Continued optimization of the MLPCN probe ML071 into highly potent agonists of the hM₁ muscarinic acetylcholine receptor

Abstract

This Letter describes the continued optimization of the MLPCN probe molecule ML071. After introducing numerous cyclic constraints and novel substitutions throughout the parent structure, we produced a number of more highly potent agonists of the M₁ mACh receptor. While many novel agonists demonstrated a promising ability to increase soluble APPα release, further characterization indicated they may be functioning as bitopic agonists. These results and the implications of a bitopic mode of action are presented.

Continued efforts to develop potent and highly subtype-selective ligands for each of the five (M_1-5) muscarinic acetylcholine receptors (mAChRs) remains a challenging, but potentially highly rewarding endeavor. Tremendous clinical promise has been shown for muscarinic activators in the treatment of Alzheimer’s disease and schizophrenia with the moderately selective muscarinic agonist xanomeline.^1,2 In the absence of truly subtype-selective activators, it has not been definitively established which mAChRs are primarily responsible for the positive outcomes observed with xanomeline. While a spectrum of mAChR-knock out (KO) mice have strongly implicated the M₁ and M₄ receptors in mediating beneficial CNS effects,³ the study of selective mAChR ligands in non-mutant species and under various disease states remains a worthy goal. The ability to develop selective mAChR agonists is complicated by the highly conserved orthosteric binding site shared across all five receptor subtypes. As a result, classic orthosteric agonists, such as xanomeline, do not display the desired levels of receptor subtype selectivity. Recently, the focus of novel mAChR activators has moved away from the orthosteric site in favor of allosteric sites.⁴ These allosteric sites have the potential to be less conserved in their amino acid sequences, across the M_1-5 receptors, thereby giving rise to much higher levels of receptor subtype selectivity. Allosteric ligands can function as purely allosteric agonists, activating the receptor independent of acetylcholine (ACh), positive allosteric modulators (PAMs), which enhance the activity of endogenous ACh, or as a combination of these two modes (an ago-PAM). Considerable progress has been made in the development of highly selective M₁ PAMs,^5,6,7 M₄ PAMs,⁸ M₅ PAMs,⁹ and M₁ allosteric agonists.¹⁰ Most recently, we have described the development and characterization of VU0364572 (Figure 1) starting from the MLPCN Probe ML071.^11,12 Concurrently, we pursued an alternative optimization of ML071 through the introduction of numerous diverse constraints and now detail those efforts in this Letter.

Figure 1.

The MLPCN Probe molecule ML071 and the next-generation highly selective, orally bioavailable, and CNS penetrant agonist VU0364572.

Continued optimization of ML071 built on the highly modular and versatile nature of the synthesis routes utilized previously. Briefly, all analogs appearing in subsequent Tables and Figures were prepared according to the general routes appearing in Scheme 1, employing routine methodology as required for elaborating protecting groups into libraries of targets. Starting materials were either commercially available or prepared according to literature procedures. Route A depicts the reductive amination step involved when the Eastern piperidine-like portion of the final target arises from a ketone building block. Route B depicts the corresponding reaction between an aldehyde and the Eastern region already containing a primary or secondary amine.

Scheme 1.

General methods for the production of novel mAChR agonists. Reagents: (a) NaHB(OAc)₃, DCE or DCM, HOAc (0-4 equiv).

