Driving antimalarial design through understanding of target mechanism

Abstract

Malaria continues to be a global health threat, affecting approximately 219 million people in 2018 alone. The recurrent development of resistance to existing antimalarials means that the design of new drug candidates must be carefully considered. Understanding of drug target mechanism can dramatically accelerate early-stage target-based development of novel antimalarials and allows for structural modifications even during late-stage preclinical development. Here, we have provided an overview of three promising antimalarial molecular targets, PfDHFR, PfDHODH and PfA-M1, and their associated inhibitors which demonstrate how mechanism can inform drug design and be effectively utilised to generate compounds with potent inhibitory activity.

Introduction

Malaria is caused by an infection with the apicomplexan parasite of the genus Plasmodium. An estimated 219 million cases and 405 000 malaria-induced fatalities occurred in 2018 alone, with children under age 5 representing two-thirds of these incidences [1]. Resistance to existing antimalarials, particularly to WHO recommended artemisinin combination therapies (ACTs), has been attributed as one of the major reasons that progress toward malaria elimination has slowed in many developing nations [1,2]. The first clinical signs of resistance to ACT in Plasmodium falciparum (Pf) malaria was reported in 2004 in Cambodia and raised concerns as there are no alternative first-line treatments of Pf available [3,4]. The recurrence of resistance, coupled with the slowly rising incident rate (2% since 2015) means that the design of novel antimalarials needs to be carried out in an efficient and rational manner.

For the last few decades phenotypic-based high-throughput screening (HTS) has been particularly useful for neglected and orphan diseases [5]. Molecules from these screens are typically first-in-class medicines often with unknown mechanism of action (MoA) that play an important role in the design of follower drugs [6]. Phenotypic screening can reduce the time between candidate identification and preclinical trials, and can guarantee that a molecule has the ability to permeate membranes [5]. These candidates can struggle in later stages of development as limited mechanistic information restricts the capacity to re-design if toxicity or selectivity issues arise [7]. In contrast, target-based campaigns rely heavily on knowledge of the target structure, function and MoA to design new inhibitors. A detailed understanding of the structure activity relationships (SAR), achievable through data from in vitro target-based assays and structural biology techniques such as X-ray crystallography and cryo-electron microscopy, allows researchers to fine-tune a drug candidate to optimise the desired properties. When structural data is available, target-based approaches can also utilise vast libraries of hit candidates via in silico screening, which require fewer resources than a quintessential HTS.

In this review, we have explored three examples of target-based campaigns, which are all currently in different stages of development, to show how knowledge of target mechanism in these programs have accelerated antimalarial design. Each program has taken advantage of structure-function data to inform the design of candidate molecules as well as to overcome disadvantages of the associated biological targets. Our first example is the pyrimethamine-resistant target dihydrofolate reductase (DHFR), which has undergone an extensive target-based program that has resulted in a number of inhibitors, including the Medicines for Malaria Venture (MMV) clinical candidate P218. Our second example demonstrates how structural data provided a route to selective and potent inhibitors of plasmodial dihydroorotate dehydrogenase (DHODH). Finally, we explore an early drug discovery program focused on the Plasmodium M1 aminopeptidases (PfA-M1) and the various structure-based campaigns which allowed for the design and synthesis of potent cross-species inhibitors. These examples aim to highlight that target-based drug design approaches play an important role in the future of antimalarial research.

Dihydrofolate reductase — A resistant enzyme made sensitive

Dihydrofolate reductase has a vital role in cell growth and proliferation by catalysing the reduction in dihydrofolic acid to tetrahydrofolic acid. Tetrahydrofolic acid is a folic acid (folate) derivative and is required in Pf for the synthesis of deoxythymidine monophosphate, a DNA monomer [8]. In protozoa, the DHFR protein encodes a thymidylate synthase (TS) domain resulting in a bifunctional enzyme, in contrast with human DHFR where the two enzymes are separate [9]. PfDHFR-TS (Figure 1A) exists as a dimer, with each monomer 608 residues in length, with an N-terminal DHFR domain of 231 residues (in cyan, Figure 1A) and a C-terminal TS domain of 288 residues (in orange, Figure 1A). The two enzymatic domains are connected by an 89 amino acid junction region (in pink, Figure 1A) [10]. The DHFR domain consists of an eight-stranded central β-sheet surrounded by seven α-helices. The wild-type (WT) active site includes Ile14, Ala16, Trp48, Asp54, Phe58, Ser108, Ile164 and Thr185, which interact with both dihydrofolic acid and NADPH, a cofactor that acts as an electron donor (Figure 1B) [10].

