Piperaquine Resistance Is Associated with a Copy Number Variation on Chromosome 5 in Drug-Pressured Plasmodium falciparum Parasites^,

Abstract

The combination of piperaquine and dihydroartemisinin has recently become the official first-line therapy in several Southeast Asian countries. The pharmacokinetic mismatching of these drugs, whose plasma half-lives are ∼20 days and ∼1 h, respectively, implies that recrudescent or new infections emerging shortly after treatment cessation will encounter piperaquine as a monotherapy agent. This creates substantial selection pressure for the emergence of resistance. To elucidate potential resistance determinants, we subjected cloned Plasmodium falciparum Dd2 parasites to continuous piperaquine pressure in vitro (47 nM; ∼2-fold higher than the Dd2 50% inhibitory concentration [IC₅₀]). The phenotype of outgrowth parasites was assayed in two clones, revealing an IC₅₀ against piperaquine of 2.1 μM and 1.7 μM, over 100-fold greater than that of the parent. To identify the genetic determinant of resistance, we employed comparative whole-genome hybridization analysis. Compared to the Dd2 parent, this analysis found (in both resistant clones) a novel single-nucleotide polymorphism in P. falciparum crt (pfcrt), deamplification of an 82-kb region of chromosome 5 (that includes pfmdr1), and amplification of an adjacent 63-kb region of chromosome 5. Continued propagation without piperaquine selection pressure resulted in “revertant” piperaquine-sensitive parasites. These retained the pfcrt polymorphism and further deamplified the chromosome 5 segment that encompasses pfmdr1; however, these two independently generated revertants both lost the neighboring 63-kb amplification. These results suggest that a copy number variation event on chromosome 5 (825600 to 888300) is associated with piperaquine resistance. Transgene expression studies are underway with individual genes in this segment to evaluate their contribution to piperaquine resistance.

INTRODUCTION

Efforts to reduce the substantial mortality and morbidity of malaria rely on effective patient management as well as interventions directed at reducing or blocking Plasmodium parasite transmission. Adequate antimalarial treatment of infected individuals not only limits the duration and severity of disease, but also has the potential of reducing transmissibility (34). In response to the spread of multidrug-resistant Plasmodium falciparum malaria, there has been a nearly worldwide adoption of clinically highly efficacious artemisinin-based combination therapies (ACTs). These ACTs combine an extremely potent, short-lived artemisinin derivative with a partner drug that possesses a longer half-life. These combinations not only increase the therapeutic efficacy and patient adherence, but also reduce the risk of selecting for drug-resistant parasites. One ACT that is now an official first-line policy for the treatment of P. falciparum malaria in several Southeast Asian countries, and under evaluation in South America and Africa, is dihydroartemisinin-piperaquine.

Piperaquine tetraphosphate (PQP) is a bisquinoline antimalarial drug that was synthesized in the 1960s at Rhone-Poulenc and independently at the Shanghai Research Institute of Pharmaceutical Industry (8). Due to the increasing prevalence of chloroquine (CQ)-resistant parasites in southern China, PQP was adopted as the first-line treatment in 1978 (8). This drug was used widely for both prophylaxis and treatment, with the equivalent of 140 million adult doses manufactured and distributed (57). Its application as monotherapy, however, resulted in the eventual emergence of PQP-resistant parasites, which diminished its use by the late 1980s. PQP was subsequently combined as part of China-Vietnam 4 (known as CV4), an ACT that achieved high cure rates and that consisted of dihydroartemisinin (DHA), trimethoprim, PQP, and primaquine (8). This combination has been revised, and PQP is presently coformulated solely with DHA. This combination has undergone successful clinical evaluation in both Africa and Asia (2, 8, 22, 23, 52). The previous use of PQP as monotherapy, however, demonstrates that selection of resistance is possible in the field, with clinical treatment failure correlating with an increase in in vitro 50% inhibitory concentrations (IC₅₀s). These values reportedly attained 1.7 μM in Hainan Island, China (14, 59, 60). The mechanism by which resistance is mediated, however, remains unknown.

Using a newly recloned line of P. falciparum Dd2, a strain prone to the acquisition of drug resistance (40), we describe the selection of P. falciparum parasites resistant to PQP and present genetic and biochemical analyses of these mutant parasites. Whole-genome comparisons between the parental clone, PQP-resistant clones, and PQP-revertant clones (cultured in the absence of drug selection pressure with eventual loss of drug resistance) led us to identify an association between PQP resistance and a copy number variation (CNV) on chromosome 5 (ch5).

MATERIALS AND METHODS

Parasites.

Experiments described in this study were performed with an isogenic line of P. falciparum Dd2, which was newly recloned using limiting dilution (19) and which is referred to as Dd2 1pa (11, 43). Parasites were cultured in an asynchronous manner in vitro using standard conditions and P. falciparum culture medium (16, 56) supplemented with 5% AlbuMAX (Invitrogen, Carlsbad, CA). Parasites were isolated from infected erythrocytes by treatment with 0.1% (wt/vol) saponin prior to nucleic acid purification.

Selection of piperaquine-resistant parasites.

