Background
The ability of the malaria parasite
Plasmodium falciparum to develop resistance to drugs motivates the critical need to deliver new development candidates to bolster the current clinical pipeline[
1]. Large-scale screens of synthetic chemical libraries have been especially successful in identifying numerous early hit compounds against cultured parasites[
2]. Compared to these phenotypic screening programmes, target-based discovery approaches have been less successful, perhaps partially due to the lack of extensively validated drug targets. Further characterization of priority hit compounds may include an investigation of their modes of action, which could also yield new potential targets for malaria drug discovery. The mode of action of a compound may be assessed by determining the effect of the compound on specific biochemical or cell biological pathways, or by a more global approach, e.g. transcriptomic, proteomic and/or metabolomic profiling[
3]. These studies require knowledge of the rate of action of a compound and should ideally be performed at early time-points when the compound starts exerting its primary effect(s), as opposed to later time-points when the primary mode of action may conceivably be obscured by non-specific secondary responses in the parasite. Moreover, a highly desirable property of anti-malarial compounds is that they should kill parasites rapidly, in order to reduce the required dosages in clinical use, minimize the likelihood of resistance development, and increase patient compliance. This requires an accurate determination of the rate of action of promising compounds against malaria parasites, particularly to enable researchers to rank compounds for further pre-clinical and clinical development.
A parasite reduction ratio (PRR) assay was recently described to predict the rate of parasite clearance
in vivo by measuring the extent to which parasites recover from drug exposure for defined periods of time in
in vitro cultures[
4]. However, for mode of action studies, the rate of action is conventionally assessed by evaluating parasite morphological changes over time during drug exposure using Giemsa-stained blood smears and light microscopy. While this technique is relatively simple to perform, it is time-consuming, highly susceptible to subjective interpretation and difficult to express quantitatively, unless the morphological changes are drastic and uniform. Making the distinction between parasites with normal
vs aberrant drug-induced morphologies is particularly challenging due to the heterogeneous morphology of individual parasites under routine culture conditions, the tendency of individual cells to display a spectrum of mild to severe morphological abnormalities, particularly at early time-points, and the challenge of preparing uniform microscopy preparations on separate occasions.
In this study, ATP quantitation and luciferase activity measurements as a means to detect the rate and severity of drug-induced stress in
in vitro cultures of
Plasmodium falciparum was explored. ATP content in cells as an indicator of metabolic status could conceivably be used to detect abnormal metabolic activity imposed by drug action, while luciferase activity in transgenic parasites was unexpectedly found to decrease rapidly and profoundly during drug exposure, which may be exploited as a novel indicator of the rate of drug-induced stress. The rate, magnitude and nature of the changes in parasite ATP content and luciferase activity levels over 10-hour incubation periods were characterized using a panel of six compounds with different modes of action: the clinical anti-malarial drugs chloroquine, mefloquine and artemisinin, and the experimental compounds DL-α-difluoromethyl ornithine (DFMO), an inhibitor of polyamine biosynthesis[
5], ritonavir, an HIV aspartyl protease inhibitor with known anti-malarial activity[
6] and gramicidin, a mixture of channel forming ionophores for monovalent cations[
7].
Methods
Parasite cultivation, morphological evaluation and drug IC50 determination
Plasmodium falciparum 3D7 cultures were maintained at 37°C in medium consisting of RPMI 1640 supplemented with 2mM L-glutamine, 25mM Hepes, 20mM glucose, 0.65 mM hypoxanthine, 60 μg/mL gentamycin, 2.5% (w/v) Albumax II and 3% (v/v) type O
+ red blood cells in flasks suffused with a mixture of 5% CO
2, 5% O
2, 90% N
2. Parasite life-cycles were routinely synchronized by the sorbitol method[
8]. Parasite morphology was assessed by light microscopy of methanol fixed and Giemsa-stained thin blood smears using a 100x oil-immersion objective. Images were captured with an Olympus BX41 upright microscope equipped with a CC12 Soft Imaging system. Drug 50% inhibitory concentrations (IC
50) were determined by measuring parasite viability after a 48h incubation with 3-fold serial dilutions of the drugs using the parasite lactate dehydrogenase (pLDH) assay[
9]. IC
50 values were derived from non-linear regression dose–response plots prepared with GraphPad Prism (v.5.02).
