Mutational analysis of an antimalarial drug target, PfATP4

Significance

Plasmodium falciparum ATP4 (PfATP4) is a Na⁺ efflux pump and represents an important target for antimalarial drugs with nanomolar potency. However, the structure of PfATP4 is unknown, prompting the development of new methodologies to investigate the structure/function relationship. Here, we introduce a dynamic homology modeling approach to a) identify key residues essential for PfATP4 function and b) provide a structural basis to understand resistance-associated mutations. To validate these predictions, we developed a genetic system to manipulate the PfATP4 gene to assess the phenotypic consequences of such changes. Our results support the utility of combining homology modeling and genetics to gain functional insights into an antimalarial drug target. A similar approach could be applied to other essential genes in malaria parasites.

Abstract

Among new antimalarials discovered over the past decade are multiple chemical scaffolds that target Plasmodium falciparum P-type ATPase (PfATP4). This essential protein is a Na⁺ pump responsible for the maintenance of Na⁺ homeostasis. PfATP4 belongs to the type two-dimensional (2D) subfamily of P-type ATPases, for which no structures have been determined. To gain better insight into the structure/function relationship of this validated drug target, we generated a homology model of PfATP4 based on sarco/endoplasmic reticulum Ca²⁺ ATPase, a P2A-type ATPase, and refined the model using molecular dynamics in its explicit membrane environment. This model predicted several residues in PfATP4 critical for its function, as well as those that impart resistance to various PfATP4 inhibitors. To validate our model, we developed a genetic system involving merodiploid states of PfATP4 in which the endogenous gene was conditionally expressed, and the second allele was mutated to assess its effect on the parasite. Our model predicted residues involved in Na⁺ coordination as well as the phosphorylation cycle of PfATP4. Phenotypic characterization of these mutants involved assessment of parasite growth, localization of mutated PfATP4, response to treatment with known PfATP4 inhibitors, and evaluation of the downstream consequences of Na⁺ influx. Our results were consistent with modeled predictions of the essentiality of the critical residues. Additionally, our approach confirmed the phenotypic consequences of resistance-associated mutations as well as a potential structural basis for the fitness cost associated with some mutations. Taken together, our approach provides a means to explore the structure/function relationship of essential genes in haploid organisms.

As the continued threat of resistance to mainstay antimalarials grows in endemic regions, significant efforts have been undertaken to discover new antimalarial drugs. Among the many antimalarials discovered over the past decades are multiple chemical scaffolds targeting PfATP4, a Plasmodium falciparum Na⁺/H⁺ pump (1–6). Two of these compounds have progressed to Phase IIb clinical trials, showing potent activity with extremely rapid clearance of circulating P. falciparum and Plasmodium vivax parasites (7, 8). Inhibition of PfATP4 results in an influx of Na⁺ in the parasite cytoplasm followed by dramatic consequences, including swelling of the parasite (9), changes to the lipid composition of the parasite plasma membrane (PPM) (10), induction of premature schizogony (10), and eryptosis (3), all of which have the potential to contribute to parasite death.

PfATP4 belongs to a superfamily of P-type ATPases that use adenosine triphosphate (ATP) hydrolysis to transport ions or metabolites across membranes (11, 12). These pumps are divided into five classes, P1 to P5, and are further divided into subclasses (A to D) based on their function and membrane localization (12–14). PfATP4 belongs to the P2-type ATPases, of which the sarco/endoplasmic reticulum Ca²⁺ ATPase (SERCA) and Na⁺/K⁺ ATPase have been extensively investigated (14). PfATP4 is further classified into the P2D subfamily, which contains pumps in fungi, lower plants, and some unicellular eukaryotes (15–17). Pumps in this family transport monovalent cations, such as Na⁺, and are called the ENA (exitus natrus) ATPases (15–17). Phylogenetic analysis of P2D-type ATPases revealed distinct clades within this subclass with strong bootstrap values (>90%), one composed of pumps from fungi, kinetoplastids, bryophytes, and Entamoeba, and the other consisting of apicomplexan, chromerid, stramenopile, chlorophyte, and dinoflagellate pumps (Fig. 1A and SI Appendix, Fig. S1 and Table S1). A previous study (18) also suggested that ATP4 pumps from apicomplexans and chromerids form a distinct fifth clade of P2-type ATPases. Biochemical and structural properties that distinguish these pumps from other P2D ATPases have not been explored.

Fig. 1.

PfATP4 models using SERCA as a template. (A) Phylogenetic analysis of P2D-type ATPases, depicted as a radial tree diagram with P2A-type ATPases as outgroups; Red dots indicate >90% bootstrap support. A rectangular phylogram version of the tree is shown in SI Appendix, Fig. S1. (B) Catalytic and intermediate states of PfATP4 shows the possibility of 4 conformational states namely 2Na⁺E1P-ADP, E2P, H₃E2P-ATP, and H₃E2-ATP (C) trajectory traces from MD simulations of the 4 states shows all states are energetically stable from 25 to 115 ns (D) three-dimensional molecular models of the 4 conformational states are shown as ribbons and colored according to their domain architecture (TM helices—green, A-domain—blue, N-domain—yellow and P-domain – red). The predicted PfATP4 inhibitor pocket is depicted as a pink circle in the TM domain. RMSF maps for each conformational state are shown below the structures, and A, N, and P domains are marked for clarity. (E) Hydrogen bonding pattern at the proximity of the Na+ coordination site in each conformational state is shown with H-bonds marked as dotted yellow lines and labeled with distance, participating residues are shown as licorice sticks and colored atom type (carbon: gray/cyan/magenta, nitrogen: blue, oxygen: red) and labeled with residue number.

At present, no structure of a P2D-type ATPase has been determined. Since PfATP4 is a validated and attractive antimalarial drug target, it is crucial to understand the structure/function relationship of this protein. So far, attempts to express functional PfATP4 in heterologous systems have been unsuccessful, preventing such investigations. To circumvent this limitation, we developed a high-confidence homology model of PfATP4 undergoing its catalytic cycle in its explicit membrane environment. Using this model, we identified several critical residues likely involved in PfATP4 function. We performed mutational analysis of these residues using a genetic system that permits assessment of their consequences on parasite biology. In addition to identifying residues involved in PfATP4 function, our model was also used to assess the structural consequences of mutations conferring resistance to different PfATP4 inhibitors. In particular, we identified a possible structural basis for the fitness defect observed with the development of certain mutations. Mutations resulting in fitness defects can reduce the spread of resistance; therefore, it is crucial to study their structural consequences.

