Extended-Spectrum Beta-Lactamase Producing-Escherichia coli Isolated From Irrigation Waters and Produce in Ecuador

In cities across the globe, the majority of wastewater – that includes drug resistant and pathogenic bacteria among other contaminants – is released into streams untreated. This water is often subsequently used for irrigation of pastures and produce. This use of wastewater-contaminated streams allows antibiotic-resistant bacteria to potentially cycle back to humans through agricultural products. In this study, we investigated the prevalence of extended-spectrum β-lactamase (ESBL)-producing Escherichia coli isolated from produce and irrigation water across 17 provinces of Ecuador. A total of 117 vegetable samples, 119 fruit samples, and 38 irrigation water samples were analyzed. Results showed that 11% of the samples were positive for E. coli including 11 irrigation water samples (29%), and samples of 13 vegetables (11%), and 11 fruits (9%). Among the 165 E. coli isolates cultured, 96 (58%) had the ESBL phenotype, and 58% of ESBL producing E. coli came from irrigation water samples, 11% from vegetables, and 30% from fruits. The blaCTX–M–55, blaCTX–M 65, and blaCTX–M 15 genes were the most frequently found gene associated with the ESBL phenotype and coincided with the blaCTX–M alleles associated with human infections in Ecuador. Three isolates had the mcr-1 gene which is responsible for colistin resistance. This report provides evidence of the potential role of irrigation water in the growing antimicrobial resistance crisis in Ecuador.


INTRODUCTION
The rise of antimicrobial resistance (AMR) is one of the most serious biological threats facing modern society, and the inability to treat bacterial infections is already occurring in many nosocomial infections (Frieri et al., 2017). The World Health (WHO) has listed extended spectrum β-lactamase-producing Enterobacteriaceae (ESBL-E) as the most critical antimicrobial resistant microorganisms, among the "Highest Priority" pathogens due to the increasing prevalence in humans and livestock (Yassin et al., 2017;Shrivastava et al., 2018;Li et al., 2019;Murray et al., 2021).
Globally, the majority of wastewater produced by urban settlements goes into streams without prior treatment. Only 20% of produced wastewater receives proper treatment (UNESCO, 2012), and the capacity to treat wastewater often depends on the income level of the country; treatment capacity is 70% of the generated wastewater in high-income countries, compared to ∼8% in low-income countries (Sato et al., 2013). This phenomenon is rising as urban populations grow and developing countries increasingly install pipes to channel wastewater away from communities, even before the development of wastewater treatment plants. The wastewater comes from diverse sources (e.g., homes, hospitals, and animal processing plants, etc.) and contains large quantities of antibiotic resistant bacteria (ARB), often carrying antimicrobial resistance to last-line antimicrobials, such as carbapenems (Lin et al., 2020).
These antimicrobial resistant bacteria (ARB) can cycle back to humans when wastewater-contaminated streams are used to irrigate produce or provide water to food animals (FAO and WHO, 2008;Leff and Fierer, 2013;Pigłowski, 2019); one recent example is the finding of New Delhi metallo-β-lactamases-type carbapenem-resistant Escherichia coli in water, domestic food animals, and humans (carbapenem, a last-line drug, is used exclusively in human medicine) (Li et al., 2019;Murray et al., 2021). Many antibiotic-resistant Enterobacterales, members of the intestinal microbiome (including E. coli), can survive and multiply in the environment (Vasco et al., 2015;Guerrero et al., 2020) and may colonize humans and domestic animals through the fecal-oral route of transmission. Plasmids and other mobile genetic elements (MGEs) carrying AMR genes promote the dissemination of AMR among intestinal bacteria in the intestine of vertebrates (Bonardi and Pitino, 2019), and this cycle is fundamentally captured in the One Health concept. Produce contamination can happen before pre-harvest (i.e., through contaminated irrigation water or manure fertilization) (Beuchat, 1996;Iwu and Okoh, 2019), as well as post-harvest (i.e., by washing, handling and processing food) with irrigation water (Murray et al., 2017).
Wastewater-impacted irrigation water has been identified as the main source of contamination for fresh produce with pathogenic microorganisms and ARB (Njage and Buys, 2015;Gekenidis et al., 2018a). The fecally contaminated produce can transfer ARB to the consumer especially when the produce is consumed fresh and uncooked (Pesavento et al., 2014;Araújo et al., 2017;Hölzel et al., 2018). Besides contributing to the spread of pathogens, irrigation water may potentially play a leading role in the dissemination of ARB (Moore et al., 2010;Hong et al., 2013;Gekenidis et al., 2018b;Vital et al., 2018).
The production of extended-spectrum β-lactamases (ESBL) is one of the most important mechanisms of antibiotic resistance in Enterobacteriaceae. ESBL genes can be divided into 4 groups: TEM, SHV, OXA, and CTX-M types (Bush and Jacoby, 2010); CTX-M type is the most prevalent of ESBLs described (Rossolini et al., 2008;Bevan et al., 2017). Enterobacteriaceae members are the most common bacterial agents causing foodborne outbreaks associated with the consumption of fresh produce (Cooper et al., 2007;Kilonzo-Nthenge et al., 2018;Al-Kharousi et al., 2019;McDaniel and Jadeja, 2019;Motlagh and Yang, 2019). Pathogenic E. coli is a key bacterium in foodborne illnesses, and commensal E. coli is a common indicator organism of fecal contamination in aquatic systems (Edberg et al., 2000;Rochelle-Newall et al., 2015;Motlagh and Yang, 2019). E. coli is also recognized as an important species in the spread of ARB, mainly due to a high aptitude to acquire genetic information through horizontal gene transfer (Grasselli et al., 2008;Hasegawa et al., 2018;Marlène et al., 2020).
In Ecuador, an upper middle-income country, wastewater is almost entirely released untreated into streams; these streams often serve as irrigation water for produce and food-animal agriculture (Ortega-Paredes et al., 2020a,b). There are few studies about the dissemination of ESBL-E. coli from irrigation water to produce (Ben Said et al., 2015;Vital et al., 2018); most of the studies have been carried out in fresh produce from retail centers and groceries (Bhutani et al., 2015;Faour-Klingbeil et al., 2016;Ortega-Paredes et al., 2018;Al-Kharousi et al., 2019;Yang et al., 2019;Colosi et al., 2020;Richter et al., 2020;Song et al., 2020). The aim of this study was to build upon the previous literature to understand the relationship between ARB in irrigation water and ARB on fresh produce obtaining samples from farms and their irrigation water. The study focused on the occurrence of extended spectrum β-lactamase producing E. coli in 17 provinces of Ecuador.

