Efflux MexAB-Mediated Resistance in P. aeruginosa Isolated from Patients with Healthcare Associated Infections

Today, one of the most important challenges for physicians is the adequate treatment of infections due to multidrug resistant organism (MDR). Pseudomonas aeruginosa is considered an opportunistic organism causing different types of healthcare associated infections (HAIs). We aimed to investigate the MDR and pandrug resistance (PDR) rate in P. aeruginosa in our region and detect efflux-pump mexAB genes and the proposed binding interactions of five different categories of antimicrobial agents with the mexB pump. A total of 180 non-duplicated P. aeruginosa strains were isolated from patients with HAIs in the Suez Canal University Hospital. Phenotypically, minimum inhibitory concentration (MIC) was done for all MDR and PDR strains before and after addition of efflux pump inhibitor carbonyl cyanide m-chlorophenyl hydrazone (CCCP). Molecular detection of mexA and mexB genes was done by using polymerase chain reaction (PCR). Most of the isolated strains (126 strains) were MDR (70%); only 10 samples (5.5%) were PDR. MexA and mexB genes were detected in 88.2% (120 strains) and 70.5% (96 strains) of stains, respectively. All PDR strains (10 stains) carried both mexA and mexB genes. Efflux mexAB genes were detected in all MDR and PDR strains (136 strains). Molecular modeling studies were performed to investigate the modes of intermolecular binding interactions between the antimicrobial agents and mexB key amino acids that resulted in MDR and PDR. The current study reported high prevalence of MDR and PDR P. aeruginosa in patients with HAIs in the Suez Canal University Hospitals.

