Chloroquine Resistance Modulated in Vitro by Expression Levels of the Plasmodium falciparum Chloroquine Resistance Transporter*

Plasmodium falciparum malaria is increasingly difficult to treat and control due to the emergence of parasite resistance to the major antimalarials, notably chloroquine. Recent work has shown that the chloroquine resistance phenotype can be conferred by multiple amino acid mutations in the parasite digestive vacuole transmembrane protein PfCRT. Here, we have addressed whether chloroquine resistance can also be affected by changes in expression levels of this protein. Transient transfection reporter assays revealed that truncation of the pfcrt 3′-untranslated region just prior to putative polyadenylation sites resulted in a 10-fold decrease in luciferase expression levels. Using allelic exchange on a chloroquine-resistant line (7G8 from Brazil), this truncated 3′-untranslated region was inserted downstream of the pfcrt coding sequence, in the place of the endogenous 3′-untranslated region. The resulting pfcrt-modified “knockdown” clones displayed a marked decrease in pfcrt transcription and an estimated 30–40% decrease in PfCRT protein expression levels. [3H]hypoxanthine incorporation assays demonstrated up to a 40% decrease in chloroquine with or without verapamil IC50 levels of pfcrt knockdown clones, relative to the 7G8 parent. Single-cell photometric analyses were consistent with an altered intracellular pH in the knockdown clones, providing further evidence for a relationship between PfCRT, pH regulation, and chloroquine resistance. Genetic truncation of 3′-untranslated regions provides a useful approach for assessing the impact of candidate genes on drug resistance or other quantifiable phenotypes in P. falciparum.

Plasmodium falciparum malaria is increasingly difficult to treat and control due to the emergence of parasite resistance to the major antimalarials, notably chloroquine. Recent work has shown that the chloroquine resistance phenotype can be conferred by multiple amino acid mutations in the parasite digestive vacuole transmembrane protein PfCRT. Here, we have addressed whether chloroquine resistance can also be affected by changes in expression levels of this protein. Transient transfection reporter assays revealed that truncation of the pfcrt 3-untranslated region just prior to putative polyadenylation sites resulted in a 10-fold decrease in luciferase expression levels. Using allelic exchange on a chloroquine-resistant line (7G8 from Brazil), this truncated 3-untranslated region was inserted downstream of the pfcrt coding sequence, in the place of the endogenous 3-untranslated region. The resulting pfcrt-modified "knockdown" clones displayed a marked decrease in pfcrt transcription and an estimated 30 -40% decrease in PfCRT protein expression levels. [ 3 H]hypoxanthine incorporation assays demonstrated up to a 40% decrease in chloroquine with or without verapamil IC 50 levels of pfcrt knockdown clones, relative to the 7G8 parent. Single-cell photometric analyses were consistent with an altered intracellular pH in the knockdown clones, providing further evidence for a relationship between PfCRT, pH regulation, and chloroquine resistance. Genetic truncation of 3-untranslated regions provides a useful approach for assessing the impact of candidate genes on drug resistance or other quantifiable phenotypes in P. falciparum.
For decades, the treatment of malaria has largely depended on the use of chloroquine (CQ), 1 a 4-aminoquinoline recognized for its rapid efficacy, low toxicity, widespread availability, and affordability. The emergence and spread of CQ-resistant strains of Plasmodium falciparum has been identified as a major factor responsible for the recent increases in malaria mortality and morbidity (1). In Africa alone, malaria-related mortality is estimated at a level of 0.5 to 2.0 million people per year, predominantly children under the age of five (2). An improved understanding of the genetic and pharmacological basis of chloroquine resistance (CQR) and its relationship to the mode of action of CQ can help stimulate and inform new approaches to tackling malaria.
Central to the action of this drug is the process of hemoglobin degradation, thought to occur primarily in the digestive vacuole (DV) of the infected red blood cell (iRBC). This serves to provide amino acids and to deplete the host cell cytosol of hemoglobin, in order to sustain parasite intracellular growth (3,4). This process liberates soluble heme (Fe(II) protoporphyrin IX), whose inherent toxicity is neutralized via its rapid incorporation, as an oxidized (Fe(III)) iron-propionic acid-coordinated hematin dimer, into a crystalline lattice known as hemozoin (4 -6). Chloroquine is thought to bind to hematin dimers, thereby interfering with heme detoxification and hemozoin formation, resulting in parasite death (6 -12). Chloroquine access to heme, the kinetics of heme detoxification and hemozoin formation, and the presence of a pH gradient across the DV membrane have all been proposed to play an important role in determining CQ accumulation and activity inside the DV (reviewed in Refs. 11 and 13). Although there is much debate about the pharmacological basis of CQR, it is clear that resistant parasites accumulate less CQ than their sensitive counterparts and that this reduced accumulation can be partially reversed by verapamil (VP) (14 -16).
