Inverse docking based screening and identification of protein targets for Cassiarin alkaloids against Plasmodium falciparum.

Various reports have shown Cassiarin alkaloids, selective in vitro activities against various strains of Plasmodium falciparum with low cytotoxicity, which indicates their possible candidature as antimalarial drug. However, poor recognition of their protein targets and molecular binding behaviour, certainly limits their exploration as antimalarial drug candidature. To address this, we utilises inverse screening, based on three different docking methodologies in order to find their most putative protein targets. In our study, we screened 1047 protein structures from protein data bank, which belongs to 147 different proteins. Our investigation identified 16 protein targets for Cassiarins. In few cases of identified protein targets, the binding site was poorly studied, which encouraged us to perform comparative sequence and structural studies with their homologous proteins, like as in case of Kelch motif associated protein, Armadillo repeats only protein and Methionine aminopeptidase 1b. In our study, we also found Tryptophanyl-tRNA synthetase and 1-Deoxy-D-Xylose-5-phosphate reductoisomerase proteins are the most common targets for Cassiarins.

Introduction

Malaria is a mosquito-borne infectious disease affecting humans and other animals caused by the protozoan parasite, Plasmodium. According to WHO 2015 statistics, 212 million clinical episodes and 429,000 deaths were reported worldwide (Bhatt et al. (2015); World_Health_Organization, 2015; Kamholz, 2016; World_Health_Organization, 2016) and nearly 3.2 billion people are at the risk of malaria, especially children under age of 5 years, pregnant women, immune compromised patients, as well as non-immune migrants (Schumacher and Spinelli, 2012; Negi, 2013; Wells et al., 2015). These large numbers are mainly subjected by Plasmodium falciparum (P. falciparum), followed by P. vivax, P. ovale, P. malariae, and to some extent P. knowlesi. Although in recent years, some profound development has been seen in antimalarial drug discovery, but higher number of resistance cases, mild to moderate selectivity/toxicity ratio of most of the antimalarial drugs, show a need of new scaffolds or new chemical entity (NCE) (Bushell et al., 2017). Moreover, the alkaloid natural product class has been found promising and useful in numerous disease states, as mentioned in these reports (Kayser et al., 2003; Frederich et al., 2008; Özçelik et al., 2011; Singla et al., 2013; Singla et al., 2014). Additionally, alkaloids, such as Quinine, Cryptolepine, Thiaplakortones A–D and their semi or synthetic derivatives (Caniato and Puricelli, 2003; Oliveira et al., 2009) are well studied as antimalarial agents (Cimanga et al., 1997; Davis et al., 2013), showing alkaloidal scaffold inheritance of antiplasmodial activity.

In recent years, various medicinal active natural compounds were reported from a plant, Cassia siamea (Leguminosae). Most of these natural compounds are either isolated from leaves (Cassiarin-A, B, G, H, J, K, 5-acetonyl-7-hydroxy-2-methylchromene, Chrobisiamone A) (Morita et al., 2007; Oshimi et al., 2008; Deguchi et al., 2012), or flower (Cassiarin C, D, E, F; 10,11-dihydroanhydrobarakol, anhydrobarakol Cassibiphenol A and Cassibiphenol B) (Thongsaard et al., 2001) (Deguchi et al., 2014), or also from twigs (Siamalkaloids A, B, C) (Wu et al., 2016), structures shown in Fig. 1. Surprisingly, Cassiarin-A and Cassiarin-B were found highly selective than Chloroquine against chloroquine sensitive P. falciparum 3D7 strain over human breast cancer cell lines (MCF7), as selectivity/cytotoxicity ratio are fairly high, ≥4348, ≥1112, 3281 for Cassiarin-A, B and Chloroquine, respectively (Morita et al., 2009). Furthermore, their antimalarial role was purposed though their vasorelaxation activity, as prompted by nitric oxide production from the endothelium, which might inhibit the host cell surface attachment of the parasite (Morita et al., 2009). In 2009, Oshimi et al. isolated Cassiarins C-E and 10,11-dihydroanhydrobarakol which showed reasonable in vitro selectivity against P. falciparum 3D7 over human leukaemia cells (HL-60 cell lines) (Oshimi et al., 2009).

Fig. 1

Chemical structures of compounds isolated from Cassia siamea (Leguminosae).

Chemistry

Isoquinoline is the basic alkaloidal core of Cassiarins, which fused with 2-methyl-2H-pyran ring at position [4, 8a], forms tricyclic ring and as prototype represented in the structure Cassiarin-C (shown in Fig. 1). Further derivatization at C position of Cassiarin-C, forms Cassiarin D, E and F. The methyl at C2 in the pyran ring of these isoforms (Cassiarin-C, D and E), can adopt 2 conformations as R or S. Every isoform has its own structure signature at C2 position, when compare to Cassiarin-C structure, which has simply a methyl group: (a) Cassiarin-D has –CH2– tethered 5-propenone-7-hydroxy-4H-chromen-4-one functionality at C2 with regards to Cassiarin-C, as shown in Fig. 1; (c) Cassiarin E is Bis-isomer of Cassiarin-C; (d) Cassiarin-F has fused with a toluene ring, to form a tetracyclic ring at position [2,3] of Cassiarin-C and also has further substructure extension in a form of 2-resorcinol propanone functionality, shown in Fig. 1.

In order, to characterize the molecular targets for these Cassiarin alkaloids, we used inverse docking, which is grown as a valuable tool in drug target identification in recent years. Also, helpful in rediscovering the molecular mechanism of polypharmacological active compounds, especially, the natural products and detecting, the possible adverse side targets of existing drugs as in toxicological studies. Previous reports on inverse docking shows implementation of various methodologies, to improve the accuracy and prioritizing the identified targets. Kumar et al. tried to address the limitations of docking scoring schemes with respect to attain confidence in theoretical binding affinities (Kumar et al., 2014). They presented a reverse approach, where they used the pharmacophore features of the ligand as interactions of complementary amino acids of protein cavities (also, called them as “pseudoreceptor”). These pseudoreceptors were then matched with the cavities/ binding sites of the selected protein dataset. They applied this approach on 3 co-crystallized ligands over 28 proteins of Zea mays and provide an application of the total probability and docking energy, in order to acquire confidence in prioritizing the probable protein targets (Kumar et al., 2014). Also, Carvalho et al. adopted a reverse screening strategy based on ligand similarity and target structure, which resulted into, a number of putative protein target candidates for quercetin polypharmacological effects and also successfully correlated them, with previously tested proteins, mainly protein kinases and poly [ADP-ribose] polymerases (Carvalho et al., 2017). In another report, Kumar et al. compared the rank list results from inverse docking and ligand-based similarity search, assist them to prioritize the chitinase as most probable target for kinetin molecule, further supported by experimental data (Kumar et al., 2015). While, few compiled literature reviews on inverse screening and its application are available, related to the drug repositioning (Kharkar et al., 2014) and available target databases/servers (Lee et al., 2016).

However, the selectivity/cytotoxicity profile of these reported Cassiarin alkaloids has been promising in P. falciparum but as their protein targets are poorly recognised, which certainly limits their further exploration as antimalarial candidature. To identify their protein targets and acquire significant confidence in prioritising the identified target, we used reverse screening on all available protein targets from protein data bank, using three different placement docking methods.

Materials & methods

Proteins set

All the protein targets for P. falciparum were searched on protein data bank, claiming 1047 structures. After filtering off the NMR and low resolved cryo-electron structures from X-ray structures, proteins were selected and arranged in the order of their crystal structure resolution as an individual target, see in Table 2. In most cases, preferences were given to co-crystallised ligand containing protein structures, otherwise the structures without co-crystallised ligand protein were also selected. Later, the self-docking on co-crystallise ligand containing protein targets, was performed to calculate the minimum RMSD values (min. RMSD values) in order, to evaluate the competency of a particular protein in accommodating of its own co-crystallise ligand (also, called ligandability) (Kumar, 2018). In those structures, which lack co-crystallise ligand, active site finder tool of MOE (Del Carpio et al., 1993; Negi et al., 2013a, Negi, 2013) was used to find the active surface patches which were saved as dummy atoms for performing the later docking. Also, in certain cases we aligned the target protein sequences with their homologous proteins of other species. These studies involved superposition of three-dimensional structure of the proteins of interest, as to see the overlapped domains and regions with comparative homologous proteins, which could be inferred into key active site residues in those proteins which were poorly studied in the past.

Table 2

The selected proteins with their PDB codes, resolution (Res.), co-crystallise ligand and self-docking RMSD values are provided in this table.

