2 MATERIALS AND METHODS
2.1 Bacterial source and PCR-Sequencing
An MDR strain of A. baumannii was isolated from the previous collection 32 and used for the detection of theade B gene by polymerase chain reaction (PCR) technique. Briefly, total genomic DNA was extracted by a DNA genomic extraction kit (ParsTous, Iran). The amplification was carried out using a common PCR core mixture (total volume, 50 µl) containing 20 µL PCR buffer, 10 µL of 200 mM each dNTP (Bioneer, South Korea), 1 U of AmpliTaq DNA polymerase (ParsTous, Iran), 20 pmol of the forward -5’ TTAACGATAGCGTTGTAACC -3’ and reverse 5’- TGAGCAGACAATGGAATAGT -3’ primer pair (Bioneer, Korea) and 50 ng of the genomic DNA as the PCR template. The ade B gene was further purified by a DNA purification kit (ParsTous, Iran) and sequenced in both directions by Sanger’s dideoxy chain termination method using ABI Prism 373 DNA Sequencer (Applied Biosystems 373/3730XL, Bioneer, Korea). The ade B sequence was then submitted to the GenBank NCBI database (https://www.ncbi.nlm.nih.gov/genbank/submit/) for accession number. The BLAST program was used to find the amino acid sequence of the AdeB protein.
2. 2 Susceptibility to antibiotics and Mastoparan-B
The susceptibility of the selected strain to 12 antibiotics was carried out by disc diffusion assay and MIC methods using broth microdilution test by the Clinical and Laboratory Standards Institute (CLSI 2019) protocol.33 At the same time, the susceptibility of the isolate to Mastoparan-B was performed by microdilution method using the Tryptic Soy broth medium, and criteria for sensitivity or resistance were measured based on our previous report.34Similarly, we performed MICs of antibiotics in the presence of sub-MIC concentration (0.5 µg/mL) of MP-B to see any synergism between peptide and antibiotics. Before the experiment, we checked the viability of the cells at this concentration, by removing 1 ml aliquots of each sample grown in the presence and absence of 0.5 ml of MP-B at each 2 h and reading the turbidity with a UV/Vis spectrophotometer (Persia Med, AE-S60-4U) at an optical density (OD) of 600 nm. The standard cultures of E. coli ATTC25922 and A. baumannii ATCC 19606 were used as the quality control strains. All antibiotics were purchased from MAST Company, Manchester, UK. Mastoparan-B powder with 99.5% purity was purchased from Shimi-Daru (Tehran, Iran) and used as described by the manufacturer.
2. 3 Expressional analyses of the ade B Gene
RNA extraction, C-DNA synthesis, and relative quantitative real-time PCR (qRT-PCR) were performed to assess the expression of the ade B gene in the presence and absence of 0.5 µg/ml of MP-B as described previously.32 Briefly, amplification was done using 25 µl of SYBR®Green dye and 2× Real-Time PCR master mix solution (BIO FACT, Korea), followed by the addition of 1 µl of ade B primer pair F: 5ʹ-AACGGACGACCATCTTTGAGTATT-3ʹ and R: 5- ʹCAGTTGTTCCATTTCACGCATT-3ʹ and 5 µl of cDNA template. Quantification of the ade B gene was performed by using the ABI Step One qRT-PCR system (Applied Biosystems, Foster City, CA, USA). The gene expression was calculated as a fold change (RQ) between the target gene and matched reference 16SrRNA levels based on the following formula RQ = DDCt. Where ΔCt is equal to the difference between the cutting point (Ct) value for the analyzed gene and Ct for the 16SrRNA reference gene as described by Livak et al.35 A. baumannii ATCC 19660 was used as the internal control strain. Differences were assessed by Student’s t-test for consideration as statistically significant.
2. 4 Classification, physicochemical parameters, and Gene Ontology
The hierarchical classification of AdeB protein to the superfamily, homologous family, predicting domains, and important functional sites were carried out by the InterPro software suite (https://www.ebi.ac.uk/interpro/). To classify proteins in this way, InterPro uses predictive models, known as signatures provided by several different alignments and produces a comprehensive resource for protein classification (https://www.ebi.ac.uk/interpro/search/sequence/). Any alignment with a TM-score ≥0.5, which is the cutoff value for a significant structural match, was used as a potential template for the target protein-peptide interaction. Furthermore, molecular weight, theoretical pI, amino acid composition, atomic composition, extinction coefficient, estimated half-life, instability index, aliphatic index, and grand average of hydropathicity (GRAVY) were computed by information stored in the Swiss-Port ProtParam database (https://web.expasy.org/protparam/). Gene Ontology at molecular and cellular functions was determined using the UniProtKB platform (https://www.uniprot.org/). We also worked on the domain-based method for the AdeB protein sequence versus structural diversity and prediction of the functional site that exploits functional sub-classification of the superfamilies by the CATH protein classification database (https://www.cathdb.info/).
2. 5 Construction of phylogenetic tree of the AdeB protein
The phylogenetic tree of the AdeB protein was constructed using maximum likelihood sequence alignment Clustral Omega version 7 in the UniProtKB pipeline, and variable sites were removed by the stringent settings.36 A bootstrapped was generated with RAxML (Randomized Accelerated Maximum Likelihood) version 7.0.3 with the confidence levels (%) generated from 1000 bootstrap trials.36 Moreover, we choose a consensus sequence manually using similar protein sequences from the UniProtKB restricted to the A. baumannii AdeB protein.
