Iron is a substrate of the Plasmodium falciparum chloroquine resistance transporter PfCRT in Xenopus oocytes

The chloroquine resistance transporter of the human malaria parasite Plasmodium falciparum, PfCRT, is an important determinant of resistance to several quinoline and quinoline-like antimalarial drugs. PfCRT also plays an essential role in the physiology of the parasite during development inside erythrocytes. However, the function of this transporter besides its role in drug resistance is still unclear. Using electrophysiological and flux experiments conducted on PfCRT-expressing Xenopus laevis oocytes, we show here that both wild-type PfCRT and a PfCRT variant associated with chloroquine resistance transport both ferrous and ferric iron, albeit with different kinetics. In particular, we found that the ability to transport ferrous iron is reduced by the specific polymorphisms acquired by the PfCRT variant as a result of chloroquine selection. We further show that iron and chloroquine transport via PfCRT is electrogenic. If these findings in the Xenopus model extend to P. falciparum in vivo, our data suggest that PfCRT might play a role in iron homeostasis, which is essential for the parasite's development in erythrocytes.

uole (1), a proteolytic organelle involved in hemoglobin degradation and a target for antimalarial drugs. Diverse PfCRT variants have emerged under drug selection in different parts of the world and are associated with altered susceptibility of the parasite to a wide range of quinoline and quinoline-like antimalarial drugs, including the former first line drug chloroquine and the currently used antimalarials amodiaquine, quinine, and lumefantrine (2)(3)(4).
PfCRT confers altered drug phenotypes by acting as an efflux carrier (5)(6)(7)(8), expelling compounds from the digestive vacuole, where these drugs are thought to interfere with endogenous heme detoxification pathways. The transport of drugs seems to occur in association with protons and to be dependent on the membrane potential, suggesting a model of PfCRT as a protoncoupled and potential-driven facilitating carrier (9 -11). The drug-transporting function of PfCRT is dependent on an amino acid substitution of threonine for lysine at position 76 that is conserved among geographic PfCRT variants (12,13). The replacement of a positively charged amino acid with an uncharged residue is thought to allow positively charged drugs, such as chloroquine, to enter the binding pocket. In total, geographic PfCRT variants harbor between 4 and 10 mutations (2). The additional mutational changes augment the drug-transporting activity and/or balance it with fitness costs (13,14).
In contrast to our deepening understanding of the role of PfCRT in drug resistance, the physiological function of PfCRT has remained largely enigmatic. Sequence analyses and homology searches have grouped PfCRT among the drug/metabolite carrier superfamily based on the protein topology, with a predicted number of 10 transmembrane domains and an internal 2-fold pseudosymmetry (15,16). However, robust in silico predictions on the physiological function of PfCRT have been hampered by a paucity of closely related homologs with annotated function in other species. An early report suggesting that PfCRT resembles bacterial voltage-gated chloride channels could not be confirmed (17). Maughan et al. (18) reported a resemblance of PfCRT with a family of thiol transporters in Arabidopsis thaliana, termed PfCRT-like transporters (CLTs) that export glutathione and its precursor ␥-glutamylcysteine from the chloroplast where ␥-glutamylcysteine is exclusively made. On the basis of this finding, and taking into account that mutations in both CLTs and PfCRT affect glutathione homoeostasis in A. thaliana and P. falciparum, a function of PfCRT in glutathione transport was proposed (18 -20). Other studies have implicated PfCRT in polyspecific cation transport and, in addition to glutathione, proposed also choline; tetraethyl ammonium (TEA); the amino acids histidine, lysine, and arginine; and the polyamines spermine and spermidine as putative substrates (21). However, these studies have remained controversial mainly because none of these organic cations were able to compete with chloroquine for transport via PfCRT in a functional heterologous expression system, based on oocytes from the South African clawed toad Xenopus laevis, which display PfCRT at their plasma membrane (5).
Here we have investigated the physiological function of PfCRT expressed in X. laevis oocytes. Our data suggest that both wild-type PfCRT and a PfCRT variant associated with chloroquine resistance confer electrogenic iron transport. PfCRT does not discriminate absolutely between ferric and ferrous iron, setting PfCRT apart from the mammalian divalent metal transporter 1 (DMT1) and other known iron-transporting systems (22,23). However, the PfCRT variant associated with chloroquine resistance displayed kinetic parameters with regard to ferrous iron that were distinct from those of the wild type.

