3 RESULTS
3.1 Bacterial source and identification
In our previous reports, 65 isolates of A. baumannii were isolated from the ICUs hospitals in Kerman, Iran.31Fifty-five of these were MDR, while 10 were carbapenems susceptible. Among them, one isolate that exhibited the highest MICs to various classes of antibiotics was selected for this study.
3.2 Antimicrobial susceptibility tests and PCR-Sequencing
The organism exhibited high MICs against gentamicin and kanamycin (64 µg/ml), ciprofloxacin and levofloxacin (125 µg/ml), ceftazidime and ceftriaxone (256 µg/ml), imipenem (125 µg/ml), colistin (8 µg/ml), piperacillin/tazobactam (125 µg/ml), and azithromycin (64 µg/ml), respectively (Figure. 1A). Furthermore, MP-B prevented the active growth of the MDR strain of A. baumannii with mean MIC=1 µg/ml, where 99.9% of the cell’s population was not grown in Tryptic Soy broth medium (p≤0.05). A significant synergy was observed when the cells were grown in the presence of both antibiotics (various concentrations) and sub-MIC of MP-B (Figure 1A). The MICs against gentamicin and kanamycin were reduced to 8 µg/ml, ciprofloxacin and levofloxacin to 4 µg/ml, ceftriaxone to 2 µg/ml, carbapenems to 4 µg/ml, and tetracycline to 2 µg/ml, respectively. However, no change in the MIC level of colistin, piperacillin-tazobactam, azithromycin, co-trimoxazole, and chloramphenicol was observed (Figure. 1A). To confirm the synergism of Mastoparan-B with antibiotics, we performed growth curve analysis of the isolate in the presence and absence of sub-MIC of MP-B. When cells grown in 0.5 µg/ml of MP-B, relatively steady growth was observed after 8 hours of incubation (Figure. 1B). The results indicated that the reduction in MIC against antibiotics was not due to the death of bacteria by Mastoparan-B but rather the inactivation of some cellular mechanism (s) such as efflux pump.
To prove this hypothesis, we performed the PCR technique by using specific primer pairs for the entire ade B gene. The AdeB efflux pump gene was then sequenced using Sanger’s dideoxy DNA chain termination method by ABI Prism DNA Sequencer, and its expression was evaluated in the presence and absence of MP-B. Interestingly, a 20-fold decrease in the expression of the ade B efflux pump gene was detected when the cells were grown at the sub-MIC level of this peptide. Furthermore, we obtained the nucleotide sequence of the ade B gene by blasting with similar sequences in the NCBI database (https://www.ncbi.nlm.nih.gov/blast). From the nucleotides, we detected the AdeB protein sequence in the GenBank database.
3.3 Physicochemical parameters and Gene Ontology
The computation of physicochemical parameters of the AdeB efflux pump revealed that the protein was typically composed of 1036 amino acid residues with an average molecular weight (M. Wt.) of 112592.02 and the molar extinction coefficient of (ε) 81945 (Table 1A). Likewise, empirical analysis of amino acid composition by the Expasy ProtParam database revealed the dominance of small aliphatic amino acids like alanine at 16%, isoleucine, valine at 10%, and glycine at 6.7% in the AdeB structure (Table 1B). Furthermore, documentation resources of protein families, active domains, and functional sites indicated the AdeB protein belonged to the RND transporter efflux (HAE1) similar to ArcB protein Acrflvin-R, and ACR_tran family. It showed the nonhomologous relationship with superfamily ArcB_ DN_DC and pore-TolC-like domains. Gene Ontology further confirmed the biological, and functional activities of the AdeB protein. The overview of the sequence/structure diversity of the RND superfamily in this study compared to other superfamilies in the CATH database is presented in Figure. 2A. The obtained functional and structural domains by CATH were similar to MDR efflux transporter ArcB transmembrane domains with two related sequence families and one structural cluster (Figure 2A). The AdeB efflux protein had 91.8% unique sequence annotations (Figure 2B) and, the lineage had 6 functional family members in the genus Acinetobacter (Figure 2C).
3.4 Phylogenetic tree analysis and amino acids alignments
The phylogenetic tree analysis of the AdeB protein convincingly demonstrated high sequence identity (99.9%) with AdeB efflux pump membrane transporter tr-AOA7L9E4V2_ACIBA, tr-A0A5P6FTF2_ACIBA, and tr- A0A3Q8ULEO_ACIBA, 99.6% sequence identity with tr-AOA373B8P5_9GAMM, and 99.41% identity with efflux pump tr-A0A086HUIO_ACIBA, respectively (Figure 3A). Moreover, we also investigated blast parameters including, identity (99%), score (Average 50300), and an E-value (5.3-11) which indicated significant accuracy in clustral pattern (Figure 3B). To prove the above results, we performed multiple amino acid alignments of the AdeB and compared the amino acids helix-5 for closely related strains, the results revealed high identity among the amino acids of these strains (Figure 3C). Overall, the high identity matrix of the AdeB protein sequence was observed in the UniProtKB database (≤99%) (Figure 3D). This indicates that a high evolutionary relationship exists among sequences in this report.
