Metabolomic Profiling of the Malaria Box Reveals Antimalarial Target Pathways

Abstract

The threat of widespread drug resistance to frontline antimalarials has renewed the urgency for identifying inexpensive chemotherapeutic compounds that are effective against Plasmodium falciparum, the parasite species responsible for the greatest number of malaria-related deaths worldwide. To aid in the fight against malaria, a recent extensive screening campaign has generated thousands of lead compounds with low micromolar activity against blood stage parasites. A subset of these leads has been compiled by the Medicines for Malaria Venture (MMV) into a collection of structurally diverse compounds known as the MMV Malaria Box. Currently, little is known regarding the activity of these Malaria Box compounds on parasite metabolism during intraerythrocytic development, and a majority of the targets for these drugs have yet to be defined. Here we interrogated the in vitro metabolic effects of 189 drugs (including 169 of the drug-like compounds from the Malaria Box) using ultra-high-performance liquid chromatography-mass spectrometry (UHPLC-MS). The resulting metabolic fingerprints provide information on the parasite biochemical pathways affected by pharmacologic intervention and offer a critical blueprint for selecting and advancing lead compounds as next-generation antimalarial drugs. Our results reveal several major classes of metabolic disruption, which allow us to predict the mode of action (MoA) for many of the Malaria Box compounds. We anticipate that future combination therapies will be greatly informed by these results, allowing for the selection of appropriate drug combinations that simultaneously target multiple metabolic pathways, with the aim of eliminating malaria and forestalling the expansion of drug-resistant parasites in the field.

FIG 1

Optimization of whole-cell metabolomic profiling using the bc₁ complex inhibitor atovaquone. (A) The P. falciparum de novo pyrimidine biosynthesis pathway. Inhibition of the bc₁ complex by atovaquone is shown in red and denotes the blockage in conversion of dihydroorotate to orotate following mitochondrial electron transport disruption. (B) Parasites were treated with atovaquone (10× IC₅₀) at the ring, trophozoite, and schizont stages of intraerythrocytic development. The resulting fold change in ion counts for N-carbamoyl-l-aspartate and dihydroorotate relative to an untreated control is shown. (C) Trophozoite stage cultures were treated with atovaquone (10× IC₅₀) and extracted either in bulk (infected RBCs and uninfected RBCs), as saponin-lysed parasites, or as magnetically purified parasites. The absolute signal was determined using the average total ion counts for N-carbamoyl-l-aspartate and dihydroorotate. (D) Magnetically purified trophozoites were treated with atovaquone at 10× IC₅₀ for 0.5, 1, 2, 4, and 8 h. The parasite response over time was assessed by measuring the fold change accumulation of N-carbamoyl-l-aspartate and dihydroorotate relative to an untreated control. (E) Magnetically purified trophozoites were treated with atovaquone at 1×, 2×, 5×, 10×, 20×, and 50× IC₅₀ atovaquone for 2.5 h. The concentration-dependent inhibition was monitored using the N-carbamoyl-l-aspartate and dihydroorotate response relative to an untreated control. For each atovaquone concentration, three independent biological replicates were assayed with 3 drug-exposed and 3 untreated samples for each. Values are experimental averages ± standard deviations (SD). The red bar in each graph represents the optimal condition chosen for the experimental workflow (see Fig. 2A). NH³⁺, ammonia; HCO₃⁻, bicarbonate; Carb-P, carbamoyl phosphate; N-Carb-Asp, N-carbamoyl-l-aspartate; GAT, glutamine amidotransferase; CPS, carbamoyl phosphate synthase; ACT, aspartate carbamoyl transferase; DHOtase, dihydroorotase; DHODH, dihydroorotate dehydrogenase.

FIG 2

Metabolomic fingerprint analysis of validated compounds with antimalarial activity. (A) Experimental metabolomic profiling pipeline used for all validation and MMV Malaria Box experiments, based on parameters determined from results shown in Fig. 1. (B) Metabolomic profiling was used to measure the parasite metabolic response following drug treatment (see Table S2 in the supplemental material). The data were analyzed using self-organizing maps and are displayed as a suprahexagonal (66) metabolic fingerprint or metaprint. Metabolite clusters were associated with eight generalized metabolic pathways using the KEGG database and were color coded within the suprahexagon base map (left). Metaprints for both atovaquone and WR99210 are displayed as examples, with specific signature metabolites denoted (right). (Mapping locations for all metabolites can be found in Table S3.) (C) We assembled metaprints (right) and clusters (Pearson-Ward distance-clustering) (left) based on the metabolomic profiles for each compound, as determined from the log₂ fold change values of targeted metabolites following drug treatment relative to an untreated (no-drug) control (see Table S2). MoA classifications were performed based on a combination of clustering, metaprint analysis (right), and a priori biochemical knowledge (Table 1). Pyr, pyrimethamine; 2-DG, 2-deoxyglucose; CQ, chloroquine.

FIG 3

Metabolomic profiling of Malaria Box drug-like compounds. Heat map of log₂ fold change values of 113 metabolites, relative to a no-drug control, for 169 drug-like compounds from the Malaria Box and several selected validation compounds from each MoA classification (top). All the displayed log₂ fold changes are the averages for three technical replicates for each compound tested. Distance clustering was performed using the Pearson-Ward method. Selected metabolites are denoted on the right, corresponding to specific metabolic signatures within a particular MoA class, and are color coded. Metaprints (bottom) are included for visualization of the metabolomic profiles associated with each MoA class and contain a validation compound (see Fig. 2C, e.g., KAF246), an MMV Malaria Box representative (e.g., MMV000662), and an average representation of all metabolites within each MoA classification cluster. Metaprint data for the MMV Malaria Box compounds match the map legend in Fig. 2A and Table S3 in the supplemental material since all data were overlaid onto the trained validation set map. The full data set and metaprints are available in Tables S3 and S4, respectively. DHO, dihydroorotate; N-Carb-Asp, N-carbamoyl-l-aspartate.

FIG 4

Metabolomic profiling of the P. falciparum sexual stage using atovaquone. The metaprint for purified stage III-IV gametocytes treated with 1 μM atovaquone (Atv) for 2.5 h (top) using the same experimental screening approach as that for asexual parasite drug treatment assays (Fig. 2A) is shown. The graph highlights the atovaquone-specific signature, which is much reduced compared to that for asexual-stage parasites (Fig. 1). Two independent biological replicates were assayed with 2 drug-exposed and 2 untreated samples for each. Values displayed are experimental averages ± standard errors. The gametocyte data are available in Table S6 in the supplemental material.