Similarities in Virulence and Extended Spectrum Beta-Lactamase Gene Profiles among Cefotaxime-Resistant Escherichia coli Wastewater and Clinical Isolates

The World Health Organization has identified antibiotic resistance as one of the largest threats to human health and food security. In this study, we compared antibiotic resistance patterns between ESBL-producing Escherichia coli from human clinical diseases and cefotaxime-resistant environmental strains, as well as their potential to be pathogenic. Antibiotic susceptibility was tested amongst clinical isolates (n = 11), hospital wastewater (n = 22), and urban wastewater (n = 36, both influent and treated effluents). Multi-drug resistance predominated (>70%) among hospitalwastewater and urban wastewater influent isolates. Interestingly, isolates from clinical and urban treated effluents showed similar multi-drug resistance rates (~50%). Most hospital wastewater isolates were Phylogroup A, while clinical isolates were predominately B2, with a more diverse phylogroup population in urban wastewater. ESBL characterization of cefotaxime-resistant populations identified blaCTX-M-1 subgroup as the most common, whereby blaKPC was more associated with ceftazidime and ertapenem resistance. Whole-genome sequencing of a carbapenemase-producing hospital wastewater E. coli strain revealed plasmid-mediated blaKPC-2. Among cefotaxime-resistant populations, over 60% of clinical and 30% of treated effluent E. coli encoded three or more virulence genes exhibiting a pathogenic potential. Together, the similarity among treated effluent E. coli populations and clinical strains suggest effluents could serve as a reservoir for future multi-drug resistant E. coli clinical infections.


Introduction
Antimicrobial resistance (AMR) is a serious global health concern. In the United States alone, over 2 million AMR infections are reported annually, resulting in approximately 23,000 deaths [1]. Although AMR is frequently associated with the overuse and misuse of antibiotics in clinical settings, resistant bacteria and their genes are widespread within human, animal, and environmental settings, indicating the need for a One Health approach [2]. Wastewaters are ideal for the study of AMR as they serve as a bridge between human, animal, and environmental settings and have been frequently implicated in the acquisition and dissemination of AMR bacteria and genes [3,4]. Additionally, wastewaters are rich in selective pressures including antibiotics, pharmaceuticals, and biocides that can increase microbial horizontal gene transfer (HGT) and mutation rates, leading to an increase in AMR [5,6].
Wastewaters contain bacteria from a variety of sources including human waste, stormwater, and hospital sewage. Due to the widespread use of broad-spectrum antibiotics for disease treatment in hospital settings, hospital wastewater frequently contains and treated effluents (n = 17) from urban wastewater reclamation facility in the same city. The Milwaukee wastewater reclamation facility services 1.1 million people within a 681 km 2 area collecting and treating on average 265 mL/day of sewage during dry days. The sewage comprises domestic urban surface runoff, industrial, and hospital wastewater. The final treated wastewater is discharged into Lake Michigan within the Milwaukee Bay area (43.02329162715021, −87.89408612678749). Similar to a global analysis of disinfected WW effluents [22], no CTX-resistant E. coli isolates were identified in post-chlorinated WW effluents. The 69 E. coli isolates were tested for resistance against twelve different antibiotics. Hospital WW displayed the same or highest prevalence of antibiotic resistance compared to all other E. coli sources except for chloramphenicol (CHL), to which all isolates were fully susceptible (Figure 1). Although antibiotic resistance was not as prevalent among WW influents and effluents, more than 25% of isolates were resistant to five antibiotics (sulfamethoxazole-trimethoprim (SXT), tetracycline (TET), ceftriaxone (CRO), ceftazidime (CAZ), and CTX) (Supplementary Table S2).
infections) from Milwaukee, Wisconsin, U.S.A. All selected isolates exhibited at least intermediate resistance to third-generation cephalosporins or carbapenems. Clinical isolates were analyzed for their antibiotic resistance profiles and were compared to cefotaximeresistant E. coli isolates from WW effluents of the same hospital (n = 22), as well as influents (n = 19) and treated effluents (n = 17) from urban wastewater reclamation facility in the same city. The Milwaukee wastewater reclamation facility services 1.1 million people within a 681 km 2 area collecting and treating on average 265 mL/day of sewage during dry days. The sewage comprises domestic urban surface runoff, industrial, and hospital wastewater. The final treated wastewater is discharged into Lake Michigan within the Milwaukee Bay area (43.02329162715021, −87.89408612678749). Similar to a global analysis of disinfected WW effluents [22], no CTX-resistant E. coli isolates were identified in postchlorinated WW effluents. The 69 E. coli isolates were tested for resistance against twelve different antibiotics. Hospital WW displayed the same or highest prevalence of antibiotic resistance compared to all other E. coli sources except for chloramphenicol (CHL), to which all isolates were fully susceptible (Figure 1). Although antibiotic resistance was not as prevalent among WW influents and effluents, more than 25% of isolates were resistant to five antibiotics (sulfamethoxazole-trimethoprim (SXT), tetracycline (TET), ceftriaxone (CRO), ceftazidime (CAZ), and CTX) (Supplementary Table S2).
Multi-drug resistance (≥3 antibiotic classes), a major health threat, was evident among over 71% (49 or 69) of isolates with the highest prevalence among hospital WW and urban influent WW (Figure 1: 91% and 74% respectively). Most CTX-resistant hospital WW E. coli were also resistant to SXT, gentamicin (GEN), and ciprofloxacin (CIP) ( Supplementary Table S2), consistent with another report [23]. Resistance to these three antibiotics (SXT, GEN, and CIP), albeit at lower levels, was persistent throughout influent (68%, 58%, and 39%, respectively) and effluent (47%, 24%, and 12%, respectively). Although antibiotic resistant profiles were quite diverse within and among the sample sources (Supplementary Table S2), resistance to SXT, GEN, CIP, TET,  were the most prevalent among hospital and urban influent WW. Similar to other reports [24], hospital WW isolates had significantly higher multiple antibiotic resistance (MAR) indices than all other sources (Figure 2A: p < 0.05).

