RESEARCH ARTICLE SUMMARY

ANTIMALARIALS

Open-source discovery of chemicalleads for next-generationchemoprotective antimalarialsYevgeniya Antonova-Koch Stephan Meister Matthew Abraham Madeline R LuthSabine Ottilie Amanda K Lukens Tomoyo Sakata-Kato Manu VanaerschotEdward Owen Juan Carlos Jado Steven P Maher Jaeson Calla David PlouffeYang Zhong Kaisheng Chen Victor Chaumeau Amy J Conway Case W McNamaraMaureen Ibanez Kerstin Gagaring Fernando Neria Serrano Korina EribezCullin McLean Taggard Andrea L Cheung Christie Lincoln Biniam AmbachewMelanie Rouillier Dionicio Siegel Franccedilois Nosten Dennis E KyleFrancisco-Javier Gamo Yingyao Zhou Manuel Llinaacutes David A Fidock Dyann F WirthJeremy Burrows Brice Campo Elizabeth A Winzeler

INTRODUCTIONMalaria remains adevastat-ing disease affecting 216 million people annu-ally with 445000 deaths occurring primarily inchildren under 5 years oldMalaria treatmentrelies primarily ondrugs that target the disease-causingasexualbloodstages (ABS)ofPlasmodiumparasites the organisms responsible for humanmalaria Whereas travelers may rely on short-termdaily chemoprotective drugs those living inendemic regions require long-termmalaria pro-

tection such as insecticide-treated nets (ITNs)and vector control However ITNs do not fullyshield individuals from malaria may lose po-tency with time and can be bulky and difficultto use Another concern is that mosquitos maybecome resistant to the active insecticides thatare used in ITNs and vector control

RATIONALE As the possibility of malariaelimination becomesmore tangible the ideal

antimalarial medicine profile should includechemoprotection Chemoprotectivemedicinestypically work against the exoerythrocytic par-asite forms that invade and develop in the liverandare responsible for the earliest asymptomaticstage of the infection Suchmedicines could beformulated to provide long-acting prophylaxissafeguarding individuals that are living near ortraveling to areas that have been cleared of

parasites Long-acting che-moprotection in endemicregions could also greatlyreduce circulatingparasitenumbersandpotentially re-place a vaccine in an elim-inationcampaignAlthough

millions of compounds have been screened foractivity against parasiteABS and somehavebeensubsequently tested for potential prophylactic ac-tivity large-scale searches thatbeginwithprophy-lactic activity havenot beenperformedbecauseofthe complexity of the assay This assay requirestheproductionof infected laboratory-rearedmos-quitoes and hand-dissection of the sporozoite-infected salivary glands frommosquito thoraxes

RESULTS Todiscover leads fornext-generationchemoprotective antimalarial drugs we usedluciferase-expressingPlasmodium spp parasitesdissected from more than a million mosquitoesover a 2-year period to test more than 500000compounds for their ability to inhibit liver-stagedevelopmentofmalaria (681 compoundsshowedahalf-maximal inhibitoryconcentrationoflt1mM)Cluster analysis identified potent andpreviouslyunreported scaffold families as well as otherseries previously associatedwith chemoprophy-laxis These leads were further tested throughmultiple phenotypic assays that predict stage-specific and multispecies antimalarial activityThis work revealed compound classes that arelikely to provide symptomatic relief fromblood-stage parasitemia in addition to providing pro-tectionTarget identificationbyuseof functionalassays in vitro evolution or metabolic profilingrevealed 58 mitochondrial inhibitors but alsomany chemotypes possibly with previously un-knownmechanisms of action somewhichmaydisrupt the host pathogen signaling

CONCLUSION Our data substantially expandsthe set of compounds with demonstrated activ-ity against two known targets of chemoprotectivedrugs cytochrome bc1 and dihydroorotate dehy-drogenase These present a rich collection of chem-ical diversity that may be exploited by membersof the community seeking to accelerate malariaeliminationwith chemoprotection and chemopro-phylaxis through open-source drug discovery

RESEARCH

Antonova-Koch et al Science 362 1129 (2018) 7 December 2018 1 of 1

The list of author affiliations is available in the full article onlineCorresponding author Email ewinzelerucsdeduCite this article as Y Antonova-Koch et al Science 362eaat9446 (2018) DOI 101126scienceaat9446

A Plasmodium vivax liver-stage schizont on a lawn of hepatocytes The parasite schizonthas been stained with antibodies to parasite HSP70 (red) and UIS4 (yellow) Cell (parasiteand hepatoma) nuclei are shown in blue This study identifies compounds that can preventthe development of these liver-stage parasites and may function as chemoprotective drugsfor malaria

ON OUR WEBSITE

Read the full articleat httpdxdoiorg101126scienceaat9446

on October 14 2020

httpsciencesciencem

agorgD

ownloaded from

RESEARCH ARTICLE

ANTIMALARIALS

Open-source discovery of chemicalleads for next-generationchemoprotective antimalarialsYevgeniya Antonova-Koch1 Stephan Meister1 Matthew Abraham1 Madeline R Luth1Sabine Ottilie1 Amanda K Lukens23 Tomoyo Sakata-Kato3 Manu Vanaerschot4Edward Owen5 Juan Carlos Jado1 Steven P Maher67 Jaeson Calla1David Plouffe8 Yang Zhong8 Kaisheng Chen8 Victor Chaumeau910Amy J Conway67 Case W McNamara8dagger Maureen Ibanez8 Kerstin Gagaring8daggerFernando Neria Serrano11 Korina Eribez1 Cullin McLean Taggard1 Andrea L Cheung1Christie Lincoln1 Biniam Ambachew1 Melanie Rouillier12 Dionicio Siegel13Franccedilois Nosten910 Dennis E Kyle67 Francisco-Javier Gamo12 Yingyao Zhou8Manuel Llinaacutes514 David A Fidock4 Dyann F Wirth23 Jeremy Burrows12Brice Campo12 Elizabeth A Winzeler113Dagger

To discover leads for next-generation chemoprotective antimalarial drugs we testedmore than500000 compounds for their ability to inhibit liver-stage development of luciferase-expressingPlasmodium spp parasites (681 compounds showed a half-maximal inhibitory concentrationof less than 1micromolar) Cluster analysis identified potent and previously unreported scaffoldfamilies as well as other series previously associated with chemoprophylaxis Further testingthrough multiple phenotypic assays that predict stage-specific and multispecies antimalarialactivity distinguished compound classes that are likely to provide symptomatic relief byreducing asexual blood-stage parasitemia from those which are likely to only prevent malariaTarget identification by using functional assays in vitro evolution or metabolic profilingrevealed 58 mitochondrial inhibitors but also many chemotypes possibly with previouslyunidentified mechanisms of action