PfCRT is an electrogenic carrier of chloroquine
To study the transport function of PfCRT, we expressed the wild-type form of the protein from the chloroquine-sensitive strain HB3 and the polymorphic variant from the chloroquineresistant P. falciparum strain Dd2 in X. laevis oocytes. The PfCRT variant from the Dd2 strain differs from the wild-type protein in 8 amino acid substitutions, including the conserved K76T mutation. Expression was verified by Western analyses and immunofluorescence assays, using a guinea pig antiserum specific to PfCRT (Fig. 1, A and B). Western blot analysis identified a protein with the expected molecular mass of PfCRT of 48.6 kDa, and immunofluorescence images revealed co-localization of PfCRT with the oolemma marker inositol 1,4,5triphosphate receptor (24) in both PfCRT HB3 -and PfCRT Dd2expressing oocytes (Fig. 1, A and B). Water-injected oocytes did not react with the PfCRT-specific antiserum (Fig. 1, A and B). Quantification of the specific signals in the Western analysis suggests comparable expression levels of both PfCRT variants, relative to the internal ␣-tubulin standard (Fig. 1A), consistent with previous reports (13).
Several studies have suggested that drug transport via PfCRT is associated with protons (10,11,25). To explore the possibility of an electrogenic transport process, we studied the electrophysiological effect of chloroquine on PfCRT HB3and PfCRT Dd2 -expressing oocytes, using the two-electrode voltage clamp technique. Water-injected oocytes were analyzed in parallel. Oocytes were kept in a low-Na ϩ -containing buffer, pH 7.5 (10 mM NaCl and 86 mM NMDG chloride, henceforth termed ND10 buffer) from the moment of RNA injection to reduce the impact of the non-selective cationic conductance activated in PfCRT-expressing oocytes on electrophysiological parameters (26). The activation of the non-selective cationic conductance would lead to a low membrane potential and an elevated intracellular free Na ϩ concentration (supplemental Table S1) (26). Prior to the electrophysiological experiments, oocytes were pre-equilibrated for 30 min in ND10 buffer, pH 6.0. This protocol fully maintained verapamil-sensitive chloroquine transport activity in PfCRT Dd2 -expressing oocytes, as verified in flux experiments, using radiolabeled chloroquine and by comparing the amount of chloroquine taken up by oocytes kept in ND10 or ND96 buffer (Fig. 1C). Moreover, oocytes incubated in ND10 buffer had intracellular ATP levels in the normal range of 1-2 mM (27,28) (Fig. 1D). These findings demonstrate that the oocytes incubated in ND10 buffer are viable and healthy, consistent with previous reports (29 -31).
Chloroquine (100 M) induced an inward current in oocytes expressing PfCRT Dd2 clamped at a membrane potential (V c ) of Ϫ50 mV (p Ͻ 0.001 compared with water-injected oocytes, Figure 1. Expression of PfCRT in X. laevis oocytes. A, Western blot analysis of total lysates from oocytes injected with water, PfCRT HB3 cRNA, and PfCRT Dd2 cRNA, using a polyclonal guinea pig antiserum specific to PfCRT and a polyclonal rabbit antiserum specific to ␣-tubulin. A size standard is indicated in kilodaltons. The hybridization signals were subsequently quantified, yielding the expression levels of PfCRT HB3 and PfCRT Dd2 relative to the internal standard ␣-tubulin. The means Ϯ S.E. (error bars) from the number of independent biological replicates N BR ϭ 4 independent Western blot analyses are shown. B, confocal fluorescence images of fixed PfCRT Dd2 and PfCRT Dd2 -expressing oocytes and water-injected control oocytes. Left, fluorescence image of InsP 3 R, using a specific rabbit antisera and an Alexa Fluor 546 anti-rabbit secondary antibody. Middle, fluorescence image of PfCRT, using a specific guinea pig antiserum and the Alexa Fluor 488 anti-guinea pig secondary antibody. Right, differential interference contrast image. Bar, 250 m. C, effect of ND10 and ND96 buffer, pH 5.5, on PfCRT Dd2 -mediated chloroquine uptake in the presence and absence of verapamil (VP; 100 M). Oocytes were incubated in the respective buffer, pH 7.5, from the moment of cRNA injection. For the flux experiments, chloroquine was added at a final concentration of 50 M unlabeled chloroquine and 42 nM [ 3 H]chloroquine, and the amount of uptake was determined after 60 min of incubation. The means Ϯ S.E. of N BR ϭ 4 independent biological determinations are shown. Statistical significance was assessed using the two-tailed t test. D, intracellular ATP levels of water-injected oocytes and oocytes expressing PfCRT HB3 or PfCRT Dd2 . Oocytes were incubated in ND10 buffer, pH 7.5, after cRNA injection, for 3 days before the intracellular ATP levels were determined. The mean Ϯ S.E. of N BR ϭ 5 independent biological determinations are shown. The statistical significance was assessed using the Holm-Sidak one-way ANOVA test or a two-tailed t test, where appropriate. n.s., not significant.
according to Kruskal-Wallis one-way ANOVA on ranks test), and not in PfCRT HB3 expressing oocytes or in water-injected oocytes ( Fig. 2A and supplemental Fig. S1A). The chosen chloroquine concentration of 100 M corresponds to approximately half the reported Michaelis-Menten constant K m of chloroquine for PfCRT Dd2 of 245 M, as determined in the oocyte system (5,6,13). The chloroquine-induced currents were sensitive to verapamil (100 M), as demonstrated in paired experiments ( Fig. 2A and supplemental Fig. S1A; p Ͻ 0.001, according to a paired t test). Verapamil is an established full mixed-type inhibitor of PfCRT and a chemosensitizer of chloroquine resistance in vitro (6,32). The finding that chloroquine induced an inward current in PfCRT Dd2 -expressing oocytes and not in PfCRT HB3 -expressing oocytes or in water-injected controls was further supported by current-voltage relationships, showing a chloroquine-induced increase in basal membrane conductance G m (Fig. 2B). The increase in basal membrane conductance was maintained in the presence of diphenylamine 2-carboxylic acid (DPC; 1 mM), an established inhibitor of the endogenous non-selective cation conductance (33, 34) (supplemental Fig. S2). These findings, together with the observation that chloroquine uptake by PfCRT Dd2 -expressing oocytes was independent of DPC (1 mM) (Fig. 2C), exclude the possibility that the chloroquine-induced currents are related to or conferred by the endogenous nonselective cation conductance activated by the heterologous expression of PfCRT. Instead, our findings are consistent with electrogenic transport of chloroquine via PfCRT Dd2 .

Iron-induced currents in PfCRT-expressing oocytes
The electrogenic nature of drug transport via PfCRT offered the opportunity to screen for putative substrates using a previously established generic approach based on exposing oocytes to a complex mixture of metabolites as is present in bacto-yeast extract (BYE) and recording substrate-induced electric signals in a two-electrode voltage-clamped set up (35). Application of BYE (0.5% in ND10 buffer, pH 6.0) immediately induced a significant inward current in clamped (V c ϭ Ϫ50 mV) PfCRT Dd2expressing oocytes, comparable with the currents induced by chloroquine ( Fig. 3A and supplemental Fig. S1B; p Ͻ 0.001, according to Kruskal-Wallis one-way ANOVA on ranks test). Interestingly, BYE also induced a negative current in PfCRT HB3 Figure 2. Chloroquine-induced currents in PfCRT Dd2 -expressing oocytes. A, water-injected oocytes (white circles), PfCRT HB3 -expressing oocytes (gray circles), and PfCRT Dd2 -expressing oocytes (black circles) were voltage-clamped (Ϫ50 mV) while superfused with ND10 buffer, pH 6.0. The currents induced upon the addition of 100 M chloroquine (CQ) were measured before and after supplementation of the medium with 100 M VP in paired experiments. The lines connecting two data points indicate the response of a single oocyte before and after the treatment. The number of oocytes (n) investigated is indicated above the x axis. The medians are indicated as red lines. The statistical significance was assessed using the Kruskal-Wallis one-way ANOVA on ranks test or the paired t test, where appropriate. The corresponding p values are indicated on the graphs. n.s., not significant. B, current-voltage relationships of water-injected oocytes (left), PfCRT HB3 -expressing oocytes (middle), and PfCRT Dd2 -expressing oocytes (right) were first obtained in ND10 buffer (red-filled circles) and then ND10 buffer containing 100 M chloroquine (blue-filled circles) before adding 100 M VP (green-filled inverted triangles). I/V curves upon withdrawal of both compounds are indicated by yellow-filled triangles. Each data point represents the mean Ϯ S.E. (error bars) from n ϭ 15 oocytes. C, PfCRT Dd2 -mediated chloroquine uptake in the presence and absence of 1 mM DPC. PfCRT Dd2 -expressing oocytes and water-injected oocytes were incubated in ND96 buffer supplemented with and without 1 mM DPC and containing 42 nM [ 3 H]chloroquine and 50 M unlabeled chloroquine for 60 min at 25°C. To calculate the amount of PfCRT Dd2 -mediated transport, the amount of chloroquine taken up by water-injected oocytes was subtracted. The data represent the means Ϯ S.E. of N BR ϭ 4 independent biological determinations. n.s., not significant according to a t test.