Study Areas
This study was carried out in the following provinces of Ecuador: Manabí, Bolívar, Cañar, Loja, Guayas, Pastaza, Tungurahua, Pichincha, Azuay, Chimborazo, Cotopaxi, Imbabura, Santa Elena, Los Ríos, Morona Santiago, Orellana, and Zamora Chinchipe provinces which are mainly agrarian (Figure 1). The samples correspond to those that are collected as part of the national surveillance program that aims to monitor microbiological indicators and pathogens in the food supply ("Programa Nacional de Vigilancia de Microorganismos de Higiene y Control de Microorganismos Patógenos, para la Vigilancia Epidemiológica de Enfermedades Transmitidas por Alimentos de Origen Agrícola y Pecuario del país -PNVCH").

Isolation of Escherichia coli From Irrigation Water and Produce
The farmers of each crop indicated the irrigation water they used, and this water (n = 37) was collected in sterile bottles and transported to the laboratory at approximately 8 • C and processed within 10 h. Five hundred milliliters of water were filtered using a 0.45 µm pore membrane filter (Millipore, United States). The filter was then incubated in Chromocult R coliform agar (Merck, Germany) overnight at 37 • C, the apparent E. coli colonies were taken and seeded on MacConkey agar (Difco, United States) supplemented with ceftriaxone (2 mg/L) to identify the lactose positive colonies (a maximum of five colonies were picked from each plate) (Richter et al., 2020), colonies of presumptive E. coli were then tested for β-glucuronidase activity using Chromocult R medium (Merck, Germany). All E. coli confirmed isolates from each sample were kept frozen at −80 • C in Tryptic Soy Broth medium (Difco, United States) with 15% glycerol.
The vegetable samples were collected aseptically and refrigerated until analysis (within 12 h). Ten grams of the fresh produce were weighed and placed in a sterile plastic bag and incubated with 90 ml of peptone water (Faour-Klingbeil et al., 2016) for 30 min at room temperature. In the case of fruits such as watermelon and melon, the surface was swabbed, and the swab was placed in peptone water (described above). The next day 100 µl of the liquid was taken and cultured on MacConkey agar (Difco, United States) supplemented with ceftriaxone (2 mg/L) (Botelho et al., 2015). A maximum of five lactose positive colonies were selected from each plate sample and placed on Chromocult coliform agar after 24 h of incubation at 37 • C, colonies of presumptive E. coli, positive for β-glucuronidase, were selected for additional analyses (Lange et al., 2013). All isolates confirmed to be E. coli from each sample were kept frozen at −80 • C in Tryptic Soy Broth medium (Difco, United States) with 15% glycerol.

