Biosynthesis of heme O in intraerythrocytic stages of Plasmodium falciparum and potential inhibitors of this pathway

A number of antimalarial drugs interfere with the electron transport chain and heme-related reactions; however, the biosynthesis of heme derivatives in Plasmodium parasites has not been fully elucidated. Here, we characterized the steps that lead to the farnesylation of heme. After the identification of a gene encoding heme O synthase, we identified heme O synthesis in blood stage parasites through the incorporation of radioactive precursors. The presence of heme O synthesis in intraerythrocytic stages of Plasmodium falciparum was confirmed by mass spectrometry. Inabenfide and uniconazole–P appeared to interfere in heme synthesis, accordingly, parasite growth was also affected by the addition of these drugs. We conclude that heme O synthesis occurs in blood stage-P. falciparum and this pathway could be a potential target for antimalarial drugs.


Results and Discussion
Reports of looming resistance to artemisinin and its derivatives are a major concern for malarial treatment 18,19 , and the fight against artemisinin non-responsive parasites, the elucidation of exploitable drug targets and the development of new drugs are becoming urgent issues. Recently, a thorough analysis of druggable targets in Plasmodium was conducted and emerging resistance markers were characterized 20 .
We have been focusing on the biosynthesis of derivatives of the isoprenoid pathway in P. falciparum 21 . Despite the fact that the intraerythrocytic parasite lives in an environment that is particularly rich in heme, it has retained its own heme synthesis pathway. However, unlike heme B, the isoprenoid-linked heme O and heme A cannot be obtained from the host. Given that they are essential in other organisms, the characterization of heme farnesylation may reveal important targets against Plasmodium 22 .

Identification of a gene encoding heme O synthase. For the characterization of heme O biosynthesis,
we identified a gene encoding a protein with sequence homology to heme O synthase (HOS) in the genome of P. falciparum 23 , an enzyme required for farnesylation of heme.
In Saccharomyces cerevisiae, HOS, also referred to as COX10 24 , is a mitochondrial inner membrane protein that forms a multimeric complex of ~300 kDa. Subsequently, heme A synthase (HAS) oxidizes heme O to yield heme A, a cofactor of cytochrome c oxidase (COX) or complex IV of the mitochondrial respiratory chain. COX has several subunits, three of which are encoded in mitochondrial DNA; these are referred to as COX1, COX2 and COX3. The stability of the COX10 oligomer seems to depend on the presence of freshly synthesized COX1 and its intermediates 25 .
The sequence identified in the P. falciparum genome that encodes a putative COX10, PF3D7_0519300, shares more than 60% amino acid similarity to previously characterized enzymes from other organisms. Furthermore, the residues considered relevant for the catalytic activity of COX10 were conserved in the P. falciparum sequence ( Supplementary Information, Fig. S2); these are N196, R212, R216 and H317 following S. cerevisiae COX10 numbering 26,27 . The sequence was scanned for potential transmembrane regions, and five were identified in PF3D7_0519300, similar to other COX10 proteins ( Supplementary Information Fig. S2). A phylogenetic tree ( Supplementary Information Fig. S3) showing the evolutionary relationship among different COX10 sequences revealed a close relationship between the P. falciparum and S. cerevisiae enzymes. The enzyme COX10 from S. cerevisiae had been characterized 28 . These data suggest that PF3D7_0519300 in fact encodes the P. falciparum version of COX10.
In addition, through the phylogenetic tree of COX10 ( Supplementary Information Fig. S3), the similarity of Plasmodium spp. COX10 with the enzyme from other organisms of the apicomplexan phylum was compared. Within the genus of Plasmodium, P. falciparum COX10 is closest to P. reichenowi COX10, what is expected given the similarities in most genes between these species 29 . First, we focused on the characterization of heme O because not all organisms biosynthesize heme A 14 .
Subcellular location of COX10. Since the data suggest that PF3D7_0519300 encodes COX10 in P. falciparum, we considered it important to confirm its subcellular location. Using GFP-tagged COX10, we conducted fluorescence microscopy experiments using MitoRed and DAPI staining as controls ( Supplementary Information  Fig. S4).
