In silico assessment of natural products and approved drugs as potential inhibitory scaffolds targeting aminoacyl-tRNA synthetases from Plasmodium

Abstract

Malaria remains the leading cause of deaths globally, despite significant advancement towards understanding its epidemiology and availability of multiple therapeutic interventions. Poor efficacy of the approved vaccine, and the rapid emergence of antimalarial drug resistance, warrants an urgent need to expedite the process of development of new lead molecules targeting malaria. Aminoacyl-tRNA synthetases (aaRSs) are essential enzymes crucial for ribosomal protein synthesis and are valid antimalarial targets. This study explores the prospects of (re-)positioning the repertoire of approved drugs and natural products as potential malarial aaRS inhibitors. Molecular docking of these two sets of small-molecules to lysyl-, prolyl-, and tyrosyl- synthetases from Plasmodium followed by a comparison of the top-ranking docked compounds against human homologs facilitated identification of promising molecular scaffolds. Raltitrexed and Cefprozil, an anticancer drug and an antibiotic, respectively, showed stronger binding to Plasmodium aaRSs compared to human homologs with > 4 kcal/mol difference in the docking scores. Similarly, a difference of ~ 3 kcal/mol in Glide scores was observed for docked Calcipotriol, a drug used for psoriasis treatment, against the two lysyl-tRNA synthetases. Natural products such as Dihydroxanthohumol and Betmidin, having aromatic rings as a substructure, showed preferential docking to the purine binding pocket in Plasmodium tyrosyl-tRNA synthetase as evident from the calculated change in binding free energies. We present detailed analyses of the calculated intermolecular interaction for all top-scoring docked poses. Overall, this study provides a compelling foundation to design and develop specific antimalarials.

Electronic supplementary material

The online version of this article (10.1007/s13205-020-02460-6) contains supplementary material, which is available to authorized users.

Introduction

Malaria continues to kill millions worldwide, with a significant proportion of those infected being children. Plasmodium falciparum is the most fatal species accounting for the majority of cases (World malaria report 2019). Over the years, many anti-malarial compounds have been developed but against most of them like chloroquine, pyrimethamine, and proguanil, resistant strains of the parasite have emerged (Gregson and Plowe 2005). Therefore, an innovative strategy is required to find novel antimalarial compounds. Traditional drug discovery involves de novo identification and validation of new molecular entities (NME), a long and expensive method (Shim and Liu 2014). Worryingly, the average time needed for drug development has increased from 7 to 13 years since the 2000s in the US and EU countries (Pammolli et al. 2011). Drug-repurposing is the identification of new manifestations from existing medicine and its application in the treatment of diseases apart from that drug’s intended disease (Pushpakom et al. 2019). This approach can indeed help overcome the initial obstacles in the drug discovery process. A remarkable example of a repurposed drug is the use of sedative thalidomide to treat myeloma. This drug was initially released in the market in 1957 to treat nausea in pregnant women, but soon it was found to cause severe congenital disabilities among children with fatal consequences (McBride 1977). After its withdrawal from the market, researchers discovered the anti-tumor activity of this hypnotic drug (D’Amato et al. 1994). Later, the FDA (Food and Drug Administration, USA) approved thalidomide in combination with dexamethasone for myeloma treatment (Ning and Gulley 2010). Though several of these repurposed drugs have been found by serendipity, the method of drug repurposing is gaining popularity with the increasing clinical application. Also, Natural products (NPs) have been a key source of chemical scaffolds in drug development (Firn and Jones 2003). The therapeutic areas of contagious diseases and oncology have gained from extensive diversity of naturally occurring molecular scaffolds, which can interact with many unique targets within the cell, and have been a source of inspiration for most of the FDA-approved drugs for many years (Mishra and Tiwari 2011).

