Resistome characterization of extended-spectrum beta-lactamase (ESBL)-producing Escherichia coli isolated from wastewater treatment utilities in Oregon

The emergence of bacterial infections resistance to antibiotics, including beta-lactams, is a global public health threat. Resistance to broad-spectrum beta-lactams, such as ampicillin and third- and fourth-generation cephalosporins, by extended-spectrum beta-lactamase (ESBL)-producing Enterobacteriaceae is an emerging concern even in areas with high restrictions on antibiotic consumption, such as Norway (Jørgensen et al. 2017). Infections with ESBL-producing Enterobacteriaceae are shown to increase the risk of hospitalization, likelihood of discharge to a chronic care facility, and in-hospital mortality (Schwaber et al. 2006). ESBL-producing species are characterized by their ability to hydrolyze a broad spectrum of beta-lactam antibiotics, including oxyimino cephalosporins, monobactams, and penicillins. ESBLs are confirmed as the resistance mechanism by verifying that beta-lactamase inhibitors, clavulanic acid or tazobactam, restore the sensitivity to the beta-lactam (Bush & Fisher 2011). The development of novel beta-lactams has been ineffective to reduce the spread of ESBL-producers; in fact, selective pressure of novel antibiotics has resulted in the evolution of more pernicious beta-lactamases (Bradford 2001).

ESBL-associated genes are found in many species of Enterobacteriaceae, especially in Escherichia coli, a primary indicator of fecal bacterial contamination (Palucha et al. 1999). Among clinical isolates of ESBL-producing E. coli from various countries, including Indonesia, Iran, Korea, and the United States, bla_CTX-M has been reported as the most common ESBL gene with bla_CTX-M-15 the most prevalent subtype (Park et al. 2009; Sidjabat et al. 2009; Severin et al. 2010; Haghighatpanah et al. 2016). In addition, many of the ESBL-producing E. coli carry virulence factors that can enhance their pathogenicity (Jiang et al. 2019). An example of a virulence factor is tellurite resistance, encoded by genes such as terC, which has been implicated in bacterial resistance to phagocytosis and oxidative stress and contributing to prolonged urinary tract infections (Valková et al. 2007; Turkovicova et al. 2016). Moreover, ESBL genes are often encoded on plasmids (i.e., mobile genetic elements) and hence, can easily spread via horizontal gene transfer (Haenni et al. 2018; Liu et al. 2019). In addition, ESBL-producers are commonly resistant to other antibiotics, including gentamicin, sulfamethoxazole-trimethoprim, and ciprofloxacin, resulting in fewer clinical treatment options (Schwaber et al. 2005). Furthermore, ESBL-producing E. coli are reportedly more likely to be multidrug resistant (i.e., resistant to three or more classes of antibiotics) than non-ESBL-producers (Blaak et al. 2015; Jørgensen et al. 2017). The resistome (i.e., entire ecology of resistance determinants) of ESBL-producing E. coli is not clearly understood and further knowledge is needed to enable the development of treatment options for ESBL-producing E. coli infections.

Although ESBL-producers have been primarily observed and described in clinical settings, ESBL genes and the bacterial strains that harbor them are also found in many environmental reservoirs including agricultural soils, livestock, and surface water (Ben Said et al. 2015; Blaak et al. 2015; Haenni et al. 2018). Wastewater treatment plants are also major reservoirs and sources of antibiotic-resistant bacteria and their determinant antibiotic resistance genes (ARGs) (Manaia et al. 2018; Alexander et al. 2020). Recently, ESBL-producing E. coli were detected in all the samples collected in a wastewater monitoring study in Germany (Schmiege et al. 2021). Similar to clinical settings, bla_CTX-M genes are common ESBL determinants in wastewater systems (Paulshus et al. 2019; Schages et al. 2020). Most of the studies on ESBL-associated genes in wastewater systems are from Europe; however, there is a limited knowledge regarding ESBL-producing E. coli in wastewater utilities in the United States. With growing attention to wastewater-based epidemiology as an important public health surveillance framework, understanding the resistome of ESBL-producing E. coli in wastewater treatment systems is critical (Riquelme et al. 2021).

This study characterizes the resistome of 11 ESBL-producing E. coli isolates collected from various wastewater treatment utilities across the state of Oregon. We used whole genome sequencing (WGS) to determine the antibiotic resistance genotypes of ESBL-producing E. coli isolates, their sequence types (STs), ARGs including ESBL-associated genes as well as virulence factors. Moreover, the phenotypes of antibiotic resistance were characterized. The significance and originality of this study is resistome characterization of ESBL-producing E. coli isolates in the U.S. wastewater treatment systems that have the potential to enter the environment and pose risks to human health.

