Knockout of the peroxiredoxin 5 homologue PFAOP does not affect the artemisinin susceptibility of Plasmodium falciparum

Artemisinins are the current mainstay of malaria chemotherapy. Their exact mode of action is an ongoing matter of debate, and several factors have recently been reported to affect an early stage of artemisinin resistance of the most important human malaria parasite Plasmodium falciparum. Here, we identified a locus on chromosome 7 that affects the artemisinin susceptibility of P. falciparum in a quantitative trait locus analysis of a genetic cross between strains 7G8 and GB4. This locus includes the peroxiredoxin gene PFAOP. However, steady-state kinetic data with recombinant PfAOP do not support a direct interaction between this peroxidase and the endoperoxide artemisinin. Furthermore, neither the overexpression nor the deletion of the encoding gene affected the IC50 values for artemisinin or the oxidants diamide and tert-butyl hydroperoxide. Thus, PfAOP is dispensable for blood stage parasite survival, and the correlation between the artemisinin susceptibility and chromosome 7 is probably based on another gene within the identified locus.


Results
QTL mapping reveals PFAOP as a potential artemisinin susceptibility factor. In a previous study we have interrogated the contribution of the genetic background on artemisinin susceptibility by investigating artemisinin responses in the progeny of a genetic cross between the P. falciparum clones HB3 (Latin America) and Dd2 (Southeast Asia). We found that loci on chromosomes 5 (including Pfmdr1), 12 and 13 contribute to altered responsiveness 12 . Here we repeated the study, but this time using the genetic cross between strains 7G8 (Latin America) and GB4 (Africa) 48 . Responses to artemisinin and artemether were determined, using a standard cell proliferation assay where parasites were exposed to the drugs for 72 h 49 . The IC 50 values of strains 7G8 and GB4 were for artemisinin 5.0 ± 0.6 nM and 5.7 ± 1.0 nM and for artemether 2.7 ± 0.4 nM and 3.6 ± 0.8 nM, respectively. Although the parental clones had comparable IC 50 values for artemisinin and artemether, the responsiveness of the 32 progeny investigated covered a range from 2.1 ± 0.3 nM to 7.9 ± 0.6 nM in the case of artemisinin and from 0.3 ± 0.1 nM to 3.6 ± 0.8 nM in the case of artemether (Fig. 1b).
To identify determinants of altered artemisinin and artemether responsiveness, we conducted a QTL analysis by correlating the IC 50 values with the genetic maps of the progeny and the parental clones 48 . One major QTL with a LOD score of 3.1 (p < 0.001) was identified for both drugs, mapping to the genetic marker C7MK11 on chromosome 7. According to the genome of the reference strain 3D7 this locus encodes 49 genes including PFAOP (PF3D7_0729200) ( Table S1). The locus is distinct from the artemisinin susceptibility loci identified in the HB3 × Dd2 cross and it is also distinct from the pfcrt locus on chromosome 7 that is associated with chloroquine resistance and altered susceptibility to quinine and quinine-like antimalarial drugs [50][51][52] . We found no link with the K13 gene on chromosome 13 7 or the PfPI3K gene on chromosome 5 15 . This finding was anticipated since the two parental clones carry identical wild type loci of these two genes.
Artemisinins are neither substrates nor inhibitors of PfAOP. In order to test a potential direct interaction between artemisinins and PfAOP, we first analysed whether recombinant PfAOP converts endoperoxides in an established coupled enzymatic assay (Fig. 2a). The detected consumption of NADPH in the presence of artemisinins or di-tert-butyl peroxide (DTBP) was comparable to the background activity of negative controls without peroxide substrate, whereas the conversion of tert-butyl hydroperoxide (tBOOH) served as a positive (c) Time-dependent irreversible inhibition assays. PfAOP (50 µM) was preincubated with 100 µM artemisinins for up to 10 min on ice before the enzyme was diluted 1:20 in a standard coupled enzymatic assay containing 100 µM tBOOH. Activities were normalized to controls that were preincubated without peroxide. A time-dependent inactivation of PfAOP by tBOOH served as a positive control (open circles) and was compared to a potential inactivation by artemisinins (closed circles). Assays containing DMSO are labelled with an asterisk. All data are the mean ± standard deviation of at least two independent duplicate assays.
control as described previously 45,47 . Thus, PfAOP does not convert endoperoxides and its enzymatic activity is restricted to hydroperoxide substrates.