Figure 2.
Antibiotic resistance among sample sources and phylogroups. (A) Multiple antibiotic resistance (MAR) index was determined by dividing the number of resistant antibiotic categories by the total number of antibiotic categories (n = 7) per isolate. (B) Phylogroup of total isolates was determined using the revised Clermont method. Each mark represents one isolate. The mean of each group is represented by a bar. Statistical differences between sources or phylogroups was determined using Tukey's multiple comparison test. * p < 0.05; ** p < 0.001.
In addition to antibiotic susceptibility, we also categorized these E. coli by phylogroup [25], which aids in identifying a possible common bacterial host or origin. Most clinical isolates (64%) were attributed to phylogroup B2 (Supplementary Table S3), which correlated with the abundance of uropathogenic strains [26] amongst our clinical population. Interestingly, B2 was also present in WW influents (7%) and treated effluent (13%) isolates, though not observed in hospital WW (Supplementary Table S3). The majority of hospital WW was comprised of phylogroup A (77%), which typically associates as human commensal origin [27]. In agreement with other studies [19,28], WW influents and effluents showed a greater diversity in phylogroups (Supplementary Table S3) likely due to multiple sources combining into one, as well as WW persistent populations [27]. Amongst all the phylogroups, isolates typed to phylogroup A were significantly more resistant than those typed to B2 and D ( Figure 2B, p < 0.05). This finding could be attributed to microbial source, as phylogroup A comprised most samples from hospital WW, similar to other reports [19,29,30].