Selection of resistant parasites was conducted as described previously (39). Briefly, 7G8 and Dd2 1pa parasites were propagated in vitro in red blood cells (RBC), and triplicate 60 ml cultures with an initial inoculum of 8.5 × 10⁹ or 3.3 × 10⁹ infected RBC were challenged, respectively, with either 47 nM or 140 nM PQP (AvaChem Scientific, San Antonio, TX) added to the culture medium. These concentrations were based on the peak drug concentrations measured in human plasma, when the compound is administered alone or with a meal high in fat, respectively (2, 50, 53–55). In addition, triplicate flasks were inoculated with 10 infected RBC of each strain as a growth control to assess when parasites would first be detected by microscopy. The medium was changed every other day, maintaining the initial drug concentration, and fresh RBC were added once per week. Cultures were maintained under continuous drug pressure for 80 days. Giemsa-stained blood smears of cultures were examined to detect parasite outgrowth, with a parasitemia of 2% infected RBC set as the standard for defining a positive culture. Clones of resistant parasites from drug-challenged cultures were obtained by limiting dilution. The stability of the PQP-resistant phenotype was determined by propagation of clonal lines in the absence of selective drug pressure.

Proliferation assays and IC₅₀ analysis.

In vitro responses to PQP and other antimalarials were calculated from 72-h [³H]hypoxanthine incorporation assays, using previously described methods (9) and low hypoxanthine medium (16). Incorporation was measured using a MicroBeta liquid scintillation counter (PerkinElmer, Waltham, MA). The percent reduction in hypoxanthine uptake (a marker of growth inhibition) was calculated as follows: reduction = 100 × [(geometric mean cpm of no-drug samples) − (mean cpm of test samples)]/(geometric mean cpm of no-drug samples). IC₅₀s were determined by nonlinear regression analyses (24). DHA and pyronaridine were acquired from AvaChem Scientific. Chloroquine, quinine, and mefloquine were purchased from Sigma-Aldrich (St. Louis, MO). Lumefantrine and monodesethylamodiaquine were generously provided by Philip Rosenthal (UCSF) and Pascal Ringwald (WHO), respectively.

Whole-genome comparative hybridization analysis.

Genomic DNA was isolated by standard phenol-chloroform extraction. Fifteen micrograms of genomic DNA from each clone and 2.5 ng each of bioB, bioC, bioD, and Cre Affymetrix control plasmids (Affymetrix, Santa Clara, CA) were fragmented with DNase I and end-labeled with biotin (58). The samples were hybridized to the microarrays overnight at 45°C in Affymetrix buffers, washed, and scanned using a modified protocol with wash temperatures of 23°C to account for the high AT content of P. falciparum (10, 11). Hybridizations were performed with a custom high-density Affymetrix microarray (11) that contains over 4.8 million probes to the sequenced 3D7 isolate. Polymorphisms and CNVs were detected using our custom MATLAB-based software package (11). Briefly, to detect polymorphisms, we searched for sets of three consecutive overlapping probes that had significantly lower hybridization in one of either the resistant or revertant clones compared to the parental reference Dd2, as determined by z tests with an empirically derived standard deviation and a P value cutoff of 1 × 10⁻⁸. Scanning for gene CNVs was performed by calculating the log₂ ratio of the mean intensity for probes to a gene from a resistant or revertant clone divided by the mean intensity of that gene in the reference strain (11). To determine the boundaries of amplification or deletion events, we scanned through the regions surrounding the amplifications and performed a paired t test comparing the probe intensities on either side of each position on a chromosome with a window size of 2 kb, as described previously (11).

Chloroquine resistance transporter genetic analysis.

Total RNA was isolated from saponin-lysed asynchronous parasites using TRIzol reagent (Invitrogen). Synthesis of cDNA was achieved using Superscript III (Invitrogen), with duplicate cDNA synthesis reactions performed for each template. PCRs to amplify pfcrt were conducted using Bio-X-Act short polymerase mix (Bioline, Taunton, MA) with the primers CF5B and BB116B (see Table S1 in the supplemental material). Amplifications were performed with 4 min of hot-start PCR at 95°C, followed by 30 cycles of denaturing for 30 s at 95°C, annealing for 30 s at 54°C, and extension for 2 min at 68°C. PCR products were purified for sequencing using a QIAquick PCR purification kit (Qiagen, Valencia, CA). Direct sequencing of the purified PCR products was conducted using the following primers: CF5B, AF12B, AF22B, BF107B, BB84, AB17, AB20B, AB25B, and BB116B (GeneWiz, South Plainfield, NJ) (see Table S1). Sequences were analyzed using Lasergene analysis software (DNAStar, Madison, WI). Multisequence alignments were performed using ClustalW (http://www.ebi.ac.uk/clustalw/).

Genomic DNA preparation and quantitative real-time PCR assays.

Genomic DNA was isolated from saponin-lysed parasites using standard phenol-chloroform extraction methods. The pfmdr1 copy number was determined as previously described (36). Briefly, reactions were conducted in triplicate on two independent genomic DNA preparations of each line (P. falciparum Dd2 1pa, PQP clone 1, PQP clone 2, PQP rev 1, and PQP rev 2, as well as 3D7 and FCB, which were included as copy number controls).

