Asexual blood-stage P. falciparum parasites were cultured according to established procedures (Trager and Jensen, 1976) in human RBCs (Australian Red Cross Blood Bank) diluted to 4% haematocrit in RPMI-1640 medium (Sigma-Aldrich) supplemented with 25 mM HEPES (Gibco), 0.37 mM hypoxanthine (Sigma-Aldrich), 31.25 μg/mL gentamicin (Gibco), 0.5% (w/v) AlbuMAX II (Gibco) and 0.2% (w/v) NaHCO₃ (Thermo Scientific). Parasite strains utilised included wild-type 3D7 parasites and 3D7 parasites with a S374R mutation in PfATP4, which confers resistance to the PfATP4 inhibitor MB14 (MB14R^S374R) (Gilson et al., 2019). To generate MB14R^S374R_Hyp1-Nluc parasites, 100 μg of a pEF blasticidin deaminase drug selection plasmid containing Hyp1-Nluc (Azevedo et al., 2013) controlled by a P. berghei ef1α promoter was electroporated into uninfected RBCs (Deitsch et al., 2001; Hasenkamp et al., 2012). The electroporated RBCs were inoculated with MB14R^S374R trophozoite-infected RBCs and transfectant parasites were selected and maintained with 5 μg/mL blasticidin S (Sigma-Aldrich). 3D7_Hyp1-Nluc parasites had been generated by transfection previously (Azevedo et al., 2014) and were maintained using 2.5 nM WR99210 (Jacobus).

MMV665878 (MolPort 002-647-671), MMV396719 (MolPort 002-620-441), MMV006239 (MolPort 002-099-136), MMV020136 (MolPort 000-136-457) and MMV020710 (MolPort 004-129-845) were purchased from MolPort and dissolved in dimethyl sulfoxide (DMSO). Cipargamin (a kind gift from Prof. Kiaran Kirk and Dr Adele Lehane, Australian National University) and WR99210 were dissolved in bicarbonate-free, AlbuMAX II-free RPMI medium. MB14 [synthesised by (Buskes et al., 2016)], Compound 1 (custom made), ML10 (LifeArc), zaprinast and artemisinin were dissolved in DMSO. E64, porcine heparin, chloroquine and blasticidin S were dissolved in Milli-Q water. Compounds were purchased from Sigma-Aldrich unless otherwise specified. In experiments in which each compound was tested at a single concentration, the concentration of DMSO used as a vehicle control matched the highest concentration of DMSO in any compound-treated well. Where compounds were serially diluted, the DMSO vehicle control concentration was equivalent to the highest DMSO concentration in the dilution series.

The compositions of the Na⁺-based and K⁺-based buffers were adapted from the work of Pillai et al. (Pillai et al., 2013). The Na⁺-based buffer consisted of 102.7 mM NaCl, 5.4 mM KCl, 28.6 mM NaHCO₃ and 5.64 mM Na₂HPO₄. The K⁺-based buffer consisted of 64.8 mM KCl, 28.6 mM KHCO₃, 5.64 mM K₂HPO₄ and 84.3 mM sucrose. Both buffers were prepared in deionised water and the pH was adjusted to ~7.2 (comparable to the pH of supplemented RPMI medium, measured at 7.17) using 85% orthophosphoric acid. The osmolalities of the Na⁺-based and K⁺-based buffers were 267 mOsm/kg and 274 mOsm/kg respectively and were within the prescribed osmolality range for RPMI medium (267-277 mOsm/kg). Buffers were supplemented with 0.25% (w/v) AlbuMAX II prior to use.

Routine synchronisation was performed using 5% (w/v) D-sorbitol (Sigma-Aldrich) to selectively lyse trophozoites and schizonts (Lambros and Vanderberg, 1979). Where specified in the text, synchronous schizonts were obtained as follows: trophozoite-stage parasites that had been sorbitol synchronised on the preceding day were incubated with the reversible egress inhibitor ML10 (25-30 nM) overnight. The resultant culture, consisting of RBC-trapped schizonts, was washed twice, resuspended in drug-free supplemented RPMI medium and agitated at 50 rpm at 37°C for 2 h to allow parasites to egress and invade new RBCs. Unruptured schizonts were then lysed by sorbitol treatment, resulting in cultures of parasites aged 0-2 hours post invasion (hpi). Parasites were grown under normal culturing conditions for 40 h. Schizonts (aged 40-42 hpi) were then harvested by Percoll (GE Healthcare) gradient centrifugation (Rivadeneira et al., 1983) and used in assays. Due to the relatively inefficient invasion of MB14R^S374R_Hyp1-Nluc parasites, a 3 h egress and invasion window was used for both 3D7_Hyp1-Nluc and MB14R^S374R_Hyp1-Nluc parasites prior to experiments involving MB14R^S374R_Hyp1-Nluc schizonts.