PfCRT transports iron
expressing oocytes, which was statistically not different from the BYE-induced currents observed in PfCRT Dd2 -expressing oocytes (according to Kruskal-Wallis one-way ANOVA on ranks test) ( Fig. 3A and supplemental Fig. S1B). The BYE-induced currents were partly but significantly prevented by verapamil in both PfCRT Dd2 -and PfCRT HB3 -expressing oocytes (p Ͻ 0.001, according to a paired t test, for both PfCRT Dd2 and PfCRT HB3 ), as shown in paired experiments ( Fig. 3A and supplemental Fig. S1B). BYE did not induce major currents in water-injected oocytes ( Fig. 3A and supplemental Fig. S1B), consistent with previous reports (35). These findings are consistent with an electrogenic transport process, with both the wild-type and the mutated PfCRT interacting with at least one component present in BYE.
Given that BYE contains several of the proposed substrates of PfCRT, including positively charged amino acids, thiols, polyamines, and oligopeptides, we tested whether one or several of them induced an inward current. However, neither arginine, lysine, histidine, nor glutathione was active in our system at a concentration of 1 mM (supplemental Fig. S3A). We further observed no responses with choline, TEA (a characteristic substrate of organic cation transporters), or glycylsarcosine (a surrogate substrate for mammalian oligopeptide carriers) at a concentration of 2 mM (supplemental Fig. S3A). We validated these negative results for each of these putative substrates in flux assays, using radiolabeled compounds at a concentration close to and ϳ10-fold below the previously reported K m of these organic biomolecules for PfCRT (21). None were taken up by PfCRT-expressing oocytes (supplemental Fig. S3, A and B). We further tested the negatively charged amino acids glutamic acid and aspartic acid and the apolar amino acids leucine and proline. Again, no electrical signals and no uptake were observed (supplemental Fig. S3, A and B). Furthermore, we found no evidence for facilitated uptake of the polyamines putrescine or spermine by PfCRT-expressing oocytes (supplemental Fig. S3,  A and B).
Considering that the digestive vacuole is the site of hemoglobin degradation and that the physiological function of PfCRT might directly or indirectly relate to this biological process, we inspected the list of BYE ingredients for other products of hemoglobin degradation that might be transported by PfCRT. An obvious candidate is iron, of which BYE contains 55.3 g g Ϫ1 dry weight (36). To test whether the PfCRT-associated currents are induced by iron, we compared, in paired experiments, the effect of BYE and BYE supplemented with the iron-chelating agent deferoxamine (DFO; 100 M) on the amplitude of the currents. DFO significantly prevented the BYE-induced inward currents in both PfCRT HB3 -and PfCRT Dd2 -expressing oocytes (p Ͻ 0.001, according to a paired t test) ( Fig. 3B and supplemental Fig. S1C).
The suggestion of the currents being iron-inducible was subsequently verified by superfusing the oocytes with BYE-free ND10 buffer containing iron at a concentration of 5 M, which corresponds to the amount of iron present in a 0.5% BYE solution. To distinguish between effects brought about by ferrous (Fe 2ϩ ) and ferric iron (Fe 3ϩ ), an ND10 buffer system was used that was supplemented with ascorbic acid or nitrilotriacetic acid (NTA) to maintain iron in its reduced Fe 2ϩ or its oxidized Fe 3ϩ state, respectively (22). The addition of 5 M iron induced, in voltage-clamped cells (V c ϭ Ϫ50 mV), an inward current in both PfCRT HB3 -and PfCRT Dd2 -expressing oocytes (p Ͻ 0.01 compared with water-injected oocytes, according to Kruskal-Wallis one-way ANOVA on ranks test), but only in the case of Fe 2ϩ and not Fe 3ϩ (Fig. 4, A and B). Current-voltage relationships confirmed the G m increase in the presence of ferrous iron in PfCRT-expressing oocytes (Fig. 4C).
At an iron concentration of 1 mM, PfCRT-associated currents were also observed for ferric iron in voltage-clamped cells (V c ϭ Ϫ50 mV), although to a lesser extent relative to the Fe 2ϩinduced currents at the same concentration (Fig. 4, D and E, and supplemental Fig. S4). Importantly, in all cases, the ferrous and ferric iron-induced currents could be partly inhibited by verapamil (100 M) (p Ͻ 0.01, according to paired t tests; Fig. 4 and supplemental Fig. S4). Water-injected oocytes did not respond to the addition of ferrous or ferric iron ( Fig. 4 and supplemental Fig. S4), consistent with previous reports (22). To exclude the possibility that the currents were carried by potassium, magnesium, calcium, or NMDG, the respective salts were replaced in the ND10 buffer by an equal osmotic concentration of mannitol in the physiological experiments using 1 mM iron.