2. 6 Protein dynamic simulation
From the primary amino acid sequence, the 3D model of AdeB protein was generated by the SWISS-Expasy (https://swissmodel.expasy.org), and I-TESSAR (https://zhanggroup.org/I-TASSER/) platforms using the homology modeling program. For each target, I-TASSER simulation generates a large ensemble of structural conformations, called decoys. To select the final models, I-TASSER uses the SPICKER program to cluster all the decoys based on the pair-wise structural similarity and reports up the top five models which correspond to the five closest structure clusters. The confidence of each model is quantitatively measured by the C-score that is calculated based on the significance of threading template alignments and the convergence parameters of the structure assembly simulations. Furthermore, for the evaluation of multiple, amino acids alignments and similarity matrix, we used the FASTA format of the targeted amino acids sequence and blasted in the UniProtKB database (https://www.uniprot.org/). The functional domains of the AdeB protein were bolded using the ROSETTA prediction tool (https://www.rosettacommons.org/) and the Hidden Markov Model program (HMMs) (https://www.ebi.ac.uk/Tools/hmmer/). The quality of obtained models was validated with a comprehensive scoring function for model quality assessment servers such as ProSA and QMEAN.38, 39 The dihedral angles including phi (φ) and psi (ψ) and backbone conformation were assessed using the PROCHECK tool at the PDB Sum server (https://www.ebi.ac.uk/thornton-srv/software/PROCHECK/). In the field of drug discovery, the use of 3D models can provide an understanding of why a certain drug compound is an inhibitor, while a related compound is not, or why certain proteins are druggable while others are not. Recently, the AlphaFold DB database resolves this problem and showed folding of protein plays a role in the druggability of targeted protein. In this study, we evaluated the accuracy of the AdeB protein structure, folding patterns of the helices and domain side chains including the interior side of the AdeB by the AlphaFold 2 database (https://alphafold.ebi.ac.uk). In addition, we computed the backbone dihedral angles (ɸ) to investigate rotameric conformational for the amino acid residues binding with ligand base on the Ramachandran plot and two other model simulation apo and ligands.40
2. 7 Prediction of the protein-peptide interaction site
Because of peptide flexibility and the transient nature of protein-peptide interactions, peptides are difficult to study experimentally. Therefore, we used computational methods for predicting structural information about protein-peptide interactions by the InterPep server (http://wallnerlab.org/InterPep/). The InterPep software is a powerful tool for identifying peptide-binding sites with a precision of 80% at a recall, of 20%. It is an excellent starting point for docking protocols or experiments investigating peptide interactions.
2. 8. Ligand preparation and molecular docking
Mastoparan-B with MF: C78H138N20O16 was retrieved from the PubChem database (Compound CID: 86289587) and used as a ligand in this study. 20 different ligand poses were selected using the LigPrep module in Maestro v11 employing the Schrödinger platform.41 The best interaction poses were visualized using the PyMOL server (https://pymol.org). The most comprehensive and accurate model of ligand-protein docking and the protein function annotation was obtained from the BioLiP database (http://zhanglab.ccmb.med.umich.edu/BioLiP/). The free energy was minimized by the Universal Force Field and converted to the pdbqt format in PyRx0.8 for virtual screening.42 Furthermore, the favorable potential drug target interactions between the selected ligand molecule and modeled receptor (AdeB) were identified by the extra precision (XP) feature of Grid-based Ligand Docking with Energetics (GLIDE) in the Schrödinger platform.42 Each simulation was repeated three times with different initial conditions to increase the precision of the simulations and to prevent any dependencies of the results on the initial conditions. Binding sites were generated using the Site Map tool version 2.4 (Schrödinger Inc.). Finally, molecular docking was performed via AutoDock/Vina suite to assign the ligands and receptors bond orders by adding hydrogen atoms.43 For docking purposes, the best docking pose was selected directly using the Glide G-Score as described previously.41 Once G-Score was obtained, it was used to build optimal binding free energy that distinguishes our sequence from the reference sequence (usually taken to be the optimal sequence). Site Map was tasked to identify the five top-ranked possible receptor suites using the default settings which include size, volume, amino acid exposure, enclosure, contact, hydrophobicity, hydrophilicity, and donor/acceptor ratio. Furthermore, the topology of the transmembrane, inside and outside α-helices was predicted by the DeepTMHMM (https://dtu.biolib.com/DeepTMHMM) and TOPCONS (http://topcons.net/) modules.
2. 9 Preparation of the contact map and domain boundaries
We predicted the AdeB protein contact map and domain boundaries using deep residual neural networks coupled with coevolutionary precision matrices (https://zhanglab.ccmb.med.umich.edu/FUpred). A protein contact map represents the distance between all possible amino acid residue pairs of a 3D structure. For two residues, the element of the matrix is 1 if the two residues are closer than a predetermined threshold, and 0 otherwise.44 To achieve reliable results, we used the FUpred score for a continuous two-domain protein, using the following formula
FUscore2c (l ) =2N 1, 2(l )[1N 1(l )+1N 2(l )]
where l is the domain splitting point of a protein,N 1 (l ) and N 2(l ) represent the number of contacts within the first and second domains, respectively, and N 1, 2(l ) =N 2, 1(l ) indicates the number of contacts between the first and second domains.45 The core idea of the algorithm was to retrieve domains boundaries locations by maximizing the number of intradomain contacts while minimizing the number of interdomain contacts from the contact map.