Aminoacyl-tRNA synthetases (aaRSs) are house-keeping enzymes that catalyze the charging of tRNA with their cognate amino acid, a critical step in ribosomal protein synthesis. This enzyme family is structurally well-characterized and, importantly, is a validated antimalarial target (Manickam et al. 2018). A multi-domain architecture with evolutionarily conserved substrate recognition motifs is a salient characteristic of aaRSs (Fig. 1). The Rossman fold domain facilitates ATP and amino-acid binding, while the anticodon binding domain recognizes the tRNA. These enzymes are categorized into two classes based on topology and conserved motifs—Class I has conserved KMSKS and HIGH motifs that mediate ATP binding. In contrast, Class II enzymes have three conserved sequence motifs 1, 2, and 3 that are crucial for ATP binding (Fig. 1). Several investigations have provided useful insights into the structure–function attributes of the aminoacyl-tRNA synthetase family in Plasmodium falciparum (Herman et al. 2015) (Goodman et al. 2016) (Pham et al. 2014). This study aims at evaluating the prospects of approved drugs and natural products to block the substrate-binding pockets of Plasmodium aaRSs. The motivation for this study comes from the fact that multiple approved drugs are ATP analogs, e.g., Kinase inhibitors (Bhullar et al. 2018). Some known inhibitors of malarial synthetases indeed have a fused aromatic ring as a substructure (Hoepfner et al. 2012). Natural products also have derivatives of similar organic scaffolds, so it would be insightful to access their binding potential within the synthetase pockets. The medicinal value of natural products is well established—about fifty percent of the marketed drugs are naturally occurring molecules or their derivatives (Newman and Cragg 2012). Importantly, decades of research efforts aimed at the design and development of antimalarials have gained immensely by the exploration of the abundant repertoire of natural products. For instance, artemisinin, which is one of the most effective antimalarials, is an active ingredient of Artemisia annua, a herb used in Chinese traditional medicine (Graziose et al. 2010; Wang et al. 2019). These clinically encouraging therapeutic outcomes acclaim natural product scaffolds as the starting point for antimalarial development (Wells 2011). Cladosporin, a fungal secondary metabolite, is a next-generation antimalarial compound that has two vital characteristics as a potential lead for an antimalarial drug. First, it inhibits the growth of Plasmodium in both blood and liver stages at nanomolar concentration. Second, it shows a stronger and specific inhibition of parasite lysyl-tRNA synthetase as compared to the human homolog (Hoepfner et al. 2012). Despite these essential pharmacological properties, Cladosporin failed as therapeutics against malarial because of its poor oral bioavailability. Nevertheless, this knowledge of a PfKRS specific molecular scaffold was propitious for designing metabolically stable synthetic derivatives (Rusch et al. 2019). The quest for discovering antimalarial compounds from natural sources is far from over. Various types of alkaloids derived from medicinal plants offer privileged scaffolds for antimalarial development (Uzor 2020). Recent in vitro and in vivo studies, using plant extracts from Artocarpus champeden, Callistemon citrinus, and Rumex crispus, show encouraging antimalarial activity (Idris et al. 2019; Larayetan et al. 2019; Widyawaruyanti et al. 2020); opening up avenues for the development of effective therapeutic interventions. It is important to emphasize that there are around 15 known inhibitors, which are secondary metabolites produced by microbes, that target at least eight different microbial aaRSs (Francklyn and Mullen 2019). Most of these inhibitors block the activity of prokaryotic enzymes, while some of these such as cladosporin, borrelidin, and halofuginone (a natural product derivative) specifically inhibit Plasmodium lysyl-, threonyl-, and prolyl- tRNA synthetases, respectively (Hoepfner et al. 2012; Jain et al. 2017; Novoa et al. 2014).

Fig. 1.

Domain architecture and conserved motifs for the three aminoacyl-tRNA synthetases from Plasmodium falciparum showing the Catalytic Domain (CD) and Anticodon Binding Domain (ABD). Prolyl-tRNA synthetase also has editing domain (EDIT) and C-terminal domain (C_T). The ligands were docked to the substrate binding pockets within the catalytic domain of the three aaRSs

