A Yeast-Based Drug Discovery Platform To Identify Plasmodium falciparum Type II NADH Dehydrogenase Inhibitors

ABSTRACT

Conventional methods utilizing in vitro protein activity assay or in vivo parasite survival to screen for malaria inhibitors suffer from high experimental background and/or inconvenience. Here, we introduce a yeast-based system to facilitate chemical screening for specific protein or pathway inhibitors. The platform comprises several isogeneic Pichia strains that differ only in the target of interest, so that a compound which inhibits one strain but not the other is implicated in working specifically against the target. We used Plasmodium falciparum NDH2 (PfNDH2), a type II NADH dehydrogenase, as a proof of principle to show how well this works. Three isogenic Pichia strains harboring, respectively, exogeneously introduced PfNDH2, its own complex I (a type I NADH dehydrogenase), and PfNDH2 with its own complex I, were constructed. In a pilot screen of more than 2,000 compounds, we identified a highly specific inhibitor that acts on PfNDH2. This compound poorly inhibits the parasites at the asexual blood stage; however, is highly effective in repressing oocyst maturation in the mosquito stage. Our results demonstrate that the yeast cell-based screen platform is feasible, efficient, economical, and has very low background noise. Similar strategies could be extended to the functional screen for interacting molecules of other targets.

TEXT

Malaria, transmitted through Plasmodium parasites, is one of the most common malignant infectious diseases threatening human health. Great efforts have been devoted to trying to resist and eradicate this disease; however, after an unprecedented period of success, the progress of global malaria control has stalled in recent years. Data for global malaria cases from 2015 to 2017 indicated that no significant improvement was made during this period; malaria still accounts for about 200 million cases and half a million related deaths every year (1). The reasons underlying this malady include the ongoing spread of drug-resistant malarial strains and insecticide-resistant mosquitoes. The emergence of new malarial strains and mosquitoes against conventional drugs or chemicals poses a great challenge for global health and also urgently calls for the development of new antimalarial compounds (2–4).

A significant part of the malarial control effort has been focused on novel drug development and target identification in the Plasmodium parasite. Among the targets, the mitochondrial electron transport chain (mtETC) has been considered a promising one for the development of antimalarial drugs (5–7). Supporting this notion is evidence that the existing antimalarial atovaquone (ATQ), which has been commonly used for more than 20 years in clinics, could inhibit electron transport at the bc1 complex (complex III) by interfering with the ubiquinol oxidation site of cytochrome b (8–10). Atovaquone treatment causes blockage of electron transport, proton pumping prevention, and collapse of mitochondrial membrane potential, leading to organelle dysfunction and the death of the parasite (5, 6, 9, 11). mtETC is largely conserved in stereotypic architecture across different species; nevertheless, a notable significant difference exists in the enzyme of NADH:quinone oxidoreductase (12–15). There are two major types of NADH:quinone oxidoreductase (also called NADH dehydrogenase), both of which catalyze the transfer of electrons from NADH to quinone. In mammals, type I NADH dehydrogenases (also called complex I), consisting of more than 40 subunits and sensitive to rotenone, are responsible for accepting electrons from NADH and pumping protons across the mitochondrial inner membrane (16). In contrast, type II NADH dehydrogenases (NDH2), as occurring in Plasmodium parasites, are a single-subunit rotenone-insensitive enzyme and transport electrons without pumping protons (17–20). NDH2 also occurs in other apicomplexan parasites or pathogenic microorganisms such as Toxoplasma gondii and Mycobacterium tuberculosis, some kinds of bacteria, fungi (including the Baker’s yeast Saccharomyces cerevisiae), and plants (21–24), but not in humans or other mammals.

The drastic difference between these two types of NADH dehydrogenases has attracted substantial attention and effort to develop inhibitors against Plasmodium falciparum NDH2 (PfNDH2). A ubiquinone analogue, HDQ [1-hydroxy-2-dodecyl-4(1H) quinolone; Fig. 1], was first isolated, followed by screenings to seek better HDQ-like compounds. These strivings resulted in the discovery of CK-2-68, which was obtained by an endpoint PfNDH2 enzymatic assay and exhibited a stronger inhibitory activity (25, 26), and RYL-552, which was aided by PfNDH2 crystal structuring and also possessed superior activity against malaria parasites at the blood stage (27). These compounds all inhibit Plasmodium falciparum at the blood stage. However, drug pressure studies with CK-2-68 and RYL-552 showed that both drugs target PfCytB (28), and genetic deletion studies demonstrated that Plasmodium NDH2 was dispensable in asexual-stage parasites (7, 29). Together, these data indicate that the target for both CK-2-68 and RYL-552 is PfCytB, not PfNDH2.