Since the apicoplast and the mitochondrion are close, it is mandatory to use specific markers to clearly differentiate them. Therefore, to identify if the GFP fluorescence signal from the Cox10-GFP fusion originated from either the mitochondrion or the apicoplast, or both, we used a mitochondrial marker and an apicoplast-specific antibody produced according to Tonkin et al. 30 (produced by FastBio, Ribeirão Preto, Brazil). With this, we verified that COX10 signal (as green fluorescence) is mostly overlapping with the mitochondrial red signal and, to a smaller extent, in the nucleus. Besides that, DAPI also stains mitochondrial DNA, which may explain a partial overlap between the DAPI and the GFP signal. In contrast, the overlapping fluorescence observed within the apicoplast is not fully overlapping with the signal of Cox10-GFP, suggesting that PfCOX10 is probably not located in the apicoplast ( Supplementary Information Fig. S4). Based on the COX10 sequence obtained from PlasmoDB, there is also no predicted signal sequence that may direct the protein to the Apicoplast (see entry for PF3D7_0519300 in www.plasmoDB.org).
In most eukaryotes, including humans and S. cerevisiae, cytochrome c oxidase (COX) is the terminal component of the mitochondrial respiratory chain. Intriguingly, neither plasmoDB.org nor the conserved domain server at NCBI predicted a mitochondrial localization signal.
A nuclear gene encodes human and S. cerevisae COX10, which is not a structural subunit but is required for heme A synthesis 31 . The human or yeast COX10 enzyme is located in the mitochondrion and is necessary for the synthesis of COX 28 . The localization of the putative plasmodial COX10 in the mitochondrion suggests that the P. falciparum cox10 gene indeed encodes the plasmodial COX10 enzyme.
Biosynthesis of heme o. We first characterized heme O using metabolic labeling with [1-(n)-3 H]-FPP (direct precursor for the formation of heme O) or [U-14 C]-glycine (the initial precursor of the heme pathway). The detection of radiolabeled heme O and heme B from schizonts showed that there is an active synthesis of heme B and heme O ( Fig. 1) which is absent in non-parasitized erythrocytes. As heme B biosynthesis has already been described, we used these data as a positive control for the experiment 32,33 . The identification of standard of heme B is shown in Supplementary Information Fig. S5, and based on data published by Brown et al. 32,33 .
To confirm the presence of heme O in unlabeled parasites, two different analyses were performed using mass spectrometry (Figs. 2 and 3). In a first step, the parasite extract was loaded on Sep-Pak C18 columns and the peak corresponding to heme O was analyzed by LC-MS/MS and MALDI-TOF/TOF. For this purpose, a LC-MS/MS method was developed, as described in the methods section, with the instrument set to assess the product ions for the compound, including m/z 839, as shown in Fig. 2B. The presented spectrum was obtained from the peak at the retention time point 11.58 minutes ( Fig. 2A) and shows a product of m/z 839, and the two most important ion products are m/z 737 and m/z 671 (Fig. 2B), coincident with heme O.
Complementarily, a MALDI-TOF/TOF analysis was carried out, and the results are presented in Fig. 3. Since the fragmentation pattern is different in the used instruments as previously described for peptides 34 , two different ion products could be visualized for the same parent ion m/z 839 of heme O.
Accordingly, Fig. 3 shows the spectrum of heme O, represented by the parent mass m/z 839.405, followed by its fragments with masses m/z 721.060 and m/z 693.  Again, it is important to mention that two different ionization methods -electrospray (ESI) for LC-MS/MS and MALDI -were applied and both techniques provided results compatible with the molecular ion and fragment pattern of heme O.
As a control, the acquired spectra from the extracts of transformed bacteria expressing the Bacillus subtilis CtaB protein, which corresponds to heme O synthase, and the spectra from the parasite extracts were compared. In both, we observed the ion m/z 617 corresponding to heme B (Supplementary Information Figs. S5-S7). This underscores that mass-spectrometric analysis is a reliable methodology.
