The Natural Ligand for Metalloproteinase-A Multifaceted Drug Target

Metalloproteinase is one of the key components of Russell viper venom and it is the root cause of edema, blood coagulation, local tissue damage, hemorrhage, and inflammation during snakebite envenoming. Hence, finding a suitable metalloproteinase inhibitor from natural source will be of great biological importance in mitigating pathological effects. In this current study, we employed computational analysis to examine the inhibition of metalloproteinase by phytochemicals present in Andrographis paniculata. Molecular docking studies revealed interaction of A. paniculata phytochemicals with the catalytic M domain’s active site amino acid residues, namely ASN203, ARG293, PHE203, LEU206, LYS199, and ALA122, similar to that of the reference compound Batimastat. 14-acetylandrographolide, 14-deoxy-11,12 didehydroandrographolide, Andrograpanin, Isoandrographolide, and 14-deoxy-11-oxoandrographolide displayed high binding energy and inhibition against the metalloproteinase. Molecular dynamic simulation analysis revealed less root mean square fluctuation of amino acid residues of metalloproteinase-14-acetylandrographolide complex than metalloproteinase-Batimastat complex indicating the high stability for metalloproteinase with the phytochemical. In silico analysis of parameters like ADME properties and drug-likeness of the phytochemicals exhibited good pharmacokinetic properties. Ligand-based virtual screening of phytochemicals to identify similarity to FDA-approved drugs and identification of their possible targets were also performed. The outcome of the current study strengthens the significance of these phytochemicals as promising lead candidates for the treatment of snakebite envenomation. Moreover, the study also encourages the in vivo and in vitro evaluation of the phytochemicals to validate the computational findings.

A 2.91 Å resolution crystal structure of SVMP was downloaded as a PDB file (PDB ID: 2E3X) from the Protein Data Bank (PDB), managed by the Research Collaboratory for Structural Bioinformatics (RCSB) (https:// www. rcsb. org/). The three-dimensional structure of the metalloproteinase is presented in Fig. 2.

Target Protein Binding Site Prediction
The binding site/pocket of SVMP was identified using metaPocket 2.0, which is a meta server for identifying the specific sites of peptides and proteins [13].

Molecular Docking Analysis
The binding efficiency of A. paniculata phytochemicals with SVMP active site was determined using Autodock 4, an in silico method. Water molecules were removed, followed by the addition of polar H-bonds and Kollman charges to the SVMP (target protein). The number of torsions was set for the ligand. Both the target protein and ligand were saved in pdbqt file format. For the ligand to bind to the target protein's active site, a grid map is assigned with x, y, z points. Docking was performed using the Lamarckian genetic algorithm. The binding energy, binding residues, and inhibition constant-like parameters were analyzed and produced as docking outputs [14].

Visualization of Protein-Ligand Interaction
A. paniculata phytochemicals and Batimastat docked complex with SVMP were visualized and analyzed using BIOVIA Discovery Studio Visualizer [15].

Molecular Dynamic Simulation
The protein-ligand complex flexibility was analyzed using the CAB FLEX 2.0 server. Each amino acid residue fluctuation of the best hit was elucidated based on the RMSF (root mean square fluctuation) value to analyze the conformational stability of the complex [16].

Ligand-Based Virtual Screening
Ligand-based virtual screening was performed for phytochemicals to find similar analogs from FDA-approved drugs using SwissSimilarity (http:// www. swiss simil arity. ch/), an online tool used to identify similar small molecules [18].

Target Prediction
The ligand's most possible macromolecular targets of Homo sapiens were predicted using SwissTargetPrediction [19]. Cytoscape software (3.8.2 version) was used to create a phytochemical-target illustration network.

