Improved Ca_v2.2 Channel Inhibitors through a gem-Dimethylsulfone Bioisostere Replacement of a Labile Sulfonamide

Abstract

We report the investigation of sulfonamide-derived Ca_v2.2 inhibitors to address drug-metabolism liabilities with this lead class of analgesics. Modification of the benzamide substituent provided improvements in both potency and selectivity. However, we discovered that formation of the persistent 3-(trifluoromethyl)benzenesulfonamide metabolite was an endemic problem in the sulfonamide series and that the replacement of the center aminopiperidine scaffold failed to prevent this metabolic pathway. This issue was eventually addressed by application of a bioisostere strategy. The new gem-dimethyl sulfone series retained Ca_v2.2 potency without the liability of the circulating sulfonamide metabolite.

N-Type calcium channels (Ca_v2.2) are expressed in the presynaptic termini of primary afferent nociceptors in the spinal cord and are key components of the pain transmission pathway. We have previously reported the discovery of a series of aminopiperidine sulfonamide state-dependent Ca_v2.2 channel inhibitors with potential utility for the treatment of chronic pain. Members of this class are orally bioavailable and efficacious in preclinical pain models.¹ In particular, 1 (Figure 1) is selective for inhibition of the Ca_v2.2 channel over the cardiovascular ion channels hERG and Ca_v1.2, with a decent preclinical PK and biodistribution profile. The compound exhibits dose-dependent efficacy in preclinical models of inflammatory hyperalgesia and neuropathic allodynia, but it has no efficacy in Ca_v2.2 gene-deleted mice. The compound is also devoid of ancillary preclinical cardiovascular or CNS pharmacology. While 1 may be used as a preclinical tool to evaluate state-dependent inhibition of Ca_v2.2 channels, it has several drug metabolism related issues that confound further development of the compound. These include moderate CYP3A4 inhibition (IC₅₀ = 5.1 μM), PXR activation (EC₅₀ = 0.43 μM), and the formation of the persistent circulating metabolite 2. The sulfonamide metabolite 2 exhibited preclinical antinociceptive activity in its own right, although it was devoid of significant Ca_v2.2 activity. In order to further improve this lead class of Ca_v2.2 inhibitors and address the drug metabolism issues associated with this series, SAR optimization studies were carried out on the amide substituent and the aminopiperidine core, and a bioisostere replacement strategy was pursued for the sulfonamide moiety.

Figure 1.

In vivo metabolism of 1 to afford the primary sulfonamide 2.

The first substituent to be investigated for further optimization was the benzoic amide moiety (Table 1). The synthesis of piperidine sulfonamide compounds listed in Table 1 was straightforward as described previously.¹ This synthetic route allowed rapid SAR studies modifying substitutions on the benzamide phenyl ring. In some instances, the desired benzoic acid partner had to be synthesized. For example, the intermediate 4-(methylsulfonyl)-2-trifluoromethoxy-benzoic acid used for 12 was prepared from commercially available 4-bromo-1-iodo-2-trifluoromethoxy-benzene as indicated in the Supporting Information.

Table 1. Effect of Benzamide Substitutions on Ca_v2.2 Inhibitory Potency and Selectivity.

compd	R	Cav2.2 FLIPR IC50, μM or % inh @ 1 μMa	MK-499% inh @ 10 μMb	CYP 3A4% inh @ 10 μM	PXR EC50 (μM)
1	2-SO2Me-5-F	0.56	32%	75%	0.42
3	2-SO2Me-5-CN	0.67	41%	59%	5.0
4	2-SO2Me-4-CN	0.45	66%	86%	8.4
5	2-SO2Me-4-CF3	0.027	43%	78%	4.3
6	3-SO2Me	15%	NDc	ND	ND
7	4-SO2Me	52%	ND	ND	ND
8	2-Cl-5-SO2Me	0.34	49%	69%	2.7
9	2-F-4-SO2Me	0.35	78%	ND	6.8
10	2-Cl-4-SO2Me	0.29	32%	78%	10.8
11	2-CF3-4-SO2Me	0.021	58%	75%	2.8
12	2-OCF3-4-SO2Me	0.048	63%	16%	1.7
13	4-CONH2	12%	ND	ND	ND
14	2-CF3-4-CONH2	0.76	32%	61%	1.9
15	2-OCF3-4-CONH2	0.39	41%	60%	>10

^aIC₅₀ values and % inhibition values are averages of n ≥ 3.

