LETTER doi:10.1038/nature10771 High-throughput decoding of antitrypanosomal drug efficacy and resistance Sam Alsford1, Sabine Eckert2{, Nicola Baker1, Lucy Glover1, Alejandro Sanchez-Flores2, Ka Fai Leung3, Daniel J. Turner2{, Mark C. Field3, Matthew Berriman2 & David Horn1 The concept of disease-specific chemotherapy was developed a century ago. Dyes and arsenical compounds that displayed selectivity against trypanosomes were central to this work1,2, and the drugs that emerged remain in use for treating human African trypanosomiasis (HAT)3. The importance of understanding the mechanisms underlying selective drug action and resistance for the development of improved HAT therapies has been recognized, but these mechanisms have remained largely unknown. Here we use all fivecurrent HAT drugs for genome-scaleRNA interference target sequencing (RIT-seq) screens in Trypanosoma brucei, revealing the transporters, organelles, enzymes and metabolic pathways that function to facilitate antitrypanosomal drug action. RIT-seq profil- ing identifies bothknown drugimporters4,5 and the only knownpro- drug activator6, and links more than fifty additional genes to drug action. A bloodstream stage-specific invariant surface glycoprotein (ISG75) family mediates suramin uptake, and the AP1 adaptin complex, lysosomal proteases and major lysosomal transmembrane protein, as well as spermidine and N-acetylglucosamine bio- synthesis, all contribute to suramin action. Further screens link ubiquinone availability to nitro-drug action, plasma membrane P-type H1-ATPases to pentamidine action, and trypanothione and several putative kinases to melarsoprol action. We also demon- strate a major role for aquaglyceroporins in pentamidine and melarsoprol cross-resistance. These advances in our understanding of mechanisms of antitrypanosomal drug efficacy and resistance will aid the rational design of new therapies and help to combat drug resistance, and provide unprecedented molecular insight into the mode of action of antitrypanosomal drugs. African trypanosomes are transmitted by the tsetse insect vector and circulate inthe bloodstream and tissue fluids of their mammalian hosts. These protozoan parasites cause HAT, also known as sleeping sickness, and the livestock disease known as Nagana. HAT is typically fatal if there is no chemotherapeutic intervention. The public health situation has improved recently with increased monitoring and chemotherapy averting more than 1.3 million disability-adjusted life years (DALYs) in the year 2000 and the estimated number of cases at less than 70,000 in 2006(ref. 7). However, therapies havemanyproblems, including severe toxicity and increasing resistance, which is a major concern owing to the absence of a vaccine or therapeutic alternatives3. The current HAT therapies are pentamidine or suramin, which are only suitable for the first stage ofthe diseasebeforecentralnervous systeminvolvement,and eflornithine, nifurtimox or melarsoprol for advanced disease3 (Sup- plementary Table 1). All of these drugs were developed well before the advent of molecular, target-based therapy and, with the exception of eflornithine, they elicit their antitrypanosomal effects by disrupting unknown targets. HAT treatment failure rates were reported to be increasing for suramin, when this drug was still in use in West Africa in the 1950s8, and melarsoprol treatment failure is a current and increasing problem9. We used genome-scale tetracycline-inducible RNA interference (RNAi) library screens in T. brucei to identify the genes that contribute to drug action. In these screens, replicating cells only persist in an otherwisetoxicenvironmentif knockdownconfersaselectiveadvantage (Fig. 1a) note that knockdown is not expected to identify drug targets. The RNAi library consists of ,750,000 clones, each transformed with one RNAi construct, and represents .99% of the approximately 7,500 non-redundant T. brucei gene set. Because each gene is identified by an average of approximately five different RNAi sequences, true leads can be identified withhigh confidenceandpotentialoff-target false leads can be minimized (see Supplementary Methods). Screens were performed using all current HAT drugs and each