The well-established effectiveness of naturally occurring molecular scaffolds against malaria motivated us to research the potential of natural products as malarial aaRS inhibitors systematically. For this research, we have virtually screened two small-molecule libraries against the crystal structures for lysyl-, prolyl-, and tyrosyl- tRNA synthetases. The catalytic domains from the Plasmodium and human homologous pairs are structurally conserved; however, there are important differences in the substrate-binding pockets that have been exploited for structure-based drug design. For instance, Cladosporin discriminates between Plasmodium and human lysyl-tRNA synthetase based on the different size of the hydrophobic cavity near the adenine binding pocket. Also, the presence of amino acids with shorter side chains such as VAl328 and Ser344 in the PfKRS compared to the corresponding residues in HsKRS contributes to the high specificity of Cladosporin towards PfKRS (Hoepfner et al. 2012). Similar variations in the binding pocket sizes and orientation of the loops around the binding site can be observed in the case of YRS and PRS homologs from the two organisms (Figure S1). Here, we outline a detailed and comprehensive analysis of the differential binding strengths of two sets of compounds with respect to aaRS homologs in humans and Plasmodium. The results presented here enabled us to identify potential lead-like molecular scaffolds which would prove useful in the structure-based design of specific antimalarial compounds.

Methods

Preparation of proteins and ligands

We retrieved the crystallographically determined aaRS structures from the Protein Data Bank (PDB) (Berman et al. 2000). All these structures are in holo conformation, i.e., their active site is occupied by either a substrate or a substrate analog (Table 1). We modeled the missing coordinates for the loop residues in PfKRS, HsPRS, and both YRS’s by homology derived restraints using Modeller (Sali and Blundell 1993) interface within UCSF Chimera (Pettersen et al. 2004). Detailed information for the missing residues is presented in supplementary table S1. Thus generated full-length structures were energy minimized using the default parameters in the protein preparation wizard of Schrodinger (Madhavi Sastry et al. 2013). We used two different ligand datasets—FDA Approved drugs (2261 molecules) (Wishart et al. 2006) and Plasmodium falciparum specific natural compounds from Natural Product Activity and Species source (NPASS) database (2012 molecules) (Zeng et al. 2018). LigPrep wizard was used for generating tautomers as well as the stereochemical and ionization variants for all the molecules. Subsequently, the conformation of all the ligands were energy minimized with optimized potentials for liquid simulations (OPLS_2005) force-field (Jorgensen and Tirado-Rives 1988). QikProp and Ligand Filtering features were used to eliminate compounds that did not follow Lipinski’s Rule of Five. We performed all the steps for the preparation of proteins and ligands using Schrödinger software (Schrödinger Release 2019-4: Maestro, Schrödinger, LLC, New York, NY, 2019).

Table 1.

Details for the aminoacyl-tRNA structures used in this study. Additional information regarding disordered residues in different aaRSs is given in supplementary Table S1

Organism	aaRS	PDB ID (reference)/resolution (Å)	Bound ligand/binding site	Grid center
Plasmodium falciparum	KRS	4YCV (Fang et al. 2015)/3.41	Cladosporin/ATP binding pocket	Around the bound ligand ‘Cladosporin’
	PRS	5IFU (Hewitt et al. 2017)/2.45	Glyburide/amino acid binding pocket	Around bound ligand ‘Glyburide’
	YRS	3VGJ (Bhatt et al. 2011)/2.21	Tyrosine/amino acid binding pocket AMP/ATP binding pocket	Specific residues (chain A)—247, 248, 250, 235–238,206, 207, 209, 210, 60–63, 69–72, 74, 77, 94, 96, 99,195,192,188,172,185,124
Homo sapiens	KRS	4YCU (Fang et al. 2015)/2.10	Cladosporin/ATP binding pocket	Around the bound ligand ‘Cladosporin’
	PRS	5VAD(Adachi et al. 2017)/2.36	Proline/amino acid binding pocket 91Y/ATP binding pocket	Specific residues (chain A)—160, 28, 162–167, 27, 41, 31, 52, 56, 283, 284, 280, 288, 281, 515, 706
	YRS	4QBT (Sajish and Schimmel 2014)/2.10	Tyrosine/amino acid binding pocket	Specific residues (chain A)—39–43, 54, 55, 77, 215, 184, 185, 187, 182, 152, 166, 170, 173, 72