Figure 1. PfDHFR and inhibitors.

(A) Cartoon model of the X-ray structure of WT PfDHFR-TS dimer bound to Pyr, where one monomer is shown with DHFR domain in cyan, TS domain in orange and junction region in pink. The dimer partner is shown in grey. (PDB ID 3QGT) (B) Zoom of Pyr (in sticks with red/blue carbon atoms) bound to wild-type PfDHFR (in sticks and carbon atoms in aqua) and NADPH (in lines, carbon atoms magenta). Hydrogen bonds indicated by black dashed lines. (PDB ID 3QGT) (C) Binding conformation of P218 (blue/grey/red sticks) with quadruple mutant PfDHFR (aqua sticks) and NADPH (in lines, carbon atoms magenta), with hydrogen bonds indicated by dashed lines. Mutant residues are indicated with violet labels. (PDB ID 4DP3) (D) Pharmacophore of potent PfDHFR inhibitors. Effective inhibitors typically feature a nitrogenous heterocycle (blue) that interacts with Asp54 and a substituted aromatic ring (red) which has the ability to form hydrogen bonds with residues conserved between WT and QM PfDHFR. These two binding groups are linked by a flexible region (grey) at least 4-atoms long and can be hydrophobic or hydrophilic. (E) 2D structures of PfDHFR inhibitors, Asp54 binding motif in blue, linker region in grey and secondary binding motif in red.

Some of the earliest synthetic antimalarials targeted DHFR [11–15]. Cycloguanil (1) and pyrimethamine (Pyr) (2), date back to the 1940s [16,17] but long-term widespread use of these molecules has led to resistance through the evolution of multiple mutations in PfDHFR [18–20]. The problematic PfDHFR ‘quadruple mutant' (QM) has four substitutions of Asn51-Ile and Cys59-Arg as well as active site residues Ser108-Asn and Ile164-Leu (Figure 1C). The spread of resistance in parasites halted further development of antifolate antimalarials, with Pyr being combined with sulfadoxine in the 1960s to boost the efficacy of pyrimethamine.

The elucidation of the PfDHFR WT and QM crystal structures allowed for a renaissance of antifolate antimalarials [10]. Utilising X-ray crystallography in addition to mutagenesis techniques, the resistance mechanisms within the active site were identified and explored [22–24]. First generation antifolates (Figure 1B,E) have a para-chlorophenyl moiety that fits into the active site when Ser108 is present. PfDHFR QM has a mutation of Ser108 to Asn108, which causes a steric clash between enzyme and inhibitor, resulting in poor binding. Inhibitors that replaced the rigid aryl group with a flexible chain avoided steric clashes with position 108 and showed greater affinity for both WT and QM PfDHFR by interacting with residues further away in the active site [25,26]. A study of Pyr-resistant PfDHFR obtained from field isolates found that no strain had a mutation at Asp54, implying that this residue may be essential for enzyme activity [25]. Further investigations revealed that mutation of Asp54 to any amino acid except Glu, resulted in complete loss of enzymatic activity, indicating that the carboxyl group of Asp54 is required for enzymatic function [24]. The identification of Asp54 as an essential active site residue enables PfDHFR inhibitors to be designed with Asp54 acting as an anchor point.

