Potent inhibitors of malarial P. Falciparum protein kinase G: Improving the cell activity of a series of imidazopyridines

A

Malaria is one of the most prevalent infectious diseases of the developing world whose causative agent in humans is the protozoan parasite Plasmodium, with most deaths caused by the P. falciparum species.Despite being largely preventable and treatable, it is currently responsible for almost 0.5 million deaths per year, with young children and pregnant women in sub-Saharan Africa particularly at risk. 1 The disease also continues to present significant public health policy, social and economic challenges in developing world countries, 2 with mounting concern about the rapid spread of resistance to current standard antimalarial drugs.The development of structurally and mechanistically novel malaria treatments is urgently required to add to the control tools and advance eradication of the disease. 3here is a growing body of evidence to suggest that members of the P. falciparum kinome play important roles in multiple stages of the parasite lifecycle. 4,5Among these, P. falciparum cGMP-dependent protein kinase (PfPKG) is attracting considerable interest as a promising antimalarial drug target.Chemical inhibitors of this enzyme have demonstrated prevention of merozoite egress 6 and invasion, 7 gametogenesis, 8 ookinete motility, 9 liver cell invasion 10 and sporozoite motility. 11These data continue to suggest that PfPKG remains an appealing drug target for developing new anti-malarial small molecules.
We have recently reported studies on thiazoles 12 (derived from compound 1 13 ) and on imidazo [4,5-b]pyridines as potent and selective inhibitors of PfPKG. 14Using the known compound 2 15 as a starting point, we prepared new bicyclic analogues which displayed potent inhibitory activity against the enzyme and in vitro blood stage antiparasite activity, good selectivity against human kinases and significant in vivo target-driven efficacy. 14However, important ADME parameters were thought to remain outside desirable ranges in some cases.Our aim was to investigate key structural motifs in 2 in ways that would address these important physicochemical property considerations, whilst maintaining cell potency and lipophilic ligand efficiency. 16e considered that three important initial areas of focus on the structure of 2 would aid the achievement of these objectives.The pyrimidine and its 2-substituent offered opportunities to influence potency and to probe the hinge binding motif against recent crystallographic data 14 (Figure 1 -A), drawing on our recent studies of a closely related sub-series based on a thiazole core. 12A second aspect involved relocating the tertiary amine substituent from its position on the bicyclic scaffold to within an extended aminopyrimidine (Figure 1 -B).Finally, alternative bicyclic cores (Figure 1 -C), some containing additional nitrogen atoms, could contribute to lowering lipophilicity.Our previous experiences with one such related scaffold [17][18][19][20] suggested that more straightforward synthetic access might also be realised.Herein we report our initial efforts in these areas of work and show how each guided the development of new SAR understanding towards improved in vitro profiles as described above.
We first prepared compounds with which to probe the proposed bidentate 2-aminopyrimidine hinge binding motif.The 4-aminopyrimidine isomer 5 was obtained from 4-thiomethyl-6-methylpyrimidine 3 21 by means of a three-step conversion to intermediate 4, transformation of the thiomethyl motif and introduction of the diaminomethyl side chain (Scheme 1).Regioselective iodination of intermediate 6 22 , followed by mesylation of the alcohol, displacement with dimethylamine and coupling with the appropriate boronic acid gave target compounds 7-10 in good yields.A larger aryl piperazine substituent 12 could be appended to aminopyridine 10 by means of palladium-catalysed arylation, followed by N-deprotection to give 11.These structural alterations significantly influenced levels of enzyme inhibition (Table 1).The regioisomeric aminopyrimidine 5 was much less active compared to 2 in both biochemical and in vitro blood stage anti-parasite activity (data not shown) assays. 23The known unsubstituted pyrimidine 7 15 showed some recovery against the enzyme.Pyridine 8 was also nearly 40-fold less biochemically active, and other heterocycles such as 9 showed no improvement. 24Interestingly, little change was observed on the re-introduction of an amino group in 10.Introduction of an aryl piperazine motif in 11 resulted in a further drop in potency, in stark contrast to our previous observations on similarly extended aminopyrimidines in the thiazole series. 12These data suggested that the 2-aminopyrimidine motif in 2 was likely to provide an optimal interaction with the hinge region of the enzyme.