Proliferation assays were performed as described previously (Gilson et al., 2019). Briefly, cultures of predominantly ring-stage parasites at 0.3% parasitaemia and 2% haematocrit were exposed to a 2-fold dilution series of compounds of interest during a 72-h incubation period at 37°C, after which the RBCs were lysed by freeze-thawing. Lactate dehydrogenase activity was then measured as an indicator of parasite biomass as described previously (Charnaud et al., 2018).

Graphs were produced and statistical analysis conducted using GraphPad Prism (version 9.1.1). Fitted curves were plotted and IC₅₀ values computed using the model “log(inhibitor) vs. response – variable slope (four parameters)”. The unpaired Student’s t-test and one-way analysis of variance (ANOVA) followed by Dunnett’s and Tukey’s multiple comparisons tests were applied as specified in the text and figure legends, using a significance level of 0.05 throughout.

The Malaria Box compounds MMV665878 and MMV396719 and the Pathogen Box compounds MMV006239, MMV020136 and MMV020710 were classified previously as likely PfATP4 inhibitors based on their effects on Na⁺ and pH homeostasis in asexual blood-stage P. falciparum ( Supplementary Figure 1 ). Specifically, all five compounds were found to induce intra-parasitic alkalinisation and Na⁺ accumulation in vitro (Lehane et al., 2014; Dennis et al., 2018b). Another identifying feature of the PfATP4 inhibitors is that parasites selected for resistance to one PfATP4 inhibitor are often cross-resistant to others (Jiménez-Díaz et al., 2014; Lehane et al., 2014; Flannery et al., 2015; Dennis et al., 2018b; Gilson et al., 2019). We investigated whether our five compounds also exhibited this property by measuring their activities against parasites selected for resistance to the PfATP4 inhibitor MB14, which carry a S374R mutation in the pfatp4 gene (MB14R^S374R) (Gilson et al., 2019). After a 72 h drug exposure window, the IC₅₀ of each compound was estimated by measuring lactate dehydrogenase activity as a proxy for parasite proliferation (Makler and Hinrichs, 1993). Two well-characterised PfATP4 inhibitors were also tested in this assay: cipargamin, a spiroindolone antimalarial currently in clinical trials for severe malaria (Rottmann et al., 2010; National Library of Medicine (U. S.), 2022b), and MB14 ( Supplementary Figure 1 ). The MB14R^S374R parasite line was significantly less sensitive than the wild-type 3D7 line to cipargamin (2.8-fold increase in IC₅₀; p<0.05; unpaired Student’s t-test; Figure 1 and Supplementary Table 1 ) and grew to ~85% of vehicle control levels at the highest concentration of MB14 administered in this assay (10 μM) ( Supplementary Figure 2 ). The MB14R^S374R parasite line was significantly cross-resistant to all five MMV compounds (p<0.03; Figure 1 and Supplementary Table 1 ). The fold increase in IC₅₀ in the MB14R^S374R parasites relative to the parental line ranged from 2.2 (for MMV396719) to 7.8 (MMV020136). In contrast, the IC₅₀ for a mechanistically unrelated antimalarial, chloroquine, was comparable for the two parasite lines (p=0.084). These data are consistent with the five MMV compounds sharing a PfATP4-dependent MoA.

