High-Content Screening of the Medicines for Malaria VenturePathogen Box for Plasmodium falciparum Digestive Vacuole-Disrupting Molecules Reveals Valuable Starting Points forDrug Discovery
Jie Xin Tong,a Rajesh Chandramohanadas,b Kevin Shyong-Wei Tana
aLaboratory of Molecular and Cellular Parasitology, Department of Microbiology and Immunology, Yong LooLin School of Medicine, National University of Singapore, Singapore
bPillar of Engineering Product Development, Singapore University of Technology and Design, Singapore
ABSTRACT Plasmodium falciparum infections leading to malaria have severe clinicalmanifestations and high mortality rates. Chloroquine (CQ), a former mainstay of ma-laria chemotherapy, has been rendered ineffective due to the emergence of wide-spread resistance. Recent studies, however, have unveiled a novel mode of action inwhich low-micromolar levels of CQ permeabilized the parasite’s digestive vacuole(DV) membrane, leading to calcium efflux, mitochondrial depolarization, andDNA degradation. These phenotypes implicate the DV as an alternative target ofCQ and suggest that DV disruption is an attractive target for exploitation by DV-disruptive antimalarials. In the current study, high-content screening of the Medi-cines for Malaria Venture (MMV) Pathogen Box (2015) was performed to selectcompounds which disrupt the DV membrane, as measured by the leakage of intra-vacuolar Ca2� using the calcium probe Fluo-4 AM. The hits were further character-ized by hemozoin biocrystallization inhibition assays and dose-response half-maximal(50%) inhibitory concentration (IC50) assays across resistant and sensitive strains.Three hits, MMV676380, MMV085071, and MMV687812, were shown to demonstratea lack of CQ cross-resistance in parasite strains and field isolates. Through systematicanalyses, MMV085071 emerged as the top hit due to its rapid parasiticidal effect,low-nanomolar IC50, and good efficacy in triggering DV disruption, mitochondrialdegradation, and DNA fragmentation in P. falciparum. These programmed celldeath (PCD)-like phenotypes following permeabilization of the DV suggests thatthese compounds kill the parasite by a PCD-like mechanism. From the drug develop-ment perspective, MMV085071, which was identified to be a potent DV disruptor,offers a promising starting point for subsequent hit-to-lead generation and optimiza-tion through structure-activity relationships.
With approximately 200 million cases of Plasmodium falciparum infection in 2016,malaria remains a global health burden. Although this striking figure represents
a 17% decrease in the overall incidence rate compared to that in 2010 (1), the spreadof artemisinin resistance had been progressively reported in the Southeast Asian regionsince 2008 (2–4). Coupled with chloroquine (CQ) resistance, which has existed since the1950s (5), there exists a dire ongoing need for novel therapeutic antimalarials. Inaddition to the canonical mechanism of action of inhibiting hemozoin formation (6),administration of low-micromolar levels of CQ was shown to permeabilize the parasite’sdigestive vacuole (DV) membrane and lead to Ca2� efflux (7), mediating downstream
Received 3 October 2017 Returned formodification 5 November 2017 Accepted 20December 2017
Accepted manuscript posted online 8January 2018
Citation Tong JX, Chandramohanadas R, TanKS-W. 2018. High-content screening of theMedicines for Malaria Venture Pathogen Boxfor Plasmodium falciparum digestive vacuole-disrupting molecules reveals valuable startingpoints for drug discovery. Antimicrob AgentsChemother 62:e02031-17. https://doi.org/10.1128/AAC.02031-17.
cellular events, such as mitochondrial outer membrane permeabilization (MOMP) andDNA degradation. These phenotypes were similar to the hallmark features of mamma-lian programmed cell death (PCD), indicating an alternative mode of action for thedrug, allowing it to be repurposed, despite the resistance to the drug. Furthermore,given that the parasite’s DV is the major organelle for metabolic activities and itscontents represent important and viable drug targets (8), the findings provide excitingpossibilities for the rational design of novel antimalarial drugs targeting the DV.
Following the success of Malaria Box, the Medicines for Malaria Venture (MMV)assembled the Pathogen Box (2015), consisting of 400 diverse compounds activeagainst the malaria parasite and pathogens responsible for a wide range of neglectedtropical diseases. A high-content screening assay capitalizing on conventional flowcytometry capabilities alongside high-resolution single-cell imaging was developedand performed to select Pathogen Box compounds that disrupt the DV membrane. Dueto the abundance of calcium in the parasite’s DV (9), Ca2� release into the cytosol couldserve as a surrogate indication of DV permeabilization (10), and DV disruption was thusquantified by measurement of the amount of intravacuolar Ca2� that was leaked usinga fluorescent calcium-binding probe, Fluo-4 AM. Following the disruption of the DV,mitochondrial dysfunction and DNA fragmentation were assayed to validate PCD-likefeatures as consequences of DV membrane perturbation.
To narrow the panel of primary hits, follow-on characterization studies were con-ducted to investigate their possible mechanisms of action and therapeutic potencies ina time- and dose-dependent manner across laboratory strains and field isolates. Thesecharacterization assays led to the identification of a potent, non-CQ- and non-artemisinin-cross-resistant, rapidly parasiticidal, and DV-permeabilizing hit from theMMV Pathogen Box repository with activity against P. falciparum, MMV085071.
RESULTSHigh-content screening of DV-permeabilizing molecules from MMV Pathogen
Box. After gating for focused and circular singlets, the trophozoite population wasdetermined in accordance with positive Hoechst staining and visual inspection ofbright-field images (Fig. 1a). Given that the DV is a calcium ion store (9), a compromisedDV membrane integrity leading to subsequent Ca2� efflux could be indicated by anincrease in the area of green Fluo-4 fluorescence upon high-affinity binding to Ca2�
(11). Quantification of the Fluo-4 fluorescence redistribution (Fig. 1b to d) was mea-sured by the application of an area feature mask built in to the analysis software. Toexclude nonspecific background staining or staining of the erythrocyte, only areas withintensities of greater than 250 pixels were incorporated for each single-cell analysis. Toqualify as a DV-disrupting hit from the primary screen, the Pathogen Box compoundhad to display a phenotype with at least a 70% efficiency of DV disruption relative tothat for CQ. Figure 1e shows the identity of the hits obtained from repeated rounds ofscreening. From the total of 400 Pathogen Box compounds, 10 of them produced atleast a 70% increase in the area of Fluo-4 fluorescence when they were used at 10 �M,giving an overall hit rate of 2.5%. Of note, MMV676380 and MMV032967 permeabilizedthe parasite DV to a greater extent than CQ (Fig. 1f). Since the mechanisms of action ofthe compounds in the library are unknown, the phenotypic screen could offer prelim-inary mechanistic insights into their potential antimalarial activities.
Downstream induction of PCD-like features. To validate PCD-like parasite celldeath following DV disruption, parasites were treated with the panel of hits for 10 h fordetection of downstream cellular responses. To this end, the ratiometric cationic dyeJC-1 was used to characterize the perturbation of mitochondrial integrity. After nor-malization to the results obtained with the drug-free control (treated with phosphate-buffered saline [PBS]), all the hits, including CQ, brought about the disintegration of themitochondrial transmembrane potential, while CQ displayed the greatest degree ofreduction in the JC-1 red-to-green ratio (Fig. 2a). Corroborative results were obtainedfrom the measurement of the proportion of parasites with a sub-G1 DNA content (Fig.2b). Of particular note, MMV687248 demonstrated a high degree of efficacy in trigger-
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 2
FIG 1 High-content screening assay of P. falciparum 3D7 for DV permeabilization. (a) Selection of focused and circular single cells based ongradient RMS, the aspect ratio, and the area of cells in the bright-field channel (M03). (b, c) Representative fluorochrome plots of Fluo-4 stainingand fluorescence intensity (Intensity_MC_Ch02) against Hoechst 33342 staining and fluorescence intensity (Intensity_MC_Ch01). Bright-fieldimages of PBS-treated drug-free control trophozoite population with intact DV (b) and the trophozoite population after treatment with 3 �M CQdisplaying a redistribution of Fluo-4 fluorescence due to DV being breached (c) were obtained. Ch01 and Ch02, fluorescence by DNA-bindingdye Hoechst 33342 and Fluo-4, respectively; Ch03, the bright-field channel. (d) Histogram overlay showing the increase in the area of Fluo-4fluorescence when parasites were treated with CQ (blue) and PBS (orange). M02 refers to the mask selection of channel 2 (Fluo-4), and240-4095 demarcates the range of Fluo-4 pixel intensity to be included for analysis by IDEAS software. (e) Consolidated results of screeningof the Pathogen Box compound repository at 10 �M. After normalization to the background (PBS), the mean area of Fluo-4 fluorescencewas compared to that obtained with CQ (positive control). The cutoff for hit selection was set at 70% relative to the fluorescence obtainedwith 3 �M CQ, and the novel hits were subjected to analysis and further characterization. (f) Identification of the 10 hits which displayedDV disruption. Mefloquine (MQ; 10 �M) was included as an internal negative control, as it does not breach the parasite DV (55). DMSO (1%)was included in the screening assays, as the final concentration of the screening compounds contained 1% DMSO diluted from 1 mM to10 �M. Error bars are SEMs obtained after performing at least 3 independent experiments. ***, P � 0.001 compared to the vehicle control(1% DMSO); **, P � 0.01 compared to the vehicle control (1% DMSO).
