Identification of potential leads against 4-hydroxytetrahydrodipicolinate synthase from Mycobacterium tuberculosis

4-hydroxy-tetrahydrodipicolinate synthase (DHDPS) is an important enzyme needed for the biosynthesis of lysine and many more key metabolites in Mycobacterium tuberculosis (Mtb). Inhibition of DHDPS is supposed to a promising therapeutic target due to its specific role in sporulation, cross-linking of the peptidiglycan polymers and biosynthesis of amino acids. In this work, a known inhibitor-based similarity search was carried out against a natural products database (Super Natural II) towards identification of more potent phyto-inhibitors. Molecular interaction studies were accomplished using three different tools to understand and establish the participation of active site residues as the key players in stabilizing the binding mode of ligands and target protein. The best phyto-compound deduced on the basis of binding affinity was further used as a template to make similarity scan across the PubChem Compound database (score > = 80 %) to get more divesred leads. In this search 5098 hits were obtained that further reduced to 262 after drug-likeness filtration. These phytochemicallike compounds were docked at the active site of DHDPS.Then, those hits selected from docking analysis that showing stronger binding and forming maximum H-bonds with the active site residues (Thr54, Thr55, Tyr143, Arg148 and Lys171). Finally, we predicted one phytochemical compound (SN00003544), two PubChem-compounds (CID41032023, CID54025334) akin to phytochemical molecule showing better interactions in comaprison of known inhibitors of target protein.These findings might be further useful to gain the structural insight into the designing of novel leads against DapA family.


Background:
Mycobacterium tuberculosis (Mtb), a brutal killer of the human population by spreading most infectious disease, tuberculosis (TB) has been avowed a big threat to public health across the globe [1]. The Global Tuberculosis Control 2015 has mentioned the statistics regarding the occurrence of 9.6 million of new cases (and 1.5 million patients deaths from TB in the year 2014, out of which 12% of the new cases were HIV-positive patient [1].The year 2015 is seen for a defining moment in the battling against TB where move has been begun from the Millennium Development Goals (MDGs) to another age of Sustainable Development Goals (SDGs), and a step ahead towards complete eradication of this disease [1]. With the advancement of technology new TB medications are presently emerging, and combination of different new compounds and even few vaccines are being tested in different phases of clinical trials. Nevertheless, availability of such kind of medication, resistance to the 'isoniazid', 'rifampicin', 'fluoroquinolone' and few second-line injectable drugs is considered one of the biggest hurdles in the way of SDGs [1,2]. Therefore, deciphering potent and effective molecular drug target enzymes for the development of new novel inhibitors with no pre-existing resistance mechanisms is an important emphasis of research.
The 4-hydroxy-tetrahydrodipicolinate synthase (P9WP25) is a key enzyme of Lysine/DHDPS biosynthetic pathway of Mtb responsible for synthesis of D, L diaminopimelic acid (meso-DHDPS) and lysine [3,4]. Apart from both components, few important metabolites viz. dihydrodipicolinate, a precursor of dipicolinate and UDP-MurNAc-pentapetide is also produced (Figure 1). Both pathway specific metabolites are respectively essential for sporulation and peptidoglycan cross-linking via covalent interaction with D-alanyl moieties of vicinal chain to generate murein polymers providing stability and rigidity to the bacterial cell wall [3][4][5]. It has experimentally shown that de novo biosynthesis of lysine is required for the survival of Mtb during infection, albeit its adequacy in the host. Inhibition of Lysine/DHDPS pathway is fatal to the survival of Mtb [3]. Therefore identification of effective inhibitors against enzymes of this pathway should provide leads for the development of new anti-TB drugs.  . These findings of proposed study would be a great help to wet-lab biology and computer-aided designing of effective drugs against the most infectious malady.

Retrieval of protein 3D structure
The crystal structure (3D) of Mtb-DHDPS (PDB ID: 1XXX) was extracted from RCSB Protein Data Bank. The coordinates of the chloride ion, magnesium ion, 2, 3-dihydroxy-1, 4-dithiobutane (DDT), and water molecules were removed to prepare the protein for molecular docking. The protein was energetically minimized using the CHARMm force field.

Retrieval of ligands 3D structure
The structure of PUB475318, a newly predicted inhibitor of DHDPS [4], and phyto-compound (SN00003544)-like ligands were obtained from the PubChem database of NCBI. The structures of PUB475318based similar phytochemicals were extracted from the Super Natural II database (http://www.uefs.br) . By applying CHARMm force, ligands were energetically minimized using the steepest descent algorithm for 500 steps at an RMS gradient of 0.01. Chemical structure of all ligands are shown in Figure 2.

Docking simulation
, MVD (http://www.clcbio.com) and AutoDock Tools 4.0 were used for molecular interaction studies of ligands and protein.

