A total of 951 different compounds from diverse chemical libraries [pathogen box (400), pandemic box (400), Azepino quinolines (140), and specialized secondary metabolites purified from C. pareira (11)] were screened for their in silico binding to PfGAP50 by via (a) Schrödinger drug discovery suite, (b) differential scanning fluorimetry (DSF) and (c) SPR. As shown (Fig-1a), employing different filters at successive stages, we shortlisted 345 (after in silico Extra Precision (XP) mode docking and DSF), 39 (after in vitro antiplasmodial assay), 8 (after consideration of both anti Pf IC50 and selectivity indices) and 3 (after checking the disruption of inner membrane complex and invasion). Upon analysis of results, it was interesting to know that of the pathogen and pandemic box molecules shortlisted as PfGAP50 binders, 20 were those that had been pre-approved by the FDA for the treatment of fungal, bacterial, and viral infections. For molecules selected from MMV, the PfGAP50 KD values (in parentheses, µM) observed were Mefloquine (40), Suramin (28.6), Bitertanol (0.333), Letermovir (19.2), Everolimus (0.257), Deferasirox (4.93), Valdecoxib (9.63), Ketoconazole (16.5), Ezetimibe (1.16), LED209 (2.62), Birinapant (12.4), Mgb Bp3 (14.8), Alexidine (27.0), Tipifarnib (52.2), Eberconazole (75.9), Ozanimod (76.5), Pentamidine (12.7), Rifampicin (0.116), Bedaquiline (9.27) and Brilacidin (2.81). The other MMV molecules identified as PfGAP50 binders are pending approval as drugs (SI_2). Although Mefloquine, Suramin, Rifampicin, and Pentamidine were identified as binders to PfGAP50, their discontinued use against malaria6–8 led us not to subject them for detailed in vitro stage-specific and IMC inhibition assays. Among 11 molecules from C. pareira (SI-1_Fig 2), Hayatinine, Curine, and Magnocurarine were identified as PfGAP50 binders with KD below 100 µM (SI-1_Table-2).
In vitro antiplasmodial potency and mammalian cell cytotoxicity
Our next step for shortlisting molecules was screening for potencies of molecules for their antiplasmodial IC50s against chloroquine-sensitive Pf3D7, chloroquine-resistant PfINDO, and artemisinin-resistant PfCam 3.1R539T strains. We shortlisted the best top eight molecules on the bais of IC50s and selectivity indices. (Fig-1b;SI_1-Table 1). It is noteworthy that among shortlisted molecules, Bedaquiline aka MMV689758 (anti-tuberculosis molecule), MMV1634402 aka Brilacidin (a mimic of host defense peptide defensin), MMV688271 MMV642550 and Hayatinine showed IC50s ≤ 1.5 µM against all the three strains of Plasmodium. USINB4-124-8 and Hayatinine were found to be more active against PfCam 3.1R539T (IC50 6.7 nM and 80 nM, respectively) than against PfINDO strain (IC50 2.42 µM and 0.41 µM, respectively). On the other hand, MMV1782353 and Curine, which showed RIs < 1 (RIINDO/3D7 0.64 and 0.34, respectively), exhibited RI >1 against PfCam 3.1R539T (RI1240/3D7 21.25 and 1.91, respectively).
Inhibition of asexual stages of P. falciparum
In its replicative phase, the malaria parasite traverses through diverse cell cycle stages like the ring (R), trophozoite (T), and the Schizont (S). These three stages have morphologically identifiable sub-stages like early (E), intermediate and late (L). During these transitions, there are distinct changes in chromatin remodelling, transcriptome, and proteome of the malaria parasite, making some targets fade away and some new targets emerge9. Hence it is not surprising to expect that IC50 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 (Table-1), there were marked differences in the observed IC50s against different stages. Taking ER to LR, ET to LT, ES to LS, and LS to ER stages into consideration, the ratio of the highest to the lowest IC50s obtained ranged from 54.3 µM (USINB4-124-8) to 0.58 µM (MMV688271). As a specific example, USINB4-124-8 showed a much higher IC50 (µM) (38.12) against trophozoites and schizonts (54.32) than against rings (7.2). In contrast, MMV1782353 and MMV642550 showed good activity (IC50 < 5 µM) against early and late schizonts while showing less potency (IC50 > 8 µM) against rings (Table-2). 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.
Kill kinetics of test molecules
To understand the time kinetics of action by these test molecules on development, egression, and invasion events of P. falciparum, we treated PfINDO rings and schizonts for variable times. Briefly, rings (6 - 12 h p.i), and schizonts (32 - 36 h p.i), were treated with test molecules (at IC50 and 2x IC50) for variable times (D+) as indicated (Fig-2a), followed by culture in test molecule free medium (D‑) for a total (D+ + D-) time of 96 h. MMV1782353 (0.4 µM), MMV642550 (1.2 µM) and MMV1634402 (0.67 µM) were found to reduce %P in cultures grown under D+16 h and D- 80 h conditions to half the %P of the control. In contrast, MMV6888271 (0.25 µM) and Hayatinine (0.41 µM) caused 90 % reduction in %P following 12 h D+ 84 h D- treatments (SI-1_Fig-9). USINB4-124-8 (2.42 µM), on the other hand, did not cause any reduction in %P by 12 h D+ - 84 h D- treatment. 24 h D+ - 72 h D- treatment did show a reduction of 25 % but drastic reduction up to 95 % was seen only when D+ treatment was for 36h. Bedaquiline (1.09 µM), while not hindering stage progression, was found to significantly inhibit invasion of fresh RBCs by merozoites (SI-1_Fig-9).