Results

Homology Modeling Using SERCA Intermediates.

P-type ATPases cycle between four states, accompanied by massive structural rearrangements of the cytosolic and transmembrane (TM) domains (Fig. 1B) (13, 19–21). The structure of each state has been solved for SERCA, a P2A-type ATPase. Since no structures are available for PfATP4 or other P2D-type ATPases, we created a homology model and refined the structures for the four conformational states of PfATP4 in the Na⁺ transport cycle using the SERCA protein as a template. Analysis of the trajectories from molecular dynamics (MD) simulations showed that all systems are well equilibrated after 25 ns of simulation and have stable trajectories for the next 100 ns (Fig. 1C). A comparison of these trajectories for the different conformational states reveals that the E2-ATP states form the most stable conformations. Mapping the root mean square fluctuations (RMSF) of the residues from different domains during the simulations showed that TM3, TM10, and the N-domain had major conformational changes in all the states while the E1 state additionally showed fluctuations in the P-domain (Fig. 1D). In the modeled structures, E409, E934, D963, and E1176 formed the coordination sites for binding Na⁺; these residues align with E309, E771, D800, and E908 of SERCA that form the coordination sites for binding Ca²⁺ (22). Previous studies on the SERCA protein indicate that the E2 to E1 transition is triggered by the deprotonation of E309 (23) and a ∼110° rotation of the A-domain (24), leading to the large movement in TM1 and TM2. We analyzed the intramolecular interactions, especially H-bond interactions, in the vicinity of E409 and the ion binding sites in the four models (Fig. 1E). In the 2Na⁺E1-ADP state, the two Na⁺ ions were stably coordinated at the ion binding sites throughout the simulation. The simulations also revealed that in addition to E409, E934, D963, and E1176 residues, the backbone carbonyl of V404 and the side chain of N959 also helped coordinate the Na⁺ ions in the 2Na⁺E1-ADP state. These two residues are also conserved in SERCA and participate in Ca²⁺ binding in the E1P state (25). The removal of the two Na⁺ ions led to the E2P state in which E409, E934, and E1176 are still deprotonated. The deprotonated E409 did not engage in an H-bond interaction with any residues (Fig. 1E). In the H₃E2P-ATP state, E409 is protonated, and it engaged in an H-bond interaction with N207 while the amine backbone of E409 engaged in an H-bond interaction with the N957 amine side chain. The dephosphorylated H₃E2-ATP state had a significantly different H-bond interaction network than the other states. The protonated E409 engaged in three H-bond interactions: the side chain carbonyls H-bond with the backbone carbonyl of V404 and the N957 side chain, whereas the backbone carbonyl made an H-bond with the N207 side chain (Fig. 1E). These findings indicate that the protonated E409 is crucial for stabilizing the H₃E2-ATP state. An E309N single mutation in SERCA protein destabilized the E2 state and transitioned to an E1-like conformation (23). The loss of H-bonding between V203 and E309 was found to be a trigger for the E2 to E1 transition in SERCA (23). Besides the above intramolecular interactions, the study suggested that the A and N domains are more compact in the E2 state than in the E1 state (26). We measured the center of mass distances between the A and N domains in our PfATP4 models, which are presented in SI Appendix, Table S2. The E2P state had the shortest distance (~27 Å), while the other 3 states had a ~50 Å distance separating the A and N domains. The displacement of ~25 Å from the E1 to E2 transition is similar to the SERCA protein E1 to E2 transition (27). The size of the binding cavity in each structure was calculated using CASTp (28) (Ver 3.0). The solvent-accessible volume of the binding cavity was the highest in H₃E2-ATP (614 ų), followed by H₃E2P-ATP (328 ų) and 2Na⁺E1P-ADP (292 ų). The size of the binding cavity was the least in the E2P state (180 ų).

AlphaFold (29) is an AI program that can predict proteins’ three-dimensional structures from a given sequence using deep-learning neural network models. The equilibrated models of PfATP4 described above were compared with the AlphaFold-predicted structure, and the results are shown in SI Appendix, Table S2. Superpositioning of the AlphaFold-generated PfATP4 model on the modeled 2Na⁺E1P-ADP, E2P, H₃E2P-ATP, and H₃E2-ATP structures resulted in root-mean-squared distance (RMSD) values of 6.24, 7.36, 7.31, and 6.96 Å respectively for the complete structure and 3.29, 5.14, 3.72, and 3.09 Å, respectively, for the TM regions. These results suggest that the AlphaFold structure is not representative of any of the 4 conformational states of PfATP4.

A Genetic System to Investigate PfATP4.

We developed a genetic system to validate our homology model’s predictions and the importance of specific residues in PfATP4. First, we generated transgenic parasites expressing endogenous PfATP4 tagged with a c-Myc epitope at the C terminal end under the control of the TetR-DOZi aptamer system (30) using single crossover recombination in the NF54attB parasite line (called NF54attBPfATP4:Myc; Fig. 2A; descriptions of all parasite lines are given in SI Appendix, Table S4). In these parasites, expression of PfATP4 is maintained in the presence of anhydrotetracycline (aTc) (30). After successful integration, the knockdown of PfATP4 by the removal of aTc was confirmed using both anti-c-Myc and anti-PfATP4 antibodies (Fig. 2B). As observed in a previous study (30), PfATP4 knockdown caused parasite demise within two cycles (Fig. 2C). Western blot analysis of membrane preparations from these parasites revealed the presence of PfATP4 as a 140 kDa protein in SDS-polyacrylamide gel electrophoresis (Fig. 2D).

Fig. 2.