Results and Discussion
Globally, snakes are the most dreaded venomous creatures since they induce morbidity and mortality in great numbers. Snakebite envenomation induces local tissue damage including myonecrosis and inflammation. Pathogenesis induced by snakebite is multifactorial and complex. Many studies reported the association of clinical symptoms with the biochemical variation of venom composition [20]. Snake venom is inclusive of a wide range of proteins and peptides that induce neurotoxic, hemotoxic, cytotoxic, inflammatory-like effects. If not treated, envenomation results in severe morbidity and even death. The main protein classes in snake venom, causing severe pathological effects in victims, are three-finger toxins (3FTXs), phospholipase A2 (PLA A2), snake venom serine protease (SVSP), and snake venom metalloproteinase (SVMP) [21]. Previously, we reported the interaction of A. paniculata phytochemicals with PLA A2 of Russell's viper. Phytochemicals have great affinity towards specific residues responsible for the myotoxic and enzymatic activity of PLA A2 [22].
Succeeding PLA A2, SVMP is the potent target for small pharmaceutical biomolecules. Apart from the hemorrhagic activity of snake venom metalloproteinase, other activities, namely fibro(ogen)olytic, prothrombin activation, activation of blood coagulation factor X, apoptosis, inhibition of platelet aggregation, pro-inflammatory and blood serine proteinase inhibitor inactivation, are also attributed to SVMPs [23]. All these factors have envisioned SVMP as a potential target. SVMP (PDB ID: 2E3X) selected for the study represents class P-IIId, a heterotrimeric class of SVMP, which has MDC (M -metalloproteinase, D -Disintegrin-like, C -Cysteinerich) domain along with snaclec domain. It is made up of a heavy chain and two light chains, namely LA and LB. Cys133, present in the C-terminal of the light chain, bonds (disulfide bond) with Cys389 present in the heavy chain's hyper variable region (HVR) [24]. Zinc and calcium are involved in the catalytic activity and structural stabilization of SVMP, respectively [25]. Russell's viper metalloproteinase is a strong activator of the blood coagulating factor X (FX) [26]. During the physiological coagulation, FX is activated upon R194-I195 bond cleavage by the factors IXa or VIIa resulting in the removal of 52 residues at the N-terminal of the FX heavy chain which is heavily glycosylated. This results in the formation of a catalytic triad (the active site in this case) of the serine proteinase domain. Thus, the activated FX (FXa), in turn, facilitates the conversion of prothrombin to thrombin which eventually forms a hemostatic plug [27][28][29]. Hence, SVMP is seen as one of the crucial proteins mediating the venomous activity of snake venom.
In cellular processes, the interaction of the protein with other molecules is needed for performing their biological function. Hence, knowledge of interacting/functional sites would help us to develop inhibitors for receptor proteins. Therefore, the primary prerequisite step for protein-ligand docking studies is the identification of ligand binding sites. Here, in our present study, the binding sites of SVMP are predicted using the metal pocket tool.
metaPocket analyzes and combines the results obtained from eight predictor tools, namely POCASA, LIGSITECS (LCS), Fpocket (FPK), GHECOM (GHE), Q-SiteFinder, ConCavity (CON), SURFNET, and PASS (PAS), to improve the prediction success rate. metaPocket combines z-score of all 8 predicted autonomous tools which run simultaneously to produce a total z-score. Based on spatial similarity, the pocket sites are determined and finally, all the clusters are arranged based on the total z-score as predicted by metaPocket. The results are presented in the form of a table highlighting the functional residues which are found in the vicinity of the pocket sites [13]. The results are summarized in Table 1, which highlights the binding site at the third pocket (C3), with values 45.169, 39.898, and − 4.220 as the x, y, and z grid points of the pocket and 1.72 as its total z-score. Pockets that are ranked first were used for further molecular docking study. Autodock 4.2 was used to study the binding affinity of various ligands towards SVMP at the molecular level. A two-dimensional structure representing the interaction of A. paniculata phytochemicals and Batimastat with SVMP is presented in Fig. 3. Details like binding energy, inhibition constant, type of interaction, and interacting amino acids are tabulated in Table 2.
