Novel GlyT1 inhibitor chemotypes by scaffold hopping. Part 1. Development of a potent and CNS penetrant [3.1.0]-based lead

Abstract

This letter describes the development and SAR of a novel series of GlyT1 inhibitors derived from a scaffold hopping approach that provided a robust intellectual property position, in lieu of a traditional, expensive HTS campaign. Members within this new [3.1.0]-based series displayed excellent GlyT1 potency, selectivity, free fraction, CNS penetration and efficacy in a preclinical model of schizophrenia (prepulse inhibition).

Significant efforts are currently focused on non-dopaminergic strategies to address the unmet medical needs in schizophrenia, and targeting N-methyl-D-aspartate (NMDA) receptor hypofunction has garnered a great deal of attention.^1–3 Elevation of synaptic glycine levels near NMDA-containing synapses, by inhibition of the glycine transporter type 1 (GlyT1), has proven to be a viable mechanism for achieving efficacy in multiple preclinical models of schizophrenia with a diverse array of GlyT1 chemotypes.^4–9 More recently, Roche reported Phase II clinical efficacy with their GlyT1 inhibitor (RG6178) in improving the negative symptoms in patients with schizophrenia,¹⁰ a symptom cluster largely unmet with available antipsychotic agents, that re-ignited the field. Furthermore, new data suggests roles for GlyT1 inhibition in alcohol dependence, addiction and pain.¹¹

Based on these data, the late-stage clinical status of GlyT1, and the crowded intellectual property space,^4–9 we elected to attempt to develop novel chemical space by scaffold hopping, a strategy we previously employed successfully for T-Type calcium channel inhibitors,¹² to accelerate a fast follower GlyT1 inhibitor program. For this exercise, we were attracted to the GlyT1 inhibitors reported from both Merck, represented by 1 and 2,^13–15 and Pfizer’s 3¹⁶ (Fig. 1), as they possessed potent GlyT1 inhibition with good DMPK profiles and efficacy in preclinical models;^13–16 moreover, homology and overlap was noted between these otherwise disparate chemotypes.

Figure 1.

Structures and activities of GlyT1 inhibitors disclosed by Merck (1 and 2) and Pfizer (3) utilized in subsequent scaffold-hopping to access new chemical space.

Our initial approach was to replace the central piperidine core of 1 and 2^13–15 with the [3.1.0] bicyclic ring system found in 3,¹⁶ to arrive at analogs such as 4 (Fig. 2). Synthetically, analogs 4 were arrived at via a seven step route in low (~4%) overall yield. Starting from commercially available (1R,5S,6r)-3-tert-butyl 6-ethyl 3-azabicyclo[3.1.0]hexane-3,6-dicarboxylate 5, two step conversion to the primary carboxamide 6 proceeded smoothly, followed by treatment with cyanuric chloride to afford the nitrile 7 (Scheme 1). Deprotonation with KHMDS and alkylation with cyclopropyl methylbromide in toluene at 0 °C provided 8 in 20% yield, as a single diastereomer (steroechemical assignment based on literature precedent¹⁷ and nOe studies). TFA-mediated removal of the Boc group, followed by sulfonylation with various sulfonyl chlorides, generated congeners 9. Finally, ‘Raney’ Ni reduction of the nitrile and subsequent acylation with 2,4-dichlorobenzoyl chloride delivered analogs 4.

Figure 2.

Initial approach to hybridize 2 and 3 to afford novel GlyT1 [3.1.0] scaffold 4.

Scheme 1.

Reagents and conditions. (a) LiOH, THF, MeOH, rt; (b) NH₄Cl, EDC, HOBt, DIEPA, DMF, 0 °C; (c) cyanuric chloride, DMF, 0 °C; (d) KHMDS, toluene, 0 °C, cyclopropyl methylbromide; (e) TFA, CH₂Cl₂, 0 °C; (f) RSO₂Cl, CH₂Cl₂, DIEPA, 0 °C; (g) Raney Ni, H₂ (45 psi), NH₄OH, MeOH; (h) 2,4-dichlorobenzoyl chloride, CH₂Cl₂, DIEPA, 0 °C.

