Antibiotic Resistance Mediated by Escherichia coli in Kuwait Marine Environment as Revealed through Genomic Analysis

Antibiotic-resistance gene elements (ARGEs) such as antibiotic-resistance genes (ARGs), integrons, and plasmids are key to the spread of antimicrobial resistance (AMR) in marine environments. Kuwait’s marine area is vulnerable to sewage contaminants introduced by numerous storm outlets and indiscriminate waste disposal near recreational beaches. Therefore, it has become a significant public health issue and warrants immediate investigation. Coliforms, especially Gram-negative Escherichia coli, have been regarded as significant indicators of recent fecal pollution and carriers of ARGEs. In this study, we applied a genome-based approach to identify ARGs’ prevalence in E. coli isolated from mollusks and coastal water samples collected in a previous study. In addition, we investigated the plasmids and intl1 (class 1 integron) genes coupled with the ARGs, mediating their spread within the Kuwait marine area. Whole-genome sequencing (WGS) identified genes resistant to the drug classes of beta-lactams (blaCMY-150, blaCMY-42, blaCTX-M-15, blaDHA-1, blaMIR-1, blaOKP-B-15, blaOXA-1, blaOXA-48, blaTEM-1B, blaTEM-35), trimethoprim (dfrA14, dfrA15, dfrA16, dfrA1, dfrA5, dfrA7), fluroquinolone (oqxA, oqxB, qnrB38, qnrB4, qnrS1), aminoglycoside (aadA2, ant(3’’)-Ia, aph(3’’)-Ib, aph(3’)-Ia, aph(6)-Id), fosfomycin (fosA7, fosA_6, fosA, fosB1), sulfonamide (sul1, sul2, sul3), tetracycline (tet-A, tet-B), and macrolide (mph-A). The MFS-type drug efflux gene mdf-A is also quite common in E. coli isolates (80%). The plasmid ColRNAI was also found to be prevalent in E. coli. The integron gene intI1 and gene cassettes (GC) were reported to be in 36% and 33%, respectively, of total E. coli isolates. A positive and significant (p < 0.001) correlation was observed between phenotypic AMR-intl1 (r = 0.311) and phenotypic AMR-GC (r = 0.188). These findings are useful for the surveillance of horizontal gene transfer of AMR in the marine environments of Kuwait.


Introduction
The excessive use of antibiotics to treat infectious diseases has legated the world to the public health hazard of antibiotic resistance [1][2][3].Aquatic environs have been identified as reservoirs of antibiotic resistancegenes (ARGs); however, little is known about their distribution, spread and migration [4,5].Marine environments are also sinks for pharmaceuticals, disinfectants, heavy metals, organic compounds, microplastics, and atmospheric dust [6][7][8][9][10][11].Among these, pollutants, pharmaceuticals, and heavy metals have been evidenced to impose selective pressure on inherent bacterial communities that often develop resistance genes against them [9,[11][12][13][14][15][16][17][18].In addition, genetic elements, for example, class 1 integrons, package ARGs into gene cassettes (GCs) and mediate their transfer to non-resistant bacterial communities via horizontal gene transfer (HGT) [19][20][21][22].Although more consideration has been given to animal and human health with respect to antimicrobial resistance (AMR) monitoring, the environment often remains ignored.The World Health Organization propelled the One Health Concept, which calls for the evaluation of overall environmental health [23].Immediate action is also warranted to monitor the threats of AMR in the Gulf Cooperation Council [24,25].
Emergency waste and unauthorized sewage discharge in Kuwait have introduced several antibiotic-resistant bacteria (ARBs) into Kuwait's marine environments [18,26].Among these, Escherichia coli could become an infectious animal and human bacterium showing resistance to almost all clinically used antibiotics.A wide profile of AMR E. coli has been previously demonstrated through conventional phenotypic antibiotic susceptibility testing (AST) in Kuwait's marine environment [27][28][29].However, data on the genes conferring resistance are lacking in these aquatic settings.In addition, class 1 integrons, GCs, and plasmids involved in their spread are unknown.Currently, microbiological methods of AMR surveillance are considered laborious, time-intensive, and provide limited phenotypic information.With the advent of molecular methods, a rapid and detailed genotypic assessment of all these components can be achieved [30][31][32][33][34].These methods are now being embraced as a novel approach to map the whole genomes of bacterial isolates, ARGs, plasmids, integrons, and GCs simultaneously to complement the AMR phenotypic assays [32,[35][36][37].The goal of this research was to examine the genomic profiles of E. coli through whole-genome sequencing and to study the prevalence of ARGs, plasmids, integrons (intl1), and GCs.

