Molecular epidemiology of antibiotic-resistant genes and potent inhibitors against TEM, CTX-M-14, CTX-M-15, and SHV-1 proteins of Escherichia coli in district Peshawar, Pakistan

The Extended Spectrum Beta-Lactamases (ESBLs) producing bacteria is an issue of concern for clinicians resulting in minimize the treatment options. To overcome resistance mechanisms, novel inhibitors with good Absorption, Distribution, Metabolism, Excretion, and Toxicity (ADMET) properties must inhibit the ESBLs resistant genes. The current study aimed to identify the antibiotic resistance genes of ESBLs producing E. coli and a single inhibitor was designed to inhibit all the resistant proteins. The results showed that 42.9% ESBL producers had CTX-M (69.9%), TEM (63.4%), SHV (34.5%) and CTX-M-14 (17.5%) genes. The ESBLs producing isolates were resistant to cephalosporins, quinolones, and sulfonamide with Minimum Inhibitory Concentration (MICs) ranging from 64 to >256 μg/ml. To design multi inhibitory ligands, RECAP synthesis was used for the de-novo discovery of 1000 inhibitors database. Protein crystal structures were retrieved from Protein Data Base (PDB). Lipinski’s rules of five were applied to the novel inhibitors database to improve the ADMET properties. The novel inhibitors database was selected for docking simulations. Placement of the ligand was used by the London dG algorithm implemented in Molecular Operating Environment (MOE), while GBVI/WSA dG algorithm was used for final refinement. Based on docking score, visual inspection of ligands interaction with key residues, binding affinity, and binding energy of ligands with proteins, ten compounds were selected for ESBLs proteins with best ADMET properties, binding energy, and binding affinity the reported ones. These hits compounds have unique scaffolds and are predicted to be a starting point for developing potent inhibitors against antibiotic-resistant proteins.


Introduction
Escherichia coli are Gram-negative bacilli belong to Enterobacteriaceae, commonly present in the human gut. It comprises nonpathogenic strains that form part of the normal flora, but some pathogenic strains can cause life-threatening infections in humans. The most common diseases caused by E. coli are; Urinary Tract Infections (UTIs) (Foxman, 2010), enteric infections, bacteremia, nosocomial pneumonia, cholecystitis, cholangitis, peritonitis, cellulitis, osteomyelitis, infectious arthritis, and neonatal meningitis (Kim, 2012). These infections are commonly treated with antibiotics, but improper use of these antibiotics has developed resistance in E. coli strains resulting in severe health issues worldwide (El Kholy et al., 2003). The prevalence and antibiogram of E. coli reveal massive global variations in different populations (Erb et al., 2007). A high rate of antibiotics resistance has been https://doi.org/10.1016/j.sjbs.2021.07.028 1319-562X/Ó 2021 The Author(s). Published by Elsevier B.V. on behalf of King Saud University. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). genes was also observed in Lahore, Pakistan (Abrar et al., 2019). In Pakistan, the high frequency of ESBLs-related infections has been reported in the last 20 years (Abrar et al., 2017).
After completing molecular characterization, a variety of new potent drugs have been discovered to overcome antibiotic resistance of ESBLs-Ec. In the drug discovery process, combinatorial chemistry is a well-established strategy for generating new lead molecules. Retrosynthetic Combinatorial Analysis Procedure (RECAP) Analysis and Synthesis is a new computational tool used to create fragments molecules and recombine these molecules in various ways to synthesize reasonable novel chemical structures (Lewell et al., 1998). It can give compounds with high hit rates during screening libraries of in-silico compounds compared to traditional large-scale screening (Shekhar, 2008). Thus, with the help of available computational tools, antibiotic resistance can be efficiently handled and will support clinicians to overcome the resistance mechanisms.
Therefore, the current study was designed to determine the antimicrobial susceptibility patterns of ESBLs producing E. coli and search for novel inhibitors, as it is necessary regularly for adequate clinical treatment and management.

Samples collection and inoculation
A total of 3278 clinical samples were collected from blood, Cerebrospinal Fluid (CSF), urine, pus, sputum, and infected wounds and transported to the microbiology laboratory of Khyber Teaching Hospital Peshawar for further processing. Patients undergoing antibiotic therapy or suffering from enteric infections were excluded from the study. The collected samples were inoculated on Cysteine Lactose Electrolyte-Deficient medium (CLED), MacCon-key agar, blood agar and incubated at 37°C for 24hrs (Cheesbrough, 2006). Specimens like CSF, pus, wound, and respiratory specimens like sputum were inoculated on blood agar and MacConkey agar plates and incubated at 37°C for 24hrs. Samples like blood, CSF, and other body fluids were inoculated in a culture medium of automated blood culture system (Versa Trek, USA). The inoculated bottles were subjected to incubation at 37°C for five days and observed for gas production. Pure isolated colonies were analyzed for their morphology and further subjected to Gram staining. Later on, these bacterial cultures were subjected to different biochemical tests using Analytical Profile Index (API) 10S strips for confirmation (Alnahass et al., 2016). Moreover, these phenotypically identified isolates were confirmed by PCR using specific primers for the uidA gene.

