Discovery of structurally distinct tricyclic M₄ positive allosteric modulator (PAM) chemotypes - Part 2

Abstract

This Letter details our efforts to develop novel tricyclic M₄ PAM scaffolds with improved pharmacological properties. This endeavor involved a “tie-back” strategy to replace the 3-amino-4,6-dimethylthieno[2,3-b]pyridine-2-carboxamide core which lead to the discovery of two novel tricyclic cores: a 7,9-dimethylpyrido[3’,2’:4,5]thieno[3,2-d]pyrimidine core and 2,4-dimethylthieno[2,3-b:5,4-c’]dipyridine core. Both tricyclic cores displayed low nanomolar potency against the human M₄ receptor.

Graphical Abstract

Muscarinic acetylcholine receptor subtype 4 (M₄) positive allosteric modulators (PAMs) continue to be important drug targets as novel treatments for various neurological disorders such as Parkinson’s disease,¹ Huntington’s disease,² and schizophrenia (both the positive and negative symptom clusters).^3–7 Classically, M₄ PAMs possess a β-amino carboxamide moiety as a key pharmacophore (circled in 2, Figure 1) which was believed to be essential for M₄ PAM activity.^8–14 As such chemotypes have been plagued with poor solubility, varying degrees of P-gp efflux, and potency differences across species, efforts have been made to develop M₄ PAM chemotypes devoid of the β-amino carboxamide moiety.^3,14–26 Clinical studies featuring xanomeline, a M₁/M₄ preferring agonists that lacks the β -amino carboxamide moiety, have validated targeting the muscarinic cholinergic system as a method for treating the psychosis and behavioral disturbances observed in both Alzeheimer’s and schizophrenia patients.^23,24 However, the lack of receptor subtype selectivity resulted in undesired side effects which ultimately lead to a discontinuation of clinical development. An effort to circumvent these adverse events led to the development of KarXT which is currently undergoing clinical trials. KarXT is a treatment in which xanomeline is co-administered with a pan-selective peripheral muscarinic acetylcholine receptor antagonist (trospium) to minimize the adverse side effects of xanomeline.²⁵ Additional work in the field led to the identification of a selective M₄ PAM which lacked the β-amino carboxamide moiety. Not only was this PAM efficacious in preclinical assays but had fewer and less severe adverse cholinergic-related side effects than observed in rats treated with the nonselective M₄ agonist xanomeline.²⁶ This indicates that receptor-subtype-selective M₄ PAMs could potentially improving safety profiles. In fact, CVL-231 (a selective M₄ PAM) is currently undergoing clinical trials.^27,28

Figure 1.

Exploration of novel tricyclic cores as M₄ PAMs. We utilized the same “tie-back” strategy to mask the β-amino carboxamide moiety (dotted circle), which contributed to the discovery of VU6017654 (3), to develop two unique M₄ PAM tricyclic chemotypes: a 7,9-dimethylpyrido[3’,2’:4,5]thieno[3,2-d]pyrimidine core (5) and a 2,4-dimethylthieno[2,3-b:5,4-c’]dipyridine core (6).

Our laboratory previously described two structurally distinct 5,6,6-tricyclic scaffolds devoid of the β-amino carboxamide moiety that afforded potent and CNS penetrant M₄ PAMs.¹⁷ In an effort to identify additional novel M₄ PAM chemotypes, we employed a “tie-back” strategy in order to mask the β-amino carboxamide moiety of one of our early lead M₄ PAM compounds, VU0152100.^10–11 When designing this new scaffold, we elected to keep the 4,6-dimethylthieno[2,3-b]pyridine core intact while masking the β-amino carboxamide as a pyrimidine moiety, a strategy utilized in the development of VU6017654 (3). This resulted in the discovery of a novel M₄ PAM chemotype containing a 7,9-dimethylpyrido[3’,2’:4,5]thieno[3,2-d]pyrimidine core, 5. Further exploration revealed a second novel, tricyclic M₄ PAM chemotype containing a 2,4-dimethylthieno[2,3-b:5,4-c’]-dipyridine core, 6. This body of work details the development of these two novel M₄ PAM chemotypes.

