Artemisinin activity-based probes identify multiplemolecular targets within the asexual stage of themalaria parasites Plasmodium falciparum 3D7Hanafy M. Ismaila,b, Victoria Bartonc, Matthew Phanchanaa, Sitthivut Charoensutthivarakulc, Michael H. L. Wongb,Janet Hemingwayb,1, Giancarlo A. Biaginia, Paul M. O’Neillc, and Stephen A. Warda,1

aResearch Centre for Drugs and Diagnostics, Department of Parasitology, Liverpool School of Tropical Medicine, Liverpool L3 5QA, United Kingdom; bVectorBiology, Liverpool School of Tropical Medicine, Liverpool L3 5QA, United Kingdom; and cDepartment of Chemistry, University of Liverpool, Liverpool L697ZD, United Kingdom

Contributed by Janet Hemingway, January 12, 2016 (sent for review December 24, 2015; reviewed by Kelly Chibale, Gary Posner, and Donatella Taramelli)

The artemisinin (ART)-based antimalarials have contributed signif-icantly to reducing global malaria deaths over the past decade, butwe still do not know how they kill parasites. To gain greaterinsight into the potential mechanisms of ART drug action, wedeveloped a suite of ART activity-based protein profiling probesto identify parasite protein drug targets in situ. Probes weredesigned to retain biological activity and alkylate the moleculartarget(s) of Plasmodium falciparum 3D7 parasites in situ. Proteinstagged with the ART probe can then be isolated using click chemistrybefore identification by liquid chromatography–MS/MS. Using theseprobes, we define an ART proteome that shows alkylated targets inthe glycolytic, hemoglobin degradation, antioxidant defense, andprotein synthesis pathways, processes essential for parasite survival.This work reveals the pleiotropic nature of the biological functionstargeted by this important class of antimalarial drugs.

artemisinin | antimalarial | bioactivation | chemical proteomics |molecular targets

Malaria is a global health problem with 214 million newcases of malaria and 438,000 deaths reported in 2015,

mostly in sub-Saharan Africa (1). The endoperoxide class ofantimalarial drugs, such as artemisinin (ART), is the first line ofdefense against malaria infection against a backdrop of multi-drug-resistant parasites (2) and lack of effective vaccines (3, 4).Given the effectiveness of the ART class, the question arises:how do these drugs kill parasites? A suggested mechanism ofaction involves the cleavage of the endoperoxide bridge by asource of Fe2+ or heme. This cleavage results in the formation ofoxyradicals that rearrange into primary or secondary carbon-cen-tered radicals. These radicals have been proposed to alkylateparasite proteins that somehow result in the death of the parasite(5). However, this proposal remains a subject of intense debate(6, 7), while these alkylated proteins are yet to be formallyidentified. So far, the proposed targets of ART action include aPfATP6 enzyme, the Plasmodium falciparum ortholog of mam-malian sarcoendoplasmic reticulum Ca21-ATPases (SERCAs)(5), translational controlled tumor protein, and heme (5). Addi-tionally, Haynes et al. (8) proposed that ART may act by impairingparasite redox homeostasis as a consequence of an interaction be-tween the drug and flavin adenine dinucleotide (FADH) and/orother parasite flavoenzymes in the parasite, leading to the genera-tion of reactive oxygen species (ROS). New approaches are re-quired for definitive identification of ART molecular targets. Thisinsight into the drug activation-dependent mechanism of action willbe invaluable in the target-led development of more potent drugswith the potential to circumvent the emergence of resistance tocurrent first-line ART-based therapies. The goal of this study was toidentify ART-targeted proteins and their interacting partners inP. falciparum. We recently adopted a proteomic approach de-veloped by Speers and Cravatt (9) to synthesize a suite of pyre-throid activity-based protein profiling probes (ABPPs) (10). Using

alkyne/azide-coupling partners through “click chemistry,” weidentified several cytochrome P450 enzymes that metabolizeddeltamethrin in rat liver microsomes (10). More recently, achemical proteomic approach was developed to identify parasiteproteins targeted by an albitiazolium antimalarial drug candidatein situ using a photoactivation cross-linking approach (11).However, this generic approach can introduce significant pro-miscuity in the proteins tagged based on the intracompartmentaldistribution of drug independent of actual mechanisms.Here, we introduced the design and synthesis of click chem-

istry-compatible activity-based probes incorporating the endo-peroxide scaffold of ART as a warhead to alkylate and identifiedthe ART molecular target(s) in asexual stages of the malariaparasite (Fig. 1). A major advantage of this strategy is that thereporter tags are introduced under “click” reaction conditionsperformed after the drug has achieved its biological effects, en-abling purification, identification, and quantification of alkylatedparasite’s proteins and their interacting partners as shown in Fig.1B. To avoid nonspecific probe-dependent tagging, a commonlimitation of these approaches, we generated the respective “con-trol” nonperoxide partners to improve the specificity and biologicalrelevance of our resultant tagged protein list.

Significance

The mechanism of action of the artemisinin (ART) class of an-timalarial drugs, the most important antimalarial drug class inuse today, remains controversial, despite more than three de-cades of intensive research. We have developed an unbiasedchemical proteomic approach using a suite of ART activity-based protein profiling probes to identify proteins within themalaria parasite that are alkylated by ART, including proteinsinvolved in glycolysis, hemoglobin metabolism, and redox de-fense. The data point to a pleiotropic mechanism of drug actionfor this class and offer a strategy for investigating resistancemechanisms to ART-based drugs as well as mechanisms of ac-tion of other endoperoxide-based drugs.