Malaria remains a devastating disease with216 million annual cases and 445000deaths primarily in children under 5 yearsold Malaria treatment relies primarilyon drugs that target the disease-causing

asexual blood stages (ABS) of the parasitePlasmodium falciparum These drugs includethe 4-aminoquinolines piperaquine and amodi-aquine the antifolates pymimethamine andsulfadoxine and the endoperoxides artemisininand its derivatives artesunate artemether anddihydroartemisinin (1) Artemisinin-based com-bination therapies (ACTs such as artemether-lumefantrine) are being used worldwide asfirst-line treatments (1)Although estimated malaria mortality rates

have decreased by 47worldwide since 2000 (1)in Southeast Asia resistance has emerged to the

artemisinins and ACT treatment failures arerising (2) In anticipation of eventual widespreadACT failure there has been a focused and co-ordinated effort to place new antimalarial can-didates into the drug development pipeline (wwwmmvorgresearch-developmentrd-portfolio) (3)However in vitro resistance can be generatedfor most of these new classes in ABS parasites(4) suggesting that the antimalarial pipelinewill require continuous replenishmentAn alternative approach is to create drugs that

prevent malaria by inhibiting parasites duringtheir initial stage of development in the liver be-fore their initiation of symptomatic blood-stageinfection Using chemotherapy for prophylaxis isa well-established tool to prevent malaria causedby different Plasmodium spp To prevent malariatravelersmay take oral atovaquone plus proguanil

doxycycline or mefloquine In endemic regionsof seasonal malaria in West Africa children aregiven SPAQ [sulphadoxine-pyrimethamine (SP)plus amodiaquine] as seasonal malaria chemo-prevention and pregnant women who are themost vulnerable group may also take SP forintermittent preventative therapy (5)Drugs that selectively affect the developing

liver stages of P falciparum and the relapsingspecies Plasmodium vivax have the potentialto engage new protein targets not present norrequired in parasite blood stages Such drugscould overcome both the problem of resistanceand compliance The number of parasites in theearly liver stage are low (hundreds versus bil-lions in the blood stage) reducing the proba-bility that drug resistancendashconferring mutationsmight emerge This feature could make theseliver-active compounds suitable for chemoprotec-tion and with sufficient demonstrated safety formass drug-administration or malaria-eliminationcampaignsTo identify chemoprotective candidates we

applied a liver-stage phenotypic screen to alibrary of gt500000 small molecules Our dataidentify new scaffold families that exclusivelytarget liver stages that may provide prophylacticprotection as well as new scaffolds that actagainst known targets such as dihydroorotatedehydrogenase (DHODH) These data comprisenew leads for antimalarial open-source drugdiscovery

Primary screening results

Previous high-throughput screens for antima-larial compounds have generally focused on theABS which can be readily cultured en masse(6ndash8) To discover hits with possible protectiveactivity blood stagendashactive compounds have beenfurther retested against malaria hepatic stagesoften using sporozoites from the rodent malariaspecies Plasmodium yoelii or Plasmodium berghei(9 10) Because there are no suitable methods foraxenically culturing sporozoites (which are respon-sible for liver-stage infection) this retesting hasrequired the production of infected laboratory-reared mosquitoes and hand dissection of thesporozoite-infected salivary glands from mos-quito thoraxes Despite this complex challengethousands of compounds have been examinedby using this general workflow progression lead-ing to previously unidentified chemoprotectivecandidates including KAF156 (11) KDU691 (12)DDD107498 (13) and BRD3444 (14) Howeverthis approach would not reveal compounds thatmight protect from infection without affecting

RESEARCH

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 1 of 8

1School of Medicine University of California San Diego 9500 Gilman Drive 0760 La Jolla CA 92093 USA 2Harvard T H Chan School of Public Health 665 Huntington Avenue Boston MA02115 USA 3The Broad Institute 415 Main Street Cambridge MA 02142 USA 4Division of Infectious Diseases Department of Microbiology and Immunology Columbia University MedicalCenter New York NY 10032 USA 5Department of Biochemistry and Molecular Biology and Center for Malaria Research Pennsylvania State University University Park PA 16802 USA 6Centerfor Tropical and Emerging Global Diseases University of Georgia 500 D W Brooks Drive Athens GA 30602 USA 7Department of Global Health University of South Florida 3720 SpectrumBoulevard Tampa FL 33612 USA 8The Genomics Institute of the Novartis Research Foundation 10675 John J Hopkins Drive San Diego CA 92121 USA 9Shoklo Malaria Research Unit MahidolOxford Research Unit Faculty of Tropical Medicine Mahidol University Mae Sot Thailand 10Centre for Tropical Medicine and Global Health Nuffield Department of Medicine University ofOxford Oxford UK 11Tres Cantos Medicines Development Campus Malaria DPU GlaxoSmithKline Severo Ochoa 2 Tres Cantos 28760 Madrid Spain 12Medicines for Malaria Venture PostOffice Box 1826 20 Route de Pre-Bois 1215 Geneva 15 Switzerland 13Skaggs School of Pharmacy and Pharmaceutical Sciences University of California San Diego 9500 Gilman Drive 0741La Jolla CA 92093 USA 14Department of Chemistry and Center for Infectious Diseases Dynamics Pennsylvania State University University Park PA 16802 USAPresent address BioSero San Diego CA 92122 USA daggerPresent address California Institute for Biomedical Research La Jolla CA USADaggerCorresponding author Email ewinzelerucsdedu

on October 14 2020

httpsciencesciencem

agorgD

ownloaded from

blood stages nor identify compounds that mod-ulate human hepatocyte targets and preventparasite liver-stage developmentRecently a screening method was developed

that was suitable for testing many compoundsagainst liver-stage parasites (10) This assay (Pbluc)relies on an antifolate-resistant P berghei rodentparasite that constitutively expresses luciferaseduring the sporozoite stage (15) Sporozoites aredissected frommosquitoes washed and layeredon confluent HepG2 cells that have been pre-incubated with compounds If the parasites areable to establish a successful exoerythrocytic stageinfection even if only 1 to 2 of the hepatocytesare infected the cultures will emit enough lightin the presence of luciferin to be detected withsensitive instruments Light intensity is propor-tional to parasite infection rate and this simpleassay can be run by using 1536-well plates TheP berghei system has a distinct advantage overthe human pathogen P falciparum in that thisparasite can infect a variety of hepatoma celllines which results in ease of use and reprodu-cibility Furthermore the development periodis only 48 hours which means that hepatocytecultures do not overgrow and special coculturesystems are not needed Sporozoite yields aregenerally higher and there are fewer biosafetyprecautions In addition because cellular detox-ification systems have been generally lost fromhepatoma cell lines compounds remain potenteasing their discovery The assay has shown to bevery predictive of causal prophylactic activityWethus sought to apply this screeningmethod to alarge library with the goal of finding a largerselection of starting points for chemoprophylacticdrugsScreening was conducted by using a library

of 538273 compounds from Charles River con-sisting of smallmoleculeswith an averageweightof 369 Da (6006 to 1298 Da) The library wasmostly drug-like (versus probe-like) with an aver-age logP of 3 an average number of hydrogenbond donors of 11 and 59 hydrogen bond ac-ceptors On average each compound had 62rotatable bonds 5270 compounds were repre-sented in duplicate or triplicate (table S1) Sub-stantial structural redundancy was included inthe library to establish structure-activity relation-ships in scaffold families In contrast to manypreviously published studies (6ndash8) all the com-pounds in the library including hits and neg-atives have structures and are commerciallyavailable facilitating use by the communityThe screen was run over an 18-month period

with ~20000 compounds screenedweekly (Fig 1)to accommodate the time-consuming productionof infected Anopheles mosquitoes Each weeka team dissected ~1000 mosquitoes from whichthe sporozoites were isolated and cleaned Ap-proximately 1000 sporozoites were added in a5 ml volume per well plate by using a bottlevalveinstrument to a lawn of HepG2-CD81 cells thathad been prespotted into the 1536-well plateswith 50 nl of compounds at 05 to 784mM (Fig 1)After a 48-hour incubation luciferin was addedand luminescence was measured Because a

compound that kills hepatocytes could resultin reduced luciferase signal (because parasitespresumably need a live hepatocyte for intra-cellular survival) we also tested a randomlyselected subset of the compounds (~179600) forhepatocyte toxicity using a Cell-Titer Glo assay(materials and methods) that measures thenumber of viable cells based on the amountof adenosine 5prime-triphosphate (ATP) that is pres-ent (HepG2tox)