Primers and hybridization probes to detect the copy number of PFE1010w (PFE1010wF-hp and PFE1010wR-hp) and PFE1085w (PFE1085wF-hp and PFE1085wR-hp) were designed using Primer Express (Applied Biosystems, Foster City, CA). Hybridization probes for PFE1010w (6-carboxyfluorescein [FAM]-AGAGGATCAAAAAAATGGGAGGAGATGTCATATG-Black Hole Quencher) and PFE1085w (FAM-AATGGCATTGATTTACCAACGGCTATTCAAC-Black Hole Quencher) were produced by Biosearch Technologies (Novato, CA). Plasmid constructs for each target and the beta-tubulin gene (as a reference) were generated by PCR amplification using P. falciparum Dd2 1pa genomic DNA as a template. The primers PFE1010-2F and PFE1010-2R were used to amplify a fragment of PFE1010w, the primers PFE1085wF and PFE1085wR were used to amplify PFE1085w, and the primers Betatub-2F and Betatub-2R were used to amplify a fragment of the beta-tubulin gene (see Table S1 in the supplemental material). Products were amplified using the Bio-X-Act short polymerase mix, with the following conditions: 3 min of hot-start PCR at 95°C, followed by 30 cycles of denaturing for 30 s at 95°C, annealing for 30 s at 54°C, and extension for 1 min at 68°C. PCR products were ligated into the pCR 2.1-TOPO vector (Invitrogen). Plasmid preparations of these constructs were treated with RNase A and purified using a Qiagen plasmid minikit (Qiagen, Valencia, CA). Plasmid yields were determined using a NanoDrop 1000 spectrophotometer (Thermo Fisher Scientific, Wilmington, DE).

Reactions were performed in 25-μl volumes with TaqMan 2× universal PCR master mix (Applied Biosystems), 300 nM each primer, 100 nM hybridization probe, and 50 ng genomic template. A three-step PCR was performed on a DNA Engine Opticon 2 (Bio-Rad Laboratories, Hercules, CA) for 35 cycles, with denaturation at 94°C for 20 s, annealing at 55°C for 20 s, and extension at 72°C for 30 s. Melting curve analysis was performed for each assay to verify that a single melting peak was produced, indicating a single specific PCR product for each reaction. A standard curve for each reaction was used to determine gene copy number, generated with 10-fold serial dilutions (in the range of 30 to 3 × 10⁵ plasmid copies) of plasmid-prepared DNA for the beta-tubulin gene, PFE1010w, and PFE1085w. Each sample was run in triplicate on three separate occasions.

trans expression of candidate genes.

Candidate genes were amplified from cDNA prepared from total RNA isolated from saponin-lysed P. falciparum Dd2 1pa asynchronous parasites. Synthesis of cDNA was achieved using SuperScript III. PCRs to amplify each candidate gene were conducted using Bio-X-Act short polymerase mix (for the respective primers, see Table S1 in the supplemental material). Forward primers were designed to contain an AvrII restriction site, and reverse primers contained a MluI restriction site for cloning into the expression vector. Amplifications were performed with 4 min of hot-start PCR at 95°C, followed by 30 cycles of denaturing for 30 s at 95°C, annealing for 30 s at 54°C, and extension for 45 s/kilobase at 68°C. Following amplification, reaction mixtures were purified using Qiagen PCR purification kits. Purified amplicons were subsequently digested with both AvrII and MluI and treated with Antarctic phosphatase (New England BioLabs, Ipswich, MA). Purified fragments were then ligated using T4 ligase into a complementary digested expression vector (pDC2-attP-BSD-1600pfcrt-gene-gfpmut2; derived from pDC [15]), with the transgene being driven by a 1.6-kb fragment of the pfcrt promoter and being fused at the 3′ end to gfpmut2 (6). Plasmids were transformed into Escherichia coli XL10 competent cells. Correct insertion into the expression vector was verified by partial end sequencing of plasmid preparations, using a pfcrt promoter-specific primer and a gfp-specific antisense primer (see Table S1). Sequences were evaluated using Lasergene 8.0 (DNAStar, Madison, WI). E. coli clones possessing correct expression constructs were then electroporated into P. falciparum, using previously published methods (1, 31). Populations of recombinant parasites that were positive for green fluorescent protein (GFP) fluorescence (indicating expression of the transgene), were analyzed by in vitro [³H]hypoxanthine incorporation assays to test for modulation of PQP susceptibility.

Piperaquine accumulation assays.

These were performed using conditions previously described for [³H]CQ (47, 48). Briefly, infected RBC were purified using magnetic columns (Miltenyi Biotec, Auburn, CA). Purified erythrocytes were then incubated with 5 nM [³H]PQP (American Radiolabeled Chemicals, St. Louis, MO). The amount of radioactivity taken up by the cells and that remaining in the extracellular medium was determined at 0, 1, 4, 10, 15, 20, 40, and 60 min after addition of the label, as previously described (the time zero sample was collected within ∼20 s of drug addition) (47, 48). The intracellular PQP concentration (PQP_in) was calculated from the amount of [³H]CQ taken up by the cells and by assuming that the volume of a trophozoite-infected erythrocyte is 75 fL (45). PQP accumulation was then expressed as the ratio of the intracellular versus extracellular PQP concentration (PQP_in/PQP_out) (47, 48).

RESULTS

Generation and drug-susceptibility characterization of PQP-resistant parasites.