Point mutations in pfcrt (P. falciparum chloroquine resistance transporter) have recently been implicated as the key determinant of CQR. Evidence in favor of this includes the following: (i) these mutations segregate with CQR in a P. falciparum genetic cross (17)(18)(19); (ii) mutant pfcrt haplotypes show an excellent association with CQR in laboratory-adapted field isolates from multiple geographically distinct regions (19 -21); (iii) pfcrt point mutations are often associated with an increased risk of CQ treatment failure (21)(22)(23)(24); (iv) microsatellite markers support a recent, worldwide sweep of mutant pfcrt haplotypes under CQ pressure (25); and (v) allelic exchange data have recently provided conclusive evidence that mutant pfcrt alleles prevalent in Asia, Africa, South America, and the Oceanic region can confer CQR to a CQ-sensitive (CQS) clone (26).
In a recent pfcrt allelic exchange study (26), one notable finding was that expression of the mutant haplotypes in the recombinant lines resulted in acquisition of CQ IC 50 values that were 70 -90% of those observed with non-transformed reference lines expressing the same pfcrt haplotypes. One possible explanation is that mutant pfcrt can account for the bulk but not all of the CQR phenotype and that other genes are required to elevate the CQ IC 50 values in the CQ-resistant lines. Another explanation is that the pfcrt-modified lines had lower CQ IC 50 values as a result of their reduced levels of PfCRT protein expression (as observed by RNA and protein analysis (26)). This presumably stemmed from the genetic modification of the functional locus, which involved removal of the introns and manipulation of the 3Ј-untranslated region (UTR).
To address the question of whether pfcrt expression levels can influence CQR independent of point mutations, we have implemented an allelic exchange strategy designed to introduce a truncated 3Ј-UTR into the functional pfcrt locus. This "knockdown" approach has been shown in higher eukaryotic systems to reduce levels of transcription and translation by decreasing mRNA stability and was used in Plasmodium berghei to assess the contribution of CS expression levels on sporozoite morphology (27)(28)(29). The impact of this change on pfcrt expression, acridine orange (AO) accumulation (a putative marker of vacuolar pH), and response to CQ and related hemebinding antimalarials is presented.

EXPERIMENTAL PROCEDURES
Luciferase Assays-The 1033-, 625-, and 148-bp pfcrt 3Ј-UTR sequences were PCR-amplified from Dd2 genomic DNA using the primer combinations p1/p2, p1/p3, and p1/p4, respectively (Table I), and cloned into pHLH-1 (30) in the place of the hrp2 3Ј-UTR sequence. The resulting plasmids (named pHLcrt/1033, pHLcrt/625, and pHLcrt/148, respectively, Fig. 1A) were tested in luciferase reporter assays, using the technique of RBC pre-loading with plasmid DNA developed by Deitsch et al. (31). Briefly, cultures rich in synchronized Dd2 trophozoites were obtained by successive rounds of sorbitol lysis of ring stage cultures (32). 24 h later, trophozoites were purified by Percoll-Sorbitol density gradient centrifugation (33), and aliquots were evenly distributed to preparations of uninfected RBC that had been electroporated with the purified luciferase reporter plasmids (100 g of DNA/10 9 RBC/transfection (34)). Cultures were expanded on day 2 post-electroporation from 5 to 10 ml with the addition of another 10 9 RBC pre-loaded with 100 g of plasmid DNA. The following day, parasites were harvested following RBC lysis with 1ϫ PBS containing 0.15% saponin. Luciferase expression was assayed using the Dual-Luciferase® reporter system (Promega), and activity was measured in luminescence units using an AutoLumat LB953 luminometer (EG&G Berthold).
Vector Construction and Parasite Transfection-The pfcrt allelic exchange fragment was PCR-amplified from 7G8 genomic DNA using the primer combination p4/p5. This yielded a 1.28-kb fragment comprising exons 8 -13 as well as the proximal 148 bp of pfcrt 3Ј-UTR. This fragment was cloned into the plasmid pHDWT, which uses as a selectable marker the human dihydrofolate reductase (hdhfr) gene flanked by promoter and terminator elements from the hrp3 and hrp2 genes, respectively (34). The resulting 7.1-kb transfection plasmid, pHD/pfcrt⌬ 3Ј-UTR, was electroporated into 7G8 P. falciparum ring stage parasites. Plasmid-transfected parasites were selected by addition of 5 nM WR99210 to the culture medium starting 48 h post-transfection (34). Parasite clones were obtained by limiting dilution and identified using the MALSTAT (Flow Inc.) assay reagent specific for P. falciparum lactate dehydrogenase (35,36).
Nucleic Acid Analysis-P. falciparum genomic DNA was purified as described previously (37). Plasmid integration into the 3Ј-UTR of pfcrt was detected by PCR using the primer combinations p6/p7 and p8/p9 (Table I and Fig. 1B). Integration into the hrp3 and hrp2 loci was assessed using the primer combinations p10/p11 and p12/p13, respectively. Amplification of non-recombinant, endogenous pfcrt sequence was assayed using the primers p6/p9. PCR conditions included a primer extension temperature of 62°C to account for the high A-T content in P. falciparum genomic DNA (38).