Next, we analysed whether PfAOP is inhibited by artemisinins. A reversible competitive inhibition was tested for the conversion of tBOOH in the presence of variable concentrations of artemisinins (Fig. 2b). Furthermore, a time-dependent irreversible inhibition was addressed by preincubating recombinant PfAOP with artemisinins, DTBP or tBOOH before transferring the enzyme to standard assays with tBOOH (Fig. 2c). No inhibitory effects were detected except for the positive control with tBOOH, which is not only a substrate but also an irreversible inhibitor that 'overoxidizes' PfAOP 45,47 . In summary, artemisinins are neither reversible nor irreversible inhibitors of recombinant PfAOP and enzymatic assays do not support a direct effect of PfAOP on artemisinins and vice versa. However, our enzymatic assays neither exclude a potential interaction between PfAOP and Fe 2+ -activated artemisinin nor indirect effects of PfAOP on the artemisinin susceptibility of P. falciparum. In order to address these aspects, we genetically manipulated the reference strain 3D7 and determined the artemisinin susceptibility of overexpressing and knockout strains.
Overexpression of PfAOP does not alter the artemisinin susceptibility. Artemisinins have been suggested to promote the formation of reactive oxygen species including H 2 O 2 18, 24-27 . The removal of H 2 O 2 or other hydroperoxides by PfAOP might, therefore, affect the artemisinin susceptibility of P. falciparum. PfAOP has a bipartite topogenic signal (BTS), which is required for correct apicoplast targeting of the dual localized protein 41 . In order to test a potential role of the cytosolic and/or apicoplast form of PfAOP, we sequenced exon 1 (which encodes the BTS) of the parental strains and selected progenies. No single nucleotide polymorphisms or other mutations were found. Next, we overexpressed cytosolic as well as apicoplast-targeted truncated or full length GFP-fusion constructs of PfAOP in 3D7 wild type cells as described previously 41 . In addition, we generated and analysed enzymatically inactive serine mutants 45 of these constructs as controls (Fig. 3a). All GFP-fusion constructs were successfully expressed in P. falciparum resulting, on average, in an approximately 2-to 6-fold increase of the total cellular PfAOP content (Fig. 3b). None of the overexpressed constructs led to an increase of the IC 50 value for artemisinin or had a dominant-negative effect ( Fig. 3c-f). In summary, irrespective of the protein localization or the presence or absence of the catalytic cysteine residue, overexpression of PFAOP does not alter the artemisinin susceptibility of P. falciparum.

PfAOP is not essential and does not protect against artemisinin or external oxidants. To test
whether the complete loss of PfAOP affects the artemisinin susceptibility of P. falciparum, we cloned the plasmid pL7-PFAOP and deleted the encoding gene in strain 3D7 by double crossover using the CRISPR/Cas9 system 53 (Fig. 4a). The genetic manipulation was monitored for three different clones following positive selection and limited dilution. All three clones had lost the endogenous gene and had integrated the selection marker after successful 5′-and 3′-crossovers as revealed by analytical PCR (Fig. 4b). Furthermore, the loss of PfAOP was confirmed by western blotting using a specific antibody 41 (Fig. 4c). Giemsa-stained blood smears of the three different knockout strains were subsequently analysed by light microscopy but revealed no suspicious morphologies. The growth rates were also identical to the wild type strain (Fig. 5). Next, we determined the IC 50 values of the three knockout strains for artemisinin as well as the oxidants tBOOH and diamide as described previously 12,54 . Loss of PfAOP affected none of the IC 50 values as compared to the wild type strain (Fig. 6). Please note that two different batches of artemisinin were used for the two independent sets of experiments in Figs 3 and 6, which presumably explains the systematic deviation of the IC 50 value for artemisinin. The IC 50 values for tBOOH in Fig. 6b (between 90 ± 8 µM and 99 ± 7 µM) and diamide in Fig. 6c (between 75 ± 4 µM and 84 ± 11 µM) were in accordance with previous measurements 54 .