Author contributions: H.M.I., P.M.O., and S.A.W. designed research; H.M.I. and M.P. per-formed research; V.B., S.C., M.H.L.W., P.M.O., and S.A.W. contributed new reagents/analytic tools; H.M.I., V.B., M.P., J.H., and G.A.B. analyzed data; H.M.I. and S.A.W. wrotethe paper; H.M.I. contributed to the initial concept, carried out the initial biological studies,validated the methodology, and wrote the initial draft of the manuscript; V.B. and S.C.were responsible for the generation of probe molecules; M.P. carried out confirmatorybiological studies independent of H.M.I.; and P.M.O. and S.A.W. conceived of the originalconcept and were responsible for the biological materials and chemistry, respectively.

Reviewers: K.C., University of Cape Town; G.P., John Hopkins University; and D.T., Universitadi Milano.

The authors declare no conflict of interest.

Freely available online through the PNAS open access option.1To whom correspondence may be addressed. Email: janet.hemingway@lstmed.ac.uk orsteve.ward@lstmed.ac.uk.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1600459113/-/DCSupplemental.

2080–2085 | PNAS | February 23, 2016 | vol. 113 | no. 8 www.pnas.org/cgi/doi/10.1073/pnas.1600459113

Dow

nloa

ded

by g

uest

on

Oct

ober

8, 2

020

Results and DiscussionRationale of ART Activity-Based Protein Profiling Probes Design andSynthesis. We synthesized the click probes using techniques op-timized and developed in house with the alkyne/azide introducedby an optimized peptide coupling protocol (more details are in SIText). We varied click handles to expand the utility of ART-ABPPsand compare the efficiency of two distinctive tagging chemistries(i.e., copper-catalyzed and bioorthogonal copper-free click chem-istry). Previous work by Meshnick and coworkers (12) showed thatsix uncharacterized malaria proteins bands were labeled withsemisynthetic derivatives of ART (e.g., [3H]-dihydroartemisinin)at pharmacologically relevant concentrations of the drug. Im-portantly, labeling was not shown using uninfected erythrocytesor when infected cells were treated with the inactive analog,deoxyether (12). Our strategy allows definitive identification oftagged proteins by comparison of active probe alkylation productswith those of their nonperoxidic equivalent controls (synthesizedas detailed in SI Text). Earlier work suggests that long peptidelinkers can introduce cytotoxicity (13). To avoid this issue, pep-tide linkers of maximum three carbons were used for our alkyneand azide probes.

Antimalarial Activity of ART-ABPPs. The antimalarial activity of theprobes and controls was determined in vitro against P. falciparum3D7 parasites (SI Text). The azide/alkyne function did not sig-nificantly affect antimalarial activity against 3D7 (Fig. 1A). TheIC50 values for probe 1 (P1) and probe 2 (P2) were 43.86 (SD =0.441 nM; n = 3) and 30.26 nM (SD = 0.314 nM; n = 3), re-spectively, compared with the IC50 of ART of 14 nM (SD = 0.41nM; n = 3). Importantly, control probes CP1 and CP2 were in-active in agreement with expectations (12, 14). This result sup-ports the essentiality of the endoperoxide bridge and validatesour probe plus control pair strategy.

Parasite Protein Labeling and Identification. As a next step, wemoved to the direct identification of ART molecular target(s) insitu as explained in Methods and Fig. 1B. The Mascot searchalgorithm was used to identify proteins from the resultant pep-tides identified by liquid chromatography (LC)–MS/MS, and theexponentially modified protein abundance index (emPAI) wasused to provide semiquantitation of protein abundance (15). Heatmap analysis (Fig. 2) shows the essentiality of the endoperoxide

in protein alkylation. In general, many essential proteins wereidentified as ART targets with P1 and P2 (Figs. 2 and 3). Theseresults are further supported by 1D gel fluorescence analysis(Fig. S1) for proteomes treated with a trifunctional azido-biotin-rhodamine tag (instead of the biotin azide tag) and subjected toprotein purification followed by in-gel fluorescence visualization.With this probe, significant protein labeling was apparent withthe active probe P1, whereas no labeling was evident with itsalkyne deoxy-ether negative control analog. This result again rein-forces the importance of the endoperoxide function for bioacti-vation (Fig. S1). Using on-bead trypsin digestion of capturedproteins with the biotin azide reporter followed by LC-MS/MSidentification confidently reported 58 proteins using P1; how-ever, four proteins were detected at very low abundance in two offive replicates in the positive control treatment (CP1) (Fig. 2 andFig. S2A). Multiple t test analysis confirmed that the levels of theseproteins tagged by the active probe were significantly higher than inthe control experiments (Fig. S2A). This result shows the impor-tance of having a positive control to avoid identification of proteinsthat bind nonspecifically to the streptavidin column. When DMSOsolvent was used as a negative control, nine proteins at very low,insignificant levels were detected in two of five replicates (Fig. 2 andFig. S2B). These observations emphasize the need to control fornonspecific binding in these complex proteomic experiments.Studies performed with copper (I) salts and copper complexes

have shown the potential for ART activation and rearrangement,similar to that generated by iron (16) under these conditions. ARTactivation based on copper was considered an unlikely problemduring the click reactions used in this study because of the mul-tiple washing steps involved and the use of a reducing environ-ment [DTT and Tris(2-carboxyethyl)phosphine] during proteinextraction and click reactions, conditions known to block antima-larial action of the endoperoxides (17, 18). Nevertheless, to de-finitively rule out this possibility and validate the results morestringently, we have also introduced the bioorthogonal copper-freeclick methodology by preparation of the azide analog of the alkyneprobe (P2) (Fig. 1). After the azide probe is introduced intotarget biomolecules after peroxide activation, the azide can betagged with reporter using several selective reactions (19). Here,we used strain-promoted [3 + 2] azide-alkyne cycloaddition toverify the results obtained from copper-catalyzed azide-alkynecycloaddition (CuAAC) with P1. The critical tagging reagent, a