E. coli isolates were collected from eight wastewater treatment utilities in Oregon as described previously (Khorshidi-Zadeh et al. 2021). Briefly, wastewater influent, secondary effluent, final effluent, and treated biosolids were collected from 17 wastewater treatment utilities over winter and summer in 2019 and 2020. To isolate E. coli colonies, collected samples were vacuum filtered or streaked directly onto m-TEC ChromoSelect agar (Sigma-Aldrich, St. Louis, MO) plates and then confirmed by fluorescence on MacConkey agar with MUG (Hardy Diagnostics, Santa Maria, CA). Over the course of the study, 1,143 E. coli colonies were isolated. Collected E. coli colonies were tested for the production of ESBL enzymes by measuring the zones of inhibition surrounding disks containing cefotaxime (30 μg), ceftazidime (30 μg), cefotaxime/clavulanic acid (30/10 μg), and ceftazidime/clavulanic acid (30/10 μg) (BD Diagnostics, Sparks, MD) (CLSI 2020). A difference in zone of inhibition of ≥5 mm diameter between antibiotic and antibiotic-acid disks indicated the production of ESBL enzymes. An internal quality control and E. coli ATCC 25922 strains were used as positive and negative controls, respectively. The phenotypic antibiotic resistance of the isolates to a series of beta-lactam antibiotics, including penicillins, first to fourth-generation cephalosporins, and carbapenems were tested using the AST-GN99 card with a VITEK 2 system (bioMérieux, Marcy-l'Étoile, France) according to manufacturer's instructions. The AST-GN99 card includes other classes of antibiotics (i.e., aminoglycosides, quinolones, tetracycline, nitrofuran, and sulfonamide) that supported the classification of multidrug resistance phenotypes. Overall, 13 E. coli wastewater isolates were originally classified as ESBL-producers (Khorshidi-Zadeh et al. 2021); however, two of these isolates could not be confirmed as ESBL-producers using the AST-GN99 cards. Therefore, a total of 11 isolates were included in the present study. Details about the samples associated with these isolates are listed in Supplementary Material, Table S1.

Freezer stocks were streaked onto tryptic soy agar (TSA; Hardy Diagnostics, Santa Maria, CA) and grown for 24 h at 37 °C before being transferred to tryptic soy broth (TSB, Hardy Diagnostics, Santa Maria, CA). DNA was extracted from TSB cultures following manufacturer's instructions for the DNeasy Blood and Tissue kit (Qiagen, Carlsbad, CA). Purified DNA was quantified and quality checked using a Qubit 4 (Invitrogen, Carslbad, CA) and Nanodrop (Thermo Fisher, Waltham, MA). DNA samples were submitted to the Oregon State University's Center for Quantitative Life Sciences (Corvallis, OR) for library preparation using the plexWell™ 96 kit (seqWell, Beverly, MA). Libraries were sequenced on a MiSeq 3000 instrument (Illumina, San Diego, CA), employing paired-end 150 base reads. Duplicate libraries of each isolate were prepared to meet the minimum sample requirement of the plexWell™ 96 kit.

Demultiplexed reads were quality assessed using FastQC v0.11.8 (Andrews et al. 2012) and sequencing adapters were removed using Trimmomatic v0.40 (Bolger et al. 2014). Duplicate reads were concatenated into single forward and reverse FASTQ files and uploaded to the PATRIC platform for whole genome assembly and annotation using the ‘Comprehensive Genome Analysis’ service (Davis et al. 2019). The resulting draft genome assemblies and annotation files were used for all downstream analyses.

STs were determined from draft genome assemblies based on internal fragments of seven core genes (i.e., adk, fumC, gyrB, icd, mdh, purA, and recA) using the Center for Genomic Epidemiology (CGE) server (Larsen et al. 2012). A phylogenetic tree was created based on the concatenated sequences of the seven ST core genes’ fragments using the PhyML 3.1 Maximum-Likelihood model with 100 rounds of bootstrapping using the SeaView platform (Wirth et al. 2006; Gouy et al. 2010; Guindon et al. 2010).

ARGs were identified using NCBI's AMRFinderPlus with minimum nucleotide identity of 90% and a minimum coverage of 50% (Feldgarden et al. 2019). Virulence factors were found using VirulenceFinder 2.0, with minimum nucleotide identity set to 90% and a minimum coverage of 60% for each (Joensen et al. 2014). Percent identities of amino acid sequences of AmpC were compared using Clustal Omega (Madeira et al. 2019).

The sequenced genomes have been deposited at the NCBI Sequence Read Archive (NCBI SRA) with the study identifier PRJNA767748.