Docking of key examples 2, 8 and 11 into a recently published PfPKG crystal structure (PDB:5DYK) 25 provides a possible rationale for understanding these SAR variations.As shown in Figure 2, the bicyclic core of each compound occupies a similar position, directing the 4fluorophenyl group into a hydrophobic pocket adjacent to the gatekeeper residue (T618), and positioning the dimethylaminomethyl group in 2 and 8 to enable charge interactions with two acidic residues at the solvent interface (E625 and D682).Classical bidentate hydrogen bond formation to hinge residue V621 contributes significantly to the high affinity of 2 and explains the lower affinity of 7 and 8.However, pyridine 8 requires a larger rotation out of plane from the bicyclic core (32.4°deviationcalculated for 8, 16.5°deviation for 2), in order to avoid steric clash between hydrogen atoms.Introduction of the larger phenyl piperazine in 11 appeared to allow an edge-face aromatic interaction with Y822 and a relatively long-range charge interaction with E625 at the pocket mouth.Formation of these interactions had been proposed as a route to obtaining a boost in potency in the thiazole series. 12In the current case, any benefit is more than offset by realignment to accommodate this bulky group; this results in weakening of other key interactions, among which are hinge binding to V621 and charge interaction between the N,N-dimethylamine group and E625.It also appears that re-introduction of a bidentate hinge-binding motif in aminopyridine11 is not sufficient to regain affinity.Though this favours a more coplanar arrangement of the bicyclic core and hinge-binding heterocycle, steric clash between the two rings and the increase in strain energy largely negates any affinity gain.This is reflected in an intermediate calculated dihedral angle (26.2°) for 11.Hence the current data suggested retaining the amino pyrimidine motif as a more appropriate choice of hinge binding motif, and one which would best allow for effective substituent growth towards the mouth of the binding pocket in future analogues.
We turned our attention to the benzylic tertiary amine group attached to the bicyclic core, and hypothesised that re-positioning it to become part of the aminopyrimidine substituent would be beneficial for metabolic stability whilst maintaining potency.In doing so, it was thought that the deletion of both a tertiary amine and a labile benzylic methylene group would remove two likely sites for primary in vitro metabolism.In addition, larger aminopyrimidine substituents had previously been shown to provide significant additional in vitro potency, 12 so we wished to explore whether a larger substituent containing a basic centre or other polar group could provide both potency and microsomal stability.The required intermediates 12 or 13 could be assembled using similar chemistry to that described above.The sulfide 12 could be oxidised and easily displaced with more reactive amines, as shown in Scheme 2. For less nucleophilic substrates, alternative acidcatalysed displacement conditions using the chloropyrimidine 13 proved to be more suitable, allowing rapid and efficient preparation of the desired analogues.
We were encouraged to find that early examples such as the simple alkyl amine 14 showed a promising level of in vitro potency (Table 2).Conformational constraint in non-basic 15 further supported this approach, improving activity and maintaining a good ADME profile.By extending further via a phenyl linker to give aryl piperazine 16, further significant boosts in both biochemical and anti-malarial activity in a blood stage hypoxanthine incorporation (HXI) cell-based assay were obtained.While metabolic stability for 16 remained reasonably good, other aspects of physicochemistry were likely to be driven by poorer solubility.Adjusting basicity (as in piperidine 17), or combining with a change in vector (e.g.pyrazole 18) provided only isolated improvements in ADME properties (eg.solubility for 18). 26The trend of microsomal stability for the aryl aminopyrimidines 16-18 was unexpected given their measured logD values, for which there was no convincing explanation.Taken together, these data suggested that this type of aryl linked analogue was in many cases able to provide very good levels of potency, but that a well-balanced overall in vitro profile was not easy to locate in compounds lacking the basic substituent on the bicyclic core.