Binding energy is better defined as the aggregate of the torsional free energy and the intermolecular energy of the compounds. In simple terms, the energy released during bond formation or the protein-ligand interaction can be defined as binding energy. However, for any favorable reaction to occur the free energy should be negative. The lower the binding energy, the greater is the protein-ligand binding. The binding energy of 14-acetylandrographolide with SVMP is higher than −9.32kCal/mol whereas the least binding phytochemical is bisandrographolide with −4.09kCal/mol as its binding energy.
Inhibition constant (Ki) is defined as the concentration of inhibitor required to produce half-maximum inhibition. Autodock calculates Ki as follows: where ΔG is the docking energy, Rcal is 1.987, TK is 298.15, and the docking energy is the summation of the ligand's internal energy and the intermolecular energy [30].
Ki= exp . △G × 1000 ∕Rcal × TK) Phytochemicals form a non-covalent bond with SVMP residues, namely GLU14, ILE57, ARG85, MET90, LYS93, SER94, HIS95, ASP96, MET119, CYS120, GLN121, ALA122, LYS197, PRO198, LYS199, CYS200, PHE202, ASN203, PRO204, PRO205, LEU206, ASP209, ARG275, ASP276, ASP279, ARG293, ASP294, GLN295, LEU296, TYR311, ASN312, GLY313, ASP314, ASP398, and PRO399. Non-covalent interactions have a pivotal role in assessing the function, structure, and dynamics of the biomolecules. They are reversible and have favorable energy at room temperature. Non-covalent interactions are both strong, sufficient to bind molecules together, and weak, enough to assemble and disassemble without the utilization of much energy [31]. Batimastat, a known inhibitor of  SVMP [32], formed conventional hydrogen bonds with only two amino acid residues-PHE202 and ASN203, whereas the phytochemicals of A. paniculata formed conventional hydrogen bonds with ARG293 and ASP294 found in the Calcium binding region along with the afore mentioned PHE202 and ASN203 residues. This conventional binding is owing to higher binding efficiency. Catalytic calcium-binding site residues interacting with the ligands are presented in Fig. 4. Among the twenty-two phytochemicals, eighteen formed conventional hydrogen bonds with PHE202, ASN203, ARG293, and ASP294 amino acids. As per previous reports, hydrogen bonding between a druggable ligand and a protein receptor is strong and has high specificity [33]. Three different effects rise from hydrogen bonds in ligand binding, namely (i) positioning of ligand by a binding partner, occasionally related to molecule conformational distortion, (ii) substrates recognition, and (iii) ligand affinity. However, they affect the molecule's physiochemical properties [34]. Protein structural flexibility plays a key role in its biological functions. Metalloproteinase and its complex flexibility with phytochemicals were simulated using Cab Flex 2.0, an online server. Structural simulation results showed that the RMSF of amino acids was comparatively low for the metalloproteinase-ligand complex than the metalloproteinase in its unbound state. 14-deoxy-11,12didehydroandrographolide, Andrograpanin, and 14-deoxy-11-oxoandrographolide had high fluctuations when compared with 14-acetylandrographolide, Isoandrographolide, Andropanolide, Andrographolide, and Batimastat. Molecular dynamic simulations of the metalloproteinase-ligand complex are presented in Fig. 5. The change in the amino acid fluctuation at the active site of metalloproteinase indicates the binding of phytochemicals, which ultimately enhance the rigidity of amino acids at the protein's active site. SwissADME a web tool predicts ADME parameters and calculates physiochemical descriptors along with pharmacokinetics properties, drug-likeness, and medicinal chemistry of small molecules. In our current study, ligands were evaluated for drug-likeness properties. Results were presented as BOILED-Egg model, which depicts HIA (passive human Fig. 4 Catalytic calcium-binding site residues interacting with the ligands gastrointestinal absorption) and BBB (blood-brain barrier) permeation by calculating lipophilicity (WLOGP) versus polarity (TPSA) [17]. In our current study, fourteen phytochemicals showed HIA property and seven phytochemicals showed BBB permeation property. The BIOLED-Egg model for A. paniculata phytochemicals in summary is shown in Fig. 6.
