Accessibility of substituted cysteines in TM2 and TM10 transmembrane segments in the Plasmodium falciparum equilibrative nucleoside transporter PfENT1

Infection with Plasmodium species parasites causes malaria. Plasmodium parasites are purine auxotrophic. They import purines via an equilibrative nucleoside transporter (ENT). In P. falciparum, the most virulent species, the equilibrative nucleoside transporter 1 (PfENT1) represents the primary purine uptake pathway. This transporter is a potential target for the development of antimalarial drugs. In the absence of a high-resolution structure for either PfENT1 or a homologous ENT, we used the substituted cysteine accessibility method (SCAM) to investigate the membrane-spanning domain structure of PfENT1 to identify potential inhibitor-binding sites. We previously used SCAM to identify water-accessible residues that line the permeation pathway in transmembrane segment 11 (TM11). TM2 and TM10 lie adjacent to TM11 in an ab initio model of a homologous Leishmania donovani nucleoside transporter. To identify TM2 and TM10 residues in PfENT1 that are at least transiently on the water-accessible transporter surface, we assayed the reactivity of single cysteine-substitution mutants with three methanethiosulfonate (MTS) derivatives. Cysteines substituted for 12 of 14 TM2 segment residues reacted with MTS-ethyl-ammonium-biotin (MTSEA-biotin). At eight positions, MTSEA-biotin inhibited transport, and at four positions substrate transport was potentiated. On an α helical wheel projection of TM2, the four positions where potentiation occurred were located in a cluster on one side of the helix. In contrast, although MTSEA-biotin inhibited 9 of 10 TM10 cysteine-substituted mutants, the reactive residues did not form a pattern consistent with either an α helix or β sheet. These results may help identify the binding site(s) of PfENT1 inhibitors.