Drug compounds used in this study
Chloroquine diphosphate, mefloquine hydrochloride, artemisinin and gramicidin from Bacillus brevis were purchased from Sigma-Aldrich. DL-α-difluoromethylornithine (DFMO) was kindly provided by P. Woster (Wayne State University, MI) and ritonavir purchased from Kinbester Co. (Xiamen, China). The compounds were prepared as 10mM stock solutions in DMSO (artemisinin, gramicidin, ritonavir), methanol (mefloquine) or water (chloroquine, DFMO). The proteasome inhibitors lactacystin and MG-132 were purchased from Merck and prepared as 10mM stocks in water and DMSO, respectively. Final concentrations of the drug compounds used in all assays in this study were: 100 nM chloroquine; 100 nM mefloquine; 100 nM artemisinin (ATP, morphology and recovery assays); 500 nM artemisinin (luciferase assay); 0.1 nM gramicidin; 10 mM DFMO; 100 μM ritonavir; 5 μM lactacystin; 300 nM MG-132.
Parasite recovery assay
An early trophozoite-stage culture was used to prepare a 2% haematocrit, 2% parasitaemia suspension in culture medium which was distributed into 6-well plates at 2.5 mL per well. The plates were treated with drug and solvent control solutions, each in triplicate. An additional plate was prepared with uninfected RBCs in triplicate wells to serve as a background control. The plates were transferred to an airtight chamber suffused with 5% CO
2, 5% O
2, 90% N
2 and incubated at 37°C for six hours. Following incubation, the contents of each well was mixed well and 800 μL was transferred to a sterile microfuge tube to pellet the infected red blood cells. The pellet was washed three times in 1 mL of culture medium pre-warmed at 37°C and then resuspended in fresh medium at a haematocrit of 1%. The suspension was transferred to a 96-well plates at 200 μL per well. Parasite lactate dehydrogenase activity[
9] was measured in the plate after it was returned to the airtight chamber, gassed and incubated at 37°C for 48 hours. The pLDH activity was used to calculate the percentage parasite viability after 48h for each respective drug, relative to solvent controls.
ATP assay
To quantitate changes in parasite ATP content during drug exposure, early trophozoite-stage cultures were used to prepare a 32 mL 5% haematocrit, 2% parasitaemia suspension in culture medium. The suspension was split into 2 x 15 mL cultures and incubated in culture flasks at 37°C with drug compounds and solvent control solutions, respectively. Three 500 μL aliquots of the culture suspensions were removed from the control and treated cultures, respectively, every 2 hours over a 10-hour period and transferred to cold microfuge tubes placed on ice. The infected red blood cells in the 500 μL samples were pelleted by centrifugation at 8,000 rpm for 30 seconds in a microfuge and lysed by adding 500 μL 0.24% (w/v) saponin, 0.1% (w/v) bovine serum albumin (BSA) in phosphate-buffered saline (PBS). Complete RBC lysis was achieved by vortexing the solution for approximately 45 seconds until translucent. The lysate was removed and pipetted onto the surface of a 300 μL mixture of dibutyl phthalate and dioctyl phthalate (5:4) in a microfuge tube and centrifuged at 14,000 rpm for 40 seconds. This resulted in the intact parasites pelleting at the bottom of the tube, while the aqueous RBC lysate settled above the intervening phthalate oil layer. The aqueous layer was aspirated off and the inner surface of the tube above the oil layer and the top of the oil layer carefully washed with 500 μL of 0.1% (w/v) BSA in PBS to completely remove all traces of RBC lysate. The oil layer was aspirated off, 150 μL ice-cold PBS added onto the parasite pellet and the sample snap-frozen in liquid nitrogen and stored at −20°C. ATP levels were subsequently measured by thawing the samples at room temperature, resuspending the parasite pellets by pipetting, transferring 50 μL to a white 96-well plate and adding 50 μL of CellTitre-Glo® reagent (Promega). The plate was briefly agitated and then incubated in the dark for 10 minutes at room temperature before measuring luminescence in Tecan Infinite F500 plate reader. Average background luminescence readings from wells containing PBS alone were subtracted from the sample readings.
