Molecular Mechanisms for Drug Hypersensitivity Induced by the Malaria Parasite’s Chloroquine Resistance Transporter

Mutations in the Plasmodium falciparum ‘chloroquine resistance transporter’ (PfCRT) confer resistance to chloroquine (CQ) and related antimalarials by enabling the protein to transport these drugs away from their targets within the parasite’s digestive vacuole (DV). However, CQ resistance-conferring isoforms of PfCRT (PfCRTCQR) also render the parasite hypersensitive to a subset of structurally-diverse pharmacons. Moreover, mutations in PfCRTCQR that suppress the parasite’s hypersensitivity to these molecules simultaneously reinstate its sensitivity to CQ and related drugs. We sought to understand these phenomena by characterizing the functions of PfCRTCQR isoforms that cause the parasite to become hypersensitive to the antimalarial quinine or the antiviral amantadine. We achieved this by measuring the abilities of these proteins to transport CQ, quinine, and amantadine when expressed in Xenopus oocytes and complemented this work with assays that detect the drug transport activity of PfCRT in its native environment within the parasite. Here we describe two mechanistic explanations for PfCRT-induced drug hypersensitivity. First, we show that quinine, which normally accumulates inside the DV and therewithin exerts its antimalarial effect, binds extremely tightly to the substrate-binding site of certain isoforms of PfCRTCQR. By doing so it likely blocks the normal physiological function of the protein, which is essential for the parasite’s survival, and the drug thereby gains an additional killing effect. In the second scenario, we show that although amantadine also sequesters within the DV, the parasite’s hypersensitivity to this drug arises from the PfCRTCQR-mediated transport of amantadine from the DV into the cytosol, where it can better access its antimalarial target. In both cases, the mutations that suppress hypersensitivity also abrogate the ability of PfCRTCQR to transport CQ, thus explaining why rescue from hypersensitivity restores the parasite’s sensitivity to this antimalarial. These insights provide a foundation for understanding clinically-relevant observations of inverse drug susceptibilities in the malaria parasite.


Introduction
Originally identified as the protein responsible for conferring resistance to the 'wonder-drug' chloroquine (CQ) [1,2], the Plasmodium falciparum 'chloroquine resistance transporter' (PfCRT) has become a key player in the malaria parasite's steadily expanding resistance to drugs [3][4][5]. The isoforms of PfCRT that confer CQ resistance (PfCRT CQR ) render the parasite less susceptible to many other compounds [6][7][8][9][10], but also simultaneously induce hypersensitivity to a subset of structurally-diverse molecules [6,[10][11][12][13][14][15][16][17][18]. This phenomenon, whereby resistance to one drug causes hypersensitivity to another, is known as 'inverse susceptibility' or 'collateral sensitivity' and has been observed in a wide range of pathogens and cancer cells [19][20][21][22]. The growing awareness of the propensity of drug-resistant pathogens to exhibit hypersensitivity to one or more other drugs has sparked interest in the potential for exploiting this Achilles' heel to combat existing drug resistance and to delay the emergence of resistance to new drugs [19]. However, it is not known how PfCRT CQR isoforms induce hypersensitivity to certain drugs and, more generally, much remains to be understood about the molecular mechanisms that underpin collateral sensitivity in pathogens and cancer cells [20][21][22].
The ability of PfCRT to affect the activity of so many compounds is likely to be a product of its location at the membrane of the parasite's digestive vacuole (DV) [1,11]; an acidic compartment in which many types of antimalarials accumulate and/or act. The parasite takes up hemoglobin and digests it within this compartment in order to grow within its host erythrocyte. This process releases heme monomers that are detoxified via conversion into the inert crystal hemozoin. Quinoline-type antimalarial drugs, including CQ, quinine, and quinidine, concentrate within the DV via 'weak-base trapping' [23], where they exert an antimalarial effect by binding to heme and arresting its detoxification [24][25][26][27]. Resistance to these quinolines is associated with reductions in the accumulation of the drugs within the DV [8,11,28] and we have previously obtained direct evidence of this phenomenon being due, at least in part, to the ability of PfCRT CQR isoforms to efflux CQ, quinine, and quinidine from this compartment. This was achieved by expressing a PfCRT CQR isoform from Dd2 parasites (PfCRT Dd2 ) and the wild-type sequence has been codon-harmonized for expression in Xenopus oocytes and encodes a retention motif-free form of PfCRT that expresses in a functional form at the oocyte plasma membrane [42]. We conducted immunofluorescence assays ( Fig 1A) to confirm the localization of each of the PfCRT K1 variants to the oocyte plasma membrane and used a semiquantitative western blot analysis [42] to establish that they were present at comparable levels in the oocyte membrane ( Fig 1B). Hence, any differences in drug transport activity between these isoforms of PfCRT can be attributed to differences in their transport properties rather than differences in expression.

Quinoline transport via PfCRT largely correlates with quinoline resistance
To understand the mechanisms underlying the inverse drug susceptibilities of the 106/1-and K1-derived lines, we characterized the transport properties of the respective PfCRT K1 variants in Xenopus oocytes. A key advantage of the oocyte system is that it allows the transport activity of PfCRT to be studied directly and in isolation, without confounding effects such as the binding of drugs to heme or interactions of the compound with other parasite targets or transporters. The direction of [ 3 H]drug transport in this system is from the acidic extracellular medium (pH 5.0-6.0) into the oocyte cytosol (pH 7.1-7.2 [30]), which corresponds to the efflux of protonated drug from the acidic DV (pH 5.0-5.5 [43,44]) into the parasite cytosol (pH 7.3 [45]). Noninjected oocytes take up [ 3 H]CQ, [ 3 H]quinine, and [ 3 H]quinidine to low levels via simple diffusion of the neutral species of the drug (Fig 2A-2C); this represents the background level of [ 3 H]drug accumulation [29,30]. We found that none of the PfCRT K1 isoforms carried by CQsensitive exhibited CQ transport activity when expressed in oocytes (Fig 2A). By contrast, the The 106/1-derived lines were generated by Cooper and colleagues [11,12,14] and the K1AM line was produced by Johnson et al. [15]. b The susceptibilities of the parasite lines to CQ, quinine (QN), and quinidine (QD) were collated from Cooper et al. [11,12,14] and Johnson et al. [15]. The drug resistance index is the IC 50 for a drug in a given strain or line divided by the IC 50 determined in the same study in a drug-sensitive reference strain (i.e., expression of PfCRT K1 variants from CQ-resistant parasites caused significant increases in the accumulation of CQ. Indeed, we observed a positive correlation between the ability of a PfCRT K1 variant to transport CQ and the magnitude of in vitro CQ resistance exhibited by the respective parasite line (R 2 value of 0.856; S2A Fig). We conclude that the mutations which suppress the quinine-hypersensitivity of the 106/1 76I line (C72R, Q352K, or Q352R) simultaneously re-sensitize the parasite to CQ by abrogating the protein's capacity to efflux CQ from the DV. Likewise, a mutation that suppresses the hypersensitivity of 106/1 76I parasites to amantadine (V369F), also caused a marked reduction in the ability of 76I-PfCRT K1 to transport CQ, which explains why the 106/1 76I,369F line displayed a relatively low level of resistance to CQ. Similar results were obtained when we examined the uptake of [ 3 H]quinine and [ 3 H]quinidine ( Fig 2B and 2C and S2 Fig). That is, the capacities of the different PfCRT K1 isoforms to transport quinine or quinidine generally correlated with the in vitro responses of the parasite lines to these drugs. Key exceptions included the quinine hypersensitivities of the 106/1 76I and 106/1 76I,369F lines, which were not readily reconciled with the low quinine transport activity of 76I-PfCRT K1 and the lack of quinine transport via 76I,369F-PfCRT K1 , respectively. The modest capacity of 76I-PfCRT K1 for quinine transport was nonetheless abolished by the Isoforms of PfCRT K1 localize to the surface of Xenopus oocytes. (A) Immunofluorescence microscopy was used to localize PfCRT in the oocyte. In each case, the expression of the PfCRT variant resulted in a fluorescent band external to the pigment layer, indicating that the protein was expressed in the oocyte plasma membrane. The band was not present in noninjected oocytes. The images are representative of at least two independent experiments (performed using oocytes from different frogs), within which images were obtained from a minimum of three oocytes per oocyte type. (B) The level of PfCRT protein in the oocyte membrane was semiquantified using a western blot method [42]. The analysis included PfCRT K1 as a positive control, to which the other band intensity values were normalized. The data are the mean + SEM of at least five independent experiments (performed using oocytes from different frogs), within which measurements were averaged from two independent replicates. There were no significant differences in expression levels between constructs (P > 0.05; one-way ANOVA); hence, all of the PfCRT variants were present at similar levels in the oocyte membrane.
doi:10.1371/journal.ppat.1005725.g001 introduction of C72R, Q352K, or Q352R. Another exception was the moderate ability of 76N-PfCRT K1 to transport quinidine, which was at odds with the sensitivity of 106/1 76N parasites to this drug.