FIG 1.

Molecular structures of ubiquinone (CoQ) and some previously reported NDH2 inhibitors.

A general drawback of conventional screenings such as those mentioned is a high background or false-positive rate. Chemicals that are inaccessible to the assumed target protein, unstable within the cell environment, or generally toxic to many cellular components could all be recovered. To overcome these issues, we developed a cell-based screening method. We constructed isogenic Pichia pastoris strains that differ only in their NADH dehydrogenases, so that a parallel screening would isolate compounds that specifically target only one type of NADH dehydrogenase, such as PfNDH2. The screening is highly efficient, economical, and has a very low false-positive ratio. The same design principle can be extended to the screening of other drug targets.

RESULTS

Construction of a yeast-based platform to screen for PfNDH2-specific inhibitors.

To circumvent the high false-positive rate issue often connected with screenings adopting enzymatic activity assays, we aimed to develop a convenient cell-based platform to search for specific P. falciparum NDH2 (PfNDH2) inhibitors. Normal Pichia pastoris, such as that in mammals, utilizes type I NADH dehydrogenase as the entry point for the oxidation of NADH in the mitochondrial electron transport chain (mtETC) and cannot grow without mitochondrial respiration. We reasoned that if we could replace the endogenous type I NADH dehydrogenase with PfNDH2, we might then be able to use the parental strain and the new strain in a comparative screening for specific inhibitors targeting these two types of NDAH dehydrogenases. To this end, we introduced PfNDH2 into the Pichia sp. strain GS115. We tested whether piericidin A (PA) could behave as an inhibitor of Pichia complex I. PA was originally identified in 1966 as a complex I inhibitor (30). It was subsequently reported that it could be used to inhibit the NADH:ubiquinone oxidoreductase activity in Yarrowia lipolytica (31). The structure of the Yarrowia lipolytica enzyme is very similar to that in Pichia; P. pastoris complex I has 39 subunits in common with Y. lipolytica complex I and 35 in common with the bovine enzyme (32). Meanwhile, it was also shown that PA cannot inhibit the type II NADH dehydrogenase in Saccharomyces cerevisiae (22). Indeed, PA is a specific Pichia complex I inhibitor, evidenced by the fact that PA could inhibit normal Pichia but not PfNDH2-transformed Pichia (Fig. 2A). This also indicates that in this transgenic yeast, PfNDH2 is functional and could replace the function of the innate Pichia NADH dehydrogenase (complex I).

FIG 2.

A yeast-based drug discovery platform. (A) Spotting assay showing the inhibitory action of piericidin A (PA) on Pichia endogenous NADH dehydrogenase on yeast extract-peptone-glycerol (YPG) medium. Yeast was spotted on agar plates with or without the drug by 10-fold serial dilutions in sterile water. PA was inefficient in inhibiting the yeast expressing PfNDH2, indicating that PfNDH2 is able to substitute the endogenous NADH dehydrogenase. (B) The mitochondrial membrane potential of the strains used in this work, as tested by the mitochondrial membrane potential probe TMRE (1 μM). CCCP (15 μM), the uncoupling agent of the mitochondrial membrane. The experiment was repeated three times. (C) A schematic diagram of the three Pichia yeast strains constructed for the screening. Expected yeast growth inhibition results for compounds that may interact with the two types of NADH dehydrogenases are shown on the right. x, growth repression.

We then deleted an otherwise essential component of type I NADH dehydrogenase, Fs1 or Fv1. These newly obtained PfNDH2 strains (Δfs1::PfNDH2 and Δfv1::PfNDH2), with the endogenous type I NADH dehydrogenase replaced, were able to grow normally in the presence of PA (Fig. 2A), confirming that the type II NADH hydrogenase from P. falciparum was indeed perfectly functional in Pichia cells. Considering that the Δfs1::PfNDH2 and Δfv1::PfNDH2 strains are virtually indistinguishable, only one of them, the Δfs1::PfNDH2 strain, was selected for subsequent screening experiments.

