Identi cation and characterization of GAP50 binders with the goal to identify novel antimalarials


 Malaria continues to be a killer disease even in the modern world. Indeed, vaccines and drugs have a lot to learn from the malaria parasite before they can be successful. Here, using a filter for glideosomal anchor protein PfGAP50, we have explored a plethora of small molecules to shortlist eight GAP50 binders with promising antiplasmodial activity (IC50 < 3 µM) that are also highly selective. Of these, Hayatinine, Curine, MMV689758 (Bedaquiline), MMV1634402 (Brilacidin), and MMV688271 with PfINDO IC50 ≤ 1 µM were found to stall merozoite invasion by inhibiting IMC formation besides increasing oxidant levels in trophozoites. Bedaquiline pre-treated and washed healthy RBCs showed prophylactic ability to prevent intraerythrocytic development of malaria parasite. Synergistic activities with ΣFIC values as low as 0.22 (Curine and Artemisinin) or 0.37 (Bedaquiline and Artemisinin) augur well for the development of drug combinations to combat malaria effectively. Interestingly, orally delivered Bedaquiline (50 mg/Kg b. wt.) showed substantial suppression of parasitemia in a mouse model of malaria.


Introduction
Notwithstanding intense efforts to discover drugs and vaccines and attempts to combat it through vector control, malaria continues to be a deadly curse in Africa, Asia, and South America. Current and past antimalarials such as quinoline and artemisinin derivatives, antifolates, and antibiotics targeted metabolic pathways in the asexual stage of Plasmodium 1 . With the emergence of artemisinin-resistant parasite strains and the inability of Artemisinin Combination Therapies (ACT) to clear them, there is a need to identify and aim antimalarials at novel targets with new mechanisms of action. In humans, the mosquito vector-borne malaria parasite traverses through different developmental stages, rst in the liver and then in the blood. Of these, invasion is an essential process for parasite survival and proliferation within the immunologically safe and nutritionally rich con nes of the host cell. This process has led to vaccine based approaches against several invasion related proteins including MSP-1 2 and MSP2 3 , AMA1 4 , Rh5 5 , and EBA175 6 . However, the limited e cacy of these vaccines and the complex life cycle of the malaria parasite with its ability to mutate proteins have been major hurdles in their successful development. Against this background, we selected the glideosomal complex which propels the process of fast invasion as a potential target for drug discovery. The proteins of this complex are involved in the proliferation of both sexual and asexual stages of Plasmodium and their disruption is known to curtail the parasite [7][8][9][10] . The glideosomal complex consists of glideosome associated proteins [11][12][13] 40,45,50, and 70 (GAP40 GAP45, GAP50, GAP70) along with GAPMs 14,15 , Myosin A (MyoA), and myosin A tail-interacting protein (MTIP) 16 . Amongst all these proteins, GAP50 which is highly conserved in Apicomplexans plays a pivotal role by acting rst as a scaffolding protein and next by anchoring the glideosomal complex 17 to the inner membrane 11,18 . Although GAP50 has been studied as a potential transmission-blocking vaccine candidate 19 , it has not till date been studied as a drug target. The 1.7 Å resolution crystal structure of GAP50 by Bosch et. al 20 has further paved the path for anti-malarial drug discovery by targeting this protein. Given the paucity of studies in this direction, here we have chosen to screen numerous small molecules of diverse chemical classes against PfGAP50. We believe that the promising molecules thus identi ed might become the starting points of development for future antimalarials having the potential to disrupt the delicate organization of the glideosomal complex.
To achieve our goal, we screened diverse compounds coming from MMV (pathogen, and pandemic box) as also laboratory synthesized quinolines (Supplementary Table 1) 21 and some phytometabolites isolated in our laboratory ( Supplementary Fig 1) 21,22 against PfGAP50. Successive compound screenings which started with initial permissive cutoffs used in diverse in silico and wet lab binding assays were followed by more stringent in vitro inner membrane complex disruption microscopic assays, in vitroP. falciparum culture based IC 50, and mammalian cells-based selectivity cut-offs.
Our successive screenings have led to the identi cation of potent antiplasmodial lead molecules endowed with high selectivity and low resistance indices that have multi-stage activities. The molecules thus identi ed can serve as starting points for the development of new antimalarials.