Preparation of luciferase transgenic parasites
To obtain an expression plasmid for stable episomal expression of luciferase, the expression and selection cassettes of pHTK were sub-cloned into the
NotI and
NcoI sites of pGEM-T Easy (Promega). The pHTK expression cassette consists of the
P. falciparum heat-shock protein
hsp86 5’-untranslated promoter region, the
Herpes simplex virus thymidine kinase coding sequence flanked by
XhoI sites and the
P. berghei 3’ termination region, while the selection cassette contains the human dihydrofolate reductase (
hdhfr) coding sequence flanked by the
P. falciparum calmodulin 5’-untranslated promoter region and the
P. falciparum histidine-rich protein 2 3’-untranslated region in a head-to-head orientation with the expression cassette[
10]. The coding sequence of the
Photinus pyralis luciferase gene, flanked by
NheI and
XhoI restriction sites, was PCR amplified from the pGL2 plasmid (Promega) and replaced the thymidine kinase sequence in the pHTK expression cassette to obtain pHsp-Luc. The plasmid was used to transfect
P. falciparum 3D7 parasites by electroporation (0.31 kV, 950μF)in a BioRad Gene-Pulser electroporator and stable lines selected by culturing in medium supplemented with 2.5 nM WR99210 according to previously described protocols[
11].
Luciferase assay
Transgenic luciferase-expressing parasite cultures at the early trophozoite stage were used to prepare 5% haematocrit, 2% parasitaemia suspensions in culture medium and 200 μL transferred to wells in a 96-well culture plate. A separate plate was prepared for each 2-hour time-point of the assay. Test drug compounds and solvent control solutions were added to triplicate wells in the plate, while uninfected RBCs at 5% haematocrit (200 μL) were added to triplicate wells as background controls. The plates were transferred to an airtight chamber suffused with 5% CO2, 5% O2, balance N2 and incubated at 37°C. At 2-hour intervals, one plate was carefully removed from the chamber without disturbing the settled RBCs and 150 μL of supernatant was removed from all wells, followed by the addition of 100 μL per well of Glo® Lysis Buffer (Promega). After 5 minutes incubation at room temperature, the contents of the wells were mixed by pipetting, 100 μL transferred to a white 96-well plate 100 μL of Bright-Glo® Luciferase Assay System reagent (Promega) added and the luminescence was measured immediately in a Tecan Infinite F500 multimode plate reader. The average luminescence readings obtained from uninfected red blood cell wells were subtracted as background from those obtained with drug-treated and untreated infected red blood cell cultures.
Discussion
In this study, two assay formats as a means for determining the timing and severity of drug-induced stress in Plasmodium falciparum parasite cultures over short incubation periods were explored. A panel of six anti-malarial drug compounds with different modes and, presumably, rates of action were used to interrogate and characterize the utility of the ATP and luciferase assays described above. The observed responses of trophozoite-stage parasite ATP content to drug exposure over a 10 hour incubation period can be broadly classified into three phenotypes: little or no change relative to untreated controls (chloroquine and DFMO); marked and sustained increase in ATP levels (mefloquine and artemisinin); rapid depletion of ATP content (ritonavir and gramicidin). These classifications may conceivably be interpreted as signifying, respectively: absence of significant drug-induced stress; altered (increased) metabolic activity to counteract drug-induced stress; severely compromised metabolic status due to drug action. This interpretation correlates with the results obtained with subsidiary assays. Morphological evaluation of the drug-treated parasites revealed mild abnormalities, mostly limited to growth retardation, in the ATP non-responders, and (with the exception of chloroquine) competence to recover from a 6h drug exposure. By contrast, increased ATP levels correlated with earlier appearance of growth-inhibited parasites and additional aberrant morphologies and a 44%-54% reduction in recovery following 6h drug exposure, while rapid ATP depletion was accompanied by the early appearance and preponderance of pyknotic (presumably irretrievably compromised) parasite forms and a significantly greater inhibition (79%-89%) of parasite recovery following 6h drug exposure. Interestingly, the timing and magnitude of the ATP changes, whether increased or decreased, also correlated with the timing of appearance of abnormal morphologies and ability to recover from brief drug exposure, although arguably more compounds will need to be assessed to determine the validity of this trend. Based on these results, the rate and severity of drug stress can be tentatively ranked as ritonavir>gramicidin>artemisinin>mefloquine>>chloroquine, DFMO.