Together, these findings revealed a strong relationship between the magnitude of CQ, quinine, or quinidine resistance exhibited by a given 106/1 or K1 parasite and the capacity of its PfCRT protein for mediating the efflux of the respective drug from the DV. However, the mechanism underpinning the quinine-hypersensitivity caused by the introduction of T76I into PfCRT K1 , and why the subsequent addition of C72R, Q352K, or Q352R negates this response, remained unclear.
High-affinity binding of quinine to 76I-PfCRT K1 causes hypersensitivity to quinine We undertook experiments to determine whether the kinetics of 76I-PfCRT K1 -and 76N-PfCRT K1 -mediated transport could explain why these proteins cause unexpected responses to quinine and quinidine, respectively. All of the PfCRT K1 isoforms we had identified as possessing significant CQ, quinine, or quinidine transport activity (Fig 2A-2C) were included for study. The resulting Michaelis-Menten plots (S3 Fig) yielded a Michaelis-Menten constant (K m ) and maximum velocity (V max ) for each drug and PfCRT isoform ( Table 2). Across all three drugs, the K m values increased in the order 76I-PfCRT K1 < PfCRT K1 < 76N-PfCRT K1 , whereas the V max values increased in the order 76I-PfCRT K1 < 76N-PfCRT K1 PfCRT K1 . Thus, we conclude that T76I increases the affinity of PfCRT K1 for its quinoline substrates, but CQ, quinine, and quinidine transport activities differ significantly between isoforms of PfCRT K1 . Noninjected oocytes accumulate low levels of CQ, quinine, and quinidine via simple diffusion of the neutral species of the drug [29,30] H]quinidine was measured at pH 5.0 and in the presence of 1 μM of the respective unlabeled drug. The rates of uptake (pmol per oocyte/h) in noninjected oocytes and oocytes expressing PfCRT K1 were 0.10 ± 0.01 and 0.55 ± 0.03, respectively, for quinine and 0.063 ± 0.010 and 0.38 ± 0.08, respectively, for quinidine. Note that the relatively low total concentration of quinine and quinidine (1.25 μM) used in these assays enabled the detection of [ 3 H]quinine transport via the 76I-PfCRT K1 isoform. However, this low drug concentration also resulted in rates of quinine and quinidine transport that were 10-15 times lower than those measured for CQ (a difference which corresponds well with the~12-fold reduction in the total concentration of quinine or quinidine relative to the total concentration of CQ). In all panels, drug uptake is expressed relative to that measured in oocytes expressing PfCRT K1 and the data are the mean + SEM of at least five independent experiments (performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. The asterisks denote a significant difference in drug uptake between the noninjected treatment and that measured in oocytes expressing a variant of PfCRT: *P < 0.05; **P < 0.01; ***P < 0.001 (one-way ANOVA). that this change is accompanied by a significant decrease in its maximum rate of transport. By contrast, T76N consistently decreased the affinity of PfCRT K1 for its quinoline substrates, and either had little effect on the V max (quinine) or caused a marked reduction in the V max (CQ and quinidine). Interestingly, the introduction of V369F into 76I-PfCRT K1 significantly reduced the protein's affinity for CQ (from 247 to 415 μM) and further reduced its maximum rate for CQ transport (from 29.3 to 13.8 pmol per oocyte/h), while also abrogating its ability to transport quinine and quinidine (Fig 2B and 2C). Our finding that 76I,369F PfCRT K1 is a very low-affinity and low-capacity transporter of CQ is again consistent with the modest level of CQ resistance exhibited by the 106/1 76I,369F parasite line.
Remarkably, we found that 76I-PfCRT K1 possesses an extraordinarily high affinity for quinine and an extraordinarily low V max for quinine transport; its K m was~70 times lower, and its V max~1 15 times lower, than the value obtained for PfCRT K1 . This combination of kinetic properties indicates that 76I-PfCRT K1 binds extremely tightly to quinine and only occasionally translocates the drug. Thus, quinine will clog the binding site of 76I-PfCRT K1 , which should greatly diminish the ability of the protein to transport its natural substrate. These findings suggest that quinine-hypersensitivity results from quinine exerting two killing effects in 106/1 76I parasites-the inhibition of hemozoin formation and the inhibition of PfCRT's normal physiological role (which is essential for parasite survival [39,40]). The low level of quinine transport mediated by 76I-PfCRT K1 was not detected following the introduction of C72R, Q352K, Q352R, or V369F ( Fig 2B). Of these mutations, Q352R and Q352K each fully return 106/1 76I parasites to quinine-sensitive status, C72R causes a substantial but incomplete rescue from quinine-hypersensitivity, and V369F confers a modest but significant suppression of the parasite's hypersensitivity to quinine (Table 1). Taken together, our data indicate that the insertion of a positively-charged residue into the substrate-binding cavity of 76I-PfCRT K1 greatly diminishes or abolishes its interaction with quinine (and its ability to transport CQ and quinidine), thereby reversing the parasite's hypersensitivity to quinine (and simultaneously re-sensitizing it to CQ and quinidine). Our finding that these effects are achieved to a much lesser degree by V369F, which instead introduces a bulky hydrophobic residue into 76I-PfCRT K1 , indicates that electrostatic repulsion is the key mechanism underpinning the dramatic decrease in the protein's ability to bind and translocate protonated quinolines. All values are the mean ± SEM of multiple independent experiments (performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. The exact n values are indicated in parentheses.
The P values determined from an ANOVA were less than c 0.001, d 0.01, or e 0.05 for comparisons with PfCRT K1 within the same drug treatment and kinetic parameter.
The P values determined from an ANOVA were less than f 0.001 or g 0.01 for comparisons between 76N-PfCRT K1 and 76I-PfCRT K1 within the same drug treatment and kinetic parameter. Our kinetic analyses also revealed that 76N-PfCRT K1 is unique in being both a low-affinity and a low-capacity transporter of quinidine. By comparison, PfCRT K1 mediated the low-affinity but high-capacity transport of quinidine, and 76I-PfCRT K1 was a high-affinity, low-capacity transporter of quinidine. Thus, the sensitivity of 106/1 76N parasites to quinidine is likely due to the inability of 76N-PfCRT K1 to decrease the concentration of quinidine within the DV to subtoxic levels.