Recombinant Expression and a nity puri cation of Pf GAP50
Pf3D7 genomic DNA was used to PCR amplify the PfGAP50 gene using PfGAP50 speci c primers (5'TTTTCATATGAAATGTCAACTACGCTTTGCG-3'-FP) and (5'-TTTGGATCCTTAATCTTTATTTCCCATGGGTCC-3'-RP). The ampli ed product was cloned in pET28a + vector ( Supplementary Fig-6) which was transformed into Rosetta 2 pLyS strain with ligation of PCR product downstream of His tag sequence. Using 1% inoculum of the primary culture, secondary culture was initiated in TB broth supplemented with CoSO 4 (2 µM), Kanamycin (50 µg/ml), and Chloramphenicol 34 µg/ml 20 . Cells at A 600 2.0 were induced with isopropyl β-D thiogalactopyranoside (0.5 mM) at 18°C for 16 hrs. Harvested cells were lysed by sonication in buffer A (50 mM Tris HCl, pH 8.3 supplemented with 10% glycerol and 300 mM NaCl). Cellular debris was removed by centrifugation (13000g, 45 min, 4°C) and the clear supernatant was loaded on to a preequilibrated Ni-NTA column. The column was washed using buffer A with 30 column volumes (CV) and the terminal elute was found to be protein-free as tested by Bradford assay. The bound protein was eluted using buffer B (50 mM Tris-HCl, pH 8.3, 500 mM NaCl, 10% glycerol, 0.5 M imidazole). The eluted proteins were subjected to hydrophobic interaction chromatography by loading them onto a 1M ammonium sulphate in 50 mM Tris-HCl pH 8.3 (Buffer C) preequilibrated phenyl sepharose column followed by washing with buffer C (30 CVs) and elution using decreasing gradient of ammonium sulphate. The puri ed protein was then dialyzed against the imidazole free buffer B and stored at 4°C (up to 2 months).
In silico docking, molecular dynamics simulation and -dG of binding The PDB crystal structure of PfGAP50 (ID: 3TGH) was prepared for docking using the protein preparation wizard 23 of Schrödinger software (version 19-3). Ligand structures drawn using ChemSketch (version 2021.1.2) software were imported into Schrödinger software. Ligprep 23 was used to prepare ligands for docking. Grid was generated to engulf the entire protein, and docking was carried out using the Glide module 24 . Docked poses were then taken for estimation of -dG of binding using the Prime MM-GBSA module 25 . Lead PfGAP50 binders (Bedaquiline, Hayatinine, Brilacidin, MMV688271, and Curine) that showed good promise also in P. falciparum culture-based study were taken for the molecular simulation study 26 . Brie y, the protein-ligand complex was solvated using the TIP4PEW model in an orthorhombic box. Na + and Clions were added to neutralize the system. OPLS3e force eld was used to simulate for 100 ns. In a quest to determine alternate targets of lead molecules, cross-docking was performed using in silico Schrödinger tool. Brie y, malaria parasite protein structures were taken from PDB, and a grid was generated around co-crystallized ligands after preparing the protein and ligand structures. Docking was done using glide module following which binding free energies of proteinligand complexes were calculated using Prime MM-GBSA tool.
Screening of molecules using differential scanning uorimetry Recombinantly expressed and highly puri ed PfGAP50 was used for in vitro small molecule binding using Differential Scanning Fluorimetry (DSF). Compounds were screened at 200 µM using 10 µM of protein. DSF experiment was set up as described by Huynh, K., & Partch, C.L. 27 . T m was calculated using nonlinear regression model of Boltzmann Sigmoidal curve using the following equation where Y: uorescence emission in arbitrary units; X: temperature; Bottom: baseline uorescence at low temperature; Top: maximal uorescence; Slope: the steepness of the curve; and T m : melting temperature of the protein. The binding of some small molecules may manifest as a decrease or increase in protein stability. Molecules that gave a ΔT m > 2.0°C over the solvent control were taken as potential hits.
Binding a nity determination using surface plasmon resonance PfGAP50 (50 µg/ml) was immobilized on a CM5 chip in acetate buffer, pH 4.5 using EDC-NHS coupling. Small test molecules were diluted in 1x PBS supplemented with 5% DMSO with 0.005% Polysorbate 20 (running buffer) to yield solutions of varying concentrations (two-fold serial dilution from 50 µM to 0.3125 µM) that were allowed to ow over the immobilized protein at a ow rate of 30 µl/min. Contact and dissociation times of 60 and 120 sec, respectively were given for each cycle. Regeneration was carried out using 10 mM glycine pH 3.0. Data obtained was tted with a 1:1 binding model to get K a (association constant), K d (dissociation constant), and KD (equilibrium binding a nity constant).