Phenotypic screening for ESBLs production
In phenotypic detection, the isolates resistant to CAZ or CTX were subjected to E-test for ESBLs production. The test organism was inoculated on Mueller-Hinton Agar (MHA) plate. The E-test strips were placed on the plates and incubated at 37°C for 24hrs. A ratio of the MICs of the cefotaxime/inhibitor (CT/CTL), ceftazidime/inhibitor (TZ/TZL), and cefepime (PM/PML) of !8 or the presence of a phantom zone was interpreted as positive for ESBLs production (Cormican et al., 1996).

Antimicrobial susceptibility testing
All isolates were subjected to antimicrobial-susceptibility testing against selected antibiotics by Kirby Bauer disc diffusion technique. The bacterial suspension was inoculated on MHA plates and the antibiotic discs were placed on plates at equal distances. The results were interpreted after overnight incubation at 37 C by measuring the inhibition zone diameter as per Clinical and Laboratory Standard Institute (CLSI) 2019 guidelines (Cusack et al., 2019).

Determination of minimum inhibitory concentrations
The MICs test strips were used for the determination of MICs of selected antibiotics. The strips containing an exponential gradient of antimicrobial agents were placed on the surface of an inoculated MHA plate. These plates were incubated for 24hrs at 37°C, and the MICs were recorded immediately from the scale (lg/mL) at the point where the edge of the inhibition ellipse intersects the MIC strip (Yan et al., 2002).

Molecular identification of E. coli
All the isolates were sub-cultured on blood agar and incubated overnight at 37 C. The genomic DNA was extracted using a GeneJet genomic DNA purification kit (Thermo Scientific, USA). The DNA integrity was determined by gel electrophoresis in 1% agarose (w/v), and purified DNA was stored at À20°C (Bashir et al., 2012;Chen and Griffiths, 1998). The E. coli isolates were confirmed by amplifying the b-D-glucouronidase (uidA) gene using specific primers. The final PCR reaction was done in 27 ml final reaction volume containing; 12.5 ml Taq master mix (Bioron, Germany), 0.5 ml of forward and reverse primers, 2 ml of DNA template and PCR grade water. Amplification was carried out using specific primer sequences under the optimized conditions (Bashir et al., 2012) (Table 1).

Gel electrophoresis
The amplified products were electrophoresed at 110 V for 40-60 min on 1.5% agarose suspended in TAE buffer (1X). Gels were stained with ethidium bromide solution, and bands were visualized and photographed by gel documentation system (Bio-Rad Milan, Italy) using GeneSnap software. The amplicon sizes were determined by matching with DNA ladder (100 bp) (Gangoue-Pieboji et al., 2006).

Electrostatic maps
This application works on the principle of nonlinear Poisson-Boltzmann Equation (Eq. (1)). For predicting electrostatically preferred locations of H-bond acceptor, H-bond donor and hydrophobic atoms in the active site, electrostatic Maps were generated for each protein using 4.5 Å receptor grids MOE2019.0102. Before calculation, the system was partially charged (qi). Furthermore, this equation yields electrostatic potential (u) and then use to produce predictive maps as under.

Acceptor map
An isocontour of the potential (qO u + vO), in which vO indicates van der Waals potential while qO indicates the partial charge of the oxygen ion.

Donor map
An isocontour of the potential (qH u + vH), in which vH indicates van der Waals potential and qH is the partial charge of the hydrogen ion and vH is the van der Waals potential of the hydrogen atom.

Hydrophobic map
An isocontour of the potential -(qO + qH) u + vC in which vC indicates van der Waals potential of a carbon atom.
r:e r £ þ This application computes suitable positions like positive, neutral, and negative charges in the ligand and binding site of protein using Eq. (1). The hot spots present in the protein pocket indicate binding. In hot spots, red indicates Acceptor/Negative, blue for Donor/Positive, and green for Neutral/Hydrophobic. The protein crystal structures were recovered from the RCSB PDB server having ligand bind in it were selected for the electrostatic map to find the binding residues pattern in the active pocket.