For stage-specific Northern blots, cultures were doubly synchronized by sorbitol treatment. Total RNA was prepared using TRIzol® (Invitrogen) and solubilized in deionized formamide (39). RNA samples (10 g per lane) were size-fractionated on a 1% agarose gel supplemented with 5 mM guanidine thiocyanate and transferred to Hybond N ϩ nylon membranes (Amersham Biosciences). Membranes were hybridized with 32 Plabeled 1.3-kb pfcrt or 1.0-kb P. falciparum EF-1␣ cDNA probes, generated using the primers p14/p15 and p16/p17, respectively. Autoradiography images were captured using a Fuji film FLA2000 phosphorimaging device and quantified for data analysis.
For reverse transcriptase-PCR assays, parasite lines were synchronized using successive rounds of gelatin flotation and sorbitol lysis. Cultures with Ͼ85% ring stage parasites were harvested by lysis with 0.15% saponin, and total RNA was extracted using TRIzol®. First strand cDNA was synthesized from 1 g of total RNA, using the Superscript TM pre-amplification system (Invitrogen) with oligo(dT) priming. For Southern blot analysis, genomic DNA was digested with EcoRI and HaeIII, electrophoresed overnight at 2.5 V/cm in 0.8% (w/v) agarose gels (ϳ1.5 g of DNA/lane) alongside a dideoxygenin (DIG)-labeled DNA molecular mass ladder (Roche Applied Science), transferred to Nylon membranes, and UV cross-linked. A 0.9-kb region of pfcrt (exons 12 and 13, plus 0.6 kb of 3Ј-UTR sequence) and a 0.6-kb region of hdhfr were PCR-amplified using the primers p18/p19 and p20/p21, respectively, and labeled with DIG (Roche Applied Science). Hybridizations were performed at 60°C, and membranes were washed at a maximum stringency of 0.3ϫ SSC/0.1% SDS at 60°C, prior to CDP-Star detection, and autoradiography.
Western Blotting and Immunoelectron Microscopy-For whole parasite extracts, synchronized parasite cultures rich in late rings and early trophozoites were lysed using 0.15% saponin. Parasite pellets were washed in 1ϫ PBS, resuspended in SDS-PAGE loading buffer and the proteins resolved on 10% (w/v) polyacrylamide gels prior to being transferred to polyvinylidene difluoride membranes. Western blots were performed using affinity-purified anti-PfCRT antibodies (19) and detected using the Renaissance Western blot Chemiluminescence Reagents (PerkinElmer Life Sciences). Autoradiographic data were analyzed by densitometry using Image version 1.6.2 (National Institutes of Health, available at rsb.info.nih.gov/nih-image/). Percent reductions, presented as mean Ϯ S.E., were determined by normalization of the data from each parasite line at the different loading amounts. DV extracts were prepared and analyzed by Western blotting as described previously (40,41).
Immunoelectron microscopy was performed as previously described (42). Anti-PfCRT antibody-labeled samples were stained with uranyl acetate and lead citrate and examined in a Zeiss CEM 902 electron microscope.
Drug Assays-In vitro drug responses were calculated from 72-h [ 3 H]hypoxanthine incorporation assays (34). Levels of incorporation were measured using a 1450 Microbeta liquid scintillation and luminescence counter (Wallac). Chloroquine (dihydrochloride "Aralen" formulation) was from Winthrop-Breon, whereas quinine (sulfate formulation) and mefloquine (hydrochloride formulation) were from Sigma. Drug inhibition concentrations were calculated by regression analysis of dose-response curves. Mann-Whitney t-tests were used to assess for statistical significance.
Intracellular pH Measurements-Internal/external AO ratios were measured for individual intraerythrocytic parasites under constant perfusion, using single-cell photometric (SCP) analysis of synchronized trophozoites following previously described methodologies and using custom equipment (43). We stress these SCP experiments are not analogous to other published experiments that follow the response of trapped fluorophores via time-dependent laser confocal microscopy (44,45). As described in detail elsewhere (43), 2 our SCP experiments use illumination power that is two to three orders of magnitude lower than laser confocal microscopy, dynamic thresholding computational methods that are unavailable for commercial laser confocal systems, and substantial recovery from illumination after each data point acquisition (3-s recovery in total darkness per 0.133-s data acquisition). These data have been interpreted in previous reports (19,20,42,43) as a valid measure of DV pH, because of the following: 1) the observed semilogarithmic Henderson Hasselbach behavior of AO diffusion under the experimental conditions; 2) the K ϩ -specific abrogation of this behavior upon addition of nigericin; 3) the reversible change in AO ratio upon subtle changes in external [HCO 3 ϩ ] that do not change internal pH, again in accordance with Henderson Hasselbach predictions; and 4) the use of illumination powers that are ϳtwo to three orders of magnitude below those shown to affect the integrity of DV membranes in a timedependent fashion (45). 2 Nonetheless, AO ratios have been suggested by others to be an unreliable marker for DV pH (44,45). Thus, so long as this controversy remains unsettled, in the present work we avoid explicit calculation of DV pH using these ratio data.