I356T increases quinine transport via 76I-PfCRT K1 We recently reported [42] that the CQ transport activity of PfCRT K1 is~20% higher than that of PfCRT Dd2 , even though the two proteins differ only at position 356 (Table 1). However, the I356T mutation also demonstrated epistasis. That is, depending on the nature and number of the mutations already present, its introduction increased, decreased, or had no effect on the ability of PfCRT to transport CQ [42]. We therefore investigated whether the introduction of I356T decreases the ability of PfCRT K1 variants to transport quinine and quinidine. We measured the uptake of [ 3 H]quinine or [ 3 H]quinidine in oocytes expressing either a PfCRT K1 variant (PfCRT K1 , 76N-PfCRT K1 , or 76I-PfCRT K1 ) or its I356T counterpart (PfCRT Dd2 , 76N-PfCRT Dd2 , or 76I-PfCRT Dd2 ). Consistent with our previous observation with CQ transport, we found that the rate of quinine and quinidine uptake mediated by PfCRT K1 was 17-35% greater than that measured for PfCRT Dd2 (S4 Fig). Moreover, we observed the same relationship, only considerably more exaggerated, between the 76N variants of PfCRT K1 and PfCRT Dd2 and, in the case of quinidine transport, between 76I-PfCRT K1 and 76I-PfCRT Dd2 . The single exception to this pattern was quinine transport via 76I-PfCRT K1 , which showed a 30-40% decrease relative to its PfCRT Dd2 counterpart. Our findings establish a key role for position 356 in the attainment of a high level of quinoline transport activity, but also confirm the epistatic nature of this position.

Isoforms of PfCRT from amantadine-hypersensitive parasites transport amantadine
We next sought to understand why CQ-resistant parasites are hypersensitive to amantadine and how certain mutations in PfCRT (e.g., T76K, S163R, or V369F) countercheck this response. Given that we had already ascertained that the PfCRT K1 variants carrying T76K or S163R do not possess significant quinoline transport activity (Fig 2A-2C), we focused our initial investigations on 76I,369F-PfCRT K1 , as it mediates a detectable level of CQ transport. To test the possibility that hypersensitivity to amantadine is due to it clogging the substrate-binding cavity of PfCRT, and that this effect is alleviated by the addition of V369F, we compared the ability of unlabeled amantadine to inhibit the transport of [ 3 H]CQ via 76I-PfCRT K1 and 76I,369F-PfCRT K1 (Fig 3A). Somewhat surprisingly, there was little difference in the resulting half-maximum inhibitory concentrations (IC 50 s; listed in Fig 3A). Moreover, these IC 50 s were much higher than those obtained in the oocyte system for established inhibitors of PfCRT CQR , such as the quinine dimer Q 2 C (1.4 ± 0.2 μM [46]), saquinavir (13 ± 1 μM [47]), and verapamil (30 ± 3 μM [30]). Since amantadine is a relatively low-affinity inhibitor of both 76I-PfCRT K1 and 76I,369F-PfCRT K1 , it is unlikely that the hypersensitivity of CQ-resistant parasites to amantadine is due to it exerting an anti-PfCRT CQR effect.
This result led us to examine whether amantadine hypersensitivity is instead a consequence of significant changes in the ability of PfCRT to transport amantadine. We found that oocytes expressing PfCRT K1 or 76I-PfCRT K1 showed a marked increase in [ 3 H]amantadine accumulation relative to noninjected oocytes (Fig 3B and 3C). By contrast, oocytes expressing 76K-PfCRT K1 or 163R,356V-PfCRT K1 failed to take up [ 3 H]amantadine above the background level ]amantadine transport attributable to PfCRT was calculated by subtracting the background level of accumulation (i.e., the average of the uptake measured in noninjected oocytes and oocytes expressing 76K-PfCRT K1 ) from that measured for each of the oocyte types. The rates of amantadine uptake (nmol per oocyte/h) in noninjected oocytes and PfCRT K1 -expressing oocytes were 5.4 ± 0.9 and 12 ± 2.8, respectively. The asterisks denote a significant difference from the noninjected control: *P < 0.05; **P < 0.01; ***P < 0.001 (one-way ANOVA). (D) Trans-stimulation of PfCRT-mediated [ 3 H]CQ transport by unlabeled amantadine. The oocytes were microinjected with a buffer control or with buffer containing amantadine, spermine, or histidine. The estimated intracellular concentrations ([compound] i ) are indicated. The asterisks denote a significant difference from the relevant buffer-injected control. (E) Concentration-dependence of the trans-stimulation of [ 3 H]CQ uptake by amantadine. The oocytes were microinjected with buffer containing amantadine to achieve an estimated [amantadine] i of 1 to 20 mM. The app K m and app V max values are the apparent kinetic parameters for the trans-stimulatory effect of amantadine. The data show CQ uptake above that measured in the relevant buffer-injected control; the total rates of CQ uptake are presented in S6 Fig. The noninjected data overlays the data obtained with oocytes expressing PfNT1, PfCRT 3D7 , or 76K-PfCRT K1 . (F) A magnified plot of the 76I-PfCRT K1 and 76I,369F-PfCRT K1 data from panel E. In all panels, the data are the mean ± SEM of five independent experiments (performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. Where not shown, error bars fall within the symbols. of accumulation, and 76I,369F-PfCRT K1 displayed only a very low (but statistically significant) level of amantadine transport activity. Moreover, [ 3 H]amantadine transport via PfCRT K1 , 76I-PfCRT K1 , and 76I,369F-PfCRT K1 was inhibited by the PfCRT CQR inhibitors verapamil and saquinavir. These findings suggest that hypersensitivity to amantadine arises from the ability of PfCRT CQR isoforms to transport amantadine, and that mutations such as T76K, S163R, and V369F reverse this response by substantially decreasing or abolishing amantadine transport activity.
The signal-to-background ratio we obtained for [ 3 H]amantadine transport in the oocyte system was around 2 (Fig 3B), which is relatively modest when compared with that obtained for quinine and quinidine transport (which typically produce ratios of 4-7; e.g., Fig 2B  We therefore interrogated the amantadine transport properties of PfCRT further by conducting a series of trans-stimulation experiments. Many transporters reorientate more quickly from one face of the membrane to the other when a substrate is bound compared with when the transporter is empty. Transporters that display this characteristic can be trans-stimulated; in the case of PfCRT CQR , the uptake of [3H]CQ from the external solution will be accelerated by the presence of an unlabeled substrate on the cytosolic face of the membrane. We therefore measured the uptake of [ 3 H]CQ into oocytes which had been microinjected with a buffer control or with buffer containing unlabeled amantadine, spermine, or histidine. Spermine (a polycation) and histidine (a weak base) do not appear to interact with PfCRT Dd2 (S5 Fig and Martin et al. [30]) and were therefore included as extra negative controls. In addition, oocytes expressing an unrelated P. falciparum transporter (the nucleoside transporter PfNT1 [48]) were included as a further negative control. We found that amantadine did not affect the accumulation of [ 3 H]CQ in noninjected oocytes or oocytes expressing 76K-PfCRT K1 , PfCRT 3D7 , or PfNT1 ( Fig 3D). Moreover, neither spermine nor histidine altered [ 3 H]CQ uptake in any of the oocyte types. By contrast, amantadine trans-stimulated the transport of [ 3 H]CQ into oocytes expressing PfCRT K1 , PfCRT Dd2 , 76I-PfCRT K1 , or 76I,369F-PfCRT K1 , albeit to different extents; the respective increases in the rate of CQ influx (in pmol per oocyte/ h) were 6.4 ± 0.2, 6.7 ± 0.3, 0.68 ± 0.04, and 0.43 ± 0.02. We extended this analysis by measuring the concentration-dependence of the trans-stimulation of PfCRT K1 , 76I-PfCRT K1 , and 76I,369F-PfCRT K1 (Fig 3E and 3F and S6 Fig). A least-squares fit of the Michaelis-Menten equation to the data yielded apparent kinetic parameters for the trans-stimulatory effect of amantadine (listed in Fig 3). These results revealed that the addition of V369F to 76I-PfCRT K1 causes a 6.2-to 7.2-fold increase in the concentration of amantadine required for the half-maximal trans-stimulation of the transporter.
Taken together, our work shows that PfCRT variants from parasites that are hypersensitive to amantadine possess the ability to transport this drug, and that the introduction of mutations that suppress amantadine hypersensitivity either abolish (T76K and S163R) or substantially decrease (V369F) the protein's capacity for amantadine transport. These findings led us to propose that amantadine, which should sequester within the DV via weak-base trapping, exerts its main antiplasmodial effect outside of the DV and that the PfCRT-mediated leak of the drug back into the cytosol results in amantadine hypersensitivity. We therefore utilized a set of parasite assays to test this hypothesis in situ.