Recent studies revealed that ring-stage survival assays can be advantageous to address a potential artemisinin resistance, for example, because of short artemisinin half-lives and drug-induced parasite dormancy [55][56][57][58] . We therefore analyzed the ring-stage survival percentage of our knockout and overexpressing strains after treatment with 700 nM artemisinin. Wild type strain 3D7 and a mutant NF54 strain that encodes the resistance factor K13 C580Y were analyzed in parallel as negative and positive controls 53 , respectively (Fig. 7). Neither the deletion nor the overexpression of PFAOP had an effect on the ring-stage survival percentage, whereas the survival percentage of the positive control around 12% was twofold increased as compared to strain 3D7. (The difference between the controls was, in our hands, less pronounced as previously reported 53, 56-58 , most likely because we used 700 nM artemisinin instead of dihydroartemisininin in accordance with our QTL analysis, which was also performed with artemisinin). Taken together, the assays in Fig. 7 indicate that PfAOP has no effect on the survival percentage of P. falciparum ring stage parasites under artemisinin pressure.
In summary, PfAOP is not essential in asexual blood stage parasites and is dispensable for the removal of external oxidants such as tBOOH and diamide. Furthermore, the loss of PfAOP taken alone does not affect the artemisinin susceptibility of P. falciparum. Thus, the correlation between the artemisinin susceptibility and chromosome 7 is either based on another gene within the identified locus or might require the interplay of several genetic factors.

Discussion
Immunofluorescence microscopy and transcriptome analyses revealed that dual localized PfAOP is constitutively expressed in blood stage parasites 33,41 . What could be the function of PfAOP in the apicoplast? Specific iron-sulfur clusters are, at least in prokaryotes, the major targets of oxidative challenges by superoxide anions and H 2 O 2 59 . The apicoplast is essential for parasite survival because of the biosynthesis of the isoprenoid precursor isopentenyl pyrophosphate 60,61 . Two enzymes of this biosynthetic pathway -the (E)-4-hydroxy-3-methylbut-2-enyl diphosphate synthase and reductase -bind iron-sulfur clusters 62,63 . Interestingly, two of the five artemisinin susceptibility factors that were identified from Southeast Asian blood samples in the study by Miotto et al. localise to the Constructs encoding cytosolic truncated as well as apicoplast-targeted full length GFP-fusion proteins of PfAOP are depicted below and were overexpressed in strain 3D7 as described previously 41 . Residue Cys117 at the active site of PfAOP was shown to be essential for the hydroperoxidase activity 45 and is labelled with an asterisk. The residue was mutated to serine in controls. (b) Western blot analyses confirming the overexpression of the GFPfusion constructs listed in panel (a). Detection of GFP in the upper blot revealed no significant degradation of the fusion proteins. Hence, detection of PfAOP at approximately 22 kDa reflects only endogenous PfAOP (endo) whereas the upper bands in the same blot reflect the GFP-fusion constructs (fusion). Representative images were obtained by standard western blotting. The estimated ratio between both protein species is summarized on top and is based on semi-quantitative western blotting using a C-DiGit Blot Scanner in parallel experiments. (ce) Artemisinin dose-response curves of strain 3D7 blood stage cultures that express the indicated GFP-fusion constructs from panel a. All data are the mean ± standard deviation of at least three independent triplicate measurements. (f) IC 50 values obtained from panels c-e. None of the differences between the IC 50 values was found to be significant (p > 0.05) based on statistical analyses in SigmaPlot 12.5 using the one way ANOVA method. apicoplast 10 . One of the apicoplast artemisinin susceptibility factors is ferredoxin, which also carries an iron-sulfur cluster 64,65 . The iron-sulfur clusters of ferredoxin and the (E)-4-hydroxy-3-methylbut-2-enyl diphosphate synthase and reductase (as well as other apicoplast proteins) are probably synthesized by an essential machinery within the stroma of the organelle 50 . Considering the presence and relevance of redox-sensitive iron-sulfur cluster-containing apicoplast enzymes, it is quite surprising that the loss of the hydroperoxidase PfAOP has no effect on parasite survival. One possible explanation could be the potential absence of major sources of superoxide anions or hydroperoxides in the apicoplast. However, the presence of an apicoplast superoxide dismutase 66 seems to contradict this null hypothesis. Another explanation for the absent phenotype of our Δpfaop knockout strains is that the cytosol and apicoplast harbor additional hydroperoxidases with redundant functions. Likely candidates in the cytosol are the dithiol Prx-isoform TPx-1, the Prx6-isoform 1-Cys-Prx and imported human Prx-2 37,39,40,67 . The only alternative hydroperoxidase in the apicoplast appears to be the glutathione peroxidase-like thioredoxin peroxidase TPx Gl 68,69 . At this point it is interesting to note that the expression of Plasmodium peroxidases might be interdependent and differ significantly depending on the growth conditions as reported for P. berghei ookinetes 70 . A third explanation for the absent knockout phenotype is that PfAOP exerts another non-essential function, e.g., as a redox sensor that interacts with transducer molecules 29 . However, attempts to trap mixed disulfide bonds between PfAOP and potential interaction partners so far only revealed false positive candidates (data not shown). In summary, the physiological function of dual localized PfAOP remains unknown and the encoding gene is fully dispensable for the growth of asexual blood stage parasites in cell culture.