Fig. 1. Rational design of the ART-ABPPs. (A) Conversion of ART to ART-ABPPs involves the addition of a clickable handle (i.e., an alkyne or azide to the ART drugpharmacophore by the peptide-coupling method illustrated in SI Text). The structures of the alkyne (P1) and azide (P2) probes and respective inactive deoxy controlsCP1 and CP2 with in vitro IC50 values are presented. (B) General workflow of copper-catalyzed and copper-free click chemistry approaches used in the identification ofalkylated proteins after in situ treatment of P. falciparum parasite with alkyne and azide ART-ABPPs. The azide- and alkyne-modified proteins are tagged with biotinazide and biotin dibenzocyclooctyne (Biotin-DIBO), respectively, via click reactions followed by affinity purification tandem with LC-MS/MS for protein identification.

Ismail et al. PNAS | February 23, 2016 | vol. 113 | no. 8 | 2081

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

8, 2

020

substituted biotin cyclooctyne, possesses ring strain and elec-tron-withdrawing substituents that together promote the [3 + 2]dipolar cycloaddition with azides installed into the P2 moleculeas illustrated in Fig. 1B (19). The Cu-free click reaction followscomparable kinetics to the CuAAC click reaction and proceeds

within minutes (19, 20). Both P1 and P2 exhibited similar la-beling patterns as indicated by pulldown experiments, including42 identical proteins (Fig. 3A). However, the pulldown exper-iment with P2 exhibited higher sensitivity compared with P1(Fig. 2). The higher sensitivity could be attributed to the higherefficiency of the strain-promoted [3 + 2] azide-alkyne cyclo-addition reaction compared with the CuAAC reaction in aqueousmedia (19, 20). More importantly, samples treated with the non-peroxidic equivalent of P2 (CP2) did not identify any proteins.Moreover, the comparison between samples treated in situ withP2 vs. cellular homogenates did not show any significant qualita-tive differences in the tagged protein profiles. Additionally, cellularhomogenates exposed to P2 with the iron-chelating agent desfer-oxamine (DFO) (100 μM) showed that ART–protein alkylationwas significantly reduced but not completely abolished by chelation(Fig. S3 and Table S1). These data may support a role for anonferrous activation process. Alternatively, although 100 μMDFO was previously shown to inhibit the antimalarial activityof this class using live parasite (14), incomplete iron chelationremains a possibility under this experimental condition usingprocessed parasite protein homogenate.Free sulfhydryl groups (thiols) of cysteine residues in the

proteins of Plasmodium parasites are sensitive to oxidation andalkylation by ART-derived radicals, generating cysteine adducts(21, 22). Kehr et al. (23) showed that the free thiols in cysteineare reversibly glutathionylated, forming blended protein gluta-thione disulfides. This posttranslational modification is believedto function primarily as a redox-sensitive switch to facilitate re-dox regulation and signal transduction (23). Metaanalysis on theprotein list identified by ART-ABPPs (P1 and P2) looking for aglutathione binding motif (Fig. 3B) indicated a high positive cor-relation (χ2 = 230.1 with a P value < 0.0001) as shown in Fig. 2.This positive correlation may be explained by the availability of freethiol groups in cysteine residues acting as targets for ART–proteinalkylation. In particular, thiol alkylation may directly contribute tothe observed specific inhibition of cysteine proteases and asparticproteases by the endoperoxides that results in decreased hemo-globin degradation in the parasite’s food vacuole (24).

Interaction of ART with the P. falciparum Hemoglobin DigestionPathway. In this study, ART-ABPPs labeled a substantial num-ber of proteases from the parasite’s digestive food vacuole (Fig. 2,Fig. S4, and Table S2). Plasmepsin-2, in particular, was identifiedin the high confidence list of the ART proteome. Together withplasmepsin-1, these two proteins belong to the aspartic proteasesthat coordinate with cysteine proteases (falcipain-2 and falcipain-3)in the process of hemoglobin degradation in the parasite’s foodvacuole (25–27) and are considered good drug targets (28). In-hibition of vacuolar digestion could contribute to the pleiotropicaction of the ART drug class. Neither falcipain-2 nor falcipain-3,considered to be the activators of plasmepsins (29), was targeted bythe ART-ABPPs, which may reflect the steric effect of the ARTstructure in selective binding of aspartic proteases involved in

Fig. 2. Heat map of the entire proteomic dataset identified by the alkyneand azide ART-ABPPs. Heat map analysis was carried out by plotting theaverage exponentially modified protein abundance index values for eachprotein. The data were acquired from multiple independent experiments foreach probe [P1, n = 5; CP1, n = 5; DMSO, n = 5; P2, n = 2; CP2, n = 2; P2*(cellular homogenate)]. Each independent replicate was injected into theLC-MS/MS four times to improve the accuracy of protein identification. Theheat map plot was clustered into four categories from high confidence tonoise according to the hit frequency in all replicates. Protein hits found inseven to nine independent replicates were denoted as being of high confi-dence, hits seen in four to six replicates were denoted as medium confi-dence, three hits were denoted as low confidence, and hits seen in only oneor two replicates were considered as noise. Proteins were manually searchedfor the presence (+) and absence (−) of a glutathione (GSH) binding motifaccording to data published by Kehr et al. (23). Complete datasets of ARTproteomes are illustrated in Dataset S1. DMSO, dimethyl sulfoxide treat-ment; MW, protein molecular weight in kilodalton; ORF, open readingframe names.

