Research ArticleMalaria

Antimalarial efficacy of MMV390048, an inhibitor of Plasmodium phosphatidylinositol 4-kinase

See allHide authors and affiliations

Science Translational Medicine  26 Apr 2017:
Vol. 9, Issue 387, eaad9735
DOI: 10.1126/scitranslmed.aad9735

A new antimalarial in the armamentarium

Paquet et al. screened a small-molecule library against the human malaria parasite, Plasmodium falciparum, and identified the 2-aminopyridine chemical class with potent activity. The optimized compound from this class, MMV390048, was active against multiple parasite life cycle stages, in both the mammalian host and the mosquito vector, and also killed drug-resistant parasites. MMV390048 killed the malaria parasite by blocking the parasite’s phosphatidylinositol 4-kinase (PI4K) and was able to protect monkeys from malaria infection. MMV390048 has potential as a new antimalarial drug that may contribute to global malaria eradication efforts.

Abstract

As part of the global effort toward malaria eradication, phenotypic whole-cell screening revealed the 2-aminopyridine class of small molecules as a good starting point to develop new antimalarial drugs. Stemming from this series, we found that the derivative, MMV390048, lacked cross-resistance with current drugs used to treat malaria. This compound was efficacious against all Plasmodium life cycle stages, apart from late hypnozoites in the liver. Efficacy was shown in the humanized Plasmodium falciparum mouse model, and modest reductions in mouse-to-mouse transmission were achieved in the Plasmodium berghei mouse model. Experiments in monkeys revealed the ability of MMV390048 to be used for full chemoprotection. Although MMV390048 was not able to eliminate liver hypnozoites, it delayed relapse in a Plasmodium cynomolgi monkey model. Both genomic and chemoproteomic studies identified a kinase of the Plasmodium parasite, phosphatidylinositol 4-kinase, as the molecular target of MMV390048. The ability of MMV390048 to block all life cycle stages of the malaria parasite suggests that this compound should be further developed and may contribute to malaria control and eradication as part of a single-dose combination treatment.

INTRODUCTION

Malaria, the disease caused by infection with protozoan parasites, primarily Plasmodium falciparum and Plasmodium vivax, still causes an estimated 236,000 to 635,000 deaths per annum in spite of the positive impact of interventions including medication, indoor spraying with insecticides to control the mosquito vector, and the use of insecticide-treated bed nets (1). Resistance to treatment regimens, especially the emergence of resistance to artemisinin drugs currently used in combination therapies (2, 3), still poses a threat and highlights the importance of developing treatments containing new chemical classes with different modes of action. The global aim to eradicate malaria brings about additional requirements of chemoprevention and transmission blocking, and the ability to eliminate dormant parasites in the liver that are responsible for relapse in the case of P. vivax infection. To this effect, new chemical entities should exhibit potency across the different life cycle stages of the parasite in both the human host and mosquito vector (4).

New chemical classes (5) with activity across different life cycle stages of the malaria parasite have emerged, and new mechanisms of action have been identified, including a class of inhibitors of the Plasmodium phosphatidylinositol 4-kinase (PI4K) (68). Here, we characterize the 2-aminopyridine MMV390048 (Fig. 1A) (9) as a representative of a new chemical class of Plasmodium PI4K inhibitors. MMV390048 was developed on the basis of a series of hits identified from a phenotypic high-throughput screen of a commercial BioFocus library (9). Although kinetic solubility of MMV390048 was not optimal at pH 6.5, it was well absorbed in rats. In vitro potency and good pharmacokinetics translated to efficacy in a Plasmodium berghei mouse model of malaria (9), which spurred further investigations into the potential of MMV390048 as an antimalarial agent.

Fig. 1. In vitro potency of the compound MMV390048.

(A) Structure of 2-aminopyridine MMV390048. (B) In vitro PRR depicting the number of viable parasites (P. falciparum 3D7 strain) over time after treatment with 10× IC50 of MMV390048 compared to other antimalarial drugs. (C) IC50 speed assay using unsynchronized culture of the P. falciparum NF54 strain (mean ± SD of n ≥ 3 independent assays), indicating activity at different incubation times. (D) Parasitemia as a function of time following once daily dosing for 4 days in the P. falciparum humanized SCID mouse model (n = 1 per dose level). Dosing was started on day 3 after infection. (All data points are included in tables S2 to S4.)

RESULTS

In vitro and in vivo activity of MMV390048 against blood-stage malaria parasites

The in vitro activity of MMV390048 against intraerythrocytic life cycle stages of P. falciparum (NF54 drug-sensitive strain) showed a steep inhibition curve with 50 and 90% inhibitory concentration (IC50 and IC90, respectively) values of 28 and 40 nM, respectively (fig. S1). Against a panel of multidrug-resistant clinical isolates of P. falciparum, the ratio of the maximum/minimum IC50 values for MMV390048 was 1.5-fold, suggesting that MMV390048 has a low risk for cross-resistance (table S1) (10).

The rate of parasite killing by MMV390048 was determined using limiting dilution experiments (11). The in vitro log parasite reduction ratio (PRR; log10 number of parasites that are killed in a single 48-hour life cycle) for MMV390048 was 2.7 at 10× IC50 with a lag phase of 24 hours (Fig. 1B and table S2). In the recently developed erythrocytic IC50 speed and stage specificity assays (12), the IC50 values of the slow-acting antimalarial drug pyrimethamine and MMV390048 were 7.8- and 4.8-fold higher, respectively, at the 24-hour time point compared to the 72-hour time point (Fig. 1C and table S3). These results indicate that MMV390048 has moderately slow killing action in vitro. The blood-stage specificity profile of MMV390048 was also similar to that of pyrimethamine, with highest activity against the young schizont stage (fig. S2).

In the P. berghei mouse efficacy model of malaria, the 90% effective dose (ED90) measured at 96 hours of MMV390048 after four oral administrations (4, 24, 48, and 72 hours after infection) was 1.1 mg/kg (ED50, 0.57 mg/kg). As previously described, a single oral dose of 30 mg/kg (24 hours after infection) was fully curative, defined as animals having no detectable parasites on day 30 after infection (9). In additional studies, the onset of action and P. berghei parasite recrudescence after a single oral dose of 100 mg/kg were determined, indicating an intermediate in vivo parasite clearance rate, similar to that of mefloquine (fig. S3) and slower than that of artesunate and chloroquine (all 100 mg/kg). MMV390048 was also curative at 100 mg/kg, without indication of recrudescence during the 30-day observation period.

The efficacy of MMV390048 was also assessed in humanized severe combined immunodeficient (SCID) mice infected with the drug-sensitive 3D7 P. falciparum strain (13). Efficacy was assessed after once daily oral administration of MMV390048 for four consecutive days, with blood parasitemia measured by flow cytometry (table S4). MMV390048 achieved an ED90 at day 7 of 0.57 mg/kg in this model. The rate of in vivo parasite clearance in this model (Fig. 1D) was comparable to that of the reference drug mefloquine (14).

From the SCID mouse efficacy study, the minimum blood concentration of MMV390048 necessary to inhibit growth was calculated. The direct effect pharmacokinetic-pharmacodynamic (PKPD) model, considering change in kill rate at higher doses from the SCID mouse study data (Fig. 1D, fig. S4, and tables S4 and S5), was used to determine the minimum parasiticidal concentration (MPC) and the minimum inhibitory concentration (MIC) of MMV390048 (tables S6 to S8). The model is referred to as the composite Emax model (15). For MMV390048, the highest kill rate was predicted to be at the dose of 1 mg/kg, which corresponded to a reduction of the parasite load in mice to <0.01% parasitemia at the end of 4 days (figs. S5 and S6). On the basis of the corresponding modeled blood concentrations, the predicted median effective concentration (EC50) for half-maximal kill rate was 96 ng/ml, the effective concentration to achieve 90% of the maximal kill rate (EC90) was 160 ng/ml, the MIC to inhibit growth was 71 ng/ml, and the in vivo PRR was 2.2 to 2.7 (table S9).