According to Ke et al. 35 , the heme biosynthetic pathway is not essential for asexual parasites stage. The reason why this pathway is still active in the parasite during the asexual stage is intriguing because free heme may be toxic to the parasite. However, the level of endogenously synthesized heme in parasites is much lower than that produced via hemoglobin digestion. It is possible that the parasite keeps this pathway active to supply heme for heme O biosynthesis, since farnesylated heme is essential for the biosynthesis of heme A (with heme O as intermediate), cytochrome c, or to supply heme for other heme-proteins. This occurs probably due to the changing availability of host-derived heme, which profoundly differs in dependence on the host at every stage. While during liver stage heme is probably scarce, an excess is present in erythrocytes. In a recent study on the role of dual heme sources in malaria parasite growth and development, the first enzyme, δ-aminolevulinate synthase (ALAS), and the last enzyme, ferrochelatase (FC), in the heme pathway of Plasmodium berghei (Pb) had been knocked out. The wild-type and knockout (KO) parasites had similar intraerythrocytic growth patterns in mice. Upon in vitro radiolabeling of heme in Pb-infected mouse reticulocytes and P. falciparum-infected human RBCs using [4-14 C]-aminolevulinic acid (ALA), parasites were found to incorporate in both hosts hemoglobin-heme and parasite-synthesized heme into hemozoin and mitochondrial cytochromes. The similar fates of the two heme sources suggest that parasites may obtain heme from different origins, emphasizing the biological importance of heme availability in the parasite 13 . Notably, during a genome wide piggyback-mediated insertion-mutation approach, the COX10 gene was deemed important for survival in blood stage parasites 36 .
Having shown the existence of heme O synthesis, we set out to determine whether the drugs INA and UNP specifically interfere with heme B and/or heme O synthesis. Both drugs perturb organelle structure and food vacuole morphology in schizonts 17,37 . In plants, INA and UNP inhibit the ent-kaurene oxidase (KO), a member of the Since a number of plant-exclusive isoprenic metabolites have also been found in P. falciparum, we investigated the possible route of gibberellin biosynthesis. However, we were unable to find hints of such pathway. The results obtained by metabolic labeling and recovery tests, where cultures were treated with INA or UNP and enriched with the theoretically inhibited end product (in this case, GA4) were negative (data not shown). Knowing that other antifungal agents such as azole act in the heme pathway 16 , we decided to investigate whether INA and UNP had their mechanism of action related to the heme biosynthesis or heme farnesylation pathway.
inA and Unp as potential inhibitors of heme o biosynthesis. First, we investigated whether INA and UNP inhibited heme B or heme O biosynthesis. We conducted an electrochemical oxidation/reduction test to determine if these drugs interfered in any way with heme. To test the differences in the redox behavior between healthy and Plasmodium-infected erythrocytes, films of both specimens were prepared on glass carbon electrodes and immersed in PBS at pH 7.2 (Fig. 4A). The voltammetry response of the uninfected erythrocyte sample was a pair of signals at −0.34 (cathodic, C 1 ) and −0.26 V (anodic, A 1 ) against Ag/AgCl, which can be attributed to the Fe(III)-heme/Fe(II)-heme one-electron interconversion. The symmetry of the peak profiles and their relatively low separation in the potential scale indicates that the electrochemical process possesses high reversibility. In contrast, the voltammograms of Plasmodium-infected erythrocytes showed remarkable peak splitting in both the cathodic and anodic regions, which was confirmed using square wave voltammetry as the detection mode (Fig. 4B). This voltammetry can be interpreted as the appearance of a second Fe(III)/Fe(II) couple (C 2 / A 2 ) corresponding to Fe-heme species different from those in healthy specimens, which formed as a result of the Plasmodium infection. As judged by the value of the midpeak potential of the C 2 /A 2 couple, this new species can be assigned to heme O 39 .
The effect of INA and UNP on Plasmodium-infected erythrocytes is illustrated in Fig. 5, in which square wave voltammograms of glassy carbon electrodes modified with (a) uninfected erythrocytes plus INA, (b) uninfected erythrocytes plus UNP, (c) Plasmodium-infected erythrocytes plus INA, and (d) Plasmodium-infected erythrocytes plus UNP are shown. The relevant point to emphasize in these voltammograms is that the signal (C 2 ) attributed to heme O in the samples of infected erythrocytes treated with INA and UNP is clearly lower than that in untreated Plasmodium-infected erythrocytes. These results suggest that INA and UNP interfere with the heme pathway, blocking the synthesis of heme O, and that this mechanism probably caused the observed decrease of parasitemia over 48 hours.