The synthesis of core 5 began with the commercially available amine 7, which was first condensed with formamide followed by formamidine acetate salt to afford the pyrimidone intermediate 9 (Scheme 1). Treatment of pyrimidone 9 with POCl₃ gave chloride 11 which readily underwent nucleophilic aromatic substitution with a variety of primary and secondary amines as well as alcohols to yield desired analogs 16 and 19. For this exercise, we chose to evaluate benzyl amines and amino azetidines that proved to be successful in our past endeavors to afford potent M₄ PAMs.^15–17,29

Scheme 1.

Synthesis of M₄ PAM analogs. Reagents and conditions: (a) i. HCONH₂, 150 °C; ii. AcOH • HN=CHNH₂, 150 °C; (b) POCl₃, 120 °C, 18h; (c) i. CuBr₂, ^tBuONO, CH₃CN, 60 °C, 56%; ii. Cs₂CO₃, Pd(dppf)Cl₂, dioxane/H₂O (10:1), 90 °C, 18h, 99%; (d) i. TFA, 100 °C, 4h, 98%; ii. NH₄OH, 100 °C, 48h, 98%; (e) amine, DIEA, NMP, 120 °C, 21–88%; (f) i. Methyl (4-hydroxymethyl)benzoate, NaH, THF, 0 °C, 17%; ii. MeMgBr, THF, 0 °C, 38%; (g) 20, Pd(PPh₃)₄, 1,4-dioxane, μW 120 °C, 0.5h, 10%.

Select analogs 16 & 19 were screened against the human M₄ (hM₄) receptor to determine potency with results highlighted in Table 1. It became very clear that the amine “tail” greatly influences potency. It was noted that minor changes, such as a fluoro-substitution on the phenyl ring (16m vs. 16n), led to a ~2.5-fold decrease in potency. Moreover, removal of one fluorine from the trifluoromethoxy group of 16d (inactive) afforded analog 16e (EC₅₀ = 4 μM), which reestablished activity at hM₄. Another intriguing finding was observed when the amine linker (16m; hM₄ EC₅₀ = 910 nM) was exchanged for an ether linker (19; hM₄ EC₅₀ = 8.7 μM) resulting in a 9.6-fold loss of potency.

Table 1.

Structures and activities for analogs 16 & 19.

Cmpd	R Group	hM4 EC50 (nM)a [%ACh Max]
16a		Inactive
VU6008258
16b		Inactive
VU6008243
16c		Inactive
VU6008261
16d		Inactive
VU6008266
16e		4,010
VU6008279		[45]
16f		1,560
VU6008248		[34]
16g		1,110
VU6008250		[47]
16h		Inactive
VU6008255
16i		2,560
VU6008257		[30]
16j		1,140
VU6008256		[41]
16k		>10,000
VU6008254		[30]
16l		1,400
VU6008283		[54]
16m		910
VU6008284		[76]
16n		2,320
VU6010076		[74]
16o		293
VU6008280		[42]
16p		6,550
VU6008452		[52]
16q		341
VU6007221		[67]
16r		1,640
VU6008278		[61]
19		8,740
VU6010153		[40]

^aCalcium mobilization assays with hM_4/Gqi5-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine, n =1 independent experiment in triplicate.

Noting the importance of the amine “tail”, we next elected to explore small aliphatic amines (Table 2) as well as small carbon-linked groups (Table 3) which were synthesized in accordance with Scheme 2. Briefly, commercially available 2-mercapto-4,6-dimethyl nicotinonitrile (22) was condensed with various α-bromo ketones in a Gewald-type reaction to afford thieno[2,3-b] pyridines 23. As previously described in Scheme 1, condensation with formamide afforded final compounds 24. The piperidine intermediate 24e could be further transformed via nucleophilic aromatic substitutions (26), HATU-assisted amide formation (28), or reductive amination (30). Finally, the synthesis of biaryl analog 21 was accomplished utilizing intermediate 11 in a Stille cross-coupling reaction as described in Scheme 1.