To identify genetic determinants capable of mediating resistance to PQP, we applied a constant drug-selection pressure on a strain of 7G8 and a freshly recloned P. falciparum Dd2 (Indochina) line (termed Dd2 1pa). Two different drug concentrations (140 and 47 nM) were used based on the peak plasma levels observed following oral dosing with high-fat or low-fat meals, respectively (2, 50, 53, 55). In cultures treated with 47 nM PQP, one of three Dd2 1pa flasks became positive for parasite growth by day 54 (see Fig. S1 in the supplemental material). Flasks without drug, inoculated with 10 infected RBC as a growth control, were positive by microscopy on day 15. In contrast, after 80 days of continuous culturing all the flasks inoculated with 7G8 and the two remaining Dd2 flasks challenged with 47 nM, as well as all Dd2 flasks challenged with 140 nM PQP remained negative for parasites (Fig. 1; see also Fig. S1 in the supplemental material). Resistant Dd2 1pa clones obtained from the 47 nM PQP-treated flask demonstrated a nearly 100-fold increase in IC₅₀s (mean ± standard error of the mean [SEM] PQP IC₅₀ values of 2,090 ± 152 nM and 1,691 ± 213 nM for clones 1 and 2, respectively) compared to the drug-sensitive parental clone (17.0 ± 2.0 nM; P < 0.0001) (Fig. 2A; see also Table S2 in the supplemental material).

Fig. 1.

Flow diagram illustrating the selection of piperaquine-resistant P. falciparum lines and derivation of the revertant clones following prolonged culture in the absence of drug pressure. Piperaquine-resistant clones 1 and 2 were cultured under 47 nM continuous piperaquine pressure and cloned by limiting dilution. All genetic and phenotypic analyses were conducted with cloned lines, other than when we assessed the stability of the piperaquine-resistant phenotype, which was measured with lines maintained without piperaquine pressure for 70 days. The parental Dd2 1pa clone, the piperaquine-resistant clones 1 and 2, and the piperaquine-revertant clone 1 were further analyzed by whole-genome hybridization.

Fig. 2.

In vitro antimalarial response of the piperaquine-resistant and revertant clones. In vitro [³H]hypoxanthine incorporation assays (72 h) were performed with the piperaquine-resistant clones, revertant clones, and the parental Dd2 line, which were tested in duplicate against each antimalarial drug on 4 to 15 separate occasions. IC₅₀s (shown as means ± SEMs) were derived by nonlinear regression analysis. Numerical values are listed in Table S2 in the supplemental material. For statistical comparisons, Mann-Whitney U tests were performed (P values of <0.05 [*], <0.01 [**], and <0.001 [***], unless indicated comparison is with the Dd2 parental line). CQ, chloroquine; DHA, dihydroartemisinin; LMF, lumefantrine; mdAQ, monodesethylamodiaquine; MFQ, mefloquine; PQP, piperaquine.

To address the stability of the resistance phenotype both clones were cultured for 70 days without drug pressure, at which time they continued to demonstrate PQP resistance (see Table S2 in the supplemental material). Further culturing, for 30 days, led to the loss of the drug resistance phenotype. The resulting line, derived from PQP clone 1, was cloned by limiting dilution, yielding the PQP rev 1 clone that produced a PQP mean ± SEM IC₅₀ of 37.5 ± 8.5 nM. Additional attempts to derive PQP-resistant isolates from the same P. falciparum Dd2 1pa clone using either “single-step” or a “stepwise” selection regimen did not yield drug-resistant parasites (data not shown), likely reflecting the low frequency of resistance to this antimalarial drug. An additional PQP-revertant clone was generated by the continued culturing of PQP clone 1 in the absence of drug pressure for 120 days, resulting in PQP rev 2 (PQP IC₅₀, 22.4 ± 5.4 nM) (Fig. 2A; see also Table S2).

To determine whether PQP resistance affected parasite susceptibility to other widely used antimalarials, we also determined IC₅₀s for chloroquine, monodesethylamodiaquine (mdAQ; the active metabolite of amodiaquine), mefloquine, lumefantrine, and DHA (Fig. 2; see also Table S2 in the supplemental material). In contrast to the Dd2 parental line that is CQ resistant (IC₅₀ of 136.9 ± 31.8 nM), both PQP-resistant clones, as well as the two PQP-sensitive revertant clones, demonstrated a CQ-sensitive phenotype (CQ IC₅₀s for PQP clone 1, 24.4 ± 2.6 nM; for PQP clone 2, 26.3 ± 6.1 nM; for PQP rev 1, 46.1 ± 7.9 nM; and for PQP rev 2, 49.4 ± 7.9 nM) (Fig. 2B; see also Table S2). In a similar manner, the PQP-resistant clones 1 and 2 demonstrated mdAQ IC₅₀s of 19.5 ± 3.6 nM and 21.6 ± 2.1 nM, respectively, which were both significantly lower than the Dd2 mdAQ IC₅₀ of 73.2 ± 4.8 nM (P < 0.0001) (Fig. 2C; see also Table S2). We note that both revertant clones displayed intermediate IC₅₀s for chloroquine, which differed significantly from both the parental Dd2 clone and both PQP-resistant clones (Fig. 2B; see also Table S2). In addition, both revertant clones possessed a significantly increased IC₅₀ for mdAQ compared to that for the PQP-resistant clones; however, only the mdAQ IC₅₀ for PQP-revertant clone 1 was significantly lower than that for the Dd2 1pa parental isolate (Fig. 2C; see also Table S2). The observed phenotypes for both resistant and revertant clones were similar for mefloquine and lumefantrine, compared to that for Dd2, with both the PQP-resistant and revertant clones demonstrating significantly increased susceptibilities (Fig. 2D to E; see also Table S2). These data provide evidence that selection for PQP resistance was associated with increased susceptibility to mefloquine and lumefantrine that was sustained despite the subsequent loss of PQP resistance. In our 72-h assays, there was no detectable modulation of DHA susceptibilities in either the resistant or revertant clones (Fig. 2F; see also Table S2). In addition, there was no significant modulation of susceptibility to pyronaridine (see Table S2) in the PQP-resistant or revertant clones.