In vitro and in vivo potential of MMV390048 to reduce parasite transmission

A series of in vitro assays involving the sexual stages of the parasite life cycle were conducted to analyze the potential of MMV390048 to reduce host-to-vector transmission, which will consequently reduce host-to-host transmission (Fig. 2A). MMV390048 inhibited the viability of erythrocytic sexual late-stage gametocytes (stages IV and V) in culture with a potency of 285 nM, as demonstrated by a decrease in parasite lactate dehydrogenase (pLDH) activity (Fig. 2B and table S10). Gametocyte viability assays were also performed on early-stage (stages I to III) and late-stage gametocytes expressing luciferase, with MMV390048 achieving potencies of 214 and 140 nM, respectively (Fig. 2C and table S11). Kinetic studies were performed on these transgenic luciferase-expressing gametocytes. Late-stage gametocytes were killed 2.5 times faster than early-stage gametocytes, with a 24-hour rate of onset of action (correlating with the rate of onset of action for erythrocytic asexual stage parasites) (Fig. 2D and table S12). In addition, MMV390048 showed similar efficacy against late-stage gametocytes (>90%; stages IV and V) produced from five clinical isolates of P. falciparum compared to laboratory-adapted strains, providing evidence of its activity against gametocytes of current clinical parasite strains (table S13). When transmitted to the midgut of the mosquito vector after a blood meal, stage V gametocytes differentiate rapidly into male and female gametes. This step was reconstituted in vitro to assess the impact of MMV390048 on the functional viability of gametocytes through the formation of gametes. Stage V male gametocytes of P. falciparum typically differentiate into eight flagellated motile male gametes by exflagellation, an event that can be observed and quantified under the microscope. After incubating P. falciparum stage V gametocytes with MMV390048 for 24 hours and inducing gametogenesis, exflagellation was inhibited with a potency of 90 nM (Fig. 2E and table S14).

Fig. 2. In vitro and in vivo transmission-blocking potential of MMV390048.

(A) P. falciparum sexual stages and the various readouts used in drug screening assays, including assays against early- and late-stage gametocytes, the male gamete formation assay, and the SMFA. (B) Gametocyte viability assay was performed by measuring LDH activity of nonrecombinant stage IV and V gametocytes of P. falciparum. MMV390048 is in black (IC50, 285 nM) and dihydroartemisinin is in blue (IC50, 12.0 nM) (n = 2). (C) Gametocyte viability assay was performed by measuring luminescence of recombinant parasites expressing luciferase. Stage I to III gametocytes are in black (IC50, 214 nM) and stage IV and V gametocytes are in red (IC50, 140 nM) (n = 3). (D) Clearance rates for stage I to III (black) and stage IV and V (red) gametocytes treated with MMV390048 at its IC50 concentration over a 3-day period indicated by the slopes of −6.09 and −15.49, respectively (n = 3). (E) Inhibition of exflagellation and gamete formation in the male gamete formation assay (n = 4). (F) Dose-response curve for MMV390048 in the indirect SMFA. The assay measures the number of oocysts that developed in the midgut of mosquitoes after feeding on P. falciparum stage V gametocyte–infected blood that was exposed to different concentrations of MMV390048 for 24 hours before feeding. In the direct SMFA (inset), mosquitoes fed on infected blood directly after it was treated with different concentrations of MMV390048 (n = 2). (G) Number of oocysts in the midgut of mosquitoes after feeding on P. berghei–infected mice that were treated with either MMV390048, vehicle, or control drugs during the mouse-to-mouse transmission study. (All data points are included in tables S10 to S12, S15, and S16).

The transmission-blocking assay most often used is the standard membrane-feeding assay (SMFA), performed by incubating the test molecule with stage V gametocytes of P. falciparum either for 24 hours before mosquito feeding (indirect SMFA) or directly at the time of the blood meal (direct SMFA). When tested in the indirect SMFA, MMV390048 inhibited the formation of oocysts in the mosquito midgut at 111 nM (Fig. 2F and table S15). When added directly to the blood meal, MMV390048 inhibited the formation of oocysts by less than 25% at 1 μM (Fig. 2F, inset, and table S16). The difference in potency observed for the indirect feeding assay supports a transmission-reducing role of MMV390048 that targets stage V gametocytes in the host blood more efficiently than subsequent forms developing in the mosquito midgut.

To test the host-to-vector-to-host transmission-blocking efficacy of MMV390048 in vivo, a model comprising mouse-to-mosquito-to-mouse transmission of P. berghei infection was used (16, 17). Briefly, the effect of a transmission-blocking drug can be expressed in terms of impact on the mosquito population and subsequent vertebrate host populations following mosquito bites at variable transmission intensities. Within the course of this study, MMV390048 (administered orally at 2 mg/kg) inhibited parasite transmission to the mosquito vector, with a 69.3 and a 30.3% reduction in oocyst intensity (mean number of parasites per midgut) and prevalence (% infected mosquitoes), respectively, observed over two replicate experiments (Table 1 and Fig. 2G). This resulted in a 37.2 and 46.5% reduction in sporozoite intensity and prevalence (Table 1). Mosquitoes previously fed on infected, drug-treated, or untreated mice were allowed to feed on uninfected animals (17). Individual uninfected mice were exposed to 2, 5, or 10 mosquito bites (biting rates simulated different transmission intensities). Over the range of mosquito biting rates, a 10.1% reduction in the number of uninfected mice that developed blood-stage infection was observed for mice bitten by mosquitoes that fed on treated versus untreated infected mice. The overall effectiveness of a drug intervention over one round of transmission (from mouse to mosquito to mouse) can be quantified by estimating the drug’s ability to reduce the basic reproductive number (R0) (16). This has been termed the “effect size” of an intervention. By fitting data from the mouse-to-mouse assay to a chain binomial model, we estimated the effect size of the intervention, assessing the ability of MMV390048 to reduce the basic reproductive number R0 (assuming 100% coverage). We estimated an effect size of 28.5% [95% confidence interval (CI), 22.8 to 33.7%], suggesting that MMV390048 is capable of acting as a transmission-blocking agent in lower transmission settings (16). Positive and negative controls (atovaquone at 0.3 mg/kg and sulfadiazine at 8.4 mg/kg, respectively) performed as expected and as in previous studies (17).

Table 1. Transmission-blocking effects of MMV390048 (2 mg/kg) measured in the mouse-to-mouse transmission model.
View this table:

In vitro and in vivo prophylactic activity of MMV390048 against Plasmodium liver stages

The prophylactic activity of MMV390048 against the liver stages of Plasmodium that precede symptomatic blood-stage infection was determined in vitro using Plasmodium cynomolgi, a simian parasite species closely related to P. vivax (18). This parasite produces in vitro and in vivo both large developing and small nondividing exoerythrocytic (liver) forms, recently validated as being schizonts and hypnozoites, respectively (19). Hypnozoites are capable of reinitiating a growth phase, causing subsequent disease relapses that are a key clinical feature of P. vivax infection. In a cell-based assay, MMV390048, administered to a primary rhesus hepatocyte cell culture 2 hours after inoculum (allowing sporozoites to invade the hepatocytes), showed potent inhibition of liver-stage development of both schizonts and hypnozoites (fig. S7). The IC50 values were 64 nM for schizonts and 61 nM for hypnozoites (fig. S7).