PfCRT effluxes amantadine from the DV of amantadine-hypersensitive parasites
The function of PfCRT can be investigated in its native environment by employing an assay that indirectly detects the movement of protonated drugs out of the DV [34][35][36]. This method uses a fluorescent pH-sensitive probe to measure an outward leak of protons, which manifests as an increase in the rate of alkalinization of the DV. A drug-induced proton leak arises when a weakbase drug enters the acidic DV in its unprotonated form and is effluxed in its protonated form. We applied this assay to a set of P. falciparum transfectants (the C2 GC03 , C4 Dd2 , and C6 7G8 lines [2]) that are isogenic except for their pfcrt allele, which encodes either PfCRT 3D7 (C2 GC03 ) or a PfCRT CQR isoform of the protein (PfCRT Dd2 in C4 Dd2 and PfCRT 7G8 in C6 7G8 ). CQ was included as a positive control and consistent with previous observations [34][35][36]46], it increased the rate of DV alkalinization in the CQ-resistant C4 Dd2 and C6 7G8 lines and was without effect in the CQ-sensitive C2 GC03 line (Fig 4). We obtained similar results with amantadine and also showed that verapamil inhibits the amantadine-associated leak of protons from the DV of C4 Dd2 and C6 7G8 parasites. These findings confirm that amantadine accumulates within the DV Amantadine is a substrate of PfCRT Dd2 and PfCRT 7G8 , and not of wild-type PfCRT, in situ. The rate of concanamycin A-induced DV alkalinization (expressed as the inverse of the half-time for DV alkalinization) in the C2 GC03 (CQ-sensitive, expressing PfCRT 3D7 ), C4 Dd2 (CQ-resistant, expressing PfCRT Dd2 ), and C6 7G8 (CQ-resistant, expressing PfCRT 7G8 ) transfectant lines was measured in the absence and presence of amantadine (AMT), CQ, or verapamil (VP). The effect of amantadine on the DV alkalinization rate was also measured in the presence of verapamil (final concentration of 50 μM). The drugs were added to suspensions of saponin-isolated trophozoite-stage parasites containing fluorescein-dextran in their DVs 4 min before the addition of concanamycin A (100 nM). CQ was added at a concentration of 2.5 μM and the concentration of amantadine was 10 μM. Consistent with previous studies, the addition of CQ elevated the rate of alkalinization in both of the CQ-resistant lines and had a small buffering effect in the C2 GC03 parasites [34][35][36]46]. The biological basis for the difference in the rate of CQ-dependent alkalinization between the two CQ-resistant lines is currently unclear, but could be due to differences in (1) the expression of PfCRT, (2) the transport properties of the two PfCRT isoforms, (3) the volume of the DV, and/or (4) the concentration of PfCRT's natural substrates within the DV. The data are the mean + SEM of five independent experiments (performed on different days). The asterisks denote a significant difference from the relevant solvent control: ***P < 0.001 (one-way ANOVA). via weak-base trapping and also provide in situ evidence of the ability of PfCRT CQR isoforms to transport amantadine back into the cytosol.
If amantadine hypersensitivity results from the PfCRT-mediated efflux of the drug from the DV, we would expect that inhibitors of PfCRT CQR would reduce this response. We tested this hypothesis by determining the susceptibility of the isogenic lines to amantadine in the absence and presence of verapamil or chlorpheniramine [49] (another established inhibitor of PfCRT CQR ). These experiments, which included CQ as a control as well as CQ-resistant (Dd2) and CQ-sensitive (3D7) reference strains, entailed using a fluorescence-based method to measure parasite growth in the presence of increasing concentrations of amantadine or CQ. The resulting IC 50 s (Table 3) provided two salient findings. First, the data verified the striking hypersensitivity of CQ-resistant parasites to amantadine. Secondly, in all cases this hypersensitivity was partially suppressed by the PfCRT inhibitors. Thus, we conclude that the hypersensitivity of the 106/1 76I line, and of other CQ-resistant parasites, to amantadine arises from the PfCRT CQR -mediated redistribution of the drug from the DV into the cytosol, where it gains better access to its main antiplasmodial target.

Discussion
Our work provides mechanistic explanations for the patterns of inverse susceptibility induced by PfCRT. First, we confirmed our previous observation [42] of there being a positive correlation between the capacity of a given PfCRT isoform for mediating CQ transport and the magnitude of CQ resistance achieved by the respective parasite (S7A Fig) and extended this relationship to include, with the notable exception of the QN-hypersensitive lines, a positive correlation between the quinine or quinidine transport activity of PfCRT and the parasite's in vitro responses to these drugs (Fig 5A and S7B Fig). Moreover, we showed that in most cases, the isoforms of PfCRT K1 that possessed CQ transport activity also transported quinine and quinidine and vice versa. Together, these findings confirm a common role for PfCRT in reducing the accumulation of CQ, quinine, and quinidine within the DV and explain the tendency of CQ-resistant parasites to exhibit decreased susceptibilities to quinine and quinidine. However, The P values determined from a one-way ANOVA were less than 0.001 for comparisons with the relevant CQ-sensitive strain or line (and within either the CQ or amantadine treatment).
The P values determined from a one-way ANOVA were less than e 0.001, f 0.01, or g 0.05 for comparisons between the control treatment and the relevant VP or CP treatment from the same strain or line. our results resolve this phenomenon further by providing a fundamental insight into PfCRTinduced drug phenotypes: whether a given isoform of PfCRT alters the parasite's susceptibility to a drug, and to what extent and in which direction, depends on the kinetics of the drug's transport via PfCRT.
A key example of this principle is the quinine-hypersensitivity conferred by 76I-PfCRT K1 . Our work reveals that the introduction of 76I into PfCRT K1 has the remarkable effect of transforming the protein into an exceedingly high-affinity, low-capacity transporter of quinine. These highly abnormal kinetic properties will cause the drug to clog PfCRT's substrate-binding cavity, which should block the transport of the natural substrate and thereby prevent the protein from fulfilling its essential physiological role (Fig 5A). Thus, we propose that hypersensitivity to quinine results from the drug exerting at least two killing effects in 106/1 76I parasitesanti-hemozoin and anti-PfCRT CQR . This hypothesis provides mechanistic explanations for two previously abstruse observations: (1) despite dramatic differences in their quinine responses, 106/1 76I and 106/1 parasites accumulate similar levels of quinine [11] and (2) quinine and CQ produce a synergistic interaction in 106/1 76I parasites, whereas a slightly antagonistic interaction occurs in the 106/1 76T line [12]. In the case of the first observation, the lack of a difference in quinine accumulation concurs with the exceedingly low capacity of 76I-PfCRT K1 for transporting quinine. In regard to the second phenomenon, the synergistic interaction in 106/1 76I is readily reconcilable with the ability of quinine to adhere to the binding cavity of 76I-PfCRT K1 , as this should both inhibit the normal function of the protein and block The drug would therefore remain in the parasite's DV where it exerts an anti-hemozoin effect that kills the parasite, which is consistent with the QN-sensitive (S) status of the respective lines. PfCRT K1 (76T) is able to transport QN out of the DV and thereby imparts low-level resistance (low-R) to QN. By contrast, 76I-PfCRT K1 has an extremely high affinity for QN coupled with an extremely low maximum rate of transport. This causes QN to clog the binding site of 76I-PfCRT K1 , thereby blocking the transport of the natural substrate. Hence, the QN-hypersensitivity (hyper-S) observed in 106/1 76I parasites results from QN exerting two killing effectsanti-hemozoin and anti-PfCRT CQR . The gain of a positively-charged residue at position 72 or 352 (76I R/K) prevents the interaction of the transporter with QN and returns the parasites to QN-sensitive status. (B) Amantadine (AMT) is a relatively poor inhibitor of both 76I-PfCRT K1 and 76I,369F-PfCRT K1 , making it unlikely that the AMT-hypersensitivity of CQ-resistant parasites is due to an anti-PfCRT CQR effect. The isoforms of PfCRT from AMThypersensitive parasites (PfCRT K1 and 76I-PfCRT K1 ) have the ability to transport this weak-base drug out of the DV (where it accumulates) whereas those from AMT-sensitive parasites either do not possess significant AMT transport activity (e.g., 76K-PfCRT K1 and 163R,356V-PfCRT K1 ; R/K) or transport AMT with low affinity and low capacity (76I,369F-PfCRT K1 ). The data therefore converge on a scenario in which AMT exerts its main antimalarial activity in the cytosol and AMT-hypersensitivity arises from the redistribution of the drug from the DV into the cytosol via a PfCRT CQR variant (e.g., PfCRT K1 or 76I-PfCRT K1 ). Mechanisms for PfCRT-Induced Drug Hypersensitivity the efflux of CQ from the DV. The slightly antagonistic effect observed in the 106/1 76T parasites is likely to result from the two drugs competing to bind to heme (CQ is the more potent inhibitor of hemozoin formation [25,27]) and/or other differences in their activities (e.g., CQ also appears to inhibit the glutathione-dependent degradation of heme, whereas quinine is thought to lack this activity [50]).