All 11 ESBL-producing E. coli isolates were resistant to at least one penicillin and several of the cephalosporin class beta-lactam antibiotics tested (Figure 1). In the penicillin class, all 11 E. coli isolates were resistant to ampicillin, and one isolate (G) was resistant to amoxicillin complexed with the beta-lactamase inhibitor clavulanic acid. Isolate G also demonstrated intermediate resistance to piperacillin/tazobactam complex, while all other isolates (n = 10) were susceptible. Six of the 11 E. coli isolates were resistant to all four generations of tested cephalosporins (i.e., first: cefazolin, second: cefuroxime, third: ceftriaxone, and fourth: cefepime). The other five E. coli isolates carried phenotypic resistance to three generations of cephalosporins, while two demonstrated intermediate resistance to the fourth-generation cefepime. None of the E. coli isolates displayed resistance to the tested carbapenems (i.e., ertapenem, imipenem, meropenem).

Concerningly, eight of the 11 E. coli isolates were phenotypically multidrug resistant (MDR) with resistances to three or more classes of antibiotics (Figure 1). Including beta-lactams, four of the E. coli isolates were phenotypically resistant to five classes of antibiotics (isolates G, I, J, and H). Tetracycline resistance was the most common detected non-beta-lactam antibiotic resistance (n = 10) and was observed in all the MDR E. coli isolates. Eight of the isolates were resistant to trimethoprim/sulfamethoxazole, but sensitive to nitrofurantoin. Five were resistant to both quinolones (ciprofloxacin and levofloxacin), and four were resistant to both aminoglycosides (gentamycin or tobramycin) tested.

Genome sizes of the ESBL-producing isolates assemblies ranged from 4.59 Mbp (B) to 5.35 Mbp (A) with GC content between 50.3% (A) and 50.8% (F and G). ST analyses identified four of the 11 isolates to be ST744 (isolates C, H, I, and J; Figure 1). The seven other isolates were identified as singleton STs, including ST34, ST46, ST69, ST70, ST224, ST328, and ST3580 (Figure 1).

A total of 46 different virulence factors were identified in the sequenced data among the 11 E. coli isolates (Supplementary Material, Table S2). All 11 E. coli isolates contained the terC virulence factor, responsible for tellurite resistance (Valková et al. 2007). All E. coli isolates also carried the gad virulence factor, a glutamate decarboxylase that increases survival in acidic regions of the gastrointestinal tract (Damiano et al. 2015). The other common virulence factors were sitA (with a role in transportation of Fe²⁺ and Mn²⁺, n = 8) (Sabri et al. 2006), ompT (an outer membrane protein enabling intracellular survival, n = 6) (Hejair et al. 2017), and traT (transfers protein which inhibits certain pathways, n = 6) (Miajlovic & Smith 2014).

At least one known ESBL-associated gene (i.e., bla_CTX, bla_TEM, bla_CMY, and bla_SHV) was found in 10 of the 11 ESBL-producing E. coli isolates (Figures 2 and 3), and five of the isolates (A, B, G, I, and J) contained two of these genes. Eight E. coli isolates contained bla_CTX-M genes (Figure 2(a)). Four of the eight isolates with bla_CTX-M gene – isolates C, H, I, and J which were all identified as ST744. These four E. coli isolates harbored arsenic resistance genes nearby the bla_CTX-M gene (labeled 2 and 3 in Figure 2(a)). Six of the bla_CTX-M genes were identified as subtype bla_CTX-M-55 (isolates B, C, D, H, I, and J), and two were identified as subtype bla_CTX-M-15 (A and K). All six of the bla_CTX-M-55 carrying isolates were flanked by the same mobile genetic elements (i.e., transposase, labeled 1 in Figure 2(a)). The notably different gene region was from isolate D, which was identified as subtype bla_CTX-M-55 adjacent to a set of the toxin-antitoxin yee genes and near a death-on-curing (doc) gene (Lehnherr et al. 1993; Brown & Shaw 2003).

bla_TEM genes were identified in five of the ESBL-producing E. coli isolates (Figures 2(b) and 3). Unlike the bla_CTX-M gene regions, the bla_TEM gene regions were not flanked by transposases. However, three of the five bla_TEM-containing isolates identified bla_TEM adjacent to a phage integrase gene (purple arrows in Figure 2(b)). Moreover, bla_CMY gene was found within only one of the isolates (G), reported as a subtype bla_CMY-42 (Figure 2(c)). In addition to the bla_CMY gene, isolate G also carried the bla_TEM-1 gene. A bla_SHV gene was found in isolate F and was reported as subtype bla_SHV-12 (Figure 2(d)). The bla_CMY and bla_SHV gene regions are similar to respective reference gene regions from other bacteria in the PATRIC database. Despite the multidrug resistance phenotype of many of the isolates, only isolate G had a multidrug resistance-related gene (efflux pump) nearby the ESBL gene (orange arrow in Figure 2(c)).