Grid generation and docking

An essential requirement for running docking calculations is to mark the region in which the sampling of the docked poses would be performed. For inhibitor bound aaRS structures, we prepared a grid by selecting the co-crystalized ligand using the default parameters. In the case of substrate-bound aaRSs, the centroid of the active site residues was used to mark the grid center. We used the receptor-grid generation module in Schrodinger to generate input files for docking. Details of the residues constituting the active site in different aaRSs are outlined in Table 1 and Figure S1 has a visual representation of the grid box for all the aaRSs. We docked the two ligand datasets prepared above to all the six aaRSs using the Glide module of Schrodinger (Friesner et al. 2006) in the flexible docking mode, where the program auto-generates different conformation for each input ligand while docking. Glide samples various orientations of the ligand within the grid box and provides a quantitative assessment of the strength of non-bonded interactions between the protein and the ligand through the Glide Score. We have compared the relative strength of ligand binding to each homologous pair of aaRSs using the Prime MM-GBSA module in Schrodinger. This method calculates the change in binding free energy (ΔG_binding) by subtracting the free energy of the components (receptor and ligand) from the free energy of the complex. Energies of each of the three components, i.e., complex, receptor and ligand were calculated using OPLS_2005 force-field and VSGB (Variable dielectric Surface Generalized Born) implicit solvation model (Bhachoo and Beuming 2017). We have also performed a detailed analysis of the drug-likeness of the top-ranking ligands using SwissADME (Daina et al. 2017). In addition, evolutionary conservation of the ligand interacting aaRS residues was done using ConSurf. PyMol and Python scripts were used to prepare structure and graph figures.

Results and discussion

Repositioning approved drug against malarial aaRSs

We have investigated the plausibility of repositioning the approved drugs and prospects of natural products as potential inhibitors targeting aminoacyl-tRNA synthetases from Plasmodium falciparum. To this end, we have performed virtual screening of 3540 molecules (1535 from FDA and 2010 from NPASS) against three pairs of aaRSs from Plasmodium and humans. The docking scores for all these protein–ligand dockings are in supplementary file 1. Following is the detailed comparative analysis of the binding profiles of the approved drugs to these aaRS homologs. Interestingly, for each of the three synthetases, we were able to identify a drug molecule that showed preferential binding to malarial aaRS as compared to the human homolog.

Lysyl-tRNA synthetase

For our molecular docking studies with lysyl-tRNA synthetases from Plasmodium and humans, we prepared a grid box around the bound Cladosporin in the crystal structures of both aaRSs. Figure 2a shows the distribution of 1273 docked drug molecules based on the Glide Score with the two proteins. Calcipotriol, a side chain analog of Vitamin D₃ hormone calcitriol and originally used to treat psoriasis, showed enhanced interaction with PfKRS as compared to the human homolog. This difference in binding score resulted from Calcipotriol docking to different regions within pockets of the two KRSs (Fig. 2b). The docked pose of Calcipotriol with PfKRS shows that the dihydroxy methylidene cyclohexane moiety of the ligand occupies the ATP-binding pocket like the isocoumarin moiety of Cladosporin and the adenine ring of ATP (Fig. 2c, d) with a docking score of − 7.13 kcal/mol. Whereas on analyzing the results in HsKRS, the docking score of Calcipotriol is just − 4.19 kcal/mol, and the docked Calcipotriol does not superimpose the inherent Cladosporin (Fig. 2e, f). Importantly, the docking score of the Calcipotriol-PfKRS complex is better than the corresponding score of the complex with the known inhibitor. The Glide score for re-docking Cladosporin to PfKRS crystal structure was − 4.21 kcal/mol. Structural analysis of the docked complex shows that the two hydroxyl groups of dihydroxy methylidene cyclohexane part of the docked ligand form hydrogen bonds with some highly conserved residues like side chain of His338 and the backbone of Asn339, in PfKRS. These interactions are similar to the Cladosporin-PfKRS interactions observed in the crystal structure. In HsKRS, Calcipotriol got docked near the surface of the ATP-binding pocket, forming hydrogen bonds with Glu426 (helix ɑ9) and Arg485 (helix ɑ10). Interestingly, the other end of the ligand extends towards the amino acid binding pocket in PfKRS but completely misses this pocket when docked with HsKRS. The docked pose of Calcipotriol in PfKRS is such that it sits just above the backbone of the substrate lysine in the HsKRS. A hydrogen bond between the hydroxyl group beside the cyclopropyl ring with the highly conserved residue Glu308 (helix ɑ7) stabilizes the docked pose. These results show that the docked pose of Calcipotriol engages both the substrate binding pockets in PfKRS, while it occupies only the ATP binding pocket in case of HsKRS, resulting in the difference of ~ 3 kcal/mol in the docking score of the ligand with respect to the two proteins. The calculated change in binding free energy (ΔG_binding) for the two complexes further substantiates that Calcipotriol shows a stronger binding to Plasmodium protein as compared to the human homolog. The ΔG_binding values for Plasmodium and human complexes were − 36.4 kcal/mol and − 29.2 kcal/mol, respectively.