The prophylactic effect of MMV390048 was also evaluated in vivo in P. cynomolgi–infected macaques (Macaca mulatta) (20). Two cohorts of malaria-naïve monkeys were used during this experiment. In cohort 1, MMV390048 (20 mg/kg) was administered orally to three study animals on day 1 before sporozoite inoculation; in cohort 2, two control monkeys were administered orally with the same volume of vehicle. On day 0, the monkeys were infected with 1 × 106 P. cynomolgi sporozoites intravenously, and blood smears were taken daily up to day 100 to detect malaria parasites that progressed through the liver stages into blood-stage forms. Detectable parasite infection occurred on day 8 after inoculation in the two control monkeys (Fig. 3A). Parasitemia continued to develop until day 11 after inoculation, at which time both control monkeys received a treatment consisting of a 7-day course of chloroquine (10 mg base/kg, orally). Parasite clearance occurred in all animals treated with chloroquine. Because chloroquine cannot eliminate hypnozoites, relapse was expected for the control group due to reactivation of the dormant forms (21). Relapse occurred at day 26 for monkey 1 and at day 29 for monkey 2 of the control group (Fig. 3A and table S17). These monkeys were then radically cured using a standard 7-day regimen of oral chloroquine and primaquine (10 and 1.78 mg base/kg, respectively). Once parasitemia was below the limit of detection for at least five consecutive days, monkeys were considered as radically cured and taken out of the study (colored arrows in Fig. 3A). In contrast to the control group, animals from cohort 1 did not present any parasitemia when observed for up to 100 days after inoculation, revealing prophylactic efficacy and full protection by MMV390048 (Fig. 3A, monkeys 3, 4, and 5).

Fig. 3. MMV390048 efficacy in P. cynomolgi–infected rhesus macaques and the phenotype of treated P. berghei liver-stage parasites.

(A) Daily parasitemias for individual P. cynomolgi–infected monkeys that were treated with MMV390048 on day −1 before infection (prophylactic study). The black arrow indicates when chloroquine treatment (10 mg base/kg orally for 7 days) was started to clear primary infection. Control group monkeys relapsed on days 26 and 29, whereas monkeys treated with MMV390048 remained clear of parasitemia for the 100 days of observation. (B) Mean plasma concentrations of MMV390048 over time for the treated monkeys in the prophylactic study after a single oral treatment (20 mg/kg) with MMV390048. The concentration of MMV390048 remained above its IC99 for 6.5 days (orange arrow). The in vivo IC99 was estimated to be 353 ± 32 nM (138 ± 13 ng/ml) based on the in vitro IC99 determined in the P. cynomolgi liver-stage assay and correcting for serum albumin (AlbuMAX) binding (54%) and plasma protein binding (86%). (C) Daily parasitemias for individual P. cynomolgi–infected monkeys in the radical cure study. Once infection was established, animals were treated with either MMV390048 or vehicle (20 mg/kg once daily for 5 days; brown arrows) with simultaneous treatment with chloroquine (10 mg base/kg orally for 7 days) to clear primary parasitemia (black arrow indicating start of chloroquine treatment). Control group monkeys relapsed on days 24 and 29, whereas MMV390048-treated monkeys relapsed on days 33 to 35. (D) Mean plasma concentrations of MMV390048 over time for the treatment group monkeys in the radical cure study. Brown arrows indicate treatment with MMV390048 (20 mg/kg oral dose daily for 5 days). The plasma concentration of MMV390048 remained above the P. cynomolgi liver-stage IC99 for 15.8 days (orange arrow) and above the blood-stage MIC (11.5 ng/ml free concentration) for 20 days (black line). The MIC was calculated from the in vivo P. falciparum SCID mouse model and corrected for protein binding (86.1%) and blood-to-plasma partitioning (0.86). The blue arrow represents the time at which monkeys 3, 4, and 5 relapsed. (E) Image of an early liver-stage P. berghei parasite treated with 460 nM MMV390048 from 2 to 8 hours after inoculation. (F) The parasite displayed mislocalization of the parasitophorous vacuole membrane protein UIS4 from the surface to the parasite interior (red; visualized with goat polyclonal antibody and Alexa Fluor–conjugated secondary antibody) when treated with MMV390048 compared to vehicle. For visualization, cytoplasmic P. berghei heat shock protein 70 is displayed in green (visualized with 2E6 monoclonal antibody and Alexa Fluor–conjugated secondary antibody). Images are single confocal sections. LLOQ, lower limit of quantification. (All data points are included in tables S17 to S20.)

Blood samples were taken at different time points to study MMV390048 plasma concentration–time curves and to relate the pharmacokinetic profile to the efficacy of the compound. A peak concentration (Cmax) was obtained shortly after oral administration, reaching 1.98 μg/ml within 24 hours. The concentration then decreased below the limit of quantification at around 320 hours after administration (Fig. 3B and table S18). The terminal half-life of the compound was ~30 hours, allowing coverage above the liver-stage IC99 (determined to be 90 nM in the in vitro P. cynomolgi liver-stage assay) for about 6.5 days (arrow in Fig. 3B).

In vitro and in vivo activity of MMV390048 against P. cynomolgi liver-stage hypnozoites

The potential for MMV390048 to eliminate liver-stage hypnozoites in the case of P. vivax infection (referred to as a radical cure) was also investigated using the same P. cynomolgi in vitro and in vivo models. In the cell-based assay, MMV390048 was incubated with P. cynomolgi–infected rhesus monkey hepatocytes at different time points of liver-stage parasite development, at days 1, 3, 4, and 5 after inoculation. The ability of MMV390048 to kill the parasites was reduced when the compound was administered to later-stage (>24-hour-old) hypnozoites, with a complete loss of activity when given at day 5 after infection (IC50 values shift from 61 nM at day 0 to >10 μM at day 5) (Table 2). The compound remained active when applied to later-stage (>24-hour-old) schizonts despite reduced potency (IC50 values shift from 64 nM at day 0 to 1.5 μM at day 5), suggesting that this compound was still active against late-stage schizonts (Table 2).

Table 2. MMV390048 radical cure (≥24-hour addition) versus prophylactic (3-hour addition) effects in the P. cynomolgi liver-stage assay.

EEF, exoerythrocytic forms.

View this table:

The observed inability of MMV390048 to eliminate later-stage hypnozoites was also confirmed in vivo in P. cynomolgi–infected macaques (20). Similar to the prophylactic experiment, two cohorts of malaria-naïve monkeys were used. On day 0, the monkeys were injected intravenously with 1 × 106 sporozoites; blood smears were taken and analyzed for parasitemia daily up to day 100. The two control monkeys in cohort 1 were administered orally with vehicle (hydroxypropyl methylcellulose trimellitate). In cohort 2, three monkeys were administered orally with MMV390048 (20 mg/kg oral daily dose for 5 days) at the same time as a 7-day course of chloroquine (10 mg base/kg) that was given to treat the primary parasitemia. Parasitemia was then monitored on a daily basis until relapse occurred, which happened at day 24 for control monkey 1 and at day 29 for control monkey 2 (Fig. 3C). On average, relapses occurred 16.7 days after primary parasitemia versus 9.5 days in the control group (P = 0.0397, unpaired t test). Blood samples were also taken during this study to relate efficacy to MMV390048 plasma concentration. A 5-day administration of MMV390048 led to a high Cmax, reaching 5697 ng/ml without any observed toxicity. The plasma concentration of MMV390048 was maintained above the IC99, as determined in vitro against liver-stage parasites, for 15.8 days (Fig. 3D). This pharmacokinetic profile demonstrated that the delay in relapse relative to the control group was likely due to suppression of blood-stage parasites. Blood-stage parasitemia due to reactivation of liver hypnozoites appeared only after MMV390048 dropped below its MIC (determined to be 11.5 ng/ml based on the in vivo P. falciparum SCID mouse model and correcting for protein binding and partitioning of MMV390048 between blood and plasma).