In vitro maintenance of P. falciparum cultures and assessment of antiplasmodial activity of test molecules used as +ve (zero growth) and -ve (100% growth) controls respectively. After 24 h treatment, 20 µL MTT (5 mg/mL in 1x PBS) solution was added to each well, and plates were incubated for four hours at 37°C. Thereafter, plates were centrifuged (700 g, 10 min), and supernatants were aspirated out using a multi pipette. This was followed by the addition of 200 µL of DMSO and incubation of the plate at 37°C for 10 min. h. This was followed by removal of test molecules by three centrifugal washes with 1x PBS and incubation of the washed cells in fresh cRPMI for 48 h. % P was calculated at the end of the 2 nd cycle using SYBR Green staining and counting 10 5 cells via FACS.

Evaluation of GAP50 binder antiplasmodial molecules as disruptors of inner membrane complex formation and invasion
Saini et al 32 have described in malaria parasite the presence of yet another multiprotein complex called Phil1 in close proximity of the glideosomal complex. Interestingly both these complexes present in the space between the plasma membrane and inner membrane have GAP50 as one of the constituents and both complexes appear to play a crucial role in the process of merozoite invasion. In this assay, we have used Phil1 and GAP50 antibodies to track the development of Schizont's inner membrane complex (IMC) in the presence vs absence of the small test molecules that bind GAP50. Early schizonts (30 to 34 h p.i) were treated with test compounds for 12 h, followed by indirect immuno uorescence assay (IFA) to assess IMC development. For this, cells were rinsed once with 1x PBS and xed using a solution containing paraformaldehyde (4% v/v) and glutaraldehyde (0.0075% v/v) in PBS for 30 min. The xed cells were treated with Triton X-100 (0.1% v/v) in 1x PBS (twice, 15 min each) and the so permeabilized cells were blocked using BSA (4% w/v in 1x PBS). Parasites were then incubated with 1°Ab at 1:50 dilution overnight at 4°C, followed by washing and incubation with 2°Ab (Alexa conjugated anti-rabbit antibody 488, Invitrogen) at 1:500 dilution for 3 h at RT. In the end, DAPI (10 µg/mL) was used to stain the nucleus (10 min at RT). Samples were imaged using Nikon A1-R confocal microscope. Similarly, segmented schizont cultures (42 to 46 h. p.i) were treated with test molecules for 8 h and evaluated via IFA for their ability to egress and the resulting merozoites to invade healthy red blood cells leading to the formation of rings.

In vitro potency of the combinations of lead anti-plasmodial molecules
To determine the effect of combining drugs on potency, different drug combinations were made to determine IC 50 of combinations against synchronous ring stage (6 to 12 h p.i, at 1% P and 2% H) of P. falciparum (MRA1240) in culture. 8x IC 50 of each test molecule in 10 % DMSO / 1x PBS was taken as stock (A) and mixed with 8x IC 50 of standard antimalarials (B) in four different molar ratios (4:1, 3:2, 2:3, and 1:4). Each ratio sample was further serially two-fold diluted in 10% DMSO/ 1x PBS. Four microliters of each dilution were mixed with 96 µl of parasite culture (1% P, 2% H). Thus, each experimental well contained a total volume of 100 µl with or without test molecules. Control (0.4% DMSO (v/v)) that is nontoxic to the parasite was used as -ve control (100% growth) whilst Chloroquine (4 µM) was used as + ve control (0% growth). The plates were incubated (37°C, 72 h) followed by an estimation of parasite proliferation using the SYBR Green I lysis method 29 . Data analysis was done as described by Thapar et al 33

In vivo acute toxicity study
In vivo acute toxicity was done as per OECD 423 guideline 35 . Brie y, mice fasted for 4 h before oral dosing of test molecules. After dosing, mice were kept under observation for the next 45 min for any acute toxicity symptoms followed by resumption of their feed and a daily check over 16 days for any long-term toxicity symptoms.