Results
For the present investigation, selected E. coli isolates (n = 23) were subjected to wholegenome sequencing.All sequences ranged between 4,577,350 and 5,103,695 bases, with a Phred quality score above 20.

MLST and Phylogenetic Analysis
The open reading frames corresponding to seven sequence tags (STs), namely, adk-fumC-gyrB-icd-mdh-purA-recA, were identified in the assembled genomes of 23 E. coli isolates.Alignment of sequences against these STs established the Enterobacteriaceae origin of the strains.Further treatment with the mafft software and UGENE software confirmed 20 isolates as E. coli, and three of these closely matched with Enterobacter cloacae (SE111, SE181, and SE158).Molecular data analyses helped to correctly identify the species of E. cloacae; hence, we found the molecular tests to be technically superior for discriminating between closely related E. coli and E. cloaca strains.Further analysis was performed on the 20 E. coli isolates.The phylogenetic relationships of the selected E. coli were distributed into two clusters (Figure 1).A close relationship was observed between the isolates from spatially distant locations, different seasons, and varied matrices (marine waters or mollusks).For example, SE25 and KHE11 isolated from Al-Salam and Khiran were closely related.Strains SE19 and SC118 were isolated from different matrices, i.e., marine water during winter and mollusks collected during summer; however, they were positioned on the same branch of the phylogenetic tree for their similarity.Similarly, SC59 and 756E0 were also analogous.It was noticed that the three isolates of E. cloacae, namely, SE111, SE181, and SE158, exist separately as a group.

Figure 1.
Phylogeny of the strains isolated from Kuwait.A maximum-likelihood tree was plotted on the distances between the MLST sequences of the selected strains.

Antibiotic-Resistance Gene Elements
We submitted the assembled E. coli sequences to the ResFinder database and found matches against 33 ARGs (Figure 2; Table S1).The percentage identity of the sequences ranged between 80 and 100%.The highest frequency (80%) was that of the mdf (A) gene found in at least 16 isolates.This was followed by blaTem-1B, present in 45% of E. coli strains.

