Diverse antimalarials from whole-cell phenotypic screens disrupt malaria parasite ion and volume homeostasis

Four hundred structurally diverse drug-like compounds comprising the Medicines for Malaria Venture’s ‘Pathogen Box’ were screened for their effect on a range of physiological parameters in asexual blood-stage malaria (Plasmodium falciparum) parasites. Eleven of these compounds were found to perturb parasite Na+, pH and volume in a manner consistent with inhibition of the putative Na+ efflux P-type ATPase PfATP4. All eleven compounds fell within the subset of 125 compounds included in the Pathogen Box on the basis of their having been identified as potent inhibitors of the growth of asexual blood-stage P. falciparum parasites. All eleven compounds inhibited the Na+-dependent ATPase activity of parasite membranes and showed reduced efficacy against parasites carrying mutations in PfATP4. This study increases the number of chemically diverse structures known to show a ‘PfATP4-associated’ phenotype, and adds to emerging evidence that a high proportion (7–9%) of the structurally diverse antimalarial compounds identified in whole cell phenotypic screens share the same mechanism of action, exerting their antimalarial effect via an interaction with PfATP4.


Multiple compounds in the Pathogen Box perturb the cytosolic pH (pH cyt ) of the parasite. The
Pathogen Box was also screened for the ability of the compounds to perturb parasite pH cyt , again using a 96-well plate format. The V-type H + ATPase inhibitor concanamycin A (100 nM), which causes an acidification of the parasite cytosol by inhibiting the plasma membrane V-type H + ATPase 12 , served as a positive control (Fig. 1b). Ten compounds from the Pathogen Box were found to induce acidification of the parasite cytosol when added at a concentration of 1 µM (Supplementary Table S1). None of these ten compounds were included in the Pathogen Box on the basis of having antiplasmodial activity; all were included in the Pathogen Box on the basis of their effect on other organisms.
The PfATP4-associated antimalarial KAE609 is known to induce an alkalinisation of the parasite cytosol, attributed to a lifting of the acid load associated with the (Na + efflux/H + influx) operation of the transporter 3,4 . When tested using the 96-well plate-based pH assay the KAE609-induced alkalinisation was barely detectable -a consequence of the relatively low sensitivity of this assay. Despite this, nine compounds (MMV020623, MMV020136, MMV020710, MMV020520, MMV006239, MMV000858, MMV020391, MMV020081 and MMV001059) appeared to cause an alkalinisation of the parasite cytosol when added at a concentration of 1 µM, as determined by an overlap of the fluorescence trace with that from parasites to which KAE609 had been added ( Supplementary Fig. S1). All nine compounds fell within the eleven compounds identified as being '[Na + ] cyt -disrupting' in the parasite [Na + ] cyt screen. The other two '[Na + ] cyt -disrupting' compounds, MMV085210 and MMV688980, did not cause a discernible alkalinisation under the conditions tested. Of the eleven 'Na + -disrupting' compounds, these two compounds -MMV085210 and MMV688980 -caused the smallest increase in [Na + ] cyt (Table 1) and were the least potent of the eleven compounds at killing parasites, with IC 50 values (i.e. the concentration of compound required to inhibit parasite proliferation by 50% in a 72 h growth assay) of 1.60 ± 0.07 µM (n = 4) and 1.12 ± 0.07 µM (n = 5), respectively ( Table 2). These IC 50 values are higher than the concentration used in this initial pH screen (1 µM).