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 3
ing DNA fragmentation that was similar to that of CQ, while the remaining hits led toa minimum of a 3-fold increase in the proportion of sub-G1 DNA-positive parasites. Allin all, these assays also validated parasite cell death after compromise of the parasiticvacuole by the compounds and were in agreement with a novel mechanism of P.falciparum cell death following the disruption of DV integrity.
Spectrophotometric investigation of �-hematin biocrystallization. With the hits’PCD-like features being validated through repeated screens for MOMP and sub-G1 DNAassays, the hits were subjected to further characterization studies to increase ourunderstanding of the novel compounds. During the intraerythrocytic stage of themalaria parasite, hemoglobin is degraded for amino acids and osmotic homeostasis asthe parasite matures in size (12). As cytotoxic heme is generated alongside thedegradation process, the parasites detoxify free monomeric heme into hemozoinbiocrystals. While many theories proposing the antimalarial mechanism exerted by CQhave been developed, it is generally accepted that the drug acts by inhibiting theformation of hemozoin, thereby leading to heme accumulation and parasite cell death(13). Despite the effectiveness of CQ and its impact on the ambitious goal of theelimination of malaria, resistance to CQ had limited its administration. Against this
FIG 2 Assessment of PCD-like features in malaria parasites after treatment with the indicated compoundsat 10 �M. Following DV disruption, 3D7 parasites were treated with the hits for 10 h and assayed formitochondrial depolarization normalized to that for the drug-free control (PBS) (a), and the percentageof parasites with sub-G1 DNA contents was determined via flow cytometry (b). As shown from the lowJC-1 red/JC-1 green ratio and the increased percentage of parasites with sub-G1 DNA peaks, all the hitsinduced downstream cell death features after having disrupted the parasite’s DV. As previously reported(55), MQ was observed to induce a high percentage of sub-G1 DNA-positive parasites but not mitochon-drial dysfunction at 10 �M. ***, P � 0.001 compared to the drug-free PBS-treated control.
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 4
backdrop, a de novo �-hematin assay was optimized to investigate the mode of actionof the hits and to distinguish the activity of the novel non-CQ-like hits from that ofCQ-like compounds, indicated by the low levels of �-hematin present in the final pelletafter washing off of the monomeric heme and heme oligomeric aggregates with analkaline bicarbonate buffer (14). Pyrimethamine (PYR) was used as a negative mecha-nistic control to indicate the non-CQ-like effect, as it is an established antimalarial drugwhose potency is attributed to its interference with DNA synthesis (15). Accordingly,absorbance readings resulting from unimpeded �-hematin polymerization were ob-served to be higher for PYR and non-CQ-like hits than for CQ, which impeded�-hematin formation and thus resulted in lower absorbance readings. MMV676380,MMV085071, and MMV687812 emerged as hits which displayed high optical densitiesat 405 nm comparable to the optical density of PYR after normalization to the resultsfor the drug-free (PBS) controls (Fig. 3). This preliminary finding suggested that thethree compounds from Pathogen Box are non-CQ cross-resistant and thus displaypotent activity against CQ-resistant P. falciparum strains.
Growth inhibition assay and potency across CQ-resistant strains. As the Patho-gen Box is an assembly of diverse novel compounds with activity against the pathogensresponsible for neglected tropical diseases of interest, meaningful antimalarial activityin the low-micromolar range had to be established prior to any subsequent hit-to-leadoptimization. Laboratory strains 3D7, which is CQ sensitive, and K1, which is CQresistant, were treated with 11-point serial dilutions of MMV676380, MMV085071, andMMV687812 for half-maximal (50%) inhibitory concentration (IC50) determination, andall the sigmoidal IC50 curves obtained had a goodness of fit above 0.97 (Fig. 4; see alsoFig. S1 in the supplemental material). The respective resistance indices after 48 h ofincubation were then evaluated by determination of the ratio of the IC50 for K1 to theIC50 for 3D7 (Table 1). The antimalarial efficacies of the hit compounds were determinedto be in the nanomolar to low-micromolar range, with MMV085071 exhibiting thelowest IC50. In particular, MMV676380 and MMV085071 were phenotypically validatedto be non-cross-resistant, with similar dosages required for activity against 3D7 and K1,thereby substantiating the preliminary findings obtained and shown in Fig. 3, while aconcentration of CQ 22 times greater was required for activity against K1. The IC50 ofMMV687812, however, was slightly significantly different from the IC50s of the other
FIG 3 Relative �-hematin formation using hemin and NP-40 detergent with 20 �M antimalarial controlsand the hits obtained from the assay whose results are shown in Fig. 1f. CQ served as the control forinhibition of �-hematin formation, and PYR, an antifolate drug which does not target the parasite’s hemedetoxification pathway (HDP), served as the mechanistic control of a non-CQ-like feature. Three com-pounds, MMV676380, MMV085071, and MMV687812, were shown to be non-CQ-like. Error bars representthe SEMs from at least 3 separate experiments. **, P � 0.01 compared to CQ; *, P � 0.05 compared toCQ. OD405, optical density at 405 nm.
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 5
compounds, with a concentration of MMV687812 approximately 2- to 3-fold greaterthan that of the other compounds being needed to inhibit K1.
Dose-dependent effects of DV destabilization and PCD-like phenotype induc-tion. Having demonstrated the lack of CQ cross-resistance, parasites were then sub-jected to various concentrations of the hit compounds to quantitate the minimum doserequired for induction of hallmark features of PCD for evaluation of their potencies.MMV085071 was shown to achieve a notable DV disruption and MOMP and to increasethe amount of sub-G1 DNA at a dosage comparable to that of CQ on the basis ofsigmoidal dose-response curves (Fig. 5a to c). Calculation of the 70% effective concen-tration (EC70) revealed the minimum doses required for 70% maximal phenotypeinduction (Table 2). CQ proved to display the greatest potency in eliciting the DVdisruption and PCD-like phenotypes in 3D7 parasites. At least 1 �M CQ was required tobe accumulated in the DV of parasites for potent permeabilization in 4 h, causingmitochondrial dysfunction and DNA fragmentation in 10 h. MMV687812, in contrast,required the highest dosage to achieve all 3 investigated phenotypes. MMV676380appeared to be the most effective in compromising parasite DNA integrity, whileMMV687812 targeted the mitochondria with the most pronounced efficacy comparedto its DV-disrupting and DNA degradation capacities.
FIG 4 Phenotypic validation of non-CQ-cross-resistant hits’ antimalarial activities and potencies againstCQ-sensitive (3D7) and CQ-resistant (K1) strains with an in vitro sensitivity IC50 evaluation. Error barsrepresent SEMs from at least 3 separate experiments. ***, P � 0.001 between CQ-treated 3D7 and K1parasites; ns, the IC50s obtained for MMV676380 (P � 0.235) and MMV085071 (P � 0.096) were notsignificantly different between 3D7 and K1; *, P � 0.05 for slight cross-resistance (P � 0.016) forMMV687812.
TABLE 1 Antimalarial efficacy of hits and determination of resistance indexa
Drug
IC50 (nM)bResistanceindex Therapeutic index3D7 K1
CQ 34.0 (32.6–35.5) 760.4 (699.8–826.2) 22.4 � 1.36 1,000MMV676380 941.1 (839.5–1,055) 1,223 (1,082.1–1,365) 1.30 � 0.2 �50MMV085071 76.3 (71.7–81.1) 126.6 (117.6–136.3) 1.66 � 0.27 �125MMV687812 510.7 (449.7 to 580.0) 1,317 (1,216–1,428) 2.58 � 0.45 6.2aThe resistance index is expressed as the ratio of the IC50 for strain K1 (CQ resistant) to the IC50 for strain3D7 (CQ sensitive) (Fig. 4). The therapeutic indices of Pathogen Box compounds are expressed as the ratioof the CC50 for HepG2 cells to the IC50 for 3D7. The CC50 for HepG2 cells was obtained from the PathogenBox data sheet provided by MMV. The therapeutic index of CQ is expressed as the ratio of the publishedCC50 for HepG2 cells (56) to the IC50 for 3D7. The data are the means � SEMs from at least 3 independentexperiments.
bThe values in parentheses are 95% confidence intervals.