Antibiotic-Resistance Gene Elements
We submitted the assembled E. coli sequences to the ResFinder database and found matches against 33 ARGs (Figure 2; Table S1).The percentage identity of the sequences ranged between 80 and 100%.The highest frequency (80%) was that of the mdf (A) gene found in at least 16 isolates.This was followed by bla Tem-1B , present in 45% of E. coli strains.
We compared the ARG summaries of these 20 strains with their AMR phenotypes (tested against a panel of 23 antibiotics).According to the genotypic assay, all these strains possessed ARGs; therefore, they were considered as potentially resistant to at least one of the drug classes.Phenotypically, the strains SC59, HE33, 7SE60, SE124, and CW140 were sensitive to all the antibiotics (Figure 3).
The alignment of sequences against the plasmid finder database revealed 13 isolates as hosts to 22 plasmids (Table S2).The most prevalent was the ColRNAI plasmid found in 11 isolates (42.3%) (Figure 4).Sequences were then concomitantly mined for integrons, recombination sites, and promoters.The genetic elements identified were attC (gene cassette recombination sites), intI (intersection tyr-integrase), Pc_1 (gene cassette promoter class), Pint_1 (integron promoter), and attI_1 (integron recombination site).The elements attC and intI were reported in 20% (4/20) of the strains, whereas Pc_1 and attI_1 were detected in 10% (2/20) of the isolates.Pint_1 was found in 15% (3/20) of the tested samples.Isolates CW138, CW121, and CW141 were positive for integron with the integron integrase (intI) and attC site nearby, whereas HE70 was positive for the latter only, lacking recombination sites (Table 1).These elements aid the mobilization of ARGs in other strains.The intl1 gene encodes the integrase enzyme, the promoter Pc ensures its expression, and the site-specific recombination takes place between the attl and attC sites.The GC is an open reading frame (ORF) without a promoter but with a recombination site.These GCs can integrate novel genes and support the bacterial strains to adapt.On the other hand, strain SC70 possessed only a cluster of attC sites without any integrase in its vicinity (Table S3).Classification of integrons and gene cassettes revealed the presence of class 1 integron (intl1) and GC1.All the genetic elements within strain CW141 are shown in Figure 5. Intriguingly, this strain possessed the fosfomycin-resistant gene (fosB1) and expressed phenotypic AMR against piperacillin (an extended spectrum beta-lactam) and trimethoprim, as well as harbored the incX1 plasmid.Colored boxes mark the presence of the plasmids and their percentage identity in the database (red > 80%; yellow > 90%; green > 100%; white < 0.0%).Plasmids were fetched from the PlasmidFinder database.

Discussion
WGS has become the most preferred method to identify ARGs and associated genetic elements [38].Recently, it identified ARGs in both highly resistant and susceptible isolates in the order of Enterobacterales [38].This indicates that a population of resistant E.coli is floating in the coastal waters of Kuwait [36,39].In addition to resistance profiling, WGS was also used presently for the identification of Enterobacteriaceae species.Interestingly, three of the isolates (13.0%) were identified as Enterobacter cloacae.A study from the Norwegian coast demonstrated that MALDI-TOF-MS (Matrix-assisted laser desorption ionization-Time of flight mass spectrometry) identified 90% of isolated strains as E. coli, whereas the remaining 10% were Klebsiella, Citrobacter, and Enterobacter [40].More robust and precise identification approaches such as molecular methods are preferred over conventional microbiological assays, especially in mollusks and other filter feeders in the marine environment that are prone to pathogenic exposure, in addition to E. coli, for accurate discrimination between closely related species [41].Further, the MLST-based phylogeny established a close relationship between the 20 isolates.This indicates a common origin of

