In-Silico Analyses of Sesquiterpene-Related Compounds on Selected Leishmania Enzyme-Based Targets

A great number of sesquiterpenes are reported in the available literature as good antileishmanial leads. However, their mode of action at the molecular level has not been elucidated. The lack of molecular studies could be considered an impediment for studies seeking to improve sesquiterpene-based drug design. The present in silico study allows us to make important observations about the molecular details of the binding modes of a set of antileishmanial sesquiterpenes against four drug-enzyme targets [pteridine reductase-1 (PTR1), N-myristoyl transferase (NMT), cysteine synthase (CS), trypanothione synthetase (TryS)]. Through molecular docking it was found that two sesquiterpene coumarins are promising leads for the PTR1 and TryS inhibition purposes, and some xanthanolides also exhibited better affinity towards PTR1 and CS binding. In addition, the affinity values were clustered by Principal Component Analysis and drug-like properties were analyzed for the strongest-docking sesquiterpenes. The results are an excellent starting point for future studies of structural optimization of this kind of compounds.


Introduction
Parasites of the genus Leishmania are trypanosomatid protozoa that cause the neglected disease known as leishmaniasis, which infects some 15 million people around the world in three clinical OPEN ACCESS sesquiterpenoids were not included in that study and LmCS and LmTryS enzymes are not targeted. Therefore, as part of our interest in finding new prototypes against leishmaniasis, in this work a study through molecular docking and residual interactions of the molecular basis of action at the in silico level of 123 sesquiterpene-related compounds (possessing antiparasitic activity) within the active site of LmPTR1, LdNMT, LmCS, and LmTryS, is described. These analyses showed very important details at the molecular level that should prove useful for the development of therapeutics from sesquiterpenebased prototypes.