In 2018, Subramaniam et al. used flow cytometry and microscopy to identify inhibitors of the schizont-to-ring transition in the MMV Malaria Box, classifying MMV396719 and MMV665878 as an egress inhibitor and an invasion inhibitor respectively (Subramanian et al., 2018). In a 2020 screen of the MMV Pathogen Box, Dans et al. classified MMV006239 as an egress inhibitor and MMV020136 and MMV020710 as invasion inhibitors (Dans et al., 2020). We interrogated these findings using 3D7 parasites transfected with a bioluminescent nanoluciferase enzyme fused to the N-terminal region of the exported Hyp1 protein (3D7_Hyp1-Nluc) (Azevedo et al., 2014). This fusion protein contains the Plasmodium export element (PEXEL) motif from Hyp1 and is therefore exported into the RBC compartment and released into the culture medium upon schizont rupture. Egress can therefore be quantified by exposing samples of cell-free culture medium to the furimazine substrate of nanoluciferase (Nano-Glo) and measuring the resultant bioluminescent signal (Hall et al., 2012; Azevedo et al., 2013). To quantify invasion, infected RBCs were chemically lysed to release intracellular nanoluciferase and exposed to Nano-Glo substrate. The bioluminescent signal intensity of the lysate was proportional to parasite biomass and therefore indicated the fraction of RBCs that were successfully invaded.

A potential explanation for the inconsistent effects of PfATP4 inhibitors during egress and invasion, as reported in the literature, was that these inhibitors blocked egress at certain concentrations and invasion at others. We investigated this possibility by generating concentration-response curves using a recently-developed assay (Dans et al., 2020) referred to herein as the “nanoluciferase egress and invasion assay”. Here, 3D7_Hyp1-Nluc schizonts were added to uninfected RBCs in the presence of serially diluted inhibitors. After 4 h, cell-free culture medium was harvested and used to quantify egress, and unruptured schizonts were lysed by sorbitol treatment (Lambros and Vanderberg, 1979), leaving intact any ring-stage parasites that had developed during the 4 h treatment window. Since young ring-stage parasites express negligible levels of nanoluciferase (Azevedo et al., 2013), invasion could not be measured immediately. The ring-infected RBCs were therefore resuspended in drug-free medium and incubated for 24 h to allow nanoluciferase expression to reach detectable levels. The bioluminescent signal of chemically lysed trophozoite cultures was then measured. Compound 1 (an inhibitor of PfPKG which stalls the parasite’s cell-cycle minutes before egress is due to take place (Gurnett et al., 2002; Collins et al., 2013)) and heparin (Boyle et al., 2010a) were included as positive controls for egress and invasion inhibition, respectively. In this assay, Compound 1, an exclusive egress inhibitor, resembled a dual egress and invasion inhibitor as it caused the invasive merozoites to become trapped inside unruptured schizonts, resulting in minimal ring formation and therefore low parasite biomass after the 24 h drug-free incubation period ( Supplementary Figure 3A ). In contrast, heparin was found to permit egress at all concentrations tested but inhibited invasion in a concentration-dependent manner ( Supplementary Figure 3B ). Due to the assay design, new rings were potentially exposed to test compounds for up to 4 h, meaning that drugs with anti-parasitic activity against rings may have been erroneously classified as invasion inhibitors. To exclude this possibility, we treated early ring-stage parasites with ~10×IC₅₀ of PfATP4 inhibitors for 4 h ( Supplementary Figure 4 ). Drugs were then removed and parasites were incubated for 24 h, until the trophozoite stage. The parasitaemia was then quantified by measuring the nanoluciferase signal of chemically lysed trophozoite cultures. Compound 1, which is inactive against intra-erythrocytic parasite growth (Taylor et al., 2010), served as a negative control; the antimalarials chloroquine and artemisinin disrupt ring-stage development (Wilson et al., 2013; Shivapurkar et al., 2018) and served as positive controls. All PfATP4 inhibitors behaved similarly to Compound 1 in this assay (p≥0.6; one-way ANOVA with Dunnett’s multiple comparisons test), indicating that these compounds do not arrest ring-stage development or that their effect is reversible.

The “nanoluciferase egress and invasion assay” was initially conducted using 3D7_Hyp1-Nluc parasites that had been routinely synchronised by sorbitol treatment (Lambros and Vanderberg, 1979) during the preceding growth cycles. Sorbitol induces osmotic lysis of schizonts and trophozoites but preserves ring-stage parasites (~0-24 hpi) (Wagner et al., 2003; Counihan et al., 2017) and is therefore a relatively imprecise synchronisation tool which results in mixed populations of early/mid/late-stage schizonts at the time of the assay. In this context, low concentrations of PfATP4 inhibitors moderately reduced nanoluciferase release relative to the vehicle control ( Figure 2 , left-hand column and Supplementary Figure 5 ). This trend was reversed at higher concentrations, with increasing drug concentration corresponding to increasing release of nanoluciferase, such that the extracellular bioluminescent signal approached (or, in some cases, exceeded) that measured in vehicle control-treated parasite samples. Despite pronounced inter-assay variability in the egress curves, the MMV compounds appeared to consistently reduce subsequent invasion in a concentration-dependent manner.