Cheminformatic analysis of theprimary screen

After collating the data and ranking the per-centage inhibition for 538273 compounds fromthe 1536-well screen 64172 compounds showedinhibition of gt50 and 21336 showed inhibitionlevels of gt75 (Fig 1) Although a higher-than-normal number of false positives and negativesis expected for a single-point screen (data maybe affected by edge effects transfer errors orvariation in dispense volumes) the initial screenwas informative For example the library included197 ABS-active compounds that had been previ-ously tested in a high-content imaging screen ofP yoelii liver stages (9) including 18 compoundswith activity of less than 10 mM (data file S1) Ofthose 15 showed gt50 inhibition in the primaryscreen Likewise themajority of compounds thatwere considered inactive in the high-contentimaging screen showed lt75 inhibition here(129 out of 163) (data file S1)To further evaluate the quality of the primary

Pbluc data we performed compound clustering(Fig 2 and data file S2) We hierarchically clus-tered all 538273 compounds in the primaryscreen and separated clusters using a minimumTanimoto coefficient of 085 similarity (materialsand methods) From this we identified 281090compound clusters with an average hit fractionof 0016 plusmn 0112 We calculated the probability ofhit enrichment (gt83 inhibition in Pbluc) bychance for each cluster (data file S2) Not un-expectedly certain scaffold families with knownactivity against Plasmodium liver stages such asthe phosphatidylinositol-4-OH kinase (PI4K)ndashinhibiting imidazopyrazine family (12) includeda high proportion of highly active compoundsFor example with an imidazopyrazine group(family 227215) 11 of the 14members were highlyactive [average 811 log10 probability of dis-tribution by chance (log10 p) = ndash1516] (Fig 2A)and these were similar to the known imidazo-pyrazine KDU691 (12) Likewise scaffold family37533 consisted of four members where all fourmembers in the library were more than 80active (average 884) All were very similar toMMV390048 another PI4K inhibitor (Fig 2B)Atovaquone-like scaffolds (cluster 68966) werealso present in the library and active at rateshigher than expected by chance (eight of ninemembers active log10 p = ndash1088) (Fig 2C) Thehit fractions of cluster 69866 37533 and 227215(089 075 and 079 respectively) are all in the top11 percentile when considering the entire librarySeveral additional scaffold families with known

activity against malaria parasites were also found

in the library and showed activity although thesefamilies would not have been discovered withcluster analysis at the similarity thresholds usedFor example cycloguanil (834 inhibition) wasfound to be active in the 538273-compoundlibrary but had only one close relative (gt 85Tanimoto similarity) which was not active(log10 p = ndash131) Relaxing the stringency how-ever revealed three other relatives one of whichwas also primary hit (log10 p = ndash242) (Fig 2E)Likewise compounds similar to DSM265 (16)and DDD107498 (Fig 2 F and G) (13) two otherpotent compounds with nanomolar activityagainst exoerythrocytic stages were also identifiedthrough relaxed similarity searching but were inclusters that might have arisen through chance(MMV011256 one of three active log10 p = ndash135MMV1478835 two of nine active log10 p ndash140MMV1370261 one of two active log10 p = ndash150) orwere singletons (MMV1386820)

Reconfirmation

To obtain a reasonable working set for reconfir-mation (9989 compounds) we selected com-pounds that showed gt83 Pbluc inhibitionand HepG2tox of less than 50 when available(data file S3) Cut off and filtering criteria did notbias physicochemical properties of primary hitswhose subset compounds had similar propertiesto the larger library with weights of 388 Da(9851 to 1362) 11 hydrogen bond donors 55hydrogen bond acceptors and 59 rotatable bondson average (fig S1 and table S2) To confirmprimary hits the location of wells with possibleactive compounds were identified and then ahit-picking instrument was used to reassemblea collection for reconfirmation (data file S3)We arrayed these hit compounds in eight-pointdose-response format (5 nM to 10 mM) andreassessed their potency and efficacy in duplicatein three different assaysmdashincluding the Pblucassay theHepG2tox assay and a counterscreenmdashto test whether a given compound would inhibitfirefly luciferase in a biochemical assay (Ffluc)Altogether complete Pbluc dose-response curvescould be fitted for 5942 compounds [half-maximal inhibitory concentration (IC50) lt 154 mMaverage 471 mM] and 681 showed complete Pblucinhibition at all tested concentrations (18 com-pounds) or an IC50 of lt1 mM (663 compounds)(data file S3) However of the 5942 reconfirmedhits 790 demonstrated moderate inhibition ofhepatocyte viability [HepG2tox CC50 (50 cyto-toxicity concentration) lt 2X Pbluc IC50 andHepG2tox CC50 lt maximum tested concentra-tion] and 465 of these also interferedwith fireflyluciferase production (Ffluc IC50 lt 2X Pbluc IC50and Ffluc IC50 ltmaximum tested concentration)(data file S3)Adding these data to our cluster analysis

(Fig 2) showed enrichment of false positivesarising from the use of luciferase as a surrogatefor parasite development For example 526 scaf-folds of the imidazopyridine-carboxamide group(128188) were found in the library of 538273compounds of which 145 were in the hit list(log10 p lt ndash100) (data file S2) Of these 116

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 2 of 8

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

showed biochemical luciferase inhibition lt10 mM(29 less than 1 mM) (data files S2 and S3) Vis-ualization showed that other less abundantscaffolds such as those in cluster 30984 or4198 also showed enrichment of luciferaseactivity at higher rates than expected by chance(Fig 2 and data file S2) Likewise some enrichedscaffold families (log10 p = ndash1445) were clearlytoxic to human cells (cluster 95979) (Fig 2 anddata file S2)To further establish the integrity of the hits

we next assembled a collection of compoundsfor third-round independent testing The non-

toxic second round hits were filtered throughthe general medicinal chemistry compoundfilters by using Biovia ScienceCloud This fil-tering removed 619 compounds that either failedLipinskirsquos rule of 5 (based on calculated proper-ties) as a surrogate for ldquodrug-likerdquo had structuralfeatures consistent with high chemical reactivityor instability had a high likelihood of nonspecificcovalent interaction such as thiourea or was anenone which would react with nucleophiles suchas thiols or disrupt disulphide bonds (17) Theremaining set was then prioritized on the basisof potency (IC50) and calculated lipophilicity

(cLogP) yielding 953 compounds prioritizedas follows 445 actives with IC50 lt 1 mM (mostpotentmdashno additional cLogP filter) 268 activesgt1 mM but clogP lt 25 (low lipophilicity com-pounds with lower potency) and 240 activesgt1 mM but 25 lt clogP lt 3 (as above with mod-erate lipophilicity)With this set of 953 compounds we inter-

rogated the databases of commercially availa-ble compounds for acquisition confirmationand further profiling 631 (repurchased valida-tion set) were available and thus further char-acterized (data file S4) These compounds were