Multiple compounds in the Pathogen Box perturb parasite volume. The Pathogen Box was screened for compounds that perturb the volume of isolated parasites, using the strategy outlined in Supplementary  Fig. S2. A compound was deemed a 'hit' in the volume assay if it caused the parasite volume to change (i.e. to increase or decrease) by ≥3% following a 20 min exposure. The average volume of isolated parasites was 43 ± 5 fL (mean ± SD; range 34-58 fL; n = 24). A 20 min incubation with the putative PfATP4-antagonist KAE609 (10 nM) caused the parasite volume to increase by 7 ± 2% (mean ± SD; n = 22). Twelve compounds from the Pathogen Box caused the parasite volume to increase by ≥3% when added at a concentration of 5 µM ( Fig. 1c; Supplementary Table S1), with the increase in the range 3-11%. Of these 12 compounds, ten had been identified as Na + -disrupting compounds in the [Na + ] cyt screen ( Fig. 1c; blue bars). The single remaining [Na + ] cyt -disrupting compound, MMV688980, caused a volume increase of 0.9 ± 0.3% (mean ± range/2, n = 2) at the 5 µM concentration tested here, below the 3% volume change threshold used to identify hits in this screen. Of the two remaining compounds that induced cell swelling, one caused parasite acidification ( Fig. 1c; orange bar) and one had no significant effect on either [Na + ] cyt or pH cyt ( Fig. 1c; unfilled bar).
Sixteen compounds from the Pathogen Box (each added at a concentration of 5 µM for 20 min) caused the volume of isolated parasites to decrease by ≥3% ( Fig. 1c; Supplementary Table S1). The volume decrease ranged from 3-20%. Of these 16 compounds, five were identified as causing a cytosolic acidification in the pH cyt screen (orange bars). The remaining nine compounds had no effect on either parasite pH cyt or [Na + ] cyt (unfilled bars).  (Table 1) were selected for further characterisation. The effect of each on parasite [Na + ] cyt homeostasis was first confirmed in SBFI-loaded isolated parasites using a PerkinElmer LS 50B fluorimeter. The resting [Na + ] cyt was 10 ± 3 mM (mean ± SD; n = 6) and the addition of DMSO (0.1% v/v; vehicle control) was without effect (Fig. 2). As was found in the 96-well plate assays, the addition of KAE609 (50 nM in DMSO) caused an immediate-onset, progressive increase in [Na + ] cyt . Each of the eleven compounds of interest caused a similar rise in parasite [Na + ] cyt when tested at a concentration 15× the IC 50 for inhibition of proliferation as determined in the laboratory Dd2 strain (Table 2; see Fig. 2 for sample traces) and also caused an increase in [Na + ] cyt when tested at 1 µM ( Supplementary Fig. S3).

Figure 1.
Results of the initial screen of the Pathogen Box for effects on [Na + ] cyt , pH cyt and cell volume in isolated asexual blood-stage 3D7 parasites. (a) Representative traces showing the effects of KAE609 (50 nM, green trace), 0.1% v/v DMSO (solvent control, red trace), the Na + ionophore gramicidin (5 µM, grey trace) and two different Pathogen Box compounds (a black trace for the 'non-hit' MMV676269 and a blue trace for the 'hit' MMV020081, each at 1 µM) on [Na + ] cyt in isolated SBFI-loaded parasites. For the initial screen each of the four hundred compounds comprising the Pathogen Box was tested for its effect on [Na + ] cyt a single time (i.e. n = 1). (b) Representative traces showing the effects of concanamycin A (100 nM, grey trace), DMSO (0.1% v/v, vehicle control, red trace), and three Pathogen Box compounds (a black trace for the non-hit MMV676401, an orange trace for the hit MMV102872 and a purple trace for the hit MMV676477, each at 1 µM) on pH cyt in isolated BCECF-loaded parasites. For (a) and (b) parasites were suspended in physiological saline at 1-2 × 10 7 cells/mL, and compounds were added 5-30 s prior to the start of the measurement. Fluorescence was measured using the Tecan fluorimeter in 96-well plate format. (c) The number of Pathogen Box compounds (each tested at 5 µM) that caused a ≥3% change in the volume of isolated trophozoites following a 20 min exposure (n = 2). Pathogen Box compounds that caused an increase in [Na + ] cyt are shown in blue and KAE609 (10 nM) is shown in green. Pathogen Box compounds that caused a decrease in pH cyt are shown in orange. Parasites were suspended in physiological saline at a density of 4 × 10 5 parasites/mL. The PerkinElmer LS 50B fluorimeter provides a more sensitive measure of parasite pH cyt than the 96-well plate reader, and it was used to quantify the effect of the eleven compounds on pH cyt in BCECF-loaded isolated parasites. The resting pH cyt in isolated parasites was 7.3 ± 0.1 (mean ± SD; n = 9) and the addition of DMSO (vehicle control; 0.1% v/v) was without effect (Fig. 3a). The addition of the V-type H + ATPase inhibitor concanamycin A (100 nM) caused a pronounced cytosolic acidification. The addition of KAE609 (50 nM) caused a cytosolic alkalinisation and, furthermore, reduced the extent of the acidification seen in response to the addition of concanamycin A (Fig. 3b,c; Supplementary Fig. S3). Both effects have been attributed to KAE609 inhibiting the influx of H + via PfATP4 3,4 . Each of the eleven Pathogen Box compounds tested had a similar effect to KAE609 on parasite pH cyt when tested at a concentration 15× the IC 50 for inhibition of parasite proliferation (see Fig. 3a for sample traces) and when tested at a concentration of 1 µM (Supplementary Fig. S3); i.e. the eleven Pathogen Box compounds caused significant alkalinisation of the parasite cytosol (p < 0.05, Fig. 3b), and significantly reduced the extent of the acidification observed following the addition of concanamycin A, when compared to the DMSO-exposed parasites (p < 0.05, Fig. 3c).