Infection with Plasmodium species parasites causes malaria.
Plasmodium parasites are purine auxotrophic. They import purines via an equilibrative nucleoside transporter (ENT). In P. falciparum, the most virulent species, the equilibrative nucleoside transporter 1 (PfENT1) represents the primary purine uptake pathway. This transporter is a potential target for the development of antimalarial drugs. In the absence of a highresolution structure for either PfENT1 or a homologous ENT, we used the substituted cysteine accessibility method (SCAM) to investigate the membrane-spanning domain structure of PfENT1 to identify potential inhibitor-binding sites. We previously used SCAM to identify water-accessible residues that line the permeation pathway in transmembrane segment 11 (TM11). TM2 and TM10 lie adjacent to TM11 in an ab initio model of a homologous Leishmania donovani nucleoside transporter. To identify TM2 and TM10 residues in PfENT1 that are at least transiently on the water-accessible transporter surface, we assayed the reactivity of single cysteine-substitution mutants with three methanethiosulfonate (MTS) derivatives. Cysteines substituted for 12 of 14 TM2 segment residues reacted with MTS-ethyl-ammonium-biotin (MTSEA-biotin). At eight positions, MTSEA-biotin inhibited transport, and at four positions substrate transport was potentiated. On an ␣ helical wheel projection of TM2, the four positions where potentiation occurred were located in a cluster on one side of the helix. In contrast, although MTSEA-biotin inhibited 9 of 10 TM10 cysteine-substituted mutants, the reactive residues did not form a pattern consistent with either an ␣ helix or ␤ sheet. These results may help identify the binding site(s) of PfENT1 inhibitors.
In many developing nations, malaria is a major public health problem killing about 500,000 people annually. Infection with obligate intracellular, eukaryotic parasites of the genus Plasmodium causes malaria. Five Plasmodium species, falciparum, knowlesi, malariae, ovale, and vivax, infect humans. The highest mortality is associated with Plasmodium falciparum infection. P. falciparum parasites have developed resistance to drugs that have been used to treat malaria (1,2). Chloroquine was the mainstay of malaria treatment for over 50 years but its use is now limited by widespread resistance. Artemisinin-based combination therapies (ACT) 5 were designated as the first line treatment in 2006, but resistance against ACT has been documented in South-East Asia since 2011 and more recently in parasites from Africa (3)(4)(5)(6). The spread of ACT-resistant malaria strains highlights the importance of developing new anti-malarials that target novel metabolic pathways and proteins in the parasite.
One novel target is the P. falciparum equilibrative nucleoside transporter type 1 (PfENT1) (7,8). Plasmodium parasites are purine auxotrophic, but can synthesize pyrimidines by de novo synthesis (9 -12). PfENT1 is the primary purine transporter for the import of purine nucleobases and nucleosides, necessary for DNA and RNA synthesis, replication, and other metabolic processes (11,(13)(14)(15). PfENT1-knockout parasites (pfent1⌬) are not viable in culture with media purine concentrations similar to those found in human blood, below 10 M (8, 16 -18). Recent experiments using random transposon integration inferred that PfENT1 is an essential gene for blood stage parasites (19). Previously, we used a yeast-based high throughput