Given the demonstrated benefits associated with the introduction of the central ring present in VU0364572 (Figure 1),¹¹ we first explored other modes of limiting molecular flexibility (Figure 2). For the compounds appearing in Figure 2, ML071 and 1 were prepared via Route A (Scheme 1), while the remaining analogs (2-5) were prepared according to Route B. Assay descriptions and experimental protocols for the functional muscarinic assays have been described previously.¹² Compound 1 shows that a potency enhancement of roughly 3-fold could be achieved by introducing the tropane scaffold into the Eastern region. This was not surprising since these types of tropane structures are commonly associated with muscarinic ligands (e.g. atropine and scopolamine). However, this was an inseparable mixture of endo- and exo- isomers (roughly 70:30 endo:exo based on proton NMR). Although the four other compounds in Figure 2 were less active than 1, the diazadecalins 4 and 5, were viewed as an instructive pair of isomers. Relative to 1, the cis-fused bicycle 4 has its central nitrogen, and the point of attachment to the Western half of the molecule, displayed in an equatorial fashion which might mimic the minor exo-isomer of compound 1. Conversely, the trans-fused bicycle 5 has its central nitrogen positioned in an axial sense, thereby mimicking the major, endo-isomer component of 1. While the potencies of 4 and 5 are not drastically different, it was encouraging to see that the axial nitrogen point of attachment in 5, which more closely mimics the endo-tropane of 1, was preferred over the equatorial sense of attachment in 4. This implied that the endo-isomer component of 1 was likely the major contributor to its improved activity, and that further efforts to improve potency should focus on the production of endo-tropane isomers.

Figure 2.

Novel, bicyclic analogs of ML071.

One straightforward method for increasing the proportion of the endo-isomer relative to the exo-isomer when preparing analogs of 1 via Route A (Scheme 1), would be to perform the reductive amination step with secondary amines, thereby relying on the increased steric interactions to provide more of the desired endo-isomer. When the analogs appearing in Table 1 were prepared, an increased preference for the endo-isomer was observed with endo:exo isomer ratios now heavily favoring the endo-isomer. In the context of a central piperidine (Table 1, n = 2), this provided the corresponding tropane analogs of VU0364572. The preference for the (R)-stereochemistry, as seen previously for VU0364572,¹¹ was maintained and was most pronounced for the n = 2 analogs (compare 6a with 6c). Although the (S)-isomers 6a and 6b appear to display hM₁ activity, this could also result from contamination with the (R)-isomer arising from the commercially available 95% ee starting material. As the ring size decreased (n = 1) this stereochemical preference for the (R)-isomer decreased beyond what could be explained by trace impurities, while the high endo:exo ratios were maintained. The achiral azetidine analog 6i possessed a similar potency and efficacy as ML071.

Table 1.

Structures and activities of tropane analogs 6a-i.

Compd	R	*	n	hM1 EC50 (μM)a	AChmax (%)b
6a	Me	(S)	2	2.1	52
6b	H	(S)	2	>10	12
6c	Me	(R)	2	0.059	75
6d	H	(R)	2	0.84	75
6e	Me	(S)	1	>10	22
6f	H	(S)	1	1.2	39
6g	Me	(R)	1	1.2	55
6h	H	(R)	1	0.32	62
6i	H		0	2.2	34

^aAverage of at least three independent determinations performed in triplicate. Standard deviations ranging from 5-18% of the value reported.

^bWhen hMi EC5o is “>10”, the reported AChmax equals an average of the % ACh obtained when the compound was tested at the highest concentration (30 μM). Standard deviations ranging from 2 -11% of the value reported.

Compound 6c represented one of our most potent hM₁ agonists prepared to date and when administered to rats exhibited moderate PK properties. A standard rat IV (1 mg/kg)/PO (10 mg/kg) study found that 6c possessed a high CL of 189 mL/min/kg, but with a V_ss of 11.8 L/kg and a t_½ of 46 minutes. Additionally, 6c was orally bioavailable with a %F of 70 (AUC_IV = 92.2 ng•h/mL, AUC_PO = 639 ng•h/mL). Unfortunately, compound 6c was found to be lacking in its muscarinic subtype selectivity profile, along with 6d. Table 5 (vide infra) shows that while both 6c and 6d were potent hM₁ agonists, they showed activity at various hM_2-5 receptors.¹³ Although 6c was about 100-fold selective over hM₂, this level of selectivity and the appearance of some level of agonist activity across all muscarinic subtypes was not desirable. Whether the emergence of hM_2-5 activity resulted from structural modifications that moved these agonists back into the orthosteric binding site at the expense of the high-affinity allosteric site on hM₁ (bitopic agonists) or if these agonists are still allosteric ligands at hM₁ but not allosteric with respect to their binding at hM_2-5 has not yet been determined.

