Enteropathogenic Escherichia coli (EPEC) was first identified in the United Kingdom in the 1940s as the cause of outbreaks of infantile diarrhea (Bray, 1945). Although such outbreaks are now rare in developed countries, EPEC strains continue to be a leading cause of diarrhea among infants from developing countries (Chen and Frankel, 2005). The most important feature of EPEC pathogenesis is its ability to produce characteristic histopathological intestinal lesions known as “attaching and effacing” (A/E) lesions that are also characteristic for some Shiga toxin-producing E. coli (STEC) (Jerse et al., 1990). The genes responsible for this activity in EPEC are encoded on a 35-kb pathogenicity island (PAI) called the locus of enterocyte effacement (LEE), whereas the LEE of STEC contains some additional genes encoded within a putative prophage designated 933L (Perna et al., 1998). Both EPEC and STEC LEE encode a type III secretion system, multiple secreted proteins, and a bacterial adhesin called intimin (Nataro and Kaper, 1998). The 5′ regions of intimin-encoding gene eae are conserved in both EPEC and STEC, whereas the 3′ regions are heterogeneous even within the same pathotype (Blanco et al., 2004). Several variants of the eae gene encoding different intimin types and subtypes are described (Oswald et al., 2000; Zhang et al., 2002; Blanco et al., 2006) and it has been suggested that different intimins may be responsible for different host tissue cell tropisms (Torres et al., 2005).

EPEC, carrying EPEC adherence factor plasmid (pEAF) is named as typical EPEC (tEPEC), while those without the pEAF as atypical EPEC (aEPEC) (Trabulsi et al., 2002; Hernandes et al., 2009). aEPEC are thought to be a concern as they can acquire Shiga toxin (Stx) converting bacteriophages, thereby obtaining the ability to cause more serious illness (Bolton et al., 2014). Besides the virulence genes encoded on LEE, EPEC strains also carry other virulence genes including astA (heat stable enterotoxin), the cdt (cytolethal distending toxin) gene cluster, efa1 (enterohemorrhagic E. coli factor for adherence), and paa (porcine attaching-effacing associated protein) (Bouzari and Varghese, 1990; Badea et al., 2003; Batisson et al., 2003; Dulguer et al., 2003).

Antibiotic resistance is a global concern due to the increased use of antibiotics, especially for the intestinal organisms as the gut is a heavily populated niche and resistance genes can be transmitted horizontally via these resistant organisms in the gut (Scaletsky et al., 2010). High prevalence of antimicrobial resistance among EPEC strains has been reported in different countries but the genetic basis for this resistance and the evolutionary consequences are rarely studied (Senerwa et al., 1991; Abe et al., 2009; Medina et al., 2011).

The serotyping of O antigens (together with the H-flagellar antigen) is used as an effective method to identify various pathogenic clones (Iguchi et al., 2011). The serotype O157 is a predominant serotype of the documented STEC related outbreaks worldwide (Nataro and Kaper, 1998; Caprioli et al., 2005) and is frequently associated with the H7 antigen (encoded by fliC H7). E. coli O157:H7 strains that do not possess Stx are presumably the Stx-lost variants or progenitors of STEC as they share the same phylogenetic lineage (Ferdous et al., 2015). However, the large and diverse O157 serogroup also includes many non-H7 serotypes that are commonly found in animals, food, and clinical samples. EPEC strains of the O157:H45 serotype have been noted as agents of diarrhea in outbreaks and in sporadic cases in Germany, Japan, Korea, and Thailand (Blank et al., 2003; Park et al., 2014). Strains of serotype O157:H8 and O157:H16 have been isolated from cases of diarrhea in human, whereas the latter serotype was also isolated from cattle, beef, and water (Feng et al., 2010; Iguchi et al., 2011). As these strains do not carry the stx gene, they are often not detected or no detailed characterization is carried out (Feng et al., 2010). Recently, two studies were performed on comparative genomics analysis and phylogenetic relationship of E. coli O157:H7 and non-H7 strains (Sanjar et al., 2015; Kossow et al., 2016). However, there is a lack of information regarding their virulence, antimicrobial resistance properties and some other molecular features. In our current study, apart from the phylogenetic and evolutionary relationship of E. coli O157:non-H7 groups, a comprehensive characterization was performed focusing on their diversity in virulence and antimicrobial resistance properties.

E. coli O157 strains are mostly known as enterohemorrhagic E. coli (EHEC) that have been associated with food born outbreaks worldwide. But they are also comprised of other E. coli pathogroups, including ones containing eae and lacking stx, which are considered as EPEC. E. coli O157:H7 isolates lacking stx gene are assumed to be members of STEC that lost the Stx encoding bacteriophages (Ferdous et al., 2015). The current study was performed to compare the molecular features and virulence properties of EPEC O157 having different H antigens (other than H7) with those of STEC O157:H7 isolates and to determine their phylogenetic relationship. STEC O157 isolates generally belong to ST11 and carry the fliC H7 gene encoding flagellar antigen H7 but the EPEC O157 isolates appeared to be from different sequence and H serotypes. The stx negative O157 isolates in this study were usually sorbitol fermenting (SF) and some of them were non-motile. Therefore, they could be misidentified as STEC SF O157:NM if proper molecular tests are not performed (Feng et al., 2010). For some draft genomes obtained from NCBI, prediction of the O157 serotyping gene wzx was difficult probably because of the breaking of contigs in the middle of the gene sequence, therefore a lower threshold of the percentage of minimum overlapping gene length was applied to confirm the genoserotype.