screen to identify small molecule inhibitors of PfENT1 that kill malaria parasites in culture (8). The inhibitors represent six distinct chemical scaffolds. In the present work, we sought to understand more about the structure of the transmembrane domain Competition to fund the performance of hit validation assays in Dr. Akabas' laboratory related to the high throughput screen GSK performed with our yeast based assay to identify inhibitors of PfENT1 as potential starting points for the development of novel antimalarial drugs. The only connection between the work supported by GSK and the current work is that they both involve PfENT1. Dr. Akabas holds patents on derivatives of the hits identified by GSK. GSK has no interests in the patents or subsequent work. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains Figs. S1-S6. 1 Both authors contributed equally to this work. 2  cro ARTICLE of PfENT1 with the long-term goal of identifying the residues that contribute to the binding site for the inhibitors.
The substituted cysteine accessibility method (SCAM) is an experimental approach to study the structure of membrane transporters, channels, and receptors (20 -22). SCAM experiments assess the reactivity of sulfhydryl-reactive reagents with cysteines (Cys) substituted systematically, one at a time, in transmembrane segments. We used methanethiosulfonate (MTS) reagents to assess reactivity. MTS reagents react by an SN2 nucleophilic attack on the Cys sulfhydryl. Their reaction rate is a billion times faster with the ionized thiolate anion (S Ϫ ) than with the un-ionized thiol (SH) (23,24). Because only sulfhydryls that are, at least transiently, on the water-accessible protein surface are likely to ionize, reaction with MTS reagents is a proxy for water accessibility of the corresponding wildtype (WT) protein residue (22). Thus, SCAM allows the identification of residues within transmembrane segments that are, at least transiently, on the water-accessible protein surface. We previously used SCAM to investigate the structure of the PfENT1-transmembrane domain (25).
PfENT1 is a member of the SLC29 equilibrative nucleoside transporter gene family (26). A glycosylation site insertion mutagenesis study supported the predicted 11-transmembrane segment transmembrane topology with a cytoplasmic N terminus, a large cytoplasmic loop between transmembrane segments 6 (TM6) and TM7, and an extracellular C terminus ( Fig.  1) (27). No high resolution structures are available for PfENT1 or for a homologous ENT. An ab initio model has been constructed for the homologous Leishmania donovani LdNT1.1 transporter and validated using disulfide cross-linking and sitedirected mutagenesis (28 -30). We used this model to choose TM segments for this study. In a previous SCAM study, we identified residues in TM11 that line the purine permeation pathway (25). In the ab initio model TM2 and TM10 are adja-cent to TM11. In this study, we used SCAM to identify the water-accessible residues within the PfENT1 TM2 and TM10 segments. Our results indicate that Cys substituted for some TM2 residues reacted with the MTS reagents. We infer that they are water accessible and may line the permeation pathway. Based on the pattern of MTS-reactive residues, much of TM2 appears to form an ␣ helix. Cys substituted for several TM10 segment residues reacted with MTSEA-biotin. The pattern formed by the TM10 reactive residues was not consistent with either an ␣ helix or ␤ sheet.