Table 5.

hM_1-5 selectivity profile for selected hM₁ agonists from an assay of calcium mobilization.

Compd	hM1 modea	EC50 or IC50(μM)b	%AChc	hM2 modea	EC50 or IC50 (μM)b	%AChc	hM3 modea	EC50 orIC50 (μM)b	%AChc	hM4 modea	EC50 or IC50 (μM)b	%AChc	hM5 modea	EC50 or IC50 (μM)b	%AChc
ML071	E	2.6	34	inactive			nactive			inactive			inactive
6c	E	0.017	59	E	1.8	66	E	>10	36	E	>10	28	E	3.7	62
6d	E	0.35	65	E	>10	27	inactive			inactive			E	>10	21
7	E	0.21	37	I	>10	15	I	>10	13	I	3.1	2	I	4.9	2
8	E	1.2	57	inactive			inactive			inactive			inactive
11a	E	0.018	46	inactive			I	>10	45	I	>10	48	E	2.8	31
11c	E	0.096	58	inactive				inactive		I	>10	40	E	4.2	22
12	E	0.27	40	I	2.7	1	I	2.2	0	I	0.70	3	I	4.6	3
13	E	0.041	81	I	>10	30	E	5.9	19	I	>10	26	E	1.0	28

^aMode of action in the calcium flux assay: E - represented an agonist response, I - represented an antagonist response and a blank represented a lack of clearly discernible activity at the receptor subtype up to the top dose of 30 μM, Experiments were single determinations conducted in triplicate, with routine error in the 1 - 25% range.

^bWhen the mode of action was that of an agonist (E) value is an EC₅₀, when an antagonist (I) value is an IC₅₀ and when an acceptable curve fit could not be generated, but a clear mode of action could be determined, “>10” is entered.

^cWhen the mode of action was that of an agonist %ACh represents the maximal activity observed as a percentage of the maximum response generated from 80% of a saturating dose of ACh. When the mode of action was that of an antagonist %ACh represents the remaining activity observed at the highest dose (30 μM) as a percentage of the maximum response generated from 80% of a saturating dose of ACh. When the IC₅₀ or EC₅₀ is “>10”, the reported % ACh equals an average of the % ACh obtained when the compound was tested at the highest concentration (30 μM).

To enable a more rigorous comparison of endo- versus exo-tropane isomers, each primary amine was individually prepared according to literature procedures¹⁴ and utilized via synthesis Route B (Scheme 1) to prepare pure endo-analog 7 and pure exo-analog 8 (Figure 3). As surmised for the isomer components of 1, the endo-isomer 7 was more potent than the exo-isomer 8. Interestingly, the less potent isomer 8 displayed a superior efficacy (%ACh_max = 49%), compared to the more potent isomer 7. When we determined the M_1-5 receptor subtype selectivity for these two compounds, the more potent endo-isomer 7 displayed antagonist activity across hM_2-5, while the exo-isomer appeared completely selective (Table 5). This sort of agonist/antagonist profile seen with 7 is similar to the profile seen with the M₁ allosteric agonist TBPB^10b (Table 5, compound 12). As for the high apparent selectivity demonstrated by 8, it remained to be seen if further optimization with respect to hM₁ potency would engender activity at the other muscarinic subtypes. Ultimately, we decided to move away from tropane analogs, where the point of attachment was displayed in an axial fashion, and focus on tropane analogs which position the central nitrogen in an equatorial manner, which might be envisioned as better approximating the lowest energy conformation of ML071.

Figure 3.