The 18 compounds causing optical effects in the 96-well SBFI assay were retested in the PerkinElmer LS 50B fluorimeter to determine whether they caused alkalinisation of the parasite cytosol in a similar manner to PfATP4-associated compounds. None of these 18 compounds induced alkalinisation of the parasite cytosol when added at a concentration of 1 µM; they were therefore discounted as potential PfATP4-associated compounds. The swelling of isolated parasites induced by each of the eleven compounds (tested at a concentration 15× IC 50 for growth inhibition) was quantified using the Coulter Multisizer 4. All eleven induced significant parasite swelling over the course of a 20 min exposure (p < 0.05; compared to the DMSO control; Fig. 4). The MMV compounds caused parasite swelling of a magnitude similar to that induced by KAE609, with the exception of MMV020391, which caused significantly greater swelling than the KAE609 positive control (p < 0.05). The compound MMV688980, which failed to reach the ≥3% swelling threshold used in the initial volume screen when tested at a concentration of 5 µM, caused significant swelling when tested at a concentration of 16.8 µM (15× IC 50 for growth inhibition; p < 0.05; compared to the DMSO control). By contrast, the unrelated antimalarials chloroquine (150 nM) and dihydroartemisinin (30 nM) were without significant effect on parasite volume (p > 0.05; compared to the DMSO control; Fig. 4).
In biochemical experiments on membrane fractions prepared from isolated parasites approximately one third of the membrane-associated ATPase activity was inhibited by KAE609 (50 nM; Fig. 5). Reduction of the Na + concentration of the medium, from 152 mM to 2 mM similarly reduced the membrane-associated ATPase activity by approximately one third (Fig. 5). On addition of KAE609 (50 nM) to membranes in the low (2 mM) Na + medium there was no further reduction of ATPase activity ( Fig. 5; p > 0.05), consistent with KAE609 and Na + -depletion both eliminating the same (Na + -dependent) component of membrane-associated ATPase activity, as has been reported previously 4 . When added at a concentration of 1 μM, each of the eleven compounds identified as disrupting parasite [Na + ] cyt regulation inhibited membrane-associated ATPase activity to a similar extent to KAE609 (Fig. 5). The unrelated antimalarials chloroquine (150 nM) and dihydroartemisinin (30 nM), and three other randomly selected Pathogen Box compounds (MMV667494, MMV687807 and MMV020982; each tested at 1 μM) had no effect on ATPase activity when compared to the solvent control (0.1% v/v DMSO; p > 0.05; Fig. 5). For all eleven compounds there was no significant additional inhibition when KAE609 was added in addition to the MMV compound, and there was no significant additional inhibition when the [Na + ] was reduced from 152 to 2 mM in the presence of each of the eleven compounds (p > 0.05; Fig. 5). Together, the data are consistent with all eleven compounds inhibiting the same Na + -dependent ATPase activity as is inhibited by the putative PfATP4 inhibitor KAE609.