Table 2.

Structures and activities for analog 16 containing small aliphatic amines.

Cmpd	R Group	hM4 EC50 (nM)a [%ACh Max]
16s		762
VU6007214		[50]
16t		456
VU6007218		[69]
16u		453
VU6007216		[75]
16v		359
VU6007220		[76]
16w		277
VU6007219		[67]
16x		172
VU6007217		[59]
16y		144
VU6007215		[84]

^aCalcium mobilization assays with hM_4/Gqi5-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine, n =1 experiment performed in in triplicate.

Table 3.

Structures and activities for C-linked analogs 21, 26, 28, & 30.

Cmpd	R Group	hM4 EC50 (nM)a [%ACh Max]
24a		6,790
VU6008673		[71]
24b		3,900
VU6008479		[62]
24c	--CH3	2,820
VU6008362		[58]
24d		152
VU6008676		[99]
24e		6,110
VU6009094		[44]
28a		620
VU6009098		[62]
28b		234
VU6009100		[72]
30		Inactive
VU6009198
26		Inactive
VU6009197
21		3,410
VU6007227		[62]

^aCalcium mobilization assays with hM_4/Gqi5-CHO cells or performed in the presence of an EC₂₀ fixed concentration of acetylcholine, n =1 independent experiment performed in triplicate.

Scheme 2.

Synthesis of M₄ PAM analogs. Reagents and conditions: (a) 22, KOH, IPA, 105 °C, 10–74%; (b) i. HCONH₂, 150 °C; ii. AcOH • HN=CHNH₂, 150 °C, 62 – 95%; (c) HCl, 1,4-dioxane/H₂O (1:1), 79%; (d) 25, DIEA, NMP, 120 °C, 72h, 92%; (e) 27, HATU, DIEA, DMF, 54–62%; (f) 29, STAB. AcOH, DCM, 18h, 95%.

Select analogs 16s-y, 21, 24, 26, 28, and 30 were screened against human M₄ (hM₄) to determine potency with results highlighted in Tables 2 & 3. Selecting small amines as our “tail” groups (16s-y) proved most advantageous as the majority of analogs in Table 2 possess hM₄ PAM functional potencies less than 500 nM, with three analogs displaying EC₅₀’s < 300 nM: 16w (hM₄ EC₅₀ = 277 nM), 16x (hM₄ EC₅₀ = 172 nM), and 16y (hM₄ EC₅₀ = 144 nM).

Conversely, similar small carbon-linked tail groups resulted in a loss of hM₄ functional potency (Table 3): 21 (hM₄ EC₅₀ = 3.4 μM), 24a (hM₄ EC₅₀ = 6.8 μM), 24b (hM₄ EC₅₀ = 3.9 μM), 24c (hM₄ EC₅₀ = 2.8 μM). Most notably, the addition of a hydrogen bond acceptor into the cyclohexyl ring of analog 24a to give the tetrahydropyran analog 24d (hM₄ EC₅₀ = 152 nM), resulted in a nearly 45-fold increase in hM₄ functional potency. Comparison of 24d to the corresponding piperidine analog 24e (hM₄ EC₅₀ = 6.1 μM) once again resulted in a loss of hM₄ functional potency (~40-fold). To investigate the importance of a hydrogen bond acceptor at this position, we further investigated the piperidine tail by eliminating its ability to serve as a hydrogen bond donor. This exercise generated amides (28a & 28b), small aliphatic C-linked piperidines (30), as well as aryl-linked piperidines (26). Interestingly, converting piperidine 24e into amides 28a (hM₄ EC₅₀ = 620 nM) and 28b (hM₄ EC₅₀ = 234 nM) led to an increase in functional M₄ potency by ~10-fold and 26-fold, respectively. Removing the carbonyl of analog 28a to afford analog 30 resulted in a complete loss of M₄ potency, further supporting our theory of the importance of a hydrogen bond acceptor at this position. Similarly, a complete loss of activity was observed when the piperidine tail of 24e was capped with 3-fluoropyridine (26). This loss of activity could be attributed to several factors including the large size of the C-linked “tail”, the position/orientation of the electron lone pair of the pyridine, and/or the effects of the fluorine substituent on the basicity of the pyridine.