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 6
Assessment of parasite killing speed. Using two-color flow cytometry, the timingof action of the non-CQ-like hits for viability and reinvasion capacity was characterizedat the 24th-, 48th-, and 72nd-hour time points after treatment of the parasites with thedrugs of interest. 3D7 parasites were exposed to 10 times the predetermined IC50
(Table S1) to avoid suboptimal drug treatment during the period of investigation (16).This concentration was optimized by Sanz et al. (16), who demonstrated that concen-trations corresponding to 10 times the IC50 warranted the maximal in vitro killingactivity irrespective of the mode of action. As such, should an effect of delayed parasitedeath be observed, it would be associated with the timing of action and the killingkinetics of the drug and not suboptimal drug exposure. Optimization and validation ofthe assay were first achieved with 4 established antimalarials that exhibit distinctive
FIG 5 Assessment of dose dependence of PCD-like phenotypes. MMV676380, MMV085071, and MMV687812 were serially diluted from 10 �M, while CQ wasdiluted from 3 �M, and the area of Fluo-4 fluorescence (a), the JC-1 red/JC-1 green ratio (b), and the percentage of parasites with sub-G1 DNA contents (c) werenormalized against those for the PBS-treated controls and plotted accordingly. Error bars represent the SEMs from 3 separate experiments.
TABLE 2 Dosage required for 70% maximal DV disruption, MOMP, and fragmentation ofDNA
Drug
EC70a (�M)
Disruption of DV Mitochondrial depolarization DNA degradation
CQ 0.97 0.88 1.45MMV676380 4.06 3.32 2.75MMV085071 1.38 1.91 1.81MMV687812 8.70 3.85 5.08aEC70 was calculated for the evaluation of potency in inducing PCD-like phenotypes, as a 70% cutoff wasapplied for the selection of hits during the high-content primary screen (Fig. 1e).
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 7
killing rate profiles on the basis of the double carboxyfluorescein diacetate succinimidylester (CFDA-SE)- and Hoechst-positive population (Fig. S2). The double-positive popu-lation correlates with viable parasites that are able to invade fresh uninfected CFDA-SE-labeled erythrocytes. The killing rate profiles obtained were in agreement with thefindings of pharmacokinetic studies, as atovaquone (ATV) and PYR have been known toexert their inhibitory effects more slowly than CQ and artesunate (ART), which arerapidly acting antimalarials (17). ATV collapses the mitochondrial membrane potentialby binding to the parasite cytochrome b complex, and despite its potent interferencewith parasite proliferation, it produces a delayed death phenotype, where a consider-able decline in the level of parasitemia could be detected only beyond the 72nd hourof the P. falciparum life cycle (17–19). In order to investigate parasite killing kinetics,flow cytometric analysis was repeated up to the 72nd-hour time point for dynamiccoverage of the parasite life cycle. Of the three non-cross-resistant hits, MMV085071was demonstrated to be a fast-acting parasiticidal agent which lowered the level ofparasitemia to 30% of that at the beginning within the first 24 h, while MMV687812 andMMV676380 displayed notable inhibitory action only after the 24th- to 48th-hour timepoints (Fig. 6).
Dose-response assays on ART-sensitive and -resistant strains. Thus far,MMV676380, MMV085071, and MMV687812 have been demonstrated to be DV disrup-tive and effective against both CQ-resistant and -sensitive strains. In particular,MMV085071 was also shown to be a fast-acting compound that targeted parasitegrowth within the first 24 h of the life cycle. Due to artemisinin resistance from thedelayed parasite clearance associated with a single nucleotide mutation in the para-site’s Kelch13 protein (20), it is pertinent to evaluate the applicability and efficacy of thethree hit compounds against ART-sensitive and -resistant strains. IPC 5202 is anART-resistant strain with an R539T polymorphism in the Kelch13 propeller domain.Isolated in 2011 from a western Cambodia malaria patient, it showed delayed clearanceand 40 to 49% survival in in vitro artemisinin-based susceptibility tests (21). Cam3.I_revis a revertant of IPC 5202 that is susceptible to ART. A 48-h reinvasion assay wasperformed, and the 3 compounds demonstrated comparable potencies (IC50s) withnonsignificant differences in activity according to the P values, regardless of thedelayed parasite clearance status (Fig. 7), thereby suggesting the lack of cross-resistance to ART. On the contrary, resistant strain IPC 5202 required an 8-fold increasein the ART concentration for a half-maximal inhibitory effect compared to the concen-tration required for activity against sensitive strain Cam3.I_rev.
FIG 6 Timing of action of three non-CQ-like hits and four reference antimalarials represented by consolidatedprofiles of parasite killing kinetics over 72 h. Error bars represent the SEMs from 3 to 6 separate experiments.
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 8
Drug discovery campaigns often rely on 2 different approaches: target-basedscreening and phenotypic screening. While well-defined mechanisms of action may beelucidated and molecular hypotheses can be investigated through target-based screen-ing, target-based screening also contributes to high attrition rates and low levels ofeffectiveness in the subsequent drug development pipeline as specific molecularpathways and interactions increase in complexity in relation to disease pathogenesisand therapy (22). Phenotype-driven screenings, on the other hand, pose challenges inthe optimization of drug candidates due to the lack of limited mechanistic studies. Yetfirst-in-class drugs with novel molecular entities and scaffolds could be developed dueto the diverse screening strategy harnessed through a phenotype-centric approach. Inthe current study, we adopted a target (DV)-based phenotypic screen which leveragesthe strengths of both approaches. As such, in the first-pass screening, the MMVPathogen Box library was screened for compounds that, at a concentration of 10 �M,produced phenotypes of DV permeabilization, MOMP, and DNA degradation (Fig. 1 and2). Even though PCD in P. falciparum had been a contentious topic, growing evidencehad substantiated the existence of apoptotic features during both the intraerythrocyticstage (23–25) and the mosquito stage (26, 27) of development. Our work presents arobust combination of target-based phenotypic screening alongside downstream val-idation using MOMP and DNA degradation assays, in accordance with the hallmarkcharacteristics of mammalian apoptosis. Altogether, these findings corroborate theemerging indication that the mitochondrial and DNA degradation PCD phenotypes,whether they are incidental or genetically encoded, can be readily observed in P.falciparum through exposure to physiological stress, drug treatment, and externalstimuli.
The MMV Pathogen Box contains 400 novel drug-like compounds active against awide range of pathogens responsible for infectious tropical diseases and pathogenichelminths. With a hit rate of 2.5% (10/400), 6 of them were from the malaria diseasecluster set, while the remaining 4 were from the tuberculosis disease set (Table 3). Asthese 4 hits exhibited activity against both Mycobacterium tuberculosis and P. falcipa-rum, it is plausible that they exert overlapping mechanisms of action. Commonenzymatic complexes of the type II fatty acid biosynthetic pathway and/or the iso-prenoid biosynthetic pathway(s) in both pathogens suggest possible putative drugtargets. This notion is supported by emerging reports on the antimalarial and antimy-cobacterial activities of antibiotics inhibiting the FabI and FabH enzymes of the type IIfatty acid synthesis (FAS) pathway, which is common to both microorganisms (28–31).Since the type II FAS pathway is the only means of de novo fatty acid biosynthesis inmalaria parasites (32), a possible antimalarial mechanism of action of the 4 hits from the
FIG 7 Evaluation of antimalarial potency against ART-sensitive (Cam3.I_rev) and ART-resistant (IPC 5202)strains. ***, P � 0.001 between ART-treated Cam3.I_rev and IPC 5202 parasites; ns, the IC50s ofMMV676380 (P � 0.113), MMV085071 (P � 0.490), and MMV687812 (P � 0.117) were not significantlydifferent between the two strains.