Genetic system to investigate PfATP4. (A) Schematic representation of the single crossover recombination strategy that was used to generate NF54attB PfATP4:Myc parasite line in which PfATP4 expression is regulated by the TetR- DOZi aptamer system at the endogenous locus. (B) Western blots of NF54attB PfATP4:Myc parasite lysate probed with anti-c-Myc and anti-PfATP4 antibody showing successful knockdown of endogenous PfATP4 upon removal of aTc within 48 hours postinvasion (hpi). (C) Growth curves of NF54attB PfATP4:Myc parasites in the presence and absence of 250 nM aTc. Cultures were split 1:2 at 48 and 96 hpi, and percent cumulative parasitemia was calculated by multiplying parasitemia and the split factor over time, representative of three biological replicates. (D) Western blots after SDS-PAGE (Panel 1) of membrane preparations from untagged and NF54attB PfATP4:Myc parasites probed with anti-c-Myc depicting a 140kd band corresponding to PfATP4. (E) Schematic representation of the merodiploid (MD) parasite line expressing aTc-regulated NF54attB PfATP4:Myc at the endogenous locus and Dd2 PfATP4:3xHA at the cg6∷attB locus expressed under the control of the calmodulin promoter. (F) Growth curves as estimated by cumulative parasitemia of wildtype and mutant merodiploid lines with and without aTc. Representative growth curves (of three biological replicates) of merodiploid PfATP4_WT parasites. (G) SDS-PAGE followed by western blotting of merodiploid NF54attB PfATP4:Myc PfATP4_(WT):3xHA parasite lysate maintained with and without aTc over 48, 96, and 144 hpi probed with anti-c-Myc and anti-HA to detect the expression of PfATP4 from the endogenous (Top) and ectopic loci (Bottom) indicating that the ectopic expression of PfATP4 remains stable upon knockdown of the endogenous PfATP4. (H) Western blots showing endogenous and ectopic PfATP4 expression from the merodiploid NF54attB PfATP4:Myc PfATP4_(WT):3xHA parasites with anti-c-myc and anti-HA showing codominant expression. (I) Proteomic analysis showing counts of peptide spectra originating from PfATP4 in two biological replicates of IPs from untagged and c-Myc-tagged parasites. Peptides common to both samples were considered as nonspecific contaminants during pulldown and were eliminated during analysis. Except for PfATP4, no other proteins were represented uniquely in c-Myc-tagged pulldowns. (J) Proteomic analysis after IP of untagged (PfATP4:ut) and merodiploid (PfATP4:wtMD) parasites showing the presence of only cognate peptide spectra spanning the polymorphic region.

The NF54attB PfATP4:Myc parasite line was used to generate a merodiploid parasite line (NF54attB PfATP4:Myc/PfATP4_(mut):3xHA) in which a copy of constitutively expressed PfATP4 (originating from the Dd2 strain and tagged with a 3xHA epitope) was integrated at the cg6∷attB ectopic locus (Fig. 2E). Conditional knockdown of endogenous PfATP4 expression by withdrawing aTc, was complemented by the presence of an ectopically expressed wild type PfATP4 (Fig. 2F). Western blot analysis using anti-Myc, and anti-HA antibodies confirmed knockdown of endogenous PfATP4 and constitutive expression of ectopic PfATP4, respectively (Fig. 2G). Immunopulldown (IP) experiments with either anti-c-Myc or anti-HA antibodies showed that the endogenous and ectopically expressed PfATP4 did not associate with each other (Fig. 2H). This was further supported by the fact that a single polymorphism, G1128 in NF54 (the endogenous allele) and R1128 in Dd2 (the ectopic allele), distinguished the individual alleles. Proteomic analysis of eluates by LC-MS/MS revealed the presence of G1128 spanning peptides only in c-Myc eluted samples and R1128 predominantly in HA eluted samples (Fig. 2 I and J). Together, these data validate our merodiploid system wherein the function of each allele of PfATP4 can be independently assessed.

Assessing Likely Na⁺ Coordination Residues in PfATP4.

As described above, our homology model identified residues E409, E934, D963, and E1176 as residues coordinating Na⁺ ions (Fig. 3A). To validate these predictions, we conducted mutational analyses by creating multiple merodiploid lines with mutations in ectopic PfATP4 using the NF54attB PfATP4:Myc parasites (Fig. 3B). Using these merodiploid lines, we assessed potential consequences to protein function and localization as well as to overall parasite growth. Of the four predicted Na⁺ coordination residues, we mutated E409 and E1176 to isoleucine residues to remove the charge and polarity while maintaining the relative size. The merodiploid parasite lines (PfATP4_E409I, PfATP4_E1176I) were grown in the presence and absence of aTc to assess whether the mutated allele could complement parasite growth. In the presence of aTc, the PfATP4 alleles from both loci were codominantly expressed (SI Appendix, Fig. S5 B and C). Withdrawal of aTc from the medium resulted in cessation of parasite growth compared to complementation by PfATP4_WT (Fig. 3 C–E).

Fig. 3.

Assessing predicted Na⁺ coordination sites in PfATP4. (A) Three-dimensional structure of the H₃E2-ATP state of PfATP4 is depicted as ribbons and colored gray in a membrane environment with residues from Na⁺ coordination sites and phosphorylation site labeled and shown as licorice sticks and colored atom type (carbon: gray, nitrogen: blue and oxygen: red) with the inlet depicting the magnified Na⁺ binding site. (B) Schematic representation of the integration strategy that was used to generate merodiploid NF54attB PfATP4:Myc/PfATP4_(mut):3xHA parasites in which endogenous PfATP4 expression can be conditionally regulated by the TetR-DOZi aptamer system. The mutated PfATP4 allele is constitutively expressed under the control of the RL2 promoter from the cg6::attB locus. (C–F) Representative growth curves of (two biological replicates) of merodiploid wild type, PfATP4_E409I, PfATP4_E1176I, and PfATP4_D451N, parasites grown in the presence or absence of 250 nM aTc. Cultures were split 1:4 at 48, 96, and 144 hpi. Parasitemia was determined by flow cytometry. Cumulative parasitemia was determined by multiplication of parasitemia and split factor (in this case, 4) over the time course. (G) Saponin sensitivity assay assessed by Western blot analysis of PfATP4_mut parasites grown with and without aTc for 60 h, treated with 0.02% saponin, and probed with an anti-aldolase antibody to measure cytosolic leakage. Parasites were grown in aTc for 60 h and then treated with 10 nM KAE609 for 2 h prior to saponin treatment were used as a positive control.

The fact that these mutations could not support parasite growth suggests that there may be a defect in Na⁺ efflux by the mutated PfATP4s from the parasite cytosol. In a previous study, we demonstrated that the influx of Na⁺ due to PfATP4 inhibition led to the accumulation of cholesterol in the PPM (10). This effect is reflected by parasites acquiring sensitivity to treatment with saponin, a cholesterol-dependent detergent. The saponin sensitivity can be assessed by the leakage of cytosolic proteins, such as aldolase, after saponin treatment (10). We used this observation as a surrogate for assessing PfATP4 function. Parasites were grown in the presence and absence of aTc, and their sensitivity to saponin was assessed via western blot using an anti-aldolase antibody. As a control, parasites were grown with aTc and treated with a PfATP4 inhibitor (KAE609) for 2 h prior to saponin treatment. Merodiploid wild-type parasites (PfATP4_WT) grown with or without aTc were not saponin sensitive, as indicated by the absence of aldolase leakage following saponin treatment (Fig. 3G; quantitation is given in SI Appendix, Fig. S6). In contrast, PfATP4_E409I and PfATP4_E1176I parasites grown without aTc showed evidence of cholesterol accumulation in the PPM, similar to parasites treated with KAE609 (Fig. 3G and SI Appendix, Fig. S6). These results suggest that these mutations confer a functional defect to the protein.