Intimin types of the isolates correlated to their genoserotypes; four of the genomes did not possess intimin and therefore were not considered to be EPEC. Our phylogenetic analysis based on cgMLST confirmed that E. coli O157:H7 belong to a different lineage than the non-H7 group, the latter belonging to diverse phylogenetic lineages (Sanjar et al., 2015). Our results support previous findings that horizontal transfer of the O157 antigen gene cluster has occurred independently among different E. coli lineages (Iguchi et al., 2011). Interestingly, isolates of O157:H45 clustered together independent of being a tEPEC or aEPEC. However, strain C844-97 was previously described to have lost its EAF plasmid (Hazen et al., 2013), and this may also be the case for strain C639-08 explaining their aEPEC features in spite of clustering together with tEPEC strains. Although the whole bfp gene cluster was present in tEPEC strains RN587/1 and ARS4.2123, they did not contain the complete EAF plasmid.

Almost all of the O157:H39 isolates included in this study belonged to recently evolved STs (e.g., ST4554 and ST5611). Alternatively, they were not reported previously, because no further characterization was performed as they did not carry the stx gene. We observed the presence of several plasmids, insertion elements and PAIs in O157:H39 isolates contributing to their virulence and resistance features. Three of the O157:H39 isolates of the new sequence type ST4554, shared an identical core genome but no epidemiological link could be identified among the patients from whom they were isolated.

The distribution of virulence genes almost perfectly correlated to the STs although slight intra-ST variations in virulence patterns were observed. The virulence genes were often carried on plasmids, PAIs, prophages, and insertion elements suggesting that these virulence factors were acquired from numerous sources via MGE, found specifically in the genome of pathogenic strains (Donnenberg and Whittam, 2001). The lineage specific acquisition of specific combinations of virulence factors may confer selective advantages contributing to the fitness of the pathogens favoring their establishment and transmission as new virulent clones (Reid et al., 2000). It was observed that ten of the O157:H16 isolates of ST10 (including one isolate from the US, one non-human isolate from France, and the rest from the Netherlands) shared almost identical virulence properties and were closely related in cgMLST with a maximum of a 41-allele difference (data not shown). A similar finding was observed in a previous study where O157:H16 strains were described as representatives of a widespread clone (Feng et al., 2010). However, the other two O157:H16 isolates, 3006 and Santai, belonged to two different STs, ST5502, and ST1011, respectively, and possessed almost no virulence genes. They were not classified as EPEC supporting the idea that the O and H serotyping might not represent the virulence status of the isolates and that the distribution of O and H antigen encoding genes is irrespective of the E. coli pathotype (Blank et al., 2003). Moreover, aEPEC isolates analyzed in this study were distributed in two completely different phylogenetic branches; one containing O157:H39 and O157:H45 (closer related to tEPEC) and another containing O157:H16 and O157:H26. This may be explained by horizontal transfer of the LEE PAI to different E. coli groups contributing to their EPEC feature (Feng et al., 2010).

Strain Santai belonging to the phylogenetic group D as does STEC O157:H7, appears to be more a multidrug resistant strain than a virulent one. The resistance genes were integrated into the chromosome of this organism through insertion elements and transposons from different species. EPEC 2827 was also resistant to several antibiotics and carried the drug resistance genes on a plasmid very similar to that of S. enterica (Hooton et al., 2014) and on an insertion element similar to a class1 integron of Proteus sp. (Guo et al., 2011). Isolates of ST4554 carried a plasmid similar to pCER1, a small, globally disseminated plasmid harboring sulphonamide and streptomycin resistance genes (Anantham and Hall, 2012). EPEC 1316 carried the resistance genes on a plasmid similar to pVR50A which also carries a tra gene cluster encoding conjugal transfer proteins (Beatson et al., 2015). Clearly, the transmission of MGEs plays a role in the evolution of new MDR lineages (Beceiro et al., 2013; Punina et al., 2015). EPEC infections can normally be recovered without antimicrobial therapy, therefore, the persistence of resistant EPEC strains is more likely to be related to selective pressure from antimicrobials applied at the population level or present in the environment (Scaletsky et al., 2010). Indeed the presence of resistance genes in E coli that originated from the environment have been described before (Hamelin et al., 2007).

In conclusion, E. coli O157:non-H7 isolates have evolved from multiple genetic backgrounds. They are distantly related to the most virulent EHEC clonal lineage EHEC O157:H7. Diversity in virulence and antibiotic resistance patterns have been observed among these O157:non-H7 isolates and mobile elements carried by these organisms could play a role in the evolution and dissemination of these virulent and resistant clones.

Study design: MF, JR, AF. Performing Experiments: MF. Acquisition of data: MF, AK, JR. Analysis and interpretation of data: MF, KZ, JR, and AF. Drafting of manuscript: MF, KZ, AK, JR, and AF. Critical revision: MF, KZ, AK, JR, and AF.

This study was partly supported by the Interreg IVa-funded projects EurSafety Heath-net (III-1-02 = 73) and SafeGuard (III-2-03 = 025).