Finally, the eleven compounds were tested for the relative efficacies with which they inhibit the proliferation of a spiroindolone-resistant parasite line (NITD609-R Dd2 -clone#2) and its Dd2 parent line 7 . The spiroindolone-resistant parasite line, carrying mutations in PfATP4, showed significant cross-resistance to all eleven compounds relative to the Dd2 parent parasites (p ≤ 0.001; Student t tests; Fig. 6; Table 2). By contrast, the spiroindolone-resistant parasite line showed the same sensitivity to artemisinin and chloroquine as its Dd2 parent line (Fig. 6).

Discussion
In this study the compounds comprising MMV's Pathogen Box were screened for their ability to affect a number of key physiological parameters in isolated asexual blood-stage malaria parasites. Eleven of the 400 compounds tested were found to induce a progressive, immediate-onset increase in [Na + ] cyt , similar to that induced by the antimalarial KAE609. Subsequent characterisation of these eleven compounds revealed that all shared the characteristics associated with a mechanism of action involving the ion-transporting ATPase, PfATP4. Specifically, all eleven compounds, in addition to causing an increase in [Na + ] cyt : caused an alkalinisation of the parasite cytosol; reduced the magnitude of the cytosolic acidification seen on addition of the V-type H +  Table 2). Parasites were suspended in physiological saline at 1-2 × 10 7 cells/mL, and fluorescence intensity was measured using a PerkinElmer LS 50B fluorimeter. The results shown are from a single experiment, representative of those obtained in at least three independent experiments.

Figure 3.
Effect of Pathogen Box compounds on pH cyt of isolated asexual blood-stage 3D7 parasites. The BCECF-loaded parasites were suspended in physiological saline at 1-2 × 10 7 cells/mL, and fluorescence was measured using a PerkinElmer LS 50B fluorimeter. (a) Traces showing the effects of DMSO (0.1% v/v; black trace) and MMV006239 (3.6 µM in DMSO; grey trace) on pH cyt . The addition of DMSO had no effect on pH cyt , whereas the addition of concanamycin A (100 nM; grey arrows) caused the cytosol to acidify to a pH well below that of the extracellular saline (pH 7.1). MMV006239 caused the cytosol to alkalinise, and reduced the acidification seen in response to concanamycin A. The results shown are from a single experiment and are representative of those obtained in at least three independent experiments. ∆pH cyt denotes the magnitude of the pH cyt increase observed following the addition of MMV006239. pH cyt (min) denotes the minimum pH reached following the addition of concanamycin A. (b) The magnitude of the alkalinisation observed following the addition of Pathogen Box compounds to parasites (i.e. ∆pH cyt ). The Pathogen Box compounds (each tested at a concentration 15× the IC 50 for inhibition of parasite proliferation; as listed below), and KAE609 (tested at 10 nM and 50 nM), caused a significant pH cyt increase relative to the solvent control DMSO (0.1% v/v). (c) The magnitude of the acidification following treatment with concanamycin A (100 nM) in parasites exposed to Pathogen Box compounds, as indicated by pH cyt (min). The pH cyt (min) measured for concanamycin A-treated parasites exposed to Pathogen Box compounds (tested at a concentration 15× the IC 50 for inhibition of parasite proliferation) was significantly higher than the pH cyt (min) measured for DMSO-exposed concanamycin A treated parasites. For panels (b) and (c) each symbol represents an individual measurement, with the mean pH cyt (min) averaged from the measurements made in at least three independent experiments indicated by the horizontal bars (shown ± SD). The concentrations tested were: 7. ATPase inhibitor concanamycin A; induced parasite swelling; inhibited the Na + -dependent, KAE609-sensitive membrane-associated ATPase activity; and showed reduced efficacy against parasites selected for resistance against KAE609 and carrying mutations in PfATP4. These features are summarised schematically in Fig. 7.  The eleven compounds are structurally diverse. One of the PfATP4-associated compounds, MMV006239, shares structural features in common with antimalarial spiroindolones (including KAE609 and the related spiroindolone NITD246 7 ). However, the remaining ten compounds share little apparent structural similarity either to one another or to PfATP4-associated compounds identified previously 13 .  