Discovery of Orally Bioavailable Ligand Efficient Quinazolindiones as Potent and Selective Tankyrases Inhibitors

Abstract

We report herein the discovery of quinazolindiones as potent and selective tankyrase inhibitors. Elucidation of the structure–activity relationship of the lead compound 1g led to truncated analogues that have good potency in cells, pharmacokinetic (PK) properties, and excellent selectivity. Compound 21 exhibited excellent potencies in cells and proliferation studies, good selectivity, in vitro activities, and an excellent PK profile. Compound 21 also inhibited H292 xenograft tumor growth in nude mice. The synthesis, biological, pharmacokinetic, in vivo efficacy studies, and safety profiles of compounds are presented.

The tankyrases (TNKS1 and TNKS2) belong to the poly ADP-ribose polymerase (PARP) family of enzymes and act via mono- or poly-ADP-ribosylation (parsylation) of substrate proteins. The tankyrases also contain the ANK domain, which contains 16–24 ankyrin repeats. The ANK domain interacts with a variety of proteins, including the telomeric protein telomere repeat binding factor-1 (TRF-1). Hence these proteins were named TRF-1 interacting, ankyrin-related ADP-ribose polymerases, or TNKS. It is well documented that TNKS play critical roles in telomere homeostasis,¹ regulation of the Wnt/β-catenin pathway,²⁻⁴ and other pathways, including fibrosis.⁵ The functions of TNKS in a diverse range of cellular functions have been the subject of many comprehensive reviews.⁶⁻⁹ These findings indicate that inhibition of tankyrases could have potential therapeutic benefit in the treatment of cancer, fibrosis, and other hyperproliferative diseases through diverse modes of action.^6,10 Herein, we will describe our effort on the TNKS inhibitors for the treatment of cancer.

Small molecule tankyrase (TNKS) inhibitors have been reported extensively in literature (Figure 1).¹¹⁻¹⁶ At the onset of our program, IWRs (inhibitors of Wnt response) were reported as Wnt/β-catenin pathway modulators.¹⁷ Subsequently, we observed that IWR-3/4 are TNKS inhibitors via a chemoproteomics study. To identify selective inhibitors of TNKS, we initiated a chemistry effort based on IWR-4 (compound 1g). Here, we report the SAR results of this novel class of compounds that yielded TNKS inhibitors that have good potency and excellent selectivity with improved ligand efficiency and PK/PD profiles. The synthesis and safety profiles of selected compounds are also described.

Figure 1.

Representatives of TNKS inhibitors.

Compound 1g, a quinazolindione analogue, shows modest TNKS inhibition activity. Although compound 1g exhibits reasonable biochemical potency against TNKS1 and 2 (IC₅₀ = 0.63 μM and ∼8 μM, respectively), it is less active in the cellular assay (β-catenin luciferase assay), presumably due to the poor physicochemical properties (Table 1). Nevertheless, compound 1g served as starting point for our chemistry effort. We aimed to improve the physicochemical properties of the series by focusing on the benzyl amide on the right-hand side of molecule. Interestingly, compound 2, with a simple replacement of the benzyl amide with a pyridine amide, exhibits potent inhibition vs TNKS 1 and 2 (IC₅₀: 4 nM and 39.8 nM, respectively) and >3 log units selectivity over PARPs (IC₅₀: PARP = 25 μM, PARP2 = 20 μM) and slightly improved ligand efficiency (LE) of 0.26.¹⁸ Moreover, compound 2 shows significantly enhanced cellular potency, with an IC₅₀ value of 5.0 nM.¹⁹ However, the PK profile of compound 2 was not optimal, exhibiting high clearance, short half-life, and limited bioavailability (vide infra), thereby justifying further optimization. Addition of a CF₃ group to the cyanobenzyl group led to compound 3, which maintained similar potency. The modification and substitution of the core ring system were largely tolerated. Monosubstitution at the 5-position of the quinazolindione ring provided equally potent analogues (4–7) and disubstitution of 5,6-positions (8–10) showed no further advantage. Replacement of the core ring with a heterocyclic ring caused a slight loss of potency (12–14, Figure 2). Replacement of pyridine in the amide side chain with pyrazine, substituted pyridine, and imidazole (15–17) maintained the potency. The replacement of CF₃ group in the benzylamine chain provided equal potent analogues (18, 19) as compound 7. The majority analogues prepared had very high molecular weight (approaching or over 600) and low LE (<0.3).