To determine whether this defect could be due to dysregulation of Na⁺ coordination or improper protein localization to the PPM, we conducted an immunofluorescence assay (IFA) using an anti-HA antibody and an antibody to a known parasitophorous vacuolar membrane (PVM) marker, PfEXP2 (31). The PPM, where PfATP4 localizes, is closely apposed to the PVM and, therefore, can be used to visualize proper protein localization. Similar to PfATP4_WT, mutant proteins from both PfATP4_E409I and PfATP4_E1176I parasites were closely associated with PfEXP2, indicating appropriate localization of the mutated PfATP4 (Fig. 5 and Dataset S2). These data indicate that PfATP4_E409I and PfATP4_E1176I parasites confer a fitness defect likely due to dysregulation of Na⁺ homeostasis since the protein appears to be properly localized to the PPM.

Fig. 5.

Immunofluorescence assays, as described in the Materials and Methods section, were conducted to observe the localization of ectopic PfATP4_Mut:3xHA from (all lines). Parasites were fixed and stained with DAPI (blue), anti-HA antibody (green), and anti-EXP2 antibody (red). Images are representative of ~20 captured images. (Scale bars denote 5 μm.)

Assessing a Potential Phosphorylation Site.

P-type ATPases derive their name from an intermediate phosphorylated state that is shared by all proteins in this family (13, 14). Multiple X-ray crystal structures of intermediate states of the catalytic cycle of SERCA have revealed massive structural rearrangements of the cytoplasmic and TM domains (13, 19, 20) These structural rearrangements are determined by both the bound ion and the phosphorylation state of the pump (13). In P2-type ATPases, the phosphoacceptor during autophosphorylation is an aspartate side chain in a universally conserved DKTG motif in the P domain (13). To transition from the E1 unbound state to the 2Na⁺E1P-ADP state, ATP must be bound and hydrolyzed to ADP. For this, ATP is coordinated between the N and P domains, where the conserved aspartate breaks phosphodiester bond of ATP to create ADP (13).

Sequence alignments predict that residue D451 in PfATP4 is homologous to D351 in SERCA, which was identified to be part of the DKTG motif. To assess this, we created a merodiploid parasite line with a D451N mutation in PfATP4 to force the pump into a stable, dephosphorylated state. We then assessed the ability of PfATP4_D451N parasites to complement endogenous PfATP4 knockdown. In the absence of aTc, endogenous PfATP4 was not expressed, while ectopic expression was maintained (SI Appendix, Fig. S5D). However, PfATP4_D451N could not complement endogenous PfATP4 deficiency compared to the PfATP4_WT (Fig. 3F). After evaluation of parasite growth, we next assessed PfATP4 function using the saponin sensitivity assay as described above. Western blot analysis demonstrated that PfATP4_D451N parasites grown without aTc showed evidence of cholesterol accumulation in the PPM similar to treatment with KAE609, indicating that the mutation conferred a functional defect to PfATP4 (Fig. 3G). Next, we conducted IFA to localize the PfATP4_D451N mutant protein. In contrast to the Na⁺ coordination site mutations described above, PfATP4_D451N showed both cytoplasmic and membrane localization (Fig. 5 and Dataset S2), suggesting a defect in efficient localization of the mutant protein lacking the conserved phosphorylation site.

Assessing PfATP4 Inhibitor Resistance-Associated Mutations.

Although PfATP4 is an attractive target for multiple chemical scaffolds, resistance to these compounds arises relatively easily in vitro. So far, over 20 different mutations conferring resistance to various PfATP4 inhibitors have been identified (1–3, 32). Most of these mutations do not seem to confer a fitness cost to the parasite. However, one exception is the P412T mutation, which confers resistance to spiroindolone and DHIQ classes of PfATP4 inhibitors (3). Previous studies generated P412T mutant parasites through continuous drug exposure. Thus, a possibility remains that the observed fitness defect may be due to a mutation elsewhere in the genome. To assess this possibility, we generated a merodiploid line bearing a P412T mutation expressed from the ectopic locus (PfATP4_P412T). The ability of the mutated allele to complement endogenous PfATP4 knockdown was tested by the withdrawal of aTc (Fig. 4A). Drug-selected resistant parasites bearing a P412T mutation were reported to remain alive, albeit with diminished growth (3). Similarly, merodiploid parasites with P412T mutation also showed reduced growth upon the removal of aTc, but continued growth under −aTc condition led to parasite death. The reasons for this discrepancy are not clear but may indicate a compensatory adjustment in parasites that were selected through continuous drug selection. We also engineered a merodiploid line expressing PfATP4 with two mutations, one known to confer resistance to pyrazoleamide (V178I) and the other to spiroindolone (G223R). Withdrawal of aTc did not affect the parasite’s growth phenotype (Fig. 4B). Immunofluorescence examination showed that in PfATP4_P412T and PfATP4_V178I+G223R parasites, PfATP4 localized to the parasite surface, indicating appropriate localization of the protein (Fig. 5).

Fig. 4.

Structural basis for resistance-associated mutations of PfATP4. (A and B) Representative growth curves of (two biological replicates) of PfATP4_P412T and PfATP4_V178I+G223R parasites grown in the presence or absence of 250 nM aTc. Cultures were split 1:4 at 48, 96, and 144 hpi. Parasitemia was determined by flow cytometry. Cumulative parasitemia was determined by multiplication of parasitemia and split factor (in this case, 4) over the time course. (C and D) Structural superpositioning of wildtype H₃E2-ATP of PfATP4 (represented as gray ribbons in all the models) and the P412T (represented as blue ribbons) (C) and V178I+G223R (represented as green ribbons) (D) mutant models are shown. In the magnified panels, the residues at the ions binding sites and the mutation sites are labeled and represented as licorice sticks and colored atom type with—nitrogen: blue, oxygen: red. The relative displacement of the helices or residues is indicated as black lines, and the displacement is measured in Å.