PfATP4 is postulated to function as a Na + /H + ATPase, exporting Na + from, and importing H + into, the parasite. Inhibition of PfATP4 results in: an increase in parasite Na + content and [Na + ] cyt resulting from the influx of Na + down its electrochemical gradient, into the parasite, via as yet unidentified pathways; a cytosolic alkalinisation resulting from a cessation of the 'acid loading' associated with the import of H + via PfATP4; an increase in parasite volume, resulting from the osmotic consequences of the increased Na + content of the parasite; and the cessation of Na + -dependent ATPase activity in the parasite plasma membrane. The simplest explanation consistent with the available data is that PfATP4 functions as a Na + -dependent (Na + export/H + import) ATPase, and that the eleven Pathogen Box compounds identified here exert their effects via a direct interaction with the protein, inhibiting the Na + export/H + import function, together with the associated Na + -dependent ATPase activity. Inhibition of Na + export results in the observed increase in [Na + ] cyt as Na + leaks into the parasite, down its electrochemical gradient via unknown pathways. The net influx of Na + is accompanied by swelling of the parasite. Inhibition of H + import, and the lifting of the associated 'acid load' results in the observed increase in the resting cytosolic pH, as well as the reduced acidification seen on inhibition of the primary H + extrusion mechanism -the plasma membrane V-type H + ATPase 12 -by concanamycin A. The finding that parasites with mutations in PfATP4 show reduced sensitivity to growth inhibition by all eleven compounds is consistent with the compounds exerting their growth-inhibitory effect via an interaction with the PfATP4 protein.
All eleven of the compounds identified here as having a PfATP4-associated mechanism of action fell within the 125 compounds that had been included in the Pathogen Box on the basis of having been identified in whole cell phenotypic screens as potent inhibitors of the proliferation of asexual blood-stage P. falciparum parasites. None of the 275 compounds included in the Pathogen Box on the basis of their activity against other organisms showed the PfATP4-associated characteristics. The eleven compounds represent 9% of the 125 antimalarial compounds in the collection. This 'hit rate' is similar to that reported in a previous study 1 in which it was found that 7% of the 400 compounds comprising the MMV Malaria Box showed an association with PfATP4. The 400 Malaria Box compounds, like the 125 antimalarial compounds included in the Pathogen Box, were selected on the basis of their activity in whole cell phenotypic screens against asexual blood-stage P. falciparum parasites. Both data sets therefore support the notion that 7-9% of the (structurally diverse) compounds identified in such screens share the same mechanism of action, exerting their antimalarial effect via an interaction with PfATP4. Why this one protein should be the target of such a high proportion of the antimalarial agents identified in whole cell screens is not clear. As noted previously 1 , this may reflect the fact that there are, in the parasite, a limited number of proteins that fulfil the requirements of a drug target, in being essential, accessible, and inhibitable by compounds at the low micromolar/submicromolar concentrations used in the phenotypic screens.
Most of the Pathogen Box compounds with activity against asexual blood-stage P. falciparum parasites have been tested for their effects on Plasmodium liver stages (P. berghei sporozoites; data are available via ChEMBL-NTD: https://www.ebi.ac.uk/chemblntd; accessed 5 th September 2017). The PfATP4-associated compounds identified here had variable, but typically poor, activity against P. berghei sporozoites, inhibiting their development by between 0 and 66% when tested at a concentration of 10 μM. This finding is consistent with previous studies that have noted the lack of liver stage activity of a spiroindolone 8 , a PfATP4-associated aminopyrazole 8 , and the PfATP4-associated compounds in the Malaria Box 14 .