Molecular Docking of Sesquiterpenes
Structural information about the test sesquiterpenoids were acquired from different published works reporting antileishmanial activity of natural-occurring sesquiterpenes. The IUPAC and common names can be found in the Supplementary Material. The compounds were further subdivided by common moieties such as sesquiterpene-coumarins, agarofurans, drimanes, pseudoguaianolides, guaianolides, xanthanolides, germacranolides, eudesmanolides and miscellaneous, whose structures are shown in Figures 1-9. Each ligand structure was separately submitted to a conformational search using MMFF and the geometry of the most stable conformer was then optimized by semi-empirical AM1 parametrization. The most stable conformer of each optimized structure was imported into Autodock/Vina to be docked into the active site of LmPTR1, LdNMT, LmCS and LmTryS enzymes. The structural data of the enzymes were imported from the Protein Data Bank (PDB) ( Table 1).  [11] a Assigned code in the Protein Data Bank (PDB); b Reference including the active site.         . Structures of the tested miscellaneous sesquiterpenes [24,[33][34][35][36][37][38][39][40][41][42].
The resulting docking energies (vina scores as affinity values in kcal/mol) of sesquiterpenoids are summarized in Table 2. Szowitsiacoumarin B (2) was found to be the strongest-docking sesquiterpene coumarin. The sesquiterpene coumarins exhibited overall stronger docking energies towards the Leishmania protein targets. This fact is observed by the mean value of the docking energies for each sesquiterpene-type skeletal ( Table 3). The RSD percentages were found to be in a 5.1%-11.4% range indicating dispersion between the affinities for each type of sesquiterpenes which might be correlated with enzyme selectivity. In contrast, pseudoguaianolides, guaianolides, and germacranolides revealed the highest-docking energies and lowest-RSD percentages. Agarofurans 48-50, 55, 90-91 also exhibited stronger docking energies, although they showed selectivity towards PTR1 and NMT.
In the case of the docking with PTR1, dimeric xanthanolide compounds 75-76 exhibited the best affinity values (-10.6 kcal/mol), with even lower values than that of the control used, DB07765 (-10.2 kcal/mol). Their lowest-energy docked poses are shown in Figure 10. Both complexes share identical H-bond interactions with Ser111, Ser227 and Arg17, being the crucial contacts for the ligand-enzyme docking. In addition, as shown in the 2D representation of protein ligand interactions (Figure 10c), a receptor contact through Phe113 contributes to the non-polar stabilization in the hydrophobic guaianolide moiety. These two compounds were reported by Schmidt et al. [27] with reasonable antileishmanial activity against L. donovani amastigotes with IC 50 values in the 22-27 µM range. Our results indicate that compounds 75-76 could serve as leads for further structural optimization studies towards PTR1 inhibition.
In addition, compounds 75-76 share as structural feature an α-methylene γ-lactone moiety. This moiety in sesquiterpene lactones (STLs) is considered to be capable for reacting in nucleophilic Michael additions (acting as Michael acceptor) with sulfhydryl-(-SH) or amino-(-NH 2 )-containing residues, such as cysteine, lysine and histidine [43,44]. Several studies have indicated that the formation of this type of Michael adducts is correlated with the biological activity of STLs [45]. Analyzing the 2D representation of protein ligand interactions (Figure 10c-d) for lowest-energy docking pose of 75-76, and other test α-methylene-containing STLs (data not shown), Lys198 was found to be the only residue within the LmPTR1 binding pocket that satisfies the structural requirement for a possible Michael addition, but due to the H-bond interaction (2.1 Å) between the lactone carbonyl group and Ser222, the Lys198 residue is placed at the other face of the molecule, which eventually might hinder the formation of a Michael adduct.   On the other hand, the bicyclic sesquiterpene coumarins, szowitsiacoumarins A (1) and B (2), also exhibited good affinity with PTR1 (-9.9 and -10.0 kcal/mol, respectively) very close to that of the control. The lowest-energy docked pose of compound 1 is shown in Figure 11. An identical H-bond interaction with Ser111 was revealed by the PTR1···1 and PTR1···2 complexes, which seems to be a key interaction for the complexes stabilization. Another polar interaction with Lys198 could also stabilize the complex as well as a receptor contact through Phe113 contributes to stabilization via the hydrophobic sesquiterpene centroid, which is located between three leucine (Leu226, Leu229, and Leu188) and one serine (Ser227) residues. These two compounds were reported by Iranshahi et al. [18]. However, their antiparasitic activity was not evaluated while compounds 4-5 were reported to be active against L. major promastigotes with IC 50 values in the 13-17 µM range, but the PTR1 affinity values were not significant.