	Protein	PDB	Res. (Å)	Co-crystallized ligand	Min. RMSD values
1.	Dihydroorotate dehydrogenase	4CQ8	1.98	5-(4-Cyano-2-methyl-1H-benzimidazol-1-yl)-N-cyclopropylthiophene-2-carboxamide	0.406
2.	Triosephosphate Isomerase	1O5X	1.10	Naf
3.	PfA-M1	3T8V	1.80	N-[(2-{2-[(N-{(2S,3R)-3-amino-4-[4-(benzyloxy)-phenyl]-2-hydroxybutanoyl}-L-alanyl)amino]ethoxy}ethoxy)acetyl]-4-benzoyl-L-phenylalanyl-N6-hex-5-ynoyllysinamide	1.354
		4ZW3	1.80	Tert-Butyl [(1S)-1-(4-bromophenyl)-2-(hydroxyamino)-2-oxoethyl]-carbamate	1.080
		4X2U	1.50	Tosedostat	0.511
		4K5L	1.91	Phosphonic Arginine	1.160
		4R5X	1.85	3-amino-N-{(1R)-2-(hydroxyamino)-2-oxo-1-[4-(1H-pyrazol-1-yl)-phenyl]-ethyl}-benzamide	1.492
		3EBH	1.65	2-(3-Amino-2-hydroxy-4-phenyl-butyrylamino)-4-methyl-pentanoic acid	0.850
4.	PfA-M17	4ZY2	2.10	N-[(1R)-2-(hydroxyamino)-2-oxo-1-(3′,4′,5′-trifluorobiphenyl-4-yl)-ethyl]-2,2-dimethylpropanamide	1.001
		4X2T	2.73	Tosedostat	0.701
		3KR4	2.00	2-(3-Amino-2-hydroxy-4-phenyl-butyrylamino)-4-methyl-pentanoic acid	0.396
		4R76	2.50	3-Amino-N-{(1R)-2-(hydroxyamino)-2-oxo-1-[4-(1H-pyrazol-1-yl)-phenyl]-ethyl}-benzamide	0.612
		5CBM	2.30	(2S)-2-{[(R)-[(R)-amino(phenyl)-methyl]-(hydroxy)phosphoryl]-methyl}-4-methylpentanoic acid	0.807
		4K3N	2.00	{(R)-amino[4-(1H-pyrazol-1-yl)-phenyl]-methyl}-phosphonic acid	1.208
		3T8W	2.00	N-((2R,3S,6S,18S,21S)-2-amino-18-(4-benzoylbenzyl)-21-carbamoyl-3-hydroxy-6-(naphthalen-2-ylmethyl)-4,7,16,19-tetraoxo-1-phenyl-11,14-dioxa-5,8,17,20-tetraazapentacosan-25-yl)-hex-5-ynamide	1.409
		3Q43	1.8	N-[(2S,3R)-3-amino-2-hydroxy-4-(4-methoxyphenyl)-butanoyl]-L-leucine	0.815
5.	GMP synthetase	3UOW	2.72	Xanthosine-5′-monophosphate	0.704
6.	Enoyl-ACP Reductase	3LT0b	1.96	4-(2,4-dichlorophenoxy)-3-hydroxybenzaldehyde	0.560
		1ZXB	2.68	3-Chloro-4-(4-chloro-2-hydroxyphenoxy)-N-methylbenzamide	0.788
		1V35	2.50	Nicotinamide adenine dinucleotide (NADH)	0.818
		4IGEb	2.15	7-(4-Chloro-2-hydroxyphenoxy)-4-methyl-2H-chromen-2-one	0.551
		2NQ8	2.50	Isoniazid-Nicotinamide adenine dinucleotide (INH-NAD)	0.914
		1NHW	2.35	2-(2,4-Dichloro-phenylamino)-phenol	0.907
		2FOI	2.50	4-(2,4-Dichlorophenoxy)-2′-methylbiphenyl-3-ol	1.020
7.	Beta-Hydroxyacyl-Acyl Carrier Protein Dehydratase	3AZB	2.60	5-chloro-8-[(3-chlorobenzyl)-oxy]-quinoline	0.611
8.	Prolyl-tRNA synthetase	4OLF	2.90	Halofuginone	1.270
		4WI1	1.65	1-(4-fluorophenyl)-3-[4-(4-fluorophenyl)-1-methyl-3-(trifluoromethyl)-1H-pyrazol-5-yl]-urea	0.501
9.	PfHAD1	4ZEV	1.80	Mannose-6-Phosphate	0.323
10.	Spermidine synthetase	4BP3	1.75	Decarboxylated S-adenosyl-methionine	0.901
		3RIE	1.90	5′-deoxy-5′-methylthioadenosine	0.818
		2I7C	1.71	S-adenosyl-1,8-diamino-3-thiooctane	0.450
		2HTE	2.00	5′-deoxy-5′-methylthioadenosine	0.512
11.	CDPK4	4QOX	2.75	3-(3-bromobenzyl)-1-tert-butyl-1H-pyrazolo-[3,4-d]pyrimidin-4-amine	0.717
12.	CDPK2	4MVF	2.00	Staurosporine	0.825
13.	HSP90	3PEH	2.75	2-amino-4-{2,4-dichloro-5-[2-(diethylamino)-ethoxy]-phenyl}-N-ethylthieno[2,3-d]-pyrimidine-6-carboxamide	0.309
		3PEJ	2.81	Macbecin	0.532
		3K60	2.30	Adenosine diphosphate (ADP)	0.599
		3IED	2.01	Adenylyl-Imidodiphosphate (AMPPN)	1.402
14.	Fructose-1,6-bisphosphate aldolase	4TR9	2.11	N′-[(E)-(2,4-dichlorophenyl)-methylidene]-3,4-dihydroxybenzohydrazide	0.373
15.	Ferredoxin-NADP+ reductase	2OK8	2.40	Nicotinamide adenine dinucleotide phosphate (NADP)	1.007
		2OK7	2.70	Adenosine-2′-5′-Diphosphate (2‘-ADP)	0.641
16.	Dihydrofolate reductase-thymidylate synthase (PfDHFR-TS)	1J3I	2.33	WR99210	0.513
		3UM8	2.6	Cycloguanil	0.998
		3QGT	2.3	Pyrimethamine	1.604
		3DGA	2.7	N-[2-chloro-5-(trifluoromethyl)-phenyl]imidodicarbonimidic diamide	1.200
		4DPD	2.5	Dihydrofolic acid	0.888
17.	D-aminoacyl-tRNA deacylase (DTD)	4NBI	1.86	3′-deoxy-3′-(D-tyrosylamino)-adenosine	0.711
		3LMV	2.83	4-(2-Hydroxyethyl)-1-piperazine ethanesulfonic acid	1.302
		3KO3	2.09	Adenosine triphosphate (ATP)	1.009
		3KO5	2.8	ADP	0.977
18.	FK506	4QT3	1.40	Rapamycin	0.221
		4J4N	2.75	N-(2-ethylphenyl)-2-(3H-imidazo[4,5-b]pyridin-2-ylsulfanyl)acetamide	0.708
		2VN1	2.35	8-Deethyl-8-[but-3-enyl]-ascomycin	0.890
19.	Lactate dehydrogenase	1T24	1.70	4-Hydroxy-1,2,5-oxadiazole-3-carboxylic acid	0.901
		1XIV	1.70	2-({4-Chloro-2-[hydroxy(methoxy)methyl]cyclohexyl}amino)ethane-1,1,2-triol	0.990
		1CET	2.05	Chloroquinein€	1.313
		1LDG	1.74	NADH	1.607
		4B7U	1.88	Bicine	0.780
20.	PfPK5	1V0O	1.90	Indirubin-5-sulphonate	0.449
21.	Plasmepsin II	1LF2	1.80	3-amino-N-{4-[2-(2,6-dimethyl-phenoxy)-acetylamino]-3-hydroxy-1-isobutyl-5-phenyl-pentyl}-benzamide	0.760
		4CKU	1.85	5-[1,1-bis(oxidanylidene)-1,2-thiazinan-2-yl]-N3-[(2S,3R)-4-[2-(3-methoxyphenyl)propan-2-ylamino]-3-oxidanyl-1-phenyl-butan-2-yl]-N1,N1-dipropyl-benzene-1,3-dicarboxamide	0.831
		2BJU	1.56	N-(R-carboxy-ethyl)-α-(S)-(2-phenylethyl)	0.451
		2IGX	1.70	5-Pentyl-N-{[4′-(piperidin-1-yl-carbonyl)biphenyl-4-yl]methyl}-N-[1-(pyridin-2-ylmethyl)piperidin-4-Yl]pyridine-2-carboxamide	1.222
22.	Plasmepsin I	3QS1	3.10	KNI-10,006	0.619
23.	Plasmepsin IV	1LS5	2.80	Pepstatin A	0.870
24.	Phosphoglycerate Kinase	1LTK	3.00	AMP	0.391
25.	Glutathione reductase	1ONF	2.60	Flavin Adenine Dinucleotide (FAD)	0.904
26.	Thymidylate Kinase	2YOG	1.50	1-[4-Chloranyl-3-(trifluoromethyl)-phenyl]-3-[[(2R,3S)-5-[5-methyl-2,4-bis-(oxidanylidene)pyrimidin-1-yl]-3-oxidanyl-oxolan-2-yl]methyl]thiourea	0.60
		2WWF	1.89	ADP	0.832
27.	Ubiquitin Carboxyl-Terminal Hydrolase 3 (Uchl3)	2WDT	2.30	Naf
28.	Purine Nucleoside Phosphorylase	2BSX	2.00	Inosine	0.456
		1Q1G	2.02	3,4-Dihydroxy-2-[(methylsulfanyl)methyl]-5-(4-oxo-4,5-dihydro-3H-pyrrolo [3, 2-d]pyrimidin-7-yl)pyrrolidinium	0.702
29.	Histo-Aspartic Protease (Hap)	3QVI	2.50	KNI-10,395	0.315
30.	Purine Phosphoribosyltransferase	1CJB	2.00	(1S)-1(9-deazahypoxanthin-9-yl)1,4-dideoxy-1,4-imino-D-ribitol-5-phosphate	0.611
		3OZG	1.99	[(3S)-4-Hydroxy-3-{[(4-oxo-4,5-dihydro-3H-pyrrolo[3,2-d]pyrimidin-7-yl)methyl]amino}butyl]phosphonic acid	0.942
31.	Peptide Deformylase	1RL4	2.18	2-{N′-[2-(5-amino-1-phenylcarbamoyl-pentylcarbamoyl)-hexyl]-hydrazinomethyl}1.4-hexanoic acid-(5-amino-1-phenylcarbamoyl-pentyl)-amide	0.664
32.	Cyclophilin	1QNG	2.10	Cyclosporin A	0.782
33.	Glutathione-S-Transferase	4ZXG	1.70	Ligandin	0.560
34.	Glyceraldehyde-3-Phosphate Dehydrogenase	1YWG	2.60	Nicotinamide Adenine Dinucleotide (NAD)	1.104
35.	Ribose 5-phosphate isomerase	2F8M	2.09	Naf
36.	MTIP	4R1E	1.98	5-{[(2-aminoethyl)-sulfanyl]methyl}furan-2-carbaldehyde	0.309
37.	Guanylate Kinase	1Z6G	2.18	4-(2-Hydroxyethyl)-1-piperazine ethanesulfonic acid	0.655
38.	ARO (armadillo repeats only protein)	5EWP	1.80	Naf
39.	cGMP-dependent protein kinase	4OFG	2.0	Cyclic Guanosine Monophosphate (cGMP)	0.831
40.	Apical membrane antigen 1	4Z0E	1.9	Naf
		4R19	1.8	Naf
41	DXR	5JAZ	1.4	[(2R)-2-{2-[hydroxy(methyl)amino]-2-oxoethyl}-5-(naphthalen-1-yl)pentyl]phosphonic acid	0.561
		4Y67	1.6	[(2R)-2-{2-[hydroxy(methyl)amino]-2-oxoethyl}lo-5-phenylpentyl]phosphonic acid	0.410
		4KP7	2.0	[(S)-({2-[hydroxy(methyl)amino]-2-oxoethyl}sulfanyl)(phenyl)methyl]phosphonic acid	0.701
		3WQR	1.97	[(1S)-4-[hydroxy(methyl)amino]-1-(4-methoxyphenyl)-4-oxobutyl]phosphonic acid	0.809
		4GAE	2.30	[(1S)-3-[acetyl(hydroxy)amino]-1-(pyridin-4-yl)propyl]phosphonic acide	0.677
				[(1R)-3-[acetyl(hydroxy)amino]-1-(pyridin-4-yl)propyl]phosphonic acide	0.719
		3AU9	1.90	3-[formyl(hydroxy)amino]propylphosphonic acid	0.884
42.	kelch protein	4YY8	1.81	Naf
43.	Pf41	4YS4	2.45	Naf
44.	Pf12	2YMO	1.90	Naf
45.	PfGrx1	4MZB	1.04	3-[N-morpholino]-propane sulfonic acid	0.698
46.	Tryptophanyl-tRNA synthetase	4JFA	2.60	Tryptophan	1.019
		4J75	2.40	Tryptophanyl-5′-AMP	0.719
47.	Tyrosyl-tRNA Synthetase	3VGJ	2.21	Tyrosyl-AMP	1.102
48.	Phosphoethanolamine Methyltransferase	4FGZ	1.99	Amodiaquine	0.359
		3UJB	1.52	S-adenosyl-L-homocysteine	0.902
49.	MIF	2WKF	2.05	Naf
50.	Peroxiredoxin	4D73	1.80	Naf
51.	Glycerol-3-Phosphate Dehydrogenase	1YJ8	2.85	Naf
52.	6-Pyruvoyl Tetrahydropterin Synthase (PTPS)	1Y13	2.20	Biopterin	0.772
53.	Phosphoglycerate Mutase	3EOZ	2.40	Naf
54.	Diadenosine Triphosphate Hydrolase	5CS2	1.65	Cyclomarin A	0.490
55.	Aspartate Transcarbamoylase	5ILQ	2.50	Naf
56.	Plasmodium Falciparum Rab6	1D5C	2.30	Guanosine Diphosphate (GDP)	0.705
57.	Acyl-COA Binding Protein	1HBK	2.00	Coenzyme A	0.667
58.	Pfg27	1N81	2.10	Naf
59.	Fab Complex Whith Plasmodium Falciparum Msp1-19	1OB1	2.90	Naf
60.	Adenylosuccinate synthetase	1P9B	2.00	6-Phosphoryl-Inosine Monophosphate	0.443
61.	Nucleoside diphosphate kinase B	1XIQ	3.05	Naf
62.	D-Ribulose 5-Phosphate 3-Epimerase	1TQX	2.00	Naf
63.	GTPase Rab6	1D5C	2.30	GDP	0.548
64.	GTPase Rab11	3BFK	1.80	GDP	1.001
65.	Rab5 protein	3CLV	1.89	GDP	0.670
66.	RabGDI	3P1W	1.85	Naf
67.	Orotidine 5′-Monophosphate Decarboxylase	3VI2	2.10	4-(2-hydroxy-4-methoxyphenyl)-4-oxobutanoic acid	0.487
		2Q8Z	1.8	6-Amino-UMP	0.861
		3S9Y	1.7	6-amino-5-fluorouridine 5′-(dihydrogen phosphate)	0.914
		2ZA1	2.65	Orotidine 5′-monophosphate	0.503
68.	Oxoacyl-Acp Reductase	2C07	1.50	Naf
69.	ClpP protease catalytic domain from Plasmodium falciparum	2F6I	2.45	Naf
70.	Glutamate Dehydrogenase	2BMA	2.70	Naf
71	glutamate dehydrogenase 2	3R3J	3.10	Naf
72.	PHIST	4JLE	2.35	Naf
73.	PF3D7_0823300 (GCN5)a,c	4QNS	1.50	Naf
	Histone acetyltransferase GCN5	5TPX	2.10	(1S, 2S)-N1, N1-dimethyl-N2-(3-methyl-[1,2,4]triazolo[3,4-a]phthalazin-6-yl)-1-phenylpropane-1,2-diamine	0.799
74.	PFA0510w (Bromodomain protein)a	4PY6	2.50	Naf
75.	PF3D7_1475600	4NXJ	2.18	Naf
76.	PF10_0328	3FKM	2.50	Naf
77.	Ubiquitin conjugating enzyme UBC9	4M1N	1.50	Naf
78.	Ubiquitin conjugating enzyme E2	2H2Y	2.80	Naf
79.	Ubiquitin carrier protein	2R0J	1.85	Naf
80.	Ubiquitin conjugating enzyme e2a	2Q0V	2.40	Naf
81.	PF10_0330 (Ubiquitin-conjugating enzyme)a	2ONU	2.38	Naf
82.	Falcilysin (protein)	3S5M	1.55	Naf
83.	Calcium-dependent protein kinase 3	3 K21	1.15	Naf
84.	Calcium-dependent protein kinasea	3MSE	2.10	Naf
85.	Pyruvate kinase	3KHD	2.70	Naf
86.	Calcium-dependent protein kinase 2	3PM8	2.00	Naf
87.	ADP-Ribosylation Factor 1	3LRP	2.50	GDP	0.758
88.	Aspartate Aminotransferase	3K7Y	2.80	Pyridoxal phosphate (PLP)	1.009
89.	PFC0360w protein (HSP90 Activator protein)	3NI8	2.50	Naf
90.	MAP-2 kinase	3NIE	2.30	Phosphoaminophosphonic acid-adenylate ester	0.417
91.	Serine/threonine kinase-1	3LLT	2.50	Phosphoaminophosphonic acid-adenylate ester	0.857
92.	Ornithine delta-aminotransferase	3LG0	2.30	Naf
93.	Aha-1	3N72	1.77	Naf
94.	Arginase	3MMR	2.14	2(S)-amino-6-boronohexanoic acid	0.637
		3SL1	1.90	6-(dihydroxyboranyl)-2-methyl-L-norleucine	1.377
95.	Malarial Clpb2 Atpase/Hsp101 Protein	4IRF	1.65	Naf
96.	ClpB protein (Green fluorescent protein)a	4XBI	1.80	Naf
97.	Maltose-binding periplasmic protein	4O2X	2.70	Naf
98.	Aquaglyceroporin	3C02	2.05	β-Octylglucoside	1.110
99.	Profilin	2JKG	1.89	Naf
100.	Microtubule-associated protein 1 light chain 3	4EOY	2.22	Naf
101.	Thrombospondin related anonymous protein	4F1J	1.73	Naf
102.	Apicoplast TIC22a	4E6Z	2.15	Naf
103.	Diphenyl Nucleoside	3T64	1.65	2′, 5′-Dideoxy-5′-[(diphenylmethyl)amino]-uridine	0.598
		2Y8C	2.10	5′-Tritylated Deoxyuridine Analogue	0.617
104.	Erythrocyte Binding Antigen Region II Region 175	4K2U	2.25	Naf
		1ZRO	2.25	α-2,3-sialyllactose	0.896
		1VYQ	2.40	2,3-deoxy-3-fluoro-5-O-trityluridine	1.006
105.	Erythrocyte Membrane Protein 1	3CPZ	2.80	Naf
106.	Erythrocyte Binding Antigen 140	4JNO	3.00	O-sialic acid	1.017
107.	Dbl6 Epsilon Domain (VAR2CSA)	2Y8D	1.84	Naf
		2XU0	2.06	Naf
108.	MSPDBL2	3VUU	2.09	Naf
109.	Erythrocyte Membrane Protein-1 (PfEMP1) variant 2 of strain MC	3C64	2.40	Naf
110.	2C-methyl-D-erythritol-2,4-cyclodiphosphate synthase (IspF)	4C81	1.56	Cytidine-5′-diphosphate	1.308
111.	Thioredoxin reductase	4J56	2.37	FAD	1.100
112.	Thioredoxin-2	4O32	2.20	Naf
113.	Thioredoxin Peroxidase 2	2C0D	1.78	Naf
114.	Peroxiredoxin	1XIY	1.80	Naf
115.	Thioredoxin like proteina	3CXG	2.00	Naf
116.	Thioredoxin like proteina	1SYR	2.95	Naf
117.	Peptidasea	5JR6	2.30	Naf
118	ATP-dependent Clp protease	4GM2	2.80	Naf
119	GAP50	3TGH	1.70	Naf
120.	Tumor protein (TCTP)	3P3K	2.55	Naf
121.	ARF GTPase activating protein	3SUB	2.40	Naf
122.	Sir2A	3U31	2.20	NAD	0.709
123.	Ser/Thr protein kinase	2PMO	2.90	Hymenialdisine	0.993
124.	Adenylate Kinase	3TLX	2.75	ADP	0.989
		1CJB	2.00	(1S)-1(9-Deazahypoxanthin-9yl)1,4-dideoxy-1,4-imino-d-ribitol-5-phosphate	0.640
		2VFA	2.80	Guanosine Monophosphate (GMP)	1.082
125.	Apicomplexan AP2 protein	3IGM	2.40	Naf
126.	Nucleosome assembly protein 1a	3FS3	2.30	Naf
127.	Nucleosome assembly protein	3KYP	2.80	Naf
128.	Falcipain (Isoform-2)	3BPF	2.90	N-[N-[1-Hydroxycarboxyethyl-carbonyl]leucylamino-butyl]-guanidine	0.613
	(Isoform-3)d	3BWK	2.43	N-2-(Morpholin-4-ylcarbonyl)-N-[(3S)-1-phenyl-5-(phenylsulfonyl)pentan-3-yl]-L-leucinamide	0.706
129.	Glycerol Kinase	2W41	2.41	ADP	0.83
130.	Malaria Sporozoite Protein Uis3a	2VWA	2.50	Phosphatidylethanolamine	1.502
131.	EBA-175 region VI	2RJI	1.80	Naf
132.	Pyrroline carboxylate reductase	2RCY	2.30	NADP	0.906
133.	Phosphatidylethanolamine-Binding Protein	2R77	1.60	Naf
134.	Internal Kinesin	1RY6	1.60	Naf
135.	Dynein Light Chain 1	1YO3	1.65	Naf
136.	Malarial Hypothetical protein	1ZSO	2.17	Naf
137.	Adenosyl-homocysteinase	1V8B	2.40	NAD	1.205
138.	Fe-Superoxide Dismutase	2GOJ	2.00	Naf
139.	Ribosomal RNA Methyltransferasea	2PLW	1.70	S-Adenosyl-L-homocysteine (SAM)	0.666
140.	Protein-L-isoaspartate O-methyltransferase β-aspartate methyltransferase	2PBF	2.00	SAM	0.480
141.	Dimethyladenosine transferasea	2H1R	1.89	Naf
142.	Plasmodial PLP Synthase	2ABW	1.62	Tetraethylene glycol	0.922
143.	Actin Depolymerizing Factor	3Q2B	1.60	D(-)-tartaric acid	0.801
144.	Glucose-6-phosphate isomerase	3PR3	2.45	Fructose-6-phosphate	0.442
145.	Methionine aminopeptidase	3S6B	1.95	Naf
146.	Nucleolar GTP-binding protein 1a	2QU8	2.01	GDP	0.575
147.	Orotate Phosphoribosyl transferase	4FYM	2.60	Naf