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 9
TABLE 3 Chemical structures, compound identification, and disease set classification ofthe 10 hits acquired from whole-cell high-content screen of DV permeabilizationa
Disease set classification Compound Chemical structure Mol wt
Tuberculosis MMV661713 435.33
MMV687749 412.49
MMV687248 320.32
MMV687812 534.51
Malaria MMV006901 306.37
MMV676380 370.79
MMV021375 280.38
MMV032967 369.35
(Continued on next page)
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 10
tuberculosis disease set may involve the common FAS pathway. This preliminary screencould henceforth promote an understanding of the pathogenesis of these two dead-liest infectious diseases. From the supporting information for the MMV data set, all 10hits have 1-octanol/water partition coefficient (clogP) values ranging from 1.9 to 3.4and pKa values ranging from 7.0 to 11.4. A high-content screen recently found thatcompounds with clogP values above 2 and basic pKa values ranging from 6.5 to 11.0are readily able to diffuse into lysosomes (33). These physiochemical properties are inagreement with the lysosomal accumulation of chemotherapeutics, such as CQ andquinacrine, due to their weak basicity and lipophilicity. Various anticancer drugs andtricyclic antidepressants with weak basicity and lipophilicity have also been shown toexert their therapeutic effects through lysosomal sequestration (34). In light of thesefindings, it can be postulated that the panel of hits permeabilized the parasite’s DVmembrane, by this means resulting in the downstream release of calcium ions andapoptosis-like cellular phenotypes.
Information previously published in the literature associates P. falciparum CQ resis-tance with P. falciparum crt (pfcrt) gene mutations which mitigate CQ’s antimalarialinhibitory effect on hemozoin biocrystallization by mediating its efflux from the DV (35).Consequently, it will be clinically relevant to discriminate novel hits whose primarymode of action does not rely upon the parasite’s heme detoxification pathway (HDP).As �-hematin is chemically and structurally equivalent to hemozoin, an optimized denovo �-hematin assay was adopted (36) and distinguished MMV676380, MMV085071,and MMV687812 as non-CQ-like hits from the array of 10 hits. These three hits inhibited�-hematin formation to a significantly lower extent than CQ (Fig. 3), and the in vitroresistance indices obtained via growth inhibition IC50 assays (Table 1) supported thisobservation. While a significantly greater dosage of MMV687812 was required to inhibitstrain K1 than strain 3D7 (Fig. 4), MMV687812 remained a promising hit according to itsresistance index relative to the resistance index exhibited by CQ. Furthermore, in vitroassays and sequencing analyses have identified a minimum threshold of a 10-folddifference in IC50s between sensitive and multidrug-resistant P. falciparum strains to bea robust cutoff against a false-positive resistance indication (37). With a resistance indexof 2.58, MMV687812, together with MMV676380 and MMV085071, thus continued tobe categorized as non-CQ-like compounds. Overall, IC50s in the low-micromolar andnanomolar range for the 3D7, K1, IPC 5202, and Cam3.I_rev strains affirmed theantimalarial abilities of the compounds, even though the efficacies were lower thanthose of the established antimalarials CQ and ART (Fig. 7), indicating the lack of in vitrocross-resistance.
Rapidly acting drugs are likely to eliminate circulating parasites faster and avoidrecrudescence and resistance development from suboptimal exposure to the drugs. In
TABLE 3 (Continued)
Disease set classification Compound Chemical structure Mol wt
MMV085071 348.41
MMV022478 432.91
aFour of the compounds were from the tuberculosis disease cluster, as they were first assayed againstMycobacterium tuberculosis during the curation of the MMV Pathogen Box, while 6 of them were assayedfor activity against P. falciparum.
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 11
vitro killing rates revealed, unsurprisingly, that there was no association betweenantimalarial potency, as indicated by the IC50, and the killing rate, given that thecompound-specific lag phase is independent of the potency of the dose (38).Metabolism-based assays often have limitations, in that some drugs, such as ciprofloxa-cin, clindamycin, and azithromycin, have been reported to display a delayed deatheffect (39), and metabolic proxies as markers of parasite viability, such as parasitelactate dehydrogenase (pLDH) or histidine-rich protein 2 (HRP-2) levels, therefore donot always truly reflect the killing rates. For instance, bioconversion studies on a cholineanalog, T3/SAR97276, developed by Sanofi-Aventis showed rapid metabolism of theprodrug within 2 h (40), but data from metabolism-based assays of pLDH and HRP-2failed to correlate with the analog’s efficacy at subnanomolar IC50s and the rapid onsetof parasite clearance observed in vivo (41). Parasite viability following intraerythrocyticreinvasion, coupled with time point-specific drug removal, is thus a more suitablereadout. By the 72nd hour of investigation, MMV085071 was able to produce low levelsof parasitemia similar to those produced by CQ and ART, thereby suggesting that eventhough the lag phase (the time taken for a compound to exert its outcome of maximalparasite killing) was longer, MMV085071 could achieve an overall efficacious abroga-tion of P. falciparum growth (Fig. 6). Conversely, MMV676380, MMV687812, PYR, andATV were slower acting, with a marked reduction in the level of parasitemia beingachieved only after the first parasite cycle. In light of this, parasite recovery from thelack of complete clearance could be expected should the duration of investigation beextended over 2 developmental cycles when the drug concentration falls below desiredlevels.
Nonetheless, moderately or slower-acting drugs, such as azithromycin, showed invitro and in vivo synergistic efficacy with ART or quinine (42), while CQ treatmentsupplemented with doxycycline demonstrated a significant improvement in parasiteclearance in patients (43). Since artemisinin resistance status is attributed to delayedclearance, therapeutically potent yet slow-acting drugs could be administered incombination to prevent recrudescence once the rapid parasiticidal action of artemisininhas worn off. Assays with tightly synchronized ART-resistant parasites at the ring stage(44) could be performed with artemisinin-based compounds to better assess thepotentiating effects of MMV676380 and MMV687812 against artemisinin resistance andreinvasion capacity, as implied by an enhanced susceptibility when pulsed in dose-specific combinations with ART or dihydroartemisinin.
According to compound report cards deposited in chEMBL (the Neglected TropicalDisease Archive), non-cross-resistant hits MMV085071, MMV676380, and MMV687812do not contain structural alerts or any violations of Lipinski’s rule of 5, both of which areimportant considerations to assess the drug likeness of uncharacterized compounds.The 50% cytotoxic concentration (CC50) of MMV687812 for HepG2 cells, however, wasreported to be 3.9 �M, while the IC50s against P. falciparum 3D7 and K1 were in thehigh-nanomolar to low-micromolar range (Table 1; see also Fig. S1 in the supplementalmaterial). With an overall therapeutic index of 6.2, this may indicate inherent cytotox-icity and weak selectivity. MMV676380 offers an improved selectivity with a therapeuticindex of over 50. Notably, MMV085071 proved to be a top lead compound from theseries of studies undertaken, as the work described here revealed its rapidly actingparasiticidal kinetics, a potency of PCD-like induction comparable to that of CQ, and itslow-nanomolar efficacy against 3D7 (IC50, 76 nM) and K1 (IC50, 127 nM). With atherapeutic index of over 125, it is reasonable to speculate that MMV085071 has asatisfactory safety profile. Surprisingly, a higher inhibitory dosage was observed for fieldisolates (Fig. 7) than for laboratory strains (Fig. 4).
MMV085071, which is categorized in the malaria disease cluster set to have potentactivity against 3D7 and W2 parasites (CQ resistant), was also previously reported byNovartis (PubChem GNF-Pf-3843). Structurally, MMV085071, also known as 2-(5-methoxy-3-pyridinyl)-6-[4-(4-pyridinyl)-1-piperazinyl]pyrazine, contains pyridine, piper-azine, and pyrazine drug-like chemical moieties, which are also basic heterocyclicamines. Pyrazine rings have long been involved in a range of antimycobacterial (in the
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 12
form of pyrazinamide), anticancer, and insecticidal effects (45). Novel analogues withpyrazine substitutes were proposed to inhibit protein kinases while modulating cancercells’ sensitivity toward apoptosis and anticancer chemotherapy (46). Piperazine ringsare popular constituents of pharmaceutical agents, and specifically, pyridinylpipera-zine and pyrimidinylpiperazine pharmacophores, similar to the pharmacophore ofMMV085071, are commonly recommended as medications for neurological disorders,such as depression (mirtazapine) and Parkinson’s disease (piribedil). These chemicalscaffolds also demonstrated anticancer and antiretroviral characteristics against Bcr-Abltyrosine kinase and HIV, respectively (47, 48).