In vivo antimalarial study
In vivo antimalarial study was carried out as per ARRIVE guidelines 2.0 with prior permission from ICGEB animal ethical committee (ICGEB/IAEC/30012021/MPB-8). Brie y, 10 5 P. berghei ANKA infected RBCs were injected into mice (6 to 8 weeks old, average weight 20 ± 2 gm) and grouped randomly with seven mice/group. After 24 h of infection, test molecules were orally administered in vehicle solution (2% (Hydroxypropyl) methylcellulose with 2% Tween 80 in Normal saline) for four consecutive days followed by daily monitor of % P, body weight, and surface temperature day 5 to day 30 36 . Chloroquine (50 mg/kg b. wt.) was taken as + ve control. In silico stabilities of complexes of PfGAP50 with Bedaquiline, Hayatinine, MMV688271, Curine, and Brilacidin

Screening of 951 compounds
Studying the ner nuances of the mode of binding of small molecule ligands to the target proteins of interest can facilitate the development of structure-based novel therapeutics. Towards this goal, of the eight promising molecules ( Fig. 1b) the ve including Bedaquiline, Brilacidin, Curine, Hayatinine, and MMV688271 which were found to be the most promising in most screens were shortlisted for MD simulation to assess the dynamic stability of their intermolecular interactions with amino acid residues constituting the respective binding pockets in PfGAP50. A brief description of molecule speci c interactions is given below: Bedaquiline was found to be interacting through hydrophobic interactions (W35, I69, H256, M278), hydrogen bonding (G65), and a water bridge (N221), (Supplementary Fig. 8 and 9). These interactions were found to be maintained over 30% of 100 ns simulation time and appeared to be helping in stabilizing the binding of Bedaquiline with GAP50, resulting in -36.9 kcal/mol binding energy (Supplementary Table 2). The sensorgram pro le obtained using one-on-one interaction with GAP50 ( Supplementary Fig. 9b) indicated a binding a nity (KD) of 9.27 µM.
The binding of Brilacidin with GAP50 ( Supplementary Fig. 10 and 11) was found to be majorly driven by water bridges mediated via the side chains and the backbone of N127, E129, N148, D201, I322, N349, E351, and L352 which were further strengthened by Hydrogen bond interactions involving E129, E351, and L352. Amongst these interactions, the ones with E129 and E351 were maintained for > 50% of 100 ns simulation time ( Supplementary Fig. 10d). The presence of as many as sixteen water bridges between Brilacidin and GAP50 may have contributed to the most promising MMGBSA dG binding of -64.82 kcal/mol (Supplementary Table 3), with KD 2.81 µM ( Supplementary Fig. 11b).
Interactions of MMV688271 (Supplementary Fig. 12 and 13) with GAP50 were found to be stabilized through Hydrogen It was interesting to observe that the structurally close Curine and Hayatinine, bind to GAP50, but each bind to its unique site (Fig. 2). Further, SPR determined KD of Curine was nearly three times higher than that of Hayatinine, suggesting Hayatinine's interaction with GAP50 is stronger than is the case with Curine ( Supplementary Fig. 14). Hayatinine was predicted to interact with G119, Q120, M146, P147, and H152 via water bridge and E133 via hydrogen bond (Supplementary Fig. 15 and 16). These bonds that were found to persist for > 30% of simulation time, might be preventing Hayatinine from moving out of its surface binding site ( Supplementary Fig. 15b). Residues interacting with Curine ( Supplementary Fig. 15   determined against different stages of the malaria parasite can be quite different. In a bid to explore the differential vulnerability of cell cycle stages of the malaria parasite to our lead antiplasmodial molecules, we subjected different stages of the parasite to each lead molecule at different concentrations. As shown (Fig-3a), there were marked differences in the observed IC 50s against different stages. Taking ER to LR, ET to LT, ES to LS, and LS to ER stages into consideration, USINB4-124-8 showed a much higher IC 50 (µM) (38.12) against trophozoites and schizonts (54.32) than against rings (7.2). In contrast, MMV1782353 and MMV642550 showed good activity (IC 50 < 5 µM) against early and late schizonts while showing less potency (IC 50 > 8 µM) against rings (Fig-3a). It is noteworthy that despite great structural resemblance (Fig-1b), Hayatinine (608.7 Da) showed a marked preference to prevent LS to ER transition whereas the preference for Curine (594 Da) was ES to LS along with ET to LT transitions.