Putative.

Both have close self-docking RMSD.

Pairwise Sequence alignment shows PF3D7_0823300 (GCN5) ¥ is a Histone acetyltransferase (GCN5).

It is isoform-3 of Falcipain and has identity of ∼66%.

co-crystallise ligand is in racemic mixture.

Na: no co-crystal ligand. Res. = Resolution of crystal structure.

Ligand set

As absolute stereochemistry at C2 position of Cassiarins is unknown, therefore we build both (R) and (S) stereoisomers, which were further minimised by MMFF94x Forcefield. Although, the energy minimisation step showed a reasonable energy difference between both the stereoisomer forms of individual Cassiarins (C, D, E & DBH), but these were used as such in our molecular modelling studies, as to avoid any pseudo positive or misleading results.

Molecular modelling

The proteins were prepared by, (a) removing of the water molecules from their crystal structures; (b) modelling the missing or breaks in their loops; and (c) protonation of the structure. Later, the co-crystallise ligand binding site or saved dummy atoms on proteins were used for docking of the Cassiarins. This inverse screening was performed by utilising 2 docking placement methods (also called, “Differential placement method based docking”). The first was the alpha triangle placement method, which generates the ligand-protein poses based on the overlapping of ligand atom triplets onto the triplets of protein point sites (are, also called alpha sphere centres). At each iteration cycle, a pose was determine based on sampling of a random triplet of ligand atoms over a random triplet of alpha sphere centres. The following setting was used for this method: minimum and maximum iterations cycles were set 800,000 and 5,000,000 respectively with timeout (6000 s). The second was triangle matcher placement method, which generate the ligand-protein poses by aligning the ligand atoms triplet on triplets of alpha spheres in a more systematic way than in the Alpha Triangle method (method setting was, total number of returned poses was set 100,000, with time out 3000 s). Lastly, we utilised another approach which was different than the previous docking methods, i.e. grid based docking, as a part of VlifeMDS suite tool (VLife, 2010) which uses the genetic algorithm for grid formation and docking. Later, we compare the docking binding affinities, resulted from these methods, to get the confidence in prioritising the most putative Cassiarin targets.