To directly test whether the present SCP experiments monitored AO localizing within the DV, co-localization studies of AO and hemozoin were performed using laser scanning confocal microscopy (cf. Fig. 5C, under "Results"). Laser confocal images were acquired in "z-axis series" mode using an Olympus Fluoview confocal microscope and an Olympus UPlanApo 100ϫ oil immersion objective (numerical aperture 1.35) as detailed elsewhere. 2 In brief, the 488-nm line of an argon laser (20% output) was used for excitation, and we acquired a set of 15 conjugate transmittance and fluorescence images upon successive 0.2-m dis-placements along the z-axis. We first focused transmittance in the absence of laser illumination until hemozoin was clearly visible within the center of the synchronized late stage trophozoite parasite DV, thus putting the focal plane within the DV. We then defined this z-axis position as 0. A series of conjugate fluorescence and Nomarski transmittance images were then rapidly acquired in successive 0.2-m z-axis displacements relative to 0, such that we were sure to scan through the parasite. After archiving the image pairs, we determined the largest hemozoin diameter in the Nomarski transmittance set via histogram analysis. The conjugate fluorescence image partner to this transmittance image was then subjected to histogram analysis using the same parameters. Each image (transmittance and fluorescence versions) defined the same ϳ0.25-m-thick optically defined slice (slightly larger than 0.2 m due to the pinhole setting) corresponding to the DV of a late stage trophozoite. All fluorescent light recorded from this slice thus resided within 0.12 m in the ϩ or Ϫ direction relative to the visualized hemozoin and within the x and y dimensions shown in Fig. 5 under "Results."

RESULTS
Effect of Truncation of pfcrt 3Ј-UTR Determined using Luciferase Assays-Initial 3Ј rapid amplification of cDNA ends (RACE) assays on the pfcrt 3Ј-UTR identified two candidate polyadenylation sites, located ϳ163 and 513 bp downstream of the pfcrt stop codon. 3 Using a P. falciparum transient transfection reporter assay (31), we tested whether 3Ј-UTR truncation prior to one or both of these sites would lead to reduced gene expression. pfcrt 3Ј-UTR sequences either 1033, 625, or 148 bp in length were cloned into pHLH-1 (30), downstream of the firefly luciferase gene and in the place of the hrp2 3Ј-UTR sequence, to form the plasmids pHLcrt/1033, pHLcrt/625, and pHLcrt/148, respectively. In these constructs, luciferase transcription initiates from an hrp3 promoter sequence. Luciferase values, assessed 72 h post-electroporation, showed that truncation of the pfcrt 3Ј-UTR from 1033 to 625 bp or to 148 bp yielded 64 or 92% reduction of luminescence units, respectively, relative to the full-length 1033-bp fragment (Fig. 1A).
Production of pfcrt "Knockdown" Clones by Allelic Exchange-Based on these reporter assays, we designed a transfection plasmid, called pHD/pfcrt⌬ 3Ј-UTR, to replace the fulllength pfcrt 3Ј-UTR in the chromosomal locus with the truncated 148-bp sequence. This plasmid contained the truncated 3Ј-UTR preceded by the last six exons of pfcrt. Allelic exchange of pHD/pfcrt⌬ 3Ј-UTR into the parasite genome via homologous recombination and single site-crossover was predicted to generate a functional pfcrt locus that retained the original pfcrt promoter and coding sequence with 3Ј-UTR regulation provided by the 148-bp truncated sequence (Fig. 1B). This approach was also predicted to generate a downstream, disrupted remnant pfcrt sequence that lacked a promoter and the first seven exons of the endogenous gene. For this experiment, we transfected the 7G8 line to test whether expression levels of the 7G8-type pfcrt allele (which appears to have originated independently in South America and Oceania (19,20,25,46)) had a direct impact on the P. falciparum CQR phenotype.
Electroporation of pHD/pfcrt⌬ 3Ј-UTR into 7G8 parasites generated episomally transformed lines that were maintained by selection with WR99210. Passaged lines were periodically screened by PCR for homologous recombination and plasmid integration into the pfcrt (Fig. 1B) or hrp2 or hrp3 loci (not shown). One line was found to contain a mixed population of parasites that had integrated into either the pfcrt or the hrp2 loci (data not shown). Clones were generated from this line (at day 90 post-electroporation) by limiting dilution (35), leading to the identification of K2C3 and K2F10. These "transcription knockdown" clones were confirmed by PCR analysis to be re-combinant solely for pfcrt integration and were PCR-negative for wild type pfcrt (Fig. 1C).