MTS reagent effects on WT PfENT1
PfENT1 has 11 endogenous Cys residues. Therefore, we first determined whether MTS reagents had an effect on WT PfENT1. We assayed the effect of a 5-min application of increasing concentrations of the three MTS reagents on [ 3 H]adenosine ([ 3 H]Ado) uptake into yeast expressing WT PfENT1. Reagent solubility issues limited the maximum concentration of MTSEA-biotin to 8 mM. We chose to limit the application duration to 5 min to avoid issues related to changing MTS reagent concentration during an experiment due to spontaneous hydrolysis of the MTS reagents. In particular, for MTSET ϩ the hydrolysis half-time is 11.2 min in aqueous solution at pH 7.0 (24). The hydrolysis half-time for MTSES Ϫ is much longer, 370 min, whereas the hydrolysis halftime for MTSEA-biotin has not been determined but is likely to be longer than the MTSET ϩ half-time (24). Concentrations of MTS reagents below 1 mM had no effect on [ 3 H]Ado uptake into yeast expressing WT PfENT1 (Fig. 2). However, MTS reagent concentrations above 1 mM caused up to ϳ25% inhibition of [ 3 H]Ado uptake into yeast expressing WT PfENT1 (Fig. 2, Table 3). We sought to quantify reactivity based on the MTS reagent concentra- All endogenous Cys residues are represented by black circles. Topology prediction was carried out using the TOPCONS web server. The topology image was generated using the TOPO2 transmembrane protein display software.

Water-accessible residues in PfENT1 TM2 and TM10 segments
tion that caused half of the maximal effect of a 5-min application (XC 50 ). Unfortunately, GraphPad Prism software was unable to unambiguously fit a four-parameter variable slope log [inhibitor] versus response relationship to the data because there was no way to define the maximum effect of MTSEA-biotin on PfENT1 function. Thus, for WT we were unable to determine an XC 50 for the MTSEA-biotin effect, but it must be greater than 8 mM, which caused 22 Ϯ 6% inhibition (Table 3).
This suggested that at high concentrations the MTS reagents were able to react with one or more of the endogenous Cys residues. We sought to identify the reactive endogenous Cys residue. We mutated each endogenous Cys residue to alanine (Ala), one at a time. All of the single Cys to Ala mutants were functional (data not shown). MTSEA-biotin caused a similar inhibition with all 11 of the individual Cys to Ala mutants (data not shown). Therefore, it is likely that multiple endogenous Cys residues reacted with high concentrations of MTSEA-biotin resulting in the small observed functional effect. Given the potential complexity of trying to identify pairs or triplets of reactive endogenous Cys, we chose to make the TM2 and TM10 Cys-substitution mutants in the WT PfENT1 background. Of note, we previously reported that when PfENT1 was expressed in Xenopus oocytes we did not observe any effects of MTS reagents (25). However, in those experiments we only applied 2 mM MTS reagent for 2 min, which may not have been a sufficient concentration and reaction time to see the effects of the reagents on WT that we observed here.