Result and discussions

Protein set and docking

All the 1047 proteins were retrieved from protein data bank, which were found to belongs to 147 different protein-types. This protein dataset was divided into 2 categorises: one which has co-crystallise ligands and other one, without co-crystallise ligands. In case of co-crystallise ligand containing protein-types, we considered all those structures for a protein where co-crystallise ligand has a diverse chemotype in its structure. Later, we performed the self-docking to filter the most suitable protein crystal structure based on the min. RMSD value for its own co-crystallise ligand. In those protein-type, where protein structure does not contain any co-crystallise ligand, the structures were chosen based on their resolution (Res.). In order, to find the active site on those structure which does not contain co-crystallise ligand, active site finder tool was used to identify the active patches for the docking. Later, alpha triangle and triangle matcher placement methods resulted in various docking poses, which were ranked by GBVI/WSA dG scoring function (results for alpha triangle and triangle matcher are provided in Supplementary information, Table 1 and Table 2 respectively). While, the grid based docking results are enlisted in Supplementary information as Table 3. The grid based docking was performed on a larger area (80 × 80 × 80), as increased size of sheared active cavity results more conspicuous differences in the docking energies which could be useful in separating closely related putative targets.

Table 1

Enlisting the in vitro antiplasmodial and cytotoxicity activities of natural compounds isolated from Cassia siamea plant.

Alkaloids	Plasmodium facliparum (IC50 = µM)	Cytotoxicity (µM)	Reference
Chloroquine	0.011a	36.1c	Morita et al. (2009)
Cassiarin A	0.023a, 0.005b	>100c, 35e	Morita et al., 2009, Morita et al., 2007
Cassiarin B	22.0a, 6.9b	>100c, >100e	Morita et al., 2009, Morita et al., 2007
Cassiarin C	24.2a	>100d	Oshimi et al. (2009)
Cassiarin D	3.6a	>100d	Oshimi et al. (2009)
Cassiarin E	7.3a	>100d	Oshimi et al. (2009)
10,11-dihydroanhydrobarakol (DHB)	2.3a	>100d	Oshimi et al. (2009)
Anhydrobarakol (ANH)	4.7a, 7.8a	>100d	Oshimi et al. (2009, Oshimi et al., 2009, Oshimi et al., 2008)
5-acetonyl-7-hydroxy-2-methylchromone (AHMC)	8.6a, 4.5a	>100d	Oshimi et al. (2009, Oshimi et al., 2009, Oshimi et al., 2008)
Chrobisiamone A	2.6a	–	Oshimi et al. (2008)
Cassiarin F	3.3b	>50d	Deguchi et al. (2011)

Chloroquine-sensitive P. falciparum strain 3D7.

P. falciparum 3D7.

MCF7 (human breast adenocarcinoma) cell line.

HL-60 Human blood premyelocytic leukaemia.

P388 mouse leukaemia cells.

Table 3

Energy profile of stereoisomers of Cassiarins isoforms.

	R-Cassiarins C	S-Cassiarins C	R-Cassiarins D	S-Cassiarins D	R-Cassiarins E	S-Cassiarins E	R-DHB	S-DHB	Cassiarins F
Stretch (EStr)	0.8104	0.8429	2.9064	2.9917	1.9958	2.0769	0.9461	0.9461	2.4203
Bend (EB)	2.7384	3.0206	9.1569	9.3554	6.0665	6.2631	2.8678	2.8678	7.9539
Stretch-Bend (EStr-B)	0.0803	0.1062	0.3036	0.3219	0.1787	0.1957	−0.0318	−0.0318	−0.0135
Torsion (Etor)	−8.4023	−7.9008	−12.7333	−12.5845	−17.7278	−17.5754	−4.5516	−4.5516	0.0139
Non-1,4 VDW (EnVDW)	−3.9005	−3.9103	−5.2953	−5.4962	−8.6554	−8.8469	−3.4075	−3.4075	−12.0544
1,4 VDW (EVDW)	15.9754	15.8132	30.3389	30.4761	32.8734	33.0109	16.7451	16.7451	28.1042
Dipole/Dipole (Ed-d)	0.2167	0.2191	−1.3802	−1.2855	−0.3444	−0.3346	−0.1735	−0.1735	−0.4288
Total Energy (E = kcal/mol)	7.5183	8.1907	23.2970	23.7788	14.3868	14.7896	12.3946	12.3946	25.9957

Total Energy (MM2) = E + E + E + E + E + E + E.

The energy minimisation step revealed the most stable conformation among the isomers of individual Cassiarins, see in Table 3. In general, we found that R-stereoisomers have more energy minimised structures than their S-isoforms and DHB stereoisomers appeared to be unaffected with the C2-stereochemistry. The large structure containing isoforms show more energy penalties than their smaller isoforms, as follow: F > D > E > DHB > C, which seems reasonably obvious, as these have more steric hindrance in their structures. Also, we found certain isoforms consist similar structural connectivity but shows significant differences in their energy levels (C vs DHB; E vs D), indicating that the subtle alteration, like replacing of nitrogen as from N1-isoquinoline by oxygen, diminishes the aromatic character, which increases the cyclic ring constrain.

Based on the binding energies, resulted from three different docking placement methodologies, we considered and compare top 25 most energy minimised Cassiarin-protein complexes and selected only those ones which were present in the result of all three docking methods (Table 4 in Supplementary information). Based on these observations, we found 16 targets as for individual Cassiarin stereoisomer, listed in Table 4. While Table 5, summarises the basic structural features of these targets. The binding mode of Cassiarins with these protein targets has been discussed descriptively in the later sections of this article.

Table 4

The most common PDBs as targets were identified for Cassiarins isoforms/isomers.

R-Cassiarins C	S-Cassiarins C	R-Cassiarins D	S-Cassiarins D	R-Cassiarins E	S-Cassiarins E	R-DHB	S-DHB	Cassiarins F
4J75	4J75	2W41	5EWP	3UOW	3LG0	3VGJ	1P9B	4j75
3VGJ	3VGJ	3PR3	2W41	3FS3		4Y67	3FS3	4Y67
3MMR	3MMR	3VUU	4Y67	2W41		1P9B	3MMR
4Y67	4YY8			4YY8		4J75	4J75
				2C07		3S6B	4Y67
				3LT0		3PR3

Table 5

Summary of individual protein target.

PDB	Protein/Target	Co-crystallise ligand	Structural features and functions
4J75	Tryptophanyl-tRNA synthetase	Trptophanyl-Adenosine monophosphate (WAMP)	• Utilises tryptophan as substrate • Contain 632 amino acids • Has KMSKS conserved loop and AIDQ motif in its active binding site
3VGJ	Tyrosyl-tRNA synthetase	Tyrosyl-Adenosine monophosphate (YAMP)	• Utilises tyrosine as substrate • Catalytic domain contains 18-260 amino acid residues • Has KMSKS conserved loop and AIDQ motif in its binding site
4Y67	1-Deoxy-D-Xylose-5-phosphate reductoisomerase	Fosmidomycin	• Homodimer in its active form • Each monomer is made up of two large domains • One large domain for NADPH • Other one for catalysis • FOSMIDOMYCIN inhibitors are most studied class
2W41	Glycerol kinase	Adenine Diphosphate	• Utilises glycerol as substrate • Participate in rate-limiting step in glycerol utilisation • Has 501 amino acid residues • Contains two domains separated by a deep cleft • Domain I (regions 1–262 and 436–471) • Domain II (regions 263–435 and 472–501) • Domain 1 is for glycerol binding; Domain II is for ADP binding. • Its kinase domain shows typical H-bond acceptor–donor triad interactions
3MMR	Plasmodium falciparum Arginase	2-(S)-amino-6-boronohexanoic acid	• Utilises binuclear manganese • Forms tetrahedral geometry with both the manganese atoms. • Close to human arginase I (28%) and II (27%)
3FS3	Nucleosome Assembly Protein	n. a	• 347amino acid long-dimer • Has two domains, domain I (mainly contain, dimerization helix-2, region started from 37 to 87) • Domain II, containing multiple α-helices and a subdomain containing four antiparallel β-strands (residues 128–185)
1P9B	Adenylosuccinate synthetase	GDP and IMP	• Consists of 19 strands (β1–β19), 12 α-helices and seven 310 helices • It has nine parallel β-strands, with a tenth antiparallel strand (β19) form a central sheet. • Has 4 subdomains • Subdomains III (residues 278–302) majorly constitute ligand binding pocket • It has 2 binding sites: orthosteric (IMP binding site) and allosteric (GTP binding site).
3PR3	Glucose-6-Phosphate Isomerase	Fructose-6-phosphate	• Orthosteric site contains key amino acids, Glu380, Lys540, Thr233, Lsy232, Thr236, Ser231 and Gly158 • Like, human glucose-6-phosphate isomerase (HsG6PI), it also has two globular domains (as large and small domains) and an “arm-like” C-terminal tail
4YY8	Kelch Motif Associated Protein of Plasmodium Falciparum	Mono-alkylated p-substituted sulphonamides	• Made up of 28 β-strands where β4 to β28 involved in the formation of its 6 kelch motifs • Every motif contains 4 β-strands in common, • 6 kelch motifs together built a propeller architecture
3LG0	Ornithine δ-aminotransferase of Plasmodium falciparum (PfOAT)	n. a	• Enzymatically active as a homodimer • High percentage of conserved residues in the active cavity, which is proximal to the interface between two subunits • Each subunit contains a pyridoxal-phosphate (PLP) binding domain and a substrate binding domain • PLP binding loop domain (region started from 287 to 293) and strictly conserved in all species
3LT0	Enoyl Acyl Carrier Protein Reductase	Triclosan and NADPH	• With its most studied class of inhibitor (triclosan derivatives), it shows formation of typically ternary complex of PfENR-NAD+-triclosan • Ring A of Triclosan binds to hydrophobic pocket and has π-stacking interaction distance with nicotinamide ring of the cofactor NAD+
2C07	Oxoacyl Acyl-Carrier-Protein Reductase	Triclosan and NADPH	• Most studies limited to triclosan and its derivatives. • Our comparative structural studies with E. coli found 4 key mutations: Ser99 (Gly41 in E. coli), Ser94 (Ala36 in E. coli), Arg95 (Thr37 in E. coli) and Ser198 (Gly137 in E. coli).
5EWP	Armadillo Repeats Only Protein of Plasmodium Falciparum	n. a	• 252 amino acid residues long dimer • Contains 15 α-helixes, form a characteristic alpha solenoid structure • Each Armadillo repeat is composed of a pair of alpha helices that form a hairpin structure
3S6B	Methionine Aminopeptidase 1b	n. a	• Comparative studies with human homologous protein: Thr156 and Ser268 mutated in place of Pro192 and Cys301 (in human),
3UOW	Guanosine monophosphate synthetase	Xanthose Monophosphate (XMP)	• It is dimeric in nature. • Each monomer is composed of two catalytic domains, an N-terminal independent GATase (1–236) and a C-terminal ATPPase domain (237–555) • Its dimer form is highly required for its activity as the interface has 108C-terminal residues of the ATPPase domain. • In this interface, two cis-prolines (Pro548–Pro549) allow a tetrahedral configuration of Asp543, Thr551, Glu553 and Arg539
3VUU	Merozoite surface proteins have erythrocyte-binding Duffy Binding Like Domains (MSPDBL2)	n. a	• Assist parasite for its initial binding to the surface receptors on the host red blood cell. • Consist of a boomerang shaped α-helical core formed from three subdomains • Subdomain 1 (region 161–225) only contains a 5-residue long α-helix (helix 1) provide a junction for subdomains 2 and 3

n.a. no crystallise protein.