Type I NADH dehydrogenase or complex I is able to pump protons, whereas the alternative NADH dehydrogenase NDH2 does not. Replacement of complex I with PfNDH2 in Pichia did not result in observable growth defect. To directly confirm that the mitochondrial membrane potential is unaffected, we used a fluorescent cationic dye, tetramethylrhodamine ethyl ester perchlorate (TMRE), to monitor a possible change. No change was observed (Fig. 2B). In addition, when carbonyl cyanide 3-chlorophenylhydrazone (CCCP), an uncoupler of mitochondrial membrane potential, was used, mitochondrial membrane potentials of all these strains dropped to a similar extent.

For the initial screen, two isogenic P. pastoris strains, the GS115 wild type (WT) and the Δfs1::PfNDH2 strain, which only differ by type I or type II NADH dehydrogenase, were used (Fig. 2C). If a compound specifically suppresses PfNDH2, we would expect the Δfs1::PfNDH2 strain, which harbors exogenously introduced PfNDH2 and has the native complex I disrupted, would not be able to grow. GS115-PfNDH2 is the normal Pichia GS115 strain with heterologous expression of PfNDH2, so it contains both types of NADH dehydrogenases. It was designed for the second-round confirmation assays, to ensure that the lead compounds are indeed working on PfNDH2 instead of just depending on it (i.e., a possible secondary effect).

Compound screening for PfNDH2 inhibitors.

A high-throughput screen was then set up. The chemicals used for the screening were from several small libraries comprising commercial resources and locally supplied ones, which include some natural products and compounds from NIH clinical collections, with 2,432 compounds altogether. We cultured both the wild-type and the Δfs1::PfNDH2 yeast strains until the log phase, then distributed them to 96-well plates that had been preloaded with compounds to be tested. Inhibition was assayed by optical density at 600 nm (OD₆₀₀) to monitoring the growth. After comparing the growth of yeast in each well, ∼60 compounds were found in various degrees inhibiting one or both kinds of yeast (Fig. 3A). Among these, only 16 compounds were inhibitive to the PfNDH2 strain, and most of them were only slightly effective. Five candidates with comparably better potency were chosen for a second round and more refined testing. Each of these five chosen compounds was individually ascertained by plate spotting assay to confirm the phenotype. The 50% inhibitory concentration (IC₅₀) was also estimated. One compound, l-methyl-2-pentyl-4(1H)-quinolinone (MPQ), turned out to be the more prominent one, and its structural formula is depicted in Fig. 3B.

FIG 3.

Compound screening for Plasmodium NADH dehydrogenase inhibitors. (A) Results of the screening are shown. All compounds were plotted on a scattergram in which the relative inhibition rates of PfNDH2 and complex I were indicated on the y axis and x axis, respectively. Each dot represents an individual compound. The red dashed line indicates the division between inhibition and promotion. For a potential lead, only inhibition of cell growth above 60% was considered. A total of 45 compounds exhibited inhibitory effects on both PfNDH2 and complex I (inhibition, >60%). Of these, 16 compounds exhibited more obvious inhibitory effects on PfNDH2 than on complex I. Four of them, associated with better potency and specificity, are highlighted in cyan and circled. (B) Molecular structure of the PfNDH2 inhibitor 1-methyl-2-pentyl-4(1H)-quinolinone (MPQ).

MPQ is a potent and specific PfNDH2 inhibitor.

