Synthesis and SAR of a novel Kir6.2/SUR1 channel opener scaffold identified by HTS

Abstract

Kir6.2/SUR1 is an ATP-regulated potassium channel that acts as an intracellular metabolic sensor, controlling insulin and appetite-stimulatory neuropeptides secretion. In this Letter, we present the SAR around a novel Kir6.2/SUR1 channel opener scaffold derived from an HTS screening campaign. New series of compounds with tractable SAR trends and favorable potencies are reported.

To create your abstract, type over the instructions in the template box below. Fonts or abstract dimensions should not be changed or altered.9.0_/0

Graphical Abstract

ATP-regulated potassium channels (K_ATP channels) are important intracellular metabolic sensors consisting of a heterooctameric assembly of four Kir6 and four SUR subunits.^1,2 Figure 1A summarizes the tissue-specific expression of various K_ATP isoforms. In particular, the Kir6.2/SUR1 subfamily channel has been considered an important drug target due to its role in glucose homeostasis through the control of insulin secretion in pancreatic β-cells.^3–7 Recent studies have also shown that Kir6.2/SUR1 channels may be involved in the secretion of appetite-stimulatory neuropeptides in hypothalamic neurons, further exemplifying the importance of the channel as a valuable therapeutic target.⁸

Figure 1.

K_ATP channel isoforms and selected Kir6.2/SUR1 channel openers. (A) Diversity of K_ATP isoforms and their tissue-specific expression. (B) Structures of representative Kir6.2/SUR1 channel openers 1–4.^{6,11,12,13,16 a}Reported EC₅₀ value; EC₅₀ was 0.75 ± 0.08 μM (Mean ± SEM, n = 6 experiments) in thallium flux assays with T-REx-HEK293 cells

Although Diazoxide (1), an FDA-approved K_ATP channel opener, is still widely used in the clinic, its off-target effects on vascular K_ATP channels limit its utility to efficiently treat abnormalities (such as hyperinsulinism, and hypoglycemia).^6,9,10 As such, there remains an unmet need in the field for alternative therapeutics. In 2003, NN414 (2) was discovered by Novo Nordisk as a next-generation K_ATP channel opener. NN414 (EC₅₀ = 0.45 μM; ref 5) was over sixty-fold more potent than Diazoxide (EC₅₀ = 30 μM), while showing good Kir6.2/SUR1 channel selectivity and giving hope to the field.^4,11 However, the clinical development of NN414 was suspended during phase 2 clinical trials due to an adverse effect on the liver.³ Besides NN414 (and its close analogs) and (±)-Cromakalim derivative BMS-191095 (3), only a limited number of Kir6.2/SUR1 channel opener chemotypes have been reported in the public domain,^6,11–16 many of which only show moderate potencies (e.g. 4, EC₅₀ = 7 μM)⁶ or contained pharmaceutically disadvantageous moieties. In this Letter, we describe the synthesis and structure-activity relationship (SAR) of a structurally distinct Kir6.2/SUR1 channel opener scaffold derived from an HTS screening campaign.

Our ongoing HTS efforts to find potent and selective Kir6.2/SUR1 channel openers have identified several hits of interest including 5-8 (Figure 2A–C); 20,480 compounds were screened at a single dose of 10 μM via high-throughput thallium flux assay and 513 hits were confirmed with 2.5% hit rate. This 20,480 compound library is a chemically diverse subset of the Vanderbilt Discovery Collection. The Discovery Collection (~100,000 total) is made up of lead-like motifs, minimum panassay interference, and maximum chemically diverse small molecules (mean MW = 378) from Life Chemicals. Compounds 5, 7, and 8 showed complete subtype selectivity against Kir6.1/SUR2B channel (a vascular K_ATP channel), while 6 showed weak activity (EC₅₀ > 10 μM). Especially, compound 5 (EC₅₀ = 1.2 μM) was about five-fold more potent than our first-generation tool compound (VU0071063 (4), EC₅₀ = 7 μM) and was selected as our new starting point. To understand the basic SAR texture around 5, we divided the molecule into three parts, the A-ring, the core, and the B-ring, and explored each part separately (Figure 2D).