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 3 of 8

Fig 1 Screen workflow (A) The Charles River library (538273compounds) was plated into 1603 384-well plates For the primary Pblucsingle-point screen compounds from four of the 384-well plates weretransferred with a pin-tool instrument (50 nl per well) into 1536-well assayplates containing seeded HepG2 cells (3 times 103 cells per well)The next dayP berghei luciferasendashexpressing sporozoites were freshly prepared frominfected Anopheles stephensi mosquitoes and ~1000 sporozoites in a5 ml volume were added to each well After 48 hours P bergheindashLucgrowth within hepatocytes was measured with bioluminescence (B) Toprepare source plates for the first round of reconfirmation (second roundof screening) the 9989 hit compounds were transferred from the original384-well library with an automated hit-picking system and serially

diluted into eight points (13 dilutions) for dose-response screening induplicate The hit compounds (50 nl per well) in serial dilutions wereacoustically transferred into assay wells containing HepG2 cells for Pblucand HepG2tox assays In addition biochemical recombinant luciferaseinhibition assay (Ffluc) were also performed (C) For final reconfirmation(third round) 631 compounds prepared from re-sourced powders andwere serially diluted (10 points 13 dilution) and plated into Aurora 1536-well compound plates Compounds (50 nl per well) were acousticallytransferred into 1536-well assay plates Multiple dose-respose assays suchas Pbluc HepG2tox Ffluc and ABS-Sybr were performed to determineIC50 in the third round of screening A P vivax liver schizont formationhigh-content assay in single-point (2X) was also performed

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

arrayed within the 1536-well plate format into10-point dose response tested in duplicate (tech-nical replicates) and counterscreened in multipledifferent assays repeated up to four times with dif-ferent sporozoite batches Two separate HepG2toxassays and one Ffluc also were performed Retest-ing showed a high reconfirmation rate (82 withan average Pbluc IC50 lt 10 mM) and 376 com-pounds with Ffluc IC50 and HepG2tox CC50 lt 2timesPbluc IC50 (fig S2) All compounds were alsotested in dose-response format against ABS ofthe multidrug-resistant P falciparum Dd2 strainusing a standard SYBR green incorporation blood-stage assay that identifies compounds that pre-vent growth and development (ABS-Sybr fourbiological replicates) (6)

Activity against P vivax

To better understand species specificity the afore-mentioned repurchased validation set (631)

reconfirmed in eight-point dose assays withPbluc was also tested against the human path-ogen P vivax For this we used a single-pointhigh-content imaging screen (PvHCI) with pri-mary humanhepatocytesmaintained in 384wellplates Hepatocyteswere inoculatedwith P vivaxsporozoites dissected frommosquitoes fed withblood obtained from malaria patients The fol-lowing day compounds were added to a finalconcentration of 10 mM and then reapplied atdays 2 and 3 and 4 After compound treatmentcultureswere fixed onday6 after infection stainedwith immunoreagents against PvUIS4 (materialsand methods) and we performed high-contentimaging andanalysis to score eachwell for growthof liver schizonts Of the 631 compounds screenedin this single-point assay 163 were confirmed toinhibit P vivax schizont growth by at least 70(normalized to positive control KDU691) (data fileS4) in at least one of two biological replicates and

91 were confirmed in both replicates (fig S2) Thisconfirmation rate was achieved despite severalcritical differences between the rodent andhuman assays including hepatic metabolismof unoptimized compounds and addition ofcompound on day 1 after infection (PvHCI)instead of preinfection (Pbluc) Furthermore12 compounds (fig S2) were considered notreconfirmed in a third round of P berghei testingbut were nevertheless still active in P vivax(for example MMV1288041 inhibited P vivaxby 100 in both replicates and had an IC50 ofP berghei 327 mM in second-round testing)highlighting how experimental design and anal-ysis constraints (for example curve-fitting com-pounds that did not fully inhibit growth in Pblucat 10 mM the highest concentration tested wouldbe considered ldquonot confirmedrdquo) could lead topossible false negatives On the other hand com-pounds could lose potency over repeated use

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 4 of 8

Fig 2 Cluster analysis (A) For display 405 compounds from 68 clustersthat show a P value le 005 cluster size ge 4 and hit fraction ge 075 arepresented (data file S2) Most common substructure (MCS) per cluster isidentified by using the top three active compounds and a hierarchical treewas constructed from the MCSs to illustrate the intergroup connectionCompound members were then added to surround MCS nodes Allcompound nodes are colored by hit status and shaped by otherannotations Primary hits are orange reconfirmation hits (Pbluc IC50 lt 10 mM)are red third-round reconfirmation set (631 compounds) is purple

and others are light blue P falciparum asexual blood statendashactivecompounds (ABS-Sybr IC50 lt 10mM) are indicated by squares luciferaseinhibitors (Ffluc IC50 lt Pbluc IC50 2) are indicated by triangleshepatocyte toxic (primary HepG2tox gt 50 or HepG2tox CC50 lt PblucIC50 2) are indicated by diamonds and the rest are shown as circles(B to G) Active compounds in selected clusters [(B) (C) and (D)] with hitfractions of less than 075 as well as examples of singleton hits that aresimilar to the known antimalarial compounds (E) cycloguanil (F)DDD107498 (13) and (G) DSM265 (data file S3) (16)

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

owing to freeze-thaw and hydration there couldbe differences in liver-stage schizogony betweenP vivax and P berghei (liver schizonts producefar greater quantities of liver merozoites oversix additional days of development in com-parison to the 72-hour liver cycle of P berghei)and the wild-type P vivax may be more drug-sensitive than P bergheindashANKA-GFP-Luc-SMCON

or P falciparumDd2 both of which are resistantto some antifolates

Activity against P falciparumblood stages

To further study the function of the third-roundcompounds we next testedwhether the repurch-ased validation set of 631 would be active againstP falciparum ABS parasites The active setshowed several known chemotypes One exampleMMV1468363 showed a high degree of similarityto tetracyclic benzothiazepines which are knowncytochrome bc1 inhibitors (18) There were sevenclosely related members of the tetracyclic benzo-thiazepines scaffold family in the larger libraryand six of the seven showed inhibition rates ofgreater than 85 in the primary screen (log10 p =ndash7969 cluster 157835) (Fig 2) Another recog-nizable scaffold was that of MMV1005663 sim-ilar to the possible FabI inhibitor trichlocarbanas well as the liver stagendashactive malaria box mem-ber MMV665852 (19) This latter compound isassociated with amplifications of the gene encod-ing phospholipid flippase pfatp2 (19) This com-pound series (cluster 84381) was representedmultiple times in the library with all five testedmembers showing strong inhibition (average 93log10 p = ndash734) in the primary screen (Fig 2)Three hundred ninety-eight liver-active com-

pounds were inactive in P falciparum ABS-Sybr(average IC50 of gt 10 mM) (fig S2) highlightingthe rich new untapped chemical space that wassampled by using this screening cascade whencompared with the ABS-first cascade Althoughsome could be due to species-specific activities130 compoundswere active against both P vivax(in at least one replicate) and P berghei schizonts(fig S2) but were inactive against ABS Althoughour use of antifolate-resistant P berghei and amultidrug-resistant P falciparum strain may in-fluence this result most of these scaffolds likelyaffect parasite or human processes that exist onlyin the hepatic stages These scaffolds generally didnot show any similarities to known antimalarialsbut several of these were supported by strongstatistical overrepresentation in the primary screen(Fig 2 and data file S2) For example seven ofthe ninemembers of the imidazoquinoline cluster(112845)were active in the primary screen (log10 p=ndash889) (Fig 2)