Decreased accumulation of [³H]PQP in drug-resistant parasites.

Under the conditions tested, the time course of PQP incorporation differed profoundly between the two parasite clones investigated (Fig. 3). In the PQP-sensitive parasite clone Dd2, the amount of internalized PQP rapidly achieved high levels and further increased with time until it reached saturation after 30 to 40 min. In contrast, the PQP-resistant parasite clone PQP clone 1 failed to incorporate appreciable levels of [³H]PQP over the entire 60-min time course.

Fig. 3.

Time course of piperaquine accumulation in P. falciparum. Cell suspensions of either Dd2 (PQP sensitive [black circles]) or PQP clone 1 (PQP resistant [orange squares]) were incubated in 5 nM [³H]piperaquine at 37°C, and the amounts of internalized piperaquine were determined at the time points indicated. Results represent the means ± SEMs of three independent experiments.

Genetic characterization of PQP-resistant parasites.

To begin to investigate the genetic basis of PQP resistance, genomic DNA from each of the two PQP-resistant clones, along with one revertant clone and the parental Dd2 1pa line, was extracted, purified, digested, labeled, and hybridized to a custom microarray that tiles through the P. falciparum genome at two- or three-base-pair spacing (NCBI Gene Expression Omnibus accession number GSM689142) (11). This microarray design allows for the detection of polymorphisms in approximately 90% of coding regions and 60% of noncoding regions, based on previous validation studies that place confidence on the identification of a single-nucleotide polymorphism (SNP) when that nucleotide is represented by at least three oligonucleotide probes that give high-quality fluorescence data (on average, each unique sequence is represented by 7 to 9 oligonucleotide probes) (11). This coverage is more limited in regions with high AT content and in repetitive regions of the P. falciparum genome. For our experiments, probe coverage for each P. falciparum gene ID is listed in Table S3 in the supplemental material.

Compared to results for the parent clone, our tiling array analysis revealed a novel SNP in pfcrt (P. falciparum chloroquine resistance transporter; MAL7P1.27) that localized to predicted transmembrane domain 2. This was the sole polymorphism that was identified in both PQP-resistant clones, and it was also observed in clone PQP rev 1 (Fig. 4B; see also Table S4 for a listing of all polymorphisms identified with a P value cutoff of 1 × 10⁻⁸).

Fig. 4.

Genetic analysis of piperaquine-resistant parasites. (A) Genomic tiling arrays identified shared-copy-number variations in the piperaquine-resistant clones compared to those of the parental Dd2 line. Log₂ ratios of probe intensities, plotted along chromosome 5 for the hybridization of genomic DNA from PQP clone 1 relative to the Dd2 parental line, demonstrated an increased copy number of ch5 with the approximate breakpoints 825600 and 888300 (probe intensity reflects the comparative abundance of genomic DNA from the two lines hybridized to the microarray; orange indicates no difference or similar genomic abundance; white/yellow indicates higher probe intensity for PQP clone 1 compared to that for Dd2 1pa; red/brown indicates lower probe intensity for PQP clone 1 or rev 1 compared to that for Dd2 1pa). In addition, there was a reduction in the copy number of the region surrounding pfmdr1 (888300 to 970100). Comparison of the PQP clone 1 and revertant clone 1 found deamplification of the region that was amplified in PQP clone 1 (825600 to 888300) and further deamplification of the region surrounding pfmdr1 (888300 to 970100). Locations of PFE1010w, PFE1085w, and pfmdr1 are noted. (B) Tiling array analysis also identified a polymorphism in pfcrt. PCR amplification and sequencing revealed this polymorphism to be a nonsynonymous change in the coding region, encoding the predicted point mutation C101F. Copy number variations identified by genomic tiling array were confirmed by quantitative PCR for PFE1010w (ch5, 831614 to 834340) (C), PFE1085w (ch5, 882373 to 884898) (D), and pfmdr1 (PFE1150w; ch5, 957885 to 962144) (E). P. falciparum 3D7 and FCB were included as controls. 3D7 has been demonstrated previously to have 1 copy of pfmdr1, and FCB has 2 copies.