Hayatinine and USINB4-124-8 decreased %P to half of the control in 16 h D+ treatment of schizonts with microscopic images showing late-stage 46 h schizonts and merozoites in treated cultures vs fully formed rings in control cultures. Curine and MMV 642550 killed 90 % of schizonts following 8 h D+ and the surviving parasites revealed non invasive merozoites (Curine) and pyknotic schizonts (MMV642550) vs rings in case of control.
MMV1634402 and MMV1782353 showed schizontocidal effects after 4 h and 8 h D+ treatments, respectively, with pyknotic schizonts seen after 12 h treatment. MMV688271 reduced %P to half of the control in 12 h D+ treatment with surviving parasites seen invading and forming rings (SI-1_Fig-10). A reduction of %P to basal level at 8 h D+ treatment and pyknotic schizonts at 16 h D+ treatment, suggested the schizontocidal action of Bedaquiline (SI-1_Fig-10).
Monitoring growth of malaria parasite in healthy RBCs “preloaded” with test molecules
Test molecules retained inside 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, healthy RBCs preincubated with different test molecules for 72 h and thereafter given centrifugal washes to remove test molecules from the medium, were exposed to mature “about to egress” schizonts. Parasitemia was monitored after 96 h. Control, Curine, Hayatinine, MMV642550, and USINB4-124-8 treated RBCs showed 100 % growth (Fig-2b). On the other hand, MMV642550, MMV1634402, and MMV688271 treated RBCs showed only 50 % growth of control indicating a decrease in growth and maturation rate of the parasite. Interestingly Bedaquiline loaded RBCs, showed necrotic trophozoites with % growth < 5 %. It appears that Curine, Hayatinine, MMV642550, and USINB4-124-8 are either not taken up or easily washed off and are not retained by healthy RBCs. On the other hand, MMV642550, MMV1634402, and MMV688271 appear to be retained partially, which causes partial retardation in parasite growth. In a prophylactic role, Bedaquiline appeared to be the best in being retained at concentrations that kill the parasite (Fig-2b).
Effect of lead molecules on the rupture of schizonts and invasion by merozoites
Molecules with IC50 < 10 µM that were preventing egress or invasion during LS à ER transition were studied to examine inhibitions of egress from schizonts and invasion by merozoites into RBCs. In the presence of MMV1782353/ MMV642550, schizont maturation was found to be stalled but once these molecules were removed schizonts were able to rupture and release merozoites. However, the released merozoites were unable to invade, leading to a sharp decrease in %P with respect to -ve control (Fig-2c). In the case of Hayatinine, and MMV1634402), 8 h treatment led to arrest at 46 to 48 h p.i schizonts with few merozoites stuck to the RBC membrane. Curine allowed schizont egress, but the level of invasion was significantly reduced. Both Bedaquiline and MMV688271 showed schizontocidal activity on mature about to rupture schizonts (Fig-2c).
Effect of test molecules on ROS levels
Molecules showing IC50 (LRàT) < 10 µM, were taken at 2 x IC50 to assess their effects on ROS levels. Bedaquiline and MMV688271 were found to elevate ROS similar to H2O2 (100 µM) while MMV642550/MMV1634402 showed ROS levels similar to ART (700 nM). Further, in all samples, ROS levels were found to be quite similar at both 8 h and 12 h, suggesting 8 h treatment is sufficient to have maximum ROS levels. Although the parasite is adept in regulating ROS via the formation of hemozoin, Glutathione, and Catalase, the large excess production stimulated by the small molecules could overwhelm the parasite's ROS homeostasis, leading to its death. Among the seven molecules selected, only two molecules viz. Hayatinine and Curine showed only moderate elevation in ROS (Fig-2d). This may be due to the combined effects of ROS stimulating and ROS scavenging10 activities of these two molecules.
Putative antiplasmodial targets of Bedaquiline, Hayatinine, MMV1634402, MMV688271, and Curine
To find if there may be alternate targets other than GAP50 for the promising molecules studied by us, we performed cross-docking against 68 different malaria proteins, involved in different metabolic pathways of Plasmodium. Each lead molecule was docked into the active sites of selected proteins, and the scores for docking and dG binding were computed. Out of the 68 proteins studied (SI-Table 2), the number of the protein-ligand complexes with negative values for dG binding score were 63 (MMV688271), 47 (Bedaquiline), 18 (MMV1634402), 31 (Hayatinine), and 42 (Curine) (SI-1_Table-9). Thirteen different proteins showed binding to all the five lead molecules (Fig-2e).