Cooper and colleagues reported a third perplexing characteristic of the 106/1 76I line that may likewise arise from quinine's ability to block the normal function of 76I-PfCRT K1 . The presence of verapamil typically re-sensitizes CQ-resistant parasites to CQ (and to quinine and quinidine) and this resistance-reversing effect was evident for all three drugs in the 106/1 76T and 106/1 76N lines [11,12]. It was also apparent in the 106/1 76I parasites, but only in the CQ and quinidine treatments. In the quinine treatment, verapamil exerted the opposite effect; i.e., it decreased the sensitivity of the 106/1 76I line to quinine and thus partially suppressed the parasite's hypersensitivity to this drug [11,12]. One possible explanation for this highly unusual observation is a scenario in which verapamil competes with quinine for binding to 76I-PfCRT K1 , and that the natural substrate is better able to access the translocation pore when verapamil is bound relative to when quinine is bound. Given that the substrate-binding site of PfCRT Dd2 has been shown to behave as a large polyspecific cavity that can bind at least two molecules simultaneously [29], it is conceivable that verapamil and quinine interact with the binding cavity differently and that verapamil presents somewhat less of an obstacle to the transport of the natural substrate via 76I-PfCRT K1 than does quinine. Such a scenario would provide a mechanistic explanation for the verapamil-induced reduction in quinine's activity against the 106/1 76I line.
We recently reported that CQ-resistant parasites are hypersensitive to dimers of quinine, and that this appears to be due to these molecules inhibiting both heme detoxification and PfCRT CQR function (the quinine dimers are potent inhibitors of PfCRT Dd2 but are not translocated by the transporter) [46]. Hence, the anti-PfCRT CQR activity of quinine in the 106/1 76I parasites is not simply an intriguing but extraneous biological oddity; it signifies that one of the parasite's key modulators of drug resistance is itself a druggable antimalarial target.
Our findings suggest that less dramatic changes in the kinetics of transport via PfCRT can also significantly affect the parasite's response to a drug. 76N-PfCRT K1 decreases the parasite's susceptibility to CQ and quinine while being without effect on its response to quinidine [11,12,14]. A simple interpretation of these observations would be that 76N-PfCRT K1 transports CQ and quinine but fails to recognize quinidine. However, we found that 76N-PfCRT K1 maintains the ability to transport quinidine, but that its affinity for the drug is slightly reduced, and that its maximum rate of quinidine transport is significantly reduced, relative to the kinetics of quinidine transport via PfCRT K1 . The low-capacity, low-affinity nature of quinidine transport via 76N-PfCRT K1 will limit the protein's ability to reduce the accumulation of the drug within the DV and hence could explain why 76N-PfCRT K1 has little net effect on the parasite's susceptibility to quinidine (S7B Fig). The hypersensitivity of CQ-resistant parasites to amantadine was first reported several decades ago [13], but the molecular basis for this phenomenon has remained unknown. We demonstrate that amantadine hypersensitivity cannot be explained by an anti-PfCRT CQR effect and is instead attributable to a second mechanism. Our work shows that amantadine sequesters within the acidic environment of the DV and that the hypersensitivity of CQ-resistant parasites to amantadine results from the ability of PfCRT CQR isoforms to efflux the drug from the DV into the parasite cytosol (Fig 5B). These findings indicate that amantadine acts on a target outside of the DV and that it exerts its antiplasmodial activity to greater effect when proteins such as PfCRT Dd2 , PfCRT 7G8 , PfCRT K1 , and 76I-PfCRT K1 leak the accumulated drug back into the cytosol. The antiviral target of amantadine is a proton channel [51] and it is possible that the drug likewise targets an essential cation channel or transporter in the parasite.
All of the mutations that introduced a positively-charged residue into PfCRT K1 abolished its ability to transport CQ, quinine, quinidine, and amantadine. Hence, in addition to confirming the importance of electrostatic repulsion in preventing interactions between PfCRT and protonated drugs, our results suggest that the transporter's ability to transport protonated quinolines out of the DV (and thereby confer quinoline resistance) is fundamentally connected to its ability to transport protonated amantadine out of the DV (and thereby induce amantadine hypersensitivity). Furthermore, the detection in field isolates of T76K, C72R, or S163R in otherwise PfCRT CQR isoforms [5] indicates that these types of changes could be occurring in the parasite population in response to selection forces exerted by other drugs and/or because, in the absence of CQ, these parasites are fitter than their CQ-resistant counterparts (most PfCRT CQR isoforms impart a fitness cost [52][53][54][55]). Such a phenomenon appears to have taken place in French Guiana; Pelleau and colleagues [56] recently showed that another mutation that introduces a positive charge into PfCRT CQR (C350R) emerged in a CQ-resistant population following the withdrawal of CQ and is attributed with reinstating these parasites to CQsensitive status.
Our findings also provide a foundation for understanding and exploiting clinically-relevant cases of reciprocal collateral sensitivity. Most of the current treatments for uncomplicated malaria are combination therapies that pair an artemisinin derivative (all of which are metabolised into dihydroartemisinin) with a quinoline-related partner drug-of which the most widely used is lumefantrine [57]. Cases of severe malaria, as well as Plasmodium-infected pregnant women, are typically treated with combinations that include an antibiotic, such as clindamycin. It is, therefore, worth noting that several in vitro studies have observed isoforms of PfCRT CQR to induce hypersensitivity to lumefantrine (with 1.5-3.2-fold decreases in the IC 50 [16,17,56,[58][59][60][61]), artemisinin and/or dihydroartemisinin (with 1.9-3.7-fold decreases in the IC 50 [6,11,17,58,59,62]) as well as to a number of antibiotics [10,[63][64][65][66][67]-including clindamycin [10,63,65,68]. Moreover, multiple clinical trials undertaken in malarious regions throughout the world have associated the wild-type protein (i.e., PfCRT 3D7 ) with significant reductions in the parasite's susceptibility to artemether-lumefantrine [16,[69][70][71], such that (1) the administration of artemether-lumefantrine was found to cause significant selection of CQsensitive parasites carrying PfCRT 3D7 and (2) the presence of parasites carrying PfCRT 3D7 prior to artemether-lumefantrine treatment was associated with an increased risk of recrudescence. It is possible that the hypersensitivity of CQ-resistant parasites to lumefantrine, the artemisinins, and/or to the antibiotics results from one of the mechanisms we describe in this study. For instance, the antibiotics are either known or expected to act on targets outside of the DV [63,72,73], but due to their weak-base nature, these drugs will sequester within the acidic environment of this organelle. Hence, it is plausible that the antibiotics gain greater access to their antimalarial target by being transported into the cytosol via PfCRT CQR . For these antibiotics as well as for lumefantrine and the artemisinins, and for the many other pharmacons that display enhanced activity against CQ-resistant parasites, the set of assays outlined in our study offer the means to determine which of the two mechanisms are involved, or whether an altogether different mechanism is responsible. Such insights have the potential to contribute to the formulation of rational approaches for maintaining and extending the useful lifespan of many antimalarials by exploiting the opposing selection forces they exert upon PfCRT.