Discussion
WGS has become the most preferred method to identify ARGs and associated genetic elements [38].Recently, it identified ARGs in both highly resistant and susceptible isolates in the order of Enterobacterales [38].This indicates that a population of resistant E.coli is floating in the coastal waters of Kuwait [36,39].In addition to resistance profiling, WGS was also used presently for the identification of Enterobacteriaceae species.Interestingly, three of the isolates (13.0%) were identified as Enterobacter cloacae.A study from the Norwegian coast demonstrated that MALDI-TOF-MS (Matrix-assisted laser desorption ionization-Time of flight mass spectrometry) identified 90% of isolated strains as E. coli, whereas the remaining 10% were Klebsiella, Citrobacter, and Enterobacter [40].More robust and precise identification approaches such as molecular methods are preferred over conventional microbiological assays, especially in mollusks and other filter feeders in the marine environment that are prone to pathogenic exposure, in addition to E. coli, for accurate discrimination between closely related species [41].Further, the MLST-based phylogeny established a close relationship between the 20 isolates.This indicates a common origin of E. coli, most probably one of the point sources of emergency outfalls.The genetic similarity between the E. coli isolates of seawater and mollusks is most probably due to the filter feeding mechanism of the latter [42].A comprehensive analysis might assist in contact tracing of these AMR strains.
Intriguingly, the E. coli isolates (n = 20) possessed genes resistant against seven drug classes.This is most likely due to the selective pressure imposed by antibiotics such as Azithromycin, Cefalexin, Ciprofloxacin, Clarithromycin, Dimetridazole, Erythromycin, Metronidazole, Metronidazole-OH, Ofloxacin, Ronidazole, Sulfamethoxazole, Tetracycline, and Trimethoprim found in seawater samples [8] and the effluent streams of waste-water treatment plants in Kuwait [43].Moreover, the presence of conjugative plasmids and integrons in these strains is indicative of ARG transmission through horizontal gene transfer [19,20,22,44,45].Nevertheless, this study only targeted the fecal coliforms, and the likelihood of other microbial communities with diverse ARG profiles is expected.The advanced method of high-throughput shotgun metagenomic sequencing would be a more valuable tool in the assessment of resistomes at contaminated and non-contaminated sites [26,38].
The presence of beta-lactam genes and their phenotypic expression in strains such as HE40, SC63, SC70, SC118, SE25, SE19, KHE11, SE138, and CW138 draws our immediate attention.Moreover, plasmid vectors were also recorded in HE40, SC70, SE25, SE19, SE138, and CW138.In addition, the strains HE40 and CW138 also possessed integrons and gene cassettes, indicating the possibility of horizontal gene transfer of beta-lactamase genes in the region.Extended spectrum beta-lactams are the last-choice drugs to treat hardcore bacterial infections [46,47].Worldwide circulation of beta-lactamases have been reported recently with more than 40% in Asian countries [48].A baseline study also documented the presence of beta-lactams in aerosols collected from a public building situated near the coast of Kuwait [49].Moreover, trimethoprim resistance was confirmed in strains KHE11 and CW138.The dominance of the MFS-type drug efflux gene (tetracycline, disinfectants, antiseptics) was also noticed in 75% of sequenced isolates.Tetracycline-resistant genes (tetA/tetB) were found in 55% of isolates.The AMR phenotypes for these antibiotics were missing and need to be taken into consideration while performing future phenotypic assays.
We also report that 35.99% of AMR E.coli held the intl1 gene, in parallel with Enterobacteriaceae found in Korea (41.4%) [50].Contrastingly, only 12.1% of Enterobacteriaceae strains were positive for class 1 integrons in an effluent handling system in Poland [51].Intl1 positive strains were maximum in locations proximal to the coastal emergency waste outfalls.Numerous storm outlets discharge emergency waste into the country's marina, leading to deteriorated water quality [24,27,52,53].These findings are in congruence with our earlier reports on the presence of AMR microbes, ARGs, integrons, plasmids, and insertion sequences near the outfalls [18,26].We attribute this to the antibiotics, pharmaceuticals, and other contaminants introduced into the seawater [8,43].A constant selection pressure is thought to elevate the horizontal gene transfer frequency within these aquatic milieus [25,33,36,45,54,55].Regression analysis predicted a relationship (7.5% variance) between intI1-GC resistance.This suggests that the class 1 integron could potentially play a role in the transmission of AMR in Enterobacteriaceae [45]; however, other contaminants, such as metals, cannot be ignored [5,[56][57][58].
In the present investigation, WGS identified ARGs in both resistant and susceptible E. coli phenotypes.The presence of intl1 and GCs indicates the HGT of AMR in the region.WGS is recommended as an accurate and precise monitoring tool for individual AMR strains perpetrating in marine ecosystems.

Sampling Locations, Collection and Isolation of Strains
A detailed overview of the sampling task is provided elsewhere [28] as they were previously collected and tested for AMR through standard microbiological methods.However, for ease of readership, a succinct description is provided in Sections 4.1.1-4.1.3.A detailed description of the genomic analysis performed in the current study has been described in Sections 4.2-4.5.