BioPredicta
It employed Genetic algorithm (GA), Piecewise Linear Pairwise Potential (PLP) and Grid algorithms energy minimization by using MMFF force fields. The Dock scoring function was used to assess the binding efficacies of ligands. This scoring function take into account the terms for van der Walls interaction, hydrophobic effects, hydrogen bonding and deformation penalty. BioPredicta tool uses following fitness function for searching rigid docking space. E = InterEq; E = InterEvdW + InterEq; E = EEPIC; Where, InterEq = intermolecular electrostatic energy of complex; InterEvdW = intermolecular vdW energy of complex; EEPIC = electrostatic potential for intermolecular complex All other required parameters were set as default during the process of molecular interactions.

MVD
It integrates highly efficient PLP and MolDock scoring function for molecular docking. Docking parameters and other required parameters were set to default values [15]. MolDock-rerank score was further employed to judge the binding affinity of ligands.

AutoDock
Polar H-atoms, Kollman united atom and atom type parameters were added and further, non-polar H-atoms were merged during generation of the protein pdbqt file. During preparation of ligand pdbqt file, polar H-atoms added, non-polar H-atoms merged, number of torsions, and rotatable bonds were defined. Cubic volume of 40 × 40 × 40 Å 3 with 0.408 Å grid points spacing and X: 3.163, Y: 39.286, Z: 70.258 centre coordinates was set to cover the entire active site and accommodate ligand to move freely. Lamarckian genetic algorithm was employed for the receptor-fixed ligand-flexible docking calculations. The conformer having lowest free energy of binding (ΔG) was considered for further analysis

Results and Discussion:
Two approaches were implemented to search and find out the potent leads against Mtb-DHDPS. Virtual screening of phytocompounds from the natural products database of the UEFS (http://www.uefs.br) was performed as first approach using recently predicted inhibitor, PUB475318 as template [4]. In the second approach, similarity search for diverse classes of compounds from the PubChem database were carried out using SN00003544 of the first approach as a template (Figure 3).

Docking of phyto-compounds
Among all phyto-compounds docked with the Mtb-DHDPS, SN00003544 was found to bind with the best efficacy in the Nterminal (β/α)8-barrel domain of Mtb-DHDPS comprises of 1-233 residues [4] as consistently reflected by scoring functions of adopted docking tools (Figure 4). Ala18, Thr54, Thr55, Tyr143, Arg148, Lys171, Ala173, Gly194, Asp195, Asp196, Ile211, and Val213 residues of N-terminal (β/α)8-barrel domain and Met251, Ser252, Gly255, and Gly256 residues of C-terminal alpha-helical domain of target protein were engaged in molecular interactions [ Table 1]. Among all residues, the active site residues Thr54, Thr55, andLys171 of N-terminal domain and Ser252, and Gly256of Cterminal were engaged in hydrogen bond formation with the best phyto-lead (SN00003544). Hydrogen bonding between DHDPS and SN00003544 provides a directionality and specificity of interaction. Furthermore, Arg148 of N-terminal is also involved in salt bridge formation and thus contributing to protein-ligand stabilization ( Figure 5, Table 2). Interaction profiling of ligand and protein in the study was carried out by using PLIP tool [16].