Prevalence of β-Lactamases among CTX-Resistant Populations
Among CTX-resistant E. coli populations from the clinic and environmental sources, we assessed the presence of seven different β-lactamases: CTX-M subgroups (1, 2, and 9), TEM, SHV, OXA, and KPC. They have become a global pandemic with ESBL prevalence increasing among commensals and clinical isolates [21,31], causing significant therapeutic challenges [32]. Women with ESBL-producing E. coli urinary tract infections are significantly more likely to contain the same ESBL-producing E. coli clone in their feces [33]. From a One Health perspective, the strongest correlation of ESBL genes among commensal and pathogenic isolates is with environmental sources, specifically surface and WW Phylogroup of total isolates was determined using the revised Clermont method. Each mark represents one isolate. The mean of each group is represented by a bar. Statistical differences between sources or phylogroups was determined using Tukey's multiple comparison test. * p < 0.05; ** p < 0.001.
In addition to antibiotic susceptibility, we also categorized these E. coli by phylogroup [25], which aids in identifying a possible common bacterial host or origin. Most clinical isolates (64%) were attributed to phylogroup B2 (Supplementary Table S3), which correlated with the abundance of uropathogenic strains [26] amongst our clinical population. Interestingly, B2 was also present in WW influents (7%) and treated effluent (13%) isolates, though not observed in hospital WW (Supplementary Table S3). The majority of hospital WW was comprised of phylogroup A (77%), which typically associates as human commensal origin [27]. In agreement with other studies [19,28], WW influents and effluents showed a greater diversity in phylogroups (Supplementary Table S3) likely due to multiple sources combining into one, as well as WW persistent populations [27]. Amongst all the phylogroups, isolates typed to phylogroup A were significantly more resistant than those typed to B2 and D ( Figure 2B, p < 0.05). This finding could be attributed to microbial source, as phylogroup A comprised most samples from hospital WW, similar to other reports [19,29,30].