The preliminary results of ciprofloxacin susceptibility test, in MDR and PDR strains, using the disk agar diffusion method, showed that 120 isolates (88.2%) were resistant. MIC for ciprofloxacin ranged from 0.25 to 256 mg/L. According to the established breakpoint values recommended by CLSI, the P. aeruginosa isolates with MIC ≥ 4 mg/L are considered as ciprofloxacin resistant. Nearly, all tested isolates were resistant to ciprofloxacin by MIC test (MIC ≥ 4 mg/L). By using CCCP as EPI, the MICs of ciprofloxacin for 100 isolates (83.3%) of the ciprofloxacin resistant isolate decreased (more than 4-fold) on the CCCP-supplemented plate. In addition, in other synergy tests by using CCCP and other antibiotics as imipenem, we observed a reduction in MIC on addition CCCP which proved the role of efflux pumps in the resistance to these two drugs (p = 0.001). On the other hand, there was no significant change in the MIC of cefotaxime and gentamicin for all the isolates after CCCP addition (p > 0.005).
Molecular detection of the 16s gene, mexA and mexB genes was done by PCR to all MDR and PDR isolates (136 stains). All strains were positive for Pseudomonas 16S gene ( Figure 2; Figure 3). MexA and mexB genes were detected in 88.2% (120 strains) and 70.5% (96 strains) of stains, respectively. Nearly, 80 strains (58.8%) were carrying both mexA and mexB genes including all PDR strains (10 stains), which were mostly isolated from burn and surgical departments (48.7% and 37.5%, respectively). On the other hand, 56 strains (41.2%) were carrying either mexA or mexB genes. Most of them were isolated from ICU, NICU and burn units.
Regarding the antibiotic resistance, we found that the strains carrying both of mexA and mexB genes were more resistant than strains carrying only one gene (the results were statistically significant, p < 0.001). All ciprofloxacin and imipenem resistant strains carried at least one resistant gene. We noticed that MDR strains (No.126) were isolated from the burn units, surgical wards, ICU and NICU (54.8%, 31.7%, 10.3%, 3.2%, respectively). Half of the PDR strains were isolated from the burn unit (50%, 5 strains) and surgical wards (30%, 3 strains), the remaining 2 strains were isolated from ICU.
The preliminary results of ciprofloxacin susceptibility test, in MDR and PDR strains, using the disk agar diffusion method, showed that 120 isolates (88.2%) were resistant. MIC for ciprofloxacin ranged from 0.25 to 256 mg/L. According to the established breakpoint values recommended by CLSI, the P. aeruginosa isolates with MIC ≥ 4 mg/L are considered as ciprofloxacin resistant. Nearly, all tested isolates were resistant to ciprofloxacin by MIC test (MIC ≥ 4 mg/L). By using CCCP as EPI, the MICs of ciprofloxacin for 100 isolates (83.3%) of the ciprofloxacin resistant isolate decreased (more than 4-fold) on the CCCP-supplemented plate. In addition, in other synergy tests by using CCCP and other antibiotics as imipenem, we observed a reduction in MIC on addition CCCP which proved the role of efflux pumps in the resistance to these two drugs (p = 0.001). On the other hand, there was no significant change in the MIC of cefotaxime and gentamicin for all the isolates after CCCP addition (p > 0.005).
Molecular detection of the 16s gene, mexA and mexB genes was done by PCR to all MDR and PDR isolates (136 stains). All strains were positive for Pseudomonas 16S gene (Figures 2 and 3). MexA and mexB genes were detected in 88.2% (120 strains) and 70.5% (96 strains) of stains, respectively. Nearly, 80 strains (58.8%) were carrying both mexA and mexB genes including all PDR strains (10 stains), which were mostly isolated from burn and surgical departments (48.7% and 37.5%, respectively). On the other hand, 56 strains (41.2%) were carrying either mexA or mexB genes. Most of them were isolated from ICU, NICU and burn units.
Regarding the antibiotic resistance, we found that the strains carrying both of mexA and mexB genes were more resistant than strains carrying only one gene (the results were statistically significant, p < 0.001). All ciprofloxacin and imipenem resistant strains carried at least one resistant gene.  Molecular modeling results showed that the position and binding interactions of ceftazidime, gentamicin, ciprofloxacin, aztreonam and imipenem could be closely replicated as observed with LMNG and mexB in the crystal structure PDB 6IIA. The five different categories of antimicrobial agents were bound to the distal binding pocket, where they were inserted into the inhibitor-binding hydrophobic pit showing that both the antimicrobial drugs and the competitive inhibitory LMNG had almost the same binding interactions in the DBP ( Figure 4      Molecular modeling results showed that the position and binding interactions of ceftazidime, gentamicin, ciprofloxacin, aztreonam and imipenem could be closely replicated as observed with LMNG and mexB in the crystal structure PDB 6IIA. The five different categories of antimicrobial agents were bound to the distal binding pocket, where they were inserted into the inhibitor-binding hydrophobic pit showing that both the antimicrobial drugs and the competitive inhibitory LMNG had almost the same binding interactions in the DBP ( Figure 4    Molecular modeling results showed that the position and binding interactions of ceftazidime, gentamicin, ciprofloxacin, aztreonam and imipenem could be closely replicated as observed with LMNG and mexB in the crystal structure PDB 6IIA. The five different categories of antimicrobial agents were bound to the distal binding pocket, where they were inserted into the inhibitor-binding hydrophobic pit showing that both the antimicrobial drugs and the competitive inhibitory LMNG had almost the same binding interactions in the DBP (Figures 4-8). The molecular binding modes can illustrate the role of mexB in P. aeriogenosa resistance to the studied antibiotics.

Discussions
The emergence of Gram-negative MDR is considered a significant public health issue [28]. The rapid rise of antibacterial resistance is still a major global health problem that can impair the antibacterial agents' effectiveness and wastes the efforts for developing new drugs. Efflux pumps have a great concern in emergence of P. aeruginosa antibacterial resistance [6,7,29].
The present study aimed to study the prevalence of MDR and PDR P. aeruginosa and its relationship with the efflux pump mexAB genes as a possible factor involved in resistance in the Suez Canal University Hospital, Ismailia, Egypt. Our data showed high prevalence of MDR among the isolated strains (70%), and only (5.5%) of the isolates were PDR. In concordance with our results, a