After extensive exploration of the amine tail, we turned our attention toward accessing the importance of the sulfur of the 7,9-dimethylpyrido[3’,2’:4,5]thieno[3,2-d]pyrimidine core. To do so, we exchanged the sulfur with an oxygen to generate a 7,9-dimethylpyrido[3’,2’:4,5]furo[3,2-d]pyrimidine core, 17. Similar to the synthesis 16, the synthesis of core 17 began by condensing formamide with ethyl 3-amino-4,6-dimethylfuro[2,3-b]pyridine-2-carboxylate (8) to afford the pyrimidone intermediate 10 (Scheme 1). Treatment of pyrimidone 10 with POCl₃ gave chloride 12 which readily underwent nucleophilic aromatic substitution with a variety of primary and secondary amines to afford desired analogs 17 in moderate to good yields.

Functional hM₄ potencies for analogs 17 were determined and the results highlighted in Table 4. Of the compounds evaluated, none possessed sub-micromolar potencies. In fact, we observed complete loss of activity in several analogs (17a, 17d, 17e, and 17f). Directly comparing analog 16m (hM₄ EC₅₀ = 910 nM) to 17e (inactive), 16w (hM₄ EC₅₀ = 277 nM) to 17a (inactive), and 16v (hM₄ EC₅₀ = 359 nM) to 17c (hM₄ EC₅₀ = 1.59 μM; 4.4-fold potency decrease) further illustrates the significance of the sulfur atom in the tricyclic ring.

Table 4.

Structures and activities for analog 17.

Cmpd	R Group	hM4 EC50 (nM)a [%ACh Max]
17a		Inactive
VU6008413
17b		2,600
VU6008838		[64]
17c		1,590
VU6008833		[44]
17d		Inactive
VU6008822
17e		Inactive
VU6009080
17f		Inactive
VU6008669
17g		1,490
VU6008806		[49]

^aCalcium mobilization assays with hM_4/Gqi5-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine, n =1 independent experiment performed in triplicate.

Finally, we evaluated the relevance of the pyrimidine nitrogen at the 5-position by synthesizing analogs 18 (Scheme 1). These analogs were synthesized from commercially available amine 7, which underwent a Sandmeyer reaction to give the corresponding bromide. The bromide intermediate was transformed into intermediate 13 via a Suzuki cross-coupling reaction. Cyclization of 13 to form the pyrone followed by exchange with ammonia afforded the pyridone 14, which was chlorinated via treatment with POCl₃ to give intermediate 15 in moderate to good yields. Chloride 15 then underwent nucleophilic aromatic substitution with varies amines to give analogs 18.

Analogs 18 were screened against hM₄ to determine potency with results highlighted in Table 5. It was immediately apparent that this modification was not detrimental to hM₄ activity. In fact, several analogs displayed increased potencies. For instance, direct comparisons between 16m (hM₄ EC₅₀ = 910 nM) and 18b (hM₄ EC₅₀ = 327 nM) resulted in a ~2.8-fold increase in potency. Similarly, comparing 16p (hM₄ EC₅₀ = 6.55 μM) and 18c (hM₄ EC₅₀ = 150 nM) resulted in a ~44-fold increase in potency. Likewise, this trend was also observed when analyzing 16v (hM₄ EC₅₀ = 359 nM) and 18g (hM₄ EC₅₀ = 85 nM) which resulted in a ~4-fold increase in functional hM₄ potency.