Phenotypic correlates of liver-stage prophylaxis with MMV390048

An in vitro model consisting of P. berghei–infected HepG2 liver cells was used to search for phenotypic correlates of liver-stage prophylactic activity. Treatment with 460 nM MMV390048 from 2 to 8 hours after infection of HepG2 cells resulted in mislocalization of a parasite protein (the parasitophorous vacuole membrane protein UIS4) from the surface to the interior of the parasite (Fig. 3, E and F). This cellular phenotype has been shown previously to be associated with efficient liver-stage parasite clearance in vitro and causal prophylaxis in vivo by the compound Torin2 but has not been observed with current clinically used drugs with activity against liver-stage parasites (22).

In vitro evaluation of P. falciparum mutant strains that are resistant to MMV390048

To identify the mechanism of action of MMV390048, drug-resistant mutants of P. falciparum were generated and genetic mutations were identified by whole-genome sequencing (23). Initial drug pressure at 46 nM (3× IC50; IC50 determined to be 16 nM in this instance) over a range of inocula (106 to 109) indicated that mutants were obtained with a frequency of about 1 per 3 × 106 infected red blood cells in vitro (table S21). To generate parasites with greater drug resistance, three parasite cultures (109 Dd2 parasites per flask) were subjected to a higher selective pressure of 76 nM MMV390048 (5× IC50). The cultures recrudesced between days 28 and 35 and were cloned by limiting dilution. Both the bulk culture and clones showed shifts in IC50 and IC90 values, with clones giving about four- to fivefold IC50 shift (70 to 90 nM) relative to the parental line (fig. S8). Three clones (E6, A2, and B9) were selected for genome sequencing.

Whole-genome sequencing of MMV390048-resistant P. falciparum parasites identified a PI4K (P. falciparum PI4K, Pf3D7_0509800) as a potential resistance determinant. Two clones from the same culture (flask 2 clones A2 and E6) harbored an A1319V mutation, whereas a third clone (flask 3 clone B9) yielded an S743T mutation in PfPI4K (table S22). The read count at both positions exceeded 120-fold coverage with no contaminating wild-type reads. Pfpi4k was the only gene mutated in all three clones selected under drug pressure, providing evidence in favor of PfPI4K being the target of MMV390048. The program BIC-Seq (24) was used to search for copy number variations (CNVs) in MMV390048-resistant mutants A2, E6, and B9, by varying input parameters empirically. No CNVs were detectable in any iteration; in particular, none was found in the Pfpi4k gene (fig. S9).

Cross-resistance of MMV390048 was tested against three transgenic parasite lines having mutations in the PfPI4K kinase domain that confer resistance to the chemically distinct imidazopyrazine (PI4K-S1320L and PI4K-H1484Y) and quinoxaline (PI4K-Y1356F) scaffolds (6). In comparison to the Dd2 multidrug-resistant parental line (IC50, 22.7 nM), parasites bearing PfPI4K mutations at H1484Y (IC50, 62.8 nM) and S1320L (IC50, 57.7 nM), which is immediately adjacent to the A1319V mutation identified in this study, were cross-resistant to MMV390048, further supporting PfPI4K as the target of this compound (fig. S10).

Chemoproteomics identifies PfPI4K as the target of MMV390048

A chemoproteomics approach was also taken to determine the mechanism of action of MMV390048. Covalent immobilization on Sepharose beads of MMV666845, an active analog containing a primary amine functionality (NF54 IC50, 0.019 μM; table S23), was used to affinity capture potential target proteins from a P. falciparum blood-stage extract. Pull-down experiments were performed in the absence or presence of excess “free” MMV390048 to delineate target proteins for which capture was competitively inhibited. In a second experiment, an analog devoid of antimalarial activity, MMV034137 (NF54 IC50, 9.2 μM; table S23), was used. All proteins captured by the beads were quantified by isotope tagging of tryptic peptides followed by liquid chromatography tandem-mass spectrometry (LC-MS/MS) analysis (25). MMV390048 competitively inhibited the binding of only a single protein, PfPI4K, to the beads (Fig. 4A).

Fig. 4. Chemoproteomic identification of Plasmodium PI4K as the target of MMV390048.

(A) Binding of MMV390048 to P. falciparum PI4K in parasite extracts. An analog of MMV390048, MMV666845, was covalently immobilized on Sepharose beads and used for affinity capture of potential target proteins from a P. falciparum blood-stage extract. The addition of excess MMV390048 (10 μM) to the P. falciparum extract competitively inhibited the binding of P. falciparum PI4K to the beads. No other protein was reproducibly inhibited. The addition of MMV034137, a closely related structural analog devoid of antimalarial activity, had no effect on the capture of P. falciparum PI4K. (B) MMV390048 binds to the ATP binding site of Plasmodium PI4K but not to human PI4Kα or PI4Kβ. A set of seven promiscuous ATP-competitive kinase inhibitors was covalently immobilized to beads (Kinobeads) and used for affinity capture of potential kinase targets from a P. falciparum blood-stage extract and from a human erythroleukemia K562 cell line. The plot illustrates pIC50 values obtained in two independent replicates of the experiment (exp. 1 and 2). The addition of MMV390048 to the extract competitively inhibited the binding of P. falciparum PI4K to the beads. PIP4K2C was the only human kinase for which binding was also competitively inhibited by MMV390048 with an IC50 value in the same range as the P. falciparum PI4K. (C) The addition of MMV390048 to the P. falciparum extract over a range of concentrations yielded an apparent dissociation constant (Kdapp) of 0.3 μM using MMV666845 for affinity capture and a calculated Kdapp of 0.1 μM using Kinobeads for affinity capture. (D) Structure-activity relationship of MMV390048 analogs revealed excellent correlation of antimalarial activity and binding to P. falciparum PI4K in parasite extracts. Fourteen compounds from the aminopyridine (MMV390048) series with increasing activity against P. falciparum were subjected to profiling using immobilized MMV666845 beads [see also (A)]. With increased potency against P. falciparum, the compounds showed increased competition for binding to P. falciparum PI4K and thus reduced its binding to the bead matrix. r, Pearson’s correlation coefficient; P, P value (calculated probability). (All data points are included in tables S26 and S27.)

A second experiment was performed in which we added free MMV390048 over a range of concentrations to establish a competition-binding curve and to determine an IC50 value (Fig. 4, B and C). The IC50 values obtained in these experiments represented a measure of target affinity but were also affected by the affinity of the target for the bead-immobilized ligand. The latter effect was deduced by determining the depletion of the target proteins by the beads such that apparent dissociation constants (Kdapp) could be calculated, which were largely independent from the bead ligand (26). The apparent Kd value was determined to be 0.30 μM for PfPI4K (Fig. 4C).