Evidence of iron transport via PfCRT
We next investigated the interaction of PfCRT with ferrous and ferric iron in flux assays, using radiolabeled 55 Fe. The uptake of iron by PfCRT-expressing and water-injected oocytes was measured in an acidic ND96 buffer (pH 5.5) containing 96 mM NaCl and either ascorbic acid or NTA. The buffers were validated using oocytes expressing rat DMT1, which exclusively transports ferrous and not ferric iron (supplemental Fig.  S5) (22).
Oocytes expressing PfCRT HB3 and PfCRT Dd2 took up significantly more iron from the external medium containing 2 M radioactive 55 Fe and 48 M non-radioactive 54 Fe, both in its Fe 2ϩ and Fe 3ϩ form, than did water-injected oocytes (p Ͻ 0.05,

PfCRT transports iron
compared with water-injected oocytes, according to Holm-Sidak one-way ANOVA test) (Fig. 5, A and B, left panels). The specific PfCRT-attributable portion of ferrous and ferric iron uptake increased in a linear fashion with time for at least the first 90 min (Fig. 5, A and B, right panels). Ferrous and ferric iron uptake by PfCRT-expressing oocytes was pH-dependent and increased as the pH became more acidic (Fig. 5, C and D).
Verapamil (100 M) inhibited the uptake of ferrous and ferric iron by PfCRT-expressing oocytes by ϳ50 -60% (Fig.  6, A and B). Furthermore, both iron species were able to block PfCRT Dd2 -mediated chloroquine uptake (Fig. 6C). The extent of iron taken up by PfCRT-expressing oocytes increased in a hyperbolic manner with the amount of PfCRT-encoding cRNA injected, as exemplified for ferrous iron (Fig. 6D). Uptake A and B, water-injected oocytes (white circles), PfCRT HB3 -expressing oocytes (gray circles), and PfCRT Dd2 -expressing oocytes (black circles) were voltage-clamped (Ϫ50 mV) while superfused with ND10 buffer, pH 6.0. Iron was maintained in divalent form or trivalent form by adding to the buffer a quadruple ratio of L-ascorbic acid or nitrilotriacetic acid, respectively. The currents induced by adding 5 M Fe 2ϩ (A) or 5 M Fe 3ϩ (B) were measured before and after supplementation of the medium with 100 M VP. The number (n) of oocytes is indicated above the x axis. The medians are indicated as red lines. The statistical significance was assessed using the Kruskal-Wallis one-way ANOVA on ranks test or paired t test, where appropriate. The corresponding p values are indicated on the graphs. C, current-voltage relationships of water-injected oocytes and PfCRT HB3 -expressing oocytes (left) and water-injected oocytes and PfCRT Dd2 -expressing oocytes (right) were obtained in ND10 buffer, pH 6.0, in the presence and absence of 5 M ferrous iron. Each data point represents the mean Ϯ S.E. (error bars) of n ϭ 12-15 oocytes. D and E, as in A and B, but this time, a concentration of 1 mM ferrous (D) or ferric (E) iron in ND10-mannitol buffer, pH 6.0, was used. n.s., not significant.

PfCRT transports iron
of chloroquine was measured in parallel assays using the ferrous and specific buffer (Fig. 6D), and the data obtained were consistent with previous findings (5,13).
Both ferrous and ferric iron are known to create reactive oxygen radicals via the Fenton reaction. Such reactive oxygen radicals can initiate lipid peroxidation, resulting in membrane damage (37). To assess whether the presence of iron during the course of an uptake experiment adversely affected the membrane integrity of the oocytes, we preincubated PfCRT Dd2 -expressing oocytes in the appropriate buffer containing increasing concentrations of ferrous or ferric iron, ranging from 0 to 400 M. After 60 min of incubation, oocytes were washed and placed in ND96 buffer containing 42 nM tritiated and 50 M unlabeled chloroquine for 60 min. No differences in chloroquine uptake were noted between iron-pretreated and untreated oocytes (supplemental Fig. S6), suggesting that pretreatment with iron, even at a dose of 400 M for 1 h, did not affect membrane integrity or the activity of PfCRT Dd2 .
We further considered the possibility of iron uptake occurring via the non-selective cation conductance that is activated in PfCRT-expressing oocytes (26). As a result of the activated non-selective cation conductance, PfCRT-expressing oocytes revealed an enhanced Na ϩ influx when incubated in ND96 buffer, relative to water-injected oocytes, as determined using 22 Na (Fig. 7A). The addition of DPC (1 mM) blocked 22 Na influx via the non-selective cation conductance (Fig. 7A), consistent with previous reports (26). In comparison, DPC did not affect uptake of ferrous or ferric iron by PfCRT-expressing oocytes ( Fig. 7B), suggesting an uptake mechanism independent of the non-selective cation conductance.
To investigate whether uptake of iron by PfCRT-expressing oocytes displays saturation kinetics, we incubated water-injected and PfCRT-expressing oocytes with a constant amount of radioactive 55 Fe (2 M) and different concentrations of nonradioactive 54 Fe, ranging from 0 to 400 M for 60 min before the amount of internalized radiolabel was determined. As seen in Fig. 8 (A and B), the amount of ferrous and ferric iron taken up by PfCRT-expressing oocytes approached a plateau as the total iron concentration increased, consistent with a saturable transport mechanism. Fitting the Michaelis-Menten equation to the kinetic data yielded estimates of the apparent Michaelis-Menten constant K m and the maximal transport velocity V max . Interestingly, the kinetic parameters for ferrous iron significantly differed between PfCRT HB3 -and PfCRT Dd2 -expressing oocytes (p Ͻ 0.001 and p ϭ 0.02 for V max and K m values, respectively, according to a two-tailed t test) ( Table 1). PfCRT Dd2 had approximately 3 times lower K m and V max values relative to PfCRT HB3 (K m ϭ 20 Ϯ 9 M (number of independent biological replicates N BR ϭ 5) and 70 Ϯ 17 M (N BR ϭ 5), and V max ϭ 4 Ϯ 0.5 pmol oocyte Ϫ1 h Ϫ1 (N BR ϭ 5) and 13 Ϯ 1 pmol oocyte Ϫ1 h Ϫ1 (N BR ϭ 5), respectively). In contrast, the kinetic parameters for ferric iron were not significantly different between the two PfCRT variants (K m ϭ 170 Ϯ 50 M (N BR ϭ 5) and 130 Ϯ 40 M (N BR ϭ 5), and V max ϭ 14 Ϯ 2 pmol oocyte Ϫ1 h Ϫ1 (N BR ϭ 5) and 20 Ϯ 2 pmol oocyte Ϫ1 h Ϫ1 (N BR ϭ 5), respectively; p ϭ 0.5 and p ϭ 0.05, respectively; according to two-tailed t-tests).