PCR Amplification for Detection of β-Lactamase Genes
When samples were positive for ESBL-producing E. coli, one to five isolates selected per sample for further analysis. A total of 96 isolates were tested for the following resistance genes: bla SHV , bla TEM , bla CTX−M , and bla OXA ( Table 1). Bacterial DNA was extracted by boiling (Dashti et al., 2009), and PCR amplification reactions were performed in a volume of 25 µl containing 12.5 µl of 2 × Qiagen Multiplex PCR Master Mix (Qiagen GmbH, Hilden, Germany), 0.2 µM concentrations of each primer, and 2 µl of DNA template. The cycling parameters were as follows: an initial denaturation at 95 • C for 15 min; followed by 30 cycles of 94 • C for 30 s, 62 • C for 90 s, and 72 • C for 60 s; and with a final extension at 72 • C for 10 min. Amplification products were observed in agarose gel electrophoresis 1.5%, stained with Ethidium bromide at 100V for 45-60 min. The size of the amplified products was compared with the commercial (Invitrogen, United States) 100-bp ladder. The band size (bp) for each gene was: bla SHV , 237; bla TEM , 445; bla CTX−M , 593; and bla OXA : 813 (Fang et al., 2008).

DNA Sequencing and Analysis
Genomic DNA was extracted from the isolates using the Wizard R Genomic DNA Purification (Promega, United States) according to the manufacturer's instructions. Sequencing was carried out at the University of Minnesota Mid-Central Research and Outreach Center (Willmar, Minnesota) using a single 2 × 250-bp dualindex run on an Illumina MiSeq with Nextera XT libraries to generate ∼30-to 50-fold coverage per genome. Genome assembly of MiSeq reads for each sample was performed using SPAdes assembler with the careful assembly option and automated k-mer detection (Bankevich et al., 2012). The identification of genus and species of the isolates was carried out using fastANI (Jain et al., 2018) with a percentage >80% of identification. Acquired AMR genes, plasmid types were identified using ABRicate tool (version 0.8.13), Resfinder was the database used for the identification of resistance genes (Zankari et al., 2012); PlasmidFinder database for plasmid replicon identification (Carattoli et al., 2014).

Phylogenetic Analysis
Pan-genomic analysis was carried out with Roary (Page et al., 2015); the core genome of the isolates analyzed was defined with at least 99%. A maximum likelihood phylogenetic tree with (1,000 bootstrap replicates) was created based on the core genomes of the isolates using RaxML-NG (Kozlov et al., 2019). The phylogenetic tree was visualized using iTOL (Letunic and Bork, 2019). Additionally, multilocus sequence typing (MLST) (Larsen et al., 2012), based on seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA, and recA) and core genome (cgMLST) (Hansen et al., 2021) were performed using the Center for Genomic Epidemiology website 1 . The isolates also were characterized by Clermont phylogenetic typing by EzClermont web (Waters et al., 2020).
One hundred percent of the E. coli isolates from vegetables and fruits were resistant to ampicillin and cefazolin, cefotaxime, and tetracycline. Ninety-one percent of E.coli isolates from vegetables were resistant to cefepime. Two ESBL isolates from irrigation water presented resistance to the critically important class carbapenems, however no carbapenemase gene was detected. Additionally, we observed 33 resistance profiles across all of the extended spectrum beta-lactamase-producing E. coli isolates. The resistance profiles with the highest number of isolates are summarized in Table 3. In addition, 94% (90 of 96) of the E. coli ESBL isolates presented multi-drug resistant (MDR) patterns, with non-susceptible to at least one antibiotic in three or more antimicrobial categories (Magiorakos et al., 2012).
The application of a cgMLST scheme showed 55 cgSTs, from which only 2, cgST86226 (banana, Manabí, n = 5; irrigation water Pichincha, n = 1) and cgST135673 (banana Manabí, n = 3; irrigation water, Zamora Chinchipe n = 1) were isolates from two different sources. Several isolates belonging to the same ST (based on 7 genes) were assigned to different cgSTs based on cgMLST and some of the isolates from the same sample had the same cgST. Additionally, we constructed a maximum likelihood tree based on the core genomes to compare the phylogeny of isolates of E. coli from the irrigation water, vegetables, and fruits (Figure 2). The phylogenetic analysis showed that all isolates with the same cgMLST and obtained from different sources differed in thousands of SNPs indicating that although the isolates were genetically close, they have been evolving apart for many years ( Table 4 and Figure 2). The genomes of ESBL-E. coli isolates from irrigation and fresh produce did not cluster apart; instead the isolates form different sources seemed to share recent common ancestry (Figure 2).
The presence of AMR genes in the genome sequences of 80 ESBL-E. coli isolates was investigated by Resfinder. Several ESBL-encoding bla CTX−M gene variants were distributed in isolates from irrigation water and fresh produce (Figure 3). Among the 80 ESBL-E. coli isolates, we identified allelic variants of bla CTX−M in 77 (96%). The most common allelic variants were bla CTX−M−55 in 49 isolates (64%) and the second most common allele was bla CTX−M−65 in 14 isolates (18%) (Supplementary Table 1).
We found some discrepancies in some ESBL-E.coli isolates that were positive by PCR for some genes but negative by  whole genome sequencing (WGS): 12 isolates for bla TEM gene, 9 isolates for bla SHV genes and bla CTX−M in one gene. Additionally, 2 isolates showed bla SHV and bla TEM using WGS, but were negative by PCR. The WGS analysis of ESBL-E. coli allowed us to identify 2 isolates of E. coli from irrigation water and 3 isolates from banana with the presence of the mcr-1 gene that confers resistance to colistin.