More recently, development of antifolates (Figure 1E) have had a heavy focus on designing inhibitors to be effective in both Pyr-resistant and Pyr-sensitive strains of Pf. These campaigns utilise information obtained from the X-ray crystal structures and molecular dynamics to inform the inhibitor design. One of the earliest modern inhibitors was WR99210 (3), which was potent against both WT and QM PfDHFR. WR99210 featured a 2,4-diaminodihydrotriazine ring, which anchored the molecule to Asp54. The inhibitory activity of WR99210 against both WT and QM PfDHFR was due to the inclusion of a flexible 5-membered chain, which bridged the dihydrotriazine ring to a 2,4,5-trichlorophenyl moiety. The flexible linker provided the trichlorophenyl ring with access to a region deeper within the active site, whilst avoiding residue 108. However, WR99210 suffered from gastrointestinal tract toxicity and poor oral bioavailability, caused by basicity of the dihydrotriazine [27,28]. Saepua et al. [29] reported a range of pyrimethamine derivatives active against both WT and QM PfDHFR, including potent analogue 4. These compounds possess a similar pharmacophore to WR99210, with the pyrimidine heterocycle interacting strongly with Asp54. The 4-atom linker of 4 extends a meta-phenoxyacetic acid group further into the active site, allowing for hydrophobic interactions between Phe58, Phe116 and the phenyl ring. The carboxylate moiety of the phenoxyacetic acid forms charge-mediated hydrogen-bonds with Arg59 and an additional interaction with Arg122 via an intermediary water molecule [29]. Other PfDHFR inhibitory studies feature chemical classes other than dihydrotriazines and pyrimidines, however, still maintain the Asp54 interaction. Thakkar et al. reported a series of triazoles (e.g. 5) that displayed potent inhibitory activity against WT PfDHFR. The thiol interacts with Asp54, whilst being supported by additional π interactions between the triazole and active site. The two para-chlorophenyl groups extend into different regions of the active site to form various hydrophobic interactions [30]. Bhat et al. reported a series of 4-aminoquinoline-triazine bifunctional hybrids that showed moderate activity against chloroquine-sensitive (3D-7) and chloroquine-resistant (RKL-2) Pf. These compounds target DHFR with the triazine moiety, whilst the quinolone group is able to inhibit the conversion of haem to haemozoin. In DHFR, one of the phenols of 6 anchors the molecule to Asp54 as well as forming π–π interactions with adjacent Phe58 [31].

The current clinical candidate P218 (7), is a potent inhibitor of both WT and QM PfDHFR that is under development by MMV in partnership with Janssen and recently proved to have a good safety profile in healthy human volunteers [21]. P218 is based on a 2,4-diaminopyrimidine scaffold (7, blue) and a 3-phenylpropanoic moiety (7, red) connected by a 5-atom linker (7, grey) [28]. The selection of the pyrimidine was based on previously reported mechanistic information, acting as the Asp54 anchor. Furthermore, pyrimidine was chosen over dihydrotriazine due to the superior oral bioavailability of pyrimidines and the toxicity associated with dihydrotriazines [27,28]. Co-crystal structures of P218 bound to WT and QM (Figure 1C) PfDHFR revealed that the 5-atom linker steers the 3-phenylpropanoic acid group away from residue 108. The carboxylate of the 3-phenylpropanoic acid moiety forms charge-mediated hydrogen bonds with Arg122 and Arg59. Arg122 is conserved across QM and WT PfDHFR, with the aliphatic portion of the side chain highly restrained by surrounding residues [28]. This restrained side-chain serves as a rigid and reliable platform for ionic interactions to be formed. The replacement of PfDHFR Met55, Cys/Arg59 and Phe116 with structurally equivalent human DHFR Phe31, Gln35 and Asn64 resulted in the Arg side-chain restriction not being observed in human DHFR, which caused P218 to adopt a radically different conformation in human DHFR and does not form interactions with equivalent Arg70, providing necessary selectivity to the P218 [28,32]. The carefully considered design of P218 resulted a highly potent inhibitor of QM PfDHFR (K_i = 0.54 ± 0.12 nM), which displays good antiplasmodial activity against both WT and QM Pf (IC₅₀ = 4.6 ± 1.6 and 56 ± 20 nM, respectively) [28], and possesses a favourable safety and tolerability profile [21]. Target-based drug design has played a crucial role in the discovery and development of this promising clinical candidate.