PfCRT transports iron
When assessing the substrate specificity of each PfCRT variant, it is apparent that PfCRT HB3 displayed comparable K m values for both ferrous and ferric iron (p ϭ 0.2, according to a two-tailed t test) but significantly different V max values (p Ͻ 0.01, according to a two-tailed t test) ( Table 1). In comparison, PfCRT Dd2 revealed both significantly different K m and V max values with regard to ferrous and ferric iron (p ϭ 0.02 and p Ͻ 0.001, respectively, according to a two-tailed t test) (Table 1). Thus, the mutations that distinguish PfCRT Dd2 from PfCRT HB3 and bring about the drug-transporting function affect the carrier's handling of ferrous iron, whereas the ferric iron transport seems to be largely unaffected by these mutational changes.

Discussion
Our data show that both the wild-type and a PfCRT variant associated with chloroquine resistance are able to transport ferrous and ferric iron when functionally expressed in X. laevis

PfCRT transports iron
oocytes. This finding is supported by both electrophysiological and flux studies.
To identify putative substrates of PfCRT, we employed a recently established generic approach based on exposing oocytes expressing a transporter of unknown functionality to the complex metabolite mixture of a diluted bacto-yeast extract solution and recording substrate-induced electric signals in a two-electrode voltage-clamped setup (35). This approach, as elegant as it is, has one precondition; the transport process must be electrogenic to record an electrical signal. Although previous studies have proposed that PfCRT transports chloroquine together with protons (10, 11), the electrogenic nature of this transport process has yet to be formally demonstrated. To record such electrical currents, we had to suppress endogenous conductances activated by the expression of PfCRT (26) (supplemental Table S1). Activation of endogenous conductances is regularly observed in oocytes expressing heterologous proteins (38 -41).
With the activity of the non-selective cation conductance reduced and intracellular Na ϩ overload prevented, we were able to show that the addition of chloroquine elicited a measureable inward rectifying current in voltage-clamped PfCRT Dd2 -expressing oocytes, and not in PfCRT HB3 -expressing oocytes or water-injected oocytes. We excluded the possibility that these chloroquine-induced currents are related to the activated non-selective cation conductance, on the basis of a lack of responsiveness to the inhibitor DPC (supplemental Fig.  S2). In contrast, the chloroquine-induced currents were significantly blocked by the addition of verapamil, the well-studied full mixed type inhibitor of chloroquine transport on PfCRT Dd2 ( Fig. 2A) (6). Thus, our findings causatively link the electrical signals elicited by chloroquine to the activity of PfCRT and support a model of PfCRT Dd2 acting as an electrogenic transporter of chloroquine. We explain the negative chloroquineinduced currents by the influx of net positive charges into the oocyte, which is consistent with PfCRT-mediated chloroquine transport in association with H ϩ in the parasite (10,11).
Establishing the electrogenic transport activity of PfCRT Dd2 enabled a global screen for physiological substrates, based on diluted BYE and the recording of BYE-induced electrical signals. Indeed, BYE induced measurable signals of a verapamilsensitive inward current in both PfCRT Dd2 -and PfCRT HB3expressing oocytes, but not in water-injected oocytes (Fig. 3A). We initially thought that these BYE-induced currents might be carried by basic amino acids, which, according to a previous study, are putative substrates of PfCRT (21). However, neither histidine, lysine, nor arginine at a concentration of up to 1 mM was able to reproduce the BYE-induced currents in PfCRTexpressing oocytes. Flux experiments conducted with tritiated histidine, lysine, or arginine corroborated the conclusion that PfCRT does not act on these positively charged amino acids. We also could not confirm previous studies suggesting a role of PfCRT in the transport of glutathione, the polyamines spermine and putrescine, choline, tetraethyl ammonium, or glycylsarcosine (supplemental Fig. S3, A and B) (20,21). The latter two substances are typical substrates of organic cation transporters and oligopeptide carriers, respectively (42,43). None of these substances induced currents in, or were taken up by, PfCRTexpressing oocytes. Whereas our findings contrast with a functional role of PfCRT as a polyspecific cation transporter, they are consistent with previous studies showing that none of these putative substrates are able to compete with chloroquine for transport via PfCRT Dd2 (5).
We suspect that these diverging findings arose from the investigation of different PfCRT constructs. Studies proposing a role of PfCRT in polyspecific cation transport used a PfCRT fusion protein that was modified at both the N-and C-terminal end by the addition of YbeL and a His 6 tag (21). YbeL is an uncharacterized E. coli protein of 120 amino acids. It is possible that fusing PfCRT at both ends to a heterologous, possibly enzymatically active polypeptide affected its conformation and membrane topology and hence its functionality and substrate specificity. In contrast, the PfCRT constructs used herein have been validated in several functional studies, providing fundamental insights into PfCRT-mediated drug transport processes (5,6,13,44,45), although, admittedly, they harbor discreet amino acid replacements needed to remove putative lysosomal trafficking and sorting motifs and ensure expression at the oocyte plasma membrane (5).
After excluding several currently discussed substrates of PfCRT, we explored the possibility of iron being a substrate of PfCRT. The rationale behind this idea was the high concentration of metabolically released iron in the parasite's digestive vacuole of 53 fg/digestive vacuole (46) and the fact that BYE contains 55.3 g g Ϫ1 dry weight of ionic iron (36), which amounts to a total iron concentration of 5 M in the 0.5% BYE bufferusedintheelectrophysiologicalexperiments.Severalindependent lines of evidence suggest that ionic iron is, indeed, a substrate of PfCRT. First, there is evidence from our electrophysiological studies, which show that the addition of the iron chelator DFO to the buffer largely eliminated the signal induced by BYE (Fig. 3B). Moreover, both the ferric and ferrous forms of iron induced inward currents in both PfCRT HB3 -and . The uptake of radiolabeled iron into water-injectedoocytesandoocytesexpressingPfCRT Dd2 orPfCRT HB3 wasmeasured in the appropriate ND96 buffer, pH 5.5, containing an extracellular concentration range of 2-400 M iron (2 M radiolabeled 55 Fe and the appropriate amount of non-radioactive iron). The amount of PfCRT-mediated iron uptake was calculated by subtracting the amount measured in water-injected oocytes from that in oocytes expressing PfCRT at each iron concentration. A least-squares fit of the Michaelis-Menten equation to the resulting data yielded the kinetic parameters compiled in Table 1. The data represent the means Ϯ S.E. (error bars) of N BR ϭ 5 independent biological determinations, with n ϭ 12-30 oocytes per treatment and biological replicate.