The effects of the antiplasmodial Pathogen Box compounds on Stage V gametocytes have also been investigated (https://www.ebi.ac.uk/chemblntd; accessed 5 th September 2017). Ten of the eleven PfATP4-associated compounds identified in this study reportedly lacked potency against Stage V gametocytes, reducing their viability by ≤60% at a concentration of 10 μM. However, one of the eleven compounds, MMV020081, reduced the viability of Stage V gametocytes (and, in a separate study, Stage IV gametocytes 15 ) with a submicromolar IC 50 . These data mirror previous findings with the PfATP4-associated Malaria Box compounds 14 , which were found to possess variable (and, on average, low) activity against late gametocytes (they were, however, more active against early gametocytes; discussed in ref. 1 ). The activity of PfATP4-associated clinical lead compounds against sexual stage parasites has also been investigated using a variety of assays. Nanomolar concentrations of KAE609 were reported to reduce the viability of Stage II gametocytes, inhibit the development of late gametocytes from Stage II gametocytes, and inhibit the development of oocysts when added to the mosquito blood meal 16 . The lead pyrazoleamide PA92 was shown to inhibit the formation of (male and female) gametes from Stage V gametocytes at concentrations <100 nM 6 . The lead dihydroisoquinolone (+)-SJ733 was found to inhibit the transmission of P. berghei from mice to mosquitoes at doses lower than those required to kill asexual parasites 5 . Taken together, the available data indicate that PfATP4-associated compounds possess activity against a number of sexual stages, with mature gametocytes not being among the most susceptible.
The identification in this study of eleven new PfATP4-associated compounds increases the range of chemical structures known to show an association with PfATP4. Whether, and if so how, such a structurally diverse range of compounds can interact with, and inhibit, this protein is as yet unclear. Resolution of this issue awaits the successful heterologous expression and, ultimately, structure determination, for the protein.
In addition to identifying eleven PfATP4-associated compounds this study identified a number of other compounds that perturbed the pH cyt and/or volume of the parasite. The molecular mechanism(s) underpinning these perturbations remain to be established.

Methods
Ethics statement. The use of human blood in this study was approved by the Australian National University Human Research Ethics Committee. The blood was provided by the Australian Red Cross Blood Service without disclosing the identities of the donors. The donors signed an agreement with the Australian Red Cross Blood Service permitting their blood to be used for medical research.
All methods used in this study were performed in accordance with the relevant guidelines and regulations.
Parasite strains and culture. Plasmodium falciparum parasites of the 3D7 strain, the spiroindoloneresistant strain NITD609-R Dd2 -clone#2 7 (which carries two mutations in PfATP4: T418N and P990R) and its matched parental Dd2 strain, were cultured in human erythrocytes as described previously 17 . Parasites were synchronised by exposure to sorbitol (5% w/v) 18  . Each of the 10 mM solutions was diluted ten-fold in DMSO to yield 1 mM stocks which were dispensed as 10 µL aliquots into the individual wells of a 96-well plate for storage. The PfATP4-associated antimalarial KAE609 (cipargamin) and additional quantities of the Pathogen Box compounds MMV020136, MMV020710, MMV020520, MMV006239, MMV000858, MMV020391, MMV020081, MMV001059, MMV020623, MMV688980 and MMV085210 were obtained from MMV. Dihydroartemisinin was purchased from SelleckChem and chloroquine and artemisinin were purchased from Sigma. The MMV compounds, artemisinin and dihydroartemisinin were dissolved in DMSO; chloroquine was dissolved in water. All compound solutions were stored at −20 °C and thawed immediately prior to experimentation. Stock solutions were subjected to a maximum of 5× freeze-and-thaw cycles before being discarded.