PfDHODH — Structural data providing a route to selective inhibition

Plasmodium parasites rely entirely on the de novo biosynthesis pathway for their pyrimidine, an essential metabolite precursor for DNA and RNA synthesis [33–36]. Inhibition of pyrimidine biosynthesis results in parasite death in both the blood and liver stages, making the pathway an attractive target for drug development [37]. Dihydroorotate dehydrogenase is involved in the fourth step of the pyrimidine pathway and utilises flavin mononucleotide (FMN) and coenzyme Q (CoQ) to catalyse the oxidation of dihydroorotate (DHO) to orotate (ORO) [38]. DHODH enzymes exist as either cytosolic enzymes (class I) or membrane-associated (class II) [39,40]. Plasmodium DHODH are class II enzymes that form a conserved catalytic α/β-barrel (Figure 2A), composed of eight-strand β-barrel (in cyan) wrapped by eight α-helices (in tan), as well as a highly sequence variable hydrophobic N-terminal extension (in grey) [39,41]. This N-terminal extension forms a domain of two α-helices, located close to the top of the barrel and a tunnel-like pocket is formed as a result of the packing [41]. The active site of the enzyme is located within the barrel and contains the binding sites for FMN (Figure 2A, shown in yellow) [39]. Inhibitors of the enzymes can bind directly in the barrel blocking substrate access [41,42] or within the tunnel-like pocket near the N-terminal extension blocking the electron transfer between FMN and CoQ [41,43]. Targeting inhibitors to the tunnel binding site provides a route to selective enzyme inhibitors due to high sequence variability and conformational mobility of this region [41]. The tunnel binding site has three regions, a hydrophilic pocket comprising highly conserved residues His185 and Arg265; a hydrophobic pocket that contains key aromatics Phe188 and Phe227 involved in inhibitor positioning and finally a small hydrophobic channel leading toward FMN lined by Val532, Ile272 and Ile263 (Figure 2B–D) [41,44].

Figure 2. PfDHODH and inhibitors.

(A) The X-ray structure of PfDHODH (PDB ID 1TV5) demonstrating the central α/β-barrel (α-helices in tan, β-barrel in cyan) and the two α-helices of the N-terminal domain in grey. Substrates FMN and ORO are shown in stick with green and yellow carbons, respectively. The species-specific site of inhibitor binding is indicated by the box in black. (B) The inhibitor binding site from the co-crystal structure of PfDHODH and A77-1726 (PDB ID 1TV5), Pf-A77-1726 is shown in blue sticks, residue sidechains are shown in sticks, hydrophobic pocket in teal, hydrophilic pocket in orange and small hydrophobic tunnel in purple. In the black box the structures of Hs-A77-1726 and Pf-A77-1726 are overlayed to demonstrate the different binding orientations; A77-17266 is shown in sticks with Pf in blue carbons and Hs in red carbons. (C) 2D chemical structure of inhibitors discussed; hydrophobic region in teal, hydrophilic region in orange and small hydrophobic channel in purple. Also showing the hit-lead evolution starting with DSM1 to get to DSM265, the clinical candidate and the residues involved in binding. (D) The inhibitor binding site from the co-crystal structure of DSM265 bound to PfDHODH (PDB ID 4RX0) coloured as per B with DSM265 in magenta and hydrogen-bonds indicated by black dashes.

Development of potent Plasmodium-specific inhibitors was challenging prior to the resolution of the PfDHODH structure [45–47]. The first X-ray crystal structure of PfDHODH was solved bound to a human HsDHODH inhibitor (A77-1726) (Figure 2B) [41]. A77-1726 binds to the species-selective inhibitor site in both enzymes and is a potent inhibitor of HsDHODH but only a moderate inhibitor of PfDHODH (IC₅₀ = 0.261 and 190 µM, respectively) [43,44,48]. In HsDHODH, the trifluoromethylphenyl moiety of A77-1726 sits in the large hydrophobic pocket, and the polar carbonyl, nitrile and enol groups hydrogen-bond to Arg136, Tyr356, His56 and Gln47, respectively [41,49]. The species-selective inhibitor site in PfDHODH is significantly smaller than the human enzyme due to the presence of sequence changes, namely Met536, Phe174 and Phe188 that close the tunnel from different sides and substitutions of two Val residues with Ile at the end of the tunnel (Figure 2B) [41]. In PfDHODH, the orientation of Phe188 pushes the trifluoromethylphenyl moiety closer to the first helix resulting in an almost 180° rotation of A77-1726 about its long axis (Figure 2B, Pf-A77-1726 in blue and Hs-A77-1726 in red). This shift disrupts hydrogen-bonding that likely contributes to potency observed toward HsDHODH. The conserved His185 and Arg265 (Figure 2B in orange) hydrogen bond with the carbonyl and enol groups of A77-1726 in PfDHODH as opposed to the long-distance water-mediated contacts in HsDHODH [41]. The variation in the binding tunnels of the two enzymes suggests that ligand binding in PfDHODH is directed by hydrophobic interactions, notably stacking interactions with Phe188 and Phe227 (Figure 2B in teal), while the conserved residues form important hydrogen-bonds [41]. The methyl group of A77-1726 extends toward the small, hydrophobic cavity (in purple) [48].