Tryptophanyl-tRNA synthetase

The cytosolic tryptophanyl-tRNA synthetase of Plasmodium falciparum (Pf-cTrpRNA, PDB: 4J75, Res.2.4 Å, (Koh et al., 2013)) belongs to aminoacyl tRNA synthetase (aaRSs) class, which charges amino acids to their cognate tRNAs during protein synthesis and requires a large conformational change during their functioning. Previous studies on bacterial and human tRNA synthetase revealed key structural features: (a) In bacteria, these have open, ligand-free state (F-state) where either Trptophan (Trp) or ATP can bind; (b) Simultaneous binding of Trp and ATP in the pre-transition state requires a conserved loop, KMSKS (492K493M494S495S496T in P. falciparum and 349K350M351S352A353S in humans) (Datt and Sharma, 2014) to close onto the active site and C-terminal domain moves toward the active site, containing Rossmann-fold domain; (c) After the intermediate tryptophanyl-adenylate (WAMP) formation, both the KMSKS loop and C-terminal domain move slightly away from the catalytic core to allow to release of the product (called, as P-state) (Datt and Sharma, 2014) (d) While in human cytoplasmic TrpRS (Hs-cTrpRS), the binding of Trp is mainly accompanied by the N-terminus and a conserved AIDQ motif. Phylogenetically, Pf-cTrpRS is more close to Hs-cTrpRS (∼44% identity) than the human mitochondrial TrpRS (∼16% identity) (Koh et al., 2013).

According to literature (Datt and Sharma, 2014), the WAMP binds to the 28 key amino acid residues of Pf-cTrpRNA (highlighted as bold single letter amino acid code): (i) 296YTGR300G and 317HXGHXI323P in the tip of a loop between β3 and α5 (ii) 341QXSXXEK (iii) 415YXXX419Q (iv) 450VPQGXD456QXXX460F (v) 481VF483M (vi) 492K493M, as shown in Fig. 2(A) and as follow: (a) The indole and adenine ring has π-π interactions with Tyr306, Phe482; (b) The free NH2 and N1-heteroatom of adenine ring has H-bond interaction with backbone of Met483; (c) The hydroxyl (OH) groups of ribose sugar shows H-bonding with β-COOH of Asp455 and backbone of Glu452; (d) The phosphate head lies within H-bond distance with side chain of Arg309 and backbone of Gly310; (e) The NH2 tethered functionality of tryptophan interacts with Gln429. Like adenine ring of WAMP, tetracyclic core of Cassiarin-F (Orange) faces vertically to the β-strand region extended from 481Val to 496Thr and α5-helix from 318Leu to 334Phe and also, has T-shaped π-π interactions with Phe482 (3.89 Å) and His320 (3.23 Å) (Sinnokrot and Sherrill, 2004), see in Fig. 2(B). The N1-isoquinoline of tetracyclic ring of Cassiarin-F faces towards a cavity consist β-strand (region from 448Cys to 453Gly), 453GID456Q conserved motif, α11-helix (region from 456Q to 470M) and has H-bond acceptor interaction with NH2 terminus of the side chain of Gln456 (3.02 Å). The resorcinol phenolic-OH inclined towards the β4-strand (region from 337Pro to 342Leu), α-helix (343Ser to 350Phe), showing H-bond donor interaction with γ-COOH acid side chain of Glu346 (2.43 Å). The extended propanone functionality aligned with β3-strand (304Tyr to 309Arg), has H-bond acceptor interaction with guanidine side chain of Arg309 (1.95 Å) and amide backbone of Gly310 (2.09 Å). While the tricyclic ring of R-Cassiarin-C shows reverse orientation: (a) phenolic ring fitted into a pocket surrounded by α11 (region from 456Q to 470M) and conserved motif GIDQ, showing H-bond donor interactions with alcoholic side chain of Thr307 (2.39 Å; β3 strand) and amide backbone of Pro451 (2.39 Å; β6: 448Cys to 453Gly), while its N1-isoquinoline ring projected outwards, as shown in Fig. 2(C). Although, the S-isomer of Cassiarin-C attained a reasonable conformational binding change, as phenolic OH projected towards the α6-helix and involved with γ-COOH group of Glu346 (2.32 Å), but its N1-isoquinoline ring has H-bond acceptor interaction with Thr307 (2.76 Å) of β-strand and utilisation of conserved motif GIDQ, shows similar binding pocket like R-isomer-C, shown in Fig. 2(D). The heteroatom (N1) replacement with oxygen atom in Cassiarin-C, brings R-DHB, which restricted the phenolic ring, aromatic character due to formation of quinone ring and also affected its binding with Pf-cTrpRNA. However, its binding resembles to the 2-resorcinol propanone substructure of Cassiarin F binding, as the carbonyl (CO) of quinone utilises ε-NH2 of Lys347 (2.48 Å) of β6-strand and have proximity with the conserved motif GIDQ, α11, α6 and β3, as shown in Fig. 2(E). Although, as it is devoid of tetracyclic ring like in Cassiarin-F therefore the interactions with of β7 region and α5 got disappeared. The S-DHB has shown a close identity in its binding pattern like R-DHB, which would therefore similarly mimic interactions as like 2-resorcinol propanone of Cassiarin F, with slight variation in its binding conformation as resulted by the modification of C2-stereochemistry, see in Fig. 2(F). The close binding behaviour of both isomers of DBH with regards to the 2-resorcinol propanone of Cassiarin-F, offers a possible bioisostere substitution on tetracyclic core of Cassiarin-F in order, to improve structure based rational design against Pf-cTrpRNA protein in P. falciparum. Additionally, we observed that S-DHB binding within the π-stacking interaction to the Tyr425 residue, which is deleted in human aminoacyl tRNA synthetase proteins and can be utilised as a structural feature in optimisation of NCE.

Fig. 2

Illustrating the interaction of the various ligands with tRNA synthetase: (A) Binding of WAMP (B) Binding of Cassiarin F (Orange) (C) Binding of R-Cassiarin C (Blue), (D) Binding of S-Cassiarin C (Violet), (E) Binding of R-DHB (Cyan); (F) Binding of S-DHB (Brown).

Tyrosyl-tRNA synthetase

Tyrosyl-tRNA synthetase (PDB: 3VGJ, Res. 2.21 Å, (Bhatt et al., 2011) belongs to the aminoacyl tRNA synthetase family proteins and therefore have similar function and catalytic motifs as seen in previous case of Pf-cTrpRNA, except its utilisation of tyrosine at the place of tryptophan, in the form of tyrosyl-AMP. It consists of a catalytic domain region started from residues 18–260, contains KMSKS and GIDQ conserved motifs. Also, as obvious, its nucleoside binding pocket interactions are similar like Pf-cTrpRNA (Fig. 3(A)), as (a) adenosine ring fits in the cavity constituted by Lys247, Leu238, Met 237, Gly236 and Met248 of complimentary K247M248SKS conserve domain (Fig. 3(B)); (b) The His70, Ala72 and Gln73 which lies at the tip of loop between α1-helix/β3-stand and Leu206, Asp209, Gln2010 of evolutionary conserved residues 207GI209D2010Q create its sugar binding pocket; (c) Asp61, Phe63, Glu64 encloses the sugar-phosphate junction and phosphate head; (d) The Gln192, Asp195 located on the top of ionised NH2 of tyrosine where the Trp94, Ala96, Ph99 composed hydrophobic pocket in order to accommodate aromatic phenol ring.

Fig. 3

(A) Superposition of active site domain of Tyrosyl-tRNA synthetase (secondary structure ribbon colour; grey colour code for co-crystallise ligand) and Pf-cTrpRNA (secondary structure in green colour; brown colour code for co-crystallise ligand); (B) interactive mode of co-crystallise ligand with Tyrosyl-tRNA synthetase.

The molecular docking of R-Cassiarin-C has shown H-bond donor interactions with Thr307 and Gln452 through its phenolic (OH) group, as shown in Fig. 4(A). As compared to R-isomer, S-Cassiarin-C shows more suitability for AMP binding pocket, as: (a) its phenolic (OH) orientated similarly like phosphate heads and has H-bond acceptor-donor interactions with side chain of Lys77 (2.72 Å) and free COOH terminus of Arg61 (2.67 Å), (b) The pyridine ring of its isoquinoline core binds like adenosine ring of AMP as it utilises His70 (T-shaped π-π interaction, 3.59 Å) and Lys247 (H-bond acceptor interaction, 2.67 Å) of KMSKS conserved motif, shown in Fig. 4(B) (Sinnokrot and Sherrill, 2004). Similar to S-Cassiarin-C, the binding conformation of R-DHB also utilises AMP binding pocket, as: (a) quinone has H-bond acceptor interaction with Gly207 backbone (2.03 Å); (b) The 2-methyl-2H-pyran ring mimic the adenosine ring binding region through H-bond acceptor interaction with the side chain of Lys247(2.02 Å), see in Fig. 4(C).

Fig. 4

Molecular binding poses: (A) R-isomer of Cassiarins-C (blue) shows its utilises the GIDQ conserved domain (presence of Asp209, Gln210); (B) S-isomer (violet) of Cassiarins-C binds to adenosine monophosphate cavity; (C) R-DHB (cyan).

1-Deoxy-D-xylose-5-phosphate reductoisomerase (DXR)

DXR (PDB: 4Y67, Res. 1.6 Å, (Chofor et al., 2015)) is a class B dehydrogenase enzyme, which exists as a homodimer in its active form (the active region started from Lys75 to Ser488) where each monomer is made up of two large domains separated by a cleft containing a deep pocket, a linker region, and a small C-terminal domain (Chofor et al., 2015). One of the large domain is responsible for NADPH binding (region started from 77 to 230), and the other domain is for catalysis (contains, metal and substrate binding, region started from 231 to 369). The catalytic domain is an α/β-type structure, consisting of five α-helices (α7–α11) and four β-strands (β8–β11) and have two different conformations, open and closed. The open conformation assist the substrate D-xylose-5-phosphate (DXP) to enter and binds to the active site (Mac Sweeney et al., 2005; Umeda et al., 2011). On the other hand, the NADPH binding site, contain conserved residues (Asp231, Glu233, Ser269, Ser270, Trp296, Met298, Ser306, Asn311, Lys312, and Glu315), which are also conserved in all human malaria parasites (Yajima et al., 2007; Kunfermann et al., 2013).

However, the most studied inhibitor class, fosmidomycin and its analogues bind in a typical fashion to DXR protein (Chofor et al., 2015), which can further categorised into three regions (a) phosphate head (PO4) binding region, which has tight H-bond interaction network with Ser270, Asn311, two water molecules, and His293; (b) hydrophobic carbon backbone binding region, which runs parallel to the indole ring of Trp296 and interacts with Met298; (c) hydroxamate binding pocket, which coordinated in cis-arrangement with metal ion (Mg2+ or Mn2+), negatively charged residues (Asp231, Glu233, and Glu315) and forms a typical distorted trigonal bipyramidal geometry for these class of compounds (Fig. 5(A)).