Parasite [Na + ] cyt and pH cyt measurements. For measurements of [Na + ] cyt isolated parasites were loaded with the acetoxymethyl (AM) ester of the Na + -sensitive fluorescent dye sodium-binding benzofuran isophthalate (SBFI, Life Technologies) as described elsewhere 4 , and suspended in physiological saline at a density of 1-2 × 10 7 parasites/mL. For the initial screen of the Pathogen Box the fluorescent signal from SBFI-loaded parasites was measured in a Tecan fluorescence spectrometer plate-reader. In subsequent experiments with those compounds identified in the initial screen as 'hits' , measurements with greater sensitivity were made using a PerkinElmer LS 50B fluorescence spectrometer with a dual excitation Fast Filter accessory. For measurements on both instruments, SBFI-loaded parasites were excited at wavelengths of 340 nm and 380 nm, and the emission was recorded at 515 nm. The ratio of the fluorescence intensity measured at each of the excitation wavelengths (340 nm/380 nm) is indicative of parasite [Na + ] cyt . Measurements performed in the PerkinElmer LS 50B fluorescence spectrometer (but not those obtained using the plate reader) were calibrated, such that measurements of the fluorescence ratio yielded an estimate of parasite [Na + ] cyt , as described previously 4 . For measurements of pH cyt , isolated parasites were loaded with the AM ester of the pH-sensitive fluorescent dye 2′,7′-Bis(2-carboxyethyl)−5(6)-carboxyfluorescein (BCECF, Life Technologies) as described elsewhere 1,4 , then suspended in physiological saline at a density of 1-2 × 10 7 parasites/mL. For the initial measurements in the Tecan plate-reader, as well as subsequent measurements in the PerkinElmer LS 50B fluorescence spectrometer, the excitation wavelengths were 440 nm and 495 nm, and emission was recorded at a wavelength of 520 nm. The ratio of the fluorescence intensity measured at each of the excitation wavelengths (495 nm/440 nm) provides a measure of parasite pH cyt . The fluorescence ratio measurements made in the PerkinElmer LS 50B fluorescence spectrometer (but not those obtained using the plate-reader) were calibrated to enable estimates of pH cyt as described previously 12 .
In the initial screen of the Pathogen Box for compounds that disrupted [Na + ] cyt and/or pH cyt , using the Tecan fluorescence plate-reader, fluorescence intensity was measured for 20-40 min with a measurement taken every 15-45 s. Compounds were loaded into individual wells of a 96-well plate at 40× the required final concentration in 5 µL of physiological saline. A 195 µL aliquot of dye-loaded parasite suspension was then added into each well, 5-30 s prior to measuring the fluorescence intensity, giving a final cell concentration of 1-2 × 10 7 parasites/mL. For the purpose of the initial screens a 1 µM concentration was used for each compound. The final concentration of DMSO in each well was 0.1% v/v.
The 'vehicle control' for both [Na + ] cyt and [pH] cyt measurements was 0.1% v/v DMSO. For [Na + ] cyt measurements, the positive controls were the spiroindolone KAE609 (50 nM), which causes parasite [Na + ] cyt to increase (by inhibiting the active extrusion of Na + via PfATP4 4,9 ), and the Na + -ionophore gramicidin (5 µM), which permeabilises the parasite membrane to Na + , resulting in a rapid influx of Na +4 . For pH cyt measurements, the V-type H + ATPase inhibitor concanamycin A (100 nM), which causes a rapid acidification of the parasite cytosol 4,12 , was included as a positive control. For both Na + and pH measurements, 15-35 Pathogen Box compounds were tested per 96-well plate. Each row of the 96-well plate held five Pathogen Box compounds and the appropriate set of control compounds.
Parasite volume measurements. The Pathogen Box compounds were screened for effects on the volume of isolated trophozoites using a Beckman Coulter Multisizer 4 with a 100 µm ' Aperture tube' , as described elsewhere 10,23 . Isolated trophozoites were suspended at 4 × 10 5 parasites/mL in glucose-containing physiological saline and incubated at 37 °C for 15-20 min in the absence of any compound to allow the parasites to recover from the saponin-isolation procedure. Parasite volume was measured at time-zero (immediately prior to the addition of compound) and 20 min after the addition of compound. Each measurement consisted of approximately 20 000 pulses, with each pulse corresponding to the passage of a cell through the aperture. The resulting parasite volume population data was fitted with a log Gaussian distribution and the median volume of the population thereby determined.
The volume measurements are relatively low-throughput (compared to the 96 well plate assays) and a screening strategy was therefore devised as follows: (1) Those compounds that were found in the initial [Na + ] cyt and/or pH cyt screens to perturb parasite ion homeostasis were tested individually (at a concentration of 5 µM) for their effect on parasite volume. (2) Those compounds that were found in the initial [Na + ] cyt and pH cyt screens not to perturb parasite ion homeostasis were combined in 'groups of five' which were then tested together; i.e. the parasites were exposed to the five compounds at the same time. Each of the five compounds was present at a concentration of 1 µM. (3) For any group-of-five compounds found to perturb the parasite volume by ≥3% in the initial measurement, the experiment was repeated a second time to confirm the original observation. (4) For those group-of-five compounds that were confirmed as perturbing parasite volume by ≥3%, each of the five compounds was tested individually for its effect on parasite volume when added at a concentration of 5 µM. (5) For any of the individual compounds from a group-of-five found to perturb the parasite volume by ≥3% in the initial measurement, the experiment was repeated a second time to confirm the finding.