We also identified deamplification of an 82-kb region of chromosome 5 that includes pfmdr1 (multidrug resistance protein; PFE1150w) from three copies in Dd2 to two copies in the resistant clones and one copy in the revertant clone (Fig. 4A). Finally, we identified the amplification of an adjacent 63-kb region of chromosome 5 (ch5) that was present only in the resistant clones. This region deamplified back to a single copy in the revertant clones, and thus represented the only genetic change in the resistant clones and the revertant clone 1 that was consistent with the gain and subsequent loss of the resistance phenotype (Fig. 4A). A genome-wide representation of the tiling array data that compared Dd2 1pa and PQP clone 1 and that shows the copy number changes on chromosome 5 is illustrated in Fig. S2 in the supplemental material. We note that this microarray depicts the direct juxtaposition of the amplification and deamplification events; however, it is unclear if this actually occurred, since the microarray is not a direct representation of the physical genome. Further studies are required to evaluate if these copy number events are immediately adjoining or physically separate events.

To confirm the whole-genome hybridization findings of a possible SNP in pfcrt, the full-length coding sequence was amplified from cDNA and fully sequenced from the parental Dd2 clone, both PQP-resistant clones, and PQP rev 1. This revealed a single nucleotide substitution, consisting of a guanine-to-thymine transversion, in both PQP-resistant clones as well as PQP rev 1 compared to the parental Dd2 pfcrt sequence (GenBank accession number JF520758). This mutation is predicted to alter the amino acid sequence of PfCRT at position 101, changing a cysteine to a phenylalanine. For PQP rev 2, only positions 458600 to 459386 were sequenced directly from cDNA, again demonstrating the same SNP and same predicted C101F mutation.

Hybridization probe-based quantitative PCR (qPCR) confirmed that indeed the ch5 segment (breakpoints, ∼888300 and 970100), which includes pfmdr1, was deamplified from three copies in the parental Dd2 isolate (copy number amplification resulting from in vitro mefloquine drug selection [33]) to two copies in both PQP-resistant clones and a single copy in both of the PQP revertant clones (Fig. 4A, as evaluated following analysis of pfmdr1).

To confirm the CNV on the upstream segment of ch5, qPCR was performed evaluating both PFE1010w and PFE1085w, genes located near the 5′ and 3′ boundaries of the predicted amplification. These experiments confirm the comparative genome hybridization analysis, indicating that this segment of ch5 (breakpoints, ∼ 825600 and 888300) was indeed amplified in the two PQP-resistant clones. However, in both independently generated revertant clones, the amplification of this region was lost, reverting to the original copy number of one.

Evaluation of candidate genes by overexpression in trans.

To evaluate the predicted open reading frames (ORFs) that were present in the CNV on ch5 associated with the PQP resistance phenotype (825600 to 888300), we attempted to express each ORF individually in trans in a PQP-sensitive background. Overexpression was either attempted as an episomal construct in PQP rev 1 or integrated into a NF54 line harboring an attB site (S. Adjalley, unpublished line) to facilitate rapid Bxb1 integrase-mediated transgene integration (31). We were unable to clone the predicted open reading frames for PFE1045c, PFE1060c, and PFE1070c due to their long length (6,165 bp, 4,581 bp, and 4,038bp, respectively), which precluded amplification from cDNA or resulted in rearrangements of the plasmid during cloning in E. coli. Of the 20 ORFs predicted in the ch5-amplified segment, we were able to clone and introduce 13 into P. falciparum, all of which were expressed in trans (as confirmed by the detection of GFP fluorescence). However, none of these revealed a significant difference in the PQP susceptibility (Table 1). Despite repeated attempts, we were unable to obtain recombinant parasites that overexpressed PFE1000c, PFE1025c, PFE1080w, or PFE1090w in trans in either the PQP rev clone 1 or the engineered NF54 attB line background. One likely explanation is that expression in trans using a heterologous promoter and/or 3′ untranslated region was toxic to the transformed parasites. Work is underway to clone the native promoters and 3′ untranslated regions from these ORFs.

Table 1.

Piperaquine IC₅₀s of parasite lines expressing candidate chromosome 5 genes in trans and predicted gene annotations^a

Gene	Recipient parasite line		Gene annotation
	PQP rev 1	NF54 attB
PFE1000c	NV	NV	Conserved Plasmodium protein; unknown function
PFE1005w	13.3 ± 0.9		40S ribosomal protein S9; putative
PFE1010w	19.0 ± 0.5		Protein phosphatase; putative
PFE1015c	27.4 ± 7.0		Conserved Plasmodium protein; unknown function
PFE1020w	18.9 ± 8.3		U6 snRNA-associated Sm-like protein Ism2; putative
PFE1025c	NV		Conserved Plasmodium protein; unknown function
PFE1030c	14.9 ± 1.3		Phosphomethylpyrimidine kinase; putative
PFE1035c	19.7 ± 1.3		Diadenosine tetraphosphatase; putative
PFE1040c	19.7 ± 1.5		Conserved Plasmodium protein; unknown function
PFE1045c	ND	ND	Conserved Plasmodium protein; unknown function
PFE1050w	17.7 ± 0.8		S-Adenosyl-l-homocysteine hydrolase
PFE1055c	NV	34.2 ± 25.0	Conserved Plasmodium protein; unknown function
PFE1060c	ND	ND	Conserved Plasmodium protein; unknown function
PFE1065w	20.0 ± 3.2		Conserved Plasmodium protein; unknown function
PFE1070c	ND	ND	Conserved Plasmodium protein; unknown function
PFE1075c	12.3 ± 1.3		Conserved Plasmodium protein; unknown function
PFE1080w	NV	NV	Ribosomal large subunit pseudouridylate synthase; putative
PFE1082c	12.1 ± 3.6		Conserved Plasmodium protein; unknown function
PFE1085w	18.7 ± 3.7		DEAD/DEAH box ATP-dependent RNA helicase; putative
PFE1090w	NV	NV	Nucleotide binding protein; putative