IMC formation and invasion inhibition using Phil1 marker
Since PfGAP50 is involved in the formation of IMC during schizogony11 and anchors key proteins involved in invasion12, we checked whether our lead GAP50 binder molecules could perturb IMC formation and invasion by merozoites. Indeed, two of the major inner membrane multiprotein complexes viz. Glideosomal complex13 and Phil1 complex14 play a key role in invasion. Interestingly pull-down experiments have demonstrated the presence of GAP50 in both these complexes2,5,11. We hypothesized that if IMC formation is disrupted, then the regular arrangement of Phil1 seen in control cells will not be seen in parasites with disrupted protein complexes. Our results showed that untreated control parasites showed distinct punctate green fluorescence marks on the periphery of schizont, indicating proper organization of Phil1 with GAP50 in IMC, and such parasites were highly invasive (Fig-3, left panel, row A). However,in Bedaquiline, Hayatinine, and MMV1634402 treatment groups, an irregular distorted, depleted pattern of Phil1 fluorescence around developing merozoites was seen, suggesting a defect in IMC formation. In all instances where IMC formation was found to be distorted (Fig 3, left panel, rows C, D, and F), we noticed a blockade of invasion as well (Fig 3, right panel, rows C, D, and F). However, MMV688271 and Curine, which did not show any effect on IMC formation, nevertheless caused inhibition of invasion (Fig-3).
In silico stabilities of complexes of PfGAP50 with Bedaquiline, Hayatinine, MMV688271, Curine, and MMV1634402
Studying the finer 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, Bedaquiline and MMV688271 (pathogen box), Hayatinine and Curine (Cissampelos pareira), and MMV1634402 (pandemic box) were shortlisted based on the in silico screening followed by MD simulation to assess the dynamic stability of their intermolecular interactions with amino acid residues constituting the respective binding pockets in PfGAP50 (Fig-4).
Bedaquiline interaction was found to be primarily through hydrophobic interactions (W35, I69, H256, M278), hydrogen bonding (G65), and water bridge (N221), as seen through MD simulation (Fig-4a). These interactions were found to be maintained over 30 % of 100 ns simulation time and appeared to help in stabilizing the binding of Bedaquiline with GAP50, resulting in -36.9 kcal/mol binding energy. The sensorgram profile obtained using one-on-one interaction with GAP50 suggested incomplete dissociation with a binding affinity (KD) 9.27 µM. Curine, that shares structural similarity with Hayatinine, was also found to bind to GAP50, but at a completely different site from that of Hayatinine (SI-1_Fig-11). Residues interacting with Curine via hydrogen bonds, hydrophobic interactions, and water bridges were with Y96, L134, D135, D137, A138, V350, and E351 (Fig-4b). On the other hand, Hayatinine was found to interact with G119, Q120, E133, M146, P147 and H152. The Hydrogen bonds, hydrophobic bonds, and water bridges with backbone and side chains of these amino acid residues persisting for more than 30 % of simulation time seem to prevent surface binder Hayatinine from moving out of its binding site. Interactions of MMV688271 (Fig-4d) with GAP50 were found to be stabilized through Hydrogen bonds (E123, H152, F344, L347, P348), water bridges (E123, H152, F344, L347), and hydrophobic interactions (H152, F344, L347, V350). MMV1634402 (Brilacidin) binding with GAP50 was found to be majorly driven by water bridges mediated via side chains and the backbone of N127, E129, N148, D201, I322, N349, E351, and L352 between ligand and protein 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 (Fig-4e). A large number (sixteen) of water bridges between Brilacidin and GAP50, may have contributed to the lowest – dG of binding -64.82 kcal/mol, with a KD 2.81 µM.
In vitro effects of the combination of lead molecules with antimalarial drugs on ART resistant P. falciparum
Drug combinations help in the assessment of synergistic, additive, or antagonistic activities. This, in turn, helps in the selection of combinations that retard the development of resistance, reduce toxicity, and increase the efficacy of the drugs used. Herein, we tested our lead molecules (Hayatinine, Bedaquiline, MMV688271, Curine, and MMV1634402) in combination with standard antimalarial drugs (Mefloquine, Pyrimethamine, and ART) for their activity against intraerythrocytic development of ART resistant P. falciparum. In different test molecule and antimalarial drug combinations, strong synergy (∑FIC < 1) was observed at a 1:4 molar ratio (Table-2). These results show the importance of our lead molecules in providing synergistic combinations for effective antimalarial therapy with a reduced probability of drug resistance.
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 (SI-1_Table-10) and till day 16 mice were healthy with respect to 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 the progressive increase in %Parasitemia with a median survival of 19 days and death of all 7 mice by day 22. These mice exhibited increased fluctuations in body temperature and a decrease in mean body weight with the progression of the disease. Interestingly, Bedaquiline (50 mg/kg b. wt.) resulted in the complete suppression of %Parasitemia by day 19, reduced fluctuations in temperature and body weight as compared to -ve control, and survival of three out of seven mice till 34 days (Fig-5). On the other hand, USINB4-124-8 (100 mg/kg b. wt.) showed no difference from -ve control in terms of survival, % rise in parasitemia, fluctuation in body weight, and temperature (Fig-5).