Prevalence of β-Lactamases among CTX-Resistant Populations
Among CTX-resistant E. coli populations from the clinic and environmental sources, we assessed the presence of seven different β-lactamases: CTX-M subgroups (1, 2, and 9), TEM, SHV, OXA, and KPC. They have become a global pandemic with ESBL prevalence increasing among commensals and clinical isolates [21,31], causing significant therapeutic challenges [32]. Women with ESBL-producing E. coli urinary tract infections are significantly more likely to contain the same ESBL-producing E. coli clone in their feces [33]. From a One Health perspective, the strongest correlation of ESBL genes among commensal and pathogenic isolates is with environmental sources, specifically surface and WW [34], in contrast to agricultural sources. One of the most common ESBLs that continues to increase in prevalence is the CTX-M family [34,35]. CTX-M enzymes originally targeted cefotaxime; however, their predominant presence in mobile genetic elements increased the variability and hydrolytic activity to other cephalosporins, including ceftazidime [36]. Furthermore, intra-and interspecies HGT within WW poses an additional threat with the evolution of chimeric CTX-M genes [37]. In our study, CTX-M-1 subgroup was the most common among ESBLs (Table 1: 81% of total samples analyzed) where 95% of cefotaxime-resistant hospital WW populations encoded bla CTX-M-1 subgroup followed by 73% of urban WW influent and 81% of treated effluent isolates (Table 1). Resistance to CRO and CTX were significantly correlated with strains encoding bla CTX-M-1 (Figure 3, p = 0.0001). Among clinical isolates, the CTX-M-1 subgroup predominated as the second most common ESBL among clinical strains, where it was encoded by 64% of isolates (Table 1). The CTX-M-1 subgroup is the most common ESBL within hospital isolates from North America comprising between 34% and 66.2% [38,39]. Another study demonstrated 79% and 67% of ESBL-producing E. coli phylogroup B2 strains from urine and feces, respectively, encoded CTX-M-1 subgroup compared to 21% and 25% CTX-M-9 subgroup [33]. The CTX-M-9 subgroup was more prevalent among urban WW isolates compared to clinical and hospital WW (Table 1, p= 0.005). Although not as common as CTX-M-1 subgroup, the CTX-M-9 subgroup members were initially found on conjugative plasmids [40]. This subgroup, along with CTX-M-1, is commonly present within hospital sinks [41]. Others have found both CTX-M subgroups 1 and 9 at similar frequencies among WW influent E. coli populations [37]. Within the United States, CTX-M subgroups 1 and 2 have been located without subgroup 9, based on total WW DNA analysis [42]. Bevan et al. reviewed 220 articles about CTX-M subgroups and identified a predominate geographic distribution of CTX-M-1 and -9 subgroups among most of the world, though some countries in South America display a stronger CTX-M-2 subgroup population [43]. Our findings agree with the worldwide CTX-M-1 and -9 subgroup pattern and no CTX-resistant isolates encoding CTX-M-2 subgroup. With the advent of sequencing technologies providing more affordable sequencing options, it is important to distinguish between the health risks associated with ESBL findings from total DNA compared to bacterial isolates.
Among the other β-lactamase genes analyzed in our study, bla TEM and bla OXA were carried by 73% of isolates (Table 1). Both ESBLs were found on a relatively high percentage in all sources, although clinical sources, like Branger et al. [44], had the highest prevalence of bla TEM compared to hospital WW isolates, which had the lowest prevalence (Table 1: 100% vs. 59%, respectively). However, this pattern was reversed with bla OXA , where 86% of hospital WW isolates encoded it, though only 45% of clinical strains carried the gene (Table 1). Further analysis of bla TEM and bla OXA genotypes with resistance phenotypes did not identify any significant correlations ( Table 2). Our study did not identify any bla SHVencoding CTX-resistant strains from either hospital or urban WW . This is in agreement with other studies worldwide reporting a predominance in TEM, OXA, and CTX-M [45,46].  Plasmid maps from hospital wastewater E. coli 1000C-3 isolate. Plasmids sequences were determined from long-read Nanopore and Illumina short paired-end reads. pMHW-1 was identified as a 40,232 bps mobilizable IncP-6-type plasmid, and pMHW-2 as a 145,456 bps conjugative IncF plasmid containing a tra operon. Arrows are color-coded exhibiting genes involved in drug resistance (fuchsia), virulence (blue), mobile genetic elements (MGE) (black), plasmid replication, maintenance and transfer (white), and opening reading frames (ORFs) (grey). When assessing the risk of ARGs within environmental bacterial populations, it is important to assess if the ARGs encoding resistance against currently used antibiotics, especially last-resort antibiotics, are found on mobile genetic elements [47]. In our study, we have identified plasmid-mediated resistance to third-generation cephalosporins (Supplementary Table S4) from WW sources. Additionally, class 1 integron-integrase gene (int1) was found among 65% of all CTX-resistant populations with similar levels amongst clinical and WW populations (Table 1: 64.4% and 64.2% respectively). Our findings agree with others who identify int1 commonly throughout WW treatment process [48], as well as harboring gene cassettes encoding ARGs [19], thus supporting its use as an indicator of arthropogenic pollution [49]. Widespread dissemination of int1 among diverse bacterial populations within mobile genetic elements, such as plasmids and transposons, supports a strong association of their presence with the rapid evolution of antibiotic resistance [49].
With carbapenems traditionally used to treat ESBL-producing Enterobacteriaceae infections, the evolution of carbapenemases pose a direct threat to therapeutic options. Enterobacteriaceae-produced carbapenemases are primarily grouped into Ambler class A (Klebsiella pneumoniae carbapenemase (KPC) and GES), class B (metallo-β-lactamases (MBL) like NDM, VIM, and IMP), or class D (OXA-48-like enzymes) [50]. Of these, KPC-producers are the most endemic throughout the Americas, parts of Europe, and China, while spreading globally [51]. Amongst our cefotaxime-resistant E. coli populations, hospital WW was the only source that carried bla KPC (Table 1: 41%, p < 0.001). The presence of bla KPC was significantly correlated with carbapenem and ceftazidime resistance and the absence of bla TEM (Table 2, p < 0.01). Other research has found that the highest prevalence of clinical bla KPC appears within inpatient facilities, peaking in long-term acute care hospitals [51].
We also identified pMHW-2 (Accession number: CP091428), a multi-drug resistant self-transmissible IncFIA plasmid in E. coli 1000C-3 (Supplementary Table S5: 99.74% coverage and 99.48% identity with E. coli plasmid F Accession number AP001918) consisting of 145,445 bps with 52% GC content and 196 predicted ORFs. IncF plasmids are commonly associated with multi-drug resistant E. coli strains from the clinic, animals, and the environment [60]. This plasmid shared homology with numerous plasmids from E. coli (Supplementary Table S7). Conjugative pMHW-2 plasmid contained a tra operon with 30 genes aiding self-transmissibility (Figure 3 and Supplementary Table S6). These genes are involved in the regulation of transcription, relaxosome formation, type 4 secretion system, pilus formation, and mating pair formation [61]. pMHW-2 plasmid also contains a Tn5403, Tn3-like transposon, which has been identified to contain enzymes capable of helping non-conjugative plasmids [62]. The self-transmissible, multi-drug resistant plasmid pMHW-2 is rich in ARGs downstream of the type 1 integron integrase (Figure 3 and Supplementary Table S6: 14,767-36,470 bps). pMHW-2 encoded two beta-lactamases, bla CTX-M-15 and bla OXA-1, with bla CTX-M-15 part of Tn2 between IS26 elements (Figure 3). Among clinical isolates, bla CTX-M-15 has been commonly identified on IncFI plasmids belonging to phylogroup A [63]. An IS26-composite transposon also contained bla OXA-1 along with type B-3 chloramphenicol O-acetyltransferase (catB3) and an aminoglycoside 6 -N-acetyltransferase-Ib (aac(6')-Ib-cr) gene (Figure 3 and  Supplementary Table S6). This plasmid also contained a macrolide resistance gene (mph(A)), tetracycline (tetA), sulphonamide (sul1), trimethoprim (dfr17), and aminoglycoside (aadA5), as well as resistance genes to disinfectants (e.g., qacE and sitABCD). The large presence of IS26 interspersed within plasmids and chromosomes of clinical and environmental isolates within the Enterobacteriaceae family supports a pivotal role in their dissemination of multi-drug resistance [64,65]. Beyond resistance, this plasmid encodes an iucABCD iutA operon responsible for the biosynthesis of the iron siderophore aerobactin. The production of aerobactin has been associated with increased virulence and relapse episodes of urinary tract infections [66,67]. Under iron-limiting and oxidative stress conditions, aerobactin has increased biofilm formation and hydrogen peroxide resistance respectively [68]. When the whole-genome sequence was analyzed with PathogenFinder [69], E. coli 1000C-3 matched 663 pathogenic families, resulting in a pathogen probability of 0.919.