An inherent limitation of the “nanoluciferase egress and invasion assay” is that it does not distinguish between two distinct mechanisms of invasion inhibition. The first mechanism corresponds to compounds that act on unruptured schizonts to reduce the invasive capacity of released merozoites and the second mechanism applies to compounds that directly target free merozoites or disrupt the interaction between merozoites and uninfected RBCs. We therefore sought to clarify the timing of action of our chosen PfATP4 inhibitors by testing these compounds for their capacity to inhibit the invasion of purified merozoites ( Figure 3A ) (Boyle et al., 2010b; Wilson et al., 2013) Briefly, magnet-purified schizonts were incubated with the cysteine protease inhibitor E64 (Glushakova et al., 2009), which prevents rupture of the host cell membrane, until discrete merozoites were visible inside the schizonts as determined by Giemsa-stained smears. The mature schizonts were then mechanically ruptured to release merozoites, which were immediately added to uninfected RBCs and allowed to invade for 30 min in the presence of ~10×IC₅₀ of PfATP4 inhibitors. The inhibitors were then removed and the infected RBCs were resuspended in drug-free medium and incubated for 24 h. Invasion was quantified as described above and the invasion efficiencies in the presence of PfATP4 inhibitors ranged from 78.5 ± 6.1% (for MMV665878; mean ± SD) to 97.6 ± 5.7% (cipargamin). No PfATP4 inhibitor differed significantly from the negative control antimalarial chloroquine, which does not affect invasion (Burns et al., 2019) (p>0.2; one-way ANOVA with Dunnett’s multiple comparisons test). Conversely, a known invasion inhibitor, heparin (Boyle et al., 2010a), reduced invasion to 29.6 ± 18% (p<0.0001). These data indicate that PfATP4 inhibitors neither block parasite invasion directly nor prevent the conversion of merozoites to rings.

P. falciparum replicates by schizogony, during which successive, asynchronous nuclear divisions are followed by a single, synchronous cytokinetic division (Rudlaff et al., 2020). This results in a segmented schizont containing ~16-32 merozoites (Salmon et al., 2001). The precise temporal regulation of merozoite egress is therefore crucial, as early egress would result in the release of immature, non-invasive merozoites. As it was possible that PfATP4 inhibition prevented invasion indirectly by accelerating or prematurely triggering egress, we investigated this by monitoring the trajectory of egress during a 4 h drug treatment period. Prior to the assay, 3D7_Hyp1-Nluc parasites (which have an asexual cell cycle of ~44 h in our laboratory) were synchronised to a 2 h age window and grown to the late schizont stage (40-42 hpi). Schizonts were then exposed to ~10×IC₅₀ of PfATP4 inhibitors and cell-free medium was harvested and the nanoluciferase signal intensity was measured after 10, 20, 30, 40, 60, 120 and 240 min ( Figure 3B ). Compound 1 and a DMSO vehicle control served as positive and negative controls for egress inhibition, respectively. By 120 min, all PfATP4 inhibitors except cipargamin had significantly reduced egress relative to DMSO (p ≤ 0.01; one-way ANOVA with Tukey’s multiple comparisons test). For cipargamin, the inhibition of egress reached significance at 240 min (p<0.0001). No PfATP4 inhibitor differed significantly from Compound 1 at any time point. The data indicate that the extent and trajectory of egress are comparable during treatment with PfATP4 inhibitors or Compound 1 and thus exclude the possibility that PfATP4 inhibition accelerates merozoite egress. The inhibition of egress measured in the nanoluciferase-based assay was subsequently validated by examination of Giemsa-stained thin blood smears prepared after treating synchronous late-stage schizonts with ~10×IC₅₀ of PfATP4 inhibitors for 4 h, by which time vehicle control-treated cultures consisted predominantly of ring-stage parasites ( Supplementary Figure 6 ). The PfPKG inhibitors Compound 1 and ML10 and the cysteine protease inhibitor E64, which acts downstream of ML10 and Compound 1 (Glushakova et al., 2009; Dans et al., 2020), served as positive controls for egress inhibition. Whereas E64-treated schizonts tended to rupture during smearing, the five MMV compounds resembled Compound 1 and ML10 in that they produced RBC-trapped schizonts which remained intact during smearing.