As shown in Fig. 4A, MPQ has a significant growth inhibitory effect on strains containing only PfNDH2, but has no effect on strains containing type I NADH dehydrogenase (complex I), including the PfNDH2 expression strain with both two types of NADH dehydrogenases (GS115-PfNDH2). The specificity of MPQ against PfNDH2 was further tested in the Baker’s yeast S. cerevisiae, a classic model organism. Similarly to Plasmodium, S. cerevisiae also contains only type II NADH dehydrogenase in its mitochondrial respiratory chain, but includes NDI1, NDE1, and the less functional NDE2 (33, 34), without the L-shaped complex I. NDI1 and NDE1 are the two main NADH dehydrogenases in the Baker’s yeast. Both are homologous to PfNDH2. We wondered whether MPQ is effective on general type II NADH dehydrogenase or if it only targets PfNDH2. We used a strain that deleted yeast-endogenous NDI1 and NDE1 (which on its own could not grow on nonfermentative medium) and heterologously expressed PfNDH2. The newly derived Δndi1 Δnde1::PfNDH2 Baker’s yeast strain was compared to normal Baker’s yeast for MPQ sensitivity. Experimental results show that MPQ specifically inhibits PfNDH2 but has no inhibitory effect on S. cerevisiae NDH2 (Fig. 4B). Even at 100 μM, MPQ does not affect normal S. cerevisiae growth. Therefore, the compound MPQ appears to be a highly specific inhibitor of PfNDH2; it not only fails to suppress Pichia complex I, but also poses little activity against Baker’s yeast NDH2.

FIG 4.

MPQ specifically inhibited PfNDH2 but not the Saccharomyces cerevisiae NDH2 or Pichia type I NADH dehydrogenase complex I. (A) Spotting assay of Pichia yeast on nonfermentable medium (YPG) containing MPQ. Pichia yeast was spotted by 10-fold serial dilutions on agar plates with various concentrations of MPQ in sterile distilled water. MPQ has a significant growth-inhibitory effect on strains containing only PfNDH2. (B) Spotting assay of Baker’s yeast, S. cerevisiae. MPQ has no inhibitory effect on yeast NDH2. (C) The inhibitory effect of MPQ on the respiration of Pichia wild-type and Δfs1::PfNDH2 strains. Relative oxygen consumption represents the respiratory ability of the yeast. Dimethyl sulfoxide (DMSO) was used as the control. AA, antimycin A (complex III inhibitor); PA, piericidin A (complex I inhibitor). (D) MPQ exerted an obvious repression effect on mitochondria isolated from the PfNDH2 strain. The curve indicates a time course of the O₂ content in the buffer. The greater the slope of the curve, the higher the respiratory rate. The dramatic fluctuations/spikes in the curves were caused by probe interference during sample addition, which soon returned to norm. These experiments were repeated three times.

For a more direct proof that MPQ inhibits PfNDH2, we tested the inhibitory effect of MPQ on the respiration of normal/wild-type Pichia and PfNDH2-transgenic Pichia (Δfs1::PfNDH2 strain) cells. We examined the effect of inhibitors on respiration both in whole cells (Fig. 4C) and on isolated intact mitochondria (Fig. 4D). The results showed that MPQ indeed has a significant selective inhibitory effect on PfNDH2. Antimycin A (AA), a respiratory chain inhibitor targeting complex III, inhibited both wild-type and PfNDH2 strains. For mitochondria purified from wild-type Pichia, addition of MPQ had little effect on O₂ consumption; however, MPQ exerted an obvious repression effect on mitochondria isolated from the PfNDH2 strain. On the other hand, piericidin A (PA), an inhibitor against complex I, strongly inhibited wild-type Pichia respiration, while it had little effect on PfNDH2 cells. Worth noting is that the PfNDH2-transgenic Pichia (Δfs1::PfNDH2) strain appeared to have stronger mitochondrial respiration than the wild type. This may be attributable to strong expression of PfNDH2 as directed by the pGAP promoter, in addition to the lack of H⁺ pumping ability of PfNDH2, which may enable electron flow with less resistance. This more robust respiration in the mtETC in turn may compensate for the inability of PfNDH2 to pump H⁺, so overall H⁺ pumping of the PfNDH2 mtETC is comparable to that of the wild type in maintaining the mitochondrial membrane potential and ATP generation.

MPQ is ineffective at inhibiting the growth of asexual-stage Plasmodium parasites.