Retrosynthetic combinatorial analysis procedure (RECAP) analysis and synthesis
To generate novel inhibitors database, 21 reported b-lactamases inhibitors were recovered from the Pubchem server (https://pubchem.ncbi.nlm.nih.gov/) with unique PubChem CID and were kept in the inhouse database in mdb format. RECAP Analysis and Synthesis implemented in MOE 2019 were applied on the abovereported inhibitors as part of a de novo discovery methodology. One thousand novel structures were generated and were saved in mdb format. All the structures in the database were energy minimized, and the hydrogens were added to it using the MOE2019.0102.

ADMET properties
To select inhibitors having good ADMET properties, Lipinski's rules of five were applied using the MOE 2019 descriptor calculator, on the output database generated from the RECAP synthesis (Lipinski, 2004).

Docking validation and simulations
Before docking the novel ligands in the proteins active site, the docking protocol of MOE2019.0102 was validated through redocking, the reported inhibitors in the crystal structures. The method of Rigid body pose generation was used for Proteinligand docking. The validation of docking was reported in RMSD of the re-docked and reported ligand superimposition. The retrieved output database from RECAP analysis and synthesis was docked in the active sites of each protein after the docking validation. To generate tens of thousands of ligand poses around the restraints to restrict the binding site region, the tool of Fast Fourier Transforms (FFT) was selected. The (rigid body) top 10 poses were refined with R-Field electrostatics, and the final Refined (top 10 poses) were generated based on equation-2 GBVI/WSA dG solvation model as output. Where the transitional and rotational entropy gain/loss average is represented by ''c", the forcefields dependent constants are ''a and b", to calculate the currently loaded charges using ei = 1 constant dielectric, represented by coulombic electrostatic E Coul , GB/VI solvation model E sol was used to calculate solvation electrostatic, E vdW showed the van der Waals contributions and the surface area exposed was calculated by SA weighted .
The docked compounds were saved in mdb format. Using the proxy triangle algorithm, 1 conformation was saved for each ligand using equation-3 London dG scoring methodology in MOE2019.0102 for refinement (Labute, 2008). Where rotational and translational entropy gain or loss is represented by ''c", ligand flexibility energy loss is E flex , hydrogen bonds geometric imperfections are f HB , ideal hydrogen bond energy is c HB , metal ligation imperfection is f M . In contrast, ideal metal-binding energy is c M , and the dissolution energy of i atom is D i .
The final ten compounds were selected based on the highest docking score, interaction with crucial residues visual inspection, binding energy, and binding affinity, potentially inhibiting all the MBLs proteins. These ten final hits have more tremendous binding energy, and binding affinity from the crystal structures reported inhibitors and can be used for experimental procedures.

Calculations of binding energy and binding affinity
The binding affinities were calculated using the equation-2 Generalized Born/Volume Integral (GB/VI) algorithm in MOE2019-0102 to determine the most potent ligand for each blactamases protein complex with ligands. All the non-bonded interaction energies are Generalized Born interaction between the protein residues, and the ligand molecules consist of coulomb electrostatic interaction. The binding energy and binding affinity were calculated and described in the unit (Kcal/Mol) for each hit after energy minimization (Wadood et al., 2017).

Molecular characterization of ESBLs gene (s)
The targeted ESBLs genes were detected in 242 (98.37%) isolates out of 246. These isolates were positive for one or more genes, while 04 isolates were negative for targeted genes. The Bla-CTX-M was the most prevalent gene, observed in 172(69.9%), TEM in 156(63.4%), SHV in 85(34.5%), while CTX-M-14 was observed in 43(17.5%) isolates (Table 2 and Fig. 1). After sequencing, the reference sequences were aligned using the NCBI Blast; Bla-TEM gene with accession number MK878892, Bla-SHV gene with accession number MH460805, Bla-CTX-M gene with accession number

Designing of potent ESBLs inhibitors
The resistance to antibiotics is an issue of concern and has involved interests for researchers worldwide. In the past, Insilco approaches have been used to predict strategies like targeting the protein crystal structures to design novel inhibitors to overcome the mechanism of resistance (Hassan Baig et al., 2016). To design multi inhibition small molecules for ESBLs proteins, crystal structures of the four selected proteins were retrieved from the RCSB PDB server. The reported ligand for the 1ERM Bla-TEM is Ethyl Boronic Acid, ruthenium (ligand) for 4XXR Bla-CTX-M-14, (2S, 5R)-1-formyl-5-[(sulfooxy) amino] piperidine-2-carboxamide (NXL) for 4HBU Bla-CTX-M-15, Meropenem for 2ZD8 Bla-SHV-1 is given in (Fig. 2).