A B

Fig. 3. Overview of ART-ABPP–tagged proteins. (A) Venn diagram showingoverlap between proteins identified with the alkyne (P1) and azide (P2) ARTprobes. (B) Trend analysis between the confidence in protein identificationas described in Fig. 2 and the proportion of those proteins carrying a glu-tathione (GSH) binding motif (χ2 = 230.1 with a P value < 0.0001).

2082 | www.pnas.org/cgi/doi/10.1073/pnas.1600459113 Ismail et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

8, 2

020

hemoglobin digestion. It also suggests that the targeting of proteinsis not promiscuous. However, neither plasmepsin-2 nor M1-familyalanyl aminopeptidase, which identified with high confidence byART-ABPPs, was significantly affected by the iron-chelating agent(DFO), suggesting that the alkylation and inhibition of proteaseswith ART may not be solely dependent on free Fe+2 (Fig. S3). Thisobservation requires additional investigation. The observation thatART-ABPPs target multiple components of this degradationpathway is credible evidence of ART activation and may implicatedisturbance in the parasites feeding process as an important aspectof drug action (Fig. S4) as indicated by work using alternative ap-proaches (7, 30–32).

Interaction of ART with P. falciparum Antioxidant Defense Systems.Hemoglobin digestion in the malaria parasite is a highly dynamiccatabolic process. Hemoglobin digestion leaves the parasite underintense oxidative stress by generating active redox byproducts, in-cluding heme and a superoxide anion radical (O• 2), with the latterbeing rapidly converted to H2O2 (33). To cope with such highlycytotoxic processes, the parasite recruits an efficient enzymaticantioxidant defense system that includes glutathione and manythioredoxin-dependent proteins. Of particular interest in this studyis the labeling of ornithine aminotransferase (OAT), a thioredoxin-dependent protein identified with ART-ABPPs in high confidenceand high abundance (Fig. 2 and Table S2). PfOAT is essential inthe coordination of ornithine homeostasis, polyamine synthesis,proline synthesis, and mitotic cell division (34). In malaria parasites,PfTrx1 regulates PfOAT redox activity by interaction with twohighly conserved cysteine residues (Cys154 and Cys163), which donot exist in any other species and may interfere with substratebinding (34). PfOAT labeling was significantly reduced after ironchelation (Fig. S3), suggesting the supportive role of free iron indrug activation and leading to irreversible binding to PfOAT.ART itself can generate ROS in the parasite through endoper-

oxide bridge activation (8, 35), and free Fe+2 can enhance thegeneration of ROS by the Fenton process (35). Therefore, it hasbeen presumed that ART-activated ROSmay lead to parasite deathby overwhelming the parasite’s antioxidant defense systems (36).Identification of stress-related proteins (Table S2) by ART-ABPPssupports the hypothesis that ART is involved in the activation andgeneration of ROS in vivo. Apart from Actin 1, none of theseproteins were significantly affected by iron chelation under thecondition used here (Fig. S3).

Interaction of ART with the P. falciparum Glycolysis Pathway. Pre-vious work by our group showed that artemether, the methylderivative of ART, negatively alters some key glycolytic enzymeswhile increasing the expression of stress response proteins afterpharmacologically relevant drug exposure in the P. falciparum(37). Increased levels of stress-associated proteins coupled with adecrease in glycolytic enzymes suggest a mechanism where theparasite designates to protect its machinery for generating itsenergy supply from disruption by endoperoxides. This hypothesisis further supported by the recent study, where ART-resistantparasites exhibited a slower growth rate through the early ringstage of the intraerythrocytic developmental cycle (38). A credible

explanation of this phenomenon, observed in ART-resistant par-asites, is that slowing down growth at ring stage by closing downaspects of energy production is a protective mechanism, whereasthe parasite experiences ART-induced stress and possible proteindamage. Normal growth resumes after up-regulation of the un-folded protein response pathways that are capable of resolvingprotein damage caused by ART (38). Interestingly, in this study, of11 enzymes participating in glycolysis, 7 enzymes were labeled anddetected with ART-ABPPs (Table S2). The disruption of themicroassembly of glycolysis in the parasite by ART should lead tothe parasite death. In support of this observation, studies inSchistosoma japonicum-infected mice exposed to artemetheralso showed an effect of drug treatment on important glycolyticenzymes in the parasite (39, 40). In general, these findings areconsistent with the vital and essential role that glycolytic en-zymes play in supporting rapid growth and proliferation, similarto that seen in other rapidly proliferating cells, such as cancercells (41, 42). The high glycolytic flux of intraerythrocytic de-velopmental cycle maintains rate-limiting glycolytic intermedi-ates to support other pathways (e.g., nucleotide and lipidbiosynthesis) (Fig. S4) (41).Overall, apart from glucose-6-phosphate isomerase, none of the

glycolytic proteomes identified by ART-ABPPs were significantlyaffected by iron chelation using cellular homogenate (Fig. S3).

Interaction of ART with P. falciparum Nucleic Acid and ProteinBiosynthesis Pathways. Plasmodium parasites cannot synthesize pu-rine nucleotides de novo and instead, rely on the purine salvagepathway that is essential for parasite growth and survival (43) andhas been exploited as a drug target. ART-ABPPs identified fiveenzymes (Table S2) involved in purine and pyrimidines synthesis(nucleic acid biosynthesis). Many of the enzymes involved in Plas-modium nucleic acid biosynthesis differ from those of their humanhosts. Accordingly, nucleic acid biosynthesis pathways have longbeen considered exploitable targets for novel antimalarial drugdesign (43–46). For example, P. falciparum bifunctional dihy-drofolate reductase-thymidylate synthase mediates the conversionof dihydrofolic acid (dihydrofolate; vitamin B9) to tetrahydrofolicacid by dihydrofolate (44). Tetrahydrofolate is needed to make bothpurines and pyrimidines, the building blocks of DNA and RNA, andsome of our most successful antimalarials, such as pyrimethamineand proguanil, target this enzyme, resulting in parasite death (44,47). Furthermore, P. falciparum hypoxanthine-guanine-xanthinephosphoribosyltransferase, an ART-ABPPs target identified withhigh confidence, is also central in purine salvage (46). Labeling ofthe P. falciparum hypoxanthine-guanine-xanthine phosphoribosyl-transferase was significantly reduced after iron chelation (Fig. S3),suggesting a role of chelatable iron in ART activation leading toalkylation of the enzyme. ART-ABPPs also identified many ri-bosomal proteins (Table S2), suggesting that protein synthesis isan additional target for ART.