Mechanism of action studies reveal anabundance of mitochondrial transportchain inhibitors

To further investigate the mechanism of actionof 13 of the most potent ABS hits (Fig 3A) weacquired additional compound from commercialvendors Because most if not all target identifi-cation methods require activity in P falciparum

blood stages all compounds selected for initialtarget discoverywere active against ABS parasiteswith an average IC50 of 465 nM (range of 13 nM to12 mM)As an initial pass we first subjected the com-

pounds to metabolic profiling (20) This liquidchromatographyndashmass spectrometry (LC-MS)ndashbased method measures several hundred metab-olites and identifies those that show statisticallysignificant increases or decreases upon parasitecompound exposure (data file S5)Whereas threegave ambiguous results 10 of the 13 analyzedscaffolds gave a metabolic profile signature anal-ogous to that of atovaquone (fig S3) indicatingthat these 10 most likely interfere with one ormore targets in themitochondrial electron trans-port chain (mETC) a known druggable pathwayfor P falciparum blood and liver stages (Fig 3A)The set of 13 compounds represents 11 distinctscaffolds (Fig 3B) so this degree of functionaloverlap would not have been predicted by struc-ture alone To our knowledge none of the mo-lecules have been previously identified as actingagainst the mitochondrial electron transportpathwayTo further confirm that these 10 compounds

(representing eight chemotypes) inhibited themETC we took advantage of a transgenic parasiteline that overexpresses the Saccharomycescerevisisae dihydroorotate dehydrogenase (Dd2-ScDHODH) (21) Unlike the type-2 P falciparumenzyme that is dependent on cytochrome bc1 forubiquinone the cytosolic type-1 yeast enzyme canuse fumarate as an electron acceptor This allowsthe transgenic parasites to bypass the need forETC activity to provide ubiquinone to PfDHODH(21) Compounds that target PfDHODH or otherenzymes along the mETC lose potency in theDd2-ScDHODH transgenic cell line As expectedthe Dd2-ScDHODHparasites showmarked (248to gt1000-fold) resistance to the 10 compoundswith the mETC metabolic profile (Fig 3C andtable S3) Furthermore a variation of this func-tional assay can distinguish between inhibitorsof PfDHODH and cytochrome bc1 Specificallythe addition of proguanil to Dd2-ScDHODH pa-rasites restores the inhibitory capabilities of cyto-chrome bc1 inhibitors however growth is notaffected in the case of PfDHODH inhibitors Threeout of six mitochondrial inhibitors tested in theseconditions were not inactivated by proguanilsuggesting a profile consistent with PfDHODHinhibition (Fig 3C and table S4) To further in-vestigate mitochondrial inhibition and becausethere are multiple potential targets we used anin vitro evolution and whole-genome analysis(IViEWGA) approach (22) to further elucidate themolecular target of several of the compoundsincluding MMV1454442 MMV1432711 andMMV142795 First three independent linesresistant to MMV1454442 were isolated aftergrowing in sublethal concentrations of com-pound The resistant clones showed an average42-fold shift in the IC50 (range of 19 to 94) (tableS5) Whole-genome sequencing of the nine clones(three each from three independent selections) aswell as the drug-sensitive parent clone to 78-fold

coverage (table S6) revealed that the resistantlines carried either a single-nucleotide variantPhe188Ile (Fig 3D and data file S6) or a copynumber variant (table S8) in P falciparumdihydroorotate dehydrogenase (PF3D7_0603300)which isawell-validateddrug target inP falciparum(16) This result is consistent with proguanil notaffecting growth inhibition in Dd2-ScDHODHparasites (Fig 3C)MMV1454442 an amino-triazolalthough somewhat similar to a pyrrole-basedDHODH inhibitor (23) is a previously uniden-tified chemotype which would not have beenpredicted with structural information alone ThePhe188 residue is located in the species-selectiveinhibitor-binding pocket of PfDHODH (24) andhas been shown to be in contact with the knownDHODH inhibitor leflunomide (25) and the tri-azolopyrimidine DSM338 (Fig 3D) (26) Further-more mutation of this residue has been shown toconfer resistance to the alkylthiophene inhibitorGenz-669178 (27) suggesting that MMV1454442likely shares the same space (Fig 3D)IViEWGA of MMV1432711-resistant parasites

revealed they had acquired one of two non-synonymous single-nucleotide variants (SNVs)in the gene encoding cytochrome b (data file S7)The amino acid mutations found Tyr126Cys andVal259Leu are located within helix C in theubiquitinol-binding pocket of cytochrome ba catalytically important subunit of the cyto-chrome bc1 complex that contains two reactionsites Q0 (ubiquitinone reduction site) and Qi

(ubiquitinol oxidation site) MMV1432711 has achemical scaffold similar to that of the Qi in-hibitors sowe used a homologymodel of PfCYTb(Fig 3E) to resolve themode of binding Dockinginto the model showed that MMV1432711 is likelya class II inhibitor The allele Y126C was prev-iously reported to confer resistance to decoquinate(28) and MMV008149 (29) To our knowledgeallele Val259Leu has not been reported in theliteratureFor compound MMV1427995 2-[(56-diphenyl-

124-triazin-3-yl)thio]-1-(pyrrolidin-1-yl)propan-1-one in vitro evolution studies yielded tworesistant lines that were cloned for further phe-notyping and whole-genome sequencing (tablesS5 and S6) Clones showed an average 26-foldIC50 shift (range of 18- to 34-fold) in susceptibilityto MMV1427995 (table S5) Sequencing revealedthat both clones carried the amino acid mutationArg95Lys in cytochrome b located in the matrix-oriented region of the protein after the secondtransmembrane domain (Fig 3E and data file S6)One clone also carried an additional Pro102Thrmutation in cytochrome c oxidase subunit 1 Thismutation is located between the second and thirdtransmembrane domain is not located in anyof the iron or copper redox centers and is toour knowledge the first described mutation incytochrome c oxidase selected for during com-pound exposure However this mutation did notinduce a higher resistance level than did theArg95Lysmutation alone andmay thus representa compensatory mutation MMV1427995 is a dif-ferent scaffold family (also overrepresentedin the initial set of screening hits) from known