^bhERG inhibition was measured by binding displacement of [³⁵S]MK-499.

^cND: not determined.

The resulting compounds were evaluated for their ability to inhibit Ca_v2.2 channels using the FLIPR calcium-influx assay that has been described previously.² Compounds were screened for % inhibition of Ca_v2.2 FLIPR activity at 1 μM concentration. Compounds showing good activity inhibiting Ca_v2.2 channels were titrated to determine their IC₅₀ and were also counter-screened against the related L-type calcium channel Ca_v1.2 using a related FLIPR assay. Potent compounds were also screened against the hERG K⁺ channel using a binding assay that measures the displacement of [³⁵S]MK-499, a well characterized hERG K⁺ channel blocker.³ Additionally, the active Ca_v2.2 inhibitors were profiled for inhibition of CYP3A4 and activation of PXR, which may indicate the likelihood of the compound to induce CYP3A4 expression in vivo.⁴ The resulting data are summarized in Table 1.

Compound 1 is a relatively potent PXR activator. Thus, reducing PXR activity while improving Ca_v2.2 inhibition was the focus of our initial SAR optimization. The addition of polar substituents to specific positions of the molecule (often at the end region of the molecule) has been shown to reduce PXR potency in other instances.⁵ We decided to apply this strategy to compound 1 by adding more polar functional groups at the meta or para position of the benzamide phenyl ring while retaining the methylsulfone group at ortho position. Replacing the fluoro group in compound 1 with a cyano group (3) had a minimal impact on Ca_v2.2 inhibitory potency and resulted in over 10-fold reduction in PXR potency. The cyano group in the para position exhibited further reduced PXR activity, with a slight gain in Ca_v2.2 potency (4). Interestingly, placing a lipophilic CF₃ group at the para position as in 5 led to a 10-fold reduction in PXR activity as compared to 1. Compound 5 also exhibited a 20-fold increase in Ca_v2.2 potency over 1. A substantial loss in potency was observed for compounds with the sulfone in the meta (6) or para (7) positions. However, in each instance the potency was restored by the addition of an ortho substituent (8–12). Among ortho substituents, the lipophilic trifluoromethy and trifluoromethoxy substituents (11 and 12) provided a greater enhancement of Ca_v2.2 inhibitory potency than the halogen substituents (8, 9, or 10). PXR activity was reduced when the methylsulfone group was introduced at the para or meta position, and the para substitution has a stronger effect (8 vs 10). Substitution with other polar functional groups was also explored. With a primary amide at the para position, compound 13 was significantly less potent as a Ca_v2.2 inhibitor. However, again in this instance, the addition of a CF₃ or OCF₃ group at the ortho position restored Ca_v2.2 inhibitory potency (14 and 15). Many of the compounds listed in Table 1 were reasonably devoid of MK-499 binding activity. However, CYP3A4 inhibition was substantially reduced only for compound 12.

In a parallel effort, the SAR investigation on the center aminopiperidine portion of the molecule was undertaken. This region of the molecule was found to be surprisingly tolerant of modification while maintaining Ca_v2.2 inhibitory potency, as illustrated in Figure 2. The addition of an ipso-methyl group at the 4-position of the aminopiperidine ring (16) resulted in only a slight loss of potency as compared to 1. The analogous 3-aminopyrrolidine (17) exhibited a small improvement in potency, even as a racemic mixture. Although with azetidine as the center scaffold compound 18 is less potent in comparison to its piperidine analogue 1, the addition of the 2-methyl group resulted in a significant increase in potency (19 and 20). In these compounds, the relative stereochemistry has little effect on potency as both cis- and trans-diastereomers have similar activity (cis-rac-19 and trans-rac-20). With 2-azaspiro[4.5]decane as the center scaffold, compound 21 was 3-fold more potent than compound 1 as a Ca_v2.2 inhibitor. From this investigation of aminopiperidine replacement moieties, we concluded that the center substituent of this lead class provided limited influence on the potency of the compounds as Ca_v2.2 inhibitors but rather was acting as a scaffold for the presentation of the distal pharmacophores of the molecules. The compounds in Figure 2 were synthesized according to the similar procedure used for the synthesis of the compounds in Table 1. The synthesis of the 4-methylpiperidine intermediate for the synthesis of 16 is described in the Supporting Information.