Structurally unrelated drugs are actively transported (effluxed) out of the cell against the concentration gradient by P-glycoprotein (P-gp), membrane-bound transporter, resulting in intracellular concentration reduction, which in turn affects the drug's oral bioavailability  Fig. 6 The BIOLED-Egg model for A. paniculata phytochemicals. Legend: BOILED-Egg model representing the HIA (passive human gastrointestinal absorption), BBB (blood-brain barrier) permeation by calculating lipophilicity (WLOGP) versus polarity of phytochemicals. Molecule 1. 14-acetylandrographolide, 2. 14-deoxy-11,12didehydroandrographolide, 3. 14-deoxy-11-oxoandrographolide, 4. Andrograpanin, 5. Andrographidine A, 6. Andrographidine C, 7. Andrographidine E, 8 [35]. BOILED-Egg model additionally indicates that among twenty-two phytochemicals, fifteen would function as P-gp + (P-gp substrate). The drug-likeness property of A. paniculata phytochemicals is shown in Table 3. 14-acetylandrographolide and Andrograpanin displays P-gp + and HIA properties, whereas 14-deoxy-11,12didehydroandrographolide displays BBB permeation along with P-gp + and HIA properties. Thus, these results show the potential use of A. paniculata phytochemicals as drug moieties. The results also suggest that ADME parameter-guided structural alterations can be made to further formulate versatile drugs. Ligand-based virtual screening (LBVS) was performed using the SwissSimilarity tool to find similar analogs from FDA-approved drugs for the phytochemicals. The LBVS predictions work on the hypothesis of "Similar molecules are prone to exhibit similar biological activity" and relies on the concept of "molecular similarity quantification" [36]. A. paniculata phytochemicals were found to be analogous to drugs with antibiotic, antiinflammatory, antihypertensive, antihypotensive, anti-arrhythmic, anticancer, antidiabetic, antilipidemic, antiviral, anti-retroviral, antimalarial, and antifungal activities, along with a similarity score ranging from 0.002 to 0.887. Ouabain, an anti-arrhythmic drug, showed the highest similarity score with the phytochemical Andropanoside. Eplerenone, an antihypertensive drug, was found to be analogous to Paniculide A, with a similarity score of 0.513. A histogram representing the phytochemical similarity score against the FDAapproved drugs is represented in Fig. 7. However, screening of 14-deoxy-11,12 didehydroandrographolide did not show any similar analogs. SwissSimilarity results with analog scores are tabulated in Table 4. Putative targets of A. paniculata phytochemicals were identified through SwissTar-getPrediction, a web tool used to find the possible target for small molecules [37]. In our current study, we found that each phytochemical would target and interact multiple proteins. This might be owing to the polypharmacological effect of the phytochemicals. The phytochemicals which targeted different classes of metalloproteinase were selected and a phytochemical-target illustration network was created using Cytoscape, a software used for visualizing network and integration of biomolecular interactions [38]. The phytochemicaltarget network results are presented in Fig. 8.

Conclusion
Our current study indicated the strong interaction of A. paniculata phytochemicals with the metalloproteinase catalytic M domain. Ligand-based virtual screening showed that phytochemicals were analogs to FDA-approved drugs having antibacterial, anti-inflammatory, anticancer, antidiabetic, antilipidemic, antifungal, antimalarial, and antiviral activity. ADME analysis showed that A. paniculata phytochemicals have a good bioavailability score and can be further evaluated as potential drug candidates. Target prediction revealed that phytochemicals would have multiple protein targets which highlights its polypharmacology effect. Molecular dynamic simulation exhibited binding of A. paniculata phytochemicals with metalloproteinase which influences its biological activities. In conclusion, phytochemicals of A. paniculata were found to be potent drug candidates targeting metalloproteinase, thereby suggesting its use as a drug to neutralize exogenous metalloproteinase in case of snakebites and endogenous metalloproteinase in case of cancers.