Subsequent studies investigated derivatives of A77-1726 and other scaffolds with alternating hydrophilic head and hydrophobic tail groups, as well as exploring the small hydrophobic cavity but resulted in a series with only mid- to sub-micromolar activity [48,50–55]. Target-based HTS campaigns identified a thiophene-carboxamide as a new starting block for PfDHODH inhibitors [52]. Structure-based drug design of the initial thiophene-carboxamide hit delivered Genz-667348 (PDB ID 3O8A) that showed selective, low nanomolar activity toward PfDHODH [56,57]. The benzimidazole of Genz-667348 sits in the hydrophobic pocket forming edge-to-face stacking interactions with Phe188 (Figure 2C in teal), while N and O atoms of the amide moiety sit in the hydrophilic pocket and form hydrogen-bonds with His185 and Arg265, respectively (in orange) [56]. The cyclopropyl group of Genz-667348 occupies the narrow hydrophobic channel (in purple), and an increase in the ring size resulted in a loss of activity [56–58]. Despite promising pan-parasitic target inhibition in vitro, efficacy in vivo and acceptable ADME/PK, these inhibitors have not advanced beyond late lead development stage [56–59].

The compound DSM1 was also identified from a target-based HTS (Figure 2C) [47,60]. Early structural studies showed that the triazolopyrimidine moiety sat in the hydrophilic pocket (Figure 2C in orange) and the naphthyl group occupied hydrophobic pocket (in teal) [44]. Within the hydrophilic pocket, the bridging nitrogen N − 1 in the triazolopyrimidine inhibitors is able to delocalise electrons into the ring, establishing dipoles in the inhibitor that permit hydrogen-bonds between the bridging nitrogen N − 1 and His185, as well as the pyridine nitrogen N − 5 and Arg265 (shown in orange). This was identified as a key factor in the potency of this inhibitor class and replacement of N − 1 atom in DSM1 with either O or S results in complete loss of activity as these atoms are unable to induce this delocalisation effect [44,61]. In the hydrophobic pocket, the position of the naphthyl group shifted Phe188 to allow for edge-to-face stacking (in teal), and thus extending DSM1 further than the trifluoromethylphenyl of A77-1726. Subsequent structure-based design investigated replacing the naphthyl group in DSM1 with substituted phenyls and showed that the hydrophobic pocket prefers bulky para-substituted phenyl groups where 4-SF₅-Ph (as demonstrated by DSM265) represents the best balance between desirable potency, metabolic stability and plasma exposure [44,60]. Additionally, it was found that ortho- or meta-substituents resulted in a loss of activity likely due to the lack of space within the pocket [44]. Replacement of 4-SF₅-Ph with CF₃-pyridinyls in the hydrophobic pocket improved drug-like properties [62], however, unexpected off-target toxicity later halted the progress of this analogue [63].

DSM1 and the bound crystal structure allowed rapid progress to identify more potent compounds with desirable physicochemical properties. The C-5 atom of DSM1 provided a route into the narrow hydrophobic tunnel, and subsequent exploration identified that small substitutions at the C-5 position pick up additional hydrophobic interactions available in this cavity [44,64]. DSM265 is the outcome of this successful program and possesses desirable potency, in vivo efficacy, long half-life and excellent oral exposure [64]. It binds PfDHODH with triazolopyrimidine moiety occupying the hydrophilic pocket (in orange), the CF₂CH₃ substitution sitting in the hydrophobic channel (in purple) while the main hydrophobic pocket encapsulates the 4-SF₅-Ph group (in teal) (Figure 2C,D) [37,64]. DSM265 is currently being developed by MMV in partnership with Takeda and has shown promise in Phase Ia/b studies as a single-dose treatment and as a once-weekly chemoprophylaxis as well as Phase IIa efficacy studies in patients with Pf and Pv malaria infections [65–68]. DSM265 has completed Phase IIa clinical trials and a number of other PfDHODH inhibitors have also been designed and synthesised utilising the understanding of target mechanism [59,69–73]. Altogether, these efforts have not only validated PfDHODH as a clinical target but have resulted in desirable clinical candidates.