Immobilization of a drug via a linker may not be compatible with binding to all of its targets. Therefore, a similar capturing experiment with “Kinobeads” was performed, which represents a combination of immobilized promiscuous adenosine triphosphate (ATP)–competitive kinase inhibitors (27, 28). As in the previous experiments, PfPI4K was the only P. falciparum protein that exhibited a dose-dependent reduction of bead binding upon the addition of MMV390048 to the extract. These studies yielded an apparent dissociation constant (Kdapp) of 0.1 μM, which was concordant with the previous Kdapp value derived with the bead-immobilized analog (Fig. 4C). The apparent Kd values obtained from the chemoproteomic experiments were about 10-fold lower compared to the potency of MMV390048 in parasite growth assays. We also assessed a set of MMV390048 analogs active against P. falciparum, with an IC50 ranging from low nanomolar to mid micromolar. The chemoproteomic competitive binding data showed a good correlation between the degree of binding to PfPI4K and antiparasitic activity (Fig. 4D and table S23).

For the identification of possible human host cell targets and off-targets, we adopted the same chemoproteomics strategy and used the same bead-immobilized ligands for pull-down experiments in human K562 erythroleukemia cells. Given the fact that MMV390048 is an ATP-competitive kinase inhibitor, the compound exhibited a markedly clean selectivity profile. We did not observe any binding to the human PfPI4K orthologs PI4KB (Q9UBF8) and PI4KA (P42356) (fig. S9). Only three human kinases showed any binding affinity for MMV390048. Human PIP4K2C (Q8TBX8) was the major human target and had a similar affinity to that for PfPI4K, whereas the other two targets, ATM (Q13315) and TNIK (Q9UKE5), showed low binding affinity (Fig. 4B and fig. S11).

Functional evaluation of the activity of MMV390048 against recombinant P. vivax PI4K

Because PI4Ks are highly homologous across P. falciparum and P. vivax species (29), MMV390048 was evaluated for inhibition of recombinant P. vivax PI4K (6). MMV390048 inhibited the activity of P. vivax PI4K with an IC50 of 0.0034 μM. Similarly, its analog, MMV034137, which was devoid of in vitro whole-cell activity (IC50, 2.8 μM), had decreased activity against P. vivax PI4K (IC50, 0.24 μM) compared to MMV390048. MMV666845, the analog that was used for affinity capturing, was equipotent (IC50, 0.0015 μM) with MMV390048. Across a set of analogs evaluated in the P. vivax PI4K functional assay, a close correlation between P. vivax PI4K inhibition and activity against the drug-sensitive P. falciparum NF54 parasite strain was observed (table S23 and fig. S12). This result further supports the genomic and chemoproteomic findings and confirms the mode of action of MMV390048 to be through inhibition of Plasmodium PI4K.

In vivo pharmacokinetics of MMV390048 in mice, rats, dogs, and monkeys

Pharmacokinetic studies for MMV390048 were conducted in mice, rats, dogs, and monkeys to aid in human pharmacokinetic and dose predictions (fig. S13). The pharmacokinetic profiles after intravenous administration were characterized by low plasma clearance (0.036 to 0.39 liter/hour per kg) and a moderate to high volume of distribution (1.3 to 3.0 liters/kg) in all species (table S24). The half-life was moderate to long in all species, ranging from 2.5 hours in mice to 7.3 hours in rats, 52 hours in dogs, and 61 hours in monkeys. After oral administration, half-lives were similar to those after intravenous dosing in all species except mice, where the oral half-life was longer (table S25). The times at which maximum concentrations were reached increased with increasing dose, characteristic of dissolution rate–limited absorption. The oral bioavailability of MMV390048 from a simple aqueous suspension formulation was >46% in all species. In both rats and monkeys, increases in the area under the curve (AUC) occurred roughly proportional to an increase in dose, with greater than proportional increases in AUC noted in mice (table S25).

Pharmacokinetic data from mouse, rat, dog, and monkey were used for scaling to human using allometric principles (fig. S14 and table S24). MMV390048 is predicted to have low plasma clearance in humans (0.02 liter/hour per kg) and a moderate volume of distribution (2.8 liters/kg). Together, these parameters led to a predicted elimination half-life of about 90 hours. A human dose for the treatment of P. falciparum was sought that could maintain blood concentrations above therapeutic levels for 8 days (four asexual parasite cycles) to bring parasitemia down by more than 6 log units. Assuming an oral bioavailability similar to that achieved using a simple suspension in animals (70%) and a typical body weight of 70 kg, a dose of around 80 to 100 mg of MMV390048 was estimated to achieve the target drug concentration in humans.

MMV390048 safety for individuals with G6PD deficiency

To assess and compare the hemolytic effects of MMV390048, 4-day treatment regimens at dose levels of 1, 10, and 30 mg/kg per day were assessed in SCID mice engrafted with G6PD-deficient human red blood cells. Seven days after treatment with MMV390048, neither a substantial decrease in human red blood cells nor a significant increase in spleen weight was observed [one-way analysis of variance (ANOVA) with Bonferroni’s multiple comparison tests], leading us to conclude that MMV390048 was not hemolytic in this model (fig. S15). MMV390048 should therefore be safe as a treatment for G6PD-deficient patients infected with Plasmodium, in contrast to primaquine, which showed a hemolytic effect in this model.

DISCUSSION

MMV390048 is a new chemical entity derived from high-throughput screening and phenotypic-based optimization. Its lack of cross-resistance with marketed and clinically used antimalarial compounds suggests a potentially new mechanism of action, characterized by activity against young schizonts in the erythrocytic stage of the parasite life cycle. Potent in vivo efficacy was shown in both P. berghei and P. falciparum mouse models, with ED90 values of 1.1 and 0.57 mg/kg, respectively. Clearance of parasites and cure was also achieved previously in the P. berghei mouse model with a single dose as low as 30 mg/kg, driven by the strong potency and excellent pharmacokinetic behavior of the compound (9). Across mouse, rat, dog, and monkey species, MMV390048 maintained a low clearance and a long half-life. Good oral bioavailability was also evident across all species, supporting the correlation between the in vitro potency and the in vivo efficacy. The consistent pharmacokinetic profiles led to a human half-life prediction of around 90 hours and a low dose of 80 to 100 mg to maintain a therapeutic concentration for 8 days (9). MMV390048, therefore, has the potential to be used as a component of a single-dose combination therapy.

In addition to its activity on asexual blood-stage parasites, MMV390048 also showed potential for interrupting transmission. Its potency against gametocytes was submicromolar and well aligned with the efficacy observed against male gametes and oocysts in mosquitoes. When assessed in the host-to-host transmission model with P. berghei, a moderate reduction in the number of mice that developed blood-stage infection was observed. Clinical studies will be necessary to assess the compound’s effects on gametocyte viability and to fully evaluate its potential for transmission blocking in patients. Nevertheless, the data demonstrating inhibition of transmission from mouse to mosquito to mouse suggest a potential role for such a molecule in malaria eradication efforts.

MMV390048 also holds promise as a prophylactic and chemoprotective agent, as demonstrated by its impact on parasite liver-stage development. Potent in vitro prophylactic activity against the P. vivax–related simian parasite species, P. cynomolgi, was evident in preventing both hypnozoite and schizont development in the liver. This prophylactic effect translated in vivo to monkeys infected with P. cynomolgi that were treated with MMV390048. In contrast, liver-stage activity was not evident against late-stage hypnozoites (>24 hours old and referred to as late stage) in vitro and in vivo. Therefore, MMV390048 did not deliver a radical cure. This was also recently observed for the PI4K inhibitor KDU691 (30). Although MMV390048 showed activity against late-stage schizonts, there was a drop in activity when the compound was added at day 5, which could imply that eventually a higher dose would be needed to kill developed liver schizonts in vivo than was needed to clear a blood-stage infection. This requires more understanding and will be the subject of future studies. The phenotype in early developing P. berghei liver stages was shown to be mislocalization of the parasitophorous vacuole membrane protein UIS4 to the parasite interior, similar to a previous observation with the compound Torin2 (22). Further investigations into the RNA and protein expression of the PI4K target will be necessary to evaluate the phenotypic correlates of activity against late-stage schizonts and hypnozoites.