DISCUSSION
In this study, we found that irrigation water, fruit, and vegetables were contaminated with ESBL-E. coli and the highest percentage was found in irrigation water (58%), which confirms the important and emerging role that irrigation water, contaminated with wastewater, has in the spread of ARB and ESBL E. coli and ESBL genes. (Gekenidis et al., 2018a;Vital et al., 2018). The major ESBL gene was the CTX−M (94 of 96 isolates) followed by bla SHV 28% (27 of 96), and bla OXA 1% (1of 96). The prevalence of bla CTX−M type ESBL genes in irrigation water E. coli was 57%, followed by 15% in banana isolates. Additionally the most abundant allelic variants of bla CTX−M found in vegetables, fruits and irrigation water (bla CTX−M55 , bla CTX−M65 , and bla CTX−M15 ) ( Table 4) are the same alleles found in children and domestic animals in Ecuador (Salinas et al., 2021), in rivers that cross cities (Ortega-Paredes et al., 2020a), and in bacteria from human infections in Ecuador (Cartelle Gestal et al., 2016;Soria Segarra et al., 2018). The presence of the same bla CTX−M alleles in isolates from different sources provides strong evidence that these sources (irrigation water, domestic animals, and humans) are connected. The allelic variants of bla CTX−M from isolates obtained from same European country, but from different (unconnected) sources, animal species or time periods, have been shown to be different (Day et al., 2019;Ludden et al., 2019).
Our genomic analysis showed that most strains obtained from irrigation water and produce were genetically different with 3 exceptions (HY1.4.3 and V427.2; HP6.1 and V661.1; HP1.4 and V662.1), however the number of SNPs between thes strains  ranged from 9,332 to 20,310 suggesting that these strains have been evolving apart for many years ( Table 4). As expected, some isolates from the same vegetable or fruit showed higher level of genetic closeness, for instance: V698.3 and V698.4 had 12 SNP; V663.4 and V663.5, 6 SNPs; V696.2 and V696.4, 13 SNPs; V1147.5 and V1147.1, 2 SNPs). Interstingly, 2 isolates obtaind from the same irrigation channel 1 month appart (HY3.5.2 and HY5.2.1) had 24 SNPs, suggesting that this strain was higly adapted to water. We did not find additional asociation of ESBL-E.coli clusters with provinces, which may indicate that different E. coli lineages have been widely distributed in the Ecuadorian territory (Figure 2). These findings may indicated that E. coli populations in the environment are highly diverse (Day et al., 2019;Ludden et al., 2019) and bla CTX−M -genes are probably diseminating in the environmet mostly by mobile genetic elements and not so much by bacterial clones. The plasmids carrying bla CTX−M -genes disseminate efficiently by conjugation, even between bacteria belonging to different genera (Cantón et al., 2012). Transposable elements (such as ISEcp1) are also very active in bla CTX−M -gene mobilization among different plasmids (Cantón et al., 2012). The activity of these MGEs conceals the source of origin of these antimicrobial resistance genes.
The majority of strains isolated from irrigation water and vegetables belonged to phylogroups A and B1 which are considered more generalists, found in most warm-blooded animals and environmental samples (Touchon et al., 2020). We found that some genetically close E.coli isolates, obtained from the same vegetable, had 1 or 2 additional antimicrobial resistance genes which may be a reflection of the dynamic process of antimicrobial resistance gene-turnover in the environment (Barrera et al., 2019).
The bla CTX−M type of ESBL gene is of increasing concern globally (Bevan et al., 2017), and is the predominant ESBL gene in both community and hospital-acquired infections (Manyahi et al., 2017;Fils et al., 2021). A troubling feature of bla CTX−Mbearing plasmids is their ability to capture additional resistance determinants, including carbapenemase genes (Partridge et al., 2012;Potron et al., 2013). Further analysis is necessary to understand whether the plasmids carrying bla CTX−M genes, in bacteria from irrigation water and produce, are the same as those circulating in bacterial isolates from human isolates.