Fig. 2.

a Docking scores for the approved drugs with PfKRS (PDB ID: 4YCV) and HsKRS (PDB ID: 4YCU). Top five ligands based on the Glide score against PfKRS are highlighted in blue, while the ligands having the largest difference in scores between the two KRS homologs are red. The details for all the annotated points in the scatterplot are given in Table S2. b Superimposition of Calcipotriol poses in PfKRS (yellow) and HsKRS (cyan). Our results show a difference of ~ 3 kcal/mol in the Glide score for these two poses. Superimposition of the docked Calcipotriol with the crystallographic Cladosporin pose (from PDB ID: 4YCV) in the case of PfKRS (c) and HsKRS (e) with corresponding intermolecular interactions for the docked pose (d, f)

Prolyl-tRNA synthetase

Prolyl-tRNA synthetase is also a homodimer similar to tyrosyl-tRNA synthetase but belongs to class II of aaRS. We have performed docking using only the monomer and have been able to dock 849 approved drugs to both the homologs (Fig. 3a). In the case of docking with PfPRS, our calculations show Raltitrexed as the top-ranking docked molecule. Raltitrexed inhibits thymidylate synthase and works as a chemotherapeutic drug for colorectal cancer. The docked pose for Raltitrexed in PfPRS superimposed well with the crystallographic pose of Glyburide within the amino acid binding pocket (Fig. 3c, d). Encouragingly, Raltitrexed showed stronger binding to PfPRS compared to Glyburide as evident from their docking scores of − 8.48 kcal/mol and − 4.09 kcal/mol, respectively. In complex with PfPRS, the carbonyl group of Raltitrexed formed a hydrogen bond with the backbone of Ser33 in the Rossmann fold domain. Also, the carboxylic group adjacent to amine forms a hydrogen bond with the backbone of the highly conserved Asn230. The other end of Raltitrexed, having two carboxylic groups is engaged with the active site via salt-bridge with the side chain of Arg232 (another highly evolutionarily conserved position with PfPRS). Interestingly, these sets of non-bonded interactions between PfPRS and Raltitrexed mimic the binding of Glyburide to the active site of the protein. On the other hand, the docked pose of Raltitrexed within HsPRS, was oriented away from the binding pockets (Fig. 3e, f) and, therefore, had a low docking score of − 2.66 kcal/mol. The docked HsPRS-Raltitrexed complex was stabilized by two hydrogen bonds with Ser33 and Lys38, which were formed by two carboxyl groups on one end of Raltitrexed, whereas the other end did not participate in any non-bonded interactions. These differences in intermolecular interactions corroborate well with the quantitative estimates for the strength of Raltitrexed binding to PfPRS and HsPRS having ΔG_binding values of − 52.64 kcal/mol and − 20.38 kcal/mol, respectively. Figure 3b shows the superimposed orientations for the docked Raltitrexed against Plasmodium and human PRSs. The observed difference in the binding profile of Raltitrexed with these two homologs and the difference of ~ 6 kcal/mol in the docking scores indicate that this molecule could be a potential lead for antimalarial drug development.