Characterization of Cys-substitution mutants
In the ab initio model of the LdNT1.1, transporter residues from TM2 and TM10 flank TM11 and appear to line the putative purine transport pathway (28 -30). The N-terminal ends of both TM2 and TM10 of PfENT1 are predicted to be located extracellularly and the C-terminal ends are predicted to be cytoplasmic ( Fig. 1). We mutated residues from the predicted extracellular end of these two membrane-spanning segments toward the cytoplasmic end. In TM2 we mutated 14 consecutive residues, one at a time, to Cys from Phe-63 through Val-79, except residue 75, which is an endogenous Cys and was left unchanged. In TM10 we mutated 10 consecutive residues to Cys, one at a time, from Ala-348 through Phe-357.
We expressed each Cys-mutant in the purine auxotrophic yeast. The purine auxotrophic yeast can grow in media containing adenine as the sole purine source because adenine can enter the yeast through the endogenous FCY2 transporter (31). In the absence of functional PfENT1, the purine auxotrophic yeast cannot grow in media that contains adenosine as the sole purine source (data not shown), because they lack an endogenous adenosine transporter (8). In contrast, PfENT1-expressing yeast can grow in media with either adenine or adenosine as the sole purine source because PfENT1 can transport adenosine into the yeast (8). Yeast expressing each of the TM2 and TM10 Cys mutants were able to grow in media containing 1 mM adenosine as the sole purine source (data not shown). This implies that all of the Cys mutants formed functional adenosine transporters that imported sufficient adenosine to meet the purine requirements for yeast proliferation.
To determine whether the Cys substitutions affected the transporter's purine affinity, we measured the affinity of the nucleobase, hypoxanthine, and nucleoside, inosine, by uptake competition experiments. In uptake competition experiments, the purines hypoxanthine or inosine could compete with a radiolabeled purine nucleoside at either the Cys-mutant PfENT1 or at a downstream metabolic enzyme (32). To avoid this potential complication in the interpretation of the experiments, because PfENT1 transports uridine (Urd) (8), we used For many of the TM2 and TM10 mutants the IC 50 values for hypoxanthine and inosine inhibition of [ 3 H]Urd were within 1 order of magnitude of the IC 50 value for WT PfENT1 (Fig. 3, Tables 1 and 2). For the nucleobase hypoxanthine, only the IC 50 value for V74C was about 12-fold higher than the hypoxanthine IC 50 value for WT. In contrast, for inosine the IC 50 values were more than 10-fold higher than WT for F63C, N66C, Q69C, L73C, V74C, and F352C. At these positions, the Cys substitutions reduced the affinity for inosine by greater than 10-fold.

Reactivity of TM2 Cys-substitution mutants with MTS reagents
We tested the reactivity of the MTS reagents with individual Cys-substitution mutants using procedures similar to those we used for WT. We determined the effect of a 5-min application of increasing concentrations of the three MTS reagents on Water-accessible residues in PfENT1 TM2 and TM10 segments [ 3 H]Ado uptake into yeast expressing the individual Cys-substitution mutants. For two of the 16 TM2 Cys-substitution mutants, V74C and S77C, none of the three MTS reagents caused effects that were significantly different from those observed with WT PfENT1 (Fig. 4, Figs. S1 and S2). In contrast, for the other 14 TM2 Cys mutants, application of at least one of the MTS reagents caused a significant concentration-dependent effect on [ 3 H]Ado uptake. MTSEA-biotin application caused inhibition of subsequent [ 3 H]Ado uptake with 10 of the mutants (Fig. 4, Table 3). At three of these mutants, N66C, I70C, and G72C, MTSET ϩ application also inhibited subsequent [ 3 H]Ado uptake (Fig. 4, Fig. S1). For F68C, only MTSET ϩ had an effect and it caused potentiation of subsequent [ 3 H]Ado uptake (Fig. 4, Figs. S1 and S2). MTSES Ϫ only inhibited subse-quent [ 3 H]Ado uptake with the G72C mutant (Fig. 4, Fig. S2). For the G72C mutant, the Hill slope of the concentration-effect relationship was shallow, about Ϫ0.6, with all three MTS reagents suggesting that inhibition involved a multistep process (Fig. 4, Figs. S1 and S2, Table 3).
The maximal effect at 8 mM MTSEA-biotin, the highest concentration tested, ranged from 90% inhibition for I70C to about 30 to 40% inhibition for F63C and Y65C (Fig. 4, Table 3). For all of the mutants more cytoplasmic than L73C (Fig. 4), the reaction did not appear to have gone to completion during the 5-min application of 8 mM MTSEA-biotin. Consistent with this observation, the MTSEA-biotin XC 50 concentration was Ͻ400 M for all mutants between F63C and I71C (Fig. 4, Table 3). In contrast, the MTSEA-biotin XC 50 concentration was Ͼ1 mM for all mutants between L73C and V79C (Fig. 4, Table 3).
Surprisingly, with four of the Cys mutants, K64C, T67C, F68C, and T71C, the MTS reagent application potentiated subsequent [ 3 H]Ado uptake (Fig. 5, Table 3). The extent of potentiation of [ 3 H]Ado uptake ranged from about a 2-fold increase   Water-accessible residues in PfENT1 TM2 and TM10 segments for K64C (Fig. 5A) to about 20-fold for T71C (Fig. 5D) Table 3. Please note that the WT data in Fig. 2A is also shown in Figs. 4 and 7 to provide a point of comparison for the effects of MTSEA-biotin on the individual TM2 and TM10 Cys mutants, respectively. a NE, no effect.