Electrostatic maps
The electrostatic map was calculated for the active pocket of protein and ligand of all the selected beta-lactamase proteins (TEM, CTX-M-14, CTX-M-15, and SHV-1) having ligand in it (Fig. 3). The color of hydrogen bond acceptor areas is red, the hydrogen bond donor atoms are blue, and the hydrophobic interaction atoms are represented by green color.

Docking validation
The docking techniques were subjected for validation by docking the reported co-crystallized ligands of all targeted b-lactamases in the active pockets of each protein before docking (Fig. 4). The docked ligands of the proteins were in suitable range (0.3274 Å) for TEM, (0.5273 Å) for CTX-M-14, (0.8047 Å) for CTX-M-15 and (0.9144 Å) for Bla-SHV-1.

Docking simulation, prediction of binding energy and binding affinity
The residues interacting with the reported ligand in the crystal structure were selected as an active pocket of the proteins. The Novel compounds generated from RECAP synthesis were docked with the active pocket residues. The docked compound of 20% based on the lowest docking score was selected for visual inspection to identify the best interacting compounds. Based on docking score, binding energy, and binding affinity, the top 10 compounds were selected, which have promising interactions with selected proteins. To rank, the final compounds, binding energy, and binding affinity were calculated for each compound with the selected proteins shown in (Table 5). 3.9. Docking score, binding energy, and binding affinity for TEM protein The highest binding energy (À80.49 kcal/mol) and binding affinity (À9.48 kcal/mol) were reported for TEM of compound 275, followed by compound 783 having binding energy (À78.66 kcal/mol), binding affinity (À9.43 kcal/mol), compound 848 with binding energy (À7084 kcal/mol), binding affinity (À8.98 kcal/mol). All the ten reported compounds showed the best binding energy and affinity (Table 5). The docking result of compound 848 for TEM protein revealed that the highest docking score of À8.08 was observed interacting with Ser130, Ser235, Tyr105, Ala237, and Arg243 of the active pocket having high binding energy (À70.84Kcal/mol) and affinity (À8.98Kcl/mol). The compound 783 for Bla-TEM also documented a high docking score (À8.03), binding energy (À78.66Kcal/mol), and affinity (À9.43Kcal/mol), which interact with Ser130, Ser235, Tyr105, Asn132, Gly236, Ala237, and Arg243 of the active pocket. The compounds 848 and 783 recorded the best interaction with critical residues of the TEM active site (Tables 5 and Fig. 5).