The lack of a notable ATP response with DFMO agrees with its accepted mode of action, which is cytostatic rather than cytocidal. It inhibits ornithine decarboxylase, a key enzyme in polyamine biosynthesis, which slows down growth and eventually blocks parasite cell cycle progression in the late trophozoite stage in a manner which is reversible by exogenous addition of polyamines to the medium[
5]. Thus, a 10h incubation with DFMO starting in the early trophozoite stage may generate insufficient stress to result in a notable disruption of ATP homeostasis. The consensus view is that chloroquine becomes ionized and trapped in the low pH environment of the parasite food vacuole, where it disrupts haemozoin formation and causes an accumulation of toxic free haem and chloroquine-haem complexes[
17]. The results of this study suggest that a 10h incubation with chloroquine from the early trophozoite stage does not generate sufficient haem complexes to exert a significant effect on parasite ATP levels and/or haem-induced toxicity is slow-acting. Interestingly, the failure of parasites to recover from the 6h chloroquine incubation in the recovery assay may provide additional evidence for the irreversible entrapment of chloroquine in the food vacuole, where it probably continues to cause haem accumulation and toxicity despite the washing away of exogeneous chloroquine in the medium. Thus, the trapping phenomenon likely contributes greatly to the effectiveness of chloroquine as an anti-malarial drug, despite the apparently slow rate at which it induces metabolic stress compared to the other compounds. An alternative explanation is that haem accumulating in the parasite during the 6h chloroquine treatment remains parasite associated after chloroquine washout and continues to exert toxicity in the following 48h incubation.
Despite decades of clinical use, the mode of action of mefloquine is still uncertain. A general assumption is that it shares chloroquines effect on haemozoin formation and causes a toxic accumulation of haem or haem complexes[
18]. However, cell biological evidence e.g. the differential effects of the two drugs on haemoglobin endocytosis, trafficking and digestion suggests otherwise[
19,
20]. In the present study, the rapid and marked effect of mefloquine on ATP levels and luciferase activity is not mirrored by chloroquine and thus strongly implies a different primary mode of action for mefloquine. As for mefloquine, the mode of action of artemisinin is uncertain and has been suggested to involve
inter alia inhibition of the parasite equivalent of the sarcoplasmic reticulum Ca
2+-ATPase (PfATP6), depolarization of mitochondrial membranes, haem adduct formation, generation of reactive oxygen species in conjunction with haem and protein alkylation[
21]. Nonetheless, the appeal of artemisinin as an anti-malarial is partly based on its perceived rapid action against parasites[
4], which is supported by the observations from this study that ATP and luciferase activity levels in treated parasites are markedly affected by the drug at the first 2h time-point of exposure. Uncertainty regarding the modes of action of mefloquine and artemisinin make it difficult to fully explain the cause for increased ATP levels in treated parasites, other than to make a general assumption that it reflects increased metabolic activity by the parasite as part of a cellular stress response to overcome detrimental drug effects. This likely necessitates increased production of ATP to fuel synthesis and activities of enzymes, substrates and co-factors involved in e.g. antioxidant defence and protein chaperone systems.
The very rapid and profound depletion of ATP in ritonavir-treated parasites was supported by the early preponderance of pyknotic parasite morphologies and highly compromised ability to recover from a 6h drug exposure. This was surprising, given that ritonavir is an HIV protease inhibitor and was proposed to act against parasites by inhibiting aspartyl proteases responsible for haemoglobin digestion[
22]. Arguably, inhibition of this process would result in a more protracted growth inhibition of parasites due to amino acid starvation, not the rapid and lethal effect observed here. This argues for a different mode of action of ritonavir, which was also proposed in a study reporting the anti-malarial interactions of HIV protease inhibitors with hemoglobin protease inhibitors, mefloquine and chloroquine[
23]. The rapid depletion of parasite ATP by gramicidin, however, is consistent with its probable mode of action. Gramicidins are lipophilic, linear peptides that form channels in membranes that are permeable to monovalent cations[
24]. The rapid disruption of cellular sodium, potassium and proton gradients through these channels should have immediate pleiotropic consequences for parasite metabolism, which may also be reflected by the extreme potency of gramicidin against parasites (IC
50 < 0.1nM – Additional file
2).