Water-accessible residues in PfENT1 TM2 and TM10 segments
due to an increase in the number of PfENT1 transporters during the short period of MTS reagent application, however, we did not explicitly test this. More likely, either the affinity for Ado increased or the transporter turnover rate increased. We tested whether MTS modification of these Cys mutants altered the Ado affinity. To determine Ado affinity, we measured the inhibition of [ 3 H]Urd uptake into yeast expressing the four Cys mutants by increasing concentrations of cold adenosine (Fig.  6). A 5-min application of 3 mM MTS reagent did not alter the IC 50 value for adenosine inhibition of [ 3 H]Urd uptake (Fig. 6). This implies that MTS reagent modification of these four Cys mutants most likely increased the transporter turnover rate.

Reactivity of TM10 Cys-substitution mutants with MTS reagents
Similar experiments, as for the TM2 mutants, were carried out for the TM10 mutants (Fig. 7, Table 4). For six of the 10 TM10 Cys-substitution mutants, M349C, L350C, A351C, N354C, G355C, and W356C, the MTSEA-biotin reaction went to completion in the concentration-time domain of the experiment as indicated by the extent of reaction plateauing at the highest three concentrations used (Fig. 7). For three of the TM10 mutants, A348C, F352C, and F357C, the MTSEA-biotin reaction did not appear to go to completion in the 5-min period of reagent application (Fig. 7). For A348C and F352C, the calculated XC 50 was Ͻ1 mM, but for F357C it was Ͼ1 mM (Table 4). Finally, the effect of the reagents on T353C appear similar to WT (Fig. 7, Table 4). Application of MTSET ϩ or MTSES Ϫ did not affect subsequent [ 3 H]Ado uptake for any of the TM10 mutants except for G355C, where 3 mM MTSET ϩ caused about 50% inhibition (Figs. S3 and S4).

Discussion
An underlying assumption of SCAM studies is that the Cys substitutions do not significantly alter the native protein structure (22,24). All 24 of the Cys mutants supported growth of purine auxotrophic yeast on adenosine as the sole purine source. Because these purine auxotrophic yeast lack an endogenous adenosine transporter, their ability to grow with adenosine as the sole purine source provides strong evidence that all of the Cys mutants retain the capacity to transport adenosine. This suggests that the overall structure of the mutants is similar to WT. However, for six mutants, F63C, N66C, Q69C, L73C, V74C, and F352C, the Cys substitution did reduce the inosine affinity by more than an order of magnitude (Tables 1 and 2). Curiously, these five TM2 mutants lie on one side of an ␣ helical wheel projection of TM2 (Fig. 8B). At four of these positions, except for V74C, the Cys mutation did not have as large an effect on hypoxanthine affinity. Because hypoxanthine is the nucleobase of inosine, this may suggest that the inosine ribose sugar might interact with this region of TM2. Two other residues lie on the same side of the putative TM2 helix, I70C, where MTSEA-biotin caused the greatest inhibition of [ 3 H]Ado uptake, and S77C, two helical turns more cytoplasmic where MTS reagents had no functional effect (Tables 1 and 3).
The residues in TM2 are poorly conserved among equilibrative nucleoside transporter homologues from different species (Fig. S5). In contrast, Gly-355 in TM10 is absolutely conserved   Table 4. Please note that the WT data in Fig. 2A is also shown in Figs. 4 and 7 to provide a point of comparison for the effects of MTSEA-biotin on the individual TM2 and TM10 Cys mutants.