Monitoring growth of malaria parasite in healthy RBCs "pre-treated" with test molecules Test molecules retained by RBCs can prevent the growth and maturation of the parasite. However, entrapment of test molecules by healthy RBCs is hitherto not a very well explored prophylactic strategy. In this experiment, we have pretreated healthy RBCs with 10X IC 50 of each test molecule and used the washed RBCs for invasion by merozoites. Our results (Fig-3b)  parasite growth. In a prophylactic role, Bedaquiline appeared to be the best in being retained at a concentration su cient to kill the parasite (Fig-3b).

Effect of test molecules on the creation of prooxidant milieu in the malaria parasite
Cells work best at optimum Redox conditions created by homeostasis mediated by the combined actions of pro and antioxidant mechanisms present in every cell. Any imbalance in this homeostasis is lethal for the cell 42 . Given the robust heme detoxi cation mechanism that converts the prooxidant toxic heme to hemozoin along with Thioredoxin and Glutathione dependent antioxidant systems in the malaria parasite's metabolically active trophozoite stage 43 , the fact that GAP50 binders showed preferential activity against trophozoites, made us examine if these test molecules in uenced the redox milieu of the malaria parasite. When Trophozoites, pre-treated with redox sensitive H2DFCDA were treated with 2x IC 50 [ET◊LT] of lead molecules such as Bedaquiline and MMV688271, the dye uorescence was found to be signi cantly increased wrt control, indicating increased oxidants in the parasite (Fig. 3c). Although the parasite is adept at regulating its redox status via diverse mechanisms, the skewed prooxidant status stimulated by the small molecules could overwhelm the parasite's redox homeostasis mechanisms and assist in killing the parasite.

Effect of lead molecules on the rupture of schizonts and invasion by merozoites
Since the invasion of merozoites is a crucial step preceding the exponential rise in % parasitemia, molecules with IC 50 < 10 µM against schizonts (Fig. 3a) were studied to examine (a) inhibition of egress of merozoites from schizonts and (b) merozoite invasion into healthy RBCs. Analysis of microscopic images indicated that in presence of MMV molecules1782353 and 642550, schizont rupture was stalled but once the test molecules were removed by centrifugal washings, schizonts were found to rupture and release merozoites. However, the released merozoites were unable to invade, leading to a sharp decrease in %P akin to heparin, a known inhibitor of invasion 44 (Fig-4a). In the case of Hayatinine, 8 h treatment of schizonts led to inhibition of egress with microscopic images showing many schizonts and just a few merozoites stuck to the RBC membrane. Curine allowed schizont egress, but the level of invasion was signi cantly reduced, suggesting that the treatment rendered the merozoites to be non-invasive. With no egress and no invasion, both Bedaquiline and MMV688271 showed schizontocidal activity on mature about to rupture schizonts (Fig-4a). The ve hit molecules with invasion inhibition comparable to control heparin were further con rmed for their invasion inhibitory activity by confocal microscopy (Fig-4b).
IMC formation inhibition using Phil1 and GAP50 as a marker Since our target-based drug discovery was based on screening of small molecules against PfGAP50 and since GAP50 is a protein involved in the formation of IMC during schizogony 13 and anchors several key proteins involved in invasion 15 , we checked whether our lead GAP50 binders (IC 50 against schizonts < 10 µM) at their sublethal concentrations could perturb IMC formation. Recent publications 32,45 have shown colocalization of GAP50 complex with Phil1 complex. Indeed both these complexes seem to work in concert to facilitate the process of invasion 32,45 . Hence, we wanted to examine if the schizont targeting molecules identi ed by us may be disturbing the delicate interaction of these two complexes which gives proper organization to the IMC. With the availability of antibodies against both GAP50 and Phil1, co-localization of Phil1-GAP50 in control vs test molecule treated schizonts were studied. As shown (Fig-5 Table 3), we observed several instances of different kinds of drug interactions ranging from strong synergy to additivity to strong antagonism. When the ∑FIC values were averaged over varying ratios of drug combinations (Fig-6) Tables-5-9). Among the shortlisted proteins, three (Plasmepsins V, Cyclophilin, and Malaria Sporozoite Protein Uis3) were common for all ve lead molecules (Fig-7b). These three targets are known to be expressed only in trophozoites and schizont stages 46 (Fig-7a) and binding of the test molecules to them might explain their cidal activity towards mature stages of P. falciparum. We also identi ed proteins such as peptide deformylase, Pyruvoyl Tetrahydropterin Synthase (PTPS), and glutaredoxin 1 which are uniquely expressed in trophozoites to be the putative targets of test molecules (Fig-7a). In an attempt to nd the binding a nity of our lead molecules to their putative target proteins, MMGBSA dG binding was computed. (Supplementary Tables 5-9). In all cases, dG binding was found to be < -30 kcal/mol (Fig-7c). This low -ve score suggests that the proteins identi ed might indeed be the targets for our lead molecules. However extensive studies will be required to con rm the status of these proteins as drug targets for these molecules.