The results obtained with the ATP assay suggest that it could represent a sensitive, quantitative means for detecting the earliest time-points of drug-induced stress to inform and complement drug mode of action studies. However, the question remains whether it could also be a useful tool for unambiguously determining the rate and extent to which parasite viability is irrevocably compromised by a particular drug. In principle, a total depletion of ATP could have been regarded as a signpost for irreversible parasite lethality. However, this is not entirely the case, as evidenced by the ability of ritonavir and gramicidin-treated parasites to recover from a 6h treatment, albeit severely limited, despite an apparent complete reduction in ATP in 2–4 hours. Conversely, artemisinin and mefloquine treated parasites actually display increased ATP levels at 6h, despite the fact that their recovery from a 6h treatment is inhibited by approximately 50%. The fact that treatment with the panel of six drugs produces three distinct phenotypes of ATP responses (increased, decreased and unchanged ATP levels) may further complicate a detailed interpretation of ATP responses to experimental drug pressure, necessitating an exploration of ATP responses with a larger drug panel before considering scale-up of the assay. In addition, 5xIC
50 drug concentrations were used in this study as a compromise to obtain sufficient drug pressure on the parasite without excessive off-target effects unrelated to the primary drug mode of action which may be prevalent at higher concentrations. However, it should be cautioned that 5xIC50 does not necessarily equate to lethal dose to the same extent for all compounds. Therefore, rate of killing studies may best be performed at lethal dose concentrations. Characterization of the assay, perhaps in conjunction with the recently described parasite recovery rate (PRR) assay[
4], would be required to more firmly define the correlates of ATP levels and irreversible parasite lethality. Nonetheless, the proposed ranking of the test drugs based on ATP responses, as discussed above, suggests that the assay in its current form may be used to assess the rate of parasite viability inhibition of experimental compounds relative to each other and standard “benchmark” drugs.
Serendipitously, it was found that luciferase activity in transgenic parasites responds rapidly and markedly to drug exposure. A highly attractive advantage of the luciferase assay is that it is entirely multiwell plate-based, requires minimal liquid handling steps and provides an extremely sensitive and robust read-out, thus making it potentially amenable to high-throughput formats. The overall trend in the results was similar to that obtained with the ATP assay. DFMO and chloroquine produced a slow, mild decrease in luciferase activity, while artemisinin, mefloquine and ritonavir profoundly compromised luciferase activity within 2 hours. The rapid decreases in luciferase activity during drug exposure could be construed as a cellular stress response in which proteolysis and amino acid release and/or selective translational inhibition is used to alter the proteome of the parasite[
25,
26]. Luciferase is particularly known to be susceptible to proteolytic degradation[
27]. Nonetheless, the activity decrease was not affected by proteasome inhibitors, even though proteasomes are principally responsible for cytoplasmic protein turnover and homeostasis in mammalian cells[
26]. Interestingly, the proteasome inhibitors on their own also produced marked loss of luciferase activity in 6 hours, suggestive of parasite stress experienced by the inhibition of their protein turnover ability. The rapid loss of luciferase activity is also not shared by all parasite cytoplasmic proteins. In contrast to luciferase, parasite pLDH activity showed only mild changes after 6 hours incubation with all the drugs, which hints that luciferase is particularly sensitive to cellular stress conditions. Caution should be exercised in interpreting the luciferase results, however. The luciferase assay requires the overexpression of a foreign reporter protein in the parasite. Conceivably, this could cause subtle alterations that may obscure drug-specific effects in subsequent drug mode of action studies (e.g. by transcriptomic, proteomic or metabolomic analyses). In addition, transgene overexpression might alter parasite sensitivity and the rate of drug-induced stress by certain compounds. For example, the detailed ranking of compound rate of action in the luciferase assay would differ from that obtained by the ATP assay. Artemisinin produced more profound and rapid changes in luciferase activity than the other drugs, while gramicidin had a milder effect than expected from the earlier assays (ATP, morphology and parasite recovery). While mindful of these caveats, the ease of the assay still suggests that it could be used as a preliminary screen for rapid vs. slow acting compounds over a 4–6 hour incubation period, especially when a large number of compounds need to be assessed.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
HCH and CJP conceived and designed the study, TK and ACvB performed the experimental procedures and data analysis, HCH and CJP were principally responsible for interpretation of the data. HCH and TK drafted the manuscript, which was read, amended and approved by all authors. All authors read and approved the final manuscript.