Sustainable malaria control dictates the integration of therapeutic strategies targeting the pathogenic asexual forms as well as the transmissible sexual gametocyte forms of malaria parasites. Drug discovery is a long, arduous and expensive process, and to avoid duplication in screening libraries for gametocytocidal compounds, especially in the face of limited resources, it would be useful to compare assay results generated in different laboratories. Currently there are often discrepancies in the data, which hampers progress in this field. Two major challenges in this regard are the lack of standardization in: 1) culturing reproducible batches of pure, stage-specific gametocytes in high yield and 2) assays for the metabolically hypo-active gametocytes.
Gametocytocidal screening assays
Endeavours to find transmission blocking anti-malarials are currently hindered by a lack of understanding of the basic biological processes governing gametocytogenesis in the human host, which makes an informed choice of an assay system problematic. In humans, gametocytes are suggested to be terminally differentiated and metabolically less active [
32]. It is proposed that there is no replication of the gametocyte genome during development and that gametocytes are arrested in phase G
0 of the cell cycle [
56]. Nucleic acid synthetic activity is likely restricted to RNA synthesis and genetic evidence shows that gametocytes are haploid. Synthesis of RNA is reported to stop after day 6 of development, and there is no haemoglobin digestion and protein synthesis in mature gametocytes [
57,
58]. Any assay for gametocytocidal activity therefore currently relies on measuring changes in metabolic status or stress responses in gametocytes, performed against the background of a metabolically ‘inactive’ cell.
Several groups have published platforms allowing the evaluation of gametocytocidal activity of a compound series. A Bill and Melinda Gates Foundation Initiative recently attempted to compare gametocyte assay platforms from various laboratories using different assay readouts (reported elsewhere). However, direct comparison of the data from these studies is fraught with difficultly due to differences in parameters such as: 1) parasite strains used; 2) different gametocyte induction protocols; 3) composition of culture medium used; 4) gametocyte isolation protocols; 5) stage of development of gametocytes; 6) assay platforms; 7) presence or absence of erythrocytes; 8) number of gametocytes per assay well; 9) panel of compounds; 10) concentration of compounds; 11) drug exposure times; 12) presentation of data, e.g. % inhibition at single concentrations or only IC50 values, etc.
The production of viable and functional gametocytes using a standardized, robust gametocyte production protocol was used here to overcome the limitations listed above. These gametocytes were subsequently used, for the first time, in a parallel, comparative interrogation of four different gametocytocidal assays (ATP, pLDH, PrestoBlue® and the luciferase reporter) for their ability to detect gametocytocidal activity of a single set of 10 compounds from MMV. The compound set was provided blinded and only unblinded after completion of all the assays and data analysis. The parallel nature of the assays performed on the same gametocyte population uniquely allowed direct comparisons of the assay platforms in this study.
Comparison of assays that evaluate different biological functions
The tricarboxylic acid (TCA) cycle is highly active in gametocytes, especially in females that are characterized by a rapidly expanding mitochondrion in preparation for gametogenesis [
59,
60]. A recent study found 15 of 16 mRNA transcripts of mitochondrial TCA cycle enzymes upregulated in gametocytes [
50]. Increased TCA cycle activity implies increased levels of cytoplasmic ATP. In the luminescent ATP assay, ATP content is used to evaluate the functional integrity of living cells, as injured/dead cells will display drastically reduced ATP levels [
61]. This platform should, therefore, be reliable for the assessment of compounds for their ability to inhibit gametocytes. However, the assay requires the downstream manipulation of gametocytes (enrichment and isolation) during which gametocyte viability is compromized. ATP levels decrease by as much as 50 % within the first 16 h after enrichment, and this level steadily decreases as gametocytes are further incubated after enrichment to a total level of only 20 % remaining at 48 h, which is the usual timeframe for assaying drug effects. This then influences the number of viable gametocytes remaining in the population used to detect drug effects. Moreover, although this manipulation results in high S/N ratios due to gametocytes being enriched above background, it could contribute to higher inter-assay variability. However, the ATP assay is possibly less prone to artefacts compared to fluorescent viability assays [
62].
