second halogen in examples 5k and 5l eliminated the measurable efficacy displayed by their mono-halogenated analogs (5h and 5b, respectively). In general it could also be concluded that substitution at the 4-position on these aryl groups was uniformly detrimental. Having learned from the initial library that proper substitution at the 3-position of the benzyl group could be beneficial, and with the bromide 5h in hand we sought to introduce various aryl rings at this location. The small library of analogs 6 appearing in Table 2 were prepared via microwave-assisted Suzuki couplings between 5h and the requisite boronic acid. Impetus for the introduction of the pyrazole moiety found in 6a–c, came from its successful incorporation into our previous M₁ PAM probe (CID 44251556) and BQCA analogs (Figure 2). Gratifyingly, the N-methyl pyrazole 6a did have measureable activity but did not impart a significant improvement in potency over the corresponding bromide (5h). Divergent from the SAR established during the development of CID 44251556, the unalkylated pyrazole 6b was now preferred, possessing an M₁ EC₅₀ = 2.19 μM. In contrast to the N-methyl pyrazole 6a, the larger sec-butyl substituent of 6c was not tolerated, and neither were the other heterocycles of the remaining analogs (6d–h).

Table 2

Structures and activities of analogs 6.

To further the development of this novel series of M₁ PAMs we focused on three of the more potent analogs (5b, 5h and 6b) and applied a fine-tuning process of introducing fluorine atoms at various locations to provide analogs 7 (Table 3). Across the series, substitution at the 4-position was uniformly not tolerated (7a–c), consistent with the SAR appearing in Table 1. Bis-fluorination of the indole ring eroded activity in the context of the bromine analog 7d but conversely augmented the activity of the pyrazole congener 7e. This type of subtle/confounding SAR was similarly observed with respect to fluorination at the R² position in analogs 7f–h. While the presence of a fluorine at R² was neutral or slightly beneficial in the context of the chlorine analog (7f), its presence resulted in clearly decreased activity for both the bromine and pyrazole analogs, 7g and 7h. Lastly, the introduction of a single fluorine at the R⁶ position could either be moderately detrimental in the context of pyrazole 7i or decidedly beneficial with respect to bromine analog 7j, where an almost 3-fold improvement in potency was observed. In this manner, both CID 44475955 (7j) and the related difluorindole analog CID 44475948 (7e) were developed and chosen for further evaluation (For clarity, these compounds are shown in Figure 5).

Table 3

Structures and activities of analogs 7.

Figure 5

The two most potent M₁ PAMs developed from the lead CID 2157678.

Both CID 44475955 and CID 44475948 were highly selective PAMs for the M₁ receptor, displaying minimal/no potentiation of the M₂–M₅ receptors up to 30 μM (Figure 6A–B). Similarly, both compounds demonstrated impressive left-ward shifts in the potency of ACh in M₁ ACh Concentration Response Curve (CRC) fold-shift experiments (Figure 6C). When tested at 30 μM, the bromine analog 7j engendered a 45-fold increase in ACh activity, while the pyrazole analog 7e increased ACh activity 98-fold. These levels of potentiation are very similar to those seen for BQCA14,15 and represent a sizable improvement over the earlier M₁ PAM probe molecule CID 44251556 which produced only a 3-fold shift in an identical experiment. Still, these experiments did not reveal a definitive superiority for either molecule.

Figure 6

A) CRCs for CID 44475955 at M₁–M₅; B) CRCs for CID 44475948 at M₁–M₅; C) Fold-Shift for CID 44475955 and CID 44475948 at 30 μM.

An examination of calculated physical properties and practical considerations were more helpful in choosing the preferred probe. The in silico values for CID 44475955 and CID 44475948 appearing in Table 4 were calculated using TRIPOS software. Also included in Table 4 are the averages from the MDDR database of compounds both entering Phase I and launched drugs. Once again both compounds were very similar across a number of parameters. However, CID 44475955 possessed superior values for both total polar surface area (TPSA) and its number of hydrogen bond donors (Hdon). These two discrepancies both point to a lower probability of crossing the blood brain barrier for CID 44475948. Although not tested in vivo, this would lead to the prediction that CID 44475955 has a better chance of being a CNS penetrant molecule, and therefore a potentially superior MLPCN probe compound for the muscarinic receptors present in the CNS. Additionally, from a synthesis standpoint, the greater ease and lower cost for the acquisition of indole (Scheme 1) over 4,6-difluoroindole (the starting material for CID 44475948) makes CID 44475955 the more practical and cost-effective probe molecule.

Table 4

Calculated Property Comparison with MDDR Compounds.

To more fully characterize this novel M₁ PAM, CID 44475995 was tested at Ricerca’s (formerly MDS Pharma’s) Lead Profiling Screen (binding assay panel of 68 GPCRs, ion channels and transporters screened at 10 μM), and was, thus far, found to not significantly interact with 16 out of the 16 assays conducted (Remaining 52 pending).24 Thus, CID 44475955 is highly selective and can be used to dissect the role of M₁ in vitro. Furthermore, when used in conjunction with our earlier M₁ PAM probe molecule, the potentially different effects caused by an ACh low fold-shift probe (CID 44251556, 3-fold) versus those of an ACh high fold-shift probe (CID 44475955, 45-fold) may reveal the importance of this parameter towards the treatment of different disease states.

In summary, from an initial MLSCN screening campaign of just 65,000 compounds a second, potent and highly selective M₁ PAM (CID 44475955) has been identified which is structurally distinct from our initial M₁ PAM probe (CID 44251556). CID 44475955, similar to our previous probe, possesses comparable potency to BQCA, but now in contrast to our first probe, also shares BQCA’s ability to produce a large left-ward shift in the activity of ACh at the M₁ receptor. CID 44475955 now represents the third known chemotype to provide potent and selective M₁ positive allosteric modulation. Further in vitro and in vivo characterization of CID 44475955 is in progress in the assay submitter’s lab, and data will be reported in due course.

Synthetic procedure and spectral data for CID 44475955

MLS#s

002700179 (Probe, 500 mg), 002700180, 002700181, 002700182, 002700183, 002700184

Solubility

Solubility in PBS (at pH = 7.4) for ML169 was 0.08 μM.

Stability

Stability (at room temperature = 23 °C) for ML169 in PBS (no antioxidants or other protectorants and DMSO concentration below 0.1%) is shown in the table below. After 48 hours, the percent of parent compound remaining was not reported, but the assay variability over the course of the experiment ranged from a low of 89% (at 15 minutes) to a high of 142% (at 1 hour).

Reactivity

As assessed through a glutathione (GSH) trapping experiment in phosphate buffered saline (with a substrate concentration of typically 5–50 μM and a GSH concentration of 5 mM, at t = 60 minutes) ML169 was found to not form any detectable GSH adducts.^*