Figure 2.

Summary of the center scaffold modifications and SAR depenence on Ca_v2.2 potency. IC₅₀ values are averages of n > 3.

In an earlier report on this lead class,¹ we discovered that a critical issue of compound 1 was formation of the long lasting sulfonamide metabolite 2 both in vitro and in vivo (Figure 1). One possible mechanism for formation of metabolite 2 is through a two-step metabolism sequence involving hydroxylation of the α-carbon of the sulfonamide group, followed by hydrolysis of hemiamino ketal. Several strategies were pursued to inhibit the formation of metabolite 2 based on this hypothesis. These strategies included replacing the piperidine with more oxidation resisting azetidine,⁶ making α-carbon quaternary to prevent hydroxylation, and using the spiro ring system to provide extra stability. A selection of structurally diverse compounds that share common 3-CF₃-benzene sulfonamide fragment are illustrated in Figure 3. These compounds were incubated with human or rat liver microsomes and analyzed for formation of metabolite 2. To our surprise, for all the compounds tested in this assay (16 and 22–27), the formation of metabolite 2 occurred at a greater rate than that of compound 1 (Table 2).

Figure 3.

Compounds tested for microsomal stability and metabolite 2 formation. Ca_v2.2 IC₅₀ values are averages of n ≥ 3.

Table 2. Formation of Metabolite 2 in Microsome Incubation Studies^a.

	% of parent remaining		conc of 2 (μM)
compd	rat	human	rat	human
1	40	30	0.01	0.03
16	NDb	ND	0.06	0.04
22	18	26	0.08	0.95
23	3.4	30	0.08	1.16
24	72	78	0.04	0.45
25	4.7	5.3	0.11	0.83
26	3.8	4.2	0.18	5.13
27	12	8.5	ND	ND

^aCompounds were incubated at 10 μM with 1 mM NADPH and 1 mg/mL rat or human microsomal protein for 1 h.

^bND: not determined.

To determine if metabolite 2 is formed in vivo as well, compound trans-rac-25 and 27 were evaluated in rat pharmacokinetics studies. Indeed, 25 and 27 gave very high concentrations of metabolite 2, as illustrated in Figure 4. While the normalized AUC for the parent compound 25 in whole blood was only 0.017 μM·hr·kg/mg after a 3 mg/kg oral dose, the normalized AUC for metabolite 2 in blood was 64 μM·hr·kg/mg. Similar results were obtained with 27. The normalized AUC for the parent and metabolite was 0.022 μM·hr·kg/mg and 218 μM·hr·kg/mg, respectively, following 3 mg/kg oral dose of 27. At this point, we were convinced that formation of the primary sulfonamide metabolite 2 is endemic to the sulfonamide substituent and that alternative strategies such as replacing the sulfonamide moiety with a bioisostere would be required.

Figure 4.

Blood levels of parent 25 (□), parent 27 (△), and metabolite 2 in rats after 3 mg/kg oral dose of 25 (■) and 27 (▲). Values are averages for n = 2 rats (±SD).