^aIC₅₀s were derived by nonlinear extrapolation of drug inhibition data generated from two independent 72-h [³H]hypoxanthine incorporation assays performed in duplicate. Values are means ± SEMs, shown in nM. PQP, piperaquine; ND, not done; NV, no viable parasites were recovered from transformation/selection regimen. Gene annotations were acquired from PlasmoDB version 7.1 (http://plasmodb.org/).

DISCUSSION

Current efforts to reduce the burden of malaria are threatened by the emergence in Western Cambodia of parasites that possess a delayed clearance time when treated with artesunate monotherapy or artesunate-mefloquine combination therapy (12). However, the apparent reduced drug susceptibilities of these parasites is not currently associated with clinical treatment failures. Factors demonstrated to increase ACT treatment failure rates include poor treatment adherence, poor pharmacokinetic profiles of the administered drugs, and parasite resistance to the ACT partner drug (37). A recent study evaluating artesunate partnered with sulfadoxine-pyrimethamine (SP) in Southern Mozambique demonstrated an overall decrease in the asexual parasite prevalence in the study population. However, there was also an increase in the prevalence of SP-resistant parasites possessing the DHFR/DHPS quintuple mutations from 11% in 2004 to 75% by 2008 (38). Indeed, the increasing prevalence of SP-resistant parasites in Mozambique has hampered efforts to introduce artesunate-SP as first-line treatment of uncomplicated malaria (38). This study underscores the necessity to understand the genetic and molecular basis of resistance to ACTs, which would permit surveillance efforts for the emergence of parasites resistant to the ACT partner drugs as a key component of the effective utilization of ACTs.

One of the major initial concerns with the use of PQP was the suspected cross-resistance with chloroquine and amodiaquine, due to the structural similarities of the compounds (PQP is a bis-chloroquine derivative) (8, 41). Because of the shared chemical structures, it was postulated that these drugs could potentially share similar modes of action (involving inhibition of heme detoxification in the parasite digestive vacuole [21]) and resistance (41). Support for similar mechanisms of action comes from investigations in the rodent parasite Plasmodium berghei that reported swollen digestive vacuoles and abnormal hemozoin clumping in PQP-treated parasites (8). In addition, both PQP and CQ possess tertiary nitrogens, which would become protonated in the acidic environment of the digestive vacuole, limiting their membrane permeability and facilitating drug accumulation. However, PQP remains active against CQ-resistant isolates (3), indicating that although PQP is closely related to CQ, resistance to these two drugs appears to be mediated by distinct mechanisms. This is supported by numerous clinical trials of DHA-PQP that show good therapeutic efficacy in areas with a high prevalence of CQ-resistant parasites (4, 22, 23, 26, 27, 29, 30, 42, 51, 52).

Using recombinant engineered lines, a study by Muangnoicharoen et al. found only a modest trend of increasing PQP IC₅₀s associated with mutant PfCRT haplotypes in recombinant parasites (3.9 nM for the CQ-sensitive haplotype compared to 11.5 nM for the CQ-resistant haplotype; this contrasted with CQ IC₅₀s of 22.9 nM and 143.8 nM, respectively [28]). Also, a recent study by Briolant et al. that assessed drug responses in clinical isolates did not find a correlation between PQP and other quinoline drug susceptibilities, or polymorphisms in pfcrt, pfmdr1, pfmrp, or pfnhe1, that have been demonstrated to modulate quinoline susceptibility (5). Nevertheless, in our study of in vitro-selected resistant parasites, PQP resistance was inversely correlated with CQ and mdAQ susceptibilities.

Analysis of the PQP-resistant parasites and aminoquinoline drug susceptibilities suggests the lack of a conserved mechanism of cross-resistance between PQP and either CQ or the amodiaquine metabolite mdAQ in this genetic background. This contrasts with observed cross-resistance between CQ and amodiaquine (20, 32, 44). Interestingly, in both independently derived PQP revertants, the mdAQ susceptibility had decreased relative to the PQP-resistant parasites, although this did not attain levels found in the Dd2 parental strain. There was also an increase in the susceptibility to both mefloquine and lumefantrine in the PQP-resistant clones, associated with the deamplification on ch5. This chromosomal segment includes pfmdr1, which has previously been associated with resistance to these two antimalarials (36, 49). This finding is also potentially relevant with previous work demonstrating that resistance to CQ and amodiaquine in P. falciparum is inversely correlated with resistance to arylamino alcohols (mefloquine and lumefantrine) (13), and the selection of resistance to mefloquine results in an increase in CQ susceptibility (7, 35). This provides further evidence for drug susceptibility interactions between quinolines and arylamino alcohols and suggests related biological roles played by the two digestive vacuole membrane-resident transporters PfCRT and PfMDR1. Of note, the PQP susceptibility of the parental Dd2 clone, which has three copies of pfmdr1, was not significantly different from the PQP-revertant clones (Fig. 2A; see also Table S2 in the supplemental material) that possess a single pfmdr1 locus, suggesting that pfmdr1 amplification mediates little to no resistance to PQP in this genetic background. Given that the only strain in which we were able to obtain resistant parasites was Dd2, which possesses three copies of pfmdr1 (the 86Y allele), in addition to the CQ-resistant Dd2 allele of pfcrt, we are not able to ascertain the importance of pfmdr1 copy number changes as potential compensatory mutations and can state only that these were insufficient to mediate PQP resistance.

