Trisubstituted thiazoles as potent and selective inhibitors of Plasmodium falciparum protein kinase G (PfPKG)

Highlights

• A series of thiazole compounds inhibit the malarial kinase PfPKG.

• SAR development of this series led to compound 23.

• 23 has a biochemical potency of 2 nM and an in vitro cellular potency of 113 nM.

• 23 also possesses good in vitro ADME and excellent kinase selectivity.

Abstract

A series of trisubstituted thiazoles have been identified as potent inhibitors of Plasmodium falciparum (Pf) cGMP-dependent protein kinase (PfPKG) through template hopping from known Eimeria PKG (EtPKG) inhibitors. The thiazole series has yielded compounds with improved potency, kinase selectivity and good in vitro ADME properties. These compounds could be useful tools in the development of new anti-malarial drugs in the fight against drug resistant malaria.

Introduction

Malaria causes up to 450 000 deaths every year, most of which occur in sub-Saharan Africa.¹ Most of the malaria mortality is caused by Plasmodium falciparum (Pf), one of the four species known to infect humans. Transmitted through the bites of Anopheles mosquitoes, the symptoms include fever, headache, chills and vomiting and if not treated quickly, anaemia, respiratory distress and multi organ failure. The population groups who are at highest risk are young children and pregnant women. Chloroquine was the first line of treatment for malaria for many years; however due to the development of resistance, its use has been substituted by drugs, the current standard of care being artemisinin-based combination therapy (ACT). However, parasite resistance to ACTs in South East Asia has been detected and is predicted to grow; therefore development of other treatments is highly desirable².

One promising antimalarial drug target that is being investigated is Pf cGMP-dependent protein kinase (PfPKG), an enzyme which is conserved to a high degree of homology in other species of therapeutic relevance such as P. vivax and P. berghei. This protein kinase is essential in all the key phases of the Pf parasite life cycle and in the blood stage, inhibiting PfPKG stops replication by preventing merozoite egress³ and invasion.⁴ First generation PfPKG inhibitors have also been shown to block gametogenesis⁵ and ookinete motility⁶ – two key developmental events required for mosquito-borne transmission. It has also been shown that PKG is essential for sporozoite motility and liver cell invasion⁷ as well as late liver stage development,⁸ making PfPKG an attractive target for new classes of anti-malarial drugs.

Herein, we will discuss PfPKG inhibitors derived from PKG inhibitors (compound 1 and 2) developed to treat Eimeria infections in poultry⁹ (Fig. 1). Both of these compounds showed low nanomolar potencies in a biochemical assay against PfPKG¹⁰ and moderate potency in an in vitro cell viability assay, the hypoxanthine incorporation assay (HXI).¹⁰

This paper is focused on the monocyclic compound 1, containing a pyrrole and an unflanked 4-pyridyl, both considered undesirable motifs for further SAR development. Furthermore, poor kinase selectivity was seen with 1, as it also showed potent activity against several other human kinases. Due to these unfavourable properties of 1, an alternative core was sought for further analogue development with the aim of enhancing anti-parasitical activity against PfPKG.

Literature searches on similar structures to compound 2 had showed examples where the imidazopyridine core could be swapped for a thiazole core, while still maintaining the interactions formed by the appended groups.¹¹ In addition, homology modelling (data not shown) indicated a good fit for our proposed compounds. With this knowledge, we swapped the monocyclic pyrrole core to a thiazole core, and the pyridine for the 2-aminopyrimidine from compound 2, to give thiazole 3 (Fig. 2).

Compound 3, when tested, showed similar biochemical potency and a slight drop in cellular potency when compared to compounds (1) and (2) (Fig. 2), which was seen as a positive result for the changed thiazole core.

To optimize the potency, we first examined the pendent 2-aminopyrimidine (Scheme 1, Table 1).

Alkylation of 4 with benzoate 5 was achieved using LiHMDS to give ketone 6. This was then reacted with (Me)₃SiCl and (ⁿBu)₄NBr to yield the α-chloro ketone in situ, which was cyclized with the Boc-protected thioamide in refluxing ethanol to give 7. Deprotection of 7, and subsequent reductive amination gave 8, and on oxidation, the resulting sulfone could be displaced by the appropriate amine to yield 9a-d. Two alternatives to the N-methylpiperidine ring were also made. 9e was synthesized following sulfone oxidation of 7, with Boc deprotection occurring in the displacement step by the amine. 9f was synthesized from tert-butyl 3-carbamothioylpyrrolidine-1-carboxylate and 6 using similar chemistry.

Introduction of a methyl group on the amino pyrimidine (9a) had little effect on potency but extending to the phenylpiperazines (9b-c), gave a 10-fold improvement in both biochemical and cellular potencies. The more basic phenylpiperidine (9d) gave a further boost in potency with an IC₅₀ of 300 pM in the biochemical assay and a cellular potency of 25 nM. Not converting the piperidine N-Boc to methyl allowed access to 9e, while swapping the piperidine group of the thiazole 2-position for the pyrrolidine (9f) gave an equipotent compound to 9b. With the generation of an apo X-ray structure we were able to model the compounds into the ATP binding site of PfPKG. This revealed the key interactions made by compound 3 and the rationale for improved IC₅₀ values of compounds 9b-f. As shown in Fig. 3, the 4-fluorophenyl is deeply buried in a hydrophobic pocket between the catalytic lysine (K570) and the small gatekeeper residue (T618).¹² The piperidine attached to the thiazole core projects out towards solvent, engaging in charge interactions with D682. The 2-aminopyrimidine forms a pair of H-bonds with the backbone of V621 on the hinge and offers a vector for growth from that amine. The pendent phenyl ring forms edge-face aromatic interactions with Y822 and positions the terminal amine to form charge interactions with D628 at the entrance to the pocket.