The initial 10-membered library of analogs 4 displayed somewhat unexpected SAR, with a significant diminution in potency relative to the piperdine-based GlyT1 inhibitors 1 and 2 (Table 1).^13–15 For example, 4a (GlyT1 IC₅₀ = 1.3 μM), the direct analog of 2 (GlyT1 IC₅₀ = 26 nM), lost 50-fold in potency, and 4b (GlyT1 IC₅₀ = 360 nM), the n-propyl congener of 1 (GlyT1 IC₅₀ = 2.4 nM), lost ~150-fold. A trend towards increased activity resulted with a cyclopropyl methyl sulfonamide 4d (GlyT1 IC₅₀ = 230 nM), but all other modifications, including aryl (4i and 4j) and heteroaryl sulfonamides (4g and 4h) led to significant loss in GlyT1 potency. However, all analogs 4 retained high selectivity versus GlyT2 (IC₅₀ > 30 μM) and preliminary in vitro DMPK profiling indicated that this core retained the favorable disposition properties of 1–3 (f_u 2–4%, IC₅₀ >10 μM vs. CYP3A4, 2D6, 1A2 and 2C9). Substitution of the 6-cyclopropyl methyl group in 4 with either phenyl or 2-pyridyl moieties, also led to inactive analogs (GlyT1 IC₅₀ > 30 μM) and represented a divergence from the SAR of the piperidine series 1 and 2.^13–15 Analogs of 4 where the optimal sulfonamides were maintained (4b and 4d), but the substitution on the benzamide moiety was varied, once again led to a large diminution in GlyT1 potency (data not shown).

Table 1.

Structures and activities of [3.1.0] analogs 4.

Compound	R	GlyT1 IC50 (μM)a	GlyT2 IC50 (μM)a
4a		1.3	>30
4b		0.36	>30
4c		3.4	>30
4d		0.23	>30
4e		5.5	>30
4f		2.9	>30
4g		1.0	>30
4h		0.89	>30
4i		>10	>30
4j		>10	>30

^aIC₅₀s represent single determinations performed in duplicate

As we began to assess and consider the data generated thus far, we were attracted to the N-methyl imidazole moiety of Pfizer’s 3,¹⁶ and with simple models could achieve an orientation in which the N-methyl imidazole moiety could fill the same space as the alkyl sulfonamides in 4b and 4d. Thus, we hypothesized that N-methylimidazole sulfonamide congeners might enhance GlyT1 potency in analogs 4.

To test this concept, we took advantage of our large supply of various 2-pyridyl containing [3.1.0] cores (inactive with alkyl or aryl sulfonamides) and prepared the N-methylimidazole sulfonamide analogs 10 and 11 (Fig. 3), as work form Merck demnstarted that the 2-pyridyl moiety was superior to the orignal 4-phenyl moiety in 1.^14,15 These analogs all displayed sub-micromolar potency at GlyT1 (IC₅₀s of 247 nM for 10 and 185 nM for 11), clogPs of 2.3, large fraction unbound in plasma (f_u 7–14%), clean CYP profiles (IC₅₀ >30 μM) and in oral brain tissue distribution studies, K_p ([brain]/[plasma]) ratios of 1.1. Interestingly, replacement of the N-methylimidazole with either imidazole or an N-methyl pyrazole led to a complete loss of GlyT1 activity. As incorporation of the N-methyl imidazole sulfonamide increased potency in the 2-pyridyl [3.1.0] core from IC₅₀s >30 μM to IC₅₀s < 250 nM, we were excited to see the impact of this modification to the already very potent cyclopropyl methyl [3.1.0] core of 4.

Figure 3.

Impact of incorporation of an N-methyl imidazole sulfonamide into the [3.1.0] scaffold to afford 10 and 11.

With a slight modification of the chemistry in Scheme 1, we were able to readily prepare a library of analogs 12, where the benzamide moiety was varied in the context of the cyclopropyl methyl [3.1.0] core containing an N-methyl imidazole sulfonamide. As shown in Table 2, this endeavor was very productive, affording potent (IC₅₀s from 4 to 119 nM) and selective GlyT1 inhibitors with excellent DMPK profiles and CNS penetration in rat with oral dosing (K_p of 0.3 to 0.8). While many analogs displayed excellent potency at GlyT1, such as 12a, 12d and 12f, analog 12d stood out as having the best balance of potency (GlyT1 IC₅₀ = 5 nM), fraction unbound in plasma (f_u 7%), and in an oral brain tissue distribution study, a K_p ([brain]/[plasma]) ratio of 0.8.

Table 2.

Stuctures and activities of [3.1.0] analogs 12.

Cmpd	Ar	GlyT1 IC50 (nM)a	clogP	fu (hum)	Kpb (rat PO)
12a	2,4-diClPh	4	2.3	5%	0.3
12b	2,4-diFPh	44	1.9	8%	0.5
12c	2,6-dFPh	28	1.9	8%	0.5
12d	2-CF3Ph	5	2.5	7%	0.8
12e	3-CF3Ph	64	2.5	4%	0.7
12f	2-ClPh	19	2.6	4%	0.4
12g	3,4-diClPh	82	3.2	2%	0.3
12h	2-ClPh	119	2.7	3%	0.4
12i	3,5-diClPh	100	3.3	2%	0.3
12j	4-ClPh	32	2.7	3%	0.3

^aIC₅₀s represent single determinations performed in dupliate.

All analogs were inactive on GlyT2 (IC₅₀ > 30 μM).