Southern blot hybridization confirmed that the endogenous pfcrt locus had been replaced by the 3Ј-UTR-truncated locus in the K2C3 and K2F10 clones (Fig. 1D). Using a pfcrt probe, hybridizing bands of 10.2/6.8 kb and 6.5/2.7 kb, respectively, were detected upon EcoRI and HaeIII digestion of genomic DNA from the knockdown clones. This contrasted with the 10.0-and 7.6-kb bands generated by these two enzymes in non-transfected 7G8 genomic DNA. The visible lack of the wild type 7.6-kb HaeIII band in the K2C3 and K2F10 clones provided further evidence for complete pfcrt allelic replacement. These restriction patterns were consistent with integration of a single plasmid copy into the pfcrt locus, because integration of two or more tandem copies would have generated an additional 6.9-kb band and a 1.5-kb band upon EcoRI and HaeIII digestion, respectively. Results obtained using an hdhfr probe also showed hybridizing bands of the expected size in K2C3 and K2F10 (data not shown).
Expression Levels in Knockdown Mutants-To assess the impact of 3Ј-UTR truncation on pfcrt expression from the chromosomal locus, we performed Northern blots on tightly syn-chronized RNA preparations from the knockdown and 7G8 parental clones. For this, we collected RNA samples from early rings, late rings, trophozoites, and schizonts. Results showed significantly reduced pfcrt transcript levels in the knockdown clones K2C3 and K2F10, relative to 7G8 (Fig. 2A). Timing of maximal transcription remained equivalent between the lines, peaking in the early ring stages and rapidly decreasing to undetectable levels by the trophozoite stage. These results suggest that the timing of transcription was largely determined by the pfcrt promoter region, which remained unchanged in these different clones. Densitometric analysis of the early ring stage results, comparing ratios of pfcrt to EF-1␣ signal intensity between the different clones, predicted a 52-55% decrease in pfcrt transcription in the knockdown clones. This assumes that EF-1␣ levels of transcription remained unchanged between these clones. The reduction in pfcrt transcript size (from 4.2 kb in the wild type locus to 3.7 kb in the knockdown clones) suggests that the wild type transcript ends about 0.65 kb downstream of the pfcrt stop codon (as compared with 0.15 kb in the knockdown clones). Given that the coding sequence is 1.25 kb, this would predict that the endogenous pfcrt transcript contains about 2.3 kb of 5Ј-UTR sequence and thus starts 0.65

FIG. 1. pfcrt 3-UTR truncation: transfection approach and generation of recombinant clones.
A, effect of pfcrt 3Ј-UTR truncation on luciferase activity. Results show the mean Ϯ S.D. of the luciferase signals obtained from synchronized P. falciparum (Dd2 strain) ring-stage parasite cultures transfected with luciferase reporter plasmids. These constructs shared a common hrp3 5Ј-UTR promoter sequence to drive luciferase expression and differed in their 3Ј-UTR terminator sequence. Luminescence units were normalized relative to the values obtained with pHLcrt/1033, which contained 1 kb of pfcrt 3Ј-UTR. B, schematic representation of pHD/pfcrt⌬ 3Ј-UTR integration into the parasite genome via homologous recombination and single-site crossover. This results in the generation of an upstream functional pfcrt gene possessing the endogenous pfcrt promoter and coding sequence plus the 148-bp truncated 3Ј-UTR. A downstream, disrupted pfcrt sequence lacks the promoter and the first seven exons. The integrated human dihydrofolate reductase (hDHFR) expression cassette confers resistance to WR99210, which is used as the selection agent. PCR primers are specified in Table  I. C, PCR analysis showing truncation of the functional pfcrt 3Ј-UTR in the knockdown clones K2C3 and K2F10. Primers p6/p7 (lanes 1, 6, and 11) and p8/p9 (lanes 2, 7, and 12), respectively, amplified the functional 3Ј-UTR-truncated and disrupted pfcrt loci in both knockdown clones but not in non-transfected 7G8 parasites. Primers p6/p9 (lanes 3, 8, and 13), specific for the wild type pfcrt locus, gave a product only with 7G8 parasites. No clones showed evidence of integration into hrp3 or hrp2, assayed using primers p10/p11 (lanes 4, 9, and 14) or p12/p13 (lanes 5, 10, and 15), respectively. M, 1-kb DNA ladder. D, Southern blot hybridization with a DIG-labeled pfcrt probe confirming plasmid integration into the pfcrt locus in K2C3 and K2F10. E, EcoRI; H, HaeIII.
kb downstream of the neighboring cg3 gene (18). Reverse transcriptase-PCR analyses of RNA prepared from synchronized late ring and early trophozoite stages gave results consistent with the Northern data (not shown).