Table 1. SAR^a.

^aThe values are the geometric mean of at least two determinations. All individual values are within 25% of the mean.

Figure 2.

Structures of compounds 12–14.

A breakthrough was realized when we tested the carboxylic acid (20), an intermediate of compound 3 (Table 2). Surprisingly, compound 20 shows a slight loss in potency in the biochemical assay (TNKS1 (IC₅₀: ∼8 nM) and TNKS2 (IC₅₀: ∼80 nM)) and in the cellular assay (IC₅₀: 63 nM) compare to the parent compound 3 but demonstrates improved physicochemical properties and improved LE (0.32). Additional truncated acid analogues (21–26) were prepared with substitutions in benzyl amide and the 5-position of the core quinazolindione ring, and all show potent and selective activities with acceptable LE. Compound 21 is the most potent among them and has an overall balanced profile. It exhibits IC₅₀ values of 4 and 63 nM against TNK1 and 2 in the enzymatic assay, respectively, and an IC₅₀ of 20 nM in the cellular assay. Moreover, compound 21 inhibits proliferation of A549 and H292 cell lines with IC₅₀ values of 39.5 and 12.8 nM, respectively. It is known that the of cyclohexane carboxylic acid can be epimerized via a plausible acyl-CoA mediated cis/trans isomerization through a cyclohexylidene (acyl-CoA) methanolate intermediate.^20,21 In this case, we also observed the epimerization of carboxylic acid of 21 in PK study. Interestingly, although the cis analogue of 21 is active in the enzymatic assays (TNKS1 and 2, IC₅₀s of 316 and 100 nM, respectively), it shows poor cellular activity (IC₅₀ > 5 μM). Therefore, the isomerization was deemed to be inconsequential to further understanding of PK/PD relationship.

Table 2. SAR (Continued)^a.

					IC50 (nM)
compd	R1	R2	MW	LE	TNKS1 FP	TNKS2 FP	β-cat luc
20	H	H	485.5	0.32	7.9	79.4	63.1
21	CH3	H	499.5	0.32	4.0	63.1	20.0
22	OCH3	H	515.5	0.30	8.1	125.9	20.0
23	F	H	503.5	0.28	39.8	316.2	63.1
24	Cl	H	519.9	0.30	15.8	125.9	63.1
25	CH3	F	517.4	0.30	6.3	79.4	20.0
26	CH3	CH3	513.5	0.29	15.8	79.4	251.2

^aThe values are the geometric mean of at least two determinations. All individual values are within 25% of the mean.

Selected compounds from this series were examined in preclinical pharmacokinetic experiments (Table 3). In rat, compounds from the quinazolindione amide subseries (e.g., 2 and 19) exhibit high plasma clearance, high volume of distribution, and short t_1/2 with limited oral bioavailability. However, the truncated carboxylic acid subseries (e.g., 20 and 21) show excellent PK profiles, low clearance, and long t_1/2. Bioavailabilities of 20 and 21 are good with excellent oral exposures. We also examined compound 21 in other preclinical species, such as mouse and dog, and the PK profiles were similarly favorable in both species.