PfCRT transports iron
PfCRT Dd2 -expressing oocytes (but not in the control waterinduced oocytes) (Fig. 4). These iron-induced signals were partially reduced by the addition of verapamil.
Flux experiments substantiated our conclusion that the ironinduced signals represented a transport of iron. As shown in Figs. 5 and 8, PfCRT expression in oocytes led to a significant uptake of both iron forms. Importantly, the uptake of both Fe 2ϩ and Fe 3ϩ was saturable in oocytes expressing either PfCRT Dd2 or PfCRT HB3 (Fig. 8). The saturable nature of the uptake kinetics is consistent with carrier-mediated transport of the iron, but it is difficult to reconcile with iron entering the oocyte via a nonspecific leak or via the endogenous non-selective cation conductance. Further excluding a role of the endogenous nonselective cation conductance, iron fluxes were independent of DPC (Fig. 7B). Importantly, iron transport via PfCRT HB3 or PfCRT Dd2 was sensitive to verapamil and could be blocked by 50 -60% (Fig. 6, A and B). Moreover, chloroquine transport via PfCRT Dd2 could be competed for by increasing concentrations of both ferrous and ferric iron (Fig. 6C).
Our finding that verapamil acted on both PfCRT HB3 and PfCRT Dd2 might be surprising. However, a functional assay to interrogate the interaction of PfCRT with verapamil has thus far existed only for the chloroquine resistance-mediating PfCRT variants, thus precluding consensus on whether or not verapamil inhibits wild-type PfCRT. Only now, with the finding of PfCRT HB3 -expressing oocytes responding to BYE and iron in both electrophysiological and flux assays, have such tools become available. On the basis of these data, we propose that verapamil binding is an intrinsic feature of PfCRT, one that is not limited to certain variants, albeit the affinity to verapamil might reveal variant-specific characteristics. This conclusion is consistent with previous reports showing that verapamil, as a partial mixed type inhibitor of PfCRT Dd2 , binds to a site distinct from that of chloroquine and other drugs (6).
PfCRT Dd2 -and PfCRT HB3 -expressing oocytes exhibit somewhat different kinetic parameters for Fe 2ϩ transport (Table 1). There was a highly significant difference between the V max values for Fe 2ϩ between the PfCRT HB3 -and PfCRT Dd2 -expressing oocytes, but this difference for Fe 3ϩ was only of borderline significance. Similarly, the K m values were significantly different, as between the PfCRT HB3 -and PfCRT Dd2 -expressing oocytes for Fe 2ϩ , but not so for Fe 3ϩ . Apparently, PfCRT Dd2 has an impaired Fe 2ϩ iron transporting activity, whereas its Fe 3ϩ activity is maintained compared with PfCRT HB3 . This might explain the fitness cost incurred by P. falciparum strains harboring PfCRT Dd2 or some other geographic PfCRT variants in the absence of a drug pressure (14,47).
The finding that the two PfCRT variants differed in their ability to transport iron links iron transport to the molecular features of PfCRT itself. That this difference is more marked in the case of Fe 2ϩ might point to ferrous iron as being the form that is a substrate of PfCRT. The individual or combined contribution of the eight amino acid substitutions distinguishing PfCRT Dd2 from PfCRT HB3 to the differential iron transport activity is not yet clear and must await crystallographic validation. However, one might extrapolate findings from DMT1 to PfCRT. In the case of DMT1, an asparagine residue seems to coordinate ferrous iron binding (48). On the basis of this observation, one might speculate that the N75E and the N326S substitutions present in PfCRT Dd2 selectively affect the ability to transport ferrous iron.
We have to qualify our findings in the sense that they are solely based on functional studies conducted on PfCRT expressed in the heterologous X. laevis oocyte system. The X. laevis oocyte system is a validated tool that has provided fundamental insights into PfCRT-mediated drug transport processes (5,6,13,44,45). However, recent studies have revealed that PfCRT is posttranslationally modified in the parasite by phosphorylation at four sites and by palmitoylation at one site (49,50). Whether these posttranslational modifications are conserved in the oocyte systems is unknown, as is the effect of posttranslational modifications on the functionality of PfCRT. We are, therefore, aware that our findings await validation in the parasite. However, the tools to demonstrate iron transport across the parasite's digestive vacuolar membrane have yet to be developed.
If PfCRT is indeed a transporter of iron, what might be its role in the parasite's physiology? The parasite metabolizes extensively the hemoglobin of its host cell and, in doing so, liberates as much as 58 fg of heme-associated iron per single trophozoite (46). 95% of the iron remains heme-associated and eventually biomineralizes to hemozoin in the parasite's digestive vacuole (46). The fate of the remaining 5% iron is less clear. It probably contributes to the labile iron pool, with most of it being chelated in inorganic and/or organic non-heme complexes. Free iron will be mostly in its ferrous state, and very little, if any, will exist as ferric iron at the acidic pH of the digestive vacuole (resting pH ϳ5.2 (51)) due to the formation of water-insoluble ferric hydroxide.
By facilitating ferrous iron transport across the digestive vacuolar membrane, PfCRT might mobilize ferrous iron derived from hemoglobin degradation and make it accessible to the parasite's metabolic activities. This model assumes a direction of iron flux via PfCRT from the acid compartment (lumen of the digestive vacuole in the parasite or external medium in the case of oocytes) to the cytoplasm. Our finding that iron flux via PfCRT is pH-dependent and increases with acid pH is consistent with PfCRT extruding iron from the digestive vacuole in a Table 1 Kinetic

parameters describing the interactions of PfCRT Dd2 and PfCRT HB3 with ferrous and ferric iron
Best-fit values of kinetic parameters were obtained from the kinetics of ferrous and ferric iron uptake using the Michaelis-Menten equation (Fig. 8). The means Ϯ S.E. of N BR (shown in parentheses) independent biological determinations are shown. Statistical significance was determined between corresponding kinetic parameters, using a two-tailed t test.