The kinase selectivity of 9c when measured against a panel of human kinases at a concentration of 100 nM (Fig. 5) showed extensive inhibition of several kinases across the panel. It was thought that selectivity could be achieved via the gatekeeper or DGF loop. Therefore, with the phenyl piperazine in place, variations were then explored around the 4-fluorophenyl portion of the molecule. It was postulated that due to the proximity of this group to the threonine gatekeeper residue,¹² which is much smaller than the corresponding residues in most human serine/threonine kinases,¹⁴ suitable changes here might enhance the selectivity of the compounds (Table 2).

Methyl derivative (10) and cyclopropyl, (11) were synthesized using the same route as shown in Scheme 1, from the appropriate ester starting material. The syntheses of 16a-e are shown in Scheme 2, starting with the methyl m-(methylthio)benzoate 12, which was elaborated via similar chemistry to intermediate 15. Compound 15 then underwent a double SMe oxidation to the bis-sulfone with hydrogen peroxide and catalytic sodium tungstate, followed by displacement of the (methylsulfonyl)pyrimidine by the requisite amine.

Replacement of the 4-fluoropenyl moiety with alkyl substituents gave rise to weakly active analogues (10, 11) which both showed a significant drop in biochemical potency when compared to 9c. The lower activity seen with the alkyl substituents could be attributed to their inability to sufficiently fill the hydrophobic pocket between the catalytic lysine (K570) and the small gatekeeper residue (T618) (Fig. 3). Despite the binding potency of 11, it showed similar cellular potency to 9c, possibly resulting from poor kinase selectivity as 11 is capable of binding to kinases in the cell with larger gatekeepers.¹⁴ Introduction of the sulfone (16a) gave a compound with comparable IC₅₀ values to 9c, but with a much improved kinase selectivity profile (Fig. 5).

To further enhance the kinase selectivity of the compounds, additional analogues were made with groups of greater polarity in an attempt to capitalize on additional interactions with the PfPKG enzyme. The new analogues were synthesized with smaller alkyl groups on the pyrimidine ring instead of the large phenyl piperazine group (Table 3).

The synthesis of sulfonamide 23 is highlighted in Scheme 3.

The nitroester (17) was hydrogenated to the aniline, and protected as the methyl carbamate to yield 18 which was subjected to subsequent ketone formation, then chlorination and cyclization to give thiazole 20. Boc deprotection and reductive amination, followed by hydrolysis with sodium hydroxide gave the unprotected aniline, 21. Sulfonylation of 21 was carried out with two equivalents of methanesulfonyl chloride to give firstly the bis sulfonamide, which on exposure to excess sodium hydroxide, yielded the mono sulfonamide 22. Oxidation to the sulfone and displacement with cyclopropylmethylamine yielded 23.

Replacement of the phenyl piperazine by a hydrogen to give the unsubstituted amino pyrimidine (16b) showed a drop in potency, due to the loss of the aforementioned charge interaction with D682. Small alkyl substituents on the amino pyrimidine with the methyl sulphone (16c-e) all showed good potency in the biochemical assay but this did not translate into good potency in the cell assay.

The fluorosulfonamide (23) was the most potent analogue in the biochemical assay and docking of this compound (as shown in Fig. 4) revealed the additional polar interactions responsible. The cyclopropylmethyl efficiently fills the small hydrophobic pocket between V621 and Y822, whilst the sulphonamide forms H-bonds with the DFG loop and a charge interaction with K570.

Additionally 23 showed very good selectivity against human kinases, likely arising from the small threonine gatekeeper residue as described earlier. Compound 23 and two additional key examples were then profiled in a number of in vitro ADME assays (Table 4). Data for 9c showed a very good overall profile, good logD and stability along with good PAMPA and kinetic solubility. Despite an otherwise excellent profile, the LogD of 16a was low when measured, potentially contributing to the poor permeability seen. Compound 23 was found to be metabolically stable in human and mouse liver microsomes, and also showed good permeability (Table 4). In transforming 9c (cLogP 3.76 pIC₅₀ 9.16 LLE¹⁵ 5.40) to 23 (clogP 2.83 pIC₅₀ 8.70 LLE 5.87), we were able to reduce reliance on potency gained from the large, lipophilic phenylpiperazine, and instead improve alternate polar interactions.

In summary, several PfPKG inhibitors have been developed and synthesized around the new thiazole core with several possessing low nanomolar biochemical potencies and cellular potencies around 100 nM. Excellent selectivity was achieved over human kinases through modification of the 4-fluorophenyl group to more polar substituents such as 23, removing previous inhibition seen against p38a MAPK and PKA kinases. Furthermore, the anti-parasitical activity of the compounds has been enhanced greatly, as 23 shows a 4-fold improvement in cellular potency over compound 1. Presently, the thiazole series is being further developed to improve anti-parasite activity.