Studies in both P. falciparum and Xenopus laevis oocyte expression systems have demonstrated that mutations in PfCRT that are linked to parasite CQ resistance also associate with decreased accumulation of CQ in P. falciparum parasites or increased transport kinetics of CQ in transfected oocytes (22, 25, 45, 46). Strikingly, in our PQP-pressured lines, PQP resistance was associated with increased sensitivity to CQ. Our observation that the novel C101F mutation in pfcrt was retained in both independently generated PQP-revertant clones, which are sensitive to both PQP and CQ, implies that this mutation alone could not have mediated PQP resistance. However, this mutation may have played a role in the loss of CQ resistance in the PQP-resistant clones. It also remains possible that the PfCRT mutation was necessary but not sufficient for PQP resistance, or that it was selected to restore some loss of fitness caused by other genetic changes that themselves contributed directly to PQP resistance. Experiments are underway to evaluate the role played by this PfCRT mutation in modulating PQP and CQ susceptibility in the context of the CQ-resistant Dd2 PfCRT haplotype. Of note, the 12 introns present in pfcrt make it difficult to amplify the entire ORF (especially from filter papers that are commonly used to sample blood from infected individuals). Accordingly, genetic epidemiological studies usually analyze only a specific sequence (typically the region containing the K76T mutation that is a sensitive marker of CQ resistance) that do not, for instance, encompass codon 101. Our results, however, suggest that other mutations in pfcrt not normally analyzed may also significantly affect parasite susceptibility to CQ.

Because the only detected genetic variation that associated with the PQP resistance phenotype was the amplification on ch5 (nucleotides 825600 to 888300), we hypothesized that overexpression of one of the genes in this genomic segment might be responsible for the observed drug resistance phenotype. To test this, we attempted to clone and express each predicted ORF in trans in a PQP-sensitive background, and evaluate PQP susceptibility in the recombinant lines. We were successful in expressing 13 of the 20 predicted ORFs in trans, as determined by the addition of a C-terminal GFP fusion (in all successful constructs, GFP was localized to the parasite cytoplasm) (Table 1). However, no recombinant line with confirmed expression of the gene fusion had any significant modulation of the PQP susceptibility. This possibly suggests that one of the gene constructs that failed to yield viable parasites or one of the untested long ORFs (>4 kb) may be responsible for the resistance phenotype. Other possibilities are that the C-terminal fusion, used to verify transcription/translation, perturbed the function of one of the evaluated gene products, or that multiple ORFs in the amplified region contribute together to the PQP-resistant phenotype. Additional work is underway to address these possibilities. Furthermore, due to the limitations in genome coverage of the tiling array analysis, specifically in AT-rich or repeated sequences, we are conducting whole-genome sequencing to obtain additional insight into genetic changes potentially associated with the PQP resistance phenotype.

Both the drug phenotype, manifesting as a 100-fold decrease in susceptibility, along with the drug accumulation that demonstrated the lack of appreciable accumulation in PQP-resistant parasites, are consistent with resistance being mediated by either the exclusion of drug or its efficient transport out of the parasite. CQ resistance mediated primarily by mutations in PfCRT manifests as an ∼6- to 10-fold shift in the CQ IC₅₀ value, and although CQ accumulates in resistant parasites this occurs at a substantially reduced amount compared to that in CQ-sensitive parasites (17, 18). Although none of the predicted ORFs present in the CNV on ch5 (Table 1) have the predicted attributes of a transporter, it is plausible that one may function in a regulatory role, affecting the accumulation of PQP in drug-resistant parasites.

Our study demonstrates that PQP-resistant parasites can be selected by in vitro continuous drug pressure, albeit at a low frequency. The PQP concentration used for our selection studies is comparable to that normally obtained in patient plasma when the drug is administered without dietary supplementation; of note, the comparable concentration obtained upon coingestion with a high-fat meal (140 nM) did not result in the selection of viable parasites, supporting the importance of this pharmacokinetic parameter (50). To our knowledge, this is the first report of the selection of highly PQP-resistant strains of P. falciparum in vitro. Here, we provide evidence that P. falciparum Dd2 resistance to PQP is associated with a CNV on ch5 (localized to 825600 to 888300). This fragment was amplified in both resistant clones and deamplified in both PQP-revertant clones. All other genetic alterations that were identified in the PQP-resistant clones were maintained in the revertant clones or further deamplified, thus associating PQP resistance in these P. falciparum Dd2 parasites with the chr5 CNV. Identification of the PQP resistance determinant in these in vitro-selected lines will permit studies into molecular/biochemical pathways that could mediate resistance in the field.