PfCRT transports iron
manner driven by the proton electrochemical gradient. Although this is a plausible hypothesis, given that chloroquine and other quinolone-like drugs are transported by PfCRT in association with protons out of the digestive vacuole (5, 6, 10, 11, 13), we cannot exclude the possibility of PfCRT extracting iron from the parasite's cytoplasm and depositing it in the digestive vacuole. Maybe PfCRT can perform both functions, depending on electrochemical gradients and whether or not there is a deficiency or a surplus of iron in the parasite's cytoplasm. Thus, PfCRT might contribute to a network of iron regulatory systems, which, in addition to PfCRT, includes a recently discovered iron transporter of the parasite's endoplasmic reticulum (52,53).

Reagents and radiolabeled compounds
Bacto-yeast extract was purchased from BD Biosciences.

Solutions
Solutions used in electrophysiological studies-ND10 (10 mM NaCl, 86 mM NMDG chloride, 2 mM KCl, 1.0 mM CaCl 2 , 1 mM MgCl 2 ) was buffered with 5 mM HEPES/NaOH for pH 7.5 and with 5 mM MES/Tris base for pH 6.0. If necessary, the osmolarity of the solution was adjusted using 1-2 mM mannitol to match the osmolarity of the ND96 buffer (see below). ND10mannitol (10 mM NaCl, the appropriate amount of mannitol (replacing 86 mM NaCl, 2 mM KCl, 1.0 mM CaCl 2 , and 1 mM MgCl 2 ) to maintain osmolarity) was buffered at pH 6.0 with MES/Tris base. Iron-containing solutions were buffered using MES, a buffering agent that, unlike most of Good's buffers, does not form metal chelates (54). In addition, the concentration of CaCl 2 was lowered to 0.6 mM when iron was present in the buffer (22). To maintain iron in the divalent or trivalent form, ascorbic acid or NTA, respectively, was added in quadruple molar ratio relative to the respective iron concentration (22).
Solutions used in transport studies-ND96 (96 mM NaCl, 2 mM KCl, 1.8 mM CaCl 2 , 1 mM MgCl 2 ) was buffered with 5 mM MES/Tris base for pH 6.0 for chloroquine transport. For transport studies using iron, the ND96 buffer was modified as described above, and pH was adjusted to 5.5. Buffers used for studies related to ferric iron transport were supplemented with 200 M ferrocenium hexafluorophosphate to block an endogenous ferrireductase activity present on the oocyte surface (22).

Immunofluorescence
Three days after cRNA injection, oocytes were fixed with 4% paraformaldehyde plus PBS for 4 h at room temperature. Fixed cells were washed with 3% BSA plus PBS and then permeabilized with 0.05% Nonidet P-40 plus PBS for 60 min. The rabbit polyclonal antiserum raised against InsP 3 R-I (H-80) corresponding to amino acids 1894 -1973 within the cytoplasmic domain of InsP 3 R-I (1:200; Santa Cruz Biotechnology) or the guinea pig anti-PfCRT antiserum (dilution 1:500) was added, and oocytes were incubated overnight at 4°C. Oocytes were subsequently washed in 3% BSA plus PBS, and the Alexa Fluor 546 anti-rabbit secondary antibody (Invitrogen) or the Alexa Fluor 488 anti-guinea pig secondary antibody (dilution 1:1000) was added and allowed to bind for 90 min at room temperature. Subsequently, the cells were rinsed four times before they were analyzed by fluorescence microscopy. Images were taken with a Zeiss LSM 510 confocal laser-scanning microscope.

Intracellular ATP determination
The intracellular ATP level of oocytes was determined as described (27). Briefly, whole oocytes were individually mixed with 500 l of boiling water and kept at 100°C for 5 min. Sam-

PfCRT transports iron
ples were then kept on ice until measurement. 2.5 l was subsequently added to 22.5 l of 10 mM HEPES-Na, pH 7.3, containing 2 mM MgCl 2 , and the amount of ATP was determined, using the ATP bioluminescence assay kit CLS II (Roche Applied Science) as instructed by the supplier. Bioluminescence was recorded using the Lumat LB9597 (EG&G Berthold).

PfCRT expression in X. laevis oocytes
Adult female X. laevis frogs (NASCO and Xenopus Express (Vernassal, France)) were anesthetized in a cooled solution of ethyl 3-amino benzoate methanesulfonate (0.1%, w/v) before a surgical partial ovarectomy. Small pieces of ovary were gently shaken at room temperature for 2-3 h in a Ca 2ϩ -free ND96 solution supplemented with collagenase 1A (Sigma-Aldrich). After adding CaCl 2 (1.8 mM) and carefully washing off the enzyme, defolliculated and healthy-looking stage V-VI oocytes were selected and allowed to recover for a few hours (up to 24 h) before being microinjected. The injected RNA was generated as follows. The codon-optimized coding sequences of PfCRT Dd2 and PfCRT HB3 were subcloned in the SP64T expression vector. The resulting SP64-PfCRT plasmid was subsequently linearized using the restriction endonuclease PstI (New England Bio-Labs GmbH, Frankfurt, Germany) and transcribed in vitro using a mMessage mMachine kit (Ambion, Germany) (6). The obtained cRNA was kept up to 8 weeks (at Ϫ80°C), and diluted in RNase-free water to the desired concentration on the day of the oocyte injection. To inject in each oocyte, the precise volume of 50 nl, precision-bore glass capillary tubes (Microcaps Drummond) were pulled on a vertical puller (PE2, Narishige, Tokyo, Japan) and graduated. These micropipettes were placed on a micromanipulator (Narishige) and connected to a microinjector (InjectϩMatic, Geneva, Switzerland). Oocytes were disposed on a mesh tissue in a Petri dish and injected under stereo microscopic control with water alone (control oocytes) or with RNA-containing water. Unless stated otherwise, 30 ng of RNA in 50 nl of RNase-free water or 50 l of water alone were injected per oocyte (InjectϩMatic). Where indicated, oocytes were injected with cRNA coding the divalent metal transporter from rat, rDMT1 (23). The rDMT1 construct was a kind gift from Professor M. Hediger (Bern University, Switzerland). The injected oocytes were incubated at 17°C for 48 -72 h in ND96 for transport assays or in ND10 medium (reduction of NaCl to 10 mM by an equimolar concentration of NMDG or choline chloride) for electrophysiological studies. The incubation media were supplemented with penicillin/streptomycin (100 units/100 g ml Ϫ1 ).