Figure 2.

Summary of HTS screening results and hit to lead optimization strategy (A) Summary of Primary Screening results revealed small molecules that increase the measurement of thallium flux in cells containing Kir6.2SUR1 channels. Hits were identified as Activators when the increase was greater than 3 Standard deviations from the mean plate response (N = 320). Hits selected for retesting in duplicate included the 327 Activators (>3 SD in blue) and 249 weaker activators (B-score > 5). Retest positives remaining after the duplicated testing included 513 of the 576 tested. Comparison to Counter screening reduced these to 494 promoting to candidates for Concentration Response Curve studies (Omit 63 retest negatives or 19 actives in comparison to Kir6.1SUR2B or HEK). Values are displayed normalized to the plate control VU0071063 (positive control in purple n = 32 per plate). Total tested small molecules were 20480. Not shown (B-score Activators (1254), B-score Inhibitors (272). (B) Concentration Response Curve of selected Compounds 5-8. Triplicate values of 9 concentrations ranging 5 nM-30 μM*, plots using GraphPad Prism 9.2 *30 μM omitted on 5. (C) Structures of selected HTS hits 5 – 8. (D) Compound Optimization Strategy.

The synthesis of 5 and its related analogs is outlined in Scheme 1. An EDCI coupling reaction between commercially available aminothiadiazoles 10 and readily available carboxylic acids 9 provided HTS hit 5 as well as A-ring and B-ring variants in low to moderate yields. Several B-ring variants were synthesized in two steps starting from 11 through either an amidation-alkylation sequence (Scheme 1b) or an alkylation-amidation sequence (Scheme 1c).

Scheme 1.

Synthesis of Kir6.2/SUR1 channel openers^a

^aReagents and conditions: (a) EDCI·HCl, HOBt, DMAP, DMF, 50 °C, o/n; (b) alkyl bromides, KOH, EtOH/H₂O (3:1), rt, 1 h.

Our initial efforts were focused on the A-ring pendant group. As shown in Table 1, the replacement of the ethyl moiety on the 4-position of the 1,2,3-thiadiazole ring with a methyl-group was well tolerated (5a, EC₅₀ = 1.8 μM), while slightly larger substituents marginally improved potency (5b-d). Interestingly, a phenyl ring on the 4-position was also well tolerated (5e, EC₅₀ = 0.78 μM). This result indicates there may be adequate room in the binding pocket to accommodate larger substitutions. As our initial screen of analogs demonstrated that the 4-position of the 1,2,3thiadiazole tolerated simple substituents, we shifted our focus to evaluate the SAR of other areas of the scaffold. The A-ring SAR could then be revisited in the context of an optimized B-ring and core.

Table 1.

Structures and activities for A-ring variants 5a-5e

Cmpd	R=	Synthetic Route	Kir6.2/SUR1 EC50 (μM) [Emax (%)]
5HTS hit		1	1.1a [93]
5a		1	1.8a,c [91]
5b		1	0.46b [82]
5c		1	0.25b [87]
5d		1	0.66b [90]
5e		1	0.78b [87]

^aThallium flux assays with T-REx-HEK293 cells; values represent means from one experiment performed in triplicate. 4 was used as a positive control. 0.48 mM Tl₂SO₄ was used.

^bThallium flux assays with T-REx-HEK293 cells; values represent means from one experiment performed in triplicate. 2 (EC₅₀ = 0.75 ± 0.08 μM) was used as a positive control.

^cValues represent means from two experiments performed in triplicate.

Holding the A-ring and the core constant (4-ethyl-1,2,3-thiadiazole and 2-amino-1,3,4-thiadiazole, respectively), our SAR focus then moved to the B-ring (Table 2). The first library (5f-5j) was aimed at replacing the thioether linker while maintaining the terminal phenyl ring. Although replacement of the sulfur atom with -SO₂- (5f, EC₅₀ = 35 μM) or -NH- (5g, inactive) were not tolerated, the regioisomeric thioether (5h, EC₅₀ = 2.0 μM), ether (5i, EC₅₀ = 3.3 μM), and simple ethylene linker (5j, EC₅₀ = 4.1 μM) maintained good potencies. We were encouraged by this result as it shows that the thioether, a well-known metabolic soft spot, can be replaced with a pharmaceutically more desirable moiety.