Comparison of the individually prepared endo- (7) and exo- (8) tropane isomers.

Viewing compound 8 as a new starting point for medicinal chemistry optimization, we began by exploring replacements for the ethyl carbamate terminus (R groups, Table 2). The structure activity relationship (SAR) in this region was remarkably steep. The methyl carbamate 9a lost greater than 10-fold potency, but remained an agonist. The only minor modification which appeared to be somewhat tolerated was the introduction of an allyl carbamate, although this substitution resulted in a 2-fold decrease in potency. The propyl- and butylamides (9c and 9d, respectively) were not tolerated and speak to the importance of having a carbamate at this location. Ureas (like 9h) and sulfonamides (9i and 9j) were similarly unproductive. Clearly the ethyl carbamate stood out as the premier moiety at this location.

Table 2.

Structures and activities of exo-substituted tropane analogs 8, 9a-j.

Compd	R	hM1 EC50 (μM)a	AChmax (%)b
8	-CO2Et	0.97	49
9a	-CO2Me	>10	21
9b	-CO2allyl	2.0	40
9c	-(CO)Et	-	-
9d	-(CO)Pr	-	-
9e	-(CO)/-Pr	5.4	23
9f	-CO2Bn	>10	7
9g	-CO2Ph	-	-
9h	-(CO)NMe2	-	-
9i	-SO2Me	-	-
9j	-SO2Et	-	-

^aAverage of at least three independent determinations performed in tri licate.“-” indicates an inactive compound (< 5% ACh_max) up to the highest concentration tested (30 μM). Standard deviations ranging from 10 -15% of the value reported.

^bWhen hMi EC5o is “>10”, the reported ACh_max equals an average of the % ACh obtained when the compound was tested at the highest concentration (30 μM). Standard deviations ranging from 3 - 21% of the value reported.

Concurrently we explored SAR around the Western end of compound 8 (Table 3). Among the substituted benzamides (10a-h), meta-substitution with a chlorine atom (10d) provided the only clear improvement over 8. Various urea termini were tolerated at this location; however, only 10k produced an increase in hM₁ potency. The two nitrogen-containing heterocycles 10o and 10p were not tolerated, whereas the thiophene analog 10q was equipotent with compound 8. A sulfonamide analog 10r was relatively inactive, as were a handful of other sulfonamides at this location (data and structures not shown), but showed a trend toward hM₁ antagonist activity. Although potencies were not improved to very low nanomolar levels with the analogs 10, we continued to explore alternate tropane scaffolds that positioned the nitrogen in an equatorial manner.

Table 3.

Structures and activities of exo-substituted tropane analogs 10a-r.

Compd	R	hM1 EC50 (μM)a	AChmax (%)b	Compd	R	hM1 EC50 (μM)a	AChmax (%)b
10a		1.1	47	10J		2.5	39
10b		2.3	37	10k		0.57	53
10c		-	-	101		>10	11
10d		0.44	63	10m		2.0	44
10e		3.6	33	10n		-	-
10f		1.9	43	10o		-	-
10g		0.83	54	10p		>10	6
10h		0.81	50	10q		1.0	33
10i		1.5	34	10r		-	-

^aAverage of at least three independent determinations performed in triplicate.“-” indicates an inactive compound (< 5% ACh_max) up to the highest concentration tested (30 μM). Standard deviations ranging from 2 -12% of the value reported.

^bWhen hM₁ EC₅₀ is “>10”, the reported ACh_max equals an average of the % ACh obtained when the compound was tested at the highest concentration (30 μM). Standard deviations ranging from 1 - 22% of the value reported.

Utilizing synthesis Route A (Scheme 1), and an alternate tropanone wherein the bridgehead was moved away from the ethyl carbamate, we easily obtained the analogs appearing in Table 4. Interestingly, the reductive amination step which sets the endo/exo-connectivity in 11 now afforded only the endo-product as a result of the bridgehead being closer to the reaction center thereby enhancing its steric influence. Furthermore, since the bridgehead had moved to the opposite side of the piperidine ring (compare 10 with 11) we could now directly obtain the desired equatorial orientation of the central nitrogen.