Interaction of ART with P. falciparum Chaperone Proteins and theCytoskeleton. ART-ABPPs identified several chaperone proteins(Table S2). The majority of chaperonins were initially identified asheat shock proteins expressed in parasites in response to elevated

Fig. 4. Gene Ontology (GO) analysis for biologicalpathways identified from ART proteome. (A) Blackbars represent the numbers of proteins identifiedwithin each biological process (pathway) using theSTITCH 4 web tool (stitch.embl.de). (B) Multiple test-ing corrected P values for enriched GO categories.The confidence view of ART–protein and protein–protein interactions networks built up using the STITCH4 web tool is illustrated in Fig. S4. A completedataset of ART interactome is in Datasets S2 and S3.

Ismail et al. PNAS | February 23, 2016 | vol. 113 | no. 8 | 2083

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

8, 2

020

temperatures, a prominent symptom of the disease, and/or othercellular stresses. In general, molecular chaperones are involved inmany cellular functions, including helping newly synthesized pro-teins to fold, protein trafficking, and signal transduction (48).TCP-1 and cytoskeletal proteins α-tubulin, β-tubulin, and actin 1were all labeled by ART-ABPPs. The fact that actin and tubulinfolding and dimerization are chaperoned by the TCP-1 chaperoninsystem gives rise to suggest a potential link between ART, theparasite cell structure, protein trafficking systems, and signaltransduction.

Interaction of ART with Transport Proteins in P. falciparum. ART-ABPPs identified the two subunits of the V-type H+-ATPase(subunits A and B) at various levels of confidence from high tomedium, respectively (Table S2). In the parasite, a V-typeATPase generates a proton motive force across the parasiteplasma membrane as well as across the digestive food vacuole(49, 50). It has previously been shown that ART results in rapiddepolarization of the parasite plasma membrane (51). Althoughthe depolarization of the plasma membrane was previouslyshown to be partially attributed to lipid peroxidation, the ART-ABPPs data presented here could be interpreted as showing thepotential for a direct inhibition of the V-type ATPase activity byART. Structural similarity between ARTs and thapsigargin, aninhibitor of SERCA (PfATP6), led to a hypothesis that PfATP6was a target for ART (5). This proposal received additionalsupport from a study connecting ART resistance in P. falciparumfield isolates from French Guiana with the S769N mutation inPfATP6 (52). The functional importance of PfATP6 in ARTaction and resistance remains controversial; however, PfATP6was detected by the alkyne version of ART-ABPPs (P1), albeit atlow yield and low confidence. In addition to metabolic pathways,ART-ABPPs identified nine proteins related to parasite–hostcell invasion and the host immune responses (Table S2). Thefunctional relevance of these as targets remains to be tested.

ART and Resistance Mechanisms in P. falciparum. Several genes havebeen associated with ART resistance mechanisms, including thechloroquine resistance gene (Pfcrt), the multidrug resistancepump (Pfmdr1), and the K13 propeller domain of the kelch-likeprotein (2, 38, 53). This study was not designed to probe re-sistance, and only a single parasite isolate has been investigated;however, the gene products of Pfcrt and Pfmdr1 (Table S3), and aprotein from the proteasomal system, Cdc48 protein (but notK13), were labeled with ART-ABPP. These findings indicate thepotential for the use of ART-ABPPs in investigating resistanceand tolerance mechanisms using appropriate parasite isolateswith well-defined resistance phenotypes.

ART Interactome. We used a bioinformatics interaction networkanalysis approach using STICH 4 software (54) to build up andpredict functional ART–protein association networks and identifymajor pathways targeted by the drug based on compiled availableexperimental evidence, database information, and the literature.Analysis of the data generated in silico connectivity for ART,with 67 proteins collectively identified with P1 and P2 termedthe ART interactome shown in Fig. S5. As predicted, the pro-teome experimentally identified with ART-ABPPs revealed astrong network association of functionally important proteins(Fig. S5). The connectivity analysis confirmed that the glycolyticpathway is a primary interacting target for ART with a P valueof 1 × 10−9 (Fig. 4). Additional pathways were identified asbeing targets for ART interaction, including the biosynthesis ofsecondary metabolites, metabolic pathways, ribosomes, phag-osomes, etc. (Fig. 4). Additional analysis to identify the cellularcomponents targeted by the ART interactome using Cytoscapesoftware supports a role for ART acting akin to the “clusterbomb” hypothesis (55), with multiple cellular targets mainly butnot exclusively in the food vacuole and cytosol (Figs. S4 and S6).The important next step will be to establish which of these

interactions has functional relevance regarding parasite le-thality and resistance.