Discussions
The emergence of Gram-negative MDR is considered a significant public health issue [28]. The rapid rise of antibacterial resistance is still a major global health problem that can impair the antibacterial agents' effectiveness and wastes the efforts for developing new drugs. Efflux pumps have a great concern in emergence of P. aeruginosa antibacterial resistance [6,7,29].
The present study aimed to study the prevalence of MDR and PDR P. aeruginosa and its relationship with the efflux pump mexAB genes as a possible factor involved in resistance in the Suez Canal University Hospital, Ismailia, Egypt. Our data showed high prevalence of MDR among the isolated strains (70%), and only (5.5%) of the isolates were PDR. In concordance with our results, a

Discussions
The emergence of Gram-negative MDR is considered a significant public health issue [28]. The rapid rise of antibacterial resistance is still a major global health problem that can impair the antibacterial agents' effectiveness and wastes the efforts for developing new drugs. Efflux pumps have a great concern in emergence of P. aeruginosa antibacterial resistance [6,7,29].
The present study aimed to study the prevalence of MDR and PDR P. aeruginosa and its relationship with the efflux pump mexAB genes as a possible factor involved in resistance in the Suez Canal University Hospital, Ismailia, Egypt. Our data showed high prevalence of MDR among the isolated strains (70%), and only (5.5%) of the isolates were PDR. In concordance with our results, a study from Iran published in 2016, reported high MDR rate in P. aeruginosa isolates (66%) from burns unit and only one strain was PDR (0.66%) [30].
We noticed increased rates of MDR and PDR in the burns unit (54.8% and 50%, respectively) followed by SSIs in surgical wards (31.7% and 30%, respectively). As a matter of fact, P. aeruginosa is the main pathogen causing burn infections and is considered as a major colonizer of burn wounds as the moist surfaces of burn wounds represents a favorable medium for its growth, and because of its ability to persist well in hospital environments [19].
The highest resistance was noticed against ciprofloxacin (70%), aztreonam (69%), cefepime (68%), ceftazidime (68%), gentamicin (65%), imipenem (62%) and meropenem (62%). A recent study in burn centers of Tehran, Iran, reported high rate of resistance to ciprofloxacin, amikacin and gentamicin (over 85%) [31] which was higher than our findings. This difference may be due to different samples sources, as in our study, we collected specimens from all cases with HAIs not from burns unit only. Burns unit, as a critical care unit, is rich in bacteria with severe antimicrobial resistance. This explains the high rates of MDR and PDR in burns unit in our study.
We noticed that more than 80% of the MDR isolates were resistant to more than seven antibiotics and more than 60% of the isolates were resistant to more than ten antibiotics. In a similar study, Pellegrino et al. reported lower percentage to co-resistance to seven or ten antibiotics, at two different centers (nearly 32% and 40%, respectively) [32]. The lower resistance rate of that study compared to ours is expected as the antibiotic resistance rates has increased radically all over the world in the past few years. On the other hand, Delpano et al. reported higher antibiotic resistance compared to our study that 100% of the isolates were resistant to ciprofloxacin, tobramycin, gentamicin, cefepime, imipenem and meropenem [33].
Overexpression of efflux pumps could be the leading cause of MDR in bacteria as it leads to a decreased intracellular concentration of antibiotics and reduced susceptibility to antimicrobial agents due to continuous expelling of structurally unrelated drugs [34]. This explains the over representation of mexA and mexB genes (70.5% and 88.2%, respectively) in our study. We noticed that all PDR isolates were carrying both mexAB genes. This over-representation of mexAB genes is matched with the fact that the antibiotics that are substrates of the corresponding transporters are used usually in patient treatment plans.
Molecular docking was done using the molecular operating environment (MOE) program, to investigate the intermolecular binding affinity between the studied antimicrobial agents and mexB which was one of the overexpressed efflux pumps in our study. This binding affinity is directly correlated to the antimicrobial resistance due to the efflux mechanisms which result in a decreased concentration of the antimicrobial agents intracellularly. Some of the key amino acid residues that had intermolecular bindings with all of the antimicrobial agents were Lys134, Gln46 and Gln176.