Fig. 3.

a Docking scores for the approved drugs with PfPRS (PDB ID: 5IFU) and HsPRS (PDB ID: 5VAD). Top five high scoring ligands based on the Glide score against PfPRS are highlighted in blue, while the ligands having the largest difference in scores between PfPRS and HsPRS are red. Green triangle represents Raltitrexed, which has the highest docking score with PfPRS (− 8.48 kcal/mol) and has a relatively poor score with HsPRS (− 2.66 kcal/mol).The details for all the annotated points are given in Table S2. b Superimposition of Raltitrexed poses in the PfPRS (magenta) and HsPRS (cyan). Superimposition of the docked Raltitrexed with the crystallographic Glyburide pose (from PDB ID: 5IFU) in the case of PfPRS (c) and HsPRS (e) with corresponding intermolecular interactions for the docked pose (d, f)

Tyrosyl-tRNA synthetase

Tyrosyl-tRNA synthetase is a class I aaRS and is present as a homodimer with each monomer having binding sites for all the substrates. For our molecular docking studies, we have used only one of the two monomers from the dimeric crystal structure. Using Glide, we could dock 1407 approved drugs to both human and Plasmodium YRSs. Figure 4a shows the distribution of these docked molecules based on the Glide Score for docking with the two proteins. Our results show that the top-ranking ligand, i.e., Cefprozil had different docking poses in the binding pockets of human and Plasmodium YRS (Fig. 4b). The docked Cefprozil was engaged in multiple non-covalent interactions within the active site of PfYRS. As shown in Fig. 4c, d, the carbonyl group in the ligand participated in hydrogen bond interactions with the residues from both the ATP binding loops YRS. Side Chains of His70 (first histidine of the HIGH loop) and Lys247 (first lysine of the KMSKS loop) were less than 3 Å from the ligand’s carbonyl functional group. The tyrosyl moiety in Cefprozil showed an analogous set of interactions as observed for the bound tyrosine in the crystal structure. The phenol ring in the ligand was anchored to the active site via bifurcated hydrogen bonding with the side chains of Tyr60 (helix ɑ10) and Asp195 (strand β2). Furthermore, the N-atom within the tyrosyl substructure participated in hydrogen bond interactions with backbones of Ile172, Gln192, and Gln210 and with the Glu64 sidechain. Also, the pi-pi and cation-pi interactions with Phe99 (a highly conserved residue in helix ɑ5) and Lys250 (second lysine of the KMSKS loop), respectively, anchored the ligand’s aromatic rings within the binding pocket. Interestingly, the docked pose for Cefprozil in the HsYRS active site showed a lesser number of stabilizing non-covalent interactions as compared to the docked pose in PfYRS. We observe that the orientation for the tyrosyl moiety in Cefprozil within HsYRS is markedly different compared to the tyrosine bound in the crystal structure. Analysis of the HsYRS-Cefprozil complex showed only three intermolecular hydrogen bonds (Fig. 4e, f). Docking calculations show that Cefprozil could not access the tyrosine binding pocket in HsYRS and instead got docked near the surface of the active site. Sidechains of Lys81 and Glu148, along with the backbone of Ala147 engaged the ligand around the HsYRS active site. Importantly, neither of the two ATP-binding loops in HsYRS participated in interaction with the docked Cefprozil. Overall, these results indicate that Cefprozil interacts rather weakly within the active site of HsYRS. These observations reconcile well with the calculated docking scores for Cefprozil to PfYRS and HsYRS, i.e., − 9.34 kcal/mol and − 3.47 kcal/mol, respectively. The difference of ~ 6 kcal/mol in the docking score with these two YRSs indicates preferential binding to the PfYRS active site compared to HsYRS. Furthermore, we observed a substantial difference in the calculated ΔG_binding for Cefprozil binding to Plasmodium (− 51.5 kcal/mol) and human (− 19.6 kcal/mol) YRSs. Overall, these quantitative measures demonstrate that Cefprozil could be a potential lead molecule for anti-malarial development.