Table 3. Pharmacokinetics of Selected Compounds^a.

		dose
species	compd	iv	po	CL	Vd	t1/2	DNAUCpo	F
ratb	2	0.5e	10	91.2	4.42	0.72	37.7	3.6
	19	0.5e		45.9	2.1	0.62
	20	2	5	1.88	0.81	5.20	1293	14.5
	21	0.5	8.6	3.48	0.97	3.89	6651	133
mousec	21	1	3	1.44	1.34	10.9	5831	37.1
dogd	21	1	2	3.15	2.12		2937	52.2

^aDose: mg/kg; CL: mL/min/kg; Vd: L/kg; DNAUC_po: (ng*h/mL)/(mg/kg); t_1/2: hour; F: %, n ≥ 2. For compound 21: rat iv formulation: 1.00 mg/mL in DMSO:PEG400:Water (10:45:45), po formulation: 2.00 mg/mL in 1% HPMC/water(1:1). Mouse iv formulation: 1 mg/mL in 20%HP-beta-CD in saline, pH = 8, po formulation: 3 mg/mL in 0.5% HPMC. Dog iv Formulation: 0.25 mg/mL in 5% DMSO:20% HP-Beta-CD in saline, po formulation: 0.5 mg/mL in 5% DMSO:6% HP-β-CD in saline, pH 10.3.

^bSprague–Dawley rats, fasted.

^cMale C57 mouse, fasted.

^dMale beagle dogs, fasted.

^eiv: cassette dosing.

Given the overall balanced potency, selectivity, and PK profiles of compound 21, it was selected for further evaluation. Compound 21 is not a PgP substrate in the MDCK cell line, nor a P450 inhibitor (IC₅₀ > 100, 84, 36, >100, >100 μM for 1A2, 2C9, 2C19, 2D6, and 3A4, respectively) and showed no time dependent inhibition (TDI). The compound is also selective in our selectivity panel that consists of >40 liability enzymes and receptors. Furthermore, compound 21 is not mutagenic when tested in a bacterial mutation screening assay (Ames test) with Salmonella typhimurium TA1535, TA1537, TA98, TA100, and Escherichia coli WP2uvrA(pKM101) in the presence and absence of S9-mix.²²

To further assess the effects of compound 21 on tumor growth inhibition in vivo, we investigated the dose response of 21 in an H292 lung cell subcutaneous xenograft model in Balb/c nude mice. In this study, mice treated with 21 (10 and 30 mg/kg poqd, the formulation used in the xenograft study is a suspension formulation similar to the formulation used in the PK study, except 2% Tween was added to improve solubility) exhibited a significant reduction of tumor volume compared to vehicle control, and compound was well tolerated at concentration tested. Cisplatin at 3 mg/kg (ipqw, used as a positive control) and 21 at 3 mg/kg (poqd) showed similar and modest tumor growth inhibition. However, no difference in efficacy was observed between 10 and 30 mg/kg, suggesting saturation of compound availability or maximum potential target engagement at 10 mg/kg by compound 21 (Figure 3). The PK analysis result indicated that compound 21 at 30 mg/kg qd was almost completely cleared from plasma between doses. The AUC value of compound 21 in plasma was positively correlated with the dose. In conclusion, compound 21 at the dose of 10 and 30 mg/kg, poqd displayed superiority in H292 xenograft tumor growth inhibition as compared with vehicle and cisplatin group without any obvious toxicity.

Figure 3.

Antitumor effect of cisplatin (3 mg/kg ipqw), compound 21, 3, 10, and 30 mg/kg poqd, against H292 xenograft tumors in nude mice. For detailed statistics of this study, please see Supporting Information.

Compound 21 was examined in the 7-day rat tolerability study. Compound 21 was delivered orally via a spray-dried dispersion with 24.71% drug load at doses of 10, 100, and 300 mg/kg/day for 7 days in male Wistar Han rats. Although compound 21 was tolerated at the low dose (10 mg/kg/day) with 8% body weight loss at day 7, higher doses (100 and 300 mg/kg/day) were not tolerated. On days 6/7 in rats dosed with ≥100 mg/kg/day, multiple signs of poor tolerability were observed, including decreased activity, loss of skin elasticity, hunched posture partially/half closed eyes, irregular breathing, staining (anal area, around mouth), and/or abnormal fecal consistency (mucoid, soft, watery). Microscopic findings in rats given ≥10 mg/kg/day were noted in multiple organs including the GI tract. Gastro-intestinal (GI) toxicity with other TNKS inhibitors have also been reported previously.^15,23,24 It was suggested that the observed GI toxicity may be due to mechanistic toxicity of TNKS inhibition. Although it may be possible to identify tankyrase inhibitors with a reasonable therapeutic window, the GI toxicity may cause challenging clinical development.