To compare PfCRT protein expression levels between parental and recombinant lines, Western blots were performed with lysates of synchronized parasite cultures and affinity-purified anti-PfCRT antibodies (19). Densitometric analysis (averaged between the different sets of loadings, Fig. 2B) predicted a reduction in PfCRT expression levels of 32 Ϯ 10% for K2C3 and 34 Ϯ 12% for K2F10 (mean Ϯ S.E.), as compared with nontransfected 7G8 PfCRT. These data were confirmed with Western blot analysis of purified DV, which showed an estimated 70 -80% reduced expression of PfCRT in the knockdown clones (data not shown). These data suggest a possibly even greater impact of the truncation on decreasing the amount of PfCRT being inserted into the DV membrane.

Localization of PfCRT to the DV of pfcrt Knockdown Para-
sites-Related studies in other organisms have found that truncation of 3Ј-UTR sequences can occasionally result in mislocalization of the gene product (27). To address this possibility, we performed immunogold electron microscopy using affinitypurified anti-PfCRT antibodies. This localized PfCRT solely to the membrane region surrounding the DV in the knockdown and 7G8 lines (Fig. 3). Quantitation of PfCRT antibody-conjugated gold particles in 10 cryosectioned parasite DVs gave an estimated 40 -80% decrease in antibody labeling and provided additional complimentary evidence for decreased PfCRT expression in the DV of knockdown clones.
Response of pfcrt Knockdown Parasites to Chloroquine, Quinine, and Mefloquine-72-h [ 3 H]hypoxanthine incorporation assays revealed that the reduction in PfCRT expression levels was accompanied by a significant attenuation of the CQR phenotype, with the knockdown clones showing up to a 38% reduction in CQ IC 50 values when compared with 7G8 (Fig. 4A). Essentially the same percentage decrease in IC 50 values was Parallel electrophoresis of duplicate samples on a separate SDS-PAGE gel, followed by Coomassie Blue staining and densitometric analysis, was used to equalize protein loadings for the different lines. The smaller apparent PfCRT size observed in these studies, compared with the observed range of 42-45 kDa previously reported (19), may relate to differences in protein molecular weight standards and/or sample processing issues. observed when CQ was tested in the presence of the CQR reversal agent VP (tested at 0.8 M, Fig. 4A). This indicated that the degree of VP reversibility was essentially unchanged in the knockdown clones, even though the level of CQR was reduced. This degree of VP reversibility was noticeably lower than in the CQ-resistant FCB line (Fig. 4A), which carries a pfcrt haplotype frequently found in Asian and African CQresistant lines (19,20). Conclusive evidence that the extent of VP reversibility is largely determined by the pfcrt haplotype was recently obtained using allelic exchange (26). The decreased level of CQR in the knockdown clones was apparent across multiple, partially inhibitory CQ drug dilutions (Fig. 4,  D and E). These clones nonetheless had IC 50 values for CQ with or without VP that were closer to 7G8 than to the VP-insensitive CQS parasite line 106/1 (Fig. 4A). These drug assay data were consistent with the observation that only 7G8 and FCB, but not the pfcrt knockdown clones or 106/1, could be propagated long term in culture medium supplemented with 80 nM CQ (data not shown).
Quantitation of the response of K2C3 and K2F10 to the other putative heme-binding antimalarials quinine and mefloquine indicated no statistically significant change in the respective IC 50 values compared with 7G8 (Fig. 4, B and C). Verapamil was found to have a partial reversal effect on the quinine response but no effect on mefloquine.

AO Perfusion Experiments with pfcrt Knockdown Parasites-
Single-cell photometry studies of P. falciparum-infected RBC with AO have consistently shown that CQ-resistant parasite lines expressing mutant PfCRT have a steeper intracellular [AO]/extracellular [AO] slope, relative to CQS lines (19,20,42,43). These results suggest but do not prove that PfCRT mutations alter DV ion transport physiology. Here, measurements of the change in AO fluorescence in single cells, as a function of time and of [AO] in the perfusate, showed a dramatic change in the knockdown clones, which now had profiles similar to the CQS line 106/1 (Fig. 5A). This contrasted with the relationship observed for 7G8 and the other CQ-resistant line FCB. Calculation of the mean fluorescence intensity data for each line, versus external [AO], confirmed that the knockdown clones grouped with 106/1, and not with 7G8 or FCB (Fig. 5B). These data implied a direct impact of PfCRT expression levels on the internal versus external AO relationship. One interpretation of these data is that lower compartmental pH accompanies reduction in the level of mutant PfCRT expression. Using different equipment and methodologies, other researchers have questioned whether AO localizes to the DV and/or whether illumination affects its subcellular disposition (Refs. 44 and 45, respectively). To address this question, we used laser scanning confocal microscopy to compare AO and hemozoin localization. Results (Fig. 5, C and D) show that the AO fluorescence signals, FIG. 4. PfCRT expression levels specifically reduce CQR, without dramatically impacting parasite response to quinine and mefloquine. A-C, IC 50 values for chloroquine (CQ), quinine (QN), and mefloquine (MFQ) with or without verapamil (VP) in 7G8 and the PfCRT knockdown clones. Mean IC 50 values for FCB, 7G8, K2C3, K2F10, and 106/1 were 221 Ϯ 