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 5 of 8

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 6 of 8

Fig 3 Target identi-fication studies(A) Chemicalstructures and IC50 ofselect antimalarialcompounds identifiedas hits Tanimotoclustering demon-strates that mostmolecules are struc-turally distinctalthough some sharesimilar scaffolds(B) Metabolomicanalysis revealsthat 10 of the 13compounds likelytarget the mETC andpyrimidine bio-synthesis pathwaysRobust increases inpyrimidine bio-synthesis precursorsN-carbamoyl-L-aspartate (CA) anddihydroorotate (DHO)are signatures ofmetabolic disruptionof de novo pyrimidinebiosynthesis Themetaprints forMMV1068987 andMMV1431916 are sim-ilar to the metaprintof the PfATP4 inhibi-tor KAE609 whereasthe metaprint forMMV011772 isinconclusive Thenumbers below eachcompound nameindicate the Pearsoncorrelation with anatovaquone profile(fig S3) (C) IC50 ofeach compound inDd2 cells expressingS cerevisiae DHODHnormalized to parentThe transgenicDd2-ScDHODH strainexpresses the cyto-solic type 1 DHODHfrom S cerevisiae(ScDHODH) and isresistant toP falciparummETC inhibitors Abla-tion of compoundactivity in this cell linerelative to its parentindicates inhibition ofDHODH or downstream effectors in the mETC such as Cytbc1 Atovaquone a known Cytb inhibitor was included as a positive control whereas otherlicensed antimalarials with mETC-independent mechanisms of action serve as negative controls (D) Location of Phe188Ile mutation found in whole-genomesequences ofMMV1454442-resistant parasites by using a crystal structure of PfDHODH (4ORM) (27) Amino acid residue 188 is highlighted inmagentaThestructure shows a known PfDHODH inhibitor DSM338 (27) cocrystalized with the protein (E) Homology model of PfCytb (from PDB 1BE3) (35) withTyr126Cys and Val259Leumutations (highlighted in magenta) fromMMV1432711-resistant parasitesThe Arg95 has previously been implicated in atovaquonebinding and resistance (36)

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

P falciparum cytochrome bc1 inhibitors and itstarget would not have been predicted throughsimilarity searchingGiven the high number of mitochondrial in-

hibitors in the dataset we further examined theset of 631 compounds (repurchased validationset) All 631 compounds were tested in du-plicate in eight-point ABS dose response in twoP falciparum D10ndashderived lines (21) one ofwhich expresses ScDHODH (fig S4 and datafile S7) in the presence and absence of 1 mMproguanil in duplicate (~80000 data points)Of the 136 compounds with ABS activity visualinspection showed that 78 were likely not mito-chondrial inhibitors and 58 showed profilesconsistent with mitochondrial inhibition (figsS4 and S5 and data file S7) Of these 10 wereclear DHODH inhibitors (including the threeshown in Fig 3) one was a potential DHODHinhibitor and 47 were likely or possible cyto-chrome bc1 inhibitors (including all nine fromFig 3 that were tested) Strong nonrandomstructure-activity relationships are evident (figS5) validating the assay For example six of theseven compounds that were more than 55similar to MMV1042937 (fig S5 fourth row) inthe set of 631 were predicted to be cytochromebc1 inhibitors (log10 p = ndash609) The seventhMMV1457596 was missed because the IC50 wereall gt10 mM but visual inspection of the curvesshowed an ~75 reduction in signal at 10 mM forthe ScDHODH line relative to the PfDHODH line(data file S7)

Discussion

Previous open-access high-throughput screenshave had a great impact on malaria researchand we anticipate that our data provide a richresource in the search for new antimalarialsDespite the reliance on a nonhuman malariaspecies for screening the repeated rediscoveryof chemotypes with known potent activity againstP falciparum blood stages and P vivax liverschizonts or clinical efficacy in humans showsthat the data from this rodent malaria model arepredictive Furthermore liver schizonticidal ac-tivity is a required component of next-generationantimalarials proposed byMedicines for MalariaVenture critical components of which includeprophylactic and chemoprotective liver-stageefficacy after single exposure (TCP4) (30) Anotheradvantage of the Pbluc assay is the reduced me-tabolic capacity of the host cells used this shouldresult in a higher hit rate withmetabolically liablecompounds in a high-throughput phenotypicscreen Last although it is possible that somecompounds were missed through the use of arodent parasite rodent malaria remains an ef-ficient and important in vivo model for drugefficacy testing compounds that do not act inthis model would be costlier to progress intoanimal causal prophylaxis testing In additionsome of the compounds that have activity heremay eventually show activity against P vivaxhypnozoites and if a radical cure chemotype isto be found its discovery would be made muchmore likely through elimination of liver schizont-

inactive compounds The use of the low-costrodent model allowed a much higher number ofcompounds to be evaluated than would be pos-sible with human parasites which can be dif-ficult and expensive to acquire The large sizeof the library used here facilitated compoundclustering and allowed a probabilistic identifi-cation of active familiesThe high number of parasite mitochondrial

inhibitors (57) that were discovered with com-bined target identification methods is perhapsnot surprising given that compounds were se-lected for target identification on the basis ofpotency and substantial work has shown thatmitochondrial inhibitors are very potent anti-malarials Nevertheless the dataset contains arich set of other ABS scaffolds that could stillbe good starting points for medicinal chemistryefforts designed to improve potency selectivityand reduced toxicity The picomolar inhibitorand clinical candidate KAE609 was derived froman ~90 nM IC50 starting point that was dis-covered in a high-throughput ABS screen (31)Although our target identification efforts

showed that several compounds with activityacross species were mostly in known targetclasses a much larger number of compoundswere active in the liver stages and had no ac-tivity in the blood stages These presumablyact against host pathways that the parasiterequires for development or alternatively theinfected cell may be weakened by the presenceof a parasite and be more susceptible to killingby compounds that target essential host cellularprocesses It is worthwhile to mention that be-cause there are no reported antimalarial com-pounds with exclusive prophylactic activity andmethods for target identification are not readilyavailable for compounds that do not act in theblood stage we are unable to conclude muchabout their mechanism of action Neverthelessthe scaffolds should still represent importantstarting points for prophylaxis stage drugs andexploring new mechanism of actionsIn theory liver-stage antimalarial compounds

could function as more cost-effective ldquochemicalvaccinesrdquo that could replace conventional vac-cines in malaria-elimination campaigns providedsufficient safety A fully protective conventionalvaccine has never been developed for malaria(32) vaccines are very species-specific (and po-tentially strain-specific) whereas chemotherapyis generally not and vaccines which typicallyconsist of recombinant protein or a heat-killedorganism require maintenance of a cold chainIn the case of the irradiated or attenuated cryo-preserved sporozoite vaccine (33) the cost ofgoods (sporozoites that are hand-dissected frominfected mosquitoes) will be much higher than asmall-molecule therapyWe estimate that it costs10 cents per well to create the 1000 sporozoitesneeded for one well of a 1536-well plate Creatingenough sporozoites to immunize a humanwouldcost many times this amountIn an analogous situation long-acting inject-

able HIV drugs have now been developed andtheir deploymentmay help to prevent the spread

of the HIV virus (34) Such intramuscular or sub-cutaneous chemoprotection injections or ldquochem-ical vaccinesrdquo can be deployed in resource-poorsettings and patient compliance is not as muchof an issue as with standard orally availabletablet drugs Because the injectable typicallyconsists of a small molecule refrigeration andmaintenance of a cold chain may not be neededThe cost of goods for an injectable that needs tobe supplied once every 1 to 3 months will mostlikely be lower than the cumulative cost of atablet that needs to be taken every day even con-sidering the cost of a syringe and a health careworker to provide delivery There are also otherlow-cost deliverymethods such as patches Thusit seems entirely feasible that a contribution tothe eradication of malaria could come throughthe use of a long-acting chemical vaccine

Materials and methods

A detailed description is provided in the supple-mentary materials

REFERENCES AND NOTES

1 World Health Organization (WHO) World Malaria Report 2017(WHO 2017)

2 B Blasco D Leroy D A Fidock Antimalarial drug resistanceLinking Plasmodium falciparum parasite biology to theclinic Nat Med 23 917ndash928 (2017) doi 101038nm4381pmid 28777791