Table 4.

Structures and activities of endo-substituted tropane analogs 11a-j.

Compd	R	hM1 EC50 (μM)a	AChmax (%)b	Compd	R	hM1 EC50 (μM)a	AChmax (%)b
11a		0.092	53	11f		1.0	37
11b		0.17	51	11g		0.69	9
11c		0.047	52	11h		3.6	28
11d		0.24	47	11i		>10	11
11e		0.20	53	11J		0.17	54

^aAverage of at least three independent determinations performed in triplicate. Standard deviations ranging from 3 - 23% of the value reported.

^bWhen hMi EC₅₀ is “>10”, the reported ACh_max equals an average of the % ACh obtained when the compound was tested at the highest concentration (30 μM). Standard deviations ranging from 2 -10% of the value reported.

This change in tropane scaffold afforded an almost uniform increase in potency for analogs 11 compared to the corresponding tropanes 10. Only a handful of analogs were prepared, but even with this small sampling of R-groups, hM₁ potencies were now in the low nanomolar range for compounds 11a and 11c. Interestingly, when central piperidines analogous to compounds 6a-d were introduced into this alternate tropane scaffold, only weakly potent agonists or wholly inactive compounds were obtained (data and structures not shown).

With another series of highly potent hM₁ partial agonists in hand, the paramount question of receptor subtype selectivity was quickly explored. Unfortunately, as shown in Table 5, both compounds displayed off target activity at subsets of the hM_3-5 receptors. Compound 11a had mixed activity as an antagonist at hM₃ and hM₄, but functioned as a partial agonist at hM₅. Similarly, 11c was an antagonist at hM₄, but a partial agonist at hM₅. We have observed this mixed profile with other hM₁ allosteric agonists in our functional calcium mobilization assays, in particular AC260584 (Compound 13).¹⁵

Having optimized hM₁ potency in three related series of tropane partial agonists, we tested select compounds for their ability to enhance the release of soluble APPα (sAPPα),¹⁶ as has been demonstrated with previous M₁ agonists¹¹ and PAMs.⁷ Gratifyingly, all five of the tropane agonists tested (6c, 6d, 7, 8 and 11a, Figure 4) were able to stimulate the release of sAPPα in TREx293-hM₁ cells to the same extent as 10 μM carbachol (CCh) and 2 μM ML071. These experiments continue to support the belief that activation of M₁ may have the potential to be disease modifying for the treatment of Alzheimer’s disease.

Figure 4.

Ability of hM₁ partial agonists to stimulate nonamyloidogenic sAPPα release in TREx293-hM₁ cells. Agonists were tested at 2 μM and the results presented were from three independent experiments.

Although these tropane agonists were very attractive from an M₁ potency aspect, the loss of selectivity accompanied by their increase in M₁ potency brings into focus a potential weakness within this series of hM₁ bitopic agonists. Although the highly selective interaction with the M₁ receptor may be allosteric at low concentrations due to a high-affinity allosteric site, the bitopic nature of these agonists may result in orthosteric interactions at higher concentrations or at the M_2-5 receptors as potency is improved at the M₁ receptor. As such, a potentially more tractable approach toward the selective activation of each individual muscarinic subtype may be realized with subtype selective PAMs. Detailed in vitro pharmacological studies around this group of bitopic¹⁷ partial agonists are in progress and will be reported shortly.¹⁸

In summary, we have expanded the SAR surrounding ML071, resulting in the development of highly potent hM₁ partial agonists containing two different tropane scaffolds. However, as potency was increased within each of the three series, off-target activity at the other muscarinic receptors surfaced. This may be related to the bitopic nature of these weak partial agonists.