For the cephalosporin ceftazidime, the 2-aminothiazol-4-yl group elongated parallel in the space above the inhibitor-binding pit, mimicking the LMNG positioning. The ceftazidime carboxylate formed strong ionic bonding with the basic amino acid Lys134. H-bonds were formed with Lys134, Gln46 and Gln176 in addition to arene-arene bindings between thiazole ring and Phe178 in the hydrophobic pit ( Figure 9). The aminoglycoside gentamicin was flexibly bound towards the exit and the entrance of DBP, forming H-bonds with Arg128, Lys134, gln176 and Arg620. Arene-H binding was also formed with Phe615 ( Figure 10). The aminoglycoside gentamicin was flexibly bound towards the exit and the entrance of DBP, forming H-bonds with Arg128, Lys134, gln176 and Arg620. Arene-H binding was also formed with Phe615 ( Figure 10). Pathogens 2020, 9, 471 8 of 13 Figure 9. Two-dimensional image of the binding interactions of ceftazidime in the mexB active site with Lys134, Gln46, Gln176, Phe178 as key binding amino acids.
The aminoglycoside gentamicin was flexibly bound towards the exit and the entrance of DBP, forming H-bonds with Arg128, Lys134, gln176 and Arg620. Arene-H binding was also formed with Phe615 ( Figure 10). The quinolone ciprofloxacin was mainly inserted in the inhibitor-binding pit and formed arene-arene bindings with Phe628 and Phe178, arene-H bindings with Phe610 and Phe178. H-bonds were also observed with Arg620 and Arg128 ( Figure 11). The quinolone ciprofloxacin was mainly inserted in the inhibitor-binding pit and formed arene-arene bindings with Phe628 and Phe178, arene-H bindings with Phe610 and Phe178. H-bonds were also observed with Arg620 and Arg128 ( Figure 11).
Pathogens 2020, 9, x FOR PEER REVIEW 9 of 13 Figure 11. Two-dimensional image of the binding interactions of ciprofloxacin in the mexB active site with Phe628, Phe178, Phe610, Arg620 and Arg128 as key binding amino acids.
The monobactam aztreonam was flexibly bound towards the exit and the entrance of DBP, forming H-bonds with Arg128, Lys134, gln176, Gln46, Lys151, Ser87 and Arg620. Arene-H binding was also formed with Phe610 and arene-arene binding with Phe178 ( Figure 12). Finally, the β-lactam imipenem formed H-bonds with Lys134 and Arg128 in addition to arene-H bindings with Phe610 and Phe628 in the hydrophobic pit ( Figure 13). The monobactam aztreonam was flexibly bound towards the exit and the entrance of DBP, forming H-bonds with Arg128, Lys134, gln176, Gln46, Lys151, Ser87 and Arg620. Arene-H binding was also formed with Phe610 and arene-arene binding with Phe178 ( Figure 12).
Pathogens 2020, 9, x FOR PEER REVIEW 9 of 13 Figure 11. Two-dimensional image of the binding interactions of ciprofloxacin in the mexB active site with Phe628, Phe178, Phe610, Arg620 and Arg128 as key binding amino acids.
The monobactam aztreonam was flexibly bound towards the exit and the entrance of DBP, forming H-bonds with Arg128, Lys134, gln176, Gln46, Lys151, Ser87 and Arg620. Arene-H binding was also formed with Phe610 and arene-arene binding with Phe178 ( Figure 12). Finally, the β-lactam imipenem formed H-bonds with Lys134 and Arg128 in addition to arene-H bindings with Phe610 and Phe628 in the hydrophobic pit ( Figure 13). Finally, the β-lactam imipenem formed H-bonds with Lys134 and Arg128 in addition to arene-H bindings with Phe610 and Phe628 in the hydrophobic pit ( Figure 13). Figure 12. Two-dimensional image of the binding interactions of aztreonam in the mexB active site with Arg128, Lys134, gln176, Gln46, Lys151, Ser87, Arg620, Phe610 and Phe178 as key binding amino acids.
Finally, the β-lactam imipenem formed H-bonds with Lys134 and Arg128 in addition to arene-H bindings with Phe610 and Phe628 in the hydrophobic pit ( Figure 13). The limitation of this study: The presented data showed the presence of the genes in the genomes and provide no insight into the level of their expression. Further studies are needed to analyze conservation of the genes sequence in the genomes (their presence) of diverse clinical isolates and correlate it (their presence) with antibiotic resistance. Our study also did not discuss the correlation between the presence of single or both mexAB and MDR phenotype as we did not detect mexAB in non MDR isolates by PCR.