One major advantage of the pLDH assay is that it is performed directly on parasite cultures, therefore minimising manipulation of gametocytes. pLDH has been shown to be present in the blood of malaria patients and to be a reliable marker for gametocyte viability as the enzyme is present at high levels throughout gametocytogenesis [
63,
64]. pLDH catalyses the conversion of pyruvate, the final product of glycolysis, to lactate. This flux is important for the regeneration of NAD
+, itself an essential cofactor for glycolysis. Only when increased fermentative glycolysis is possible, do cells exhibit increased proliferation through the anabolic capacity of glycolysis [
65]. However, late stage gametocytes exhibit decreased expression of genes responsible for glycolysis, protein biosynthesis and haemoglobin catabolism [
50]. As mentioned, terminally differentiated gametocytes make use of a canonical TCA cycle, and less glucose is metabolized by fermentation to lactate. Reduced glycolytic activity suggests that less pyruvate is converted to lactate by LDH, which might explain the relatively low S/B values obtained for the pLDH assay. A direct comparison of the pLDH activity in asexual parasites, early and late stage gametocytes will be valuable in confirmation of assay reliability. Additionally, the continued presence of pLDH activity even after parasite death has to be taken into account [
34].
Redox reactive (oxidoreductive indicators) cell permeable dyes like alamarBlue
® and PrestoBlue
® have previously been reported as robust assays amenable to high-throughput screens, more sensitive than traditionally used tetrazolium dyes [
31,
41]. These assays enable ease of use when screening gametocyte populations. However, care must be taken with these assay platforms in the case where compounds target the redox state of the parasite, as it may interfere with the assay readout. Moreover, compounds routinely used like methylene blue cannot be used in this assay platform due to colourimetric interference.
Despite the drawbacks of each assay, the ATP, pLDH and PrestoBlue® assays are useful for assaying gametocytocidal activity of compounds on non-genetically modified lab strains as well as clinical isolates. This information will become increasingly important in drug development programmes, enabling early detection of cross-resistance or efficacy failure of lead gametocytocidal compounds. Assay cascades may also be influenced from an economic point of view, with the PrestoBlue® the least expensive (~$1 per compound) followed by the pLDH and lastly the luminescence ATP and luciferase assays (up to a 7-fold increase in cost).
Compared to the assays above, the bioluminescent luciferase reporter lines enabled stage-specific gametocytocidal activity detection through endogenous luciferase production under stage-specific reporters, again with minimal interference of the luminescent signal by the compounds screened. This assay platform was robust in all cases resulting in high S/N ratios and in quantification of the internal signal. This platform has recently been further optimized to a dual-colour assay to simultaneously and quantitatively assay the viability of different stages of gametocyte populations [
66].
In addition to the above, the blinded nature of the study and the fact that the assays were performed in parallel allow for a situation where assay platforms can be compared directly. The data were validated since the DHA control included in the study correlated very well for all assay platforms with the blinded DHA control included in the compound panel.
As reported [
35,
38], the endoperoxides were able to equally target both the early- and late-stage gametocytes and were the most active compounds tested with IC
50s in the nM range in some assay platforms. The equipotency of the endoperoxides to early- and late-stage gametocytes was confirmed here particularly with the luciferase reporter assay (for artemether, DHA and OZ439). These data are also comparable to the late-stage gametocytocidal activity of endoperoxides (artemether and DHA) [
34]. The endoperoxides are amongst the most potent anti-malarials, fast acting and thought to act through alkylation of haem or other biomolecules [
67] and requiring iron-mediated activation of the endoperoxide bridge. The ability of artemisinins particularly to oxidize cofactors of parasite flavoenzymes contributes to generating cytotoxic metabolites and reactive oxygen species, resulting in oxidative damage to cells [
68,
69]. The PrestoBlue
® assay was able to detect comparable levels of activity to the luciferase reporter assay particularly for DHA, indicating that the oxidoreductive dye is able to measure drug response in the context of oxidative cellular stress, at least for the compounds tested here. It has however been postulated that another target for the endoperoxides may be direct interaction with PfATP6, interfering with ATP synthesis [
69]. As mature gametocytes do not metabolize haemoglobin effectively, the latter may indeed be the physiological mode of action of these compounds in the sexual parasites. Interestingly, the ATP assay indicated poor activity of the endoperoxides compared to the luciferase and PrestoBlue
® assays, previously reported as well [
21,
33]. Within the endoperoxide group, OZ439 was shown to have potent gametocytocidal activity with the luciferase reporter assay in the low nM range, however, none of the other assay platforms were able to detect this activity. This confirmation of the ability of the endoperoxides to target both immature and mature gametocytes sheds light on artemisinin-based combination therapy as transmission blocking drugs, an effect that should not solely be ascribed to the extremely rapid clearance of asexual parasites and young gametocytes. Surprisingly, the endoperoxides seem to exclusively target male gametocytes preventing male gamete formation; the exact reason for this is unclear [
70]. The ATP assay has been reported to be a poor indicator of the gametocytocidal activity of endoperoxides [
21,
33] and this was confirmed here with IC
50s of 15 μM obtained for DHA compared to nanomolar IC
50s seen with the pLDH, PrestoBlue
® and luciferase assays. Alternatively, gametocytes may stay viable in the presence of endoperoxides or at least maintain their ATP pool and LDH activities while some gene promoters are not as active, explaining the difference between reporter gene and metabolic readouts.