3.5 Prediction of the 3D structure of AdeB protein
The obtained AdeB trajectory showed an inner membrane protein with a 3-fold asymmetrical axis positioned perpendicular to the membrane surface with the typical RND-like folds. The predicted RND efflux pump exhibited three components; i) a 30Å U shape funnel channel in the AdeC outer membrane region composed of both α- helices and β-sheets (Figure 4A), ii) a 40 Å large periplasmic membrane fusion AdeA region with the coil-coil conformation, and iii) a 50Å transmembrane AdeB transporter helical structure in the form of helix-turn-helix with a pore (Figure 4A). The large periplasmic domain of the AdeB molecule is created by two extracellular loops that link together TM1 with TM2 and TM7 with TM8. This periplasmic domain can be divided into six subdomains, PN1, PN2, PC1, PC2, DN, and DC (Figure 4B). These features fit with the extreme variety of antibiotics recognized by AdeB. The N-terminal TM1 leads to the PN1 subdomain that connects the PN2. Similarly, TM7 leads to the PC1 subdomain that was connected to the PC2 (Figure 4B). Nevertheless, a pore connected the inner membrane cleft to the periplasmic cavity (Figure 4C). The amino acids analysis of the pore displayed a high quantity of Lys, Phe, and Ala residues (Ala residue relatively stabilized the pore channel) (Figure 4C). The combination of deep neural network learning with completed I-TASSER assembly significantly improved by threading template quality and therefore boosted the accuracy of the final model through optimized fragment assembly simulations with C-score = 1.41, TM- score =0.99, RMSD (Å) = 4.4 ± 2.9, and P-score = 1, respectively. Nevertheless, the validity of the AdeB structure was further confirmed by the Ramachandran plot (Figure 5A) and local similarity estimate of amino acid residues in chains A, B, and C (Figure 5B). The results indicated that over 99% of the amino acid backbones remained in the Ramachandran favored regions and less in the outlier. The plot also illustrated torsional angles at low clash point ψ (0.87%) and high Ramachandran favored ɸ (97.92%) conditions. The obtained results of the local quality estimate by the Swiss-Expasy database for all three chains illustrated consistency in our work. Furthermore, the structural validation of the secondary protein structure using the PROSA website (Figure 5C) showed the generated model has a Z-score = 2.3, which is within the acceptable range of −10 to 10.
3.6 Molecular docking analysis
For docking of Mastoparan-B with the AdeB protein, we prepared 20 poses and used them to determine the suitable binding site on the AdeB efflux pump. The comparison of the molecular docking simulation before and after attachment of MP-B showed MP-B attached exclusively with the helix-5 inner membrane via Val 499, Phe 454, Thr 474, Ser 461, Gly 465, and Tyr 468 residues, respectively (Figure 6A). Upon docking of the ligand, there was a change in energy minimization (glide score of 0.8 to 0.2), and a shift in the orientation of dihedral angles including phi (φ), and psi (ψ) by distances of 9.0 Å, 9.3 Å, and 9.6 Å (Figure 6B). The black arrows indicate the bond interaction potentials: bond-stretching, VB, angle-bending, VA, dihedral (out-of-plane), Vdh. The protein folding changes were detected by AlphaFold DB software. Here, we used the Alfafold database for the first time for the determination of the AdeB domains and detailed information about folding and the position of helix-5 in the interior side of the AdeB protein (Figure 7A). In addition, the attachment of MP-B to this helix and amino acids is involved in molecular docking studied by this software represented in Figure 7B. Finally, the overall docking process, the positions of the amino acids, and H-bonds with the ligand molecule are shown in Figure 7C. Moreover, comparing the AdeB molecule with template 7cz9A as the model indicated a high score in the ligand-receptor interaction (z-score = 33.510 cutoff = 6.9).
3.5 Prediction of the position and topology of transmembrane helices
The topology of AdeB protein was predicted by both DeepTMHMM and TOPCON tools using Z-coordinates and predicted ΔG-values across the sequence and the Wilcoxon rank-sum test to get a quantified comparison between the two reliability scores independent of the overall prediction accuracy. The majority of transmembrane helices were arranged at positions between 300-500 and 900-1000. The place of binding ligand to the receptor is shown by a circle with a Z score of 5.8 (Figure 8A). Furthermore, analysis by TOPOCONS further confirmed the above results (Figure 8B), and estimated the percent of α-helix inside, membrane, and outside of the AdeB protein (Figure 8C). The transmembrane helices (TMHs) topology constituted the highest percentage in the AdeB structure. Furthermore, employing Pyr2 software we determined 47.7% of amino acid residues displayed α-helical structure, 21% β- sheets, 11.5% turns, and 24.5% coil with the Z score +2.52. As a result, the most frequent topology in AdeB protein was α-helix, followed by a random coil and extended β-strand barrel.
3.8 Contact map of AdeB protein
A protein contact map represents the distance between all possible amino acid residue pairs of a three-dimensional protein structure. Here we carried out the contact map analysis of AdeB transporter by using the FUpred server. There were three overlapping D1-3, D6-2 and D6-1 at the N-terminal segment with a confidence score of 8.76 (Figure 9A). The contact maps derived from the application of a distance cutoff of 9 to 11Å around the Cβ atoms constitute the most accurate representation of the visualized 3D structure. Nevertheless, the domain boundaries were shown by a continuous domain curve (Figure 9B). This was followed by heat map scores which showed a low domain shift point for AdeB protein (Figure 9C). Overall, the predicted continuous domain protein secondary structure of the AdeB suggested a confidence score of >8.6, indicating our prediction was accurate.