In vivo toxicity and antimalarial study of Bedaquiline, and USINB4-124-8 USINB4-124-8 at 100 mg/kg b. wt. did not show any signs of acute toxicity (Supplementary Table 10) and till day 16 mice were as healthy as the control mice. Bedaquiline acute toxicity was not carried out as this molecule is already FDA approved for tuberculosis treatment. Treatment of P. berghei ANKA infected mice with vehicle solution (-ve control) led to increased uctuations in body temperature and a decrease in mean body weight concomitant with the progressive increase in % Parasitemia with a median survival of 19 days and death of all 7 mice by day 22. Likewise, USINB4-124-8 (100 mg/kg b. wt.) fed mice also showed no difference from -ve control in terms of survival, % rise in parasitemia, uctuation in body weight, and temperature (Fig-8). In contrast, Bedaquiline (50 mg/kg b. wt.) resulted in the complete suppression of % Parasitemia by day 19, reduced uctuations in temperature and body weight as compared to -ve control, and survival of three out of seven mice till 34 days (Fig-8).

Discussion
Control measures against the vector Anopheles mosquito and extensive use of artemisinin-based combination therapy (ACT) against the parasite did result in a marked decrease in malaria cases in the last decade 48 . However, the surge in malaria cases in Asia as also in Africa which have reported resistance to CQ, sulfadoxine-pyrimethamine, ACT, and other antimalarial combinations, demands new antimalarials against new targets to combat the threat of drug-resistant strains. An emerging strategy to kill Plasmodium is to target its Achilles' heel, viz invasion by sporozoites into hepatocytes, merozoites into erythrocytes, and ookinetes into mosquito gut wall. As many proteins of the merozoites bear similarity with proteins found in sporozoites and gametocytes 15,49 , a small molecule preventing invasion by binding to a common target could block entry of malaria parasite into the safe milieu of different cells in man and mosquito. This could enable the immune system to kill the "out of cell" malaria parasite and not allow it to create havoc by the exponential rise in its numbers seen upon successful invasion into a cell. In the event of rapid death, resistance is not possible either. Thus, we aimed to block invasion so as not to give the parasite a chance for the exponential rise in percent Parasitemia that malaria parasite ensures from within the safe con nes of the host cell. Among the major surface proteins of merozoite that have been targeted for both vaccine and drug development, the highly conserved glideosomal proteins have remained largely neglected 50 . These motor proteins are involved in the meticulously organized development of the inner membrane complex (IMC) that includes the IMC resident multiprotein complexes that are known to facilitate the highly e cient and extremely fast process of invasion by merozoites, sporozoites, and ookinetes. Among these proteins, glideosome associated protein 50 aka GAP50 is highly conserved in apicomplexan parasites and acts as an anchor for other motor proteins. Further GAP50 bears low homology to human protein homologs 47 which makes it a suitable target for drugs.
The presence of conserved regions on the surface of GAP50 20 , zygote motility inhibition by anti-GAP50 antibodies 19 , and the ability of GAP50 to bind human complement factor H 51 to suggest that this protein is of vital importance to the parasite. Hence GAP50 is well suited as a target for screening the next generation of novel drugs against malaria.
The di culty of targeting speci c sites on GAP50 stems from the fact that since GAP50 is an anchor protein with no enzymatic activity, there is no knowledge of ligand binding sites and no known inhibitors against GAP50. In the absence of this vital knowledge, blind in silico docking of 951 compounds was done in parallel with screening these compounds using DSF. These two tools helped in hit enrichment in a high throughput manner allowing us to narrow down from 951 to 345 compounds for in vitro SPR and in vitro antiplasmodial activity. Further, even as the structural homology between GAP50 and hPAP (human purple acid phosphatases) was only ~30% 20 , we selected only those compounds which showed Selectivity Indices > 100 against HUH 7 and HEK293 human cell lines. This stringent criterion helped eliminate all those compounds that might be toxic.