The analysis of sequenced data via AMRFinderPlus identified a total of 31 different ARGs belonging to 11 classes of antibiotics (Figure 3). All the 11 ESBL-producing E. coli isolates contained ampC. AmpC beta-lactamases are cephalosporinases of clinical importance as they confer resistance to cephalothin, cefazolin, cefoxitin, most penicillins, and beta-lactamase inhibitor-beta-lactam complexes (Jacoby 2009). Results show very conserved location for ampC. The average nucleotide identity matrix between the E. coli isolates for the ampC is shown in Supplementary Material, Table S3. Overall, ampC genes ranged from 96.30 to 100.0% nucleotide identity. Unsurprising, isolates C, H, I, and J that were all identified as ST744 showed 100.0% similarity for ampC. Moreover, nine of the 11 E. coli isolates contained genes encoding for aminoglycoside resistance, where aadA5, aph(3′′)-Ib, and aph(6)-Id were the most common genes (n = 6; Figure 3). Those same nine isolates with resistance to aminoglycosides also contained resistances to sulfonamides and tetracyclines. The most common sulfonamide resistance gene was sul1 (n = 7), followed by sul2 (n = 5). The most common tetracycline resistance gene was tet(A) (n = 8). Seven E. coli isolates harbored a trimethoprim resistance gene, six of which harbored drfA17 and only one carried drfA12. Genes encoding resistances to fosfomycins, chloramphenicols, quinolones, and lincosamides were also detected among the E. coli isolates (Figure 3).

Comparing the resistance phenotypes and genotypes of the 11 ESBL-producing E. coli isolates, some similarities and discrepancies were observed. All isolates harbored phenotypic resistance to the tested first-, second-, and third-generation cephalosporins (i.e., first: cefazolin, second: cefuroxime, third: ceftriaxone; Figure 1), while all but one (isolate E) carried at least one of the known ESBL genes (Figure 3). The lack of any ESBL-associated gene in isolate E suggests the presence of another novel mechanism supporting cephalosporin resistance. Moreover, the two E. coli isolates G and F that remained susceptible to cefepime (fourth-generation cephalosporin) did not carry the bla_CTX-M gene (Figure 3), which suggests a correlation between the prevalence of bla_CTX gene and fourth-generation cephalosporins. Moreover, all 11 E. coli isolates harbored the ampC gene (Figure 3) and AmpC beta-lactamases commonly mediate resistance to beta-lactamase inhibitor-beta-lactam combinations; however, only one isolate (G) showed resistance to amoxicillin/clavulanic acid, and one isolate (B) had intermediate resistance (Figure 1). Comparing resistances to aminoglycosides, results suggest the aadA5 gene (Figure 3) as a good indicator for the prevalence of phenotypic resistances to gentamicin and tobramycin (Figure 1). For resistance to tetracycline, we observed phenotypic resistance of all but one E. coli isolate (K) (Figure 1) and found one of three resistance genes (tet(A), tet(B), and tet(C)) in all but isolates E and K, which shows these three tetracycline resistance genes as good indicators of their associated resistance phenotype.

The global spread of pathogens resistant to antibiotic treatments is a major public health issue. According to the most recent CDC report on antibiotic resistance, over 2.8 million cases of infections in the U.S. are caused by antibiotic-resistant bacteria with over 35,000 per year associated deaths (CDC 2019). Resistance via ESBL-producing bacteria is especially concerning because of the vast number of antibiotics they confer resistance to (i.e., penicillins and cephalosporins). It is estimated that healthcare complication risk is increased by 50% if infections are caused by ESBL-producing Enterobacteriaceae (CDC 2019). Last-resort beta-lactam antibiotics like carbapenems are currently used for the treatment of ESBL-producing E. coli infections (Hawkey & Livermore 2012); however, certain ESBL-associated genes (e.g., bla_VIM and bla_KPC) have been shown to mediate resistance to these last resort antibiotics (Hoelle et al. 2019).

Few studies have characterized ESBL determinant genes in E. coli isolated from wastewater treatment utilities in the U.S. One such study, limited to isolates resistant to imipenem, found a variety of ESBL-related genes including bla_CTX-M, bla_SHV, and bla_TEM to be prevalent in wastewater samples from seven states (Hoelle et al. 2019). Another study reported that ESBL-producing E. coli isolated from Colorado wastewater treatment plants expressed distinct genetic profiles compared to sewage, with wastewater samples more likely to carry bla_CTX type genes and more likely to be MDR (Haberecht et al. 2019). Our results agree with those findings.