PfA-M1 — Flexible substrate pockets provide key to potency

The Pf M1 aminopeptidase (PfA-M1) is a clan MA zinc-dependent metalloaminopeptidase (MAP) which has been a potential antimalarial target of interest for over a decade [74–76]. PfA-M1 is involved in the terminal stages of haemoglobin digestion by facilitating the final stages of haemoglobin peptide hydrolysis into single amino acids [77–79]. Haemoglobin digestion is a semi-ordered enzymatic cascade that is essential for parasite proliferation. Plasmodium spp are unable to synthesise amino acids, so rely on this process heavily during the trophozoite and schizont phases of the intra-erythrocytic life-cycle. These phases require an abundance of amino acids to ensure sufficient parasite growth and to prevent premature haemolysis [80]. PfA-M1 is an essential enzyme [74] and is highly expressed throughout the parasite cytosol and digestive vacuole [81].

In Pf, PfA-M1 remains inactive until a 196 residue extended N-terminus is hydrolysed and the enzyme is released [82]. The X-ray crystal structure of the soluble, active enzyme has been extensively studied, with the tertiary structure being divided into four domains, consisting of 26 α-helices and seven β-sheets arranged in a bacterial aminopeptidase N-fold (Figure 3A) [81]. PfA-M1 possesses the two signature M1 aminopeptidase family motifs, the zinc-binding and substrate coordination motifs [74]. The substrate pool of PfA-M1 has also been investigated, with a broad specificity for bulky, hydrophobic and basic P1 residues being observed [83]. Early work to identify the effects generated by the inhibition of PfA-M1 was conducted using compounds from a bestatin-based library [79]. Selective inhibition of PfA-M1 in vitro resulted in a distinctive digestive vacuole swelling [79]. The change in vacuole morphology was dose-dependent and hypothesised to be caused by an excess of oligopeptides, which are too large to be exported to the cytosol. The outcomes of selective PfA-M1 inhibition in vivo have not yet been determined, however, studies using inhibitors that can target both PfA-M1 as well as the M17 aminopeptidase have shown that inhibition of these two enzymes can control parasite growth in murine models [84,85].

Figure 3. PfA-M1 and inhibitors.

(A) The X-ray structure of PfA-M1 (PDB ID 4ZX4) demonstrating four domains in yellow, green, red and dark blue. (B) A closeup of the co-crystal structure of PfA-M1 and MIPS1700 (12) showing the inhibitor binding site with the important amino acids involved in forming interactions; MIPS1700 in purple, S1 pocket in orange, S1′ pocket in blue, ZBG and catalytic triad in pink. (C) A number of inhibitors reported in literature; colour-coded per respective region. (D) The chemical evolution starting with the arginine mimetic (10) and ending with the most recent published inhibitor (12).

PfA-M1 is a zinc-dependent enzyme and to date, all inhibitors produced have targeted the essential Zn²⁺ in the active site. The binding pose of the inhibitors is largely determined by the zinc binding group (ZBG) coordination. The co-crystal structure of the early inhibitors show that the Zn²⁺ itself is co-ordinated to the enzyme by a catalytic triad of residues His496, His500, and Glu519 while the ZBG of the inhibitor is co-ordinated to the Zn²⁺ through metallo-bonds (Figure 3B, in pink) [74]. For instance, bestatin (Figure 3C, 8) [79] utilises its hydroxyl and carbonyl oxygen while other peptide-based inhibitors [85–87] utilise their carboxylic acid moiety for coordination. Other studies explored the use of phosphonic acids and aminophosphonates (e.g. 10 and 14) [88–91] and while they exhibited excellent ZBG properties for coordination, these moieties often suffer from permeability issues and were produced largely as tool compounds. As a result, more recent compounds utilise groups like hydroxamic acids (9, 11, and 12, this type of coordination is demonstrated in Figure 3B) [90–95] and geminal-diol (13) [96–98] for the coordination of Zn²⁺. Although targeting the Zn²⁺ is essential, the approach raises challenges for selectivity of the parasite target over human homologues — parasite specific inhibitors therefore require the employment of specific interactions with the PfA-M1 S1 and S1′ subsites [81].