Genetic ablation studies have shown that parasite NDH2 is dispensable for parasites at the blood stage; the absence of NDH2 does not change the survival status of Plasmodium (7, 29), suggesting that the parasite mtETC remains functionally competent without NDH2 at this stage. Interestingly, previously developed HDQ drugs against PfNDH2 exhibited potent repression of malaria at the blood stage (25, 26, 28, 35). We reasoned that if MPQ is specific against PfNDH2, we would expect MPQ to be an inefficient inhibitor of Plasmodium at the blood stage. Indeed, MPQ poorly inhibits the growth of Plasmodium at the asexual stage (Fig. 5A); the IC₅₀ is about 50 μM, much higher than that in the Pichia yeast. This is in contrast to the data for HDQ and CK-2-68; the IC₅₀ values for HDQ and CK-2-68 are between 50 to 70 nM, against both WT and PfNDH2-knockout asexual blood-stage parasites (29).

FIG 5.

Inhibitory effects of malaria parasites by MPQ. (A) Parasites were cultured for 72 h in the presence of various concentrations of drugs. Parasite viability was measured by fluorometric detection after SYBR green I staining. RYL-522 and artemisinin (ART) were used as the controls. n = 3. (B, C) Inhibition by MPQ of rodent malaria in vitro. The number of merozoites per schizont was counted under incubation with drugs. Microscopic analysis of parasite development revealed a significant decrease in the mean number of merozoites formed per schizont in Plasmodium berghei ANKA cultured in medium supplemented with ART or MPQ. (B) Average curve; (C) scatterplot of the original results. n = 100.

MPQ effectively arrests Plasmodium sporozoite maturation in Anopheles mosquitoes.

PfNDH2 is dispensable for the growth of P. falciparum at the blood stage (29). Although its role during the sexual stage has not been examined, lacking PbNDH2 in Plasmodium berghei parasites resulted in an arrest of oocyst development and maturation, which in turn affects the spread and transfection of malaria (7). We therefore examined the efficacy of MPQ on parasites in its mosquito vector.

We tested the efficacy of MPQ on P. berghei. PbNDH2 displays a high level of identity to PfNDH2 (64% identity across the whole mature protein, compared to 30% with respect to S. cerevisiae Ndi1/NDH2). Blood-stage rodent malaria parasites can be maintained in vitro for one developmental cycle; parasites can replicate inside erythrocytes by schizogony to generate the infectious merozoite forms. One parasite can normally develop 15 or 16 merozoites per schizont on average. When suppressed by drugs, the number will decrease or even approach zero. The numbers of merozoites under the incubation of artemisinin (ART) or MPQ were counted and, as shown in Fig. 5B and C, in the blood stage, MPQ’s inhibition of P. berghei was nearly 4 orders of magnitude worse than that of the control drug, ART, consistent with the results for P. falciparum stated previously.

We subsequently tested the efficacy of MPQ against malaria parasites in mosquitoes. The drug was fed by dissolving it in sucrose solution, and the development of the malaria parasite in the infected mosquitoes was analyzed. Midguts from Anopheles stephensi mosquitoes fed with dimethyl sulfoxide (DMSO; control group) or MPQ (test group) were compared. Treatment with MPQ did not affect oocyst formation, with comparable numbers of oocysts in both the control and test groups (Fig. 6A and B). However, quantification of the oocyst size and the number of sporozoites in the midgut revealed significantly reduced numbers in sporozoite formation for the drug-treated group (Fig. 6C and D). This reduction resulted in a much-diminished infection rate of the mosquitos. These data are consistent with the previously reported phenotype of PbNDH2⁻ parasites during the insect stage (7), suggesting that MPQ could suppress malaria spread through specific inhibition of NDH2 at the mosquito stage.

FIG 6.

Effects of MPQ on P. berghei infection in Anopheles stephensi (A). Oocyst staining in the midgut of mosquitoes 8 days postinfection. Oocysts were stained with 0.5% mercurochrome (orange) and observed at ×200 magnification. ATQ and DMSO were used as positive and negative controls, respectively. Images are representative of at least 10 individual mosquito midguts. Bars, 100 μm. (B). Oocyst numbers in MPQ-treated mosquitoes. Median oocysts number is indicated by the horizontal black bar. (C). Oocyst size of MPQ treated mosquitoes. (D). Sporozoite numbers and infection prevalence in salivary glands of mosquitoes treated with MPQ. (B and D) Each dot represents an individual mosquito. (C) Error bars indicate standard error (n = 12). Significance was determined by Mann-Whitney test in (B and D), and by Student’s-t test (C). ns, no significance; *, P < 0.05; ***, P < 0.001; ****, P < 0.0001.