36, 136 Ϯ 10, 84 Ϯ 6, 89 Ϯ 5, and 32 Ϯ 4 nM, respectively, for CQ, and 61 Ϯ 12, 98 Ϯ 7, 59 Ϯ 5, 62 Ϯ 7, and 32 Ϯ 6 nM for CQ plus VP. The differences in CQ with or without VP IC 50 values in the knockdown clones compared with 7G8 were statistically significant (*, p Ͻ 0.05, Mann-Whitney t test). Values were calculated from four independent assays performed in duplicate. Control experiments performed on 7G8 lines episomally transformed with the pHDcrt/⌬ 3Ј-UTR construct, prior to PCR-detectable plasmid integration and allelic exchange, showed no shift in IC 50 values for CQ with or without VP (data not shown). D and E, representative dose-response assay for CQ with or without VP, measured as means Ϯ S.E. for each dilution point measured in triplicate. For clarity, only the knockdown lines and 7G8 are shown. as measured in successive 0.2-m optical sections along the z-axis, coincide with the hemozoin localization to within 0.5 m. Thus, hemozoin is contained within an irregular sphere defined by AO fluorescence that extends no more than 0.5 m in any direction from hemozoin. This is well within the confines of the DV that can attain a diameter of up to 2.5-3.0 m in the mature, trophozoite-stage intra-erythrocytic parasite (4). DISCUSSION The recent finding that the inheritance of CQR in the progeny of a P. falciparum genetic cross is closely associated with point mutations in pfcrt (19) has stimulated a wave of molecular and epidemiological investigations. These have typically revealed a very high correlation between CQR and the PfCRT K76T point mutation in laboratory-adapted lines and clinical samples from around the world (reviewed in Ref. 47). This degree of association is higher in vitro than in vivo, presumably because the latter is significantly influenced by additional factors, including partial immunity (21). In the present study, we now show that the degree of CQR can also be affected by changes in the level of PfCRT expression. These results could explain why, in a recent study, parasites engineered to express mutant pfcrt alleles were only 70 -90% as CQ-resistant as the haplotype-matched, non-transformed lines (26). In that study, PfCRT expression levels were reduced ϳ30 -70% relative to the non-transformed parental line.
Our data suggest that differences in pfcrt expression levels may account in part for the two to 4-fold differences in CQ IC 50 values that have been reported between CQ-resistant lines carrying identical pfcrt haplotypes (18 -20, 46, 48 -50). Undoubtedly, other genes are also involved, including pfmdr1, which was recently shown by allelic exchange to be able to modulate the CQR phenotype in a parasite line expressing mutant pfcrt (51). Interestingly, several studies have found altered levels of pfmdr1 expression in mutant lines selected under continued drug pressure (52)(53)(54). pfmdr1 also contains several point mutations that in some, though not all studies, show a correlation with CQR (reviewed in Ref. 55). Thus, point mutations and expression levels of both pfcrt and pfmdr1 have now been found to have an impact on the degree of CQ susceptibility. Both genes encode proteins localized on the membrane of the DV (19,42,56), the organelle wherein CQ exerts its activity, implying that altered functional properties of the DV membrane are key to the CQR mechanism.
How does mutant PfCRT contribute to CQR? One possibility could involve an effect on pH values and pH gradients within the intact intra-erythrocytic parasite that might critically alter CQ accumulation or pH-dependent binding of CQ to relevant targets. In the present study, we found a significant decrease in accumulation of the pH sensor AO in the pfcrt knockdown clones. These clones, designed to differ from the CQ-resistant  42). Extrapolation of these curves suggests a reduced DV pH in the knockdown clones, aligning them more closely with the CQS reference line 106/1 than with the CQR reference lines FCB and 7G8. Data were collected from 26 -53 individual parasites for each line. In C and D we show colocalization of AO-and DV-entrapped hemozoin (as described under "Experimental Procedures"). The x axis denotes the distance along the x,y plane, the left y axis denotes AO fluorescence intensity, and the right y axis denotes relative transmittance. The top line in each panel thus shows a distinct negative transmittance peak corresponding to optically dense hemozoin, whereas the bottom line shows a positive peak corresponding to AO fluorescence emission. As described in the text, this emission must originate within 0.12 m of hemozoin (in the z axis sense) and within 0.5 m of hemozoin (in the x,y sense; see x axis each panel), thus putting the majority of this emission within the DV. C is for parasites exposed to perfusate harboring 1 M AO, and D is for parasites exposed to 0.5 M AO. In both cases the histograms are for a single parasite and show that AO-and DV-localized hemozoin are virtually superimposable under the conditions used in our SCP experiments (43). These data are representative of histogram analysis performed from z slice series of at least 20 individual parasites. A more extensive analysis of colocalization and other AO-staining characteristics will be published elsewhere. 