To assess the impact of altering the bicyclic core, we selected and prepared several matched-pair analogues in related [6,5-] ring systems.The imidazopyrimidines 19 and 20 could be synthesised from the regioisomer of 3 and an appropriate aminopyrimidine in the same way as described for 5, with separation of the regioisomeric bicyclic intermediates.Benzimidazole 23 could be prepared by oxidation and displacement of intermediate 21 23 , followed by reduction of the Weinreb   The imidazopyridazine 26 could be accessed by varying a previous chemical approach. 17Conversion of bromoketone 24 15 to intermediate 25 was followed by oxidation of the thiomethyl group and reaction with ammonia (Scheme 4).Then displacement of the 2-chloro group with the correct amine gave the target 26.Finally, condensation of the alkynylpyrimidine 27 27 with an N-aminopyridinium salt gave intermediate 28 and its (undesired) regioisomer.Transformation of the desired regioisomeric product using known chemistry provided the pyrazolopyridine 29.
These modifications to the bicyclic core once again resulted in considerable variation in in vitro activity and ADME profile.The imidazopyrimidine examples 19 and 20 could not in either case match the biochemical activity, anti-parasite activity or lipophilic efficiency of 2, though both possessed similar ADME profiles (Table 3).Some potency was regained in benzimidazole 23, as were further improvements in lipophilicity and metabolic stability, though passive permeability was lower.Whilst offering improvements in synthetic tractability and molecular diversity, 17 imidazopyridazine 26 was significantly less potent and passively permeable.Encouragingly, the pyrazolopyridine 29 regained the best levels of in vitro activity and lipophilic efficiency, though further improvements to a promising ADME profile proved difficult to locate in additional examples, particularly with respect to metabolic stability.
Computational calculations again suggested a plausible framework by which to understand these data.Each bicyclic system bears a conserved nitrogen atom in the 5-membered ring, but this does not appear to engage the side chain of residue K570 in a bonding interaction in any case.A similar observation was made for a crystal structure of the P. vivax PKG enzyme in complex with 2. 14 Generation of electronegative field isosurfaces 28 revealed subtle influences on the hinge binding interaction between the pyrimidine nitrogen atom and the backbone of residue V621.This important interaction was relatively unperturbed for 2 (and also for 29 -data not shown), but appeared to be significantly weakened in the case of benzimidazole 23 (Figure 3).Although a smaller effect seemed to affect the imidazopyrimidines 19 and 20, higher desolvation penalties due to increased polarity could explain the lowering of affinity for those molecules.The imidazopyridazine 26 showed an electronegative field isosurface very similar to 2 and 29, but two additional factors may contribute more significantly to its lower  affinity.Firstly, the electrostatic repulsion between the nitrogen atoms on the core and pyrimidine ring appears likely to deplanarise that part of the molecule, weakening key interactions in doing so.Secondly, the pendent amine employed may result in less optimal positioning of the charged group, despite increased polarity and flexibility.This report has described our initial efforts to develop a series of bicyclic PfPKG inhibitors, based upon a previously described chemical starting point.By focusing on key sub-structural motifs, we have both confirmed and enhanced important SAR and physicochemical trends, resulting in key examples such as 16 and 29.These two compounds showed good levels of biochemical and in vitro anti-malarial activity; 29 in particular possessed equivalent mLogD, improved passive permeability and retained a good level of lipophilic ligand efficiency.Docking and computational analyses provided additional insight into the origins of the observed structure activity relationships.Studies to refine, expand and apply this new knowledge to the preparation of further improved compounds will be reported in due course.

Figure 2 .
Figure 2. Compounds 2 (left), 8 (centre) and 11 (right) docked into PfPKG, with protein surface coloured by electrostatic potential.Charge-based and H-bonding interactions and distances are shown in purple, the dihedral angles between the bicyclic core and the hinge-binding heterocycle are shown in red, and the edge-face interaction between Y822 and the phenyl ring in compound 11 is shown in light blue.

Table 1
Examining the hinge binding motif.
a nt = not tested.

Table 2
Relocating the polar bicyclic group.
a nt = not tested.b % remaining after 30 mins.