Prevalence of Virulence Genes among CTX-Resistant Populations
Antimicrobial resistant opportunistic pathogens pose an even bigger health risk. Another way to assess the potential pathogenic nature of a bacterial isolate is to identify encoded virulence factors, which are involved in an array of activities including attachment to host tissues, immune evasion, and production of toxins capable of tissue damage [70]. Frequently, these virulence factors are found on pathogenicity islands and other mobile genetic elements that allow for gene transfer [71]. We analyzed seven virulence genes comprising both cell surface and secreted virulence factors [70]. CTX-resistant clinical strains exhibited significantly more virulence genes compared to WW effluent isolates ( Figure 4A, p = 0.0007), where 64% of clinical isolates displayed three or more virulence genes compared to only 31% among WW effluent populations. Comparatively, the WW effluent isolates appeared to be less virulent; however, all encoded at least one virulence gene ( Figure 4A). logroup B2 has been associated with more virulence factors compared to other phylogroups [74,75]. In addition, we found that 75% of the B2 phylogroup isolates (from both clinical isolates and WW effluents) co-expressed aerobactin and afa adhesions (afa), a virulence factor commonly associated with UPEC [71]. However, due to the diversity and abundance of virulence genes in bacteria [76], it becomes difficult to attribute specific virulence genes to a pathogenic lineage, especially since they can be found among commensal strains [76]. Despite this limitation, our study detected similar virulence genes in both clinical samples and WW effluents including afa adhesions ( Figure 4B) commonly associated with UPEC populations. These findings are consistent with a previous study identifying WW E. coli strains containing the same virulence factors as clinical UPEC isolates after whole-genome analysis [77]. Together, these findings implicate a potential pathogenic linkage emphasizing a health risk among treated effluent populations.  The virulence factor, aerobactin, was present in all clinical and effluent E. coli isolates. Aerobactin is frequently found among UPEC strains, is involved in iron sequestration, and is commonly associated with multi-drug resistance residing both on chromosomes and within plasmids [72]. Typically, aerobactin is associated with the presence of other virulence genes, such as type 1 fimbriae (fimH), hemolysin (hlyCA), and P fimbriae (papC) [73]. For our clinical strains, aerobactin and the adhesion gene, fimH, were co-expressed; similar findings were evident among effluent strains where co-expression occurred among 93% of the population ( Figure 4B). Overall, virulence genes were significantly more abundant in the B2 phylogroup, regardless of source, compared to non-B2 phylogroups ( Figure 4C, p < 0.05). This finding is consistent with other reports where phylogroup B2 has been associated with more virulence factors compared to other phylogroups [74,75]. In addition, we found that 75% of the B2 phylogroup isolates (from both clinical isolates and WW effluents) coexpressed aerobactin and afa adhesions (afa), a virulence factor commonly associated with UPEC [71]. However, due to the diversity and abundance of virulence genes in bacteria [76], it becomes difficult to attribute specific virulence genes to a pathogenic lineage, especially since they can be found among commensal strains [76]. Despite this limitation, our study detected similar virulence genes in both clinical samples and WW effluents including afa adhesions ( Figure 4B) commonly associated with UPEC populations. These findings are consistent with a previous study identifying WW E. coli strains containing the same virulence factors as clinical UPEC isolates after whole-genome analysis [77]. Together, these findings implicate a potential pathogenic linkage emphasizing a health risk among treated effluent populations. All samples were immediately placed on ice until it was processed within 8 h. Various water volumes (≤100 mL) were filtered over a 0.45 µm filter and placed on modified mTEC with cefotaxime (4 µg/mL) to obtain resistant E. coli populations. Random violet color colonies underwent further isolation by performing a four-quadrant streak on TSA with cefotaxime (2 µg/mL) to ensure isolation. From these plates, 58 WW isolates were obtained from hospital WW (n = 22), as well as influents (n = 19) and treated effluents (n = 17) with a maximum of 10 colony forming units (CFUs) per source per date. No CTX-resistant E. coli populations with the volumes tested were evident from one effluent sampling date and one hospital WW source on one date. Additionally, no CTX-resistant violet colonies were evident from post-chlorinated effluents in the volumes tested. Frozen glycerol stocks were made of all isolates for further analysis.