Generation of PfCRT coding sequences and synthesis of cRNA
The coding sequences of the different isoforms of PfCRT were generated via site-directed mutagenesis using the primer pairs listed in S1 Table and an approach described previously [42]. The mutations were introduced into a codon-harmonized version of the PfCRT sequence that had been inserted into the pGEM-He-Juel oocyte expression vector [74]. This sequence encodes a version of PfCRT that is free of endosomal-lysosomal trafficking motifs and which is therefore expressed at the plasma membrane of Xenopus laevis oocytes [30,42]. All of the resulting coding sequences were verified by sequencing (undertaken by the ACRF Biomolecular Resource Facility, ANU). The plasmids were linearized with SalI (ThermoFisher Scientific) and 5'-capped complementary RNA (cRNA) was synthesized using the mMessage mMachine T7 transcription kit (Ambion), and then purified with the MEGAclear kit (Ambion).
Harvest, preparation, and microinjection of X. laevis oocytes Ethical approval of the work performed with the X. laevis frogs was obtained from the Australian National University Animal Experimentation Ethics Committee (Animal Ethics Protocol Number A2013/13) in accordance with the Australian Code of Practice for the Care and Use of Animals for Scientific Purposes. Oocytes were harvested and prepared as described in full elsewhere [9]. Briefly, sections of ovary were harvested from adult female frogs (purchased from NASCO) via a minor surgical procedure and single, de-folliculated oocytes were prepared using collagenase D (Roche). Stage V-VI oocytes were microinjected with cRNA (20 ng per oocyte) encoding PfCRT or PfNT1 and were stored at 16-18°C in OR 2+ buffer (82.5 mM NaCl, 2.5 mM KCl, 1 mM MgCl 2 , 1 mM Na 2 HPO 4 , 5 mM HEPES, 1 mM CaCl 2 , and 50 μg/mL gentamycin; pH 7.8).

Oocyte membrane preparation and western blot analysis
The preparation of oocyte membranes and the semi-quantification of PfCRT protein was carried out using a protocol described in detail elsewhere [42]. Protein samples prepared from oocyte membranes were separated on a 4-12% Bis-Tris SDS-polyacrylamide gel (Life Technologies) and transferred to a Protran 0.45 μM nitrocellulose membrane (Amersham, GE Healthcare Life Sciences). The membranes were probed with rabbit anti-PfCRT antibody (concentration of 1:4,000; Genscript) followed by horseradish peroxidase-conjugated goat anti-rabbit antibody (1:8,000; Life Technologies, cat. no. 656120). Validation of the specificity of the anti-PfCRT antibody has been published in detail elsewhere [42]. The PfCRT band for each variant was detected by chemiluminescence (Pierce), quantified using the Image J software [75], and expressed as a percentage of the intensity measured for the PfCRT K1 band. Total protein staining was used to evaluate sample loading and efficiency of transfer as outlined previously [42]. Between five and seven independent experiments were performed (on oocytes from different frogs), and in each experiment measurements were averaged from two independent replicates.

Immunofluorescence of oocytes expressing PfCRT
Immunofluorescence analyses were performed on oocytes three days post-injection using a method adapted from Weise et al. [76]. Unless specified otherwise, both the incubation and wash steps were conducted at room temperature with gentle shaking or rotation. The volume of the incubation solution was 500 μL and the washes were performed with 1 mL of the specified solution. Six oocytes from each treatment type were fixed for 30 min in a solution of phosphate-buffered saline (PBS) and 4% v/v paraformaldehyde and then washed (10 min) three times in PBS. The oocytes were permeabilized with 100% methanol for 20 min (-20°C, without shaking) and washed (10 min) three times in PBS. A blocking solution (4% w/v bovine serum albumin (BSA), 2% v/v normal goat serum (Life Technologies), and 0.1% v/v Triton X-100 in PBS) was applied for 2 h, after which the oocytes were incubated overnight at 4°C in a second blocking solution (4% w/v BSA and 2% v/v normal goat serum in PBS). The samples were then incubated for a further 4 h at room temperature before the blocking solution was replaced with a solution containing the rabbit anti-PfCRT antibody (1:100 in 1.5% w/v BSA and 0.01% v/v Triton X-100 in PBS) and the samples were incubated for 4 h at room temperature and then overnight at 4°C. Three 10-min washes were performed in PBS supplemented with 1.5% w/v BSA and all of the remaining steps were undertaken in the dark (and at room temperature). The Alexa Fluor 488 donkey anti-rabbit antibody (1:500; Molecular Probes, cat. no. A-21206) was incubated with the samples for 4 h in PBS supplemented with 4% w/v BSA and 2% v/v normal goat serum, after which three 10-min washes were performed in PBS.
The oocytes were post-fixed with paraformaldehyde (3.7% v/v in PBS) for 30 min, washed twice (15 min) in 2 mL of PBS, and then dehydrated with a series of incubations in solutions of increasing ethanol content. The solutions (in order of administration) were: 30% v/v ethanol in PBS, 50% v/v ethanol in PBS, 70% v/v ethanol in ultrapure water, 90% v/v ethanol in ultrapure water, and 100% ethanol. In each case, the samples were briefly washed in the ethanol-containing solution before a 15-min incubation was performed. A further two incubations in 100% ethanol were conducted before the oocytes were embedded in an acrylic resin using the Technovit 7100 plastic embedding system (Kulzer). Briefly, the samples were incubated with 500 μL embedding solution (50% v/v Technovit 7100 in 100% ethanol) for 2 h, after which the oocytes were incubated overnight in 500 μL of a second embedding solution (1% w/v Technovit 7100 'hardener 1' in Technovit 7100). A further two incubations (!2 h each) were performed in the second embedding solution before all of the solution was removed and 800 μL of a third embedding solution (6.66% v/v Technovit 7100 'hardener 2' and 1% w/v Technovit 7100 'hardener 1' in Technovit 7100) was added. After the samples had set (!4 d), a microtome was used to obtain~4 μm slices, which were dried on microscope slides. Coverslips with a drop of Pro-Long Gold Antifade Mountant (Life Technologies) were placed over the slices and sealed with nail polish.
Images of the slices were obtained with a Leica Sp5 inverted confocal laser microscope (Leica Microsystems) using the 63x objective. Excitation was achieved with a 488 nm argon laser and the emissions were captured using a 500-550 nm filter. Images were acquired using the Leica Application Suite Advanced Fluorescence software (Leica Microsystems). At least two independent experiments were performed (on oocytes from different frogs) for each oocyte type, within which slices were examined from a minimum of three oocytes. All of the slices taken from oocytes expressing a PfCRT variant displayed a fluorescent band above the pigment layer (i.e., consistent with the localization of PfCRT to the plasma membrane) that was not present in noninjected oocytes.

Drug transport assays
The radiolabeled drugs were purchased from American Radiolabeled Chemicals ( mL polystyrene round bottom tube (Falcon) and washed twice with 3.5 mL of ND96 buffer, with the residual buffer removed by pipette. Influx commenced with the addition of 100 μL of ND96 buffer supplemented with the radiolabeled and unlabeled drug and, where specified, an unlabeled inhibitor (e.g., verapamil, saquinavir, or CQ). The assay was terminated by removing the reaction buffer with a pipette and washing the oocytes twice with 3.5 mL of ice-cold ND96 buffer. Each oocyte was transferred to a separate well of a white 96-well plate (NUNC or PerkinElmer), incubated overnight at room temperature in 30 μL of 10% SDS, and mixed on an orbital shaker the following day for approximately 5 min. The lysed oocyte was then combined with 150 μL of MicroScint-40 microscintillant (PerkinElmer), the plate covered with a TopSeal-A (PerkinElmer), and the radioactivity measured with a PerkinElmer MicroBeta 2 microplate liquid scintillation analyzer.