Neither traditional growth inhibition assays nor ring-stage survival assays revealed an altered artemisinin susceptibilty of PFAOP overexpressing and knockout strains. Since non-essential PFAOP on its own does not affect the artemisinin susceptibility, PFAOP might either contribute to a polygenic trait, or another gene within the identified locus on chromosome 7 might be the cause for the detected effects in Fig. 1. Which of the other genes from Table S1 might be relevant within the identified locus? Among the seven genes that were recently discovered as (potential) artemisinin susceptibility factors by chemogenomic profiling 11 , only one gene (PF3D7_0727100) is located on chromosome 7. PF3D7_0727100 encodes a conserved Plasmodium protein of unknown function, which was also identified in our QTL analysis (Table S1). It is therefore possible that PF3D7_0727100 is the crucial factor within the artemisinin susceptibility locus. Two other candidates from Table S1 -namely PF3D7_0727400, which encodes a putative proteasome subunit alpha type-5, and PF3D7_0729900, which encodes a putative dynein heavy chain -were also identified by mass spectrometry screens using modified artemisinin probes and click chemistry 16,17 . PF3D7_0727400 might be of particular interest because of recent models that link artemisinin susceptibility to proteasome-dependent protein turnover 15,18,20,71 . Future analyses are necessary to address the relevance of the candidates PF3D7_0727100, PF3D7_0727400 and PF3D7_0729900 for artemisinin susceptibility.

Materials. ART, DHA, ATS and DTBP were purchased from Sigma. ATM was from Sciphar. Recombinant
PfAOP with an N-terminal MRGSH 6 GS-tag replacing the first 59 amino acid residues of the full length protein as well as P. falciparum glutathione reductase (PfGR) and P. falciparum glutaredoxin (PfGrx) were produced in Escherichia coli and purified by affinity chromatography as described previously 45,72 . Plasmid pUF1-Cas9 for the expression of Cas9 and plasmid pL6 for guide RNA expression and homologous recombination in P. falciparum 53 were kind gifts from J.J. Lopez-Rubio. Primers were from Metabion. WR99210 was a gift from Jacobus Pharmaceutical Company.
Steady-state kinetics. Coupled enzymatic assays containing recombinant PfAOP, PfGR and PfGrx were carried out at 25 °C using a thermostatted Jasco V-650 UV-visual spectrophotometer as described previously 45,47 . All assays were performed in assay buffer containing 0.1 M Tris-HCl, 1 mM EDTA, pH 8.0. Stock solutions of 4 or 6 mM NADPH, 2 or 4 mM tBOOH, 40 mM GSH and 10 mM ATS were freshly prepared in assay buffer before each experiment. Stock solutions of 10 mM ART, ATM, DHA and DTBP were dissolved in DMSO. Whether PfAOP converts endoperoxide substrates was tested in standard assays containing 150 µM NADPH, 1 U/mL PfGR, 1 mM GSH and 5 µM PfGrx. The consumption of NADPH was monitored at 340 nm (ε = 6.22 mM −1 cm −1 ). A baseline was recorded for 60 sec before the addition of 100 µM ART, ATM, DHA, ATS or DTBP. All assays were subsequently started by adding 2.5 µM recombinant PfAOP. Assays containing 100 µM tBOOH with or without the same volume of DMSO served as positive controls. Initial activities were corrected by subtracting the final slope (ΔAbs/min) of the baseline using the Spectra Analysis program (Spectra Manager version 2, Jasco).