As the egress inhibition observed in schizont rupture time-course assays and in Giemsa-stained smears contradicted the apparent “enhanced rupture” phenotype seen in the “nanoluciferase egress and invasion assay”, we repeated the latter assay using 3D7_Hyp1-Nluc parasites that had been synchronised to a 2 h age range and grown to the late schizont stage (40-42 hpi). The treated parasite populations now consisted primarily of well-segmented schizonts, with minimal early/mid-stage schizonts. In this context, the five MMV compounds resembled Compound 1 ( Supplementary Figure 3A ), reducing egress and invasion in a concentration-dependent manner without exclusively inhibiting invasion at any concentration ( Figure 2 , right-hand column). However, whereas the mean bottom plateaus of the egress and invasion curves for Compound 1 did not differ significantly (p=0.4; unpaired Student’s t-test), the invasion concentration-response curve generated by each of the five MMV compounds reached a significantly lower bottom plateau than the corresponding egress curve (p ≤ 0.03).

Taken together, these data suggest that the effects of PfATP4 inhibition on schizont rupture depend heavily on schizont age. Specifically, the PfATP4 inhibitors appeared to block egress from late schizont-infected RBCs; however, in the presence of younger schizonts, high drug concentrations enhanced nanoluciferase release, possibly by inducing lysis of infected RBCs (examined in a later section).

Several PfATP4 inhibitors have been reported to induce swelling and lysis of trophozoite- and schizont-infected RBCs (Jiménez-Díaz et al., 2014; Vaidya et al., 2014; Dennis et al., 2018a; Gilson et al., 2019), likely due to the osmotic influx of water that accompanies intra-parasitic Na⁺ accumulation. We therefore speculated that the “enhanced rupture” phenotype observed when high concentrations of PfATP4 inhibitors were added to minimally segmented schizonts or schizont populations spanning a broad age range was attributable to Na⁺-dependent osmotic lysis of the host cell. To test this hypothesis, we treated 3D7_Hyp1-Nluc trophozoites (~26-32 hpi) with PfATP4 inhibitors in K⁺-based buffer, Na⁺-based buffer and standard RPMI medium ( Figures 6A–E ). Drugs were administered at concentrations of ~5×, 10× and 20× the IC₅₀, estimated using the lactate dehydrogenase growth assay. Artemisinin, a potent but non-lytic antimalarial (Gilson et al., 2019), served as a negative control ( Figure 6F ). After a 4 h drug treatment period, the growth medium was harvested and the extent of cell lysis was computed by measuring the nanoluciferase released into the medium and expressing this as a percentage of the total nanoluciferase content of the whole culture (medium and cells, with parasites and host cells chemically lysed to release nanoluciferase). The percentage lysis at 4 h in DMSO-treated samples was subtracted from the percentage lysis in drug-treated samples to yield the compound-dependent cell lysis.

In K⁺-based buffer, compound-dependent lysis was close to negligible for all MMV compounds, ranging from -0.16 ± 0.25% (for MMV396719 at 10×IC₅₀; mean ± SD) to 0.66 ± 0.38% (MMV665878 at 20×IC₅₀), and displayed no discernible concentration dependence. In Na⁺-based buffer and RPMI medium, the MMV compounds induced nanoluciferase release in a concentration-dependent manner. MMV665878 was the most lytic of the five compounds, causing 15 ± 1.5% lysis and 11 ± 5.2% lysis at 20×IC₅₀ in Na⁺-based buffer and RPMI medium respectively. At all concentrations tested, the MMV compounds induced slightly but not significantly less lysis in RPMI medium than in Na⁺-based buffer (p>0.21; unpaired Student’s t-test). As anticipated, artemisinin caused minimal lysis in all conditions, consistent with the induced release of nanoluciferase being a consequence of PfATP4 inhibition.

