Escherichia coli is a Gram-negative enteric species that contains both commensal and pathogenic members [1]. Among the pathogenic strains, the shiga-toxin producing variants (STEC) account for an approximately 73,000 annual cases of reported human illnesses in the United States in an estimate made in 2002 [2]. The pathogenic forms are typically disseminated via incorrectly prepared food items, including ground beef products and produce, or contaminated water sources, including potable and recreational waters [3-5]. In addition to its role in human diseases, E. coli is also the preferred indicator organism for fecal contamination because it is the only member of the coliform group that is always tied to fecal matter from warm-blooded animals [6].

The Environmental Protection Agency’s data reviewed by Macler and Merkle [7] showed that 40,000 out of 156,000 ground water systems in the United States examined by either total coliform or fecal coliform count have exceeded the mandated level of Total Coliform Rule maximum. The presence of coliforms in over 25% of groundwater systems suggests that there is a widespread dissemination of enteric organisms in the environment. The typical sources of these contaminations include concentrated point sources (e.g. defective septic systems, leaky sewer lines, and discharges from wastewater treatment plants) and non-point sources (e.g. runoff from livestock farming and urban and rural runoff containing feces from companion animals or wildlife). After E. coli enters the water system, its dissemination is dependent on various geohydrological factors, such as the amount of particulates and the flow rate [8]. Recent studies have shown that once present in the environment, E. coli is capable of persisting and replicating under many different conditions [9]. In particular, when the environment is enriched with fecal material, E. coli is apparently able to flourish in localized areas and persist [10, 11]. The presence and persistence of E. coli in the environment raise significant concerns in tracking the dissemination pattern and following up on source-tracking efforts during disease outbreaks because they act as multiple reservoirs where rainfall or irrigation can serve to re-introduce fecal contaminants into the surrounding. As a result of this characteristic of E. coli, there is also concern about the spread of antibiotic resistance. In a study of water sources associated with an area with cattle farming, 24% of the 600 isolates tested showed antibiotic resistance to at least one antimicrobial agent, and many were resistant to more than one [12]. In a different study that examined the Karstic aquifers in France, the percentage of E. coli isolates that showed antibiotic resistance to at least one antimicrobial agent ranged from 17 to 70%, depending on the rainfall and seasonal factors [13].

Taken together, the presence of E. coli in the environment poses both immediate concerns due to the widespread cases of disease outbreaks associated with water sources contaminated with pathogenic variants of E. coli, as well as a long-term concern over the persistence of this organism as it relates to the spreading of antibiotic resistance. Therefore, a better understanding of the characteristics of E. coli in surface water sources will be beneficial in order to manage public health concerns and shed light on the dynamic nature of the underlying processes governing the distribution and persistence of this organism in our environment. In this study, surface water sources in Illinois and Missouri, U.S.A., were surveyed for the presence of E. coli. Isolates that were obtained were then assayed for antibiotic resistance and their genetic makeup on four loci related to pathogenicity was determined using a multiplex PCR procedure.

In this study,a survey of unrelated urban and rural surface waters (e.g. ponds, streams/creeks, lakes, and rivers) revealed the widespread distribution of E. coli across diverse sources. Historically, these results would suggest point and/or non-point sources of fecal contamination [16-20]. However, recent studies have shown that some E. coli strains may persist and replicate in natural environments such as soils, sediments, and surface waters [11, 12, 21-28]. Therefore, the presence of E. coli alonemay no longer represent recent fecal contamination and instead, may indicate the persistence of some strains in the environment. Future studies using longitudinal monitoring [29, 30] and microbial source tracking [31-33] might help shed more light on the persistence and origin, respectively, of E. coli at the sites surveyed in our study. Nevertheless, E. coli in surface waters is a public health concern, and our data showed that the isolates from different surface water sources possess different phenotypic and genotypic traits (Tables 3, 4, and 6; Fig. 1). Consequently, with their diverse E. coli populations, these surface waters have the potential to serve as reservoirs for the dissemination of currently existing strains as well as the emergence of new variants. In this regard, some of the phenotypic and genotypic attributes related to public health concerns include (i) metabolic richness, which may promote survival of different strains in the environment; (ii) possession of antibiotic resistance; and (iii) harboring of virulence factors.

Barcodes (also referred to as fingerprints or profiles), normally assigned to microorganisms based on phenotypic or molecular analyses, have been used over the years to identify, compare, and track microorganisms [34-39]. In the present study, the diversity of metabolic capabilities among isolates was revealed using the combined API 20E and Enterotube II identification systems, where 18 unique barcodes (i.e., metabolic fingerprints) were identified from the 45 isolates tested (Table 3). No apparent correlation was detected between isolate barcodes and the types of surface water sampled (Table 4). Furthermore, 6 isolates yielded a negative ONPG test result in the API 20E system, suggesting the absence of β-galactosidase activity; yet, these same 6 isolates were positive for lactose utilization in both the Enterotube II and EC broth assays. This discrepancy may be due to the slow conversion of the ONPG substrate under API 20E conditions or to the fact that some strains of E. coli are ONPG negative [40, 41].