3 M A Phillips et al Malaria Nat Rev Dis Primers 3 17050(2017) doi 101038nrdp201750 pmid 28770814

4 E L Flannery D A Fidock E A Winzeler Using geneticmethods to define the targets of compounds with antimalarialactivity J Med Chem 56 7761ndash7771 (2013) doi 101021jm400325j pmid 23927658

5 J R Matangila P Mitashi R A Inocecircncio da LuzP T Lutumba J P Van Geertruyden Efficacy and safety ofintermittent preventive treatment for malaria in schoolchildrenA systematic review Malar J 14 450 (2015) doi 101186s12936-015-0988-5 pmid 26574017

6 D Plouffe et al In silico activity profiling reveals themechanism of action of antimalarials discovered ina high-throughput screen Proc Natl Acad Sci USA105 9059ndash9064 (2008) doi 101073pnas0802982105pmid 18579783

7 W A Guiguemde et al Chemical genetics of Plasmodiumfalciparum Nature 465 311ndash315 (2010) doi 101038nature09099 pmid 20485428

8 F J Gamo et al Thousands of chemical starting points forantimalarial lead identification Nature 465 305ndash310 (2010)doi 101038nature09107 pmid 20485427

9 S Meister et al Imaging of Plasmodium liver stagesto drive next-generation antimalarial drug discoveryScience 334 1372ndash1377 (2011) doi 101126science1211936pmid 22096101

10 J Swann et al High-throughput luciferase-based assayfor the discovery of therapeutics that prevent malariaACS Infect Dis 2 281ndash293 (2016) doi 101021acsinfecdis5b00143 pmid 27275010

11 K L Kuhen et al KAF156 is an antimalarial clinical candidatewith potential for use in prophylaxis treatment andprevention of disease transmission Antimicrob AgentsChemother 58 5060ndash5067 (2014) doi 101128AAC02727-13 pmid 24913172

12 C W McNamara et al Targeting Plasmodium PI(4)K toeliminate malaria Nature 504 248ndash253 (2013) doi 101038nature12782 pmid 24284631

13 B Baragantildea et al A novel multiple-stage antimalarial agentthat inhibits protein synthesis Nature 522 315ndash320 (2015)doi 101038nature14451 pmid 26085270

14 N Kato et al Diversity-oriented synthesis yields novelmultistage antimalarial inhibitors Nature 538 344ndash349(2016) doi 101038nature19804 pmid 27602946

15 B Franke-Fayard et al Murine malaria parasite sequestrationCD36 is the major receptor but cerebral pathology is unlinked tosequestration Proc Natl Acad Sci USA 102 11468ndash11473(2005) doi 101073pnas0503386102 pmid 16051702

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 7 of 8

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

16 M A Phillips et al A long-duration dihydroorotatedehydrogenase inhibitor (DSM265) for prevention andtreatment of malaria Sci Transl Med 7 296ra111 (2015)doi 101126scitranslmedaaa6645 pmid 26180101

17 J B Baell J W M Nissink Seven Year Itch Pan-AssayInterference Compounds (PAINS) in 2017-Utility andLimitations ACS Chem Biol 13 36ndash44 (2018) doi 101021acschembio7b00903 pmid 29202222

18 C K Dong et al Identification and validation of tetracyclicbenzothiazepines as Plasmodium falciparum cytochrome bc1inhibitors Chem Biol 18 1602ndash1610 (2011) doi 101016jchembiol201109016 pmid 22195562

19 A N Cowell et al Mapping the malaria parasite druggablegenome by using in vitro evolution and chemogenomicsScience 359 191ndash199 (2018) doi 101126scienceaan4472pmid 29326268

20 E L Allman H J Painter J Samra M CarrasquillaM Llinaacutes Metabolomic profiling of the malaria box revealsantimalarial target pathways Antimicrob Agents Chemother60 6635ndash6649 (2016) doi 101128AAC01224-16pmid 27572391

21 H J Painter J M Morrisey M W Mather A B VaidyaSpecific role of mitochondrial electron transport in blood-stagePlasmodium falciparum Nature 446 88ndash91 (2007)doi 101038nature05572 pmid 17330044

22 M R Luth P Gupta S Ottilie E A Winzeler Using in vitroevolution and whole genome analysis to discover nextgeneration targets for antimalarial drug discovery ACS InfectDis 4 301ndash314 (2018) doi 101021acsinfecdis7b00276pmid 29451780

23 V Patel et al Identification and characterization of smallmolecule inhibitors of Plasmodium falciparum dihydroorotatedehydrogenase J Biol Chem 283 35078ndash35085 (2008)doi 101074jbcM804990200 pmid 18842591

24 N A Malmquist R Gujjar P K Rathod M A Phillips Analysisof flavin oxidation and electron-transfer inhibition inPlasmodium falciparum dihydroorotate dehydrogenaseBiochemistry 47 2466ndash2475 (2008) doi 101021bi702218cpmid 18225919

25 D E Hurt J Widom J Clardy Structure of Plasmodiumfalciparum dihydroorotate dehydrogenase with a boundinhibitor Acta Crystallogr D Biol Crystallogr 62 312ndash323(2006) doi 101107S0907444905042642 pmid 16510978

26 X Deng et al Fluorine modulates species selectivity in thetriazolopyrimidine class of Plasmodium falciparumdihydroorotate dehydrogenase inhibitors J Med Chem 575381ndash5394 (2014) doi 101021jm500481t pmid 24801997

27 L S Ross et al In vitro resistance selections forPlasmodium falciparum dihydroorotate dehydrogenaseinhibitors give mutants with multiple point mutationsin the drug-binding site and altered growth J Biol Chem289 17980ndash17995 (2014) doi 101074jbcM114558353pmid 24782313

28 T G Nam et al A chemical genomic analysis of decoquinatea Plasmodium falciparum cytochrome b inhibitor

ACS Chem Biol 6 1214ndash1222 (2011) doi 101021cb200105dpmid 21866942

29 V C Corey et al A broad analysis of resistance developmentin the malaria parasite Nat Commun 7 11901 (2016)doi 101038ncomms11901 pmid 27301419

30 J N Burrows R H van Huijsduijnen J J Moumlhrle C OeuvrayT N C Wells Designing the next generation of medicinesfor malaria control and eradication Malar J 12 187 (2013)doi 1011861475-2875-12-187 pmid 23742293

31 B K Yeung et al Spirotetrahydro beta-carbolines(spiroindolones) A new class of potent and orally efficaciouscompounds for the treatment of malaria J Med Chem 535155ndash5164 (2010) doi 101021jm100410f pmid 20568778

32 A Ouattara et al Designing malaria vaccines to circumventantigen variability Vaccine 33 7506ndash7512 (2015)doi 101016jvaccine201509110 pmid 26475447

33 B Mordmuumlller et al Sterile protection against human malariaby chemoattenuated PfSPZ vaccine Nature 542 445ndash449(2017) doi 101038nature21060 pmid 28199305

34 R J Landovitz R Kofron M McCauley The promise andpitfalls of long-acting injectable agents for HIV preventionCurr Opin HIV AIDS 11 122ndash128 (2016) doi 101097COH0000000000000219 pmid 26633643

35 S Iwata et al Complete structure of the 11-subunit bovinemitochondrial cytochrome bc1 complex Science 281 64ndash71(1998) doi 101126science281537364 pmid 9651245