To understand the structural consequences of P412T and V178I+G223R mutations, the H₃E2-ATP structure from MD, which was previously simulated for 1 μs to apparent equilibrium, was used. The homology models of the mutants were simulated for an additional 100 ns. A comparison of the RMSD plots suggested that the structures were stably equilibrated throughout the simulations (SI Appendix, Fig. S2A). Superpositioning of these mutant models onto the wild-type H₃E2-ATP structure showed RMSDs of 2.91 and 3.51 Å for the P412T and V178I+G223R models, respectively (SI Appendix, Table S2). Analysis of the mutant models showed significant movement of TM3 toward TM1 by 5.5 Å in the V178I+G223R model (Fig. 4 C and D). This movement may lead to a decrease in the volume of the PfATP4 inhibitor binding cavity, which is lined by several residues from TM1 and TM3 regions, and may explain the loss of efficacy and drug resistance observed with these mutants. G223 lies in the cytoplasmic region of the PfATP4 structure, and the double mutation V178I+G223R results in the formation of a new salt bridge interaction between D160 and R223. This is a significant structural change that might impact the overall conformation of the E2 state, affecting drug sensitivity.

To understand the effects of these mutations on the Na⁺ coordination sites, the relative displacements of E409, E934, D963, and E1176 were measured (Fig. 4 C and D). A previous study on SERCA showed that the movement of E309 (corresponding to E409 in PfATP4) is critical for the E2 to E1 transition (23). In the P412T mutant, the displacement of E409 was ~2 Å, which was greater than the shift seen in the V178I+G223R mutant. The movement of E409 in the mutants could alter the H-bond network in the vicinity of E409, particularly with V404, N959, and N207, and affect the stability of the E2 state (SI Appendix, Fig. S2B). The H-bond distances between the protonated side chain of E409 and the V404 backbone were 2.2, 2.8, and 1.8 Å for the wild type, P412T, and V178I+G223R, respectively (SI Appendix, Fig. S2 B–D). There was a significant change in the H-bond distances between the carbonyl backbone of E409 and N207 in the P412T mutant compared to other mutants. This H-bond exists only in the E2 state and is broken in the E1 state in the PfATP4 models (SI Appendix, Fig. S2C). Thus, based on these modeling results, we predict that the P412T mutation may decrease the stability of the E2 state and affect the E1/E2 equilibrium, potentially affecting parasite fitness.

Response of merodiploid parasites to antimalarials.

We used these merodiploid lines to assess their responses to three different antimalarial compounds (KAE609, PA21A092, and artemisinin) in the presence and absence of aTc. As shown in Table 1, Na⁺ coordination mutations (PfATP4_E409I and PfATP4_E1176I) showed hypersensitivity to treatment with PfATP4 inhibitors in the absence of aTc. PfATP4_D451N parasites showed hypersensitivity to KAE609 in presence as well as absence of aTc, which remains unexplained. Susceptibility of all mutants to artemisinin, however, was the same in the presence and absence of aTc.

Table 1.

Drug susceptibility of merodiploid lines to KAE609, 21A092, and artemisinin were quantified using the hypoxanthine incorporation assay

Mutation	Condition	Compound (EC50) nM
		KAE609 (Spiroindolone)	PA21A092 (Pyrazoleamide)	Artemisinin
Parental line (NF45attB PfATP4:Myc)		+aTc	0.313 ± 0.092	5.69 ± 0.61	n/a
Merodiploid WT	+aTc	0.605 ± 0.034	8.6 ± 0.42	14.30 ± 0.4
	−aTc	0.245 ± 0.021	3.09 ± 0.35	14.0 ± 0.57
E409I (Na+ coordination)	+aTc	0.181 ± 0.018	2.97 ± 0.32	12.79 ± 0.8
	−aTc	0.026 ± 0.024*	0.53 ± 0.38*	14.88 ± 1.7
E1176I (Na+ coordination)	+aTc	0.246 ± 0.042	5.3 ± 0.96	13.71 ± 0.73
	−aTc	0.049 ± 0.020*	0.82 ± 0.05*	13.53 ± 0.44
D451N (Phosphorylation cycle)	+aTc	0.049 ± 0.020	0.82 ± 0.05	13.53 ± 0.44
	−aTc	0.077 ± 0 .004	3.62 ± 0.36*	14.69 ± 0.56
P412T (Resistance to spiroindolones and DHIQ’s)	+aTc	0.405 ± 0.031	8.6 ± 0.21	13.46 ± 1.7
	−aTc	0.061 ± 0.005*	1.56 ± 0.21*	24.79 ± 1.4
V178IG223R (Resistance to pyrazoleamides and spiroindolones)	+aTc	2.38 ± 0.053	54.23 ± 8.3	12.9 ± 1.1
	−aTc	4.28 ± 0.18	135.8 ± 5.6	11.3 ± 4.0

The values are means from three biological replicates exposed to a range of compound concentrations with SE. * indicate >threefold difference in EC₅₀ values for parasites grown in +aTc or −aTc conditions with P value of <0.05 as determined by the Student t test.

Resistance-associated mutations were also assessed for their response to drug treatment. PfATP4_V178I+G223R parasites were resistant to both KAE609 and PA21A092, as expected. However, PfATP4_P412T parasites, which express the P412T mutation, showed hypersensitivity to both KAE609 and PA21092. This discrepancy could be due to a compensatory adjustment in drug pressure-selected parasites bearing a P412T mutation, which would not be present in our merodiploid line.

Discussion

Malaria parasites encode thousands of proteins, a large percentage of which are essential for parasite survival (33, 34). In the search for new antimalarials to counter the ever-present threat of resistance emergence, structural and functional details of potential targets are of great value. However, at present, such details are available only for a handful of proteins encoded by malaria parasites. Here, we built a robust homology model for a validated antimalarial drug target PfATP4 using extensive MD simulations based on four principal catalytic states of the Ca²⁺ pump, SERCA (Fig. 1 B–D). This dynamic model predicted motions of various domains of PfATP4 as it undergoes the ion pumping cycle, as well as critical residues involved in its function. While the AI-based tool AlphaFold is remarkable in providing models of proteins for which homologous structures are not available, the models it predicts are static and do not take into consideration the explicit environment in which the proteins exist. Therefore, the approach we used here has a significant advantage in gaining structural information on proteins for which authentic structures are not available. Clearly, since these are models, their predictions would need to be experimentally validated.