yielded a population of cells displayinganinducibledrugresistancephenotypeaftereightorfourteen days of selection (Fig. 1b and Supplementary Fig. 1). Genomic DNA from these cells was subjected to RIT-seq10 to create profiles of RNAi targets associated with increased resistance and to identify the genes that contribute to drug susceptibility. Genome-wide association maps show read density for 7,435 T. brucei genes (Fig. 1c). We defined genes with ‘primary signatures’ as those associated with two or more independent RIT-seq tags, each with a read density of .99 the screens yielded 55 of these signatures (Fig. 1c see Supplementary Methods and Supplemen- taryData1.PreviousworklinkedtheP2adenosinetransporter1 (AT1) to melarsoprol uptake4,11–13, an amino acid transporter family member (AAT6) to eflornithine uptake5,13,14 and a nitroreductase (NTR) to nifurtimox activation6,14. Each of these genes is identified on the appro- priate genome-wide association map (Fig. 1c), providing validation for our screens and indicating excellent genome-scale coverage in the RNAi library. Selected read-density signatures that establish new genetic links to drug susceptibility are shown in Fig. 1d. The known eflornithine transporter is the only primary signature from the eflornithine screen. By contrast, the suramin screen revealed 28 genes associated with primary signatures (Fig. 1c and Supplemen- tary Data 1). Suramin, which has been used for HAT therapy since the 1920s15, is a colourless sulphated napthylamine related to trypan red. Because this drug has a strong negative charge, it cannot cross lipid membranes by passive diffusion. Genes that are linked to the action of suramin encode ISG75, the function of which is unknown16, four lysosomal proteins (the cathepsin L (CatL) and CBP1 peptidases, p67 and Golgi/lysosomal protein 1 (GLP1)), all four subunits of the adaptin complex (AP1), which are involved in endosomal, clathrin-mediated trafficking, and multiple spermidine and N-acetylglucosamine biosynthetic enzymes (Supplementary Fig. 2 and Supplementary Data 1). Eight of these genes were selected forfurther analysis. We assembled multiple independent inducible RNAi strains for each gene and con- firmed that knockdown (Fig. 2a and Supplementary Fig. 3) increased suramin resistance in every case (Fig. 2b and Supplementary Fig. 4). We then determined subcellular localization for the putative major facilitator superfamily transporter (MFST) the tandem of three closely 1 LondonSchoolofHygieneandTropicalMedicine,KeppelStreet,London,WC1E7HT,UK. 2 TheWellcomeTrustSangerInstitute,Hinxton,Cambridge,CB101SA,UK. 3 Departmentof Pathology,Universityof Cambridge, Tennis Court Road, Cambridge, CB2 1QP, UK. {Present address: Oxford Nanopore Technologies, 4 Robert Robinson Avenue, Oxford, OX4 4GA, UK. 0 0 M O N T H 2 0 1 2 | V O L 0 0 0 | N A T U R E | 1 Macmillan Publishers Limited. All rights reserved ©2012

related MFST genes gave the strongest read-density signature in the suramin screen and the greatest half-maximum effective concentra- tion (EC50) increase (. tenfold) following knockdown (Fig. 2b). In contrast to a putative ubiquitin hydrolase (UBH1) identified by the screen, MFST and a member of the endomembrane EMP70 family partitioned into the T. brucei membrane fraction, as expected (Fig. 2c), and MFST localized to the lysosome as did the major lysosomal type I membrane glycoprotein, p67 (ref. 17), which was also identified in the screen (Fig. 2d). Because ISG75 trafficking is ubiquitin dependent18, we investigated whether UBH1 influenced ISG75 expression. UBH1 knockdown reduced ISG75 but not ISG65 expression (Fig. 2e), suggest- ing that de-ubiquitination by UBH1 specifically affects ISG75 copy number clearly this mimics the direct effect of RNAi against ISG75. A vacuolar protein sorting factor, Vps5, which positively controls ISG75 expression19, and a second putative ubiquitin hydrolase, were alsoidentified bythescreen (see Supplementary Fig. 2 and Supplemen- tary Data 1),suggestingthatISG75 copynumber is highly connected to suramin resistance. To investigate whether ISG75 contributes to suramin binding, we performed whole-cell binding assays using 3[H]-labelled suramin. Cells that were depleted for