Materials and Methods
This cross-sectional descriptive study was carried out during the period from January 2019 to December 2019. A total of 180 strains of P. aeruginosa were isolated from patients with Healthcare Associated Infections in the Suez Canal University Hospital. Samples were processed in microbiology laboratory, Faculty of Medicine, Suez Canal University.

Bacterial Isolation and Identification
The collected samples were cultured on blood, MacConkey agar, cetrimide agar (Oxoid, UK) and Pseudomonas P agar media. Colonies were identified as P. aeruginosa by colony morphology, Gram stain and different biochemical reactions [35]; (P) aeruginosa ATCC 27853 was used as a reference strain. The isolates were preserved at −80 • C in glycerol 15% in brain heart infusion broth (BHIB, Oxoid, Basingstoke, UK) and then subculturing in BHIB at 37 • C for 24 h.

Antimicrobial Susceptibility Testing
Antibiotic susceptibility testing was performed by Kirby-Bauer disk diffusion method and interpreted according to the Clinical Laboratory Standard Institute (CLSI) guidelines [36]. The antibiotics tested were ciprofloxacin, aztreonam, cefepime, tobramycin, ceftazidime, gentamicin, amikacin, imipenem and meropenem (Oxoid, Basingstoke, UK). Interpretation of susceptibility testing was performed according to CLSI [36].

Role of Efflux Pump in Antibiotic Resistance
The MICs of ciprofloxacin, imipenem, ceftazidime and gentamicin were evaluated for all MDR and PDR P. aeruginosa isolates. Then, CCCP was added in a concentration of 10 µM to each Mueller-Hinton agar plates containing 0.5 to 128 µg/mL concentration of each antibiotic used. The CCCP final concentration in the Mueller-Hinton agar was 25 µg/mL. Then, MIC was repeated. A plate with CCCP and no antibiotics was used as control. The reduction in MIC at least 4-fold of any antibiotics with CCCP indicates the presence of efflux pump in isolates [37].

Molecular Detection of mexA and mexB Genes by PCR
DNA extraction was done using Qiagen DNA Mini kit 51,304. We used primers for detection of 16s gene. PCR conditions were adjusted as described by Pirnay and his colleagues [38]. MexA and mexB genes were detected as described before [39]. The PCR products were analyzed by agarose gel electrophoresis on 1.5% agarose (w/vol.) containing 0.5-mg/mL ethidium bromide (Qiagen, Germany) using a 100-bp DNA ladder was used as the size marker (Roche, Germany). Reaction mixtures without a DNA template served as negative controls. All primers and product lengths are shown in Table 1.

Statistical Analysis
All statistical analyses were performed using the statistical package for social sciences program (SPSS version 22 for windows). Statistical significance was considered at p-value ≤ 0.05.

Molecular Modeling and Docking
Docking studies were performed using the molecular operating environment (MOE) software and mexB of P. aeruginosa co-crystallized with the LMNG (PDB 6IIA) [40,41]. All minimizations were done until an RMSD gradient of 0.01 kcal/mol/A was obtained. The MMFF94 forcefield and partial charges were automatically calculated. The alpha triangle placement was chosen to determine the poses. The free energy of ligand binding from a given pose was estimated by the London ∆G scoring function. Refined results were rescored using the London ∆G scoring function. The output database dock file was created with different poses for each ligand and arranged according to the final score function (S).

Conclusions
In conclusion, the current study recognized the high MDR and PDR P. aeruginosa rates in burn and wound samples obtained from the Suez Canal University Hospital. In all MDR and PDR isolated strains Efflux mexAB genes were detected. The studied binding interactions of different drug categories help in the molecular study of the mexB. The molecular modeling could show the key amino acids in