Screening of anti-Acinetobacter baumannii plant-based compounds, based on potential inhibition of OmpA and OmpW functions
Mohammadnejad, Parvin; Shahryari, Shahab (2021), Screening of anti-Acinetobacter baumannii plant-based compounds, based on potential inhibition of OmpA and OmpW functions, Dryad, Dataset, https://doi.org/10.5061/dryad.7d7wm37tf
Considering the adverse effect of antimicrobial resistance (AMR) crisis on human life, there is an immediate need for finding new alternatives for treatment of emerging infectious diseases. Therapeutic options, including last-line or combined antibiotic therapies for Acinetobacter baumannii, as an emergent multi-drug resistant (MDR) human pathogen responsible for severe nosocomial and several other infections, are apparently ineffective. The outer membrane protein A (OmpA) and outer membrane protein W (OmpW) are two porins known as virulence factors with different cellular functions. Identification of natural compounds with potentials to block these putative virulence factors can possibly attenuate the growth of the bacteria and control the relating diseases. The current work aimed to screen the therapeutic potential of a library of 371 phytochemicals, as multi-blockers of OmpA and OmpW in A. baumannii. Although the anti-virulence activities of these biomolecules are reported previously, no evaluation on A. baumannii has been performed so far. Moreover, there is no safety screening and early alerts of these compounds. In this study, hits were initially selected based on their physicochemical, absorption, distribution, metabolism, excretion, and toxicity (ADMET) drug-like properties. Afterwards, the selected ligands were subjected to standard docking calculations against predicted three-dimensional structure of OmpA and OmpW in A. baumannii. We identified five phytochemicals (Amorphigenin (PUBCHEM CID 92207), Bisdemethoxy-curcumin (PUBCHEM CID 5315472), Dalbinol (PUBCHEM CID 44257412), Epicatechin gallate (PUBCHEM CID 72276) and Nordihydroguaiaretic acid (PUBCHEM CID 4534)) bearing appreciable binding affinity towards the selected binding pocket of OmpA and OmpW.
Material and methods
Tools and servers
A list of all tools and servers used in this work with a brief description is illustrated in Table S1.
The complete amino acid sequences of OmpA (BAN86529 accession number) and OmpW (BAN89067 accession number) were retrieved from the National Center for Biotechnology Information (NCBI) protein database (http://www.ncbi.nlm.nih.gov/protein/) in FASTA format.
Signal peptide prediction
Secretary nature of selected proteins was predicted using SignalP 4.1 server (http://www.cbs.dtu.dk/services/SignalP/), which integrates a prediction of cleavage sites and signal or non-signal peptide prediction based on a combination of several artificial neural networks. LipoP 1.0 server (http://www.cbs.dtu.dk/services/LipoP) used to predict probable signal peptide within the sequence and signal peptidase I or II cleavage sites within the protein.
Physiochemical properties of OmpA and OmpW
To calculate different critical properties of the proteins including theoretical isoelectric point (pI), molecular weight (Mw), and total number of positively and negatively charged residues, instability and aliphatic indices, the ProtParam server available at (https://web.expasy.org/protparam/) was used. Table 1 presents each criterion with its values through this part of the study. The proteins were classified as stable with instability index of 34.31 and 16.64 for OmpA and OmpW, respectively.
Homology Modeling, model refinement, and energy minimization
As no three-dimensional structure of A. baumannii OmpA and OmpW were found in the Protein Data Bank (PDB), homology models of the proteins were built. To generate three-dimensional (3D) structure of the selected proteins, RaptorX web server (http://raptorx.uchicago.edu/StructurePrediction/predict/) was used. The server predicts the absolute global quality and comparable global quality for each of the residues of the query sequence. In order to check the 3D structure, we used PyMOL molecular graphics system version 22.214.171.124 (Schrödinger, LLC, and Portland, OR, USA). To perform structure refinement and energy minimization of the 3D modelled protein structures, the online server GalaxyRefine (http://galaxy.seoklab.org/) and YASARA software were used, respectively. GalaxyRefine employs the CASP10 assessment to refine the query structure, improving the structural and global quality of the 3D model. This method initially rebuilds side chains and performs side chain repacking and subsequently uses molecular dynamics simulation to achieve overall structure relaxation.
Validation of the 3D structures
To validate the refined and optimized 3D structures, three freely available web tools of RAMPAGE (http://mordred.bioc.cam.ac.uk/rapper/~rampage.php), ProSA-web (https://prosa.services.came.sbg.ac.at/prosa.php), and ERRAT (https://servicesn.mbi.ucla.edu/ERRAT/) were used. The server RAMPAGE evaluates Ramachandran plot through applying PROCHECK principles. The ProSA validation method evaluates model accuracy and statistical significance with a knowledge-based potential. It plots overall excellence scores of the errors calculated in the query 3D structure. Further ERRAT server was used to check the reliability of modeled proteins and their overall quality factors.
