Shiga toxin-producing Escherichia coli (STEC) are an intestinal pathotype of E. coli, currently comprising more than 400 serotypes (Blanco et al., 2004) their common feature being the production of Shiga toxin (also called Vero-cytotoxin, Stx), a potent cytotoxin inhibiting protein synthesis in the affected cell lines. A highly virulent subset of STEC strains termed enterohemorrhagic E. coli (EHEC) produce intimin apart from Stx, an adhesin that is responsible for attaching/effacing lesion on intestinal epithelial cells (Kaper et al., 2004). In addition to hemorrhagic colitis, EHEC cause a severe systemic complication called hemolytic-uremic syndrome (HUS), which results in acute renal failure in affected patients (Bielaszewska and Karch, 2005).

The primary reservoir of EHEC isolates is cattle, with isolates being present in the feces of the animals (Newell and Ragione, 2018). Human infection can result from direct contact with the animals or with their feces but occurs more frequently as a result of the consumption of undercooked meat, or of fresh vegetables, drinking water or meat contaminated with cattle feces (Chekabab et al., 2013; Terajima et al., 2017). The domestic animal reservoir, the severity of the disease, and low infective dose (<100 cells/person, Pulz et al., 2003), make EHEC serious and deadly zoonotic and foodborne pathogens. Enteropathogenic E. coli (EPEC) is the first described pathotype of E. coli and closely related to EHEC, responsible for a high proportion of children’s diarrhea mainly in countries with poor infrastructures (Ochoa and Contreras, 2011). Its key virulence factor is intimin, but unlike EHEC they do not produce Stx (Croxen et al., 2013). A subset of EPEC termed atypical EPEC (aEPEC) which are isolated from both humans and animals, do not carry the gene encoding the bundle-forming pilus (bfp). These isolates are genetically more closely related to EHEC, and it was recently suggested that subsets of aEPEC strains are possible progenitors of EHEC strains (Ferdous et al., 2015; Eichhorn et al., 2018).

An important feature of STEC, EHEC and EPEC is that their major virulence factors are carried on mobile genetic elements (MGE). Genes encoding stx are generally carried by lambdoid prophages, which are frequently inducible and transmissible (Ohnishi et al., 2002; Asadulghani et al., 2009; Rahman et al., 2018). In the case of both EHEC and EPEC, the eae gene encoding intimin is part of a 35–43 kb long pathogenicity island (PAI) termed locus of enterocyte effacement (LEE; reviewed by Jores et al., 2004), which carries genes encoding the type III secretion system, as well as effector proteins including translocated intimin receptor (tir) gene.

Highly virulent EHEC strains comprise members of the O157:H7 serotype (Hayashi, 2001; Perna et al., 2001; Paletta et al., 2020). Monitoring cattle for the presence of these strains has become a necessity, in order to understand the factors and risks associated with the transmission of EHEC O157 strains from animals to humans, as well as for epidemiological tracing.

The availability of affordable whole-genome sequencing (WGS) enables rapid assessment of genetic information encoded by a large set of strains, shedding light on of yet unknown genetic variants, additional virulence genes, and associated MGEs (Kulasekara et al., 2009). Genomic studies characterizing E. coli O157 strains revealed a remarkable genetic diversity of the serogroup (Ogura et al., 2006; Eppinger et al., 2011) especially regarding their prophage content (Shaaban et al., 2016; Sharma et al., 2019). A WGS-based study revealed that a specific subset of strains belonging to the serogroup, are frequently carried by cattle and responsible for the serious human outbreaks (Lupolova et al., 2016). It is therefore necessary to investigate cattle herds for the presence of these highly virulent strains and use the genome-based information to explore their genetic variability and assess their zoonotic potential. As the example of the recent outbreak caused by O80:H2 strains representing hybrid pathotype of STEC and ExPEC has shown (Nüesch-Inderbinen et al., 2018), the emergence of new, highly virulent pathotypes or clones is an ever-present possibility.

A previous explorative sampling of Hungarian cattle farms, which provided the first information on the prevalence of pathogenic strains of the O157 serogroup among cattle herds in Hungary used comprehensive PCR-based virulence profiling, and a classification based on pulsed-field gel electrophoresis (PFGE) together with multi-locus sequence typing (MLST) profiling of the isolated strains (Tóth et al., 2009). In addition to EHEC O157:H7, EPEC isolates of the same serotype, as well as those with other H types and atypical virulence gene repertoires are also present in cattle (Sváb et al., 2016).

The main aim in this study was to obtain further information regarding the prevalence of potentially zoonotic E. coli on Hungarian cattle farms, with a focus on the O157 serogroup and the STEC, EHEC and EPEC pathotypes. Using a WGS-based approach, we aimed to reveal potentially new genetic variants, virulence gene sets and MGEs, as well as to compare these data obtained from cattle to those of strains isolated in the earlier study together with those originating from human patients. In addition to assessing the occurrence of strains with high zoonotic potential, we also aimed to reveal potentially new genotypes, comprehensively map their MGEs, as well as their integration sites in the respective genomes.