Note that noninjected oocytes and oocytes expressing PfCRT 3D7 take up [ 3 H]CQ to similar (low) levels via simple diffusion of the neutral species of the drug; this represents the 'background' level of [ 3 H]CQ accumulation in oocytes (refer to Martin et al. [30] for full data and discussion). There is also no detectable difference in the accumulation of [ 3 H]quinine between noninjected and PfCRT 3D7 -expressing oocytes, nor does the accumulation of [ 3 H]quinidine differ between these two oocyte types, and this background level of uptake has likewise been attributed to simple diffusion [29,30].
The kinetic parameters for CQ, quinine, and quinidine transport via different isoforms of PfCRT (Table 1 and 50 )c] to the data, where y is PfCRTmediated CQ transport, y min and y max are the minimum and maximum values of y, and c is a fitted constant. PfCRT-mediated CQ transport was calculated by subtracting the uptake measured in the control oocytes (oocytes expressing 76K PfCRT K1 ) from that in oocytes expressing 76I-PfCRT K1 or 76I,369F-PfCRT K1 .
A subset of experiments measured the ability of unlabeled amantadine to trans-stimulate the uptake of [ 3 H]CQ into the oocyte. Immediately prior to the commencement of the experiment, the oocytes (days 3-4 post-injection of the cRNA) were microinjected with either amantadine (estimated intracellular concentrations of 1, 2.5, 5, 10, 15, or 20 mM) or a control treatment (buffer, spermine, or histidine; the estimated intracellular concentrations of spermine and histidine were 5 and 10 mM). The volume injected was 50 nL and the intracellular concentrations were calculated using previous estimates of the volume of stage V-VI oocytes (~400 nL [77]). The resealed oocytes were then incubated at 16-18°C in OR 2+ buffer for approximately 5 min and the influx of [ 3 H]CQ was measured as described above. For each amantadine concentration, the rate of CQ influx above that measured in the relevant buffer-injected control was calculated and a least-squares fit of the Hill equation (y = V max [amantadine] i n /(K m n + [amantadine] i n ) to the data was performed in SigmaPlot, where y is the reaction velocity and n is the Hill coefficient. This analysis yielded Hill coefficients of 1.63 ± 0.12, 1.49 ± 0.07, and 1.46 ± 0.05 for PfCRT K1 , 76I-PfCRT K1 and 76I,369F-PfCRT K1 , respectively. These values indicated that CQ and amantadine either bind independently of one another, or are slightly cooperative (i.e., the binding of one drug enhances the affinity of the transporter for the second drug). The Michaelis-Menten equation was then fitted to the data to derive the kinetic parameters for the transstimulation of [ 3 H]CQ transport by amantadine.
In most cases, at least five independent experiments were performed (on different days and using oocytes from different frogs), and within each experiment measurements were made from 10 oocytes per treatment.

Culture of P. falciparum-infected erythrocytes
The use of human blood in this study was approved by the Australian National University's Human Research Ethics Committee. The CQ-sensitive strain '3D7' (isolated from the Netherlands but probably of African origin [78]), the CQ-resistant strain 'Dd2' (isolated from Southeast Asia), and three pfcrt transfectant lines [2] (C2 GC03 , C4 Dd2 , and C6 7G8 ) were cultured and synchronized as described previously [79,80]. In the C4 Dd2 and C6 7G8 lines, the wild-type pfcrt allele of the CQ-sensitive 'GC03' strain has been replaced with the pfcrt allele from Dd2 or from the CQ-resistant '7G8' strain (isolated from Brazil), respectively. C2 GC03 is a CQ-sensitive recombinant control that retains the wild-type pfcrt allele (i.e., PfCRT 3D7 ). C6 7G8 contains an additional mutation (I351M) in PfCRT that is not present in 7G8 parasites [36]. The parasite lines were maintained in the presence of the selection agents blasticidin (5 μM; Sigma-Aldrich) and WR99210 (5 nM; Jacobus Pharmaceuticals). These selection agents were not present during the experiments.

P. falciparum H + efflux assay
Saponin-isolated trophozoite-stage parasites containing the membrane-impermeant pH-sensitive fluorescent indicator fluorescein-dextran (10,000 MW; Life Technologies) in their DVs were prepared as outlined elsewhere [36]. The isolated parasites were washed and suspended in a saline solution (125 mM NaCl, 5 mM KCl, 1 mM MgCl 2 , 20 mM glucose, 25 mM HEPES; pH 7.1) at a density of 1-3 × 10 7 cells/mL. The fluorometry experiments were performed as described previously [35]. The pH of the DV was monitored at 37°C using a PerkinElmer Life Sciences LS50B fluorometer with a dual excitation Fast Filter accessory (excitation 490 and 450 nm; emission 520 nm). The experiments entailed monitoring the alkalinization of the DV upon addition of the V-type H + -ATPase inhibitor concanamycin A (100 nM; Sigma-Aldrich), in the presence or absence of the drugs of interest. Half-times for DV alkalinization (t1/2) were calculated as outlined elsewhere [34]. In all cases, five independent experiments were performed on different days.

P. falciparum drug susceptibility
Parasite proliferation was measured in 96-well plates using a fluorescent DNA-intercalating dye [81] and a protocol described in detail previously [82]. Briefly, cell suspensions containing erythrocytes infected with ring-stage parasites (hematocrit and parasitemia of approximately 2% and 1%, respectively) were incubated at 37°C for 72 h. The samples were frozen, thawed, and then processed by the addition of 100 μL (per well) of SYBR Safe DNA Gel Stain (Molecular Probes; 0.2 μL/mL) in a lysis buffer (20 mM Tris, 5 mM EDTA, 0.008% w/v saponin, and 0.08% v/v Triton X-100; pH 7.5). The fluorescence emanating from each well was measured immediately using a Tecan Infinite M1000 PRO microplate reader (excitation 490 nm; emission 520 nm) and the average fluorescence from wells containing the highest concentration of the drug was subtracted from the resulting values. The level of parasite proliferation in the presence of each drug concentration was expressed as a percentage of the proliferation measured in the absence of the drug for which the IC 50 was being determined. The IC 50 s were determined in SigmaPlot by a least-squares fit of the equation y = a/ [1 ([drug]/IC 50 )c] to the data, where y is the percent parasite proliferation, a is the maximum change in the percent parasite proliferation, and c is a fitted constant. In all cases, five independent experiments were performed (on different days), and within each experiment measurements were averaged from three replicates.

Statistics
All errors cited in the text and shown in the figures represent the SEM. Statistical comparisons were made using one-way ANOVAs in conjunction with Tukey's multiple comparisons test.