Table 5.

Structures and activities for analog 18.

Cmpd	R Group	hM4 EC50 (nM)a [%ACh Max]
18a		>10,000
VU6010206		[90]
18b		327
VU6010225		[101]
18c		150
VU6010205		[80]
18d		233
VU6008889		[98]
18e		202
VU6009049		[82]
18f		124
VU6009203		[71]
18g		85
VU6009048		[90]

^aCalcium mobilization assays with hM_4/Gqi5-CHO cells performed in the presence of an EC₂₀ fixed concentration of acetylcholine, n =1 independent experiment performed in triplicate.

Of these compounds, 16o, 16w, 16x, 16y, 28b, 18b, 18d, and 18g were advanced into a battery of in vitro DMPK assays and our standard rat plasma:brain level (PBL) cassette paradigm (Table 6).¹⁸ These compounds were chosen to move forward based on their M₄ potency (EC₅₀ < 350 nM) as well as their chemical diversity across subseries. In regard to physicochemical properties, these analogs all possessed molecular weights less than 450 Da possessed molecular weights less than 450 Da with 16o, 16w, and 16x having the most attractive CNS xLogP values (2.79 – 3.39).^30–31 All analogs tested displayed high predicted hepatic clearance in at least one species based on microsomal CL_int data (rat CL_heps > 56 mL/min/kg). While analogs 16o and 28b displayed moderate human predicted hepatic clearance based on microsomal CL_int data (human CL_heps of 11–12 mL/min/kg) all other analogs tested displayed high human predicted hepatic clearance based on microsomal CL_int data (human CL_hep > 15 mL/min/kg).

Table 6.

In vitro DMPK and rat PBL data for select analogs 16, 18 & 28.

Property	16o VU6008280	16w VU6007219	16x VU6007217	16y VU6007215	28b VU6009100	18b VU6010225	18d VU6008889	18g VU6009048
MW	398.5	284.38	270.35	312.39	380.51	377.5	283.39	283.39
xLogP	3.39	3.24	2.79	1.85	3.75	4.89	3.69	3.92
TPSA	84.8	41.9	41.9	51.1	59	58	29	37.8
In vitro PK parameters
CLINT (mL/min/kg), rat	176	1896	741	594	367	334	2098	3859
CLHEP (mL/min/kg), rat	50	68	64	63	59	58	68	69
CLINT (mL/min/kg), human	28	134	296	129	25	50	161	345
CLHEP (mL/min/kg), human	12	18	20	18	11	15	19	20
Rat fuplasma	0.024	0.054	0.071	0.206	0.017	0.005	0.016	0.009
Human fuplasma	0.002	0.016	0.021	0.056	0.017	0.001	0.003	0.005
Rat fubrain	0.011	0.004	0.009	0.019	0.011	0.003	0.002	0.004
Brain Distribution (0.25 h) (SD Rat; 0.2 mg/kg IV)
Kp, brain:plasma	0.45	5.36	6.49	4.47	0.54	2.27	37.1	10.4
Kpuu, brain:plasma	0.21	0.65	0.27	0.40	0.35	1.16	4.87	4.18