Fig. 5

Interactive binding mode: (A) Co-crystallise fosmidomycin derivative (gold); (B) Cassiarin F (Orange); (C) R-Cassiarin-C (Blue).

The molecular docking of Cassiarin-F shows its resorcinol phenolic functionality utilises majorly hydroxamate binding region, as characterised by its H-bond interaction with the backbone of Ser269 (3.18 Å), His341 (2.65 Å) and side chain (OH) of Ser232 (2.66 Å), as shown in Fig. 5(B). Also, its hydrophobic tetracyclic ring aligns parallel to the Trp296 and Met360, as same like the hydrophobic carbon backbone of the fosmidomycin analogues. Similarly, like Cassiarin-F, the R-Cassiarin-C phenol ring fits into the hydroxamate binding region as through the H-bond acceptor/donor interactions with free –NH2 side chain of Lys205 and –COOH functionality of Asp231 respectively. While, the isoquinoline ring of R-Cassiarin-C participates in π-π interaction with Trp296 (3.86 Å & 4.45 Å), which shows its ring orientation different from the F-isoform, in Fig. 5(C). This indicates the pivotal role of additional aryl ring system (i.e. the extended propanone-resorcinol structure of Cassiarin-F), is a key difference for the such binding conformation of Cassiarin-F when compared to the binding conformation of R-Cassiarin-C.

Structurally, Cassiarin-C and Cassiarin-D are only different at C2-position substitution with 5-propenone-7-hydroxy-4H-chromen-4-one. The molecular docking of S-Cassiarin-D shows the 5-propenone functionality imitate like phosphate head and binds to Ser270 (side chain (1.61 Å) and backbone (1.92 Å)), backbone of Gly271 (2.54) and side chain (OH) of Ser269 (2.37) via H-bond acceptor interactions. While, the non-aromatic quinone interact with NH2-terminal of Glu233 (2.83 Å) through H-bond acceptor interaction. It appears that C2 extension from Cassiarin-C to Cassiarin-D pushed the tricyclic isoquinoline core more towards the NADPH binding pocket region (as provided with the presence of Ser88 and Ile89 from the of NADPH binding site) and, also aligned with the hydrophobic backbone patch (presence of Trp296 and Met360), shown in Fig. 6(A). However, H-bond acceptor interaction of N1-isoquinoline with NH-indole side chain of Trp296 (1.90 Å), shows the tricyclic ring tossed up from the cavity, which could be interesting to observe in case of co-binding of NADPH as its adenine ring would be close within 4.5 Å distance for π-π interaction with isoquinoline ring of S-Cassiarin-D. While in case of R-DHB, the quinone ring binds to phosphate head region via H-bond acceptor interaction with side chain of Ser270 (3.08 Å), backbone of Ser269 (2.03 Å) and NH2-terminus of side chain of Lys312 (2.34 Å), in Fig. 6(B). Additionally, the incorported Oxygen atom of 2H-dihydropyran core of R-DHB shares a H-bond acceptor interactions with side chain of Ser232. Also, it has been found interesting that the interactive mode of R-DBH couldn’t find characteristic hydrophobic backbone binding residues like Trp296 and Met360 within 4.5 Å. However, in case of molecular binding of S-DHB, quinone ring interacts (NH2 terminal of Lys 205 (2.63 Å) and Asn311 (2.89 Å)), via H-bond acceptor interactions, similarly like quinone ring of R-isomer, shown in Fig. 6(C) and Fig. 6(B), respectively, but its C2 associated pyran ring proximity to the Trp296 (hydrophobic interactions, 3.58 Å and 4.05 Å) and Met360, shows the critical role of C2 stereochemistry in their binding to the DXR protein, as it was not seen in R-isomer. We also found these Cassiarins binding lacks the utilisation of hydroxamate cavity of DXR protein and hence unable to coordinate with Mn2+ ions and to form typical trigonal bipyramidal geometry, as seen in fosmidomycin class inhibitors.

Fig. 6

Binding mode: (A) S-Cassiarin-D (green); (B) R-DHB (cyan); (C) S-DHB (brown).

Glycerol kinase (PfGK)

PfGK (PDB: 2 W41, Res. 2.41 Å, (Schnick et al., 2009)) phosphorylate the glycerol, which is a rate-limiting step in glycerol utilisation in parasite metabolism (Schnick et al., 2009; Naidoo, 2013). Deletion of host gene shows no effect on gametocyte development, suggesting that these life cycle stages do not utilize host-derived glycerol as a carbon source (Schnick et al., 2009). The structural architecture of PfGK contains 501 amino acid residues, arranged in two domains separated by a deep cleft. Each domain is constructed around a α/β core (characterise as βββαβαβα) that is characteristic of the sugar kinase/Hsp70/actin superfamily proteins (Bork et al., 1992; Hurley, 1996). The Domain I (regions 1–262 and 436–471) comprises β3β2β1α1β5α6β12α9 and domain II (regions 263–435 and 472–501) consists of β16β14β13α12β19α13β20α14 (Schnick et al., 2009).

However, domain 1 is responsible for glycerol binding and domain II is for ADP binding. The adenine base of ADP slipped into a pocket of domain II and the ribose-phosphate functionality pointed towards the interdomain cleft. The nucleoside (adenine-sugar) shows typical, kinase domain triad interactions (H-bond acceptor/donor/acceptor interactions (Negi et al., 2013b)) with kinase domain (Asn418, Ser332 and Gly313) of PfGK. The phosphate binding pocket surrounded by Gly267, Thr268, Gly413 and Gly414 are profoundly required in the ternary closed complex with ATP. While, sugar binding pocket constituted by residues like Gly313, Ser314 and Met415 (Hurley et al., 1993; Schnick et al., 2009), see in Fig. 7(A).

Fig. 7

Interactive mode (A) co-crystallise ADP (gold); (B) R-Cassiarin D (purple); (C) S-isomer of Cassiarin-D (green); (D) R-Cassiarin-E (pink).

The phenolic (OH) and propanone functionalities of 5-propenone-7-hydroxy-4H-chromen-4-one of R-Cassiarin-D utilises triad interaction like adenine, with backbone of Asn418 (2.35 Å) and side chain of Ser332 (2.26 Å), as shown in Fig. 7(B). Although, S-isomer of Cassiarin-D also interacted with Asn418 (2.15 Å) in a similar fashion like R-isomer, but as because C2 stereochemistry changes in both isomer, the remaining halves of both isomers oriented differently, as R-isomer extended towards Leu320, over Thr268 while S-isomers twisted in L-shaped into the vicinity of Thr268, shown in Fig. 7(C). However, the presence of Pro328 and Met415 parallel to the C2 tethered backbone in both stereoisomers binding conformations, showed their utilisation of sugar binding region. Also, R-Cassiarin-E follows the same trend as like Cassiarin-D isomers binding with kinase region, as its phenolic OH involved in H-bond donor interactions with backbone of Asn418 (2.17 Å). However, R-Cassiarin-E doesn’t have 5-propenone-7-hydroxy-4H-chromen-4-one functionality like Cassiarin-D isomers, but has bis-tricyclic system, which encloses most of the Cassiarin-D isomers binding cavity residues, Fig. 7(D).

Plasmodium falciparum Arginase (PfAI)

PfAI (PDB: 3MMR, Res. 2.14 Å, (Dowling et al., 2010)) has close resemblance with its human homologous proteins, human arginases I (HsAI, 28%) and II (HsAII, 27%) and also utilises binuclear manganese (Müller et al., 2005). The interactive mode of co-crystallise ligand (2(S)-amino-6-boronohexanoic acid, ABH) of the PfAI illustrated the key important residues of the active site, as it has H-bond interactions with Glu368, Asp274, Ser229, Asn222 and forms tetrahedral geometry with both the manganese atoms (Wells et al., 2009; Dowling et al., 2010), see in Fig. 8(A).

Fig. 8

Interactive binding mode: (A) co-crystallise ABH (gold); (B) R-Cassiarin-C (blue); (C) S-Cassiarin-C (violet); (D) S-DHB (brown).

The molecular docking of these Cassiarins shows their utilisation of the ABH binding cavity. Whereas, the R/S isomer of Cassiarin-C flipped their orientations, displaying the influence of their C2 stereochemistry and relatively small molecular size with regards to the cavity size, shown in Fig. 8(B) and (C), respectively. Their flipping in orientation can be further understood based on their interactions as the phenolic (OH) and O-pyran ring of R-Cassiarin-C has H-bond acceptor interactions with His218 (2.77 Å) and Thr337 (2.80 Å) while, the phenolic (OH) of S-Cassiarin-C has H-bond donor interactions with Asp274 (2.24 Å) and Glu277 (2.57 Å) and its O-pyran ring has H-bond acceptor with Asn222 (2.44 Å). On the other side, the S-DHB shows H-bond acceptor interactions with side chain of Ser229 (2.99 Å) and Asn222 (2.08 Å), as shown in Fig. 8(D). These isoforms binding utilises only one manganese metal ion for coordination in their 4.5 Å, which is irrespective to the conventional inhibitors as their binding utilises two manganese atoms. This point could be useful in developing a hepatic antimalarial drug in future, as PfAI has been critical for malarial parasites during their liver stage development.

Nucleosome assembly protein

P. falciparum contains two nucleosome assembly proteins termed PfNapL and PfNapS (Chandra et al., 2005). PfNapL (PDB: 3FS3, Res. 2.3 Å, (Gill et al., 2009)) is a 347-amino acid dimer, cytoplasmic localised protein and has a central core of ∼250 residues that are thought to be responsible for histone binding. PfNap composed of two domains, domain-I (consists, dimerization helix-2, region started from 37 to 87) and domain-II, containing multiple α-helices and a subdomain containing four antiparallel β-strands (amino acid residues 128–185) (Gill et al., 2009). The dimerization helix-2 of domain-I forms the distinguishing shape of PfNapL, where two backbone helices cluster in an antiparallel manner to form the dimer using mainly hydrophobic interactions and salt bridges or hydrogen bonds (Gill et al., 2009).

The R-Cassiarin-E binds in a V-shaped, as shown in Fig. 9(A), where both aromatic cores go inside to the cavity composed of α2 (Arg68, His72 and Tyr 75) of dimerization helix-2, α7 (Tyr272, Pro271, Lys266 are the residues involved), Gly145 and Phe146 at the tip of β2, and Met169, Val179 are in the loop in between β3 and β4, and exposes its CH2-tethered backbone to the surface, as seen in Fig. 9(A). Whereas, its O-dihyropyran ring and phenolic OH interacts with NH terminus of Lys266 (3.20 Å) and NH-imidazole side chain of His72 (2.28 Å), respectively. While, in case of S-DHB, the pyran ring faces towards Cys133, Tyr79, encloses into a cavity composed of Lys266, Glu267, Ile270 and Pro271 on one side and Arg68, His72 and Tyr135 from other side (see, in Fig. 9(B)).

Fig. 9

Interactive binding mode: (A) R-Cassiarin-E (pink); (B) S-DHB (brown).