The observed heterogeneity in metabolic potentials (barcodes) among our isolates may reveal stochastic events unrelated to selective pressures or it may be seen as evidence for differences in microhabitats among water sources serving as divergent selective pressures. Regardless, this diversity may broaden the ecological adaptability of the species. For instance, 28 isolates represented by 5 barcodes had the ability to utilize sucrose, which is not a trait identified in some reference strains such as E. coli B, C, and K-12 [42] and E. coli ATCC 25922 (this study; Table 3). This ability can provide an obvious survival advantage in environments where sucrose becomes available. Further, in light of the new studies suggesting that E. coli strains persistin the environment [11, 21-28], metabolic heterogeneity can become fodder for natural selection. Indeed, recent studies have shown that E. coli populations in surface waters are genetically diverse and often different than those in sediments or soils [12, 22, 23, 28].

Of the 45 isolates identified in this study, 42% (19 isolates) showed resistance to at least 1 of the 9antimicrobial drugs. In comparison, a survey of both sediment and surface water in California and a survey of a karstic system in France showed that 24% (out of 600) and 20%(out of 392) E. coli isolates carried at least 1 resistance, respectively [12, 13]. In our survey, 4 different drugs were ineffective against at least 1 isolate: cephalothin, triple sulfa, ampicillin, and tetracycline. The 4 most common ineffective antimicrobial drugs in the California and France surveys were rifampicin, erythromycin, tetracycline, and ampicillin and tetracycline, chloramphenicol, amoxicillin, and ticarcillin, respectively [12, 13]. Moreover, amoxicillin, cephalothin, ticarcillin, and sulfa drugs were also found to be ineffective against manynon-E.colienteric isolates [43], suggesting that the resistance to these drugs is widespread among members of the Enterobacteriaceae.

While our survey identified 4 isolates with resistance to 3 different antimicrobial drugs (ampicillin, cephalothin, and triple sulfa), isolates that are resistant to up to 9 different drugs have been discovered [13]. The 4 isolates in our survey were found in 4 different barcode groups and originated from 4 different geographical areas. Notably, two of the barcodes that contained isolates with resistance to 3 antimicrobial drugs also contained isolates with resistance to fewer than 3 antimicrobial drugs. At the same time, 19 of the 37 sites sampled yielded at least 1 isolate with antibiotic resistance. Taken together, our data suggested that there was no direct correlation between the antibiotic resistance of isolates and either their geographic origin or metabolic capabilities. Nonetheless, the presence of resistant organisms in natural surface waters is a concern for both public health issues and for their potential impact on natural microbiota due to their role as natural reservoirs of resistance phenotypes, especially when the tendency of these organisms to replicate and persist in the environment is also taken into consideration [44]. With resistant E. coli types present in the environment, the chance of lateral gene transfer events, like the acquisition of a broad-spectrum β-lactamase determinant bla(CTX-M-15) by a pathogen, is higher [45]. Of the three mechanisms known to contribute to horizontal gene transfer (conjugation, transformation, and transduction), the conjugal transfer of plasmids between microorganisms plays a pivotal role in the evolution of antibiotic resistance and virulence among bacteria including the emergence of novel variants of E. coli [45-48]. Whether plasmids (conjugal or otherwise) contributed to the antibiotic resistance among ourE. coliisolates is unknown. Studies in the future which target the plasmid profiles of our isolates would help provide some insight into the nature of the observed resistance.

In this regard, recent comparative genomic analyses have shown that several virulence factors in pathogenic E. coli strains originated from lateral gene transfer events [49, 50]. Multiplex PCR targeting 3 known virulence genes showed that none of the isolates in our survey contained the stx₁/stx₂ or hlyA genes, while 2 isolates with different barcodes for metabolic capacities possessed the eaeA gene (Fig. 1). These findings appear to rule out the possibility that any of our isolates were enterohemorrhagic (EHEC) or uropathogenic (UPEC) strains of E. coli, but the detection of eaeA indicated that the enteropathogenic E. coli (EPEC) pathotype may be present [51-53]. In comparison, one study reported that the intimin gene (eaeA) was present in almost 90% of all E. coli isolates identified in different estuaries, while the stx₁and stx₂ genes were found in about 25% of isolates [54]. Interestingly, in that study, all the isolates carrying the shiga toxin genes also carried the intimin gene.

The uidA gene has been used in PCR assays for E. coli as a control amplicon due to the shared ability to cleave the Colisure^® indicator 4-methylumbelliferyl-β-D-glucuronide (MUG) among different strains [15]. In some cases, uidA has even been detected in E. coli strains that tested negative for MUG utilization [15]. Of the 45 isolates, 44 produced a uidA amplicon (Fig. 1) of the correct size and were MUG positive in Colisure^® tests (samples fluoresced under long-wave UV light). However, one isolate (LC; Fig. 1, lane 4) did not produce a uidA amplicon despite being MUG positive. This aberrant result may be due to sampling methods where only one colony per water sample was assayed with PCR while there may have been multiple strains present in the sample. Furthermore, despite positive and negative PCR controls, failed PCR amplifications as a result of unidentified inhibitory residues in the DNA extract or sufficient sequence divergence at the primer-annealing sites cannot be ruled out. Lastly, the possibility also exists that this is a new variant capable of cleaving MUG with a novel mechanism.

In summary, our results demonstrated the widespread dispersion of E. coli strains of diverse characteristics (metabolic capabilities, antimicrobial resistance, and presence of pathogenic loci). This heterogeneity in natural E. coli populations has direct implications for interpreting the impact of fecal coliforms on our natural waters and public health. Further studies that can provide more detailed analyses of host origins, geographical sources, persistence behaviors, and dispersal patterns of E. coli strains will be required. Examinations on the fitness and survival qualities of these variants will also be beneficial.