Compounds within the tricycle series of core 6 (18b, 18d, & 18g) were, in general, highly bound to plasma protein (rat f_us plasma 0.005 – 0.016; rat f_us brain 0.002 – 0.004; human f_us plasma 0.001 – 0.005). By contrast, compounds within the tricyclic series of core 5 (16o, 16w, 16x, 16y, 28b) exhibited reduced protein binding profiles (rat f_us plasma 0.017 – 0.21; rat f_us brain 0.004 – 0.019; human f_us plasma 0.002 – 0.056). Interestingly, compounds 16y and 28b displayed the best overall protein binding profiles (rat f_us 0.017 – 0.21; rat f_us brain 0.032 – 0.053; human f_us 0.011 – 0.019); however, these analogs were not progressed forward due to their high predicted hepatic clearance in rat and/or human. Analogs 16o (rat brain:plasma K_p = 0.45, K_p,uu = 0.21) and 16x (rat brain:plasma K_p = 6.49, K_p,uu = 0.27) proved to have low CNS distribution of unbound drug, while analogs 16w, 16y and 28b displayed moderate CNS distribution (rat K_p,uus = 0.35 – 0.65). On the other hand, analogs 18d (rat brain:plasma K_p = 37.1, K_p,uu = 4.87) and 18g (rat brain:plasma K_p = 10.4, K_p,uu = 4.18) displayed excellent CNS penetration; however, these compounds suffered from low fraction unbound.

In summary, a scaffold hopping exercise utilizing a “tie-back” strategy based on M₄ PAM 3 proved to be a successful strategy in converting an early lead compound, VU0152100, into potent tricyclic M₄ PAM analogs devoid of the classic β-amino carboxamide moiety. Analogs within the tricycle series containing the 2,4-dimethylthieno[2,3-b:5,4-c’]dipyridine core (6) gave many potent M₄ PAMs (Table 5; hM₄ EC₅₀ = 85–327 nM); however, these analogs displayed very poor fraction unbound in regards to brain and plasma protein (f_us < 0.01) as well as high human (CL_heps ≥ 15 mL/min/kg) and rat (CL_heps ≥ 60 mL/min/kg) predicted hepatic clearance. Thus, additional analogs of core 6 were not pursued further. Directing our effort toward the 7,9-dimethylpyrido[3’,2’:4,5]thieno[3,2-d]pyrimidine core (5) showed benzyl amine linked analogs generally provide only moderately potent analogs (Table 1). Changing the linker to small aliphatic amine linkers provided several potent M₄ PAMs (Table 2; hM₄ EC₅₀ = 144–453 nM). On further evaluation, this subseries of analogs generally exhibited improved plasma protein binding profiles. Unfortunately, the small aliphatic amine linker analogs suffered from very high human (CL_heps ≥ 18 mL/min/kg) and rat (CL_heps ≥ 62 mL/min/kg) predicted hepatic clearance and further work on this subseries was abandoned. Further exploration revealed that carbon-linked groups (Table 3) showed improved brain and plasma f_us when compared to the 2,4-dimethylthieno[2,3-b:5,4-c’]dipyridine core (6) as previously discussed; however, due to poor human and rat predicted hepatic clearance, further work on this subset of analogs was not pursued.

Our efforts to mask the β-amino carboxamide moiety of an early lead M₄ PAM compound, VU0152100, resulted in the discovery of two new tricyclic cores (5 & 6) which provided potent M₄ PAM analogs. This endeavor also provided analogs VU6007215 & VU6009048, which display 3.5- fold and 6-fold increase in human M₄ activity, respectively, when compared to parent compound VU0152100. Moreover, VU6007215 exhibits a reduced protein binding profile in relation to the parent compound VU0152100. In comparison to the previously reported VU6017654 (rat f_us plasma 0.007; rat f_us brain 0.003; human f_us plasma 0.021), VU6007215 shows a much-improved protein binding profile (rat f_us plasma 0.206; rat f_us brain 0.019; human f_us plasma 0.056). Additionally, VU6007215 displayed higher CNS distribution of unbound drug (rat brain:plasma K_p = 4.47, K_p,uu = 0.40) than that of VU6017654 (rat brain:plasma K_p = 0.25, K_p,uu = 0.11); however, a ~2-fold loss in hM₄ and rM₄ functional potency was observed with VU6007215. Although this excursion did not deliver PAMs with DMPK profiles to warrant advancement as development candidates, it did garner insights for future scaffold designs toward that goal. These refinements will be reported in due course.