An ideal malaria inhibitor would pose little toxicity to mammals, including humans. Finally, we tested MPQ cytotoxicity to mammalian cells. No significant inhibition was observed for MPQ at less than 200 μM (IC₅₀ > 400 μM) in HeLa cells (Fig. 7). We then checked in vivo whether MPQ is toxic to mice. Mice were injected with MPQ drugs for 5 days (50 mg/kg). Little abnormality of the mice was observed during the whole process; the body weight and appearance of organs were similar to those of the control group. These initial results suggest that mammalian cells are highly tolerant of MPQ and that it may act as a potential lead compound for the further development of more potent NDH2 inhibitors.

FIG 7.

Cytotoxicity of MPQ in mammalian cells. HeLa cells were exposed to various concentrations of MPQ for 48 h, followed by MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide] assay. No significant inhibition was observed for MPQ below 200 μM concentrations (IC₅₀, >400 μM). Assays were done in triplicates. n = 3; n.s., no significance; *, P < 0.05; ***, P < 0.001.

DISCUSSION

In this work, a high-throughput screening system based on Pichia pastoris was established, and PfNDH2 was used as an example to demonstrate its applicability. Among nearly 2,500 selected compounds, MPQ showed superior inhibitory effects. Based on the inhibitory results of blood-stage and insect stage Plasmodium, as well as a preliminary cytotoxicity study in mammalian cells and mice, the isolated MPQ appears to have a rather specific ability to suppress the function of PfNDH2.

About 0.2% positive hits were obtained in this pilot screen. This is far below the expected hit rates for normal biochemical screens based on enzymatic activity assays. In our cell-based assay, compounds with poor membrane permeability or intracellular bioavailability and with nonspecific activity were expected to be selected against. The obtained compounds were largely confirmed to be positive in the second confirmation assay. This system thus affords several advantages that are important to consider to initiate a high-throughput screen, namely, low background rate, convenience, and low cost.

NDH2, or type II NADH dehydrogenase, differs fundamentally from the architecture of type I NADH dehydrogenase or complex I such as that in mammalian cells. This has inspired researchers over the past decade to search for NDH2 inhibitors as possible avenues of repressing relevant infectious organisms. As the starting enzyme of mtETC, the inhibition of NADH dehydrogenase was thought to result, possibly, in arrest of mitochondrial respiration and also that of pyrimidine biosynthesis, an essential pathway for malaria parasites. It turns out that NDH2 is only essential at the mosquito stage, but not in the asexual blood stage (7, 29), likely due to other components of the mtETC feeding electrons to coenzyme Q, bypassing PfNDH2 at the blood stage. Notably, we also found a link between the mechanism of the antimalarial drug artemisinin and mtETC in our Saccharomyces cerevisiae model; regulating the expression level of NDH2 could alter the inhibitory effects of artemisinin (36).

Currently, the most known NDH2 potential inhibitor is HDQ; it was first reported to be active in the obligate aerobic yeast Yarrowia lipolytica in 2005 (37). HDQ was later found to display good inhibition effects on malaria parasites and Toxoplasma gondii, making NDH2 a great potential drug target (24, 38). HDQ analogs were further developed subsequently by medicinal chemistry studies (25, 26, 35, 39). Since it was shown that the parasite NDH2 can be deleted by targeted gene disruption without interfering with the growth of Plasmodium in the pathogenic asexual blood stages (7, 29), it is expected that a specific PfNDH2 inhibitor would not be effective in suppressing Plasmodium during the pathogenic blood stages, and therefore HDQ may have other intracellular targets. Indeed, the HDQ derivative CK-2-68 was later reported working through PfCytB, which is its sole target (28, 29, 35).

Compound MPQ has a similar structure to the classic drug HDQ, but the specificity of MPQ against PfNDH2 seems much higher than those of HDQ and its derivative CK-2-68. We observed order-of-magnitude difference in their IC₅₀ values for inhibition of blood-stage Plasmodium (IC₅₀ values of HDQ and CK-2-68 were ∼12 nM, while the IC₅₀ value of MPQ was about 50 μM). Therefore, although these compounds share the quinolone core, certain changes on the side chain affect the structure-activity relationship that leads to the off-targeting of HDQ and derivatives.