Chemoproteomic pull-down experiments based on different types of immobilized ligands provided compelling evidence that Plasmodium PI4K is the target of MMV390048, because MMV390048 competitively inhibited the binding of only a single protein, PfPI4K, to the beads. The lower Kd values obtained from the chemoproteomics assay compared to the potency of MMV390048 in parasite growth assays can be ascribed to a number of possible reasons. Many kinase inhibitors exhibit higher affinity for their target kinases in activated conformations (31, 32), whereas the bulk of the P. falciparum PI4K protein present in the parasite extracts might have adopted an inactive, lower-affinity conformation. Alternatively, the compound might accumulate at the site of action in the parasite. Single-nucleotide mutations in the gene encoding P. falciparum PI4K of MMV390048-resistant parasite strains further confirmed PI4K as the target. Although the frequency of resistance to MMV390048 appeared to be high, combination treatment partners could be identified to prevent or delay the emergence of resistance in field isolates. The correlation between phenotypic effect and inhibition of P. falciparum PI4K was evident for the 2-aminopyridine series of compounds during competitive binding experiments and also with inhibition studies using recombinant P. vivax PI4K.

Inhibitors of Plasmodium PI4K will be important as treatments and as chemoprotective and transmission-blocking agents but will be unlikely to radically cure P. vivax infection based on the findings with P. cynomolgi. As a chemoprotective agent, based on the in vivo P. cynomolgi data and the SCID mouse hemolytic data, MMV390048 has the potential to be a prophylactic for all human malaria infections without any known risk of hemolysis in G6PD-deficient patients. As a kinase inhibitor, MMV390048 is selective for PI4K and did not bind to other P. falciparum and human kinases apart from human PIP4K2C, thus alleviating potential kinase-mediated safety concerns. The consequences of PIP4K2C inhibition in the host are unknown, but it is interesting to note that this kinase represents an off-target of the marketed chronic myeloid leukemia drug imatinib (Gleevec) (27). MMV390048 showed an acceptable preclinical safety profile and is currently undergoing clinical assessment. This compound holds the potential to clinically validate Plasmodium PI4K as a target in the treatment of malaria, as well as to establish the role of a PI4K inhibitor in combination therapy.

Notwithstanding the promise shown by MMV390048 based on the data presented, there are limitations to our studies. Parasite viability and inhibition studies were based on in vitro analysis. In the case of liver-stage investigations, simian and rodent parasite species were used and not human malaria parasite species. Likewise, in vivo models for transmission, prophylaxis, and radical cure were conducted in animals using the appropriate parasite species, namely, P. berghei for mouse-to-mouse transmission (related to P. falciparum) and P. cynomolgi for liver-stage infections in monkeys (related to P. vivax). PKPD estimations were also based on preclinical species data. Finally, although MMV390048 is a highly selective P. falciparum PI4K inhibitor, the full implications of blocking the PIP4K2C human kinase, including when used in combination with other antimalarial drugs, have not yet been elucidated and need further investigation.

MATERIALS AND METHODS

Study design

The objective of this study was to fully characterize the potential of MMV390048 for treatment, prophylaxis, and radical cure of malaria in mice and monkey models and to elucidate the molecular target of the compound. Efficacy was assessed across blood, liver, and transmission stages of the Plasmodium life cycle using both in vitro and in vivo models of infection. The PKPD relationship was modeled on the basis of compound exposure during efficacy studies and pharmacokinetic profiles in preclinical species. Target identification was conducted through genetic and chemoproteomic means. In vitro generation of MMV390048-resistant parasites indicated a possible drug resistance mechanism linked to a target protein. The protein target was identified through chemoproteomic pull-down “Kinobead” experiments and competitive binding assays and was further confirmed by functional enzyme assays.

In vivo transmission inhibition assessed by a mouse-to-mouse population model

MMV390048 was tested in duplicate at a concentration of 2 mg/kg. The diluting agent for all test drugs, 1% methyl cellulose (Sigma-Aldrich no. M7140), was used as a negative control (referred to as “no drug”). Sulfadiazine (Sigma-Aldrich no. S6387) and atovaquone (Sigma-Aldrich no. A7968) were used as additional negative (that is, not transmission-blocking) and positive drug controls, respectively (17). All treatments consisted of a single dose of drug. Treatments were delivered by oral gavage except sulfadiazine and atovaquone, which were administered by intraperitoneal injection.

For each treatment regime, mouse-to-mouse transmission was performed in duplicate with general parasite maintenance carried out as previously described (16, 17), with 10 mice and 1000 mosquitoes used per individual treatment. Briefly, for each drug treatment group, five female TO mice (6 to 8 weeks old) were infected with P. berghei clone 507cl1 by syringe inoculation (intraperitoneal). At day 9 after infection, tail blood drops were examined for the presence of exflagellation, and mice were given the drug or control treatment by the indicated route. Blood-stage infections were monitored on Giemsa-stained tail blood smears before and 24 hours after treatment, as previously described. Ten days after infection and 24 hours after treatment, mice were anesthetized and exposed to 500 starved female Anopheles stephensi (line SD 500) mosquitoes. Mosquitoes that did not take a blood meal were discarded, and the remaining mosquitoes were maintained on 8% (w/v) fructose and 0.05% (w/v) p-aminobenzoic acid at 19°C and 80% relative humidity. On day 10 after feeding, midguts were dissected from a random sample of 50 mosquitoes per cage, and oocyst prevalence (% infected mosquitoes) and intensity (mean number of parasites per midgut) were recorded. For each treatment group, oocyst prevalence and mean oocyst intensity were compared to the no drug control group to calculate “classical” inhibition of transmission (based on oocyst production inhibition). All mosquito dissections to assess oocyst intensity and prevalence were performed under randomized and double-blind conditions.

The remaining mosquitoes were maintained until 21 days after infection, when salivary gland sporozoites were at their peak of infectiousness (16). Individual, anesthetized naïve mice were exposed for 20 min to predetermined numbers of potentially infectious mosquitoes, randomly selected from the appropriate mosquito population. For each treatment group, five individual naïve mice were exposed to either 2, 5, or 10 mosquito bites. This range of mosquito biting rates (simulating different transmission intensities) is necessary to calculate effect size (see below). Successful feeding was confirmed by the presence of blood in the mosquito abdomen. Where necessary, additional mosquitoes were given the opportunity to feed until the required number of successful bites was achieved. After feeding, all mosquitoes were individually dissected to determine the prevalence of salivary gland sporozoites. Glands were directly dissected onto a glass slide and covered with a coverslip, and intensities were scored on a log scale (score of 0, no sporozoites visible; 1, 1 to 10 sporozoites visible; 2, 10 to 100 sporozoites visible; 3, 100 to 1000 sporozoites visible; 4, 1000+ sporozoites visible). For each treatment group, sporozoite prevalence and mean sporozoite intensity were compared to the no drug control group to calculate inhibition. The “bitten” mice were allowed to recover and maintained for 10 days after feeding. Daily tail blood smears were performed from days 4 to 10 to establish patency, parasitemia, and gametocytemia. Generalized linear mixed models were used to estimate the overall effectiveness of the different interventions combining data from all repeat replicates (13). Efficacy of the treatment (as compared to the no drug control) was included as a fixed effect, whereas the mouse-to-mosquito transmission parameters were included as random effects. We assumed a binomial error structure for the oocyst and sporozoite prevalence data, whereas we used a zero-inflated negative binomial distribution for oocyst, asexual parasite, and gametocyte intensity. The impact on sporozoite intensity was assessed by looking at differences in sporozoite score, which was assumed to follow a Poisson distribution. Ninety-five percent CIs were estimated by bootstrapping, and the model was selected using a likelihood ratio test. The overall effectiveness of an intervention over one round of transmission (from mouse to mosquito to mouse) can be quantified by estimating its ability to reduce the basic reproduction number. This has been termed the effect size. If it is assumed that all infectious mosquitoes are equally infectious, this can be estimated by fitting a chain binomial model. A full description of the methodology has been reported previously (17). The models were fitted to the data using maximum likelihood methods, and the 95% CI estimates were obtained from the likelihood profile. Programs used for data analysis were R and GraphPad Prism. Animals were selected randomly for each group but were not blinded.