Acquisition of CTX-Resistant E. coli Isolates
Clinical E. coli isolates (e.g., urinary tract (7), blood (2), wound (1), and surgical tissue (1)) from inpatient (n = 5) and outpatient (n = 6) samples were acquired from the same medical facility as the hospital WW. Presumptive E. coli colonies were confirmed using Bruker MALDI-TOF mass spectrometry [19]. Potential ESBL-and carbapenemaseproducing E. coli isolates were further identified using the BD Phoenix susceptibility system interfaced with the BD EpiCenter System V6.41 following Clinical & Laboratory Standards Institute (CLSI) guidelines. Frozen glycerol stocks were made of all isolates for future analysis.

Plasmid Isolation and Transformation
Plasmids from CTX-resistant populations were isolated using a mini prep kit according to the manufacturer's instructions (Zymo Research, Irvine, CA, USA). Isolated plasmid DNA was then separated on a 0.7% agarose gel at 80 V for 3 h. Gel images were captured after ethidium bromide staining for further analysis.
To identify resistant phenotypes associated with plasmids, they were transformed into Lucigen E. cloni ® 10G chemically competent cells according to the manufacturer instructions (Lucigen, Middleton, WI, USA). Briefly, 4 µL of plasmid DNA was added to 40 µl of the competent cells and incubated on ice for 30 min. Heat shock was performed for 90 s at 42 • C followed by ice for 2 min. Recovery medium was added to the cells and incubated for 1 h at 37 • C while shaking at 175 rpm. The reaction was plated on cefotaxime (2 µg/mL) plates and incubated overnight at 35 • C. Transformants were further tested for antibiotic susceptibility to determine antibiotic resistant phenotypes transmissible via plasmids.