In view of the possible interference of UNP and INA with the heme group, as demonstrated by electrochemistry, we conducted a metabolic assay using schizonts labeled with [1-(n)- 3   www.nature.com/scientificreports www.nature.com/scientificreports/ that at the end of the treatment, a number of viable parasites remain, and partial inhibition of parasite growth still occurs. Then, each extract of the labeled schizonts was analyzed using C18 Vac columns. INA and UNP seemed to inhibit heme B synthesis (75% for INA and 54% for UNP, Fig. 6A). However, this effect was not statistically significant (p > 0.05) due to an uneven labelling of untreated parasites (see Supplementary Information Fig. S8). Since the parasites are able to acquire heme from the host, the effect of this treatment may not be a hindrance to the parasite 15,40 . Therefore, if the parasite's heme biosynthesis is inhibited, the parasite could still recover heme from the host. Nevertheless, the biosynthesis of heme O in schizonts decreased significantly (Figs. 6B and 7), and the inhibition was 76% for INA and 64% for UNP when labeling with [U-14 C]-glycine and 58% for INA and 52.5% for UNP when labeling with [1-(n)-3 H]-FPP. This indicates that the inhibition mediated by INA and UNP are possibly not directly related to the inhibition of heme B or heme O synthesis but rather bind to heme accumulated in the food vacuole as previously also shown by Toyama and colleagues 17 .    . If INA and UNP inhibited the formation of heme O, it is possible that they also interfered with the mitochondrial potential. Therefore, we performed an experiment with the mitochondrial potential marker JC-1.
Parasites treated with INA and UNP for 48 hours apparently present a dose-dependent reduction in mitochondrial JC-1 incorporation. Therefore, the decrease in heme O may also be triggered by drug interference of mitochondrial function or by parasite death ( Supplementary Information Fig. S9), suggesting that in this assay occurred non-specific parasite death rather than selective inhibition of heme O synthesis. The control culture treated with 15 nM chloroquine also showed a decrease of the mitochondrial function, in consequence of the parasite's incapacity to polymerize heme and the concomitant production of reactive oxygen species 41 .
After the first 24 hours, UNP treatment caused a (dose dependent) decrease in mitochondrial function in relation to the control. Since UNP treatment is also related to the decrease of heme O, it may be hypothesized that the drug is acting in the mitochondrial potential. Notably, in the first 24 hours, no decrease in parasitemia was observed.
For INA treatment, there was also no appreciable effect after the first 24 hours. We cannot exclude that INA exerts an effect on the mitochondrial potential. Possibly, the dynamics of INA inhibition become visible only after more than 24 hours of treatment (Fig. 8).
Therefore, the probable inhibition of heme O by INA and UNP may in consequence affect the transport of electrons in the mitochondrial chain. As shown by Painter and colleagues, inhibition of mitochondria prevents the regeneration of ubiquinone in P. falciparum and this impaired regeneration of ubiquinone also necessary as the electron acceptor for dihydroorotate dehydrogenase, an essential enzyme for pyrimidine biosynthesis 42 .
In order to show that INA and UNP treatment also leads to decreased ubiquinone levels, a rescue experiment was performed where parasites were simultaneously treated with either INA or UNP and increasing concentrations of external decyl-ubiquinone. If INA and UNP at least partly inhibit parasite growth by decreasing internal ubiquinones, then addition of decyl ubiquinone should restore growth. This was in fact observed when 5 or 10 µM decyl-ubiquinone were added (Supplemental Fig. S10). www.nature.com/scientificreports www.nature.com/scientificreports/ Heme conversion catalyzed by COX10 appears to be a limiting step in the rate of heme A formation 25 . Thus, the regulation of the abundance and activity of COX10 is likely to modulate the heme A biosynthesis pathway. It was established that the cofactor CoA2 acts together with COX10 in the hemylation of COX1. COX10 assembles into active oligomers in normal cells 26 , and the oligomerization of COX10 appears to be a key feature of the enzymatic activity of HOS. Thus, if heme O biosynthesis is at least partially inhibited by the drugs, the lower heme O levels are directly related to the low levels of heme A. This, in turn, may affect the mitochondrial potential, because the components of the complex IV are not biosynthesized and the parasite cannot obtain neither heme O nor heme A from the host.