Supporting Information S1 Fig. Predicted topology of PfCRT showing the mutations present in the PfCRT K1 variants characterized in this study. PfCRT is predicted to contain 10 α-helical transmembrane domains (TMDs) and to be orientated in the digestive vacuole (DV) membrane with the Nand C-termini extending into the parasite cytosol [83]. The positions of the mutated residues in PfCRT K1 are indicated with black circles. The key CQ resistance-associated mutation (K76T) is represented as a red square. The purple circles show the locations of the additional residues that are mutated in the variants of PfCRT K1 . The box attached to each polymorphic residue lists the (non-wild-type) amino acid(s) that occur at that position. The predicted roles of the TMDs are as follows: 4 and 9 (outlined in dark green) are implicated in the binding and translocation of substrates, TMDs 3 and 8 (boxed in light green) are thought to assist in the binding and translocation of the substrate and may also influence the substrate-specificity of the transporter, TMDs 1, 2, 6, and 7 (boxed in black) may be involved in recognizing and discriminating between substrates, and TMDs 5 and 10 (outlined in mid-green) are thought to play a role in the formation of homo-dimers [83]. The CQ, quinine, or quinidine transport activity of a given PfCRT K1 isoform (calculated from the data presented in Fig 2) was plotted against the in vitro resistance index for the relevant drug and parasite line (listed in Table 1). Where not shown, error bars fall within the symbols. There was a positive correlation between the in vitro response of a parasite line to CQ and the CQ transport properties of the corresponding isoform of PfCRT K1 (panel A: R 2 = 0.856). This is consistent with our previous observation of a positive correlation (R 2 = 0.858; [42]) in an analysis performed with seven field isoforms of PfCRT, but contrasts with the work of Roepe and colleagues, who have not detected a relationship between PfCRT CQRmediated CQ transport and the parasite's response to CQ [84,85]. In panel B, the quinine transport activities of many of the PfCRT K1 isoforms correlated positively with the in vitro responses of the parasites to this drug. However, two key exceptions included the data for the highly QN-hypersensitive lines 106/1 76I and 106/1 76I,369F . Likewise, in panel C, a positive correlation between the capacity of a given PfCRT K1 isoform to transport quinidine and the relevant parasite's in vitro quinidine resistance index held for all of the parasites bar the QD-sensitive line 106/1 76N . Refer to the main text for a discussion of these outlying points. Note also that other genetic elements, such as the amplification or mutation of the parasite's multidrug resistance protein 1 (PfMDR1) and the altered expression of other genes, can contribute to the quinoline resistance phenotype [7,[86][87][88][89][90]. Hence, the CQ resistance index obtained for the 106/1 76I line, which is higher than what might be expected from the CQ transport activity of 76I-PfCRT K1 , may be due to changes in the expression of one or more genes, including pfmdr1 [14,89]. High-level CQ resistance has been associated with reductions in the pfmdr1 copy number and a corresponding decrease in the expression of pfmdr1 [91,92]. In this regard it is worth noting that the 106/1 76I line contains a single copy of pfmdr1, whereas the 106/1, 106/ 1 76T , and 106/1 76N lines possess two copies, and this difference has been confirmed to result in a lower level of pfmdr1 expression in the 106/1 76I parasites [14,89]. Hence, it is likely that the relatively high level of CQ resistance displayed by 106/1 76I is the product of the effects of the CQ transport activity of 76I-PfCRT K1 , the reduced expression of PfMDR1, and perhaps changes that have occurred elsewhere in the parasite's transcriptome. Thus, while this study and our previous work [42] indicate that the quinoline transport activities of a number of PfCRT variants correlate positively with the in vitro responses of the respective parasites to these drugs, changes elsewhere in the genome and transcriptome can modulate the level of quinoline resistance conferred by PfCRT CQR . (TIF) . In all cases, the rate of PfCRT-mediated drug uptake was calculated by subtracting the rate measured in oocytes expressing 76K-PfCRT K1 from that measured in oocytes expressing PfCRT K1 , 76N-PfCRT K1 , 76I-PfCRT K1 , or 76I,369F-PfCRT K1 at each drug concentration. The Michaelis-Menten equation was fitted to the resulting data using nonlinear regression. The rates of drug uptake are the mean ± SEM of multiple independent experiments (performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. Where not shown, error bars fall within the symbols. The exact n values, as well as the kinetic parameters derived from these data, are presented in Table 2 The rates of uptake (pmol per oocyte/h) in noninjected oocytes and oocytes expressing PfCRT Dd2 were 0.10 ± 0.01 and 0.44 ± 0.04, respectively, for quinine and 0.05 ± 0.01 and 0.30 ± 0.07, respectively, for quinidine. In both panels, drug uptake is expressed relative to that measured in oocytes expressing PfCRT Dd2 . The rates of quinine and quinidine uptake mediated by PfCRT K1 were 1.2-1.3 times that measured for PfCRT Dd2 , whereas the 76N-PfCRT K1 protein possessed 3.1 times the quinine transport activity, and 3.2 times the quinidine transport activity, of 76N-PfCRT Dd2 . Moreover, the rate of quinidine transport mediated by 76I-PfCRT K1 was 1.5 times that measured for 76I-PfCRT Dd2 . The single exception to this trend was the rate of quinine transport via 76I-PfCRT K1 , which was 0.55 times that measured for its PfCRT Dd2 counterpart. The data are the mean + SEM of at least five independent experiments (performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. The asterisks denote a significant difference in drug uptake between the noninjected treatment and that measured in oocytes expressing a variant of PfCRT: Ã P < 0.05; ÃÃ P < 0.01; ÃÃÃ P < 0.001 (one-way ANOVA).  1 and 2 mM). The assays were conducted at pH 5.5 and in the presence of 15 μM unlabeled CQ. The rates of CQ uptake (pmol per oocyte/h) in noninjected oocytes and PfCRT Dd2 -expressing oocytes were 1.41 ± 0.02 and 24 ± 1.3, respectively. The data are the mean + SEM of four independent experiments performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. 'ns' denotes no significant difference from the PfCRT Dd2 control (P > 0.05; one-way ANOVA). (TIF) S6 Fig. Amantadine trans-stimulates the uptake of [ 3 H]CQ into oocytes expressing PfCRT in a concentration-dependent manner. Control oocytes (noninjected oocytes and oocytes expressing PfNT1) and oocytes expressing a variant of PfCRT (PfCRT 3D7 , 76K-PfCRT K1 , PfCRT K1 , 76I-PfCRT K1 , or 76I,369F-PfCRT K1 ) were microinjected with buffer containing amantadine to achieve an estimated intracellular concentration ([amantadine] i ) of 1 to 20 mM. A control was also performed in which the oocytes were microinjected with buffer alone. The rates of CQ uptake are the mean ± SEM of five independent experiments (performed using oocytes from different frogs), within which measurements were made from 10 oocytes per treatment. Where not shown, error bars fall within the symbols. The data presented in Fig 3E  and 3F were calculated by subtracting the rate of CQ uptake measured in the buffer-injected control from that measured in each of the corresponding amantadine treatments (and within the same oocyte type). The noninjected data overlays the data obtained with oocytes expressing PfNT1, PfCRT 3D7 , or 76K-PfCRT K1 . (TIF) S7 Fig. Mechanisms for the CQ and quinidine susceptibilities conferred by PfCRT isoforms in the malaria parasite. (A) The variants of PfCRT K1 that contain 72R, 76K, 163R, 352K, or 352R (R/K) do not possess significant CQ transport activity. The drug would therefore remain in the DV where it exerts an anti-hemozoin effect that kills the parasite, which is consistent with the CQ-sensitive (S) status of the respective lines. By contrast, PfCRT K1 (76T) and 76I-PfCRT K1 transport CQ out of the parasite's DV, thereby conferring CQ resistance (R). The addition of 369F to 76I-PfCRT K1 significantly reduces its affinity and capacity for CQ transport. Hence, 76I,369F-PfCRT K1 imparts a relatively low level of resistance (low-R) to CQ. (B) The R/K variants of PfCRT K1 do not possess significant quinidine (QD) transport activity whereas PfCRT K1 , 76N-PfCRT K1 and 76I-PfCRT K1 each have the ability to transport QD out of the DV, albeit to varying degrees. These differences in QD transport activity explain, at least in part, the susceptibilities of the corresponding 106/1 parasite strains to QD. 76N-PfCRT K1 has a slightly lower affinity for QD then does PfCRT K1 , and also has a much lower maximum rate of QD transport. The low capacity and low affinity of QD transport via 76N-PfCRT K1 explains why this protein has little net effect on the parasite's sensitivity to QD. By contrast, the relatively high capacity of PfCRT K1 for QD transport is consistent with the decreased susceptibility of 106/1 76T parasites to QD (low-R). 76I-PfCRT K1 has a low maximum rate of QD transport, but this characteristic is counterbalanced by a 3-fold increase in its affinity for QD, and the net capacity of the protein for QD transport appears to be sufficient to reduce the parasite's susceptibility to QD (at least under the conditions of the in vitro parasite proliferation assays). (TIF) S1 Table. Primer sequences used to introduce mutations into the PfCRT coding sequence via site-directed mutagenesis. (DOCX)