Conclusion and Future Perspective. Using a chemical proteomicapproach, we have identified with a high level of confidence keyproteins that are alkylated by ART; the biological functions ofthe tagged proteins are consistent with many of the presumedmechanisms proposed to explain ART drug action, in whichthe ART causes extensive damages to key proteins that have awide spectrum of cellular activity. Collectively, copper-based andcopper-free ART-ABPPs approaches present significant evi-dence in support of the argument that ART can target two im-portant sources of carbon/energy and amino acid supply (i.e.,glycolysis and hemoglobin digestion pathways) within the eryth-rocytic stage of infection. ART-ABPPs also provide additionalevidence that ART targets other essential parasite metabolicpathways, including DNA synthesis, protein synthesis, and lipidsynthesis. The application of ART-ABPPs to monitor the activityof protein target(s) as well as their abundance in different par-asite lines could have significant implications for understandingthe problem of ART resistance, especially combined with ge-nomic approaches. The data generated here are a significant stepforward in identifying the complete array of protein targets thatare susceptible to alkylation with ART. The next key step will beto define formally which of these alkylation reactions is func-tionally most relevant to antimalarial drug action and which maybe modified to confer drug resistance.

MethodsSynthesis of Endoperoxide Probes and Respective Controls. Endoperoxideprobes and corresponding deoxy ether analogs (controls) were synthesized asdetailed in SI Text.

Parasite Cultures and Drug Sensitivity Assays. P. falciparum 3D7 parasiteswere cultured following the method illustrated in SI Text. Drug activities ofART-ABPPs were measured using a standard fluorometric DNA bindingmethod as detailed in SI Text.

In Situ Parasite Treatment for Chemical Proteomic Pulldown Assays. Ten flasksof synchronized trophozoite-stage parasite cultures at 10–12% parasitemiaand 4% hematocrit (0.5 L culture) were treated with ART-ABPPs P1 and P2and their respective controls (100 μL 5 mM stock) to give a final concen-tration of 1 μM probe per treatment, 10-fold the IC90 (100 nM) at shortpulse time to ensure full inhibition of parasite growth and maximal alkyl-ation (53). All treatments were carried out with a minimum of two repli-cates per treatment (noted elsewhere). Probe-treated parasites weremaintained in culture at 37 °C in an atmosphere of 3% CO2, 4% O2, and93% N2 for 6 h, an incubation period that has already been shown to causeirreversible parasite lethality (53). An equivalent culture without drugprobe (five replicates) was treated with 100 μL DMSO as a control treatmentmaintained under identical culture conditions. Parasite proteins were se-quentially extracted using a modification of the protocol described byMolloy et al. (56) (SI Text).

ART-ABPPs Labeling and the Effect of Iron Chelation. Parasite protein extracts(cellular homogenates) were prepared from 1 L untreated in vitro culturedparasite as described earlier. Protein extracts were adjusted to 2 mg/mL inEppendorf tubes and treated with P2 at 10 μM in the presence or absence ofthe iron-chelating agent DFO at 100 μM for 1 h at 37 °C. Subsequent proteinprocessing was as described in SI Text.

Identification of ART-ABPPs Alkylated Proteins. Alkylated proteins from allexperiments were subjected to click chemistry reaction with relevant biotinreporter as detailed in SI Text. As the next step, biotin-tagged proteins wasaffinity-purified using streptavidin agarose beads, and the captured proteinswas identified using LC-MS/MS after bead trypsin digestion for the purifiedproteins as described in SI Text.

ACKNOWLEDGMENTS. This work was funded by the Wellcome Trustthrough its Strategic Support Fund award, Medical Research Council Confi-dence in Concept funding, and through Mahidol-Liverpool ChamlongHarinasuta PhD Scholarship.

2084 | www.pnas.org/cgi/doi/10.1073/pnas.1600459113 Ismail et al.

Dow

nloa

ded

by g

uest

on

Oct

ober

8, 2

020

1. WHO (2014) World Malaria Report 2014 (WHO, Geneva).2. Winzeler EA, Manary MJ (2014) Drug resistance genomics of the antimalarial drug

artemisinin. Genome Biol 15(11):544.3. RTS,S Clinical Trials Partnership (2015) Efficacy and safety of RTS,S/AS01 malaria vac-

cine with or without a booster dose in infants and children in Africa: Final results of aphase 3, individually randomised, controlled trial. Lancet 386(9988):31–45.

4. Rts SCTP; RTS,S Clinical Trials Partnership (2014) Efficacy and safety of the RTS,S/AS01malaria vaccine during 18 months after vaccination: A phase 3 randomized, con-trolled trial in children and young infants at 11 African sites. PLoS Med 11(7):e1001685.

5. Eckstein-Ludwig U, et al. (2003) Artemisinins target the SERCA of Plasmodium falci-parum. Nature 424(6951):957–961.

6. O’Neill PM, Barton VE, Ward SA (2010) The molecular mechanism of action of arte-misinin–the debate continues. Molecules 15(3):1705–1721.

7. Mercer AE, Copple IM, Maggs JL, O’Neill PM, Park BK (2011) The role of heme and themitochondrion in the chemical and molecular mechanisms of mammalian cell deathinduced by the artemisinin antimalarials. J Biol Chem 286(2):987–996.

8. Haynes RK, et al. (2010) Facile oxidation of leucomethylene blue and dihydroflavinsby artemisinins: Relationship with flavoenzyme function and antimalarial mechanismof action. ChemMedChem 5(8):1282–1299.

9. Speers AE, Cravatt BF (2004) Profiling enzyme activities in vivo using click chemistrymethods. Chem Biol 11(4):535–546.

10. Ismail HM, et al. (2013) Pyrethroid activity-based probes for profiling cytochromeP450 activities associated with insecticide interactions. Proc Natl Acad Sci USA 110(49):19766–19771.

11. Penarete-Vargas DM, et al. (2014) A chemical proteomics approach for the search ofpharmacological targets of the antimalarial clinical candidate albitiazolium in Plas-modium falciparum using photocrosslinking and click chemistry. PLoS One 9(12):e113918.