Discussion
The frequency of ESBLs is increasing rapidly, develops an alarming condition because the ESBLs-Ec is challenging to inhibit due to resistance to antibiotics, cephalosporins, penicillins, and monobactam (Aboderin et al., 2009). The current study observed ESBLs production in 42.9% of E. coli isolates. The same results (35.8%) of ESBLs-Ec isolates were reported in a study conducted in Saudi Arabia (Hassan and Abdalhamid, 2014). Another study also noted the high prevalence of ESBL-Ec (36%) among clinical isolates in Pakistan, which aligns with our results (Pitout and Laupland, 2008). Another study conducted at Agha Khan University observed the exact prevalence (30%) of ESBLs-Ec isolates, according to the current research (Jabeen et al., 2005). Out of 246 isolates, 39.0% of patients were males, while 61% were females. Most of the ESBLs positive isolates were detected in hospitalized patients (53.2%) as compared to OPD patients (46.7%). The prevalence rates of ESBLs-Ec are different in different geographic areas among clinical isolates starting from low rates 3-8% in Sweden, Japan and Singapore to much higher rates 30-60% in America, 34% in Portugal, and 58% in Turkey (Paterson and Bonomo, 2005). Another study conducted in Pakistan also reported the high incidence (64%) of ESBLs-Ec isolates, supporting the current results. A much higher prevalence of ESBL detection (89.5%) was reported in Pakistan among clinically isolated E. coli. The results showed that most of the isolates were recovered from UTIs (61.4%) patients, followed by wound infected patients (pus) (29.3%), wound (1.6%), others (4.1%), upper and lower respiratory tract infections (3.3%) and chronic patients those developed ascites (1.22%). The results of the different clinical isolates of ESBLs-Ec revealed that 61.4% of patients had UTIs, 35.3% patients had SSIs, while 3.3% patients had severe pulmonary infections (PIs). However, these findings were in accordance with some other studies where the identical clinical isolates were recovered different specimens like; 46% from urine, 29% from pus, 11% from various fluids, and 4% from blood; while 12 isolates were obtained from throat swabs, catheter tips and sputum Mumtaz et al., 2011).
The antibiogram of the current study revealed that ESBLs-Ec isolates were 100% sensitive against TGC and MEM antibiotics, while high resistance was observed against FEP, CTX, ATM, CAZ, SXT, CIP, LEV, and AMC. The same findings were observed in which all the E. coli isolates were susceptible to IPM and MEM. The high resistance (89%) was recorded against CTX followed by CIP and FEP, which strongly supported the current study . The present study observed MICs of antimicrobial agents tested against ESBLs-Ec isolates, which showed resistance to blactam and non-b-lactam drugs. Lower MICs were also reported in Athens (Greece), in which the majority of the strains also showed resistance to b-lactam and non-b-lactam drugs, while for TGC, MIC 50 and MIC 90 were 2 and 4 mg/ml, respectively (Galani et al., 2012). The same findings were also observed in a study conducted in South India in which a high resistance profile of the isolates was obtained; the MIC 50 was high for cephalosporins and ATM, while MIC 50 was low for TZP, FOS, IPM, MEM, and AK (8/4, 1, 0.5, 0.12, 0.06 and 2 mg/ml), respectively (Rajenderan et al., 2014).
The CTX-M gene was observed in 69.9%, TEM in 63.4%, SHV in 34.5% while CTX-M-14 in 17.5% isolates. The high incidence of 57.7% was reported in CTX-M, 20.3% in TEM and 15.4% in SHV in another reported study, making it in line with current findings (Sastry et al., 2013). The results of ESBL positive isolates showed a high rise in CTX-M (97.4%) compared to SHV (23.1%), which supports the current study Abdalhamid, 2014, Bindayna et al., 2010). It was reported that 44.08% of the isolates harbored the CTX-M genes alone, while both CTX-M and SHV genes were detected in 20.43% isolates. The increased incidence of the E. coli (59.32%) isolates harbored the CTX-M genes, which support the current study (Kaur and Aggarwal, 2013). It was observed that comparatively, a higher level of ESBL-producers was reported in the current study. Other different studies revealed the prominently different prevalence of ESBLs resistance genes in various geograph- ical areas. This might be due to differences in drug selection and volume of ingestion of antibiotics and alterations in time of isolates collection.
In insilico drug drugs designing, the resistance to antibiotics is an issue of concern and has involved interests for researchers worldwide. In the past, insilico approaches have been used to predict strategies like targeting the protein crystal structures to design novel inhibitors to overcome the mechanism of resistance (Hassan Baig et al., 2016). The results of new compounds as inhibitors revealed that the reported ligand for the bla-TEM Ethyl Boronic Acid has binding energy (À56.72Kcal/mol) and affinity (À8.43 kcal/mol). The highest binding energy (À80.49 kcal/mol) and affinity (À9.48 kcal/mol) were reported for bla-TEM of compound 275, followed by compound having 783 binding energy (À78.66 kcal/mol), affinity (À9.43 kcal/mol), compound 848 with binding energy (À70-84 kcal/mol), affinity (À8.98 kcal/mol). These three compounds showed binding energy above À70 kcal/mol. All the ten reported compounds showed the best binding energy and affinity compared to the reported ligand Ethyl Boronic Acid. In the docking simulations, the final docking hits observed the best interactions with each protein (essential amino acids). The docking result of compound 848 for TEM protein revealed the highest docking score of À8.08, high binding energy (À70.84Kcal/mol), and affinity (À8.98Kcl/mol). The compound 783 for Bla-TEM also documented a high docking score (À8.03), binding energy (À78.66Kcal/mol), and affinity (À9.43Kcal/mol). The compounds 848 and 783 recorded the best interaction with crucial residues of TEM active site than the reported ligand Ethyl Boronic Acid (Table 5).

Conclusions
The ESBLs-Ec is an issue of concern for clinicians to treat resistant bacteria due to misuse and self-medications. Due to the emergence of resistance mechanisms, it is challenging for medical practitioners to treat patients with a limited choice of antibiotics. The researchers designed many new drugs for ESBLs-Ec to overcome the resistance mechanism. The proteins of ESBLs are the cause of different infections in both hospitalized and non-hospitalized patients. Thus it is necessary to overcome the resistance mechanism to treat the patients. In the current study, the in-silico approach was used to discover novel and potent inhibitors for TEM, CTX-M-14, CTX-M-15, and SHV-1 pro- teins. The study results revealed that 40 compounds were designed as final hits (10 for each protein). These hits compounds have unique scaffolds and are predicted to be starting points for developing novel and potent inhibitors for four antibiotic-resistant proteins.

Ethics approval
The current study was approved by the Institution Research and Ethical Review Board (IREB) of Khyber Medical College, Peshawar (Document No. 122/ADR/KMC).