In this study, 11 ESBL-producing E. coli isolated from multiple wastewater treatment utilities throughout Oregon were characterized for their genetic determinants. These E. coli isolates were collected from wastewater influent, secondary effluent (after activated sludge or biological nitrogen removal), biosolids (after anaerobic digestion), or chlorine-disinfected final effluent. Various studies have shown wastewater treatment processes to impact the antibiotic resistance load in treated wastewater and biosolids differently. Results of such studies, however, are not conclusive (Wuijts et al. 2017). The small sample size in our study (11 E. coli isolates collected from four different wastewater flows (i.e., influent, secondary effluent, final effluent, and biosolids) obtained from eight different wastewater treatment utilities) limits the ability to make any conclusions on differences between sample types or wastewater utilities. Eight of the 11 E. coli isolates in our study harbored MDR phenotypic resistances and carried determinant genes for a wide array of antibiotics in addition to beta-lactams including tetracycline, aminoglycosides, and quinolones. Wastewater treatment utilities are an excellent setting for the study of antibiotic resistance because they represent a composite sample of the community. In this paper, the common MDR phenotypes in most of the isolates (8 out of 11) is concerning. Moreover, four of 11 E. coli isolates (isolates C, H, I, and J) that were all identified as ST744 carried 100.0% similarity between their average nucleotide identities for ampC while also having very similar synteny for the bla_CTX-M-55 gene. Moreover, findings showed one isolate (E) with an ESBL-associated phenotype (resistance to first-, second-, and third-generation cephalosporins) without harboring any of the known ESBL genes. This discrepancy shows the potential limitation of annotation tools for uncharacterized novel genes that confer beta-lactam resistance. As shown by the isolates encompassing multiple STs, the ESBL-positive E. coli isolated from Oregon wastewater treatment plants were phylogenetically diverse. However, the same few ESBL genes were found across multiple STs with high similarities between their average nucleotide identities of ampC. The dissociation of core genome diversity and resistance genotypes is indicative of horizontal gene transfer. This is also supported by the colocalization of mobile genetic elements with the ESBL genes, especially bla_CTX-M types. The spread of beta-lactam resistance is likely not limited to conjugation via a plasmid as determinant genes could be transposed to other plasmids or chromosomes. The co-occurrences of the ESBL-associated genes with other ARGs as well as several virulence factors in these ESBL-producing E. coli isolates is concerning. Given that these isolates were collected from multiple wastewater treatment utilities in Oregon, these results indicate that ESBL resistance occurs in Oregon's wastewater systems, and the associated resistance is potentially transferable to other reservoirs, including receiving rivers and agricultural zones utilizing biosolids. Results confirm the need for further characterization of ESBL-producing Enterobacteriaceae resistome, especially in wastewater systems where confluence of municipal and clinical waste streams provide an ideal environment for the spread of ARGs. In addition, since wastewater streams are reservoirs for antimicrobial agents other than antibiotics, such as heavy metals, future research should explore the impact of other bactericidal compounds, such as metals, on the induction and proliferation of antibiotic resistance determinants. Finally, comprehensive health risk assessments, including quantitative microbial risk assessments, should be performed to capture the holistic environmental dimension of ESBL resistance prevalence and the associated risks to human health.

This study characterizes the genotypes and phenotypes of antibiotic resistance in 11 ESBL-producing E. coli isolates collected from Oregon wastewater systems. Known ESBL-associated genes (i.e., bla_CMY, bla_CTX, bla_SHV, bla_TEM) were identified in 10 of the 11 E. coli isolates with five isolates carrying two of these genes. The ampC gene and virulence factors were observed in all the E. coli isolates. Moreover, 31 different ARGs were identified across all the isolates. All E. coli isolates harbored phenotypic resistance to beta-lactams with eight isolates demonstrating MDR phenotypes. Six of the 11 isolates were sourced from wastewater products (biosolids and final effluents). These results demonstrate the risks of ESBL-producing E. coli in wastewater systems that can enter the environment posing a critical threat to public health.

This work was supported by the USDA National Institute of Food and Agriculture, Agricultural and Food Research Initiative Competitive Program, Agriculture Economics and Rural Communities, Grant No. 2018-67017-27631, and in-kind supplement from the Oregon State University's Center for Quantitative Life Sciences.

All relevant data are available from an online repository or repositories: NCBI SRA link for Data Availability: www.ncbi.nlm.nih.gov/bioproject/PRJNA767748/.