Identification of Potential inhibitors for Hematopoietic Prostaglandin D2 Synthase: Computational Modeling and Molecular Dynamics Simulations

To design a new therapeutic agent for Hematopoietic Prostaglandin D2 synthase (hPGDS), a set of 60 molecules with different molecular scaffolds were (range of pIC50 values are from 8.301 to 3.932) considered to create a pharmacophore model. Further, identification of potential hPGDS inhibitors were carried out by using virtual screening with different databases (from 15,74,182 molecules). The Molecular screening was performed using different sequential methods right from Pharmacophore based virtual screening, molecular docking, MM-GBSAstudies, ADME property analysis and molecular dynamics simulations using Maestro11.9 software. Based on the best pharmacophore model (ADRR_1), the resultant set of 18,492 molecules were screened. The preliminarily screened molecules were subjected to molecular docking (PDB_ID: 2CVD) methods. A set of 27 molecules was screened from the resultant molecular docking outcomes (360 molecules) based on binding free energy (ΔGbind) and Lipinski’s rule of five. Out of 27 molecules, 4 were selected visual data analysis and further subjected to molecular dynamics (MD) simulation study. Outcomes of the present study conclude with three new proposed molecules (SP1, SP2 and SP10) which show a good range of interaction with human hPGDS enzyme in comparison to the marketed compounds i.e., HQL-79, TFC-007, HPGDS inhibitor I and TAS-204.

DP2/CRTH2 receptor-mediated response including initiation and potentiation of immune cell migration, respiratory burst, type 2 cytokine productions and histamine release [3]. PGD2 is a potent target for inflammation; its influence strongly depends on whether it acts in the early or late phase of inflammation. On the one hand, in the early phase of inflammation, acute inflammation, i.e., dermatitis [9] and colitis [10], lipopolysaccharide-induced pulmonary inflammation [11] as well as in anaphylactic shock [12], PGD2 seems to have protective effects.
In this study, a set of 60 molecules were considered to create a pharmacophore model that can efficiently explain the essential features required for the inhibition of hPGDS and to generate a model that alleviates in distinguishing molecules that have good efficiency. These different molecular scaffolds inclusive of indole (A1-8), pyridine (B9-17) [33], benzaldehyde (C18-22) [34], thiophene (D23-29) [35], benzimidazole (TAS-204 derivatives) (E30-41) and pyrimidine (TFC-007 derivatives) (F42-60) [36] based molecules have been taken as a premise against hPGDS in this study (Scheme 1, detailed structures are shown in supporting information Table   S1). Ligand-based pharmacophore generation, pharmacophore-based virtual screening, molecular screening, ADME property analysis and molecular dynamics simulations have been employed to study and identify the new candidate having better interaction and binding affinity with the human hPGDS enzyme from the various available molecular database (Zinc15, chEMBL, Asinex, Decoy molecules, ChemDiv, and Specs) sources.

Methodology
Computational studies were performed using Maestro 11.9 module [37]. The process includes pharmacophore-based molecular screening, molecular docking, MM/GBSA, ADME analysis, and molecular dynamics (MD) simulations. The flowchart of the experimental work has been depicted in scheme 2.

Dataset and Protein preparation
A set of 60 molecules based on a literature survey taken into consideration (Table S1), having a spanned range of pIC 50 values (8.301 to 3.932).3D structures of all the molecules were generated in Maestro 11.9 and optimized in the 'LigPrep' module [37] by using the OPLS_2005 force field [38]. For structural optimization of the molecules, no tautomers were considered as well as only one stereoisomer (retaining specified chiralities) was generated per ligand and remaining were set as default.
3D crystal structure of the human HPGD2 enzyme (PDB ID-2CVD) [26] was downloaded from protein data bank having 1.45 Å resolution. Protein preparation is carried out by using the 'Protein preparation wizard' module. In the preprocessing of protein, missing hydrogens were added, deleted water molecules which are beyond 5 Å from hetero-group and 'ionization and tautomeric' states were generated at pH range (7.0 ± 2.0) by considering Epik (Empirical pKa Prediction). Hydrogen bonds were optimized at pH 7.0, as determined from a pKa prediction by PROPKA and finally, minimization was executed.

Pharmacophore based virtual screening
The pharmacophore model [39] was generated by using the 'phase module' of Schrodinger

MM-GBSA
In

ADME Property analysis
Pharmacokinetics parameters and physicochemical properties of the molecules were calculated by using the ADME descriptors algorithm. ADME properties of all 38 molecules (based on MM-GBSA outcomes) were analyzed by using the 'Qikprop' module of Maestro 11.9 (Ligand-based ADME).Lipinski's rule of five is used to evaluate drug-likeness and filter the molecules on the basis of druggability [41,42]. ADME explored the details of rule of five (like mol_MW< 500,

Pharmacophore based virtual screening
The Pharmacophore Hypothesis generated 14 models (table 1)

Molecular docking
The  Table 2.

ADME analysis
The drug-likeness properties of the molecules studied by using well-known ADME analysis i.e.,  Table 3) whereas remaining 11 molecules were violating above parameters (shown in the bold in Table   3). (detailed structures are shown in Table S2).

Visual Inspection
The selected 27 molecules (Table 4)

Molecular dynamics simulations
MD simulations were performed to study the physical movements of atoms & molecules and the dynamic evolution of the entire system. The RMSD is a quantitative parameter to estimate the stability of the protein-ligand system. (Fig.5   which is exactly similar as histogram results. Ligand-Protein contacts were explored by using the ligand-receptor diagram generated by MD simulation. (Fig.7) For this study amino acid-ligand interactions, more than 4% were considered.

Conclusion
This work helps in identifying a more effective drug candidate against the human hPGDS by performing pharmacophore-based virtual screening as a measuring tool. The pharmacophorebased virtual screening, molecular docking, MM_GBSA, ADME property analysis combinedly

Supplementary Material:
Supplementary material associated with this article can be found in the online version.

Notes
The authors declare no competing financial interest.