The SAR of the S1 and S1′ pockets of PfA-M1 have been thoroughly explored within various campaigns. The S1 pocket (Figure 3B) is made up of several hydrophobic residues as well as Glu572, a flexible and polar residue in the pocket that permits conformational changes upon ligand binding to allow for larger side chains in this pocket [74,86,99]. In 2006, Flipo et al. [91] reported a series of malonyl-hydroxamic acids, similar to 9, probing the active site. Although the importance of metal chelation was understood, the lack of crystal structures at this stage made the efficient design of potent inhibitors challenging. To develop the understanding of the binding mode and the subsite SAR, a study by Deprez-Poulain et al. [100] utilised the PfA-M1 crystal structure in conjunction with docking studies to identify the key interactions formed in the S1 and S1′ pockets. Furthermore, Sivaraman et al. studied a small library of phosphonic acid arginine mimetics. The biaryl pyrazole-phenyl scaffold (11) was found to outperform the arginine moiety (10), caused by hydrophobic interactions between pyrazole and Met1034 [89]. Further SAR studies by the same group, showed improved interactions with fluoro-substituted biphenyl ring systems (12) [94,95]. X-ray crystal structures demonstrate that this system allows the 3,4,5-trifluorophenyl ring to sit deeper in the S1 pocket and enter an intricate network of water-mediated hydrogen-bonds (Figure 3B, shown in orange). More recently, a series of aminobenzosuberone inhibitors (e.g. 13) with selective PfA-M1 inhibition over other MAPs were reported [96–98].

The S1′ pocket consists of predominantly hydrophobic residues, however, it also allows for some hydrophilic interactions (Figure 3B) [74,95]. Early studies involving bestatin derivatives indicated that the S1′ pocket favours smaller aliphatic side chains [86,89]. SAR studies of different inhibitors probing this pocket and the comparison of their crystal structures indicate that the spatial differences in the S1′ cavity is the major contributor to the enzyme selectivity of PfA-M1 over other MAPs (PfA-M17) [94,95]. Computational studies have shown that measurements of pocket size from crystal structures are influenced by crystal packing [101]. These two facts highlight some of the challenges of compound design and the strong need to consider mechanism and how the target responds to inhibition. The development of these PfA-M1 inhibitors has relied heavily on target-based drug design. Combining in silico and structure-based drug development principles has allowed for the efficient design, synthesis and biological evaluation of potent cross-species dual-inhibitors. The design of future PfA-M1 inhibitors should exploit existing structural data, with focus on forming interactions deep within the S1′ region. This site has usually been regarded as a hydrophobic pocket, however, compounds that target hydrophilic residues such as Arg489 and Thr577 have largely been unexplored. To advance potent inhibitors with desirable drug-like properties, these structural changes should be made while taking PK and ADME properties into consideration.

Perspectives

• Importance of the field: Malaria remains a global health burden and the emergence of resistance reduces the effectiveness of current treatment options, prompting an urgency for the discovery of novel antimalarial agents.

• Summary of current thinking: Target-based drug design relies on understanding of target mechanism, and takes advantage of available knowledge of the structural and functional data to inform the design of candidates while also overcoming challenges associated with the targets. PfDHFR, PfDHODH, PfA-M1 and their inhibitors highlight how this process can be used to design potent inhibitors.

• Future directions: With the advances in antiparasitic target-identification and recent developments of techniques such as cryo-electron microscopy, structure-based drug design is positioned to be far more prominent in future antimalarial campaigns.

Abbreviations

ACT

artemisinin combination therapy

ADME

absorption, distribution, metabolism and excretion

CoQ

coenzyme Q

DHFR

dihydrofolate reductase

DHO

dihydroorotate, DHODH, dihydroorotate dehydrogenase

FMN

flavin mononucleotide

HTS

high-throughput screen

MAP

metalloaminopeptidase

MMV

medicines for malaria venture

MoA

mechanism of action

NADPH

nicotinamide adenine dinucleotide phosphate

nM

nanomolar

ORO

orotate

PDB

protein data bank

Pf

Plasmodium falciparum

PfA-M1

P. falciparum M1 aminopeptidase

PK

pharmacokinetics

Pv

Plasmodium vivax

Pyr

pyrimethamine

QM

quadruple mutant

SAR

structure-activity relationship

TS

thymidylate synthase

WHO

World Health Organisation

WT

wild-type

ZBG

zinc-binding group

Open access

Open access for this article was enabled by the participation of Monash University in an all-inclusive Read & Publish pilot with Portland Press and the Biochemical Society under a transformative agreement with CAUL.

Competing Interests

The authors declare that there are no competing interests associated with the manuscript.

Author Contributions

P.P.S.C. and M.M. contributed equally. All authors contributed to concept and writing of manuscript. All authors declare they have no conflict of interest.