Fig. 4.

a Docking scores for the approved drugs with PfYRS (PDB ID: 3VGJ) and HsYRS (PDB ID: 4QBT). Top five high scoring ligands based on the Glide score against PfYRS are highlighted in blue, while the ligands having the largest difference in scores between the two YRS homologs are red. The green triangle represents Cefprozil which has the highest docking score with PfYRS (− 9.34 kcal/mol) and a relatively poor score with HsYRS (− 3.47 kcal/mol). The details for all the annotated points are given in Table S2. b Superimposition of the docked Cefprozil poses in PfYRS (yellow) and HsYRS (blue). Superimposition of the docked Cefprozil with the crystallographic substrate pose (from PDB ID: 3VGJ) in the case of PfYRS (c) and HsYRS (e) with corresponding intermolecular interactions for the docked pose (d, f)

Natural products as potential inhibitors of aminoacyl-tRNA synthetases from Plasmodium

Natural products constitute a class of organic compounds that are synthesized in living organisms, generally as secondary metabolites. Using a targeted set of compounds from the Natural Products Activity and Species Source (NPASS) database, we have screened their binding properties against the three pairs of aaRS homologs (Fig. 5). Docking scores for all the compounds against these aaRSs are in supplementary file 1. Details of nonbonded interactions for the top docked poses for different aaRSs are outlined below. Interestingly, our comparative analysis of the dockings demonstrates that at least two compounds showed stronger binding to the PfYRS compared to HsYRS. Similar comparative analysis for KRS and PRS, however, did not show any outstanding molecule based on the docking scores.

Fig. 5.

Comparison of the docking scores for natural products with lysyl-, prolyl-, and tyrosyl-tRNA synthetases from Plasmodium and humans. The top five molecules, based on the Glide score, against Plasmodium are blue, and the molecules with the largest difference in the docking scores with the two homologs are red. In the case of YRS, two of the top five ranking molecules had similar docking scores resulting in overlap of corresponding blue dots. Green squares represent an overlap of blue and red circles; these are Isocycloheximide, Folic acid, and Dihydroxanthohumol (top), and Betmidin (below) in the case of KRS, PRS, and YRS, respectively. Supplementary Table S3 has additional details for all the annotated points

In the case of PfYRS, Dihydroxanthohumol showed stronger binding to the active site compared to the binding in HsYRS. This compound is isolated from Humulus lupulus (Common hop), which belongs to the Cannabaceae or hemp family and is a known inhibitor of Nitric Oxide synthase. The docked Dihydroxanthohumol participated in hydrogen bonds in both amino-acid (Tyr60 or Asp195) and ATP (Tyr60 or Asp195) binding pockets of PfYRS (Fig. 6a). The hydroxyphenyl group near C5 of the propane-1-one moiety imitates the binding of Tyrosine in the amino-acid binding pocket. The docked pose superimposed well with the crystallographic pose of the bound amino-acid and AMP. The phenyl group on the other end of propan-1-one, having a dihydroxy and methoxy group, forms hydrogen bonds with Lys247 (the first lysine of KMSKS loop), Asp209, Gly62, and Glu64 (Fig. 6a). Dihydroxanthohumol also engages in pi–pi interaction with Phe99, which is a highly conserved residue in PfYRS. In HsYRS, Dihydroxanthohumol got docked near the ATP binding pocket but away from the amino-acid binding pocket. Our results show that the hydrogen bonding with the backbone of Gly38, Gly182, and Val212 and Tyr39 side chains stabilized the docked ligand (Fig. 6b). The presence of these multiple non-bonded interactions between the docked Dihydroxanthohumuol and PfYRS resulted in the Glide score of − 9.82 kcal/mol, while for HsYRS the score was only − 6.35 kcal/mol. The comparison of calculated ΔG_binding shows a drastic difference in Dihydroxanthohumuol interaction with the YRSs. The Plasmodium complex had a ΔG_binding of − 72.4 kcal/mol, while for human complex, it was + 29.8 kcal/mol. These differences the docking scores and calculated ΔG_binding values along with the conformational analysis indicates that Dihydroxanthohumol has preferential binding to the malarial YRS as compared to the human homolog. The docked pose for Betmidin (Myricetin 3-alpha-l-arabinofuranoside) in the PfYRS active site forms a bifurcated hydrogen bond with the side chains of Tyr60 and Asp195 (Fig. 6c). Additionally, the docked pose formed a hydrogen bond with the side chain of Tyr188, which is a native interaction between substrate tyrosine and the enzyme, as observed in the crystal structure. Similarly, native contacts of bound AMP in the crystal structure were also observed in the docked pose of Betmidin. The hydroxyl group on the oxolan ring interacts with the backbone of Glu64 and the three hydroxyl groups on trihydroxyphenyl form separate hydrogen bonds with the backbone atoms of Asp61 and Asp209 and to the side chain of His235. These multiple protein–ligand interactions resulted in a docking score of − 8.68 kcal/mol. Contrastingly, Betmidin showed a rather weak binding to the HsYRS forming just one hydrogen bond with the enzyme, i.e., with the sidechain of Glu148 (Fig. 6d) such that the docking score of Betmidin binding to HsYRS had a low value of − 0.66 kcal/mol. Our ΔG_binding calculations accord well with the observed differences in the docking scores. Betmidin-PfYRS complex had very strong intermolecular interaction with ΔG_binding of − 50.0 kcal/mol, whereas in case of the Betmidin-HsYRS complex the corresponding value was + 67.5 kcal/mol. This substantial difference in the intermolecular binding energy for Betmidin to the malarial and human YRSs indicates the potential of Betmidin as a privileged molecular scaffold for designing antimalarial compounds. In the case of KRS and PRS, the top-ranking docked ligands in the Plasmodium enzyme were Isocycloheximide and Folic acid, respectively. Both these compounds on docking with human aaRSs showed comparable docking scores to the corresponding homologs. Figure S2 shows the intermolecular interactions observed in the top-scoring docked molecules for KRS and PRS. For the top-ranking molecules against the PfaaRSs we have also calculated their drug-likeness using different criteria. We observed that all the small-molecules followed most of these criteria in addition to Lipinski's rule of five (Supplementary Table S3).