Tracking along the database details in PubChem and chEMBL, MMV085071/GNF-Pf-3843 was identified from a high-throughput screen (1.7 million compounds) of theNovartis compound library in 2008. Substitution of the pyridine ring with an alkyl aminegroup in MMV085071/GNF-Pf-3843 gave rise to another compound, GNF-Pf-1610,which showed a significant reduction of potency against 3D7 and CQ-resistant strainW2 compared with that of MMV085071. A more than 7-fold increase in the dosage ofGNF-Pf-1610 compared with that of MMV085071 was required for activity against both3D7 and W2 parasites (49). Accordingly, this revealed preliminary insights into thestructure-activity relationship (SAR) of MMV085071/GNF-Pf-3843, suggesting that thepyridine moiety is essential for its parasiticidal activity. The identification and charac-terization of MMV085071 thus indicate that it is a potential novel candidate fordownstream hit-to-lead optimization.
In summary, the MMV Pathogen Box was subjected to a robust, high-contentphenotypic screen for compounds that permeabilize DV, followed by flow cytometricanalysis of the compounds for MOMP and DNA degradation. From that screen, 10 hitswere subjected to further characterization studies. To narrow the panel of hits inducinga PCD-like phenotype, dose-response experiments on parasites with PCD phenotypesand CQ and ART resistance were performed, and MMV676380, MMV085071, andMMV687812 were found to be active against the CQ-resistant K1 strain and theART-resistant IPC 5202 strain. MMV085071, however, exhibited the lowest IC50 and hadefficacy comparable to that of CQ in the induction of PCD-like phenotypes in P.falciparum. While its precise mode of action is not yet elucidated, low-micromolarconcentrations of MMV085071 were able to trigger DV disruption and induce down-stream PCD-like features and parasite cell death within a short incubation time of 4 h,suggesting that its mode of action is via a PCD-like mechanism. Additional studies onkilling kinetics uncovered that it is a fast-acting compound, further reinforcing hepossibility that it is a novel drug candidate with activity against the human malariaparasite. Quantitative SAR analysis and hit-to-lead optimization of MMV085071 throughanalogues can be conducted, thus providing a strong basis for in-depth mechanisticstudies and drug development.
MATERIALS AND METHODSIn vitro culturing and synchronization of P. falciparum parasites. Laboratory CQ-sensitive strain
3D7 (MRA-102; Malaria Research and Reference Reagent Resource Center [MR4]; ATCC, Manassas, VA,USA), CQ-resistant strain K1 (MRA-159; MR4; ATCC), and field isolates artemisinin-sensitive strainCam3.I_rev (MRA-1240; MR4; ATCC [contributed by David A. Fidock]) and artemisinin-resistant strain IPC5202 (MRA-1252; MR4; ATCC [contributed by Didier Ménard]) were cultured and maintained in 25-cm2 or75-cm2 nonvented cell culture flasks at 1.25% hematocrit with complete malaria culture medium (MCM),comprising a filter-sterilized homogeneous mixture of RPMI 1640 (Life Technologies, USA), 2.2 g/litersodium bicarbonate, 0.5% (wt/vol) AlbuMAX II (Gibco, Thermo Fisher Scientific, MA, USA), 0.005% (wt/vol)hypoxanthine (Sigma-Aldrich, MO, USA), 0.03% (wt/vol) L-glutamine (Sigma-Aldrich, MO, USA), and 25�g/liter gentamicin (Gibco, Thermo Fisher Scientific, MA, USA). The pH was adjusted to 7.4 using 1 NNaOH. Type O-positive human erythrocytes in citrate-phosphate-dextrose with adenine (CPDA-1) anti-coagulant were obtained from the Interstate Blood Bank (Memphis, TN, USA) and tested negative forinfectious agents according to FDA guidelines. The flasks were incubated at 37°C without light andgassed with 5% CO2, 3% O2, and 92% N2. Routine PCR-based mycoplasma testing was conducted on amonthly basis to ensure the lack of contamination in cultures. Parasites were synchronized with 5%(wt/vol) sorbitol (Merck, Darmstadt, Germany) at 37°C for 10 min for the selection of ring-stage parasites,following which the cells were washed twice and resuspended to 1.25% hematocrit in MCM.
Drug preparation and MMV Pathogen Box library storage. CQ, pyrimethamine (PYR), atovaquone(ATV), artesunate (ART), and mefloquine (MQ) were all purchased from Sigma-Aldrich. CQ was dissolved
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 13
in sterile 1� phosphate-buffered saline (PBS), while the remaining drugs were dissolved in steriledimethyl sulfoxide (DMSO) to stock aliquots of 10 mM, frozen at �20°C, and protected from light. Freshworking drug concentrations were prepared after dilution with sterile PBS or cold MCM to ensure thatthe final assay mixture contained less than 0.5% DMSO. The MMV Pathogen Box was received from boththe Singapore University of Technology and Design, Singapore, and the Medicines for Malaria Venture(MMV; Switzerland) in Costar 96-well round-bottom microtiter plates. Upon receipt, each plate was sealedwith 96-well X-Pierce adhesive film (Sigma-Aldrich, MO, USA), and all wells, each of which contained10 �l of 10 mM drug, were reconstituted with DMSO so that the final stock concentration was 1 mM in100 �l, according to MMV’s instructions. Similarly, the plates were stored in �20°C and kept protectedfrom light.
Parasite drug treatment and staining. 3D7 parasites were synchronized for 2 cycles prior to thestart of the assay to achieve a high level of parasitemia of over 10% with mid-trophozoite-stage parasites(28 to 30 h postinvasion), which were resuspended to 2.5% hematocrit in a 96-well flat-bottom plate.After drug treatment and incubation at 37°C with a mixture of gases in a humidified hypoxic chamberfor 4 h (for the DV disruption experiments), cells were transferred to a 96-well V-bottom plate toconcentrate the pellet and stained with 1 �M Fluo-4 AM (Thermo Fisher Scientific, MA, USA) and 1 �g/mlHoechst 33342 (Sigma-Aldrich, MO, USA). Following incubation in the dark for 30 min, the cells werewashed thrice with prewarmed MCM, resuspended in 100 �l of HyClone PBS (GE Healthcare, Chicago, IL,USA), and sealed with X-Pierce adhesive film to prevent evaporation from the wells during dataacquisition. For assays validating a PCD-like response with the parasites’ mitochondria and assays of DNAintegrity, the trophozoites were incubated for 10 h under the same atmospheric conditions describedabove, stained with 6 �M JC-1 (Thermo Fisher Scientific, MA, USA) and 0.8 �g/ml Hoechst 33342, andsubjected to the same washing steps described above.
DV permeabilization screen with a flow cytometer. The ImageStream X MkII flow cytometer(Amnis, Darmstadt, Germany) is designed to have imaging properties for high-content single-cellanalysis. In addition to the analysis of cell populations based on fluorescence, bright-field and two-dimensional fluorescent images of each event could be captured and visualized. Leveraging on its cellimaging capabilities, a high-content primary screen was developed and optimized (50) to detect andquantitatively assay for the increase in the area of Fluo-4 fluorescence as a consequence of drug-mediated DV disruption and Ca2� redistribution from the DV to the parasite cytosol. Using a 60�objective and extended depth of field (EDF) element, each 96-well plate was assayed with the autosam-pler plate reader for high-throughput data acquisition. Hoechst 33342 is a lipophilic DNA-bindingfluorescent stain that is excited by a 375-nm UV laser at 10 mW, while Fluo-4 AM is a cell-permeantcalcium-sensing dye which is excited by an argon ion 488-nm laser at 200 mW. Stringent gating wasapplied during data acquisition to select for focused, round Hoechst 33342-positive singlet parasites onthe basis of the area and aspect ratio of the cells in the bright-field channel and the fluorescence intensityof Hoechst 33342. Having acquired 2,000 parasite events, data analysis was performed with IDEASsoftware (version 6.1). The gating strategy is elucidated in Fig. 1. All of the acquired events were gatedfor focused cells on a gradient root mean square (RMS) for image sharpness histogram, followed bydevelopment of a scatter plot of the aspect ratio against the area for the bright-field images (Fig. 1a). Thisensures that the parasite images and events analyzed are sharp and focused singlets. Gating fortrophozoites was based on fluorescence intensity and visual inspection of the captured images. Singletrophozoites which matured to the schizont stage during the 4-h duration of data acquisition displayedmultiple punctate forms of Hoechst-stained DNA-containing merozoites, on the basis of the Hoechststaining intensity and as verified with visual images. Since schizonts mature with a larger DV during theparasite life cycle, they were not included in the analysis. Once applied, the gating for trophozoites wasmaintained in all subsequent analyses. Untreated drug-free PBS-treated controls were included in allexperimental runs, and healthy PBS-treated parasites displayed an intact DV with little Fluo-4 redistri-bution, as Ca2� remained sequestered in the DV (Fig. 1b). To quantify single-cell DV disruption, an areafeature mask incorporated in IDEAS software was applied to measure the increase in the Fluo-4fluorescence area upon binding to the effluxed intravacuolar Ca2� (Fig. 1c). Raw feature values werenormalized, and the degree of DV disruption achieved with the test compounds was compared to thatachieved with CQ. Single-stain (Hoechst 33342 and Fluo-4 AM) controls were prepared fresh for every runof the experiment, and a compensation matrix was generated using the compensation algorithm of thesoftware.