Prediction of potential ligand binding pocket
Prior to molecular docking analysis, in order to avoid blind docking, improve the accuracy, and to identify the most important ligand-protein interactions, the top three possible binding sites of the proteins were identified using METAPOCKET version 2.0 (https://projects.biotec.tu-dresden.de/metapocket/index.php). As a META server, METAPOCKET is a popular consensus method, which combines eight methods of asLIGSITEcs, PASS, QSiteFinder, SURFNET, Fpocket, GHECOM, Concavity and POCASA to predict binding sites based on 3D protein structures. PyMOL was used for visualization of the residues in binding pockets. Drug score of the pockets was also predicted using Proteins Plus web server at (https://proteins.plus/).
Natural-compound ligand library
In silico screening of large databases is a cost-eﬀective and time-saving approach towards drug discovery. Ligands selected for the protein interactions in this study were based on a review by Silva et al. . A library of 371 phytochemicals (Supplementary, Table S2) with significant potential antibacterial activity against the major well-recognized virulence factors in a set of pathogenic bacteria (including Pseudomonas aeruginosa, Chromobacterium violaceum, and Staphylococcus aureus) are reviewed in their work. However, no inhibitory effect of these compounds have been screened against the OmpA and OmpW of A. baumannii. The Simplified Molecular-Input Line-Entry System (SMILES) of the compounds were retrieved from PubChem database at (https://pubchem.ncbi.nlm.nih.gov/). The 3D structures of all the molecules were prepared using UCSF Chimera 1.14. The ligands were energy minimized using UCSF Chimera prior to docking, in order to obtain 3D ligand structures with proper bond lengths between atoms. They were finally saved in mol2 format for further docking studies.
Pre-filtration and pharmacokinetic analysis of phytochemicals
A compound has to be passed through multiple filters to be considered as a novel drug. All of the bioactive candidates used in this study were filtered out on the basis of their important physicochemical properties using SwissADME (http://www.swissadme.ch/) online server. Initially, those compounds satisfying Lipinski’s rule of five (RO5) were selected and those that violated the rule were eliminated from downstream analysis. Pharmacokinetic properties of absorption, distribution, metabolism, excretion, and toxicity (ADMET) with crucial role in the development of drug design were predicted using pkCSM web server at (http://biosig.unimelb.edu.au/pkcsm/) to decrease the failure rate of the compound for further analysis. This online machine-learning platform provides information on water solubility, intestinal absorption (human), skin permeability, volume of distribution, total clearance, maximum tolerated dose, skin sensitization, chronic toxicity, hepatotoxicity etc.
Molecular docking analysis
Docking studies provide crucial information on the possible locations of the ligand in the binding pocket of the target. Following receptor and ligand preparation, molecular docking analysis was performed using UCSF Chimera’s built-in AutoDock Vina tool. AutoDock Vina generates different ligands conformers using a Lamarkian genetic algorithm (LGA), which is carried out with an adaptive local method search. The scoring function is based on docking energy, which includes short-range van der Waals and electrostatic interactions, loss of entropy upon ligand binding, hydrogen bonding and solvation. The ﬁltered library of the compounds were docked against selected binding pocket of both OmpA and OmpW proteins to screen them based on their orientation and binding aﬃnities towards targets. The docking was carried out by setting the grid box to encompass binding pockets to provide a search space for the molecules. Each of the compounds were docked into the protein targets. The obtained interactions and generated poses were visualized using PyMOL. The docking results were evaluated for their best orientation and higher binding aﬃnity (kcal/mol). We, then, filtered out only the best compounds that were concomitantly plug both OmpA and OmpW binding pockets.
Drug-Target Interactions and Mechanism of Binding
The mechanism of binding between drug-target complexes were profiled using Proteins Plus server (https://proteins.plus/). The server has its focus on protein-ligand interactions, which provides support for the initial steps when dealing with protein structures, namely structure search, quality assessment, and preprocessing. Advanced options, such as protein pocket detection, ensemble generation or prediction of metal coordination are also supported. Hydrogen bond interactions as well as hydrophobic interactions between potential drug molecules and the protein target were studied via 2D interaction diagrams of the docked drug-target complexes.
BOILED-Egg for Prediction of GI Absorption and Brain Penetration
To reveal the capability of GI absorption and permeability of the Blood-Brain Barrier (BBB), the BOILED-Egg model of the potential drugs were predicted using SWISSADME (http://www.swissadme.ch/).
Science and Engineering Research Board, Award: SR/WOSA/CS-97/2016