Fig. 6.

Our results for virtual screening of a natural compounds database against aaRS homologs from Plasmodium and humans show Dihydroxanthohumol and Betmidin binds more strongly to the PfYRS compared to HsYRS. Intermolecular interaction between docked poses of Dihydroxanthohumol and a PfYRS and b HsYRS, and Betmidin and c PfYRS and d HsYRS. The difference between the Glide scores for the docked Dihydroxanthohumol and Betmidin against the two homologs was ~ 3 kcal/mol and ~ 8 kcal/mol, respectively. Supplementary figure S3 shows intermolecular interactions for top ranking docked poses against KRS and PRS from the two organisms

Conclusion

The rapid emergence of drug resistance against the existing anti-malarial drugs warrants the need for short-circuiting the rather long and treacherous process of drug discovery. Disrupting Plasmodium’s protein biosynthetic machinery is a promising approach to kill the pathogen. Aminoacyl-tRNA synthetases are crucial enzymes for protein translation, and are validated antimalarial targets. Availability of experimentally determined tertiary structures for different aaRSs from Plasmodium, and human homologs, foster the avenues for computer-aided discovery of potential lead molecules. We report a meticulous assessment of predicted binding strength of approved drugs against lysyl-, prolyl-, and tyrosyl- tRNA synthetases. Also, we describe the ranking of a customized subset of natural products in terms of their preferential binding the Plasmodium aaRSs. Our in silico results show that drugs such as Calcipotriol, Cefprozil, and Raltitrexed bind strongly and preferentially to the PfKRS, PfPRS, and PfYRS, respectively. Our molecular docking calculations accentuate natural products such as Dihydroxanthohumol and Betmidin that showed stronger binding to PfYRS as compared to the corresponding human homolog. Overall, our study underscores specific molecular scaffolds complementary to the active sites within the three PfaaRSs. The outcomes of this study would prove useful in guiding the astute design of potential antimalarial compounds rapidly.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

MD ideated the study and designed the methodology. KD and NP performed the experiments. MD, KD, and NP analyzed the results and prepared the figures. KD and NP prepared a draft for the manuscript which was substantially revised by MD. All the authors approve the submitted manuscript.

Funding

None.

Compliance with ethical standards

Conflict of interest

The authors declare that they have no competing interests.