Regarding docking with LmCS, compounds 2, 30, and 75 exhibited the lowest docking energies (−11.8, −12.1, and -11.4 kcal/mol, respectively), possessing significant lower affinity values than that of control used in this study, ZINC01690699 (−9.6 kcal/mol). The agarofuran 30 was the most affine ligand for LmCs. The most-stable pose of 30 is shown in Figure 12. Two H-bond-type interactions are present in the CS···30 complex when the ligand interacts with Gly186 and Ser274 through cinnamoyl and benzoyl oxygens at C-9 and C-8 carbons (2.4 and 2.1 Å, respectively). A structural requirement was observed on comparing the affinity values of isomers 29 and 30. These compounds differ solely by the geometrical isomerism of the cinnamoyl double bond (Z-29 and E-30). However, the docking energy of 30 was lower than that of 29 (−12.4 vs. -11.4 kcal/mol). The cynnamoyl aromatic ring is apparently necessary for a hydrophobic contact with Gly184 and Thr190 in 30. The presence of the two cinnamoyl and benzoyl ester groups in 30 seems to be an essential requirement for the docking, since the best pose has these substitutions oriented towards the binding pocket, formed by Val50, Pro301, Phe273 and Ser274.  In addition, the germacranolide neurolenin A, 83, also showed a better affinity value (−10.6 kcal/mol) than that of control for docking with LmCS. Compounds 81 and 83 (which differ by the position of the isovaleoryl group attached at C8 and C9 carbons) exhibited affinities which were significantly different. On comparing the ligand interactions of these germacranolides, it is evidenced that the location of this ester group, although it is not determinant in the docking, is important since in the germacrene moiety in 83 allows suitable H-bond interactions with Gly186, Ser274 and Lys51 (ca. 2 Å) ( Figure 13), while the lowest-energy pose of 81 is oriented in such a way that H-bond interactions with Gly186, Ser274 and Asn278 are presented. Seemingly, the hydrophobic proximity with Phe273, Asn278 and Glu211 contributes to the better stability of the LmCS···83 complex. Our results indicate that compounds 30 (in fact, several agarofurans) and 83 are promissory leads for further studies towards CS inhibition.   Agarofurans 22, 27, and 39 exhibited the best docking energies (−6.9 kcal/mol) with LdNMT, but the vina score of the control used, DDD64558, was substantially better (−8.0 kcal/mol). A pi-interaction was found between nicotinoyl moiety and Arg179 for all nicotinoyl-containing agarofurans evaluated in this study ( Figure 14). The presence of this substituent was found to be critical to the docking of the agarofuran-related sesquiterpenes. Similarly, sesquiterpene coumarin 2, exhibited almost identical affinity (−6.8 kcal/mol) with LdNMT to that of 22. However, after detailed analysis of the vina scores and binding modes of sesquiterpene···LdNMT complexes, this kind of compounds could be considered as non-suitable ligands for NMT inhibition.
Sesquiterpene coumarins also exhibited good affinity values in the docking with LmTryS, with compound 1 being the strongest-docking sesquiterpene coumarin (-8.9 kcal/mol). Similarly, pseudoguaianolides 51 and 64 and germacranolides 80 and 89 exhibited good vina scores in the −8.5-−8.7 kcal/mol range. The complexes of all these compounds displayed lower docking energies than the control used, DDD66604, (−8.2 kca/mol). Figure 15 shows the 2D representation of protein ligand interactions of compounds 1, 64 and 80. The residue Gln5 was found to be a substantial polar contact in order to stabilize the complex for compounds 64 and 80, while Lys590 resulted as the indispensable side chain donor contact for compound 1. In addition, the lowest-energy pose of 80 has Lys590 very close to the α-methylene γ-lactone moiety. A Principal Component Analysis (PCA) was performed using the vina scores data of the test compounds with LmPTR1, LdNMT, LmCS, and LmTryS enzymes. The two first components of the PCA explain the 81% of the total variance. The resulting score plot (PC1 vs. PC2, Figure 16) revealed the clusters grouping the most affined ligands (red ellipses), on the one hand, and the ligands possessing the highest docking energies (green ellipses), on the other. The blue ellipse grouped the compounds with selectivity against one or two enzymes, but lowest docking values for the other enzymes. The resulting loading plot (PC1 vs. PC2) indicated that the docking values with LmTryS have a substantially different behavior than other enzymes, whereas vina scores values with LmPTR1, LdNMT, and LmCS are more highly correlated. Thus, compounds 75 and 76 are separated due to the higher docking values with LmTryS but good affinity with the other enzymes, while sesquiterpene coumarins 1-2 are clustered since they exhibited overall stronger docking energies towards the Leishmania protein targets as mentioned above. These PCA results indicated the good in silico information-based discrimination as a reasonable tool for ligand design, structural optimization and combinatorial purposes. Additionally, the majority of sesquiterpene lactones is located at the center of the score plot, indicating that, although they may have a certain affinity level towards the test enzymes, in general their structural features not fit for competition of the active sites. In order to observe the relationship between published IC 50 values against Leishmania parasites and the docking energies for ligand design purposes, a correlation analysis was performed between Log(1/IC 50 ) (IC 50 in µM) versus vina scores (kcal/mol). However, due to the assay conditionsdependent variability of the data source and the limitation of the accuracy of docking calculations to predict in vitro activity, no significant correlation was then obtained (R 2 < 20%). Such limitations in docking studies are well-known by the construction of the scoring functions through the postulation that solvation, entropy and electrostatics are largely applicable to several enzyme systems, adding compound absorption, plasma protein binding, tissue distribution, excretion and other biological events, which limit the prediction of biological activity. Nevertheless, there are several cases that distinguish between active and inactive compounds in docking studies and these procedures remain a powerful tool to facilitate the process of drug discovery [46,47].