Such experimental validation for PfATP4, however, poses challenges since expression of this protein in its functionally active form in heterologous system has not been possible thus far. Since PfATP4 is an essential gene in P. falciparum, its mutational analysis could be impeded due to the haploid nature of the parasite genome. To circumvent this, we developed a system to assess the phenotypic consequences of site-directed mutagenesis of PfATP4 by generating merodiploid parasites in which two alleles of PfATP4 are present (Fig. 2E). The endogenous allele on chromosome 12 was engineered to be conditionally expressed using the TetR-DOZI/aptamer regulation, whereas the second allele was constitutively expressed from an ectopic locus on chromosome 7 of P. falciparum. The endogenous allele was tagged with a c-Myc epitope, and the second allele was tagged with a 3x-HA epitope. In addition, we used the NF54 parasites for generating transgenic lines with ectopic PfATP4 allele from the Dd2 parasites that had a nonsynonymous polymorphism, which permitted distinction between the encoded proteins when subjected to immunoprecipitation using epitope-specific antibodies (Fig. 2 I and J). Lack of coimmunoprecipitation, as well as proteomic analyses, showed that PfATP4 molecules encoded by the two alleles did not interact with each other (Fig. 2H). Thus, the function of the complementing PfATP4 allele could be independently evaluated when expression of the endogenous allele was suppressed by withdrawing aTc from the culture medium.

Our model predicted acidic residues within PfATP4 that coordinate Na⁺ as it is being pumped across the parasite plasma membrane. We tested these predictions in two merodiploid lines, each expressing mutated PfATP4 bearing single amino acid change involved in Na⁺ coordination (E409I and E1176I). Unlike the wild-type PfATP4 allele, these mutant alleles failed to complement the knockdown of endogenous PfATP4 (Fig. 3 C–E). Immunofluorescence analysis showed the mutated protein localized to the parasite surface, indicating that there was no apparent defect in the transport of the protein (Fig. 5). We also examined the mutation of the predicted conserved aspartate (D451N) involved in the phosphorylation cycle of the pump. These parasites exhibited a relatively more severe growth inhibition phenotype when expressing this allele in the absence of the endogenous allele (Fig. 3F). Interestingly, unlike the mutations involving Na⁺ coordination residues, this mutation resulted in both cytosolic and plasma membrane localization of PfATP4 (Fig. 5), suggesting a possible link between the phosphorylation state of the protein and its efficient transport to the parasite surface.

In a previous study, we have shown that inhibition of PfATP4 for a short 2 h period results in massive changes in parasite morphology and lipid homeostasis (10). One such change involves reversible accumulation of cholesterol in the parasite plasma membrane, which can be assessed as leakage of cytosolic proteins following treatment with cholesterol-dependent detergents such as saponin (10). We have also found that the inhibition of another pump, PfNCR1 (P. falciparum Niemann-Pick type C1-related), also results in a similar accumulation of cholesterol in the parasite plasma membrane observed as saponin sensitivity (35). Since PfNCR1 has a sterol-binding domain, it is predicted to be involved in cholesterol homeostasis in the parasite and, thus, likely to be secondarily inhibited following PfATP4 inhibition (36). As shown in Fig. 3G, mutations that are predicted to affect PfATP4 function also result in the induction of saponin sensitivity in merodiploid parasites. We hypothesize that PfNCR1 requires a Na⁺ gradient across the parasite plasma membrane to maintain cholesterol homeostasis and that a collapsed Na⁺ gradient in PfATP4 mutant parasites leads to inhibition of PfNCR1, resulting in cholesterol accumulation in the parasite plasma membrane.

The utility of the homology model we have built was tested to assess the structural consequences of drug resistance-associated mutations in PfATP4. We engineered two merodiploid lines expressing such mutations and assessed their growth drug resistance phenotypes. Parasites bearing V178I+G223R mutations did not show changes in their growth phenotype and showed increased resistance to drugs when grown in the absence of aTc. Parasites bearing the P412T mutation, however, showed a severe growth defect when grown in the absence of aTc and were hypersensitive to PfATP4 inhibitors. MD simulations of this mutation revealed an impact on the relative stability of the intermediate states, which could explain the phenotype observed.

In summary, we have combined MD simulations and genetic manipulations of an antimalarial drug target to gain insights into its function. This approach also provides insights into the structure–activity relationship for hundreds of compounds synthesized as part of an effort to develop antimalarials that target PfATP4, which will be described elsewhere (Nepal et al.)

Materials and Methods

Homology Modeling.

The amino acid sequence of PfATP4 was obtained from the Uniprot database (37) (accession number: Q9U445) and is composed of 1,264 residues. A position-specific BLAST suggested the human SERCA protein as the closest structural homolog. Since SERCA undergoes significant conformational changes across its pumping cycle (Fig. 1B), we modeled the analogous conformational states of the PfATP4. The structure of each PfATP4 conformational state was generated using the homology modeling module of Molecular Operating Environment software [MOE, ver 10.1 (38)] using the corresponding crystal structures of the SERCA as templates. The 2Na⁺E1P-ADP was modeled with the SERCA structure [PDB:3BA6 (25)] with the Na⁺ ion coordinated by E409, E934, D963, and E1176, which were modeled in the deprotonated state. In this state, D451 is phosphorylated, and one ADP molecule is bound in the nucleotide-binding domain. ADP molecule was mimicked by AMP phosphoramidate as in the SERCA crystal structure. The E2P conformational state was modeled with SERCA E2P structure [PDB:3B9B (25)] as a template with D451 in the phosphorylated state but without any Na⁺ ions. The H₃E2P-ATP conformational state was modeled with the corresponding SERCA structure [PDB:3FGO (39)] with D451 as phosphorylated and with one ATP molecule bound to the nucleotide-binding domain and E409, E934, and E1176 in their protonated states. The H₃E2-ATP conformation was modeled with the corresponding SERCA structure [PDB:2C88 (40)] as a template. In this state, D451 was dephosphorylated and bound to the ATP binding domain wherein ATP was modeled as the phosphomethylphosphonic acid adenylate ester (ACP) and E409, E934, and E1176 were protonated.

All-Atom MD Simulations.