Adenylosuccinate synthetase (PfAdSS)

Each subunit of PfAdSS (PDB: 1P9B, Res. 2.0 Å (Eaazhisai et al., 2004)) consists of 19 strands (β1-β19), 12 α-helices, seven 310 helices and 6 loops (L1-5). Nine parallel β-strands (β9, β7, β5, β2, β10, β1, β11, β15 and β18) along with a tenth antiparallel strand (β19) forms a central sheet. This sheet is bordered by four subdomains: (a) subdomain-I (residues 54–65) comprises of only two β-strands (β3 and β4); (b) subdomain-II (residues 114–206) mainly involved in interface interactions; (c) subdomains-III (residues 278–302) majorly constitute ligand binding pocket; (d) subdomain-IV (residues 339–418). It has 2 binding sites: orthosteric (IMP binding site) and allosteric (GTP binding site). Both sites are close to each other. The previous studies highlighted various structural features and key residues of active site, as summarised here, (Eaazhisai et al., 2004): (a) Lys31 (which is a conserved residue in active site); (b) Lys62 (forms H-bonds with ribose hydroxyls in PfAdSS but absent in the other homologous AdSS proteins); (c) Lys29 is highly involved in phosphate head binding of GDP and shares a typical H-Bond character (Low Barrier Hydrogen Bond, LBHB, (Cleland and Kreevoy, 1994)); (d) phosphate binding pocket majorly constituted by the residues, like Asp26, Lys29, Gly53, His54 and Asn232; (e) Asp26 is believed to be a key residue which gets protonated and later coordinated to Mg2+ (Choe et al., 1999; Iancu et al., 2002). (f) Asn232, which interacts with IMP, similarly present in the E. coli and mouse AdSS complexes, (g) LBHB interaction of His54 and O2 of 6-phosphoryl of IMP (2.54 Å), is parallel to the mouse synthetase complex (2.46 Å) (Cleland and Kreevoy, 1994; Iancu et al., 2002).

The molecular docking of R-DHB shows, its pyran ring utilise the NH-guanidine side chain of Arg313 of β13 (2.16 Å & 2.30 Å) and alcoholic (OH) side chain of Thr307 of most conserved segment of loop L5 (2.62 Å) via H-bond acceptor interactions, see in Fig. 10(A). However, its binding pocket shows conserved helix α1(G28L29G30K31), L5 region containing {(H303Y305T307), β13 (R313) and L6 (P428) of the PfAdSS. However, S-DHB also shows similar binding like R-DHB to the pocket, like α1(L29G30K31), β3 (H54), L5 (E304T307), β13(R313). While, its quinone ring and 2H-pyran ring has H-bond acceptor interaction with NH of backbone of Gly30 (2.15 Å) and NH side chain terminus of Lys339 (2.44 Å), as seen in Fig. 10(B).

Fig. 10

Interactive mode: (A) R-DHB (cyan); (B) S-DHB (brown).

Glucose-6-Phosphate isomerase (PfG6PI)

As no further structural information for PfG6PI protein (PDB: 3PR3, Res. 2.45 Å, (Gileadi et al., 2011)) was available, hence we evaluated its own co-crystallise ligand (fructose-6-phosphate) binding, to allocate the key residues in its active site. The binding shows its phosphate head has H-bond network with Ser159, Ser231, Lys232, Thr233 and Thr236, while polar heads of fructose sugar has H-bond interaction with Gly158, Glu380 and Lys540, see in Fig. 11(A). Furthermore, compared with human protein (HsG6PI, PDB:1JLH, Res. 2.1 Å (Cordeiro et al., 2003)), Pf-G6PI found to have two globular domains (as one, large and other, small domains) and an “arm-like” C-terminal tail, similar like HsG6PI of humans. Both the large and the small domain have a central core of a β-pleated sheet flanked by α-helices to form a typical α/β folding motif. The large domain contains 6 β-strands (β1:40I to 42K; β2: 46F to 52R; β7:357N to 362P; β8:400V to 402F; β9:425V to 430F; β10: 495S to 500F) and small domain has 4 β-strands (β3: 150N to 154I, β4: 201N to 205L; β5: 225T to 230I; β6: 264M to 267V).

Fig. 11

(A) co-crystallise fructose-6-phoaphate (gold); (B) R-DHB (cyan); (C) R-Cassiarin-D (purple).

The molecular docking of R-DHB shows its binding complimentary to the fructose-6-phosphate as the presence of residue 156–159 and 231–239 shows the phosphate binding pocket of fructose-6-phosphate, which is situated in between β3/4 and β5/6, respectively, as shown in Fig. 11(B). While, R-Cassiarin-D shows similar binding orientation to the cavity, situated in β5/6: Lys232, Thr233, Thr236, flanked α-helix (Gly293, Arg294), β7/8 (Gln376, Glu380), C-terminal tail (Lys540) and enclosed within H-bond distance with charged side chains of Glu380 (2.09 Å), Arg294 (2.33 Å) and backbone of Gly293 (2.92 Å), shown in Fig. 11(C).

Kelch motif associated protein of Plasmodium falciparum (PfKEAP)

It is BTB domain containing 389 residues long dimer (PDB: 4YY8, Res. 1.81 Å, (Jiang et al., 2015), constituted by 28 β-strands where β4 to β28 involved in the formation of its 6 kelch motifs: Every motif contains 4 β-strands in common, except motif 5, which has unusual 6 β strands (: βL444-I448, β460-464, β469-472, βSer720-Ala724; β484-489, β491-495, β508-512, β517-520; β532-536, β539-543, β555-559, β564-567; β579-583, β586-590, β17601-605, β18610-613; β622-624 β627-630, β633-637, β640-642, β650-654, β659-663; β674-678, β682-685, β696-700, β705-709). These 6 kelch motifs together built a propeller architecture, shown in Fig. 12(B). Further comparative sequence and structure studies with human (HsKEAP, PDB: 4XMB, Res. 2.43 Å) (Jain et al., 2015) and mouse (MmKEAP, PDB: 4ZY3, Res. 1.80 Å) (Saito et al., 2016) kelch motif containing proteins, show key residues associated with the orthosteric binding site, Fig. 12(A).

Fig. 12

(A) Multiple sequence alignment (MSA) with human (HsKEAP1) and mouse (MmKEAP1) kelch proteins; (B) Superpose of human (yellow) and P. falciparum (brown) kelch protein; (C) Human kelch protein co-crystallise ligand (brown) utilising the similar cavity of P. falciparum kelch protein (yellow).

The co-crystallise ligand (mono-alkylated p-substituted sulphonamides) of HsKEAP1 fits in between cavity surrounded by kelch repeats K3/4/5: F451Y456Y482N498R529 N530Y546I551S576S577T593G595E596R597S623S624 of PfKEAP, shown in Fig. 12(C). The cross docking of co-crystallise ligand of hsKEAP1 on PfKEAP, shows distinctive kelch motif features in PfKEAP as compare to the kelch motifs of human proteins, as further supported by minimum RMSD value (4.84 Å) and free energy (−7.86) for PfKEAP than minimum RMSD (1.07 Å) and free energy (−7.34) for HsKEAP. While in our observation, we found R-Cassiarin-E (−15.2877) and S-Cassiarin-C (−12.22) are more profoundly forming energy-stable complexes with PfKEAP. Also, their interactive mode has similar binding pattern as –OH groups of R-Cassiarin-E interacts with the backbone of region through H-bond acceptor-donor interactions, as shown in Fig. 13(A). While, the other half, fits in the hydrophobic cavity constituted by aromatic amino acids. On the other side, the S-Cassiarin-C uses multi kelch motifs as compared to R-Cassiarin-E ( (β7), (β15), (β19 β20), (β25), (β29)), for its binding to PfKEAP via H-bond acceptor/donor interaction with Ser720 (2.67 Å) and Phe674 (2.41 Å) respectively, as shown in Fig. 13(B).

Fig. 13

(A) MSA of PfOAT with TsOAT and HsOAT. (B) Interactive domain of S-Cassiarin-E (yellow).

Ornithine δ-aminotransferase of Plasmodium falciparum (PfOAT)

PfOAT (PDB:3LG0, Res. 2.3 Å, (Jortzik et al., 2010)) is active as a homodimer. Based on the comparative sequence alignment with other OATS (Human: HsOAT, PDB: 2OAT, Res. 1.95 Å (Storici et al., 1999); Toxoplasma: TsOAT, PDB: 5E3K, Res. 1.73 Å (Filippova et al., 2016)) (shown in Fig. 14(A)), we observed high percentage of conserved residues in the active cavity, which is close to the interface of two subunits. Each subunit contains a pyridoxal-phosphate (PLP) binding domain and a substrate binding domain (Jortzik et al., 2010). The PLP binding loop domain (region started from 287 to 293) is strictly conserved in all species. The S-Cassiarin-E binds significantly with the PLP-binding loop domain via H-bond donor interaction with amide backbone of Pro286 (2.13 Å) and His289 (2.23 Å); π stacking interaction with imidazole ring of His289 (4.24 Å). Furthermore, N1-isoquinoline (2H) has H-bond acceptor interaction with NH-guanidine side chain of Arg83 (2.10 Å). While the 2H-isoquinoline ring folded towards a hydrophobic cavity (comprises V106L107M108M109) which allows the Cassiarin-E to undergo the specific binding conformation, shown in Fig. 14(B).

Fig. 14

Interactive binding mode: (A) R-Cassiarin-E (pink); (B) S-Cassiarin-C (violet).

Enoyl acyl carrier protein reductase (PfENR)

Previous studies on the protein, enoyl acyl carrier protein reductase obtained from different origins (P. falciparum, E. coli, B. napus, M. tuberculosis, H. pylori) show overall identical structural homology (Pidugu et al., 2004). This analysis also provides the key features, related to the substrate binding loop region, which were further correlated with the affinities of its conventional inhibitor class, Triclosan derivatives (Belluti et al., 2013). The Triclosan derivatives are primarily contain the Biphenyl ether scaffold (Ring A and Ring B separated by an oxygen atom). In case of PfENR, the binding of Triclosan (PDB: 3LT0, Res. 1.96 Å, (Maity et al., 2010)) typically forms a ternary complex as PfENR-NAD+-Triclosan, where ring A of Triclosan settles into a hydrophobic pocket (composed of Tyr277, Tyr267, Gly313, Pro314, Ile323, Phe368, Ile369, and Ala372) and has π-stacking interaction distance with nicotinamide ring of the cofactor NAD+ (Maity et al., 2010), in Fig. 15(A). While, Ring B of triclosan has close proximity with ribose-phosphate functionalities of NAD+, substrate-binding loop residues (like Ala319, Ala320, and Ile323) and a conserved loop (containing Ala217, Asn218, Ala219, and Val222) (Pidugu et al., 2004).

Fig. 15

Interactive binding mode (A) co-crystallise ligand (gold); (B) R-Cassiarin-E (pink).

The molecular modelling of Cassiarin-E advocates the binding to the co-factor binding site irrespective to the expected substrate binding site. The tricyclic ring of R-Cassiarin-E forms the sandwich-type π-π interactions with indoyl moiety of Trp131 (3.82 Å, 4.08 Å, 4.58 Å, 3.73 Å, 3.89 Å & 4.29 Å). While, the remaining part of the molecule twisted towards the ribose-phosphate sugar pocket of NADH, which was also a binding pocket of Ring B of Triclosan, see in Fig. 15(A). This tricyclic ring of this twisted half also has H-bond donor/acceptor interaction with Asp107 (2.34 Å) and NH backbone of Ala217 (2.90 Å) of cavity domains (comprising G106D107N109G110 and S317R318A319) on one side and A217N218 on other side, respectively, as shown in Fig. 15(B).