Assays for validation of PCD-like phenotype by MOMP and loss of DNA content. The ratiometricJC-1 dye is dependent on the mitochondrial membrane potential and was used for the assessment ofmitochondrial depolarization by flow cytometry. The shift of the fluorescence emission from JC-1-stainedgreen monomers (529 nm) to JC-1-stained red aggregates (590 nm) is indicative of a high membranepotential in healthy functional mitochondria, while mitochondrial degradation is denoted by a reductionin the percentage of JC-1-stained red aggregates to JC-1-stained green monomers. Using an Attune NxTflow cytometer (Invitrogen, CA, USA), a 405-nm violet excitation laser fitted with a 417LP 440/50BP filterwas used to excite and detect Hoechst 33342, while a blue light excitation laser (Alexa Fluor 488 nm)fitted with a 503LP 530/30BP filter and a yellow light excitation laser (561 nm) fitted with a 577LP585/16BP filter were used to excite and detect JC-1. At least 100,000 events were acquired from eachsample. Gating for sub-G1 DNA peaks, as an indication of DNA degradation, relative to that for theuntreated Hoechst-positive controls was determined.
Spectrophotometric determination for inhibition of �-hematin formation. To mature andproliferate, P. falciparum biocrystallizes toxic heme by-products into hemozoin crystals through theparasite’s heme detoxification pathway (HDP) during hemoglobin degradation. CQ had been widely
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 14
proposed to inhibit parasite growth by upsetting the biocrystallization process. �-Hematin was previ-ously established to be the chemical equivalent of hemozoin (51). As this study was interested in thediscovery of novel potent antimalarials, the hits were subjected to a preliminary assessment of possibleCQ cross-resistance on the basis of �-hematin inhibition. To characterize the 10 hits, CQ was included asa control for a CQ-like mechanism of inhibition of the HDP. PYR served as a non-CQ-like mechanisticcontrol because its mechanism of action primarily disrupts folate and DNA synthesis (52). In accordancewith the rationale, a spectrophotometric assay was performed to select for hits which did not inhibit�-hematin formation. Nonidet P-40 (Roche, Basel, Switzerland) was used as a biological mimic of parasitelipid factors to catalyze the formation of �-hematin from hemin (53). Drug-free controls were treated withPBS, while sample tubes were treated with the compounds of interest (CQ, PYR, and the 10 hits) at a finalconcentration of 20 �M. The final reaction mixtures were maintained at 400 �l with bovine hemin(Sigma-Aldrich, MO, USA) at 1 mM, 30.5 �M Nonidet P-40 in acetate buffer at pH 5.0, akin to the acidicpH of the parasite DV (54). The reagents used for the sodium acetate buffer (100 mM, pH 5.0), Tris-HClbuffer (100 mM, pH 8.0) with 2.5% (wt/vol) SDS, and sodium bicarbonate buffer (100 mM, pH 9.2)preparations were purchased from Sigma-Aldrich. After a 14-h incubation at 37°C and 195 rpm, the tubeswere centrifuged at 32,000 � g for 10 min at 4°C and resuspended in prewarmed Tris-HCl buffercontaining 2.5% (wt/vol) SDS for 30 min at 37°C to remove unbound proteins, followed by prewarmedalkaline sodium bicarbonate buffer to solubilize heme aggregates and precipitate the �-hematin (36).The final �-hematin pellet was washed with distilled water and resuspended in 100 �l 1 N NaOH and 900�l 2.5% (wt/vol) SDS. The samples were then read with a Tecan Infinite 200 Pro microplate reader at 405nm for detection of �-hematin. Relative absorbance readings were normalized to the backgroundreading (that for the PBS-treated drug-free controls). CQ and hits which inhibited �-hematin formationwould have lower absorbance readings by this spectrophotometric assay, while PYR and hits which didnot inhibit �-hematin formation would exhibit higher absorbance values.
A 48-h reinvasion assay for IC50 determination. Ring-stage parasites synchronized for 1 cycle priorto the start of the 48-h reinvasion assay for half-maximal (50%) inhibitory concentration (IC50) determi-nation were incubated for 48 h at 1.25% hematocrit and 0.7% parasitemia (for strain 3D7) or 1%parasitemia (for strains K1, Cam3.I_rev, and IPC 5202) with 11-point serial dilutions of the referenceantimalarial drugs and the compounds of interest. Parasite labeling was carried out with 1 �g/ml Hoechst33342 for 30 min at 37°C, and at least 100,000 events were collected by the Attune NxT flow cytometer,as elaborated above. IC50s were then determined by GraphPad Prism (version 5) software using avariable-slope logistic curve. Values were obtained from at least 3 separate triplicates.
Kinetics of P. falciparum parasite killing by compound hits. To determine if the compounds ofinterest were rapid or slower acting, parasites were treated with the drugs of interest at concentrationsthat were 10 times the IC50s predetermined prior to this assay (see Table S1 in the supplementalmaterial). This concentration ensured that any observed delayed killing effect was not due to suboptimalexposure of the parasites to the drug treatment but was attributed to the slower killing kinetics of thedrug against P. falciparum. From the point of drug treatment (time zero), parasites were harvested at 24h, 48 h, and 72 h. During parasite harvesting, the drugs were washed off with MCM at least 3 times. Inthis assay, parasites were treated with reference antimalarial drugs (CQ and ART for rapid-acting drugs,PYR and ATV for slower-acting drugs) and the Pathogen Box hits (MMV676380, MMV085071, andMMV687812) at 5% parasitemia and 0.5% hematocrit under conditions similar to those used for the IC50
assay. After harvesting and drug removal, 10 �M Vybrant CFDA-SE (Thermo Fisher Scientific, MA,USA)-prelabeled erythrocytes at 2.5% hematocrit were added to the parasites and further incubatedunder the same conditions described above for an additional parasite cycle (48 h) to allow invasion intothe labeled erythrocytes. The level of parasitemia in each well was determined with Hoechst 33342labeling and acquired using the Attune NxT flow cytometer. Parasite viability and the reinvasion capacityat all time points were indicated by a double-positive population (positive for both CFDA-SE and Hoechst33342 staining) and were normalized to those for the drug-free PBS-treated controls at time zero. At least3 separate experiments were performed at all time points (0 h, 24 h, 48 h, and 72 h).
Statistical tests and analyses. Graphs were prepared and statistical tests were performed withGraphPad Prism (version 5.0) software, which was also used to obtain P values by two-sample Student’st tests. Error bars represent the standard errors of the means (SEMs) from at least 3 separate independentexperiments.
SUPPLEMENTAL MATERIAL
Supplemental material for this article may be found at https://doi.org/10.1128/AAC.02031-17.
SUPPLEMENTAL FILE 1, PDF file, 0.1 MB.
ACKNOWLEDGMENTSWe are indebted to the Medicines for Malaria Venture (MMV; Switzerland). R.C.
provided the MMV Pathogen Box chemical library. Parasite strains were provided by theMalaria Research and Reference Reagent Resource Center (MR4).
K.S.-W.T. and J.X.T. thank the Yong Loo Lin School of Medicine of the NationalUniversity of Singapore and the National Medical Research Council, Singapore (NMRC/1310/2011 and NMRC/EDG/1038/2011), for support and grants. R.C. acknowledges
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 15
funding support from the Agency for Science, Technology and Research (A*STAR)through the Indo-Singapore Joint Science and Technology Research Cooperation grant(RGAST1503).
REFERENCES1. World Health Organization. 2016. World malaria report 2016. Global
Malaria Programme, World Health Organization, Geneva, Switzerland.2. Phyo AP, Nkhoma S, Stepniewska K, Ashley EA, Nair S, McGready R, ler
Moo C, Al-Saai S, Dondorp AM, Lwin KM, Singhasivanon P, Day NPJ,White NJ, Anderson TJC, Nosten F. 2012. Emergence of artemisinin-resistant malaria on the western border of Thailand: a longitudinal study.Lancet 379:1960 –1966. https://doi.org/10.1016/S0140-6736(12)60484-X.