12. Asawamahasakda W, Ittarat I, Pu YM, Ziffer H, Meshnick SR (1994) Reaction ofantimalarial endoperoxides with specific parasite proteins. Antimicrob AgentsChemother 38(8):1854–1858.

13. Liu Y, Wong VK, Ko BC, WongMK, Che CM (2005) Synthesis and cytotoxicity studies ofartemisinin derivatives containing lipophilic alkyl carbon chains. Org Lett 7(8):1561–1564.

14. Stocks PA, et al. (2007) Evidence for a common non-heme chelatable-iron-dependentactivation mechanism for semisynthetic and synthetic endoperoxide antimalarialdrugs. Angew Chem Int Ed Engl 46(33):6278–6283.

15. Ishihama Y, et al. (2005) Exponentially modified protein abundance index (emPAI) forestimation of absolute protein amount in proteomics by the number of sequencedpeptides per protein. Mol Cell Proteomics 4(9):1265–1272.

16. Bousejra-El Garah F, Pitié M, Vendier L, Meunier B, Robert A (2009) Alkylating abilityof artemisinin after Cu(I)-induced activation. J Biol Inorg Chem 14(4):601–610.

17. Krungkrai SR, Yuthavong Y (1987) The antimalarial action on Plasmodium falciparumof qinghaosu and artesunate in combination with agents which modulate oxidantstress. Trans R Soc Trop Med Hyg 81(5):710–714.

18. Meshnick SR, et al. (1989) Activated oxygen mediates the antimalarial activity ofqinghaosu. Prog Clin Biol Res 313:95–104.

19. Sletten EM, Bertozzi CR (2009) Bioorthogonal chemistry: Fishing for selectivity in a seaof functionality. Angew Chem Int Ed Engl 48(38):6974–6998.

20. Baskin JM, et al. (2007) Copper-free click chemistry for dynamic in vivo imaging. ProcNatl Acad Sci USA 104(43):16793–16797.

21. WuWM, Chen YL, Zhai Z, Xiao SH, Wu YL (2003) Study on the mechanism of action ofartemether against schistosomes: The identification of cysteine adducts of both car-bon-centred free radicals derived from artemether. Bioorg Med Chem Lett 13(10):1645–1647.

22. Wu Y, Yue ZY, Wu YL (1999) Interaction of qinghaosu (artemisinin) with cysteinesulfhydryl mediated by traces of non-heme iron. Angew Chem Int Ed Engl 38(17):2580–2582.

23. Kehr S, et al. (2011) Protein S-glutathionylation in malaria parasites. Antioxid RedoxSignal 15(11):2855–2865.

24. Pandey AV, Tekwani BL, Singh RL, Chauhan VS (1999) Artemisinin, an endoperoxideantimalarial, disrupts the hemoglobin catabolism and heme detoxification systems inmalarial parasite. J Biol Chem 274(27):19383–19388.

25. Banerjee R, et al. (2002) Four plasmepsins are active in the Plasmodium falciparumfood vacuole, including a protease with an active-site histidine. Proc Natl Acad SciUSA 99(2):990–995.

26. Sijwali PS, Rosenthal PJ (2004) Gene disruption confirms a critical role for the cysteineprotease falcipain-2 in hemoglobin hydrolysis by Plasmodium falciparum. Proc NatlAcad Sci USA 101(13):4384–4389.

27. Klemba M, Gluzman I, Goldberg DE (2004) A Plasmodium falciparum dipeptidylaminopeptidase I participates in vacuolar hemoglobin degradation. J Biol Chem279(41):43000–43007.

28. Rosenthal PJ, McKerrow JH, Aikawa M, Nagasawa H, Leech JH (1988) A malarialcysteine proteinase is necessary for hemoglobin degradation by Plasmodium falci-parum. J Clin Invest 82(5):1560–1566.

29. Drew ME, et al. (2008) Plasmodium food vacuole plasmepsins are activated by falci-pains. J Biol Chem 283(19):12870–12876.

30. Klonis N, et al. (2011) Artemisinin activity against Plasmodium falciparum requireshemoglobin uptake and digestion. Proc Natl Acad Sci USA 108(28):11405–11410.

31. Zhang S, Gerhard GS (2008) Heme activates artemisinin more efficiently than hemin,inorganic iron, or hemoglobin. Bioorg Med Chem 16(16):7853–7861.

32. Meshnick SR, Thomas A, Ranz A, Xu CM, Pan HZ (1991) Artemisinin (qinghaosu): Therole of intracellular hemin in its mechanism of antimalarial action. Mol BiochemParasitol 49(2):181–189.

33. Becker K, et al. (2004) Oxidative stress in malaria parasite-infected erythrocytes: host-parasite interactions. Int J Parasitol 34(2):163–189.

34. Jortzik E, et al. (2010) Redox regulation of Plasmodium falciparum ornithine δ-ami-notransferase. J Mol Biol 402(2):445–459.

35. Krishna S, Uhlemann AC, Haynes RK (2004) Artemisinins: Mechanisms of action andpotential for resistance. Drug Resist Updat 7(4-5):233–244.

36. Hunt NH, Stocker R (1990) Oxidative stress and the redox status of malaria-infectederythrocytes. Blood Cells 16(2-3):499–526.

37. Makanga M, Bray PG, Horrocks P, Ward SA (2005) Towards a proteomic definition ofCoArtem action in Plasmodium falciparum malaria. Proteomics 5(7):1849–1858.

38. Mok S, et al. (2015) Drug resistance. Population transcriptomics of human malariaparasites reveals the mechanism of artemisinin resistance. Science 347(6220):431–435.