Oxoacyl acyl-carrier-protein reductase (FabG)

Most of the interactive domain information for 3-oxoacyl acyl-carrier-protein reductase (PDB:2C07, Res. 1.50 Å, (Wickramasinghe et al., 2006)) was limited to Triclosan orthosteric site inhibition, while co-factor binding was highly underestimated. Therefore, we explore the co-factor binding site (NADPH binding site) with 3-oxoacyl acyl-carrier-protein reductase of E. coli (PDB: 1Q7B, Res. 2.05 Å, (Price et al., 2004)), found RMSD of their backbone (1.455 Å, for 237 amino acids), identity (47.1%) and similarity (68.0%) and with 4 key mutations, as indicated, in Fig. 16(A): Ser99 (Gly41 in E. coli), Ser94 (Ala36 in E. coli), Arg95 (Thr37 in E. coli) and Ser198 (Gly137 in E. coli). The molecular modelling studies revealed the binding of R-isomer of Cassiarin-E, majorly utilises the binding cavity of nicotinamide functionality of NADP, in Fig. 16(B). However, its H-bond donor/acceptor interaction with backbone (–NH) of Ser98 (2.68 Å) and phenolic (OH) of Tyr212 (2.01 Å) side chain, evident its binding to the phosphate binding cavity of NADP. The tricyclic aromatic ring of R-Cassiarin-E has T-shaped π-π interaction of Phe244 (3.55 Å & 3.79 Å) (Sinnokrot and Sherrill, 2004), as shown in Fig. 16(B).

Fig. 16

(A) Superpose of FabG protein of P. falciparum (residues labelled in brown) with E. coli utilising identical ligand (triclosan derivative) binding orthosteric site; (B) Interactive mode of R-isomer of Cassiarin-E.

Armadillo repeats only protein of Plasmodium falciparum (PfARO)

PfARO is poor studied protein (PDB: 5EWP, Res.1.8 Å (Peifer et al., 1994; Brown et al., 2016)), therefore we compare its structure with the truncated structure of cell adhesion protein of Caenorhabditis elegans (PDB:4R11, Res. 2.79 Å) (Choi et al., 2015), see in Fig. 17(A). Our investigation found PfARO is 252 amino acid residues long, right handed super helix dimer of 15 α-helixes, which forms a characteristic alpha solenoid structure (Peifer et al., 1994). However, each Armadillo repeat is composed of a pair of alpha helices that form a hairpin structure (involving alpha-helixes: α137-48, α251-57, α381-93, α4100-107, α5110-118, α6123-136, α7141-149, α8153-159, α9165-179, α10183-191, α11194-200, α12215-228, α13236-243, α14247-254, α15260-273; β-strands: β1229-230 and β2233-234). The molecular docking of S-Cassiarin-D isomer shows its binding dependent on the turn, as enclosed by the interface of α2/3 region (containing A67A70H72P73W74A75A76D77). Also, its OH group shows H-bond donor interaction with His72 (2.57 Å) and Asp77 (2.02 Å) with the mentioned interface of α2/3 region, in Fig. 17(B).

Fig. 17

(A) Superpose with cell adhesion protein of Caenorhabditis elegans (grey colour backbone); (B) Interactive mode of S-Cassiarin D (green).

Methionine aminopeptidase 1b (pfMetAP)

As no structural information related to the pfMetAP (PDB: 3S6B, Res. 1.95 Å, (Wernimont et al., 2011a, Wernimont et al., 2011b)) was disclosed by the previous studies, therefore we performed comparative studies of its structure with its human homologous protein (HsMetAP, PDB: 2G6P, Res. 1.9 Å, human methionine aminopeptidase Type 1) (Hu et al., 2006). The superimposition of pfMetAP and HsMetAP shows the coverage of 81%, RMSD of their backbone (1.253 Å, for 301 amino acids) and identity (54%), as structures shown in Fig. 18(A). Although, the binding of co-crystallise ligand with pfMetAP, shows high resemblance in their orthosteric site, with subtle mutation that could be exploited for selective drug designing and targeting against P. falciparum in future, as follow: Thr156 and Ser268 mutated in place of Pro192 and Cys301 (in human), respectively provide H-bond donor/acceptor interaction, see in Fig. 18(A). However, the tricyclic ring of R-DHB binds in the hydrophobic core (containing T156Y159F162C167H176 & H270F276H277 W320) and its quinone functionality secured the polar interface of pfMetAP constituted by D193D204E303E334 (see in Fig. 18(B)).

Fig. 18

(A) Comparison of orthosteric site of pfMetAP (residues with brown backbone) and HsMetAP (residues with grey backbone) show complimentary evolutionary mutations at Thr156, Ser268; (B) Binding mode of R-DHB with pfMetAP: utilises the major residues and shows tricyclic core spatially coplanar oriented over Phe276.

Guanosine monophosphate synthetase (GMP synthetase)

GMP synthetase (PDB: 3UOW, Res. 2.72 Å, (Wernimont et al., 2011a, Wernimont et al., 2011b)) is dimeric in nature. Each monomer is composed of two catalytic domains, an N-terminal independent GATase (1–236) and a C-terminal ATPPase domain (237–555) (Ballut et al., 2015). Its dimer form is highly required for its activity as the interface has 108C-terminal residues of the ATPPase domain. In this interface, two cis-prolines (Pro548–Pro549) allow a tetrahedral configuration of Asp543, Thr551, Glu553 and Arg539 (Ballut et al., 2015). The binding of co-crystallise xanthose monophosphate (XMP) (Gileadi et al., 2011) shows the key residues of ligand binding site of GMP synthetase, shown in Fig. 19(A): (a) Arg336, Pro436, Gly435, Pro436 covering the purine face of XMP; (b) the steric hindrance of bulky hydrophobic residues Phe510, Pro432 twisted the ribose sugar towards the polarised domain (constituted by Arg441, Gln476); (c) phosphate heads enclosed by the Lys547, Glu553, Thr551 and Ile552. However, binding of tricyclic ring of R-Cassiarin-E fitted to the purine-ribose sugar binding cavity (π-π interactions, 3.69 Å & 3.48 Å with Phe510) while the methylene (–CH2–) tethered substructure of the molecule (Dihydro-isoquinoline) popped out from the XMP binding cavity, see in Fig. 19(B).

Fig. 19

Interactive binding mode (A) XMP (gold); (B) R-Cassiarin-E (pink), with guanosine monophosphate synthetase.

Merozoite surface proteins duffy binding like Domain-2 (PfMSPDBL2)

The merozoite surface proteins of P. falciparum has duffy binding like domains (PfMSPDBL1 and PfMSPDBL2), which helps the merozoite for their initial binding to the surface receptors on the host red blood cell (Wickramarachchi et al., 2009). The duffy binding like (DBL) fold (PfMSPDBL2; PDB: 3VUU, Res. 2.09 Å, (Hodder et al., 2012)) consists of residues from 161 to 457 residues, which has a boomerang shaped α-helical core (9 α-helixes) formed from three subdomains (Hodder et al., 2012): (a) subdomain-1 (region 161–225) has only contain 5 residue long α-helix (α1) and provide a stable junction for subdomain-2 and subdomain-3 by a H-bond network (involving Arg-207 (from subdomain 1), Asp-266 (from subdomain-2), and Glu-352 (subdomain-3)); (b) Subdomain-2 (residues 226–341) composed of four structurally conserved helices (helices 2–5); (c) subdomain-3 (344–460) is a helical bundle composed of two long α-helices (α6 and α7) and two smaller α-helices (α8 and α9) (Hodder et al., 2012). Moreover, a disulfide linkage between Cys441 and Cys444 brings helices α8 and α9 are near each other in an anti-parallel manner. The R-Cassiarin-D molecular binding mode clearly shows no involvement with subdomain-1 and binds inside the cavity formed by subdomain-2 and subdomain-3: N252E254K255R261 of α3, T335G336Y337G338I340D443 are in end-tip between helix α5/6, R351T355E359 in α6; P442E443C444K445 in loop between α8/9. Also shows, the H-bond donor-acceptor interactions with COOH of Glu443(2.97 Å) and NH2 terminal of Lys445(2.69 Å), see in Fig. 20.

Fig. 20

Interactive binding pose of R-Cassiarin-D (purple) with PfMSPDBL2.

In our interest, we evaluated the physicochemical properties of these isoforms with regards to the known quinine alkaloid analogues (Chloroquine, Primaquine, Amodiaquine, Mefloquine). However, most of the Cassiarins show equivalent physicochemical properties with respect to the quinine analogous, while Cassiarin-C found to be the closest candidate with Primaquine, as shown in Table 6.

Table 6

Physiochemical evaluation of various isoforms along with the quinine alkaloids.

Molecule	a_acc	a_don	a_don_acc	a_aro	logP(o/w)	logS	Mol_wt
Chloroquine	2	1	3	10	4.28	−3.78	319.8
Primaquine	3	2	5	10	2.21	−2.24	259.3
Amodiaquine	3	2	5	16	4.57	−4.49	355.8
Mefloquine	3	2	5	10	4.27	−4.53	378.3
Cassiarin-A	2	1	3	10	2.18	−3.18	213.2
Cassiarin B	2	0	2	0	2.90	−4.04	313.3
Cassiarin B (R2 = Phenyl)	1	0	1	6	4.22	−5.25	289.3
Cassiarin C	3	1	4	10	2.24	−2.67	215.2
Cassiarin D	6	2	8	16	3.40	−5.75	445.4
Cassiarin E	5	2	7	20	4.15	−6.13	426.4
Cassiarin F	5	3	8	22	4.84	−7.45	427.4
Cassiarin G	5	1	6	10	2.30	−3.13	259.2
Cassiarin H	5	0	5	0	2.59	−3.88	359.3
Cassiarin J	6	2	8	10	2.10	−2.89	333.3
Cassiarin K	2	1	3	10	2.85	−4.01	247.6
DHB	2	0	2	0	2.12	−3.18	216.2

a_acc: number of H-bond acceptor atoms; a_don: number of H-bond donor atoms; a_don_acc: number of H-bond acceptor + donor atoms; a_aro: number of aromatic atoms; logP(o/w): Log water/octanol partition coefficient; logS: log solubility in water; mol_wt: molecular weight of molecule.

Conclusion

The search for new antimalarial scaffold still have valuable weightage. Current research identified the most putative targets for Cassiarin alkaloids in P. falciparum. We also produce a series of top 25 putative targets for individual Cassiarin isoforms against P. falciparum (provided in Supplementary information). Also, found that the monomer forms (like Cassiarins C and DHB) have comparatively more cavity fitting to these proteins, as attributed by their smaller surface area than their Bis-forms (D, E, and F) (provided in Table 2). However, their multi-mode interactions with their putative protein targets also indicate their synergistic pharmacological mode of action against P. falciparum strains. We also disclosed various comparative studies of identified protein targets with their homologous proteins, especially human homologous proteins, which were never studied before and therefore draws several key structural features and differences that could be further exploited in designing and selective targeting against these identified proteins, as provided in case of Oxoacyl acyl-carrier-protein reductase, Kelch motif associated protein, Armadillo repeats only protein and Methionine aminopeptidase 1b. We also found that the screening based on inverse docking, using three different docking methods, quite helpful in filtering the pseudo-positive results which are usually generated from one docking method. This kind of methodology could be useful in, the exploration and target identification for polypharmacological active compounds or validating the side targets of a particular drug.

Conflicts of interest

The authors declare no conflict of interest.