The compounds in Table 1 were prepared straightforwardly according to the route depicted in Scheme 1. The benzoxazinedione core (27) was prepared by the action of triphosgene with aminobenzylic acid. Addition of benzyl amine and following up by triphosgene treatment afforded the key intermediate 29. Fisher esterification provided compound 30. Alkylation with either the commercially available or readily accessible benzyl bromide yielded the ester derivatives 31. Hydrolysis of ester delivered the desired carboxylic acid analogues (32). The amide analogues can be easily prepared by coupling the acids 32 with the corresponding amines.

Scheme 1. General Synthetic Route.

In summary, the compounds disclosed in this paper represent a new structural class of tankyrase inhibitors with excellent potency on TNKSs and selectivity against PARP1/2 enzyme. The truncated quinazolindione carboxylic acid subseries possess good potency in cell, PK properties, and excellent selectivity. Compound 21 exhibited excellent potencies in cell and antiproliferation studies, great selectivity, in vitro selectivity, and an excellent PK profile. In the in vivo animal model, compound 21 also showed better antitumor activity in H292 xenograft tumor growth inhibition in nude mice as compared to cisplatin groups. However, compound 21 exhibited significant multiorgan toxicity in a 7-day rat tolerability study which precluded its further developed. Nonetheless compound 21 can serve as an excellent in vitro and in vivo tool molecule, along with other structurally distinct TNKS inhibitors, to further investigate the biological role of TNKS inhibition and its therapeutic utility in various diseases.

Glossary

Abbreviations

TNKS

tankyrases

PARP

poly ADP-ribose polymerase

TRF-1

telomere repeat binding factor-1

FP

fluorescence polarization

PK

pharmacokinetics

IC₅₀

half-maximal inhibitory concentration

SAR

structure–activity relationship

IWR

inhibitors of Wnt response

CL

clearance

Vd

volume of distribution

DNAUC

dose normalized area under curve

IV

intravenous

IP

intraperitoneal

po

oral

qd

once a day

qw

once a week

HPMC

hydroxypropyl methylcellulose

PEG

polyethylene glycol

HP-β-CD

2-Hydroxypropyl-β-cyclodextrin

GI

Gastro-intestinal

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.1c00160.

• The TNKS1 and 2 fluorescence polarization (FP), β-catenin luciferase assay conditions, and synthesis and data for compound 21 (PDF)

Author Present Address

^∥ D.Q.: Phaeno Therapeutics Co., Shanghai, 200131, China.

Author Present Address

^⊥ S.L., J.E.: Janssen Pharmaceuticals, Spring House, Pennsylvania 19002, United States.

Author Present Address

^# S.K.: Novel Human Genetics Research Unit, GSK, Collegeville, Pennsylvania 19426, United States.

Author Present Address

^∇ J.F., K.R.: IVIVT-Non-Clinical Safety, NCS, GSK, Collegeville, Pennsylvania 19426, United States.

Author Present Address

^○ A.O., M.C.: R&D, GSK, Collegeville, Pennsylvania 19426, United States.

Author Present Address

^◆ D.S.S.: SciNeuro, Philadelphia, Pennsylvania 19106, United States.

Author Present Address

^□ R.K.: RMK Drug Discovery Consulting, LLC., Pennsylvania 19333, United States.

The authors declare the following competing financial interest(s): All GSK authors are employees or formal employees of GSK and GSK shareholders.