In vivo prophylactic and radical cure efficacy studies in the P. cynomolgi monkey model

The prophylactic and radical cure activity of the test compound was investigated using a P. cynomolgi–infected rhesus monkey model, as described (33). The model consists of injecting 1 × 106 sporozoites (B strain) into monkeys and assaying for blood-stage parasitemia by serial microscopy blood smear evaluation. Test compounds were added 1 day before infection of rhesus monkeys to test for their prophylactic activity or upon primary infection (one per day for 5 days) in the radical cure protocol, in combination with a regimen of chloroquine for control animals (10 mg/kg per day orally for 7 days). The animals used were Indian-origin M. mulatta ranging in age from 2 to 10 years and in weight from 2.5 to 8.5 kg. All animals were Malaria-naïve. Each experiment included two control animals, which did not receive any test compound. In the event of primary infection, when parasitemia reached 5000 parasites/μl, animals were treated with either a regimen of chloroquine at 10 mg/kg per day orally for 7 days (control animals) or test compound for 5 days in the radical cure study. Test compound–treated animals in the prophylactic study were monitored for up to 100 days after drug treatment. If the animals were free of relapse at day 100, they were considered to have been protected against infection. In the event of relapse in either the prophylactic or radical cure study, animals were treated with a uniformly curative regimen of chloroquine at 10 mg/kg per day orally for 7 days, in combination with primaquine at 1.8 mg/kg per day for 7 days.

P. berghei liver-stage phenotyping

HepG2 human hepatoma cells (American Type Culture Collection) were cultured in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum, 2 mM Glutamax, and penicillin-streptomycin mix (100 U/ml) (all from Gibco/Thermo Fisher). Fifteen thousand P. berghei sporozoites were added to 100,000 HepG2 human hepatoma cells seeded 1 day before infection on glass coverslips and allowed to invade. Extracellular sporozoites were washed off at 2 hours after invasion, and the cells were incubated with 460 nM MMV390048 or an equivalent amount of the dimethyl sulfoxide vehicle in the control. Cells were fixed 8 hours after invasion and immunolabeled as in the study of Hanson et al. (22). Images were acquired on a Leica SP8 confocal microscope.

P. falciparum growth conditions and protein extraction for chemoproteomics

P. falciparum parasites (sensitive strain 3D7) were obtained from the Malaria Research and Reference Reagent Resource (www.mr4.org). Red blood cells were obtained from the Spanish Red Cross Blood Bank. P. falciparum strain 3D7 was grown according to the method previously described by Trager and Jensen (34). P. falciparum cultures were maintained at 37°C in RPMI 1640 medium (Gibco) supplemented with 5% AlbuMAX II (Invitrogen) and 150 μM hypoxanthine (Sigma-Aldrich) in a 5% CO2, 90% N2, and 5% O2 atmosphere using fresh uninfected red blood cells at 2% hematocrit. The parasites were grown in vitro until they matured into trophozoites and reached a parasitemia of 12 to 15%. Parasitemia levels were monitored by Giemsa-stained smears of the cultures.

Parasite cultures at trophozoite stage were harvested and washed two times in RPMI 1640 medium (Gibco) and lysed by 0.1% saponin (Sigma) for 5 min at room temperature. Free parasites were sedimented by centrifugation (3500 rpm for 10 min), washed three times with phosphate-buffered saline, and stored at −80°C. Parasites were resuspended in cell lysis buffer [50 mM tris-HCl (pH 7.5), 5% glycerol, 1.5 mM MgCl2, 1.5 mM NaCl, 1 mM Na3PO4, 25 mM NaF, 0.8% NP-40, EDTA-free protease inhibitor cocktail tablets (Roche Diagnostics), and 1 mM DTT] and disrupted by sonication (Sonics Vibra-Cell) using two 30-s cycles on ice at half-amplitude. After centrifugation (140,000g for 1 hour at 4°C), soluble proteins were recovered from the supernatants, and protein concentration of each sample was estimated using the Lowry-based DC assay (Bio-Rad) according to the manufacturer’s instructions. Typical protein concentrations of the lysate were in the range of 5 to 10 mg/ml. Samples were stored at −80°C until use. Preparation of human cell extracts was performed as previously described (21).

Chemoproteomics

Kinobeads were prepared as described (27, 28). Sepharose beads were derivatized with MMV666845 at a concentration of 0.5 mM, as described (26). Beads (35 μl) were washed and equilibrated in lysis buffer incubated at 4°C for 1 hour with 1 ml (2.5 mg) of P. falciparum extract or 1 ml (5 mg) of K562 cell extract, which was preincubated with compound or buffer. Beads were transferred to disposable columns (Mobitec), washed with lysis buffer, and eluted with SDS sample buffer. Proteins were alkylated, separated on 4 to 12% NuPAGE (Invitrogen), stained with colloidal Coomassie, and quantified by isobaric mass tagging and LC-MS/MS. Digestion, labeling with TMT isobaric mass tags, peptide fractionation, and MS analysis were performed essentially as described (27).

To create the fasta file for use in Mascot searching, the “PfalciparumAnnotatedProteins_PlasmoDB-7.1.fasta” file of P. falciparum was downloaded from PlasmoDB (http://plasmodb.org/common/downloads/release-7.1/Pfalciparum/fasta/data/, release date 22 October 2011). This file was supplemented with protein sequences of bovine serum albumin, porcine trypsin, and mouse, rat, sheep, and dog keratins. Decoy versions of all proteins were created and added. The database contains a total of 14,538 protein sequences, 50% forward, 50% reverse.

Criteria for protein quantification were as follows: a minimum of two sequence assignments matching to unique peptides was required (false discovery rate for quantified proteins < 0.1%), Mascot ion score > 10, signal-to-background ratio of the precursor ion > 4, and signal to interference > 0.5 (35). Reporter ion intensities were multiplied with the ion accumulation time, yielding an area value proportional to the number of reporter ions present in the mass analyzer. Peptide fold changes were corrected for isotope purity as described and adjusted for interference caused by coeluting nearly isobaric peaks as estimated by the signal-to-interference measure (36). Protein quantification was achieved using a sum-based bootstrap algorithm (37). Apparent dissociation constants were determined by taking into account the protein depletion by the beads as described (37).

Statistics

In general, sample sizes were small, and all data points are shown in either the figures or tables. Where applicable, details of statistical analysis and the methods used are described elsewhere in Materials and Methods.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/9/387/eaad9735/DC1

Materials and Methods

Fig. S1. Activity of MMV390048 against NF54 asexual blood-stage parasites.

Fig. S2. Lack of growth of schizonts compared to rings after treatment with MMV390048.

Fig. S3. Onset and recrudescence experiments in the P. berghei mouse model using a single dose of 100 mg/kg compound.