We thus shifted our focus to the identification of sulfonamide replacement bioisosteres, and several of these efforts are presented in Table 3. Although amide is a common bioisostere to sulfonamide, replacing sulfonamide in 12 with the corresponding amide resulted in complete loss of potency (28). Similarly, replacement of the sulfonamide moiety with amine, ether, alcohol, and ketone substituents also afforded compounds lacking significant inhibitory potency for the Ca_v2.2 channel (data not shown). The sulfone group, however, proved to be a good replacement for the sulfonamide substituent. Although sulfone compound 29 was 8-fold less potent than the analogous sulfonamide 12, the sulfone remained a moderately potent Ca_v2.2 inhibitor with an IC₅₀ of 0.36 μM. The addition of one methyl group to the α-position of the sulfone resulted in a 4-fold improvement in potency for one enantiomer (30en-1) and no change for the other (30en-2). The gem-dimethyl analogue 31 retained the potency of the most active monomethyl enantiomer of 30 and was four times more potent than the des-methyl analogue 29. The sulfone analogue bearing the analogous benzamide substitution as compound 1 was synthesized. Compound 32 is equally potent to 1 with Ca_v2.2 IC₅₀ of 0.51 μM (Table 3). By comparison of the sulfonamide analogues (1 and 12) to their corresponding gem-dimethyl sulfone derivatives (31 and 32), we concluded that the gem-dimethyl sulfone was an adequate bioisostere for further investigation. The synthesis of the critical gem-dimethylsufone piperidine intermediate is described in the Supporting Information.

Table 3. Ca_v2.2 Inhibitory Potency of Sulfone Compounds.

^aIC₅₀ values and % inhibition values are averages of n ≥ 3.

Compounds 31 and 32 were further profiled, and the results are summarized in Table 4. Both compounds maintain excellent selectivity against Ca_v1.2 and hERG channels (IC₅₀ > 10 μM), and both are weak CYP3A4 inhibitors with IC₅₀ values of about 10 μM. Although 32 showed significant improvement in PXR activation over 1, it is still a moderate PXR activator with an EC₅₀ of 2.5 μM, while 31 is devoid of significant PXR activity. It is critical that the Ca_v2.2 channel blockers can access the CNS and are not substrates for the Pgp efflux mechanism since the Ca_v2.2 channel is localized in the synapse of spinal cord neurons.⁷ Thus, compounds 31 and 32 were also tested to determine if they were Pgp substrates. With the same substitution groups on the benzamide as compound 1, 32 was found not to be a Pgp substrate (human BA/AB ratio: 1.8).⁸ However, 31 is a Pgp substrate with measured human BA/AB ratio of 4.2. Both compounds are highly permeable, with Papp of 35 × 10^–6 cm/sec for each compound. Both compounds exhibit good metabolic stability with percentage of remaining parent compounds ranging from 62% to 73% after incubation with human or rat liver microsomes (1 mg/mL protein and 1 μM substrate) for 1 h.

Table 4. Metabolism Profiles of 31 and 32.

		microsomes %a
compd	PXR activation EC50	human	rat	hPgp BA/AB ratio
31	>30 μM	62	73	4.2
32	2.5 μM	64	68	1.8

^aCompounds were incubated at 10 μM with 1 mM NADPH and 1 mg/mL rat or human microsomal protein for 1 h.

Compounds 31 and 32 were evaluated for their pharmacokinetic and efficacy properties in rats, and the results are shown in Table 5, along with corresponding data for 1 and the metabolite 2.¹ Sulfone 31 has good rat PK properties with moderate clearance and half-life (Cl = 27 mL/min/kg; t_1/2 = 2.3 h) and excellent bioavailability (F = 95%). Compound 32 has similar moderate clearance (Cl = 29 mL/min/kg) but lower bioavailability (F = 24%). Compounds 31 and 32 were evaluated for antinociceptive activity in a rat inflammatory pain model, and the results are shown alongside the previously reported results for 1 and the corresponding metabolite 2 (when dosed orally as 2).¹ Inflammatory pain was induced by intraplantar injection of complete Freund’s adjuvant (CFA) into the hind paws of rats.⁹ Mechanical allodynia was assessed at 1 and 3 h post a single oral dose. Although 1 exhibits significant efficacy in this model, it is impossible to assess the contribution from the efficacious metabolite 2, which is devoid of significant Ca_v2.2 activity and acts via an undetermined mechanism.¹ Compound 31 exhibited modest efficacy (21% and 30%, respectively) following a 10 mg/kg dose. Compound 32 showed robust activity following a 10 mg/kg oral dose, with 36% and 49% reversal of mechanical allodynia at 1 and 3 h time points, respectively. This activity increased to 57% and 54% reversal at 1 and 3 h, respectively, following a 30 mg/kg oral dose. Therefore, compound 32 achieved similar efficacy to 1 in the rat CFA model without the confounding contribution of metabolite 2.