The application of the strategy is illustrated schematically in Supplementary Fig. S2. The 'vehicle control' for volume measurements was 0.5% v/v DMSO for the initial screening process and 0.1% v/v DMSO for all other measurements.
Membrane ATPase activity assays. Membranes were prepared as described previously 4 , with minor modifications. Briefly, saponin-isolated trophozoites were lysed by first sedimenting the cells by centrifugation (12,000 × g, 30 s) then resuspending them in ice-cold sterile deionised H 2 O (containing a 1/500 dilution of Protease Inhibitor Cocktail Set III; Merck Millipore), thereby inducing cell lysis. The crude membrane preparation was then washed three times in ice-cold water (containing the protease inhibitors for the first two washes; 14,000 × g, 10 min) at 4 °C and the protein content measured using a modified Bradford Assay 24 .
ATPase activity was measured using the PiColorLock Gold Phosphate Detection System (Innova Biosciences) in a 96-well plate format. The reaction mixture in each well consisted of: (1) a pH 7.4 solution containing KCl, MgCl 2 and Tris-HCl (yielding final concentrations of 20 mM, 2 mM and 50 mM, respectively); (2) parasite membrane (at a final concentration of 50 μg protein/mL); (3) NaCl or choline chloride (yielding a final concentration of 150 mM); and (4) ATP (Na 2 ATP.3H 2 O; MP Biomedicals; added last at a final concentration of 1 mM to initiate the reaction). It was necessary to use Na 2 ATP as the commercially available K + and Mg 2+ salts of ATP both have significant phosphate contamination, which interferes with the (phosphate-based) ATPase assay. The use of the Na 2 ATP salt (1 mM) resulted in a 2 mM concentration of Na + in the reaction solution.
Following a 10 min incubation at 37 °C, 100 µL of each reaction mixture was added to 25 µL of the colourimetric reagent ('Gold mix'), which terminated the reaction. 'Stabiliser' (10 μL) was added 2.5 min later and the plate was then incubated for a further 1 h before absorbance was measured at 635 nm. Blank values (from wells containing only water, Gold mix and stabiliser) and control values (from wells containing all components, but for which ATP was not added until after the membrane was exposed to Gold mix) were subtracted from the data.
The effects of the compounds of interest on ATPase activity were assessed both in high-Na + medium (containing 152 mM NaCl) and low-Na + medium (containing just the 2 mM Na + originating from the Na 2 ATP salt), and in the presence and absence of the putative PfATP4 inhibitor KAE609 (50 nM). The final concentration of DMSO in each well was 0.2% v/v. Parasite proliferation assays. The effect of compounds of interest on the proliferation of P. falciparum parasites was measured over 72 h with a starting parasitaemia of 1% ring-stage parasites and a haematocrit of 1%. Parasite proliferation was monitored using the fluorescent DNA-intercalating dye SYBR Safe as described previously 25 . Statistical analysis. Unless stated otherwise, statistical comparisons were made using one-way Analysis of Variance (ANOVA) on pre-normalised data, blocked by experiment. The ANOVA was followed by Bonferroni's post-hoc test when appropriate.
The percentage coefficient of variation for the parasite volume estimates made in the course of the screen of the Pathogen Box compounds was calculated by dividing the average of the standard deviations for the parasite volume estimates made in replicate experiments by the mean value for the parasite volume, then multiplying this value by 100. The inter-assay percentage coefficient of variation for the parasite volume screening assay was determined by calculating the mean of the percentage coefficients of variation for the positive control (parasite volume following 20 min exposure to KAE609) and the negative control (parasite volume following 20 min exposure to DMSO). This yielded a value of 12%.
For the purpose of the Pathogen Box screens the [Na + ] cyt and pH cyt assays were used on a qualitative basis, thereby precluding the calculation of an inter-assay coefficient of variation.