ISG75 displayed significantly and specifically reduced suramin binding (Fig. 2f). We observed a greater than fourfold increase in EC50 after knock- down of the CatL-like protease known as brucipain, another abundant lysosomal protein20, and an orthogonal assay using a dual-specificity CatL–CatBinhibitor revealedinhibitor antagonism(Fig.2g),indicating that protease activity enhances suramin toxicity. Taken together, the 120 100 80 60 40 20 0 –RNAi +RNAi RNAi: ISG75 MFST P = 0.036 a b f MFSTGFP p67 Phase MFST12MYC Rab11 + DNA Phase c e UBH1 p67 RNAi: – + – + ISG75 95 72 55 72 55 43 72 55 ISG65 d EMP7012MYC MFST12MYC GFPUBH1 kDa 130 95 95 72 55 GFP MYC g 2 Eflornithine FIC50 Suramin FIC 50 1 0 0 1 ISG75 GFPUBH1 EMP7012M MFST12M p67 CatL RNAi: – + – + – + – + – + – + – + – + AP1γ GLP1 1 0 0 1 2 3 4 5 FMK024 FIC 50 Suramin FIC 50 CatL None RNAi: 12 ISG75 UBH1 MFST p67 EMP70 AP1 β GLP1 10 8 6 4 2 0 1 + DNA DNA ISG75 ISG75 RNAi EC 50 change (+tet versus –tet) S W P S W P S W P Relative 3 [H]-suramin (%) Figure 2 | A network of proteins link ISG75, endocytosis and lysosomal functions to suramin action. a, Western blots demonstrate knockdown Coomassie stains serveas loading controls. Tags, green fluorescent protein (GFP) and 123MYC epitope (12M). See Supplementary Fig. 3 for growth curves. b, Endosomal and lysosomal factors and ISG75 contribute to suramin action. Error bars, s.d. from independent RNAi strains (see Supplementary Fig. 4). c, MFST and EMP70 are membrane associated. The western blots show supernatant (S), wash (W) and pellet (P membrane fraction). d, MFST co- localizes with lysosomal protein, p67, but not recycling endosomes (Rab11). Dashed boxes, areas magnified in fluorescent images. e, Knockdown of UBH1 specificallydecreasesISG75expression. f,ISG75mediatessuraminbinding.Error bars, s.d. from duplicate experiments. P value from Student’s t-test. ISG75 knockdown is shown. Scale bar, 5 mm. g, The CatL–CatB, and ODC inhibitors FMK024 and eflornithine, respectively,antagonize suramin action. Isobolograms showing 50% fractional inhibitory concentrations (FICs). The solid lines indicate antagonism. The dashed lines indicate expected outcomes for no interaction. Introduce into bloodstream-form T. brucei RNAi library + Tet: RNAi + Drug A + Drug B RIT-seq or b Eflornithine AAT6 NTR Pentamidine Suramin Chromosome 1 5,000 500 50 50 5,000 500 50 5,000 500 5,000 500 50 5,000 500 50 AAT6 A T T 6 6 6 6 Frequency of mapped reads AT1 Drug-resistant growth (% tet-dependent) 100 80 60 40 20 0 a c d Tb09.160.4630-v2.0030 Tb927.7.7230 MFST T[SH]2 synthesis Kinase +2 LATS1-like P-ATPase +1 Protein phosphatase Aquaglyceroporin Lysosomal protein ISG75 Nitroreductase UQ9 synthesis UQ9 synthesis Pentamidine Melarsoprol Nifurtimox 1246 6331 440446 970 Suramin 406739 2745 41687 14312 83104 836 55834 178748 Tb927.10.15310 Tb927.10.14160-70 Tb927.5.350-400 Tb09.160.4300 Tb927.10.1910 Tb927.4.4510 Tb927.5.1810-30 Tb927.10.9900 Tb927.2.4370 Tb927.10.12500-10 2 3 4 5 6 7 8 9 10 11 +3 +3 +3 +1 Nifurtimox Melarsoprol Eflornithine Melarsoprol Nifurtimox Pentamidine Suramin Figure 1 | Identification of drug efficacy determinants in T. brucei. a, A schematic showing the RNAi library screening approach. The expected outcomes are given for RNAi targets that fail to affect drug resistance (black), increase resistance to drug A (blue), drug B (orange) or both (green). b, Each screen yielded a population displaying tetracycline (Tet)-inducible (RNAi- dependent) drug-resistance (see Supplementary Fig. 1). The plot indicates the proportion of the resistance phenotype that is tetracycline inducible. c, Genome-wide RIT-seq profiles. Each map represents a non-redundant set of 7,435 protein-coding sequences. Red bars represent ‘primary’ read-density signatures. Black bars represent all other signatures of .50 reads (see Supplementary Data 1). All three expected ‘hits’, AAT6, AT1 and NTR, are indicated. d, Selected signatures. Each peak represents a unique RIT-seq tag. ‘1’, numbers of additional genes identified in each category. See Supplementary Fig. 2 for details and additional signatures. RESEARCH LETTER 2 | N A T U R E | V O L 0 0 0 | 0 0 M O N T H 2 0 1 2 Macmillan Publishers Limited. All rights reserved ©2012