Table 5. Rat PK Profile and Rat CFA Efficacy of 1, 2, 31, and 32.

				rat CFA pain modele
	rat PK profilea			% reversal of allodynia
compd	AUCNb	CLc	F%d	dosef	1 h	3 h
1	0.86	35	99%	10	43%	38%
				30	66%	57%
2	111	0.31	46%	30	53%	44%
31	0.97	27	95%	10	21%	30%
32	0.27	29	24%	10	36%	49%
				30	57%	54%

^aDose 3 mg/kg PO, 1 mg/kg IV; PK parameters based on whole blood concentrations.

^bNormalized AUC (PO, μM·h·kg/mg).

^cClearance (mL/min/kg).

^dBioavailability.

^eAverage percent reversal are calculated as (postdose - prdose)/(preinjury predose) for each rat (n = 6).

^fOral dose in mg/kg.

It is interesting that despite the 6-fold greater potency for compound 31 over 32, and higher oral exposure for 31 in rat PK studies, 31 was considerably less efficacious in rat CFA model. Although the full biodistribution for these compounds into the CSF and spinal cord is not available, we infer that biodistribution differences may contribute to the discrepancy in efficacy based on in vitro studies. Sulfone 31 is a more robust Pgp substrate than 32, which may lead to lower CNS exposure and thus lower efficacy. In addition, 31 was determined to have a lower unbound fraction in 100% rat plasma (1.3%) than 32 (6.8%), which may also contribute to lower unbound exposure of 31 for biodistribution into the CNS.

Although neither 31 nor 32 were suitable to be considered as a development candidate due to the issues of Pgp substrate and low in vivo efficacy (for 31) and PXR activity (for 32), these compounds demonstrated significant metabolic advantage over sulfonamide series and their potential utility as potent and selective Ca_v2.2 inhibitors. With these modifications, we have demonstrated that the combination of hydrophilic substitutions such as methylsulfone and lipophilic substitutions of CF₃ and OCF₃ on the benzamide phenyl ring resulted in significant improvement of Ca_v2.2 potency, as well as mitigating PXR activation. In the sulfonamide series, formation of the primary sulfonamide metabolite 2 in vivo was endemic to the lead class. This issue was eventually addressed successfully by employing a bioisostere replacement strategy. The corresponding gem-dimethyl sulfone compounds exhibited similar Ca_v2.2 in vitro potency and efficacy to their sulfonamide analogues, but without the confounding liability of long lasting and efficacious sulfonamide metabolite formation. Ongoing work to further optimize the overall profile and enhance in vivo efficacy in this new series will be published in the due course.

Glossary

Abbreviations

AUC

area under the pharmacokinetic exposure curve

Ca_v

voltage-gated calcium channel

CFA

complete Freund’s adjuvant

CL

pharmacokinetic clearance

CNS

central nervous system

conc

concentration

CSF

cerebrospinal fluid

CYP

cytochrome P450 enzymes

FLIPR

fluorometric imaging plate reader

hERG

human ether-a-go-go-related gene

IV

intravenous

NADPH

nicotinamide adenine dinucleotide phosphate hydrogen

Pgp

P-glycoprotein transporter

PK

pharmacokinetic

PO

per os or by mouth

PXR

pregnane X receptor

SAR

structure–activity relationship

Supporting Information Available

Details for the Ca_v2.2 FLIPR assay, and the synthesis and characterization of compounds 12, 31, and 32. This material is available free of charge via the Internet at http://pubs.acs.org.

The authors declare no competing financial interest.