The models were equilibrated with all-atom MD simulations in a membrane environment using NAMD software (41). A membrane patch composed of 70% 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine, 20% 1-palmitoyl-2-oleoylphosphatidylethanolamine, 5% cholesterol, and 5% phosphatidylinositol 4,5-bisphosphate (PIP2) was simulated. The number of lipids and the corresponding lipid patch dimension for each system are detailed in SI Appendix, Tables S2 and S3. The system was modeled with an ensemble with constant moles, pressure and temperature using the Nose–Hoover thermostat and the Martyna-Tobias-Klei barostat, and the protein, lipid, and other heteroatoms representing ATP and adenosine diphosphate (ADP) were represented using the CHARMM36 Force field (42)and TIP3P model for water, and the entire system was neutralized with 0.15 M KCl. MD simulations for each system were carried out for a production run of 115 ns with 2 fs time step. The H₃E2-ATP structure was simulated for an additional 1 µs on the Anton2 supercomputer (43) with the same force field and time step. The force-field parameters for AMP phosphoramidate and phosphomethylphosphonic acid adenylate ester (ACP), which mimics ADP and ATP molecules, were generated from the CGennFF program (44). P412T and G223R+V178I mutants were constructed from the H₃E2-ATP state using the final conformation of the 1 µs simulation. The mutated systems were then simulated for an additional 100 ns on the Anton2 supercomputer (43). In the wild-type and mutant model simulations, the ACP molecule, which mimics ATP, was constrained in its position with a harmonic restraining potential of 5 kcal/mol, which was released in the final steps of the simulation. Protein Data Bank (PDB) files of the models generated have been deposited in a publicly accessible database (45).

P. falciparum Culture Maintenance.

Asexual P. falciparum parasites were cultured in commercially obtained Human O+ erythrocytes and maintained at 2.5% hematocrit in RPMI 1640 medium supplemented with 0.5% w/v Albumax II (Gibco by ThermoFisher Scientific), sodium bicarbonate (2.1 g/L, Corning by ThermoFisher Scientific), HEPES (15 mM, MilliporeSigma), hypoxanthine (10 mg/L, Fisher Scientific), gentamycin (50 mg/L, VWR), and further supplemented with selection drugs as needed. Parasite cultures were maintained at ~5% parasitemia or lower (unless otherwise specified) under an atmosphere of 90% N₂, 5% O₂, and 5% CO₂. Cultures were maintained at 5% parasitemia or lower unless otherwise specified. Periodically, Mycoplasma testing was done to ensure that cultures were Mycoplasma negative.

Generation of NF54attB PfATP4:Myc Transgenic Line.

To generate conditional knockdown of endogenous NF54 PfATP4:Myc tagged parasites, we utilized the TetR-DOZI system. The original pMG75-ATP4 vector (30) was a gift from Jacquin Niles (MIT, Boston, MA) in which 1,042 base pair of PfATP4 covering 2,751 to 3,792 nucleotides of the gene (primer sequences shown in SI Appendix, Table S5 with the plasmid map in SI Appendix, Fig. S3). To allow appropriate single cross-over recombination at the endogenous locus and to facilitate the development of merodiploid constructs, the original vector was modified to remove the attP site as described by Ke et al. (46). This plasmid was used to transfect NF54attB parasites as previously described (47). The transfected parasites were cloned by limited dilution in the presence of aTc (250 nM) and in the absence of the selection drug, blasticidin S (BS-2.5 μg/mL).

Generation of Merodiploid Parasite Lines.

To generate merodiploid parasite lines, full-length PfATP4 was PCR amplified from Dd2 wild-type parasite genomic DNA and cloned into pSC-B, a PCR blunt-end cloning vector, using the StrataClone Blunt PCR cloning kit (Agilent Technologies). After Sanger sequencing verification of selected clones, various mutations of PfATP4 were made using the oligonucleotides, as shown in SI Appendix, Table S5. Following sequence confirmation of the mutations, the pSC-B PfATP4_mut was digested (AvrII+BsiWI) and ligated into a pLN-yDHOD attP vector (modified from pLN-ENR-GFP) (48) digested at the same sites and having either a calmodulin or RL2 promoter to drive expression of PfATP4 (SI Appendix, Fig. S3). The final transfection vectors were verified by whole plasmid nanopore sequencing. Merodiploid parasite lines were generated by transfection of NF54attB PfATP4:Myc parasites with pLN vectors containing mutations of PfATP4 along with an integrase plasmid. Transfection vectors were coprecipitated with the integrase plasmid (48) using 0.5 M sodium acetate (pH 7) and ethanol. Transfections were carried out as previously described (46). After electroporation, parasites were cultured in media supplemented with 250 nM aTc for 48 h, at which time, G418 (10 μg/mL), DSM1(1.5 μg/mL), and BS (2.5 μg/mL, InvivoGen) were added. After 1 wk, G418 was removed, and parasites were routinely maintained in media supplemented with aTc, BS, and DSM1. Proper integration at the attB site was assessed using the primers listed in SI Appendix, Table S5 (48) and depicted in SI Appendix, Fig. S4.

The generation of the PfATP4_V178I+G223R parasite line differed slightly from the other merodiploid lines. Genomic DNA from Cmpd 2-1 resistant parasites was PCR amplified to obtain the PfATP4_V178I mutant gene (2). This DNA was then ligated into a PSC vector at the AvrII+BsiWI sites. Following this, primers listed in SI Appendix, Table S5 were used to engineer a G223R mutation in PfATP4, resulting in a V178I+G223R double mutant parasite line (PfATP4_V178I+G223R).

Parasite Proliferation Assays.

To assess for phenotypic defects due to PfATP4 mutations, transgenic parasites were synchronized using 0.5 M alanine. Trophozoite-stage parasites were isolated by Percoll gradients and washed twice in RPMI 1640 medium by centrifugation at 1,500×g for 10 min to remove residual aTc. Parasites were then split into three replicates, each containing a starting parasitemia of 0.5 to 1%, and grown in media with or without aTc for 8 d. Cultures were split at the trophozoite stage, with 75% of the culture collected for Western Blot analysis and 25% retained to continue parasite growth. The collected samples for growth assay were fixed with 4% formaldehyde and stored at 4 C. The samples were washed with 1× PBS to remove the fixative and stained with SyBr Green in 1× PBS for 30 min at room temperature. The stained pellet was washed 3× to remove the unbound stain and analyzed with BD-Acuri C6 flow cytometer. 100,000 cells were counted in each sample and data were analyzed with vendor-provided BD-Acuri C6 software. Cumulative parasitemia was calculated from day 0 to day 8 and plotted using GraphPad Prism.

PfATP4 Antibody Generation.

PfATP4 sequences encoding amino acids 449 to 760 were amplified by PCR and cloned into pET28a to generate His-tagged protein in Escherichia coli BL21Codonplus (DE3) RIL bacteria (Agilent Technologies) bacteria. The protein was isolated using a Ni-NTA column (Qiagen). 25 μg of isolated protein was used to immunize mice by subcutaneous injection with complete Freund’s adjuvant and boosted three times with incomplete Freund’s adjuvant to generate PfATP4 antibody.

Western Blot Analysis.