The absence of PfNDH2 disrupts oocyst development and sporozoite generation. Considering the indispensability of NDH2 in the mosquito stage for malaria parasites, PfNDH2 inhibitors may find use in controlling malaria in areas where mosquitos are endemic. Along this line of thinking, it is encouraging to see that mosquito exposure to the mtETC cytochrome b inhibitor atovaquone, when deposited on a bed net or a glass substrate surface, could cause full parasite arrest in the midgut and prevent transmission of infections (40). It is hoped that more potent MPQ derivatives perfected by future chemistry studies may find applications under similar schemes.

MATERIALS AND METHODS

Yeast strains, culture, and chemicals.

Standard yeast medium and growth conditions were used. Pichia pastoris GS115 (his4) and Saccharomyces cerevisiae BY4742 (MATΔ his3Δ1 leu2Δ0 lys2Δ0 ura3Δ0) strains were used as wild-type controls. Yeast was normally plated on yeast extract-peptone-dextrose (YPD; 2% glucose as carbon source) or yeast extract-peptone-glycerol (YPG; 2% glycerol as carbon source) agar plates, supplemented with different concentrations of drugs when necessary. Screening experiments were performed in liquid YPG medium, and OD₆₀₀ was measured to reflect the yeast growth.

Chemical libraries for the screening were purchased from the National Compound Resource Center, including the BBP-1440NPs library (1,440 compounds), NIH clinical collection library (446 compounds), Endocannabinoid Library (60 compounds), and Natural Products Library (486 compounds). The link is http://www.chemicallibrary.org.cn/.

Plasmid construction and yeast transformation.

For Pichia, we used the commercial insertion expression plasmid pGAPzαc (select marker zeocin) to generate the PfNDH2 expression construct under the pGAP promoter. The PfNDH2 codon was partially optimized and gene synthesized for expression in yeast. The mitochondrial targeting sequence (ATGCTATCGAAAAATTTGTATAGTAACAAGAGGTTGCTCACCTCGACAAATACGCTAGTCAAATTCGCTTCCACCAAA) is from the yeast BY4742 Ndi1 gene. Yeast transformation is via electroporation, with procedures strictly following the standard protocol recommended by the vendor’s manual (part no. 25-0174, p 16 to 18; Invitrogen).

For the Saccharomyces cerevisiae Δndi1 Δnde1::PfNDH2 strain, the double-knockout mutant (Δndi1 Δnde1) in the BY4742 background was constructed previously in the lab (36), and a synthesized PfNDH2 gene (after some codon optimization) was transformed into the strain through homologous recombination and Leu⁺ selection. Introduction of PfNDH2 restored the respiration capability of the Δndi1 Δnde1 mutant so that it could grow on nonfermented medium (e.g., YPG). Transformation was performed with the polyethylene glycol (PEG)-lithium acetate chemical method.

Parasite maintenance and culture.

P. falciparum 3D7 parasite was used in this study. Parasites were maintained in continuous culture, kept at 37°C in human erythrocytes with complete medium (one pack of solid powder RPMI 1640 supplemented with 5 g Albumax II, 50 mg hypoxanthine, 5.96 g HEPES, 2.2 g sodium bicarbonate, and 25 mg gentamicin; Invitrogen, Thermo Fisher Scientific). Cultures were kept in a CO₂/N₂ incubator filled with 5% O₂, 5% CO₂, and 90% N₂. Prior to assay initiation, the level of parasitemia was measured by light microscopy of an aliquot of the stock culture following Giemsa staining. All malaria parasites were synchronized. The early-ring-stage parasites were used for the experiment.

P. berghei ANKA was collected from the tails of infected mice at 0.5 to 2% parasitemia, washed in warmed incomplete RPMI 1640, and added to 96-well plates with complete culture medium (RPMI 1640, 20% fetal bovine serum [FBS], 25 mM HEPES, 0.85 g/liter NaHCO3, and 25 mg/1iter gentamicin) at 1% hematocrit (final volume, 200 μl). Plates were incubated at 37°C under an atmosphere of 5% O₂, 5% CO₂, and 90% N₂ for 22 to 30 h. Mature schizonts were visualized by microscopy after Giemsa staining and quantified manually. Experiments were scored in a blind manner by different investigators.