The 4-aminoquinolines (chloroquine) were confirmed to be more active against asexual parasites and early stage gametocytes, confirming their targeting of haemozoin formation in these stages of the parasite. Whilst the PrestoBlue® and pLDH assays did seem to indicate some activity against late-stage parasites, the ATP assay is possibly more informative for this chemotype. Comparatively, the 8-aminoquinolines tested here (primaquine and tafenoquine) performed poorly against both early and late stage gametocytes. However, for these compounds, the ATP assay was able to detect low inhibitory activities, implying differences in the mode of action between the 4- and 8-aminoquinolines on mature gametocytes or interference of the compounds tested (primaquine and tafenoquine) with the assay platform.
The 8-aminoquinolines are known to be metabolically activated by liver enzymes, hence eliciting activity against liver stage hypnozoite forms of malaria parasites [
71]. However, in the assay systems employed here, such metabolic activation is not possible but could be resolved by pre-exposure of the drugs to liver cell extracts before analysing their gametocytocidal activity. Needless to say, such metabolic activation is not considered in medium- to high-throughput screening assays of unknown compounds. However, the PrestoBlue
® assay seems to report ‘enhanced’ activity of particularly tafenoquine compared to the other assay systems used, but this is not the case for primaquine. Primaquine is known to have activity against liver stages of
P. falciparum,
Plasmodium vivax and
Plasmodium ovale and is, therefore, of interest in transmission blocking strategies. Primaquine has been shown to be gametocytocidal against all
Plasmodium species for late stage gametocytes through targeting the parasites’ mitochondria, but is not clinically useful against
P. falciparum asexual stages. However, at in vitro gametocytocidal IC
50 values of 1-15 μM [
15,
72], this compound would be identified as not active, as confirmed by the data presented here (<50 % inhibition observed on all assay platforms). This discrepancy between activity observed in vivo and the lack thereof in vitro supports the notion of metabolic activation of primaquine in vivo and thus hampers the use of in vitro assay systems for this class of compounds.
The naphthoquinone atovaquone was able to inhibit ~50 % of particularly immature gametocytes with previous proposed actions including the targeting of the electron transport chain through the cytochrome b ubiquinol oxidation site [
73]. Gametocytes do however have active mitochondria [
8,
65] and according to the ATP levels measured, these parasites are still 50 % viable and may indicate static arrest of the gametocytes after atovaquone treatment. The PrestoBlue
® assay was unable to detect this inhibitory capacity at a primary screening concentration of 1 μM, which may be a concern when such assay platforms are solely used to derive chemical signatures of libraries for gametocytocidal activity [
35,
42]. As this assay supposedly provides a direct readout of cellular respiration, this is either a more sensitive probe of decreased glycolysis and respiration of the parasite upon atovaquone treatment or an indication of pluri-pharmacology of atovaquone in gametocytes. When atovaquone was rescreened against the PrestoBlue
® assay at 10 μM, gametocytocidal activity was however noted for this compound (71 +/- 1.9 % inhibition), highlighting the need to define an optimal concentration threshold for primary screens that minimizes both false-negative and false-positive hit rates. In the case of synthetic compound library screens, the effects of false-negative losses may be lessened if related compounds of the same basic scaffold are identified to be active [
74].