In our study fruits, such as bananas, we hypothesize that their contamination was due to post-harvest processes in which the food is often washed in contaminated water and reused to wash several batches of the product. Although it is true, the skin of the product protects the fruit, the transmission of resistant bacteria can occur through contact and inadequate consumer hygiene (Harris et al., 2003;Hong et al., 2013;Kawamura et al., 2017;Murray et al., 2017;Hölzel et al., 2018).
We also found a higher prevalence of ARB in vegetables in farms than in retail markets in Ecuador (Ortega-Paredes et al., 2018). However, other reports from the Philippines, Lebanon, and Portugal have documented even higher levels (Faour-Klingbeil et al., 2016;Araújo et al., 2017;Vital et al., 2018). In most of the studies, the collection of produce samples has been carried out in groceries and wholesale markets, which makes it difficult to analyze sources of contamination (Bhutani et al., 2015;Yang et al., 2019;Colosi et al., 2020;Richter et al., 2020;Song et al., 2020). In this study, we collected produce and water from farms and their respective irrigation systems, which allowed us to study contamination at the source (i.e., not due to handling, transport, distribution, and processing). We found that MDR isolates were more prevalent in irrigation water isolates compared to fresh produce. Similar results were observed in the Philippines, where 58% of the E. coli isolates from irrigation water were MDR (Paraoan et al., 2017). The resistance to these antibiotics was also observed in E. coli isolates from irrigation water in other studies (Pignato et al., 2009;Ben Said et al., 2015;Vital et al., 2018).
Our study had some limitations; the number produce and fruit samples obtained in each location may not be representative of produce from other agricultural settings in Ecuador. Additionally, long-read sequencing of plasmids could not be carried out due to budgetary limitations.
We found evidence that fresh produce constitutes an important source of ESBL-E. coli and represents a route for the dissemination of resistance genes through the consumption of raw products (Rasheed et al., 2014;Hölzel et al., 2018;Al-Kharousi et al., 2019). We hypothesize that the main source of ABR contamination is irrigation water used for the cultivation of produce, which has been suggested by others as well (Pignato et al., 2009;Gekenidis et al., 2018b). In Ecuador, the lack of sewage treatment may lead to contamination of the food supply with ARB, mainly belonging to the Enterobacteriaceae family (Caicedo-Camposano et al., 2019;Ortega-Paredes et al., 2020a). Antibiotic resistant E. coli can transfer antibiotic resistance determinants not only to other strains of E. coli, but also to other species of potentially pathogenic bacteria within the gastrointestinal tract (Grasselli et al., 2008;Huddleston, 2014).

CONCLUSION
We found a high prevalence of ESBL-E. coli on produce and in irrigation water; bla CTX−M was the main ESBL gene in these isolates. Allelic variants of the bla CTX−M gene found in irrigation channels and vegetables were the same as those observed in commensal E. coli from domestic animals, and commensal and pathogenic E. coli from humans, suggesting connection between these different sources. This paradigm poses the potential risk of further spreading ARB that are resistant to last-line antibiotics such as carbapenems, which are used exclusively in serious infections in hospitals (Sheu et al., 2019). In this case, resistance goes full circle, from humans to vegetables and fruits (potentially meat and dairy), and back to human populations (Murray et al., 2021). Greater investments are needed to support the development and installation of wastewater treatment systems throughout Ecuador, as well as in other low-and middleincome countries.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
JI and LM: isolation of the Escherichia coli strains. LM: writingoriginal draft. JG, PC, and GT: review and editing. GT and LM: study design. All authors contributed to the article and approved the submitted version.

FUNDING
This work was partly supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under Award Number R01AI135118. The rest of the funding was provided by USFQ-COCIBA grants.