Egress is preceded by a complex signal transduction pathway. One strategy for determining the point along this pathway at which a novel compound acts involves arresting the pathway at a known point using a well-characterised, reversible enzyme inhibitor. After removal of the reversible inhibitor, the novel compound is added, and we can observe whether egress is now able to proceed. If egress resumes, it is inferred that the test compound imposes a ‘barrier’ upstream of the well-characterised inhibitor. Conversely, if egress remains blocked, the novel compound must act downstream of the well-characterised inhibitor. This strategy was implemented using ML10, a potent inhibitor of PfPKG (Ressurreição et al., 2020). We incubated trophozoite-stage parasites with 25 nM ML10 for ~15 h, during which time the parasites matured to the late schizont stage but were unable to egress. The ML10 was then removed and PfATP4 inhibitors, Compound 1 and ML10 were immediately added to separate aliquots of the schizont culture ( Figure 7A ). The trajectory of egress was monitored by periodically sampling the culture medium and measuring its nanoluciferase signal. In this context, nanoluciferase release (normalised to the final timepoint in DMSO-treated wells) was negligible at 10, 20 and 40 min. At 120 min, the normalised nanoluciferase signals in the presence of PfATP4 inhibitors ranged from 74 ± 12% (for MMV665878; mean ± SD) to 96 ± 0.79% (cipargamin) and were significantly higher than the nanoluciferase signals in wells treated with ML10 (8.6 ± 2.0%) and Compound 1 (6.4 ± 0.19%) (p<0.0001; one-way ANOVA with Tukey’s multiple comparisons test). These data indicate that PfATP4 inhibitors act upstream of PfPKG and are not effective during the brief window between PfPKG activation and merozoite release.

Egress is thought to be triggered when the intra-parasitic cGMP concentration exceeds a certain threshold required to activate cGMP-dependent PfPKG, the deployment of which initiates a proteolytic signalling cascade that culminates in the rupture of the parasitophorous vacuole membrane and RBC membrane (Tan and Blackman, 2021; Dvorin and Goldberg, 2022). Guanylate cyclase α (PfGCα) and phosphodiesterases, which produce and degrade cGMP respectively (Collins et al., 2013; Nofal et al., 2021), regulate the timing of this event. While normal egress likely occurs when an unknown stimulus enhances the activity of PfGCα, egress can also be triggered prematurely by small-molecule phosphodiesterase inhibitors, such as zaprinast, which rapidly elevate cGMP levels, causing PfPKG activation and parasite egress. In 2013, Collins et al. reported that when the PfPKG inhibitor Compound 1 was co-administered with zaprinast, egress was inhibited because PfPKG remained inactive despite the zaprinast-induced rise in intra-parasitic cGMP (Collins et al., 2013). Data obtained using our 3D7_Hyp1-Nluc parasites support this finding. Compound 1 in combination with zaprinast inhibited egress to an extent indistinguishable from Compound 1 alone (p≈1; one-way ANOVA with Tukey’s multiple comparisons test; Figure 7B ). Conversely, while the PfATP4 inhibitors alone reduced egress relative to the DMSO vehicle control, co-administration of zaprinast with PfATP4 inhibitors increased egress to between 145 ± 52% (MMV020136; mean ± SD) and 182 ± 24% (MMV020719) of the vehicle control values. For five of the six PfATP4 inhibitors tested (MMV665878, MMV396719, MMV006239, MMV020710 and cipargamin), the difference between ‘zaprinast + PfATP4 inhibitor’ and ‘PfATP4 inhibitor alone’ reached significance by 60 min (p ≤ 0.026; Figures 7C (i-v)). MMV020136 followed a similar trend to the other MMV compounds but did not reach significance due to higher variability between replicates ( Figure 7C (vi)). Overall, these data corroborate a model in which PfATP4 inhibitors act upstream of the cGMP checkpoint, possibly by (directly or indirectly) inhibiting the unknown stimulus that normally upregulates PfGCα activity in late schizonts. By arresting the breakdown of cGMP, zaprinast sustains high cGMP levels and partially overcomes the egress blockade imposed by PfATP4 inhibitors, resulting in a level of egress between that occurring in the presence of the vehicle control and that achieved by zaprinast alone.