36 D Birth W C Kao C Hunte Structural analysis ofatovaquone-inhibited cytochrome bc1 complex reveals themolecular basis of antimalarial drug action Nat Commun 54029 (2014) doi 101038ncomms5029 pmid 24893593

ACKNOWLEDGMENTS

We thank A Rodriguez and the New York University for providingP bergheindashinfected mosquitoes the Pennsylvania State UniversityMetabolomics Core Facility (University Park Pennsylvania) forcritical analytical expertise and H Painter for valuable input on themetabolomics experimental setup We also thank G LaMonte forassistance with parasite culture and P Hinkson (Harvard ChanSchool) for technical support and we thank S Mikolajczak(Center for Infectious Disease Research) for the PvUIS4 antibodiesDNA sequencing was performed at the University of CaliforniaSan Diego (UCSD) with support from the Institute of GenomicMedicine Core The red blood cells used in this work were sourcedethically and their research use was in accord with the terms ofthe informed consents The compound library was acquired fromCharles River We thank the Shoklo Malaria Research Unit staffinvolved in this study especially P Kittiphanakun S N HselN Keereepitoon and S Saithanmettajit for their help in mosquitorearing and P vivax infection Funding EAW is supported bygrants from the NIH (5R01AI090141 R01AI103058 andP50GM085764) and by grants from the Bill amp Melinda GatesFoundation (OPP1086217 and OPP1141300) as well as fromMedicines for Malaria Venture (MMV) MLL received support fromBill amp Melinda Gates Foundation Phase II Grand Challenges(OPP1119049) This work was also funded in part by National

Institutes of Health grant R01AI093716 supporting DFW AKLand TSK DEK and SPM are supported by grants fromthe Bill amp Melinda Gates Foundation (OPP1023601) the GeorgiaResearch Alliance and the Medicines for Malaria Venture (RD150022) The Human Subjects protocol for the P vivax study wasapproved by the Oxford Tropical Medicine Ethical Committee OxfordUniversity England (TMEC 14-016 and OxTREC 40-14) The ShokloMalaria Research Unit is part of the Mahidol Oxford Research Unitsupported by the Wellcome Trust of Great Britain The HumanSubjects protocol for the UCSD P falciparum study was approved byThe Scripps Research Institute Institutional Review Board as partof the Normal Blood Donor Service Author contributions EAWdesigned experiments wrote the manuscript and analyzed dataYAK performed exoerythrocytic luciferase ABS and toxicity dose-response assays analyzed data and wrote portions of themanuscript MA and CMT performed ABS assays and analyzeddata SM MA ME CG KG DP MI and CWM performedprimary and secondary Pbluc activity screens YAK KE JC andBA performed secondary exo-erythrocytic form activity screensCL analyzed data YZ KC and YZ performed cheminformaticsanalysis MLL and EO performed the metabolomics assay andanalyzed the data SPM AJC VC FN and DEK performedP vivax studies MR BC and JB sourced compounds anddesigned experiments DFW AKL FJG and FNS performedmitochondrial inhibitor studies DFW AKL JCJR TSK MVand DAF performed in vitro evolution experiments and analyzedthe data ML analyzed sequence data and performed ScDHODHexperiments and SO sourced compounds analyzed data andwrote the manuscript Competing interests EAW is on thescientific advisory board of the Tres Cantos Open Lab FoundationData and materials availability DNA sequences have beendeposited in the shortread sequence archive (wwwncbinlmnihgovsra) under accession no SRP134381 Metabolomics datahas been deposited in the NIH Metabolomics Workbench underaccession no PR000719 All other data are available in themanuscript or the supplementary materials This work is licensedunder a Creative Commons Attribution 40 International (CC BY40) license which permits unrestricted use distribution andreproduction in any medium provided the original work is properlycited To view a copy of this license visit httpcreativecommonsorglicensesby40 This license does not apply to figuresphotosartwork or other content included in the article that iscredited to a third party obtain authorization from the rightsholder before using such material

SUPPLEMENTARY MATERIALS

wwwsciencemagorgcontent3626419eaat9446supplDC1Materials and MethodsFigs S1 to S4Tables S1 to S8References (37ndash46)Data Files S1 to S7

26 April 2018 accepted 18 October 2018101126scienceaat9446

Antonova-Koch et al Science 362 eaat9446 (2018) 7 December 2018 8 of 8

RESEARCH | RESEARCH ARTICLEon O

ctober 14 2020

httpsciencesciencemagorg

Dow

nloaded from

Open-source discovery of chemical leads for next-generation chemoprotective antimalarials

David A Fidock Dyann F Wirth Jeremy Burrows Brice Campo and Elizabeth A WinzelerMelanie Rouillier Dionicio Siegel Franccedilois Nosten Dennis E Kyle Francisco-Javier Gamo Yingyao Zhou Manuel LlinaacutesFernando Neria Serrano Korina Eribez Cullin McLean Taggard Andrea L Cheung Christie Lincoln Biniam Ambachew Zhong Kaisheng Chen Victor Chaumeau Amy J Conway Case W McNamara Maureen Ibanez Kerstin GagaringSakata-Kato Manu Vanaerschot Edward Owen Juan Carlos Jado Steven P Maher Jaeson Calla David Plouffe Yang Yevgeniya Antonova-Koch Stephan Meister Matthew Abraham Madeline R Luth Sabine Ottilie Amanda K Lukens Tomoyo

DOI 101126scienceaat9446 (6419) eaat9446362Science

this issue p eaat9446 see also p 1112Sciencedevelopmentmodes of action but solely targeting the parasites liver stages emerged as promising drug leads for further

mitochondrial electron transport were identified Excitingly several new scaffolds with as-yet-unknownPlasmodiumThree rounds of increasingly stringent screening were used From this regime several chemotypes that inhibit

Plasmodium bergheiand Goldberg) To do so they devised a luciferase-reporter drug screen for the rodent parasite investigated the possibilities of drugs against liver-stage parasites (see the Perspective by Phillipset alAntonova-Koch

parasite As an alternative strategyPlasmodiumfrontline combination therapies which target the blood stages of the Malaria parasites are evolutionarily prepared to resist drug attack Resistance is emerging to even the latest

A path to tackle liver-stage parasites

ARTICLE TOOLS httpsciencesciencemagorgcontent3626419eaat9446

MATERIALSSUPPLEMENTARY httpsciencesciencemagorgcontentsuppl201812053626419eaat9446DC1

CONTENTRELATED

httpstmsciencemagorgcontentscitransmed10460eaap9128fullhttpstmsciencemagorgcontentscitransmed10431eaan6007fullhttpstmsciencemagorgcontentscitransmed10447eaar3619fullhttpstmsciencemagorgcontentscitransmed10463eaat6176fullhttpsciencesciencemagorgcontentsci36264191112full

REFERENCES

httpsciencesciencemagorgcontent3626419eaat9446BIBLThis article cites 45 articles 12 of which you can access for free

PERMISSIONS httpwwwsciencemagorghelpreprints-and-permissions

Terms of ServiceUse of this article is subject to the

is a registered trademark of AAASScienceScience 1200 New York Avenue NW Washington DC 20005 The title (print ISSN 0036-8075 online ISSN 1095-9203) is published by the American Association for the Advancement ofScience

Science No claim to original US Government WorksCopyright copy 2018 The Authors some rights reserved exclusive licensee American Association for the Advancement of

on October 14 2020

httpsciencesciencem

agorgD

ownloaded from