Two-electrode voltage clamp experiments
Oocytes were preincubated in an acidic ND10 (buffered with MES/Tris base at pH 6.0) solution for 30 min before being examined by two-electrode voltage clamp experiments. Oocytes were subsequently placed in a microchamber and superfused with ND10 (pH 6.0) while being punctured with two low-resistance (0.5-2.0-megaohm), 3 M KCl-filled microelectrodes. The microelectrodes were connected to a current-voltage amplifier (Axoclamp 2B or Dagan CA-1B), and the circuit was closed using an Ag-AgCl pellet. Solution change was commanded electronically, using a laboratory-made device. Whole-cell currents were recorded on a multichart recorder (Arc en Ciel, Sefram, Servofram, France) at a holding membrane potential V c ϭ Ϫ50 mV. Current-voltage relationships (I/V) were obtained in the range Ϫ100 to ϩ60 mV (voltage steps of 20-mV increments, 5-s duration from the resting potential), using a Clampex9generated protocol. The membrane conductance G m (slope conductance) was calculated from a linear part of the curves. Results were interfaced with Digidata 1322A to a computer and were analyzed with the P-Clamp9 software program (Axon Instruments). In all cases, oocytes from at least three different frogs were analyzed, with the total number of oocytes investigated indicated in the respective figure.

Intracellular Na ؉ activity
Intracellular Na ϩ activity ␣(Na ϩ ) i was measured in oocytes by double-barreled ion-selective microelectrodes (ISMs), using the sodium ionophore mixture A (Selectophore), as described previously (26,56). Briefly, after gentle beveling of the tip, the ISM slope was S ϭ 50 -59 mV between a pure 100 mM NaCl solution (which also filled the selective barrel) and a mixed (10 mM NaCl plus 90 mM KCl) solution. ISMs were connected to a high-impedance electrometer (FD 223, WPI, UK), and the circuit was closed by a 1% agar, 1 M KCl bridge. The non-selective barrel was filled with 1 M KCl. ␣(Na ϩ ) i was calculated according to (Na ϩ ) i ϭ (Na ϩ ) ref ϫ 10exp(V Na Ϫ V m )/S, where (Na ϩ ) ref is the Na ϩ activity in the superfusate (by taking an activity coefficient of 0.79) and V Na is the measured electrochemical potential difference or Na ϩ .

Transport assays
Chloroquine flux assays were performed as described previously (5,6). Briefly, oocytes were incubated for 60 min at 25°C in ND96 buffer, pH 6.0, containing 42 nM [ 3 H]chloroquine and 50 M of unlabeled chloroquine before the amount of chloroquine taken up by each oocyte was determined. Where indicated, chloroquine flux assays were performed in ND10 buffer, pH 5.5. The ND96 buffer system, pH 6.0, was also used for all transport assays pertaining to the biomolecules shown in supplemental Fig. S3 (A and B). Uptakes of iron were performed using 55 Fe (2 M) in a HEPES-free, calcium-reduced ND96, pH 5.5, to avoid metal chelation by HEPES (22) and to prevent a putative competition between divalent metals (23, 54) (see above). The amount of unlabeled iron added is indicated in the figure legends. To stop uptake, the radioactive medium was removed, and oocytes were washed three times in ice-cold buffer. Each oocyte was transferred into a scintillation vial to be lysed by adding 200 l of a 5% SDS solution. The radioactivity of each sample was measured using a ␤-scintillation counter. The direction of transport in these assays is from the acidic extracellular medium (in most cases pH 5.5) into the oocyte cytosol (pH ϳ7.2), which corresponds to the efflux of drug from the acidic digestive vacuole (pH 5.2) into the parasite cytosol (pH 7.2) (51). Water-injected oocytes were analyzed in parallel, providing information regarding uptake of a particular substance by diffusion and/or endogenous permeation pathways. This value represents the "background" against which the amount of uptake by PfCRT-expressing oocytes was compared. Where

PfCRT transports iron
indicated, the specific PfCRT-mediated uptake was determined by subtracting the corresponding background uptake measured in water-injected oocytes from that of PfCRT-expressing oocytes. Uptake was performed at 25°C for 60 min, if not indicated otherwise. In the case of iron uptake, the raw data were not normally distributed. The raw data did, however, follow a log-normal distribution, and we could therefore compute a mean of the logarithms of the individual measurements. We report the data as the antilogarithms of these means of logarithms. In all cases, at least three separate experiments were performed (on oocytes from different frogs), and in each experiment, measurements were made from 8 -10 oocytes/treatment.

Statistical analysis
Except where stated otherwise, results are expressed as means Ϯ S.E. of n oocytes obtained from N BR independent biological replicates (different frogs). In the case of electrophysiological studies, n is indicated on the graphs, and N BR ϭ 2-7. In the case of uptake assays, except for the kinetics of iron uptake, n ϭ 24 -50 oocytes from 3-5 different frogs. In the case of iron uptake kinetics, n and N BR are indicated in the figure legend. Oocytes obtained from a single frog are considered an independent biological determination. The data were analyzed for statistical significance using the two-tailed t test, the paired t test, the Kruskal-Wallis one-way ANOVA on ranks test, or the Holm-Sidak one-way ANOVA test, where appropriate. p values Ͻ 0.05 were considered significant. Statistical analysis was performed using Sigma Plot version 13.0.