39. Xiao S, et al. (1998) Effect of artemether on hexokinase, glucose phosphate isomeraseand phosphofructokinase of Schistosoma japonicum harbored in mice. Zhongguo JiSheng Chong Xue Yu Ji Sheng Chong Bing Za Zhi 16(1):25–28.

40. Xiao SH, et al. (1998) Effect of artemether on glyceraldehyde-3-phosphate de-hydrogenase, phosphoglycerate kinase, and pyruvate kinase of Schistosoma japoni-cum harbored in mice. Zhongguo Yao Li Xue Bao 19(3):279–281.

41. Salcedo-Sora JE, Caamano-Gutierrez E, Ward SA, Biagini GA (2014) The proliferatingcell hypothesis: A metabolic framework for Plasmodium growth and development.Trends Parasitol 30(4):170–175.

42. Olszewski KL, Llinás M (2011) Central carbon metabolism of Plasmodium parasites.Mol Biochem Parasitol 175(2):95–103.

43. Ting LM, et al. (2005) Targeting a novel Plasmodium falciparum purine recyclingpathway with specific immucillins. J Biol Chem 280(10):9547–9554.

44. Yuthavong Y, et al. (2012) Malarial dihydrofolate reductase as a paradigm for drugdevelopment against a resistance-compromised target. Proc Natl Acad Sci USA109(42):16823–16828.

45. Cassera MB, Zhang Y, Hazleton KZ, Schramm VL (2011) Purine and pyrimidine path-ways as targets in Plasmodium falciparum. Curr Top Med Chem 11(16):2103–2115.

46. Sarkar D, Ghosh I, Datta S (2004) Biochemical characterization of Plasmodium falci-parum hypoxanthine-guanine-xanthine phosphorybosyltransferase: Role of histidineresidue in substrate selectivity. Mol Biochem Parasitol 137(2):267–276.

47. Yuthavong Y, et al. (2005) Malarial (Plasmodium falciparum) dihydrofolate reductase-thymidylate synthase: Structural basis for antifolate resistance and development ofeffective inhibitors. Parasitology 130(Pt 3):249–259.

48. Hartl FU, Bracher A, Hayer-Hartl M (2011) Molecular chaperones in protein foldingand proteostasis. Nature 475(7356):324–332.

49. Hayashi M, et al. (2000) Vacuolar H(+)-ATPase localized in plasma membranes ofmalaria parasite cells, Plasmodium falciparum, is involved in regional acidification ofparasitized erythrocytes. J Biol Chem 275(44):34353–34358.

50. Saliba KJ, Kirk K (1999) pH regulation in the intracellular malaria parasite, Plasmo-dium falciparum. H(+) extrusion via a V-type H(+)-ATPase. J Biol Chem 274(47):33213–33219.

51. Antoine T, et al. (2014) Rapid kill of malaria parasites by artemisinin and semi-syn-thetic endoperoxides involves ROS-dependent depolarization of the membrane po-tential. J Antimicrob Chemother 69(4):1005–1016.

52. Jambou R, et al. (2005) Resistance of Plasmodium falciparum field isolates to in-vitroartemether and point mutations of the SERCA-type PfATPase6. Lancet 366(9501):1960–1963.

53. Ariey F, et al. (2014) A molecular marker of artemisinin-resistant Plasmodium falci-parum malaria. Nature 505(7481):50–55.

54. Kuhn M, et al. (2014) STITCH 4: Integration of protein-chemical interactions with userdata. Nucleic Acids Res 42(Database issue):D401–D407.

55. Ridley RG (2003) Malaria: To kill a parasite. Nature 424(6951):887–889.56. Molloy MP, et al. (1998) Extraction of membrane proteins by differential solubilization

for separation using two-dimensional gel electrophoresis. Electrophoresis 19(5):837–844.57. Trager W, Jensen JB (1976) Human malaria parasites in continuous culture. Science

193(4254):673–675.58. Lambros C, Vanderberg JP (1979) Synchronization of Plasmodium falciparum eryth-

rocytic stages in culture. J Parasitol 65(3):418–420.59. Smilkstein M, Sriwilaijaroen N, Kelly JX, Wilairat P, Riscoe M (2004) Simple and in-

expensive fluorescence-based technique for high-throughput antimalarial drugscreening. Antimicrob Agents Chemother 48(5):1803–1806.

60. Bradford MM (1976) A rapid and sensitive method for the quantitation of microgramquantities of protein utilizing the principle of protein-dye binding.Anal Biochem 72:248–254.

61. Speers AE, Cravatt BF (2009) Activity-based protein profiling (ABPP) and click chem-istry (CC)-ABPP by MudPIT mass spectrometry. Curr Protoc Chem Biol 1:29–41.

62. Voronin D, et al. (2014) Wolbachia lipoproteins: Abundance, localisation and se-rology of Wolbachia peptidoglycan associated lipoprotein and the Type IV Secre-tion System component, VirB6 from Brugia malayi and Aedes albopictus. ParasitVectors 7:462.

63. Gardner MJ, et al. (2002) Genome sequence of the human malaria parasite Plasmo-dium falciparum. Nature 419(6906):498–511.

64. Saito R, et al. (2012) A travel guide to Cytoscape plugins. Nat Methods 9(11):1069–1076.

65. Bindea G, et al. (2009) ClueGO: A Cytoscape plug-in to decipher functionallygrouped gene ontology and pathway annotation networks. Bioinformatics 25(8):1091–1093.

Ismail et al. PNAS | February 23, 2016 | vol. 113 | no. 8 | 2085

BIOCH

EMISTR

Y

Dow

nloa

ded

by g

uest

on

Oct

ober

8, 2

020