Using the lter of selectivity index, fourteen pathogen and pandemic box molecules were identi ed with resistance indices Before understanding the modes of action of the above mentioned fourteen molecules, their IC 50s against different asexual stages of parasite were determined and compounds with multi-stage activity particularly towards trophozoites and schizonts were prioritized for further investigations. This multi-stage activity assessment with special reference to the lowest IC 50 observed at [T◊S and S◊R] transitions led us to look for various mechanisms that can explain their cidal activity. In this attempt, we observed that treatment with some of these compounds (Fig. 2) led to a signi cant increase in oxidant levels similar to what was observed upon treatments with ART/DHA. This suggested that these small molecules may target proteins involved in oxidant enhancing pathways like hemoglobin digestion which releases heme, mitochondrial electron transport chain or may cause inhibition of superoxide dismutases and thioredoxin-dependent peroxidases 43,55 . Interestingly, between the structurally similar Hayatinine and Curine, it was only the latter that enhanced oxidant levels in the parasite. Challenge with Bedaquiline is known to cause an electroneutral uncoupling of respiration-driven ATP synthesis 56 , which can explain its effect in causing an increase in oxidant level in the parasite. Further, based on our cross-docking studies, Bedaquiline was seen binding with Glutaredoxin I protein (docking score -2.2, and MMGBSA dG binding -30.82 kcal/mol) which might be one of the mechanisms by which it may cause an increase in oxidant level. Brilacidin's prooxidant effect may be related to its ability to cause cellular stress by facilitating the accumulation of misfolded proteins in the endoplasmic reticulum 57 . Indeed the upregulation of various chaperones and proteases by Brilacidin 57 indicates that cytoplasmic protein misfolding stress may be a contributor to the mechanism of action of this drug. On the one hand, since an increase in cellular oxidant levels can cause damage to cellular proteins, lipids, and nucleic acids, which is fatal to the parasite 58,59 , Plasmodium has evolved diverse ways to maintain this redox homeostasis 60 . On the other hand, a recent study by Egwu et. al 61 has indicated superoxide to have a major role in the mechanism of action of several essential antimalarial drugs. Therefore, compounds showing a tendency to increase oxidants to levels that are beyond the capacity of the parasite's homeostatic ability can indeed increase cellular stress leading to the death of the parasite.
Since our screening of compounds was done using GAP50, which is expressed in the late trophozoite and schizont stages (30 -35 h p.i) of Plasmodium and is involved in IMC formation 13 , we checked whether the inhibitors (Bedaquiline, Hayatinine, Curine, MMV688271, and Brilacidin) identi ed by us were able to prevent successful schizogony and egress of merozoites. Their inhibitory activity towards schizont rupture and invasion by merozoites was further con rmed by indirect immuno uorescence assay which showed pyknotic and surface attached merozoites (Fig. 3). Further, to nd if an invasion was inhibited due to disruption of IMC, we did immuno uorescence-based confocal microscopy of developing schizonts in presence of sublethal (2x/1.5x IC 50 ) concentration of compounds to study IMC formation using Phil1 and GAP50 as IMC markers. Our results indicated that Brilacidin, and Hayatinine, which inhibited invasion, also caused deformation in IMC formation, leading to the fall in GAP50-Phil1 co-localization coe cient from 0.8 in untreated control to below 0.6 in treated cultures. On the other hand, the schizontocidal compounds Curine, Brilacidin and MMV688271, did not perturb IMC formation (P-value ~ 0.88). MMV1782353 and MMV642550 which were found to prevent schizont rupture but not an invasion, were seen to prevent Phil1 and GAP50 association (P-value < 0.6).
Data on the in silico binding lead molecules, i.e. MMV688271, Hayatinine, Curine, Bedaquiline, and Brilacidin with GAP50 has shed light on the binding site residues, which can help in the chemical synthetic tweaking of these inhibitors to increase their potency and selectivity. Interestingly while Hayatinine and MMV688271 binding sites on GAP50 were found to consist predominantly of non-polar residues, the sites that recognized Bedaquiline, MMV642550 and Brilacidin were rich in polar residues. In contrast, the MMV688271 binding site contained an equal proportion of polar and non-polar residues. The involvement of water molecules in stabilizing Hayatinine, Curine, MMV688271, MMV642550, and Brilacidin binding in their respective grooves shows the importance of the water bridges present in these regions. It is tempting to hypothesize that binding by the above-mentioned lead compounds may perturb the protein-protein interactions that are the hallmarks of the multiprotein complexes found in the IMC. Also, it was interesting to observe that with Curine and MMV688271 as also MMV642550 and USINB4-124-8 sharing the same binding sites, the eight different GAP50 binders were found to bind at six different binding sites The binding of all schizontocidal molecules (Curine, Brilacidin, and 688271) at the top left edge of the molecule and binding of invasion inhibitory molecules (Hayatinine and Bedaquiline) at bottom right seems to suggest anatomical zones in GAP50 catering to speci c functions. However, this needs further con rmation from crystallographic studies (Fig. 2).