conclusion
In this study, we demonstrated that P. falciparum has an active biosynthetic pathway for the production of heme O, the intermediate for heme A (unique cofactor of all known eukaryotic cytochrome c oxidases). Furthermore, we provided hints that INA and UNP are possibly responsible for the inhibition of heme B farnesylation besides other pleiotropic effects. In consequence, with less heme O, a decrease in parasitemia is observed. We propose that this pathway could be a valid target for novel drugs. The regulatory mechanisms that orchestrate heme metabolism of parasites remain poorly understood and are a frontier for future discovery. Subsequently, the parasite cultures were centrifuged at 800 × g for 10 min at 22 °C. The intraerythrocytic stages were purified using magnetic columns 46 and submitted to different extraction methods of gibberellin, heme B or heme O, which are further described below. As a control, the same procedure was carried out with uninfected erythrocytes. The herein used erythrocytes were older than 15 days. Therefore, no reticulocytes which may contain trace amounts of heme synthesis could be detected. extraction of gibberellin. For the extraction of gibberellin, schizont stages labeled with [1-(n)-3 H]-GGPP as described above were purified using magnetic columns and homogenized with 1.0 M Tris-HCl, pH 7.2, containing 2% Triton X-100 and incubated for one hour at 4 °C. After centrifugation (25 min at 8.000 × g and 4 °C), the supernatant was adjusted to pH 2.5. The separation procedure between the remaining aqueous extract and the diethyl-ether phase was conducted using the method described by Barthe and Bulard,47 , and this fraction was called PfexG.
Rp-HpLc. The separation of gibberellin by gradient RP-HPLC was adapted from the isocratic system described by Bhalla and Singh 48 , where the solvent was initially 90% acetonitrile (ACN) and 10% H 2 O at a flow of 0.5 mL/min for 45 min at 30 °C over a YMC C18 ODS-A column (5 µm, 300 Å, 3.0 mm × 150 mm). The wavelength used for analysis was 206 nm, and commercial gibberellin 3 (GA3) and 4 (GA4) (Sigma-Aldrich, Darmstadt, Germany) was used as a standard.
PfexG was co-injected with standard GA4, and the components were chromatographically separated by RP-HPLC. The samples were dried at 50 °C, resuspended in 0.5 mL of scintillation liquid (BetaplateScint, Perkin Elmer, Groningen, Netherlands) and analyzed in a Beckman ® scintillator unit.
Scientific RepoRtS | (2019) 9:19261 | https://doi.org/10.1038/s41598-019-55506-y www.nature.com/scientificreports www.nature.com/scientificreports/ Immunofluorescence and localization of GFP-tagged COX10. For immunofluorescence analysis, parasites were fixed with 4% EM grade paraformaldehyde and 0.0075% EM grade glutaraldehyde in PBS for 30 min as described by Tonkin et al. 30 . Parasites were then washed with PBS and blocked with 3% BSA for 1 h. Afterwards, cells were incubated with the anti-apicoplast primary antibody (dilution 1:1000, formulated according to Tonkin et al. 30 , and developed by FastaBio Ltda, Ribeirão Preto, Brazil) diluted in 0.1% BSA, 0.001% saponin in PBS for 1 hour at 24 °C. Cells were then washed three time with PBS and incubated with the secondary antibody Alexa Fluor ® 488 and DAPI 2 μg/ml also diluted in 0.1% BSA and 0.001% saponin in PBS for 45 min at 24 °C. After three washes with PBS the slides were mounted and sealed. The images were acquired in a fluorescence microscope Axio observer Z1 and processed using the Photoshop version 5.
For mitochondria localization parasites were incubated with Mitotracker, (Molecular Probes, Oregon, USA -dilution 1:10,000) for 30 min at 37 °C. Afterwards, parasites were incubated with DAPI final concentration of 2 µg/mL at 37 °C for 1 h. Finally, parasites were washed 3 times with PBS/saponin. Images were also acquired in a fluorescence microscope Axio observer Z1 and processed using the Photoshop version 5.
Vector construction. To obtain the transfection vector containing the desired cox10 gene of P. falciparum, the cox10 ORF was PCR-amplified (primers 5′-agatctATGGGATTTAATAAGATTTTTCC and 5′-ctgcagcTTTAAATGTTCTTTTTATCAAATGTAGG, program 94 °C, 40 s, 54 °C, 40 s, 65 °C 1 min 30 s, 30 cycles) on genomic DNA from PF3d7, using Elongase enzyme mix (Thermo Scientific/Invitrogen, Carlsbad, CA, USA) and cloned into pGEM ® T-easy (Promega, Wisconsin, USA). The fragment was sequenced and a fragment containing the correct sequence was excised using via the introduced BglII and PstI sites and ligated in the pRESA-GFP/HA vector 52 and cut with the same enzymes.