To test whether endoperoxides are reversible inhibitors of PfAOP, standard assays containing 50 µM tBOOH were supplemented with 0-100 µM ART, ATM, DHA, ATS or DTBP before starting the assay with 2.5 µM recombinant PfAOP. The activity was determined and normalized to the negative controls without endoperoxide. A potential irreversible inhibition of PfAOP was analysed in a time-course experiment. Samples containing 50 µM recombinant PfAOP with or without 100 µM ART, ATM, DHA, ATS, DTBP or tBOOH were preincubated on ice for up to 10 min before transferring 50 µL of the mixture to a standard 1 mL assay with 100 µM tBOOH as a substrate. The residual activity was determined in the standard assay and normalized to the activity of negative controls that were preincubated in parallel without peroxide.
Cloning of expression-and knockout constructs. The generation of pARL plasmids for the expression of full length, truncated and mutated fusion constructs between PfAOP and GFP has been described previously 41 . Plasmid pL7-PFAOP was generated as follows: 1) A suitable guide sequence was identified in PfAOP with a confidence value of 0.010 using the protospacer software (http://www.protospacer.com). 2) The 3′-homology region of PfAOP was PCR-amplified using primers PfAOP/3′MCS/pL6/ KO/s (5′-GATCGAATTCCAAACGATTTTACTTCAATAGATAC-3′) and PfAOP/3′MCS/pL6/KO/as (5′-GATCCCATGGCTGATTATTTTTTAAAAACTCTTTTAC-3′) and plasmid pQE30/PFAOP C117S as a template 45 . The PCR product was first subcloned according to the manufacturer's protocol using the TopoTA cloning kit (Invitrogen) and subsequently cloned into pL6 using the restriction sites EcoRI and NcoI. 3) Figure 5. Growth curve analysis. The parasitemia in standard blood stage cultures of strains Δaop#1-3 from Fig. 4 was determined by counting parasites from Giesma-stained blood smears. Wild type strain 3D7 served as a control. All data are the mean ± standard deviation of triplicate measurements for each strain. None of the differences between the parasitemias was found to be significant (p > 0.05) based on statistical analyses in SigmaPlot 12.5 using the one way ANOVA method.
Scientific RepoRts | 7: 4410 | DOI:10.1038/s41598-017-04277-5 The 5′-homology region of PfAOP was PCR-amplified from genomic DNA of strain 3D7 using primers PfAOP/5′MCS/KO/s (5′-GATCCCGCGGCTGTTCTTTTTATTATATGAATGAAGAG-3′) and PfAOP/5′MCS/ KO/as (5′-GATCTCTAGAGGATTCATAAACTTTTTGGGAAAACC-3′). The PCR product was first subcloned into the Topo cloning vector and subsequently cloned into pL6 using SacII and the compatible restriction sites generated by XbaI (insert) and SpeI (pL6). 4) The guide sequence was cloned into pL6 yielding pL7-PFAOP as described previously 53 . Plasmid pL6 containing the 5′-and 3′-homology regions was digested with AvrII Parasite culture, QTL analysis and genetic manipulation. P. falciparum strain 3D7 was cultured at 37 °C according to standard protocols 73 using fresh human A + erythrocytes at a hematocrit of 3% in RPMI 1640 medium that was supplemented with 0.45% (w/v) albumax II, 0.2 mM hypoxanthine and 2.7 μg/mL gentamicin. Progeny of the 7G8 × GB4 cross and the two parental clones were maintained at a hematocrit of 5% in RPMI 1640 medium supplemented with 10% inactivated human A + serum instead of albumax II. Parasites were cultured at 80% humidity, 3 or 5% CO 2 , 5% O 2 and 90% or 92% N 2 . Synchronization of ring stage parasites was carried out with 5% (w/v) sorbitol 74 . The F1 progeny from the GB4 × 7G8 cross were obtained from MR4 48 . The identity of all progeny was verified using eight polymorphic microsatellite markers 51 . QTL analysis was performed according to the method by Haley and Knott 75 in line with previous independent studies 12,51 . The genetic maps have been published 48 .