In a bid to explore what proteins other than GAP50 may bind to these lead compounds, we have found that these compounds may have alternate protein targets besides GAP50. The in silico binding of our lead compounds to one or more of these putative targets may well be the explanation for their multi-stage activity. Coupled with over hundred-fold selectivity, the presence of multi targets in malaria proteome is a plus point for these compounds since multi-targeting drugs could have low chances of losing e cacy due to the development of resistance 62 .
The longer half-life of small molecules in vivo is usually associated with their ability to bind with serum proteins or uptake in cells from where they are released in small amounts over a period of time 63 . Since malaria parasites target RBCs, we explored if healthy RBCs could be preloaded with some of our lead compounds such that these "loaded" RBCs could gain competence to restrict the growth of malaria parasites. Such retention of compounds in RBCs offers a prophylactic strategy that can be useful for passengers prone to traveller's malaria. Our data showed Bedaquiline as the most promising molecule that is retained by RBCs in concentrations su cient at preventing the growth of the parasite. This result is not surprising since Bedaquiline is known to bind with plasma proteins which increases its half-life to > 5 months 64 . Similar but slightly less potent effects were also observed in RBCs preloaded with MMV642550, Brilacidin, and MMV688271.
The nding of potent antiplasmodial activity in Bedaquiline is highly signi cant since it is a drug already in use against MDR TB 65 . This molecule has a selectivity of over 20,000 for bacterial ATP synthase 66 over mammalian ATP synthase. Our nding of Bedaquiline's ability to prevent schizogony and subsequent invasion by merozoites became more interesting when we found that Bedaquiline could cure the mice of malaria infection with mitigation of associated symptoms like pyrexia and weight loss. However, in some Bedaquiline fed mice, there was development of malaria leading to death. This is most likely due to pharmacokinetic differences 64 in laboratory mice with some mice reducing the drug level faster than others. Bedaquiline is known to be metabolized by CYP3A4 into N-monodesmethyl (M2) which has less therapeutic activity as compared with Bedaquiline 67 . Since Bedaquiline availability is dose dependent 64 , its e cacy against malaria could be increased using a higher dose of this molecule than the 50 mg/Kg used by us.
As a bonus, we found that combinations of MMV688271, Brilacidin, Hayatinine, Bedaquiline, and Curine with standard antimalarial drugs in speci c ratios showed impressive synergistic to additive to antagonistic effects. This seems to open new prospects for use of novel synergistic combinations for the formulation of new ACTs. In conclusion, we have identi ed GAP50 binder molecules and predicted ligand binding sites on GAP50, which can be taken for future drug optimization and development as novel drugs against malaria. Further, the identi cation of FDA approved Bedaquiline and Brilacidin as antimalarials might expedite their development as partner drugs with ART in Artemisinin combination therapy (ACTs).  Representation of predicted binding sites of lead molecules against GAP50. Eight most promising molecules identi ed at the end of all lters used were found to bind to GAP50 at six different sites. As shown in Results, Curine, Brilacidin, and MMV688271(all binding at the top left edge of the molecule) were found to be schizontocidal, Hayatinine and Bedaquiline (binding at the bottom right) caused inhibition of invasion, while 1782353 and 642550 were schizontostatic since their withdrawal allowed the treated schizonts to egress and merozoites to invade.  Brie y, parasites were treated with test molecules selected from panel (a), followed by xation and staining with DAPI (5 µg/mL) and anti-Phil1 antibody raised in rabbit followed by staining with Alexa 488 uorophore-conjugated anti-rabbit antibody. Notice the Phil1 staining inside the cell in control vs on the periphery in Bedaquiline and Hayatinine treated cultures and the remaining three schizontocidal molecules (Curine, Brilacidin, and MMV688271) exhibiting massive schizont staining.