Marine Waters
Seawater was collected from Al-Ghazali (29  387640 E) in an earlier campaign [28].Six replicates were sampled in 200 ml sterile bottles from each site in both summer (Jul-Aug 2015; 34.0 • C ± 2.0) and winter (Dec-Feb 2016; 17.2 • C ± 2.0) seasons (Table S4).In situ measurements of pH (8.2 ± 0.2), salinity (42 ± 2.0), and temperature were made employing a hand-held, multi-probe water quality meter (Hanna instruments, Smithfield, RI, USA).Samples were transported on ice to the Environment Public Authority (EPA) Kuwait laboratories.Samples were processed at EPA laboratories following an established protocol [36].Briefly, 10 mL of seawater was pipetted into 1 /4 strength Ringer's solution and serially diluted (10 to 10 5 ).Serial dilutions were passed through 0.45 µm Merck filters (Merck Life Sciences Limited, Dorset, UK).The filters were placed on Tryptone Bile X-Glucuronide chromogenic media (TBX) selective for E. coli.The Petri dishes were placed inverted at 30 • C for 4 h and 44 • C for 21 h.The plates were examined for the growth of blue-green colonies marking the presence of E. coli.A total of 351 strains were isolated from the collected water samples.

Mollusk Samples
Concurrent to the seawater sampling, mollusk samples were gathered in the summer (Jul-Aug 2015; 34.0 • C ± 2.0) and winter seasons (Dec-Feb 2016; 17.2 • C ± 2.0).Shells (n = 55-60) were hand-picked, sorted in autoclaved bags (4 • C), and transported to the College of Science, Kuwait University laboratories for further processing [28].The mollusc shells were removed to collect the flesh and fluid in a pre-autoclaved beaker (Borosil, Poole, UK).Approximately 20 g was macerated with 0.1% peptone water (Sigma, Plymouth, UK) and serially diluted [28].The dilutions were passed through the 0.45 µm nitrocellulose membrane filters and placed on TBX-agar for E. coli, as mentioned previously.In total, 247 mollusc strains were isolated.

Antimicrobial Susceptibility Testing
Antibiotic susceptibility assays for all these isolates were performed, and minimum inhibitory concentration (MIC) was recorded and presented in Al-Sarawi et al. [28].Briefly, the isolates were tested for susceptibility and resistance against a panel of 23 antibiotics as per the guidelines of Clinical and Laboratory Standards Institutes [59].The micro-dilution method was employed to determine the MIC by inoculating a loopful of isolate into 10 mL of Cation-Adjusted Muller Hinton (CAMBH) broth (incubated on shaker incubator for 24 h at 37 • C).Turbidity was adjusted with 0.5 M MacFarland solution and loaded to custom, dehydrated, 96-well Sensititre TM plates (GN2F, Thermo Scientific, Paisley, UK), which were further incubated at 25 • C for 24 h.Strains were classified as resistant, intermediate, or susceptible based on the MIC breakpoints standardized by CEFAS [27].

Whole-Genome Sequencing (WGS) and Filtering of ARGEs
From the 598 isolates, selected strains (n = 23, Table S5) with interesting AMR phenotypes, i.e., resistant to 2 or more antibiotics Al-Sarawi et al. [28] or susceptible to all, were subjected to WGS.These isolates were also chosen to represent varied spatiotemporal variability (three sites and two seasons).Whole-genome DNA was isolated from an over-nightly activated broth.Approximately 15 mL of Tryptic Soy Broth was seeded with a loopful of E. coli and incubated at 37 • C. The culture tube was centrifuged at 13,000× g to pellet the organisms and subjected to DNA purification as per the standard protocol of Promega Wizard ® Genomic DNA extraction Kit (Promega, Chilworth Southampton, Hampshire, UK) [60].Fluorometric quantification of DNA was conducted on a Quantus fluorometer (Promega, Chilworth Southampton, Hampshire, UK).Sequencing libraries were prepared following the Nextera XT kit (Illumina, USA).The libraries were purified using AMPure ® XP (Beckman Coulter, Brea, CA, USA) magnetic beads.Libraries were normalized and pooled before loading on a MiSeq V3 sequencing cartridge.Sequencing was performed on an Illumina MiSeq platform available at the Centre for Environment Fisheries and Aquaculture Science (CEFAS), Weymouth laboratories through paired-end (2 × 300 bp) chemistry [36].Post-sequencing, the Illumina adapters were removed by Trimmomatic v 0.36 [61] and a quality check was performed on the web version of FastQC v 0.11.5 [62].Raw reads were thereafter aligned in Spades v 3.10.1 for prokaryotic assembly [63] and annotated via Prokka v1.11 [64].The ABRicate tool [65] was used to screen the 'assembled fasta' for antibiotic-resistant genes (ARGs), employing the ResFinder v 4.0.0 [66] database (accessed on 2 April 2023).The sequences were also mined for plasmids through the plasmid finder database [67].The integron finder database v 2.0.2 was used to annotate the complete integron with integron integrase, attC recombination sites, promoters, and nearby gene cassettes [68].