Transfections were performed as described 76 using 100 μg of plasmid DNA per construct. Transfectants were usually detected in Giemsa-stained blood smears between two and four weeks post transfection. Parasites that were transfected with pARL plasmids were selected with 5 nM WR99210 and the expression of GFP-fusion constructs was confirmed by live cell imaging as described previously 41 . Parasites that were transfected with plasmid pUF1-Cas9 were selected with 100 nM atovaquone. After removal of atovaquone the strain was transfected with plasmid pL7-PFAOP and selected with 5 nM WR99210. Clonal parasite lines were obtained after limiting dilution and analysed by PCR and western blotting. Genomic DNA was isolated with phenol/chloroform as described previously 77 . Primers for analytical PCR reactions 1-3 in Fig. 4 were 3D7/P1/PfAOP/s (5′-ATATGATATATCTTATCGGTCCC-3′), probe/hDHFR/as (5′-CCTTTCTCCTCCTGGACATC-3′), probe/hDHFR/s (5′-CATGGTTCGCTAAACTGCATC-3′) and 3D7/ P4/PfAOP/as (5′-TGGTATTACAAATAAGGGAAGAC-3′). Western blots against PfAOP were performed using the purified peptide antibody αPfAOP-5 as described previously 41 with slight modifications. Parasites were isolated without magnetic cell separation by treatment with 0.05% (w/v) saponin 78,79 in ice-cold buffer containing All data are the mean ± standard deviation of at least three independent triplicate measurements. Wild type strain 3D7 served as a control. None of the differences between the IC 50 values was found to be significant (p > 0.05) based on statistical analyses in SigmaPlot 12.5 using the one way ANOVA method.
In vitro growth inhibition (IC 50 ) assays. Cell proliferation assays in the presence of ART, ATM, tBOOH and diamide were performed using the SYBR Green method 49 as described previously 12,51,54 . Synchronized ring stage parasites of overexpressing and knockout strains were incubated in 96-well plates with 1:3 serial drug dilutions. Drug concentrations ranged from 300 nM to 46 pM for ART and from 3.33 mM to 1 µM for tBOOH and diamide. Wells containing uninfected erythrocytes and infected erythrocytes without drug served as controls. Following incubation at 37 °C for 72 h, plates were frozen and stored at −80 °C. The plates were subsequently thawed at room temperature and the parasites were lysed with 2x lysis buffer containing 20 mM Tris-HCl, pH 7.5, 5 mM EDTA, 0.08% Triton X-100, 0.008% saponin and 0.12 µL/mL SYBR Green 1. The fluorescence was measured using a FLUOstar plate reader (BMG Labtech) with a gain set at 60 and an excitation and emission wavelength of 485 nm and 535 nm, respectively. Dose-response curves and IC 50 values were computed using the four parameter Hill function in SigmaPlot 12.5. As a control, the PfAOP content of overexpressing parasites was analyzed by standard western blotting and quantified using SuperSignal West Femto chemiluminiscent substrate (Thermo Scientific), a C-DiGit Blot Scanner and the software ImageStudioLite (LI-COR).
Ring-stage survival assays. The in vitro ring-stage survival assay was adapted from Witkowski et al. 56,57 .
Tightly sorbitol-synchronized early ring stage parasites (parasitemia 0.5-1%, hematocrit 2%, culture volume 2 mL) were treated in six well plates with 700 nM artemisinin for 6 h, washed five times with 5 mL RPMI, resuspended in complete medium and cultured for another 66 h. Blood smears were prepared, stained with Giemsa and approximately 10.000 erythrocytes were counted by light microscopy. Four independent measurements were conducted for each strain. Ring-stage survival percentages were calculated by comparing the number of viable parasites between drug-treated and mock-treated controls.

Figure 7.
Ring-stage survival assays. Highly synchronous ring stage parasites were treated with or without 700 nM artemisinin for 6 h, washed and further incubated for 66 h. The parasitemia was subsequently determined from Giesma-stained blood smears. The parasite survival percentage was calculated for wild type strain 3D7 (3D7 WT), a strain overexpressing GFP-tagged full length PfAOP from Fig. 3 (PfAOP), a knockout strain from Fig. 4 (Δaop#1) and a positive control that carries the mutation for K13 C580Y (NF54 K13C580Y ). All data are the mean ± standard deviation of four independent experiments. Statistical analyses were performed in SigmaPlot 12.5 using the one way ANOVA method. Results are summarized on top of the bar chart (n.s., not significant; **p < 0.01; ***p < 0.001).