3. Hien TT, Thuy-Nhien NT, Phu NH, Boni MF, Thanh NV, Nha-Ca NT, Thai LH,Thai CQ, Van Toi P, Thuan PD, Long LT, Dong LT, Merson L, Dolecek C,Stepniewska K, Ringwald P, White NJ, Farrar J, Wolbers M. 2012. In vivosusceptibility of Plasmodium falciparum to artesunate in Binh Phuoc Prov-ince, Vietnam. Malaria J 11:355. https://doi.org/10.1186/1475-2875-11-355.
4. Dondorp AM, Nosten F, Yi P, Das D, Phyo AP, Tarning J, Lwin KM, ArieyF, Hanpithakpong W, Lee SJ, Ringwald P, Silamut K, Imwong M,Chotivanich K, Lim P, Herdman T, An SS, Yeung S, Singhasivanon P, DayNPJ, Lindegardh N, Socheat D, White NJ. 2009. Artemisinin resistance inPlasmodium falciparum malaria. N Engl J Med 361:455– 467. https://doi.org/10.1056/NEJMoa0808859.
5. Payne D. 1987. Spread of chloroquine resistance in Plasmodium falciparum.Parasitol Today 3:241–246. https://doi.org/10.1016/0169-4758(87)90147-5.
6. Slater AF. 1993. Chloroquine: mechanism of drug action and resistancein Plasmodium falciparum. Pharmacol Ther 57:203–235. https://doi.org/10.1016/0163-7258(93)90056-J.
7. Ch’Ng J, Liew K, Goh AS, Sidhartha E, Tan KS. 2011. Drug-inducedpermeabilization of parasite’s digestive vacuole is a key trigger of pro-grammed cell death in Plasmodium falciparum. Cell Death Dis 2:e216.https://doi.org/10.1038/cddis.2011.97.
8. Olliaro P, Goldberg D. 1995. The Plasmodium digestive vacuole: meta-bolic headquarters and choice drug target. Parasitol Today 11:294 –297.https://doi.org/10.1016/0169-4758(95)80042-5.
9. Biagini GA, Bray PG, Spiller DG, White MRH, Ward SA. 2003. The digestivefood vacuole of the malaria parasite is a dynamic intracellular Ca2� store.J Biol Chem 278:27910 –27915. https://doi.org/10.1074/jbc.M304193200.
10. Rohrbach P, Friedrich O, Hentschel J, Plattner H, Fink RH, Lanzer M. 2005.Quantitative calcium measurements in subcellular compartments ofPlasmodium falciparum-infected erythrocytes. J Biol Chem 280:27960 –27969. https://doi.org/10.1074/jbc.M500777200.
11. Gee KR, Brown KA, Chen WNU, Bishop-Stewart J, Gray D, Johnson I. 2000.Chemical and physiological characterization of Fluo-4 Ca2�-indicatordyes. Cell Calcium 27:97–106. https://doi.org/10.1054/ceca.1999.0095.
12. Esposito A, Tiffert T, Mauritz JMA, Schlachter S, Bannister LH, KaminskiCF, Lew VL. 2008. FRET imaging of hemoglobin concentration in Plas-modium falciparum-infected red cells. PLoS One 3:e3780. https://doi.org/10.1371/journal.pone.0003780.
13. Combrinck JM, Mabotha TE, Ncokazi KK, Ambele MA, Taylor D, Smith PJ,Hoppe HC, Egan TJ. 2013. Insights into the role of heme in the mecha-nism of action of antimalarials. ACS Chem Biol 8:133–137. https://doi.org/10.1021/cb300454t.
14. Gorka AP, Alumasa JN, Sherlach KS, Jacobs LM, Nickley KB, Brower JP, deDios AC, Roepe PD. 2013. Cytostatic versus cytocidal activities of chloro-quine analogues and inhibition of hemozoin crystal growth. AntimicrobAgents Chemother 57:356–364. https://doi.org/10.1128/AAC.01709-12.
15. Schlitzer M. 2007. Malaria chemotherapeutics part I: history of anti-malarial drug development, currently used therapeutics, and drugs inclinical development. ChemMedChem 2:944 –986. https://doi.org/10.1002/cmdc.200600240.
16. Sanz LM, Crespo B, De-Cózar C, Ding XC, Llergo JL, Burrows JN, García-Bustos JF, Gamo F-J. 2012. P. falciparum in vitro killing rates allow todiscriminate between different antimalarial mode-of-action. PLoS One7:e30949. https://doi.org/10.1371/journal.pone.0030949.
17. Le Manach C, Scheurer C, Sax S, Schleiferböck S, Cabrera DG, Younis Y,Paquet T, Street L, Smith P, Ding XC, Waterson D, Witty MJ, Leroy D,Chibale K, Wittlin S. 2013. Fast in vitro methods to determine the speedof action and the stage-specificity of anti-malarials in Plasmodium fal-ciparum. Malar J 12:424. https://doi.org/10.1186/1475-2875-12-424.
18. Painter HJ, Morrisey JM, Vaidya AB. 2010. Mitochondrial electron trans-
port inhibition and viability of intraerythrocytic Plasmodium falciparum.Antimicrob Agents Chemother 54:5281–5287. https://doi.org/10.1128/AAC.00937-10.
19. Wilson DW, Langer C, Goodman CD, McFadden GI, Beeson JG. 2013.Defining the timing of action of antimalarial drugs against Plasmodiumfalciparum. Antimicrob Agents Chemother 57:1455–1467. https://doi.org/10.1128/AAC.01881-12.
22. Swinney DC, Anthony J. 2011. How were new medicines discovered? NatRev Drug Discov 10:507–519. https://doi.org/10.1038/nrd3480.
23. Ch’ng JH, Kotturi SR, Chong AGL, Lear MJ, Tan KSW. 2010. A pro-grammed cell death pathway in the malaria parasite Plasmodium fal-ciparum has general features of mammalian apoptosis but is mediatedby clan CA cysteine proteases. Cell Death Dis 1:e26. https://doi.org/10.1038/cddis.2010.2.
25. Rathore S, Datta G, Kaur I, Malhotra P, Mohmmed A. 2015. Disruption ofcellular homeostasis induces organelle stress and triggers apoptosis likecell-death pathways in malaria parasite. Cell Death Dis 6:e1803. https://doi.org/10.1038/cddis.2015.142.
26. Arambage SC, Grant KM, Pardo I, Ranford-Cartwright L, Hurd H. 2009.Malaria ookinetes exhibit multiple markers for apoptosis-like pro-grammed cell death in vitro. Parasit Vectors 2:32. https://doi.org/10.1186/1756-3305-2-32.
27. Engelbrecht D, Durand PM, Coetzer TL. 2012. On programmed cell deathin Plasmodium falciparum: status quo. J Trop Med 2012:646534. https://doi.org/10.1155/2012/646534.
28. Kuo MR, Morbidoni HR, Alland D, Sneddon SF, Gourlie BB, Staveski MM,Leonard M, Gregory JS, Janjigian AD, Yee C, Musser JM, Kreiswirth B,Iwamoto H, Perozzo R, Jacobs WR, Sacchettini JC, Fidock DA. 2003.Targeting tuberculosis and malaria through inhibition of enoylreductase: compound activity and structural data. J Biol Chem 278:20851–20859. https://doi.org/10.1074/jbc.M211968200.
29. Lu H, Tonge PJ. 2008. Inhibitors of FabI, an enzyme drug target in thebacterial fatty acid biosynthesis pathway. Acc Chem Res 41:11–20.https://doi.org/10.1021/ar700156e.
30. Waller RF, Ralph SA, Reed MB, Su V, Douglas JD, Minnikin DE, CowmanAF, Besra GS, McFadden GI. 2003. A type II pathway for fatty acidbiosynthesis presents drug targets in Plasmodium falciparum. Antimi-crob Agents Chemother 47:297–301. https://doi.org/10.1128/AAC.47.1.297-301.2003.
31. Surolia N, Surolia A. 2001. Triclosan offers protection against bloodstages of malaria by inhibiting enoyl-ACP reductase of Plasmodiumfalciparum. Nat Med 7:167–173. https://doi.org/10.1038/84612.
32. van Schaijk BCL, Kumar TRS, Vos MW, Richman A, van Gemert G-J, Li T,Eappen AG, Williamson KC, Morahan BJ, Fishbaugher M, Kennedy M,Camargo N, Khan SM, Janse CJ, Sim KL, Hoffman SL, Kappe SHI, Sauer-wein RW, Fidock DA, Vaughan AM. 2014. Type II fatty acid biosynthesisis essential for Plasmodium falciparum sporozoite development in themidgut of Anopheles mosquitoes. Eukaryot Cell 13:550 –559. https://doi.org/10.1128/EC.00264-13.