Fig. S4. Whole-blood concentrations of MMV390048 after the first dose of treatment during the PfSCID mouse efficacy study.

Fig. S5. Observed and predicted blood PK profiles from PfSCID mouse efficacy study.

Fig. S6. Observed (symbols) and predicted (lines) parasite load in mouse based on the direct effect model and PfSCID mouse efficacy data.

Fig. S7. Prophylactic efficacy of MMV390048 in the P. cynomolgi liver-stage in vitro assay.

Fig. S8. MMV390048-resistant clones show a four- to fivefold shift in IC50 relative to the parental Dd2 strain.

Fig. S9. CNV analysis using the program BIC-Seq (23).

Fig. S10. IC50 of MMV390048 against ZFN-modified parasites with point mutations in PI4K or the parental Dd2 strain.

Fig. S11. Phylogenetic tree of the human (blue) and Plasmodium (red) lipid kinase family.

Fig. S12. Correlation between whole-cell NF54 activity and inhibition of PvPI4K across a selection of MMV390048 analogs.

Fig. S13. Pharmacokinetic profiles after intravenous and oral administration of MMV390048 to mice, rats, dogs, and monkeys.

Fig. S14. Allometric scaling of clearance from different species as a function of body weight.

Fig. S15. Assessment of hemolytic toxicity.

Table S1. Activity against a panel of resistant strains to determine potential for cross-resistance.

Table S2. In vitro parasite reduction data in support of Fig. 1B.

Table S3. Speed assay IC50 results in support of Fig. 1C.

Table S4. Efficacy data in the PfSCID mouse model in support of Fig. 1D.

Table S5. Summary of exposure data in the PfSCID mouse model after oral administration of MMV390048 once a day for four consecutive days.

Table S6. Estimated PK parameters from PfSCID mouse exposure with Ka fixed at 0.5/hour.

Table S7. Estimated PKPD parameters from PfSCID mouse parasitemia and exposure study.

Table S8. Estimated PKPD parameters from PfSCID mouse parasitemia and exposure study indicating highest net kill rate (red).

Table S9. Estimated EC50, EC90, and MIC based on modeling of highest net kill rate.

Table S10. Late-stage gametocyte viability measured by pLDH activity in support of Fig. 2B.

Table S11. Early- and late-stage gametocyte viability measured by luminescence in support of Fig. 2C.

Table S12. Early- and late-stage gametocyte clearance rates in support of Fig. 2D.

Table S13. Gametocytocidal activity.

Table S14. Inhibition of exflagellation in support of Fig. 2E.

Table S15. SMFA indirect mode results in support of Fig. 2F.

Table S16. SMFA direct mode results in support of Fig. 2F (inset).

Table S17. Parasitemias during the prophylactic P. cynomolgi–infected monkey study in support of Fig. 3A.

Table S18. Plasma concentrations of MMV390048 during the prophylactic P. cynomolgi–infected monkey study in support of Fig. 3B.

Table S19. Parasitemias during the radical cure P. cynomolgi–infected monkey study in support of Fig. 3C.

Table S20. Plasma concentrations of MMV390048 during the radical cure P. cynomolgi–infected monkey study in support of Fig. 3D.

Table S21. Number of triplicates at different inocula that became positive after MMV390048 treatment.

Table S22. Whole-genome sequence analysis of three cloned Dd2 parasite lines selected for resistance to compound MMV390048.

Table S23. Correlation between the degree of binding to PfPI4K, activity against PcPI4K, and antiparasitic activity of MMV390048 analogs.

Table S24. Plasma PK parameters after a single intravenous dose of MMV390048 in different species.

Table S25. Plasma PK parameters after a single oral dose of MMV390048 in different species.

Table S26. Chemoproteomics data (provided in Excel).

Table S27. Proteomics data (provided in Excel).

References (3853)

REFERENCES AND NOTES

  1. Acknowledgments: We thank K. Kammerer and T. Rudi for sample preparation, M. Boesche for operating LC-MS instruments, T. Mathieson for the generation of the Plasmodium search database, and P. Papastogiannidis and J. Kamber for assistance in performing the P. berghei in vivo assays. We thank L. D. Shultz and the Jackson Laboratory for providing access to nonobese diabetic SCID IL2Rgc null mice (NSG mice) through their collaboration with GlaxoSmithKline Tres Cantos Medicines Development Campus. Funding: This work was supported by the Medicines for Malaria Venture (MMV) (project MMV09/0002 to K.C.). K.C. is also supported by the University of Cape Town, South African Medical Research Council, and South African Research Chairs Initiative of the Department of Science and Technology through the South African National Research Foundation. D.A.F. is supported by NIH (grants R01 AI109023 and R01 AI103058). A.-M.Z. and C.H.M.K. are supported by MMV and the Wellcome Trust (translational research grant WT078285), and M.J.D. is also supported by MMV. Author contributions: K.C., L.J.S., D.W., J.B., and M.J.W. supervised the work respectively as project director, leader, coordinators, and consultant providing intellectual input and direction. D.G.C., C.L.M., T.P., and Y.Y. were responsible for data compilation and analysis. D.G.C., C.L.M., and T.P. were also responsible for manuscript writing and coordination. P.P.H., T.S.A., M.C.S.L., R.B., and D.A.F. were responsible for MMV390048-resistant mutant generation and sequencing. S.G.-D., M.J.L.-M., M. Bantscheff, and G.D. performed chemoproteomic studies. D.L. was responsible for transmission blocking and P. falciparum SCID mouse efficacy experiments and related pharmacokinetic studies. M.J.D. and A.R. performed gamete formation assays. L.-M.B., J.R., M. Botha, and K.J.D. performed gametocyte assays. K.J.D. and R.W.S. established the SMFA platform at TropIQ. A.M.B. and S.E.Z. established and performed mouse-to-mouse transmission studies. B.C., C.D., A.-M.Z., C.H.M.K., A.T., and P.V. were responsible for in vitro and in vivo P. cynomolgi studies. S.A.C., K.L.W., D.M.S., J. Mannila, J. Morizzi, and K.C.M. performed and interpreted PK studies across all species. S.W. and C.S. performed in vitro P. falciparum and in vivo P. berghei efficacy studies. L.M.S. and F.J.G. were responsible for in vitro PRR analysis. S.M.S. modeled and estimated inhibitory concentrations of MMV390048. M.B.J.-D., I.A.-B., M.S.M., and S.F. did all P. falciparum SCID mouse efficacy and related PK studies, respectively. C.B. and C.S.L. conducted P. vivax PI4K enzyme assays. K.K.H. was responsible for liver-stage P. berghei phenotyping. R.R. was responsible for G6PD experiments. D.A.F., M.C.S.L., G.D., B.C., D.L., S.A.C., and S.W. also contributed to manuscript writing. Competing interests: M.J.W. and MMV are co-inventors on patent no. WO 2011086531 A3 “New anti-malarial agents.” The University of Cape Town and MMV are co-inventors on patent no. WO 2013121387 A1 “Anti-malarial agents.” M.J.W. is a member of the MMV Expert Scientific Advisory Committee and has been an independent paid consultant to the University of Cape Town. S.M.S. is a paid consultant to MMV on PKPD modeling. After retiring as project director from MMV, D.W. was a paid consultant to both MMV and the University of Cape Town. L.M.S., F.J.G., M.B., and G.D. are employees of and hold shares in GlaxoSmithKline. K.J.D. is a shareholder of TropIQ Health Sciences, and C.B. holds shares in Novartis AG. The other authors declare that they have no competing interests.
View Abstract

Navigate This Article