Docking study of PubChem compounds and their comparison with known inhibitors
To search and identify a diverse classes of ligands having antitubercular potential, 3-D similary search (similarity score > = 80 %) against the PubChem compound database was carried out using best phyto-ligand (SN00003544) as template. Initially SN00003544akin 5098 compounds were retrieved. These molecules were subjected to RO5 filtration before going for docking studies. The 268 molecules out of 5098 were succeeded the RO5 filtration. Molecular docking studies of these compounds were performed for the best binding orientation prediction into the active site of Mtb-DHDPS using the same docking procedure and parameters as mentioned earlier for phyto-compounds. Out of 268 only 50 compounds exhibited plausible binding along with the formation of H-bond with the active site residue Lys171 of Mtb-DHDPS. Further, out of 50 only 10 ligands were observed consistent as bestowed by all three adopted computational tools [6, 7, 14]. Similar to the best phyto-lead, H-bonding was found to be more prominent interactions withThr55, Arg148, Lys171, and Gly256 residues. The remaining 218 out of 268 compounds exhibited feeble molecular interactions and also failed to form H-bond with the active site residue Lys171, depicting their least antitubercular potential.
A comparison of top 10 PubChem hits were made with the five experimentally known inhibitors for example piperidine-2,6dicarboxylic acid (CID557515), dimethylpiperidine-2,6dicarboxylate (CID12265924), pyridine-2,6-dicarbxylic acid (CID10367), 1,4-dihydro-4-oxopyridine-2,6-dicarboxylic acid (CID11390199), and dimethyl-1,4-dihydro-4-oxopyridine-2,6dicarboxylate (CID68297515), and a novel predicted inhibitor PUB475318 of Mtb-DHDPS in order to screen the best phyto-leadlike chemical agents. Only 4 out of 10 hits exhibited stronger binding affinity in comparison of 5 experimentally known inhibitors. Furthermore, out of four only two compounds (CID54025334 and CID41032023) were depicted as stronger inhibitors in comparison of PUB475318 as shown by scoring functions of adopted docking tools (Figure 6). Docking scores, hydrogen bonding residues, residues involved in molecular interactions of top four PubChem hits and known inhibitors are summarized in Table 3.   The active site residues Thr54, Thr55, Arg148, Lys171 of N-terminal domain and Gly256 of C-terminal domain were stabilized the molecular interaction of first potent PubChem ligand (CID41032023) and protein (Mtb-DHDPS) through hydrogen bond formation. Apart from H-bonding, Arg148 is also engaged in salt bridge formation and enhancing the stability of complex (Table 4).   Likewise, higher binding affinity of second potent PubChem compound (CID54025334) towards Mtb-DHDPS was attributed by the five hydrogen bonding (Asp195, Asp196, Ile211, Val213, and Ser252), two hydrophobic interactions (Thr54, and Val213), and one salt bridge formation (Lys171) (Figure 8, Table 5) demonstrating stronger inhibitory potential of ligand in comparison of known inhibitor (PUB475318). Docking complex of known inhibitor and Mtb-DHDPS and their binding pattern are respectively shown in (Figure 9 and Table 6).

Validation of docking protocol
Validation of docking procedure adopted in the study was accomplished by superimposing all the ligands showing stronger binding activity into the active site of Mtb-DHDPS [18][19][20]. The best phyto-lead (SN00003544), phyto-lead like PubChem compounds (CID41032023 and CID54025334) and known inhibitor (PUB475318) were docked into the same binding orientation of target protein and thus favoring the adopted docking procedure (Figure 10).

Conclusion:
In the present study, we employed two virtual screening approaches towards the identification and elucidation of novel drug leads against one of the oldest malady of humankind. In the first approach we screened out a potent natural compound (SN00003544) from the UEFS (http://www.uefs.br) database that bestowed strong binding affinity with Mtb-DHDPS as shown by five hydrogen bonding (Thr54, Thr55, Lys171, Ser252, and Gly256) and one slat bridge formation (Arg148). In the second approach two compounds (CID41032023, CID54025334) akin to phyto-lead with extremely different scaffold from template molecule were identified. These two compounds demonstrated better binding mode into the active site of Mtb-DHDPS and establishing strong bonded and non-bonded molecular interactions (e.g.; hydrogen bonds, salt bridges and hydrophobic interactions) as compared by known inhibitors. In hydrogen bonding distance between donor and acceptor atoms (<4.1 Å), and angle between donor, acceptor and hydrogen atoms (>100°) were found in significant range. Similarly in salt bridges, distance between centers of charge (<5.5 Å) and in hydrophobic interactions, distance between interactions carbon atoms (<4.0 Å) were found significant [16]. Since all three leads predicted in the study have ability to inhibit the activity of target protein by blocking the active site residues via three different important interacting forces (viz. H-bond, salt bridge, and hydrophobic interaction) that determine the stability of biomolecular interactions. Due to strong blockage of active site residues of target protein, de novo biosynthesis of lysine and other secondary metabolites might be impeded during infection and survival of the pathogen threatened. Albeit the wet-lab studies are indispensable to validate the in silico findings of the study, however, predicted leads would certainly help the experimental designing of more potent anti-tubercular agents.   1 distance between hydrogen and acceptor atoms, 2 distance between donor and acceptor atoms, 3 angle between donor, acceptor and hydrogen atoms, 4 distance between centers of charge, 5 functional group in the ligand providing the charge  3.20 153.38 1 distance between hydrogen and acceptor atoms, 2 distance between donor and acceptor atoms, 3 angle between donor, acceptor and hydrogen atoms, 4 distance between centers of charge, 5 functional group in the ligand providing the charge, 6 distance between interactions carbon atoms, 7 ID of ligand carbon atom, 8 ID of protein carbon atom 102.01 1 distance between hydrogen and acceptor atoms, 2 distance between donor and acceptor atoms, 3 angle between donor, acceptor and hydrogen atoms, 4 distance between centers of charge, 5 functional group in the ligand providing the charge, 6 distance between interactions carbon atoms, 7 ID of ligand carbon atom, 8 ID of protein carbon atom 3.12 3.87 135.39 1 distance between hydrogen and acceptor atoms, 2 disatance between donor and acceptor atoms, 3 angle between donor, acceptor and hydrogen atoms