(b)
Finally, in order to expand our study for identification of sesquiterpene-base leads, drug-like properties, such as partition coefficient (LogP), molar refractivity (MR), polar surface area (PSA), number of H-donors (H-D), H-acceptors (H-A) and rotatable bonds (RB), were calculated for the strongest-docking sesquiterpenes (Table 4). On applying the rule of five (RO5) and its variants [48], epimers 22-27 are violating more than one of RO5, such as MR, PSA, number of atoms and H-acceptors whereas agarofuran isomers 29-30, agarofuran 24 and xanthanolides 75-76 exhibit an greater MR value to that of RO5 criteria. In contrast, sesquiterpene coumarins 1-2 have good drug-like properties, but the LogP values are near to the upper limit of lipophilicity. Within the strongestdocking sesquiterpenes, compounds 64, 81 and 83 displayed the best lead-like properties, suggesting these compounds deserve additional studies for enzyme inhibition.

Protein Preparation
The X-ray crystallographic structure of LmPTR1, LdNMT, LmCS, and LmTryS proteins were obtained from the Protein Data Bank at a resolution of <2.5Å. Water molecules, ligands and other hetero atoms were removed from the protein molecule along with the chains excepting chain A. Addition of hydrogen atoms to the protein was performed. Energy minimization was performed with the CHARMm force field by using conjugate gradient method with an RMS gradient of 0.01 kcal/Å mol on Discovery Studio 2.5 software [49].

Preparation of the Ligands
Structural information of 123 test sesquiterpenoids was acquired from a literature survey on natural sesquiterpenes with antileishmanial activity report (see supplementary material for the IUPAC and common names). A Monte-Carlo randomized conformational search, without any geometrical restrictions, was made using the Merck Molecular Force Field (MMFF94) included in the SPARTAN software [50] with a limit of 500 conformers. Energetically lowest stable conformers within a 6 kcal/mol energy range were geometrically optimized using the semi-empirical AM1 parametrization implemented in the software package. The minimized protein and ligands were saved in PDB format for further docking analysis.

Docking Analyses
Docking experiments were performed using the Autodock/Vina (1.1.2) package under Linux in a Python 2.7 environment, using the AMBER force field [51]. The protein and ligand molecules were prepared as described above. The docking experiment on test enzymes was carried out between the energy-minimized ligand into the active site through a cube at the geometrical center of the native ligand present in the evaluated PDB structure, with the dimensions 24 × 24 × 24Å, covering the ligand binding site with a grid point spacing of 0.375Å. The docking poses are ranked according to their docking scores (as free energy of binding) and both the ranked list of docked ligands and their corresponding binding poses were exported as a CSV file for further analysis. The docked structures were viewed using PyMOL (1.3r2) [51].

Ligand Interactions
Resulting docking information-containing files (in .pdbqt format) were imported into the Accelyrs Discovery Studio 2.5 software package. Once the docked enzyme-ligand complex was imported, an analysis of the binding site was performed in order to create a ligand interaction 2D diagram.

Drug-Like Properties Calculations
Drug-like properties of each sesquiterpene were separately calculated using the Molecular Topology and ChemProPro plugins available into the ChemBio3D ® Ultra 11.0 software package (Cambridge Soft Corporation, Cambridge, MA, USA) starting from the 3D structure.

Statistics
Regression and PCA analyses were carried out using the algorithms included into the software R (3.0.1 version) [52].

Conclusions
A plethora of antileishmanial naturally-occurring compounds active against promastigotes and amastigotes have been identified and evaluated, but their mode of action has not been elucidated. A great number of sesquiterpenes remains in the available literature considered as good antileishmanial leads, e.g., sesquiterpene lactones. However, the lack of molecular studies is a barrier in the route for improving the sesquiterpene-based drug design. In the present in silico study, we have identified molecular details of the binding modes of a set of antileishmanial sesquiterpenes against four drugenzyme targets (LmPTR1, LdNMT, LmCS, and LmTryS). Through molecular docking it was found that two sesquiterpene coumarins are good ligand candidates for the PTR1 and TryS inhibition purposes as well as some xanthanolides exhibited better affinity towards PTR1 and CS binding. The ligand interactions allowed establishing the critical contacts between the each ligand and the active site of the enzymes. The present study is an excellent starting point for future studies of structural optimization of these compounds, whose observed details at the molecular level allowing opening a working route towards developing therapeutic sesquiterpenes-based prototypes against Leishmania.