Table 2.

Structures and activities for B-ring variants 5f-5ab

Cmpd	R=	Synthetic Route	Kir6.2/SUR1 EC50 (μM) [Emax (%)]	Cmpd	R=	Synthetic Route	Kir6.2/SUR1 EC50 (μM) [Emax (%)]
5 HTS hit		1	1.1a [93]	5q		3	1.14b [105]
5f		1c	35a [150]	5r		3	3.1a,e [82]
5g		1d	Inactiveb	5s		1	0.88b [83]
5h		1	2.0a [89]	5t		1	0.55b [85]
5i		1	3.3a[100]	5u		2	0.97b [106]
5j		1	4.1a [96]	5v		2	0.94b [98]
5k		1	6.5a [101]	5w		1	0.56b [105]
5l		2	1.0a [69]	5x		1	Inactivea
5m		1	0.53b [83]	5y		1	Inactivea
5n		2	0.71b [105]	5z		2	1.1a [77]
5o		1	0.71b [100]	5aa		3	0.31a,e [54]
5p		3	1.00b [109]	5ab		3	0.57b [111]

^aThallium flux assays with T-REx-HEK293 cells; values represent means from one experiment performed in triplicate. 4 was used as a positive control. 0.48 mM Tl₂SO₄ was used.

^bThallium flux assays with T-REx-HEK293 cells; values represent means from one experiment performed in triplicate. 2 (0.75 ± 0.08 μM) was used as a positive control.

^cIntermediate, 5-(benzylsulfonyl)-1,3,4-thiadiazol-2-amine, was prepared using 2.2 equiv. of m-CPBA.

^dIntermediate, N²-benzyl-1,3,4-thiadiazole-2,5-diamine, was prepared by heating a mixture of 5-bromo-1,3,4-thiadiazol-2-amine (1 equiv.), benzylamine (1.5 equiv.), and triethylamine (2.5 equiv.) at 66 °C for 6 h.

^eValues represent means from two experiment performed in triplicate.

Because the des-phenyl analog (5k, EC₅₀ = 6.5 μM) gave about 4-fold loss in potency, our attention turned to phenyl ring decoration instead (5l-5w). Although m- and p-mono methylated analogs (5m, EC₅₀ = 0.53 μM and 5n, EC₅₀ = 0.71 μM) were favored compared to an o-methylated compound (5l, EC₅₀ = 1.0 μM), dimethylated analogs (5u, EC₅₀ = 0.97 μM and 5v, EC₅₀ = 0.94 μM) were not as potent as the mono methylated analogs. Among synthesized analogs, the only unexpected outlier to this trend came out from the 2,5-dichlorinated analog 5w (EC₅₀ = 0.56 μM). As expected, we were able to use halogens as methyl surrogates (5o, 5p, 5s, and 5t). However, halogenated compounds were slightly less potent than their methylated comparators (5m and 5n). Interestingly, polar substituent-containing analogs also showed comparable potency to HTS hit 5 (5q, EC₅₀ = 1.14 μM and 5r, EC₅₀ = 3.1 μM).

Our efforts were then directed towards the replacement of the phenyl ring, exploring both aromatic heterocycles (such as pyridine and thiophene) and aliphatic carbocycles (cyclohexane and cyclopentane). 2- and 3-pyridyl compounds (5x and 5y, respectively) proved inactive. However, the more hydrophobic thiophene analog (5z, EC₅₀ = 1.1 μM) showed comparable potency to HTS hit 5. Interestingly, both carbocycles gave a slight boost in potency (5aa, EC₅₀ = 0.31 μM and 5ab, EC₅₀ = 0.57 μM). This is noteworthy in that this substitution provides an opportunity to increase the sp3 character of the current chemotype while maintaining its potency. To that end, we designed our next-generation library based on the cyclopentyl moiety from 5ab.