33. Nadanaciva S, Lu S, Gebhard DF, Jessen BA, Pennie WD, Will Y. 2011. Ahigh content screening assay for identifying lysosomotropic com-pounds. Toxicol In Vitro 25:715–723. https://doi.org/10.1016/j.tiv.2010.12.010.
Tong et al. Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 16
34. Zhitomirsky B, Assaraf YG. 2015. Lysosomal sequestration of hydropho-bic weak base chemotherapeutics triggers lysosomal biogenesis andlysosome-dependent cancer multidrug resistance. Oncotarget6:1143–1156. https://doi.org/10.18632/oncotarget.2732.
35. Fidock DA, Nomura T, Talley AK, Cooper RA, Dzekunov SM, Ferdig MT,Ursos LM, Naudé B, Deitsch KW, Su X-Z. 2000. Mutations in the P.falciparum digestive vacuole transmembrane protein PfCRT and evi-dence for their role in chloroquine resistance. Mol Cell 6:861– 871.https://doi.org/10.1016/S1097-2765(05)00077-8.
36. Tripathi AK, Khan SI, Walker LA, Tekwani BL. 2004. Spectrophotometricdetermination of de novo hemozoin/�-hematin formation in an in vitroassay. Anal Biochem 325:85–91. https://doi.org/10.1016/j.ab.2003.10.016.
37. Chugh M, Scheurer C, Sax S, Bilsland E, van Schalkwyk DA, Wicht KJ,Hofmann N, Sharma A, Bashyam S, Singh S, Oliver SG, Egan TJ, MalhotraP, Sutherland CJ, Beck H-P, Wittlin S, Spangenberg T, Ding XC. 2015.Identification and deconvolution of cross-resistance signals from anti-malarial compounds using multidrug-resistant Plasmodium falciparumstrains. Antimicrob Agents Chemother 59:1110 –1118. https://doi.org/10.1128/AAC.03265-14.
38. Li Q, Hickman MR. 2015. Pharmacokinetics and pharmacodynamics ofantimalarial drugs used in combination therapy. Bentham Science Pub-lishers, Sharjah, Unite Arab Emirates.
39. Dahl EL, Rosenthal PJ. 2007. Multiple antibiotics exert delayed effectsagainst the Plasmodium falciparum apicoplast. Antimicrob Agents Che-mother 51:3485–3490. https://doi.org/10.1128/AAC.00527-07.
40. Vial HJ, Wein S, Farenc C, Kocken C, Nicolas O, Ancelin ML, Bressolle F,Thomas A, Calas M. 2004. Prodrugs of bisthiazolium salts are orallypotent antimalarials. Proc Natl Acad Sci U S A 101:15458 –15463. https://doi.org/10.1073/pnas.0404037101.
41. Wein S, Maynadier M, Tran Van Ba C, Cerdan R, Peyrottes S, Fraisse L, VialH. 2010. Reliability of antimalarial sensitivity tests depends on drugmechanisms of action. J Clin Microbiol 48:1651–1660. https://doi.org/10.1128/JCM.02250-09.
42. Miller RS, Wongsrichanalai C, Buathong N, McDaniel P, Walsh DS, KnirschC, Ohrt C. 2006. Effective treatment of uncomplicated Plasmodiumfalciparum malaria with azithromycin-quinine combinations: a random-ized, dose-ranging study. Am J Trop Med Hyg 74:401– 406.
43. Taylor WR, Widjaja H, Richie TL, Basri H, Ohrt C, Tjitra, Taufik E, Jones TR,Kain KC, Hoffman SL. 2001. Chloroquine/doxycycline combination versuschloroquine alone, and doxycycline alone for the treatment of Plasmo-dium falciparum and Plasmodium vivax malaria in northeastern IrianJaya, Indonesia. Am J Trop Med Hyg 64:223–228. https://doi.org/10.4269/ajtmh.2001.64.223.
44. Witkowski B, Khim N, Chim P, Kim S, Ke S, Kloeung N, Chy S, Duong S,Leang R, Ringwald P, Dondorp AM, Tripura R, Benoit-Vical F, Berry A,Gorgette O, Ariey F, Barale J-C, Mercereau-Puijalon O, Menard D. 2013.Reduced artemisinin susceptibility of Plasmodium falciparum ring stagesin western Cambodia. Antimicrob Agents Chemother 57:914 –923.https://doi.org/10.1128/AAC.01868-12.
45. Dolezal M, Zitko J. 2015. Pyrazine derivatives: a patent review (June2012-present). Expert Opin Ther Pat 25:33– 47.
46. Fokas E, Prevo R, Pollard JR, Reaper PM, Charlton PA, Cornelissen B, VallisKA, Hammond EM, Olcina MM, Gillies McKenna W, Muschel RJ, BrunnerTB. 2012. Targeting ATR in vivo using the novel inhibitor VE-822 resultsin selective sensitization of pancreatic tumors to radiation. Cell Death Dis3:e441. https://doi.org/10.1038/cddis.2012.181.
47. Voorman RL, Maio SM, Hauer MJ, Sanders PE, Payne NA, Ackland MJ.1998. Metabolism of delavirdine, a human immunodeficiency virustype-1 reverse transcriptase inhibitor, by microsomal cytochrome P450in humans, rats, and other species: probable involvement of CYP2D6 andCYP3A. Drug Metab Dispos 26:631– 639.
48. Talpaz M, Shah NP, Kantarjian H, Donato N, Nicoll J, Paquette R, CortesJ, O’Brien S, Nicaise C, Bleickardt E, Blackwood-Chirchir MA, Iyer V, ChenT-T, Huang F, Decillis AP, Sawyers CL. 2006. Dasatinib in imatinib-resistant Philadelphia chromosome-positive leukemias. N Engl J Med354:2531–2541. https://doi.org/10.1056/NEJMoa055229.
49. Plouffe D, Brinker A, McNamara C, Henson K, Kato N, Kuhen K, Nagle A,Adrián F, Matzen JT, Anderson P, Nam T-G, Gray NS, Chatterjee A, JanesJ, Yan SF, Trager R, Caldwell JS, Schultz PG, Zhou Y, Winzeler EA. 2008.In silico activity profiling reveals the mechanism of action of antimalar-ials discovered in a high-throughput screen. Proc Natl Acad Sci U S A105:9059 –9064. https://doi.org/10.1073/pnas.0802982105.
50. Chia WN, Lee YQ, Tan KS-W. 2017. Imaging flow cytometry for the screeningof compounds that disrupt the Plasmodium falciparum digestive vacuole.Methods 112:211–220. https://doi.org/10.1016/j.ymeth.2016.07.002.
51. Egan TJ, Mavuso WW, Ncokazi KK. 2001. The mechanism of �-hematinformation in acetate solution. Parallels between hemozoin formationand biomineralization processes. Biochemistry 40:204 –213.
52. Triglia T, Cowman AF. 1994. Primary structure and expression of thedihydropteroate synthetase gene of Plasmodium falciparum. Proc NatlAcad Sci U S A 91:7149 –7153. https://doi.org/10.1073/pnas.91.15.7149.
53. Sandlin RD, Fong KY, Stiebler R, Gulka CP, Nesbitt JE, Oliveira MP, OliveiraMF, Wright DW. 2016. Detergent-mediated formation of �-hematin:heme crystallization promoted by detergents implicates nanostructureformation for use as a biological mimic. Crystal Growth Des 16:2542–2551. https://doi.org/10.1021/acs.cgd.5b01580.
55. Lee YQ, Goh ASP, Ch’ng JH, Nosten FH, Preiser PR, Pervaiz S, Yadav SK,Tan KSW. 2014. A high-content phenotypic screen reveals the disruptivepotency of quinacrine and 3=,4=-dichlorobenzamil on the digestive vac-uole of Plasmodium falciparum. Antimicrob Agents Chemother 58:550 –558. https://doi.org/10.1128/AAC.01441-13.
56. Verhaeghe P, Dumètre A, Castera-Ducros C, Hutter S, Laget M, FersingC, Prieri M, Yzombard J, Sifredi F, Rault S. 2011. 4-Thiophenoxy-2-trichloromethyquinazolines display in vitro selective antiplasmodialactivity against the human malaria parasite Plasmodium falciparum.Bioorg Med Chem Lett 21:6003– 6006. https://doi.org/10.1016/j.bmcl.2011.06.113.
Screen for Plasmodium DV-Disrupting Molecules Antimicrobial Agents and Chemotherapy
March 2018 Volume 62 Issue 3 e02031-17 aac.asm.org 17