Trophozoite stage parasites were lysed using 0.05% Saponin in 1× PBS supplemented with 1 μg/mL protease inhibitor cocktail (P8215, MilliporeSigma). Parasites were spun down at 1,500×g for 10 min at room temperature. Parasite pellets were washed twice in 1× PBS until the lysate was clear. After washing, the pellets were resuspended in 100 µL of a solution containing 1% sodium dodecyl sulfate (SDS), 5% β-mercaptoethanol, and 1% bromophenol blue, mixed by vigorous pipetting and stored at −20 °C. The lysate was spun down at 17,000×g for 10 min, and the resulting supernatant was used for SDS-polyacrylamide gel electrophoresis (SDS-PAGE). 40 μL of each sample was loaded into each lane, and electrophoresis was performed using a standard protocol. Membranes were blocked with 5% w/v fat-free milk powder for 90 min. Blots were then incubated with monoclonal anti-mouse HA antibody (F-7, sc-7392, Santa Cruz Biotechnology, Lot # 12923) at 1:10,000 dilution or monoclonal anti-mouse Myc antibody (9B11, Cell Signaling 2276S) at 1:10,000 followed by incubation with either secondary goat pAb to mouse IgG HRP (ab97040, Abcam, Lot # GR3219575-4) at 1:10,000 or secondary mouse antibody to rabbit IgG HRP (Santa Cruz Biotechnology, Lot # L1819). As a loading control, blots were later probed with a rabbit anti-PyEXP2 primary antibody (a gift from James Burns, Drexel University) at 1:10,000, followed by a secondary mouse antibody to rabbit IgG HRP (see previous) at 1:10,000. Blots were then imaged on a Bio-Rad ChemiDoc™MP Imaging System.

Saponin Sensitivity Assay.

Ring-stage parasites were tightly synchronized with 0.5 M alanine and split into three T-25 flasks; aTc was removed from one of these. After 60 h, one flask grown in the presence of aTc was treated with 10 nM of KAE609 for 2 h at 37 °C. To monitor sensitivity to saponin, trophozoite stage parasites were released from erythrocytes using 0.02% Saponin in 1× PBS supplemented with 1 μg/mL protease inhibitor cocktail (P8215, MilliporeSigma). Parasites were centrifuged at 1,500×g for 10 min at room temperature. Parasite pellets were washed three times in 1× PBS until the lysate was clear. After washing, the pellet was resuspended in ~five volumes of RIPA buffer supplemented with 1 µg/mL of protease inhibitor cocktail (P8215, MilliporeSigma). The samples were further analyzed via western blot for leakage of cytosolic aldolase. Blots were incubated with anti-rabbit Plasmodium aldolase HRP antibody (ab38905, Abcam, Lot # GR289448-4) at 1:10,000 dilution and processed as described above.

Immunofluorescence Assays.

Trophozoite stage parasites were fixed in 4% paraformaldehyde + 0.001% glutaraldehyde and rotated at 37 °C for 1 h and then 4 °C overnight. Samples were washed with 1× PBS + 0.001% Tween and spun down at 3,000 rpm for 5 min. Samples were then permeabilized with 0.25% Triton X-100 and rotated at room temperature for 10 min. Following permeabilization and washing, samples resuspended at ~100 mg/mL were reduced with 0.1 mg/mL NaBH₄ while rotating at room temperature for 5 min. Following washing, samples were blocked w/v 3% BSA for 90 min and then incubated with monoclonal anti-mouse HA antibody (F-7, sc-7392, Santa Cruz Biotechnology, Lot # 12923) at 1:500 and rabbit anti-PyEXP2 primary antibody (a gift from James Burns, Drexel University) at 1:500 followed by goat anti-mouse IgG H+L Alexa Fluor™Plus 488 (Invitrogen, A32723, Lot # SG251135) at 1:500 and goat anti-rabbit IgG H + L AlexaFluor™ 568 (Invitrogen, A11036, Lot # 2447870) at 1:500. Samples were then equilibrated and resuspended in Gold Anti-Fade + DAPI (S2828, ThermoFisher Scientific) and mounted onto slides for microscopy. Parasites were then visualized, and z-stack images were captured using a Nikon Ti microscope. Images were then 2-dimentionally deconvoluted and processed using Nikon NIS Elements (5.30.02) Imaging Software.

Phylogenetic Analysis.

Sequences of P2D-type ATPases from species representing a broad range of eukaryotic groups were identified from previous studies (15, 18) and BLAST (49) searches of NCBI (50) and EupathDB (51) databases. The sequences chosen for alignment are listed in SI Appendix, Table S1. Details of the software packages and methodology used for phylogenetic analysis are given in SI Appendix, Supplemental Methods.

Growth Inhibition Assay.

All parasite growth inhibition assays were performed in triplicate in 96-well plates as described by Desjardin et al. (52). P. falciparum-infected erythrocytes at 1.0% initial parasitemia and 1.5% hematocrit were exposed to various concentrations of the indicated drug/inhibitor 24 h and then pulsed with 0.5 μCi of ³H-hypoxanthine for 24 h. Untreated parasitized and unparasitized red blood cells were incubated concurrently with the treated parasites as controls for growth and radioactive precursor incorporation. The 96-well plates were then frozen at −80 °C for 24 h to promote erythrocyte and parasite lysis. Following freezing, the plates were warmed to 37 °C, and parasites were harvested onto EasyTab™-C Self-Aligning Glass Fiber Filters (Packard, Meridian, CT). The filters were then dried completely and placed into an Omnifilter™ 96-well plate filter case (Packard, Meridian, CT). In order to measure the beta-radiation of incorporated ³H-hypoxanthine, 30 μL of OmniScint™ (Packard, Meridian, CT) scintillation fluid was added to each well of the Omnifilter case. The Omnifilter was then covered with a TopSeal and counted with a TopCount™ radiation counter. The incorporation of radioactivity into nucleic acids served as a measure of cell proliferation and is calculated by TopCount as counts per minute (cpm). The total cpm of the nonparasitized red blood cells was subtracted from the cpm of all other wells. The percent growth was plotted using GraphPad Prism software by graphing the cpm versus the drug/inhibitor concentration. The cpm of each sample was converted to percent by dividing it by the cpm of the untreated control. The best-fit curve was calculated by adjusting the lowest percent to equal zero and the highest percent to 100. From this best-fit curve, the concentration of the inhibitor that inhibits 50% of the parasite growth (IC₅₀) is calculated.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (PDF)

Dataset S02 (PDF)

Data, Materials, and Software Availability

PDB files of the models generated are available at Figshare (45). All other data are included in the article and/or supporting information.