^bK_p = partitioning coefficient = [brain]/[plasma]

After assessing ancillary pharmacology in a Eurofin Lead Profiling panel of radioligand binding assays against 68 GPCRs, ion channels and transporters, our focus narrowed on 12d, as it displayed no inhibition >50%@10 μM against any target in the panel, yet was a potent and selective GlyT1 inhibitor (GlyT1 IC₅₀ = 5±0.5 nM (N=3), GlyT2 IC₅₀ >30 μM). Eadie-Hoffstee plots, where the reaction rate is plotted versus the ratio of the reaction rate and substrate concentration, provide useful insight into the mechanism of enzymatic inhibition, with competitive and noncompetitive enzymatic inhibition demonstrating distinct patterns. An Eadie-Hoffstee plot (Fig. 5) of the effect of this series, represented by 12c, on the enzyme kinetics of [¹⁴C]-glycine transport showed that this series is competitive with respect to glycine, in accordance with the known mechanism of action for 1 and 2^13–15, and distinct from the non-competitive mechanism of action of the sarcosine-derived GlyT1 inhibitors, such as NFPS.^5–8 In addition, 12d possessed a clean CYP profile (IC₅₀ >30 μM against 1A2, 2C9 and 2D6; 14.9 μM against 3A4) and good unbound fraction in both human (f_u = 6.8%) and rat (f_u = 44.8%). We assessed stability in rat plasma for 4 hours at 37 °C, and 12d was stable, indicating the free fraction in rat plasma is truly high. In vitro intrinsic clearance experiments suggest 12d posssesses moderate to high predicted clearance for both human (Cl_HEP = 17.7 mL/min/kg) and rat (Cl_HEP = 43.0 mL/min/kg). A rat IV PK study was conducted with 12d and it displayed moderate in vivo clearance (0.5 mg/kg) (Cl_p = 31 mL/min/kg) with a short half-life (t_1/2 = 14.5 min). This is in-line with data reported for Merck’s 2, which was moderate to high clearance in rat, but low clearance in dog. We evaluated 12d in two separate rat brain tissue distribution studies: one with subcutaneous (10 mg/kg s.c. in 10% tween 80) dosing and one with oral (10 mg/kg p.o. in 0.5% methocellulose) dosing. Both dosing routes exhibited good exposure (s.c. plasma AUC_0-6hr: 976 nM*h; s.c brain, AUC_0-6hr: 431 nM*h (or K_p = 0.44); p.o. plasma AUC_0-6hr: 1156 nM*hr, p.o. brain, AUC_0-6hr: 956 nM*h (or K_p = 0.83)) with unbound concentrations above the GlyT1 IC₅₀ at 6 hr. Due to the higher exposure, for an in vivo proof of concept study in a preclinical model of schizophrenia, we proceeded with oral dosing.

Figure 5.

(A) Saturation [¹⁴C]-glycine transport in the presence of vehicle (red squares) or 40 nM 12c (blue triangles). (B) An Eadie-Hoffstee diagram for 12c and [¹⁴C]-glycine.

Based on the precendent with other GlyT1 inhibitors such as 1,¹³ we evaluated both 2¹⁵ and 12d for their ability to enhance prepulse inhibition (PPI) of the rodent acoustic startle response, a measure of sensorimotor gating known to be deficient in schizophrenic patients.^18,19 In this study (Fig. 6), both 2 and 12d were dosed orally at 30 mg/kg (a dose known to engender >90% occupancy for 2),^15,20,21 and evaluated against four increasing prepulse intensities (70–88 dB). Both 2 and 12d showed a statistically significant enhancement in prepulse inhibiton at the 82 and 88 dB prepulse intensities, with no effect on basal startle amplitude during no-stimulus trials. Thus, 12d (VU0240391), derived from a scaffold-hopping exercise employing 2 and 3, led to a novel [3.1.0]- based GlyT1 inhibitor with in vitro and in vivo properties comparable to other advanced GlyT1 inhbitors in short order, and for which a U.S. patent was issued.²²

Figure 6.

The effect of vehicle, 2 and 12d on PPI in rat at 30 mg/kg oral (p.o.) dosing in 0.5% methylcellulose. *p<0.05 in comparison to vehicle by Dunnett’s test (n=6–7).

In conclusion, we were able to scaffold hop and merge elements from both the Merck piperidine-based series of GlyT1 inhibitors, represented by 1 and 2,^13–15 and Pfizer’s 3,¹⁶ into a novel, patented series of [3.1.0]-based N-methylimidazole sulfonamides 12. Members of this series displayed exceptional GlyT1 potency, DMPK profiles, CNS penetration and comparable in vivo efficacy to advanced GlyT1 inhibitors without the need for an HTS to enable a fast-follower program. Additional scaffolds developed during the course of this scaffold-hopping program will be reported in due course.

Figure 4.

Concentration response curves (N=3) for glycine and 12d in the [¹⁴C]-glycine uptake assay.