The construct was transfected into P. falciparum 3D7 as described 53 . Integrated parasites were checked for green fluorescence. Given that the 5′ end or the promoter region of the cox10 ORF was not modified, the fused protein COX10-GFP-HA is supposed to be expressed at levels similar to the unmodified locus 54 . The vector construction is shown in Supplementary Information Fig. S1. Note that in this construct, only parasites with recombinant loci show green fluorescence, since there is no promoter included in the 5′ position of the COX10-GFP-HA fusion gene. Then, a mixture of acetone/HCl (95:5) was added in a proportion of 40% of the lysate/60% acetone/HCl, and the solution was kept on ice for 90 min. Then, the parasite or bacterial lysate was centrifuged at 8,000 × g for 10 min at 4 °C to extract heme B and heme O. Uninfected erythrocytes labeled with the same radioactive tracers and processed under equal conditions were used as a control. In total, 1 × 10 9 schizont stage parasites and the same quantity of uninfected erythrocytes were analyzed with mass spectrometry in the same way.

Separation of heme B and heme o.
The separation of both types of heme was performed using C18 Vac columns. The column was pre-washed with 2 mL of 25% ACN. The supernatant from the heme extraction was collected and passed through this column. Then, the column was washed with 5 mL of 25% ACN. Then, 80% ACN was used to elute heme B, and DMSO was used to elute heme O 33 . For bacteria, the volume of solvents used was three times higher.
Mass spectrometry. LC-MS/MS. Heme samples were analyzed using a LC-MS/MS system of an ACCELA 600 quaternary pump LC connected to a triple quadrupole mass spectrometer, the TSQ Quantum Max (Thermo Scientific, Bremen, Germany) via an ESI ionization source.
The mobile phase used to separate the compounds contained water with 0.02% of acetic acid (A) and ACN with 0.02% of acetic acid (B). The composition of solvent B varied as follows: 0-2 min, held at 10%; 2-20 min, from 10 to 90% linear gradient; and 20-30 min, held at 10%. The flow rate was 0. www.nature.com/scientificreports www.nature.com/scientificreports/ reflector 2 9.35 kV; and mass range, 500 to 2500 Da). For each spectrum, 5000 shots in 500-shot steps were summed from different positions of the target, collected and analyzed. All spectra were calibrated using a Peptide electrochemical study. Electrochemical measurements were performed at 298 ± 1 K in a thermostatic cell with CH 660I equipment. A BAS MF2012 glassy carbon working electrode (GCE) (geometrical area 0.071 cm 2 ), a platinum wire auxiliary electrode and an Ag/AgCl (3 M NaCl) reference electrode were used in a conventional three-electrode arrangement. Cyclic and square wave voltammetry (CV and SWV, respectively) were used as detection modes. Eventually, semi-derivative convolution of the data was performed in order to increase peak resolution.
Thin films of lyophilized erythrocytes and parasitized erythrocytes treated with 2.0 µM INA (Sigma-Aldrich, St. Louis, MO) or 20 µM UNP (Sigma-Aldrich, St. Louis, MO) or untreated were prepared on glassy carbon electrodes following a previously reported procedure 55 . Essentially, 20 µL of the dispersed erythrocyte material (1 mg/mL) in ethanol were applied, and the solvent was allowed to evaporate. As a result, a uniform, fine coating adhered to the basal electrode. Aqueous 0.10 M potassium phosphate-buffered saline (PBS) at a physiological pH of 7.2 that was previously degassed by bubbling with argon for 10 min was used as a supporting electrolyte.
inhibition tests with inA and Unp. Toyama et al. 17 determined the IC 50 values for INA and UNP in P.
falciparum. Three concentrations of each plus a lower and a higher concentration as well as a concentration at the IC 50 value, were evaluated and confirmed in our laboratory, and two controls, one without the drug and another with the vehicle, were used.
Another test was conducted in order to evaluate whether the inhibition of parasite growth caused by INA and UNP was reversible. We also assessed whether the parasite recovered by acquiring exogenous gibberellin contained in the medium upon INA or UNP treatment. For this, different concentrations (10 µM-150 µM) of gibberellin (G4) were added to cultures. These experiments were carried out in triplicate in independent biological triplicates. After the confirmation of the IC 50  Statistical analysis. For variables with a normal distribution, Student's t-test was applied to compare means between untreated infected erythrocytes and INA or UNP treated infected erythrocytes. Statistical analyses were completed using the GraphPad PRISM ® 5.3 software.