Water-accessible residues in PfENT1 TM2 and TM10 segments
from human to Arabidopsis to Plasmodium and the two residues on either side of Gly-355 are highly conserved. Thr-353 is either a threonine or serine, and Asn-354 is absolutely conserved except in a Culex ENT homologue where there is a serine at the aligned position (Fig. S6). Thus, despite the conservation of these residues, the Cys substitutions were tolerated.
A goal of SCAM studies is to determine the water-surface accessibility of the corresponding WT residues. In most SCAM studies, covalent reaction between an MTS reagent and a sub-stituted Cys residue is detected by a change in the protein's function. In the present study, reaction was detected by its effect on the rate of radioactive substrate uptake into yeast expressing the Cys mutant. The inference that MTS reactive Cys residues are on the water-accessible protein surface is based on: 1) MTS reagents react with the ionized thiolate (ϪS Ϫ ) form of Cys Ͼ10 9 times faster than with the un-ionized thiol (ϪSH) (22)(23)(24), 2) in free solution, the Cys thiol pK a is ϳ8.5, although in a protein, the thiol pK a can depend on the local electrostatic environment, and 3) the MTS reagents used in these experiments have very limited membrane permeability (22). Cys thiols in a membrane-spanning segment will only ionize to any significant extent if they are, at least transiently, on the water-accessible surface of the protein, which could include being in the substrate-translocation pathway through the interior of a membrane-transport protein or ion channel or might be through a crevice extending from the extracellular solution into the membrane-spanning domain. It is important to recognize that as a membrane-transport protein goes through its conformational changes, residues may move from the wateraccessible surface to a buried or lipid-facing position depending on the conformational state of the protein. So MTS reagent reactivity does not imply that the corresponding WT residue is always on the water-accessible surface.
Somewhat surprisingly, there were only three positions, V74C, S77C, and T353C, at which we did not detect any functional effect of the MTS reagents (Fig. 8). For positions where MTS reagent application has no effect, one cannot assume that the Cys is nonreactive because covalent modification might have no detectable effect on protein function (33). Alternatively, local steric factors may prevent the MTS reagent from reacting, but the Cys might still be, at least transiently, on the water-accessible surface. So lack of functional evidence of MTS reaction must be interpreted with care.
For the remaining 21 Cys mutants that appeared to react with at least one of the MTS reagents, we can assume that at least transiently the substituted Cys residue was on the water-accessible protein surface. It is unknown whether within an individual transmembrane segment all of the residues are simultaneously on the water-accessible surface or only a subset at one time as the protein moves through the conformational changes associated with its transport cycle. One question is, given the large number of MTS reactive residues, is there evidence to support an ␣ helical secondary structure. For TM2, on an ␣  Water-accessible residues in PfENT1 TM2 and TM10 segments helical wheel plot, we can identify two functionally distinct faces. One face referred to above is distinguished because most of the Cys mutants on that face caused a greater than 10-fold reduction in inosine affinity (Fig. 8B). The other face contains the four positions, K64C, T67C, F68C, and T71C, where MTS reagent reaction increased the rate of [ 3 H]Ado uptake (Fig. 8B). It is notable that at these four positions, the smaller charged MTS reagents, MTSET ϩ and MTSES Ϫ caused significant functional effects similar to the larger MTSEA-biotin. In contrast, at most of the other positions, neither of the smaller, charged reagents caused a functional effect. We do not know whether they reacted with no functional effect or whether the charge prevented access to the sites. We think the latter is less likely, especially at the more extracellular sites in TM2. The increased [ 3 H]Ado uptake was presumably due to an increase in the transporter turnover rate, rather than to a change in adenosine affinity, which we ruled out, or to a rapid increase in the number of transporters in the plasma membrane during the 5-min application period. The fact that the extent of the increase in [ 3 H]Ado uptake depends on the specific position modified on that face provides additional evidence that the increased uptake is not due to an increase in the number of transporters. Differences in the rates of MTSEA-biotin reaction as a function of position along the length of a putative helix support the transmembrane orientation with the N-terminal end being extracellular. The MTSEA-biotin XC 50 values were all less than 400 M for the TM2 residues more extracellular than T71C (Fig. 8A). This implies that the reaction rate was higher with these substituted Cys residues. Whether they are on the water-accessible surface for a greater fraction of the time or whether steric factors limit the ability of the MTS reagents to gain access to the residues more cytoplasmic than G72C is uncertain. In total, our evidence would support the hypothesis that TM2 has a largely ␣ helical secondary structure.
In the case of TM10, our results do not preclude TM10 being ␣ helical but they also do not provide evidence to support an ␣ helical secondary structure. Our results do not provide any support for a ␤ strand structure either.
The high degree of MTS reactivity around the putative circumference of the TM2 and TM10 helices stands in marked contrast to the limited MTS reactive face seen in our previous SCAM study of the PfENT1 TM11 transmembrane segment (25). We do not know whether this is due to the different expression systems used in the two studies. For TM11 we used the Xenopus laevis oocyte expression system. In the current work we used Saccharomyces cerevisiae. It is possible that differences in membrane lipid composition might affect the conformational structure of the transporter. Alternatively, the absolutely conserved GXXXG motif in TM11 may mediate a tight interaction with another part of the protein that limits access to that side of TM11 (34,35).
These results provide new insights into the structure of the transmembrane domain of PfENT1. In the absence of high resolution X-ray or cryo-EM studies, they may help to identify the binding site of PfENT1 inhibitors and the conformational changes associated with transport.