Identification of Beta-Lactamases and Virulence Genes
A Promega Genomic DNA Purification kit was used to isolate genomic DNA of resistant strains. β-lactamase genes were further amplified using 2X GoTaq Green PCR Master Mix (Promega Corp., Madison, WI, USA) and their associated primers (Supplementary Table S1). Isolated DNA from a Gram-negative carbapenemase detection panel (CDC & FDA AR Isolate Bank, Atlanta, GA, USA) served as positive controls for ESBLs.

Phylogroup Identification
Phylotyping was performed according to the revised Clermont method [25]. Briefly, a quadruplex PCR was performed using primers (Supplementary Table S1) specific to the following genes: chuA, yjaA, TspE4.C2, and arpA to determine its associated phylogroup. Further characterization of some phylogroup C and E strains was performed using two additional allele-specific PCR primer pairs (Supplementary Table S1) to arpA and trpA as described by Clermont et al. All PCR products were visible after gel electrophoresis using a 1.0% agarose gel with subsequent ethidium bromide staining.

Whole Genome Sequencing and Analysis
Genomic isolation was performed on E. coli 1000C-3 isolated from hospital WW that exhibited antibiotic resistance to third-generation cephalosporins and carbapenems. The Oxford Nanopore platform and NextSeq 2000 platform (Illumina, San Diego, CA, USA) were used to provide long and pair end reads (2 × 151 bp), respectively, for whole-genome sequencing. Illumina bcl2fastq 2.20.0.445 software and Porechop 0.2.3_seqan2.1.1 were performed on pair end reads and long reads, respectively, to perform quality control and adapter trimming. Unicycler 0.4.8 was subsequently used to create a hybrid assembly between the two reads and provide closed genomes. The WGS is available at NCBI under BioProject ID SUB10978992 with accession numbers CP091427-CP091429.
The genome was submitted to JGI Integrated Microbial Genomes and Microbiomes (IMG/MER) for annotation and further analysis [80,81], including identifying putative horizontally transferred genes. Further analysis of assembled contigs were performed using the following web-based tools: PATRIC [82] for a comprehensive genome analysis; Res-Finder 4.1 from the Center for Genomic Epidemiology [83] and Comprehensive Antibiotic Resistance Database (CARD) [84] to identify ARGs in the plasmid and chromosome; Plas-midFinder 2.1 and pMLST 2.0 to identify incompatibility types of plasmids [85]; MobileEle-mentFinder v1.0.3 and IS Finder to identify mobile genetic elements [86,87]; SerotypeFinder 2.0 for serotyping [88]; oriTfinder 1.1 to identify oriT sites and MGEs [89]; and Pathogen-Finder 1.1 to determine the pathogenic potential [69]. Plasmid maps were comprised using the above tools and SnapGene software (from Insightful Science, San Diego, CA, USA; available at snapgene.com, accessed on 8 December 2021).

Statistics
Stata 12.1 software (College Station, TX, USA) and GraphPad Prism version 9.1.0 (GraphPad Software, San Diego, CA, USA) were used to perform all statistical tests. Fisher's exact test was used to determine differences in the prevalence of ARGs and virulence genes among E. coli from various water sources. One-way ANOVA with Tukey's post hoc test determined statistical differences between MAR indices. Correlations between antibiotic resistance phenotypes and genotypes were evaluated by Pearson's correlation test. Variation among total virulence factors used Welch's t-test.

Conclusions
With antibiotic resistance threatening global health, it is essential to understand the range and mechanisms of antibiotic resistance transmission [90] across diverse ecosystems. From this study, we can conclude the following: (1) hospital wastewater provides a rich source of multi-drug resistant E. coli populations and can serve as reservoirs of mobilizable ESBLs and carbapenemase entering sewer systems; (2) wastewater treatment and chlorine disinfection are critical mitigators in controlling AMR; (3) the prevalence and similarity of virulence factors among CTX-resistant E. coli in treated WW effluents are comparable to pathogenic strains in the same geographical area, further emphasizing the potential public health risk of treated wastewater into recreational waters. While this study demonstrates similarities in phenotypic and genotypic resistance across different ecosystems, additional sampling and more complete whole-genome sequencing would further support the evolutionary linkage between populations and enhance understanding the potential pathogenic nature of WW isolate populations.