Statistical Analysis
Statistical analysis was performed in SPSS 23.0 (IBM, SPSS, Hampshire, UK).The normality of data was tested through the Shapiro-Wilk test.A t-test was run to check for the no-normality violation and homogeneity assumptions.Data were considered significant at a confidence interval of 99.95% (p < 0.05).Regression tests applying rectilinear models were employed to establish the correlation between intl1-GC resistance.Homogeneity of variance, data normality, and dimensions of linearity were also explored.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/antibiotics12091366/s1.Table S1: Antibiotic resistance genes filtered through ResFinder; Table S2: Plasmids found in selected isolates of E. coli; Table S3: Integrons and associated elements filtered from the integrall database.;Table S4: GPS coordinates and physicochemical parameters of sampling sites; Table S5: Metadata of E. coli isolates used for sequencing.

Figure 1 .
Figure 1.Phylogeny of the strains isolated from Kuwait.A maximum-likelihood tree was plotted on the distances between the MLST sequences of the selected strains.

Figure 2 .
Figure 2. ARGs distributed in the E. coli genomes.Colored boxes are positive for the respective ARG listed on the left-hand side panel.The values in each box show the percentage identity in the ResFinder database v4.0.0 (red > 80%; yellow > 90%; green > 100%; white < 0.0%).

Figure 3 .
Figure 3. Genotypic versus phenotypic resistance in selected isolates of E. coli.Phenotypes are presented on the left-hand side panel.The color codes represent Red-Resistant, Green-Susceptible, and White-Intermediate.Corresponding genotypes are shown on the right-hand side.

Figure 3 .
Figure 3. Genotypic versus phenotypic resistance in selected isolates of E. coli.Phenotypes are presented on the left-hand side panel.The color codes represent Red-Resistant, Green-Susceptible, and White-Intermediate.Corresponding genotypes are shown on the right-hand side.

Figure 4 .
Figure 4.The plasmids found in selected isolates of E. coli revealed by whole-genome sequencing.Colored boxes mark the presence of the plasmids and their percentage identity in the database (red > 80%; yellow > 90%; green > 100%; white < 0.0%).Plasmids were fetched from the PlasmidFinder database.

Figure 4 .
Figure 4.The plasmids found in selected isolates of E. coli revealed by whole-genome sequencing.Colored boxes mark the presence of the plasmids and their percentage identity in the database (red > 80%; yellow > 90%; green > 100%; white < 0.0%).Plasmids were fetched from the PlasmidFinder database.

Figure 5 .
Figure 5. Graphical representation of integrons, recombination sites, and promoters present in E. coli strain CW141.

Figure 5 .
Figure 5. Graphical representation of integrons, recombination sites, and promoters present in E. coli strain CW141.

Figure 6 .
Figure 6.(A) Prevalence of class 1 integrons and gene cassettes in E. coli isolated from the Kuwait marine waters and molluscans.(B) The occurrence of class 1 integrons in E. coli isolated during summer and winter seasons.(C) AMR-susceptible E. coli with intl1.

Figure 6 .
Figure 6.(A) Prevalence of class 1 integrons and gene cassettes in E. coli isolated from the Kuwait marine waters and molluscans.(B) The occurrence of class 1 integrons in E. coli isolated during summer and winter seasons.(C) AMR-susceptible E. coli with intl1.