PfENT1 constructs and Cys substitution mutants
All mutants were made in a yeast codon-optimized pfent1 gene with a C-terminal hemagglutinin epitope tag in the pCM189m S. cerevisiae yeast expression vector as described previously (8). The gene encodes the 3D7 parasite strain PfENT1 amino acid sequence (PlasmoDB ID PF3D7_1347200), subsequently referred to as wildtype (WT) PfENT1. In one set of mutants, each of the 11 endogenous Cys residues was mutated to alanine, one at a time. Single Cys substitution mutants were generated, one at a time, in TM2 from positions 63 to 79, and in TM10 from positions 348 to 357 in the WT PfENT1 background. TM2 residue 75 is an endogenous Cys and was left unchanged. All mutations were generated using the PCR-based QuikChange II Site-directed Mutagenesis kit (Agilent). Primers were designed using the online QuikChange primer design tool. Mutations were verified by DNA sequencing of the entire coding region (Genewiz).

Yeast growth media
Purine auxotrophic yeast were grown in synthetic defined media (SDM) that contained 2% (w/v) glucose, 0.5% (w/v) ammonium sulfate, 0.17% yeast nitrogen base (U. S. Biologicals), 0.02% (w/v) yeast dropout mix lacking uracil, adenine, histidine, and tryptophan (U. S. Biologicals), 40 mg/liter of tryptophan, and 40 mg/liter of histidine. Media was supplemented with 1 mM adenosine (Ado) as the sole purine source. For initial plating, yeast transformations were plated on solid media plates that contained SDM (lacking uracil) plus 1 mM adenine in 2% agar. For selection and propagation of PfENT1expressing yeast, solid media plates contained SDM (lacking uracil) with 1 mM adenosine in 2% agar. For all uptake experiments, PfENT1 Cys mutant expressing yeast were grown to early mid-log phase, shaking overnight at 225 rpm at 30°C in SDM containing 1 mM adenosine. Cells were pelleted and washed 3ϫ in PBS ϩ glucose (150 mM NaCl, 10 mM KH 2 PO 4 , 40 mM K 2 HPO 4 , 11 mM glucose, pH 7.4). For all the assays described below, cells were resuspended in PBS ϩ glucose to a suspension density of 2 ϫ 10 8 cells/ml. Yeast cell density was determined by measuring optical density at 600 nm (A 600 ) (Bio-Rad Benchmark Plus).

Characterization of hypoxanthine and inosine transport by the Cys mutants
To ascertain whether the individual Cys substitutions altered transporter function, we determined the purine affinity of each Cys-mutant. We measured the concentration-dependent inhibition of [ 3 H]Urd uptake by the purine nucleobase, hypoxanthine, and nucleoside, inosine. PfENT1 transports Urd. We used competition with [ 3 H]Urd uptake to avoid the potential for competition at downstream purine metabolic enzymes (37).
For most of the Cys mutants, we also characterized inosine uptake by competition with [ 3 H]Ado uptake. A 96-well plate was preloaded with 50 l/well of serially diluted (1:2) cold inosine (highest concentration, 10 mM). 50 l of 200 nM [ 3 H]Ado ( [2, H]adenosine, 35 Ci/mmol, Moravek Biochemicals) was added to each well. 100 l of yeast suspension (2 ϫ 10 8 cells/ml) was added and uptake was measured at 15 min. At the end of each experiment, cells were harvested and counted as described above. All experiments were repeated at least three times on different days.

Substituted cysteine accessibility experiments
We assayed the reactivity of the TM2 and TM10 segment Cys-substitution mutants with MTS reagents. The MTS reagents used, in order of increasing size, were MTSET ϩ , MTSES Ϫ , and MTSEA-biotin (Biotium). To ensure that the MTS reagents did not hydrolyze, MTSEA-biotin was prepared as an 800 mM stock solution in DMSO, and MTSET ϩ and MTSES Ϫ were prepared as 300 mM stock solutions in DMSO.
A 96-well plate was preloaded with 100 l of yeast suspension (2 ϫ 10 8 cells/ml). 1 l of MTS reagent serially-diluted 1:3 in DMSO was added. Plates were incubated for 5 min at room temperature. 100 l of 100 nM [ 3 H]Ado was then added. After a 15-min uptake, cells were harvested and counted as above. Data were normalized to the lowest concentration of MTS reagent applied after background subtraction. All experiments were repeated at least three times on different days.

Effect of MTS treatment on Cys mutant adenosine affinity
For some Cys-mutants, tritiated substrate uptake was potentiated after MTS reaction. To determine whether the potentiation was a consequence of increased substrate affinity or increased rate of uptake, we measured the adenosine affinity. We measured the effect of increasing concentrations of cold adenosine on the uptake of [ 3 H]Urd before and after MTS modification. A 96-well plate was preloaded with 50 l of 100 nM [ 3 H]Urd, and 50 l/well 1:2 serially diluted adenosine (highest concentration, 10 mM). A yeast suspension (2 ϫ 10 8 cells/ml) was split into two aliquots. To one aliquot 3 mM MTS reagent was added. An equal volume of buffer without MTS reagent was added to the second aliquot. After 3 min, 100 l of yeast suspension was added to each well. After 15 min, yeast were harvested and counted as described above to measure uptake.
Adenosine IC 50 values for the inhibition of [ 3 H]Urd uptake were calculated as described above for hypoxanthine and inosine. All experiments were repeated at least three times on different days.