Cell culture.

HeLa cells were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% FBS, penicillin (100 U/ml), and streptomycin (100 μg/ml) at 37°C in a humidified atmosphere of 5% CO₂.

Yeast, parasite, and mammalian cell inhibition assays.

For yeast (P. pastoris) liquid growth assays or drug screening assays, yeast was inoculated in YPG with an OD₆₀₀ of 0.01, cultured until the log phase (OD₆₀₀ = 1.2), and then incubated with or without drugs at an OD₆₀₀ of 0.05. Incubation was performed at 30°C for 24 h, and the growth was monitored by OD₆₀₀ measurement to determine the inhibitory effect of the drugs. For growth testing on agar plates (spotting assay), yeast was 10-fold serially diluted with double-distilled water (ddH₂O) and then spotted on YPG plates with or without drugs for 60 h.

Parasites P. falciparum growth was as described previously (41); compounds were added to 2% hematocrit cultures at 2% parasitemia in a total assay volume of 200 μl in 96-well plates, and incubated at 37°C in a humidified air incubator containing 5% CO₂. After 72 h of growth, fluorescence was measured by a microplate fluorometer (Fluoroskan Ascent; Thermo) after adding the probe SYBR green I, with excitation and emission wavelengths centered at 485 and 538 nm, respectively.

Mammalian cell inhibition followed a previously described method (41). Briefly, cells were inoculated into 96-well plates in a 100-μl volume and, after incubating for 24 h, treated with freshly prepared compounds diluted in medium (100 μl), to obtain a final volume of 200 μl and drug concentrations ranging from 0 to 600 μM. Following 48 h of compound exposure, an MTT [3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide] assay was used to measure the cell viability. The absorbance at 570 nm was measured with a multiwell plate reader (Multiskan GO; Thermo).

Isolation of yeast mitochondria.

Experimental procedures followed a protocol described previously (42) with little modification. Crude mitochondria were purified from Pichia yeast cells by Zymolyase-20T (MP Biomedicals) digestion and differential centrifugation from 1,000 to 10,000 × g. The mitochondrial fractions were further purified by a sucrose gradient centrifugation.

Oxygen consumption.

After growth to an OD of 1.00, yeast cells were incubated with different compounds for 5 min before centrifugation, washed, and resuspended in respiration buffer. Oxygen consumption was measured using a Clark-type oxygen electrode (OXELP; World Precision Instruments) in a total volume of 1 ml. For isolated intact mitochondria, the respiration substrates, ADP, and inhibitors were directly added into the chamber in order.

Mosquito rearing, treatment, and Plasmodium infection.

Mosquitoes (Anopheles stephensi) were reared at 28°C and 80% humidity and maintained on 10% sucrose. Adult females were fed on BALB/c mice for a blood meal and allowed to lay eggs on wet filter paper. Newly emerged mosquitoes were provided with a sugar meal containing MPQ (0.1 mM and 0.4 mM) and ATQ (0.04 mM) and equivalent DMSO dissolved in 10% sucrose for 5 days. Plasmodium berghei ANKA parasites constitutively expressing green fluorescent protein (GFP) were maintained in BALB/c mice by serial blood passage and regular mosquito transmission (43, 44). Mosquitoes treated for 5 days were allowed to feed on mice with 3 to 5% parasitemia. Mosquitoes that were not fully engorged were removed 24 h post blood meal. After imbibing a blood meal, mosquitoes were maintained at 21°C and fed with the same diet as before for 8 days and 21 days for oocyst and sporozoite assays, respectively. For the oocyst assay, the midguts of mosquitoes were dissected, and oocysts were counted under the microscope 8 days postinfection. For measurement of oocyst size, midguts were stained with 0.5% mercurochrome (Shanghai Winguide Huangpu Pharmaceutical, China) as described (45) and photographed under a microscope (×200 magnification). The images were analyzed by Fiji to measure the size of the oocysts (46). For the assay of sporozoite number in salivary glands, salivary glands of individual mosquitoes were dissected, and the infected salivary glands were ground in 30 μl phosphate-buffered saline (PBS). Sporozoites in 10 μl homogenate were counted using a hemocytometer, and the total number of sporozoites was calculated.