2 7G8 line only at the pfcrt locus, now closely aligned with CQS parasites (Fig. 5). These results support the general idea that PfCRT might be involved in ion-dependent DV transport processes that affect intracellular pH (19,41). They are also consistent with earlier reports predicting that the DV of CQresistant parasites was more acidic (19,20,42,43), however, we stress this suggestion awaits confirmation by a non-AO-dependent method for quantifying DV pH. Heme aggregation and crystallization assays have provided evidence that a more acidic DV pH would significantly reduce CQ accumulation by (i) increasing the rate of conversion of heme to hemozoin (10) and (ii) promoting acid aggregation of free heme (43). These two acid-activated processes could effectively titrate out the soluble hematin receptor for CQ and thus confer resistance. Independent assays are required to unequivocally prove this interpretation, which has been challenged by others based on AO measurements made with different methods (44). Along these lines, recent ratiometric DV pH measurements with the fluorophore NERF, which do not suffer from the same limitations that accompany AO-based analyses, provide additional evidence that CQS parasites have a higher DV pH than do CQ-resistant parasites. 2 Regardless, importantly, our data also reveal that changes in AO partitioning do not strictly correlate with changes in the degree of susceptibility, because the knockdown parasites remained CQ-resistant, albeit at borderline levels, with CQ IC 50 values of 75-80 nM. These values are slightly below the estimated CQR lower limit of 80 -100 nM (58,59). An alternative interpretation of our photometric data is that mutant PfCRT might transport AO (60,61) and that the rate of AO transport can be affected by point mutations and expression level changes in this protein.
Another possible explanation for the contribution of pfcrt to CQR is that the point mutations might enable PfCRT to directly interact with the drug, thereby perhaps competing with hematin for CQ binding and reducing intracellular CQ accumulation. This would be expected to result in reduced CQhematin interaction and enable the CQ-resistant parasite to continue its normal process of heme detoxification. The ability of VP to reverse CQR could reflect VP binding to mutant PfCRT and interfering with PfCRT-CQ interactions. Efforts are underway to discriminate between the hypothesis of an indirect role for PfCRT based on its effect on intracellular physiology (including pH changes), versus the acquisition of a drug transport function. These efforts include expressing PfCRT in yeast vesicles (41) and Xenopus laevis oocytes 4 to assay for endogenous PfCRT function and possible interactions with drug.
We note that, in this present study, reduction of mutant PfCRT expression rendered the parasites more susceptible to CQ yet had no major impact on their degree of susceptibility to either mefloquine or quinine (Fig. 4). These results contrast with previous reports that PfCRT point mutations can significantly alter parasite susceptibility to all three compounds (26,42) and suggest that expression differences and sequence changes affect parasite response to heme-binding antimalarials in separate ways. These data also indicate that the means by which PfCRT affects response to mefloquine and quinine is distinct from its capacity to confer CQR.
In parallel with this experiment, we attempted to disrupt pfcrt function using "knockout" transfection constructs. Although episomally transformed parasites were obtained, pfcrt disruption was never detected above trace levels using sensitive PCR screens (K. Waller and D. Fidock, data not shown). Our findings suggested that disruption of pfcrt is deleterious to normal parasite viability and may indicate that this gene is critical for parasite propagation. We also note that, despite multiple attempts, we were unable to isolate and clone knockdown parasites in the CQ-resistant Dd2 line, even though these could be detected in the bulk cultures, probably because of a slow growth phenotype. Slower growth was also apparent in the 7G8 knockdown lines described herein (data not shown). These combined observations suggest an important role for PfCRT in helping maintain normal parasite growth inside the iRBC. Microarray and proteomic studies are underway to test whether reduced PfCRT expression has led to compensatory adjustments in the expression levels of functionally related genes, as an approach to identifying genes that work with PfCRT in helping to regulate parasite intracellular physiology.
To our knowledge, this is the first report in which genetic modification of a Plasmodium gene has been employed to specifically study the effects of gene expression levels on drugresistant phenotypes. Application of this technology offers the attractive possibility of probing gene function in the absence of generating gene knockouts, which may significantly reduce parasite viability or in fact be lethal to the parasite, and obviates the prima facie need to identify and target individual mutations or domains. RNA interference represents a less labor-intensive alternative, and two reports provide evidence that this may work in P. falciparum (62,63). However, concerns remain because of the lack of critical RNA interference proteins in the P. falciparum genome (64), including the RNAdependent RNA polymerase Dicer. Alternative forms of RNAdependent gene regulation may nonetheless exist in P. falciparum, as evidenced by the discovery of unique RNA-protein interactions that were found to affect antimalarial antifolate susceptibility (65) as well as the finding of abundant antisense transcripts in this organism (66). An improved understanding of P. falciparum gene regulation and its application to the study of drug resistance and pathogenesis can facilitate efforts to translate post-genomic research into practical solutions to combat malaria.