Maintaining the B-ring and the core as constants (cyclopentane and 2-amino-1,3,4-thiadiazol, respectively), we revisited the A-ring SAR (Table 3). However, synthesized analogs containing heterocycles other than a 4-substituted 1,2,3-thiadiazole were all inactive (5ac-5ae). The only exception to this trend was 5af (EC₅₀ > 10 μM). Interestingly, only 5af contained a nitrogen atom (bolded) that could potentially mimic the 3-position nitrogen atom of 1,2,3-thiadiazole. Based on this result, we suspect that the 3-position nitrogen atom of the 1,2,3-thiadiazole ring may play an important role in binding. It certainly warrants a follow-up SAR study.

Table 3.

Structures and activities for alternative A-rings 5ac-5af^a

Cmpd	R =	SyntheticRoute	Kir6.2/SUR1 EC50 (μM) [Emax (%)]
5ac		1	Inactive
5ad		1	Inactive
5ae		1	Inactive
5af		1	>10 [97]

^aThallium flux assays with T-REx-HEK293 cells; values represent means from one experiment performed in triplicate. 2 (0.75 ± 0.08 μM) was used as a positive control.

Lastly, we altered the heterocyclic core (Figure 3). Although our attempt to replace the 1,2,3-thiadiazole core with a pyridazine core was unsuccessful (15, inactive), the 1,2,4-thiadiazole 14 was well tolerated and showed notable improvement (~10-fold) in potency compared to 5c and HTS hit 5 (14, EC₅₀ = 0.10 μM).

Figure 3.

Structures and activities for core variations. ^aPrepared via Scheme 1a. ^bPrepared through an SNAr between 2-amino-6-chloropyridazine and benzylamine followed by amidation.

Table 4 presents the detailed ADME profiles as well as the subtype selectivity of 14 and 5ab. Although both compounds showed favorable on-target potencies (0.10 μM and 0.57 μM respectively), their free fraction (both plasma and brain) were quite low (rat, human f_u < 0.01) indicating obvious room for improvement. Rat IV PK cassette study suggests both compounds were peripherally restricted (K_p = 0.03 and 0.04 respectively). Interestingly, both compounds showed in vitro-in vivo (IVIV) disconnect; the observed in vivo clearance (CLp, 3.02 and 2.43 mL min⁻¹ kg⁻¹, respectively) was significantly lower than the predicted clearance in the in vitro assay (Predicted Cl_hep, 66 and 55.3 mL min⁻¹ kg⁻¹, respectively). Although the low free fraction (f_u) may contribute to this IVIV disconnect, follow-up experiments are warranted to elucidate the origin of the disconnection.

Table 4.

Tier 1 DMPK profile data for the selected compounds

Property	14, VU6051545	5ab, VU6050332
Kir6.2/SUR1 EC50 (μM)	0.10 ± 0.01 (n=3)	0.57 ± 0.02 (n=3)
Kir6.1/SUR2B EC50 (μM)	inactive	inactive
MW	377.5	355.5
cLogP	2.99	2.54
TPSA	81	81
Fu, plasma (rat)	<0.01	<0.01
Fu, plasma (human)	<0.01	<0.01
Fu, brain (rat)	<0.01	0.01
Kp	0.03	0.04
Clint (mL/min/kg)	120 (h), 1168 (r)	363 (h), 264 (r)
Predicted Clhep (mL/min/kg)	17.9 (h), 66 (r)	19.9 (h), 55.3 (r)
Rat IV PK cassette
CLp (mL/min/kg)	3.02	2.43
t1/2 (h)	3.11	4.89
MRT (h)	2.12	4.11
Vss (L/kg)	0.38	0.60

In summary, we discovered a novel Kir6.2/SUR1 channel opener series from an HTS screening campaign. Our initial SAR exercise demonstrated tractable SAR textures that provide helpful insight for next-generation analog designs. Several compounds within this series showed improved potencies compare to NN414 (and Diazoxide). Potent and pharmaceutically favorable compounds within this series will be the subject of comprehensive subsequent pharmacological studies related to insulin secretion from the pancreatic β-cells, which will be published in due course.