In this study, representative sampling of Hungarian cattle farms was carried out to detect STEC, EHEC and EPEC. The prevalence of aEPEC isolates was 8.3% (19/228), while that of STEC it was 8.7% (20/228), and that of EHEC alone 3.9% (9/228). These values are congruent with earlier findings and suggest that Hungary has an occurrence of STEC, EHEC and EPEC isolates that is comparable to other countries of the developed world (Hussein, 2007; Tóth et al., 2009; Dong et al., 2017). It must be noted however, that our method for isolation was optimized for the isolation of strains of the O157 serogroup, therefore strains of all pathotypes representing other serogroups could be underrepresented in our collection. In the earlier study of Tóth et al. (2009), twice as many EPEC isolates were detected as compared to EHEC among the O157:H7 serotype. Two such aEPEC isolates (64-2 and 65) from the earlier study had highly similar virulence gene arrays to EHEC isolates in the current study as well as from humans suggesting that there are several aEPEC ancestral isolates of EHEC, essentially only lacking the Stx prophage (Ferdous et al., 2015; Eichhorn et al., 2018). At the same time, it cannot be excluded that these aEPEC strains might have lost their Stx prophage(s) (Bielaszewska et al., 2008). As none of the EPEC in our current study carried the bfp gene, all of them fall into the category of aEPEC. It is also apparent that such strains are present in similar prevalence among Hungarian cattle like the STEC strains. There was a tendency of strains originating from the same farms to harbor key virulence gene carrying MGEs in the same integration sites, as well as similar sets of additive virulence genes and prophages.

The genomic variability of the investigated pathotypes is most significant with respect to virulence gene repertoires. Regarding the stx genes, only the stx1a subtype was found in all stx1+ strains, while stx2, where present, showed greater variability (stx2a, n = 3; stx2c, n = 8; stc2d, n = 1; Table 1). Almost all stx2 subtype a and c were detected in O157:H7 strains (2/3 and 6/8, respectively), whereas the single subtype d was detected in an O157:H4 serotype isolate (ML2/7C, ST1794). In the case of the strain ML2/7C the integration site for the Stx2 prophage could not be identified, as all the other characteristic integration sites were either intact or occupied by other prophages. This finding suggests the presence of an alternative, hitherto unknown integration site for the Stx2 prophage in this strain, a possibility that has been recently raised for STEC strains (Henderson et al., 2021). It is also noteworthy that the aEPEC strains had at least two of the typical Stx prophage integration sites unoccupied, therefore they can be considered as possible recipients of this MGE and have potential of becoming EHEC (Tables 3, 4).

The eae genes represented five types (β1, γ1, ε1, ζ3, θ). The most common subtype was γ1 (n = 11) and was detected almost exclusively in O157:H7 strains (exception: 1511,: H26) irrespective of their source (human/animal). The second most common subtype was β1, which was detected only in O26:H11 strains with the “KP” designation, originating from the same farm. These findings confirm the notion that the carriage of intimin subtypes is associated to serotypes of EHEC and EPEC (Oswald et al., 2000; Tarr and Whittam, 2002).

A notable difference could be observed in the integration sites used by the LEE when comparing aEPEC and EHEC strains. In EPEC strains, LEE was almost exclusively (5/6) integrated in pheV (exception: selC), while for EHEC strains, LEE was integrated either in selC (n = 9) or pheU (n = 3). This difference could be attributed to the different clonal lineages of strains representing the two pathotypes (Ingle et al., 2016).

Besides the stx genes and LEE-encoded effectors, additional virulence-related genes were identified all the strains. Except for ML2/7C, ML5/5A and ML5/10A, all EHEC strains as well as the aEPEC strain 65 carried the ehxA encoding enterohemolysin, which suggests to the presence of the pO157 plasmid in their case. This distribution is also congruent with the results of Arimizu et al. (2019), who found that STEC strains retained ehxA more frequently than stx-negative ones. The presence of at least one pO157-associated effector gene suggests the carriage of this virulence plasmid (Lim et al., 2010) by all the WGS strains in this study, which belonged to the O157:H7 serogroup. The majority of the aEPEC and EHEC strains harbored effectors of the OI-122 genomic island (efa1 and the non-LEE effectors nleABCD) widespread in these pathotypes (Karmali et al., 2003; Konczy et al., 2008) as well as additional adhesines, siderophores and other fitness factors (Table 3). Out of these, the LEE-negative strains (both STEC, commensal and the ExPEC ML2/9C) tended to carry alternative adhesin encoding genes: the iha, genes (Tarr and Whittam, 2002), and lpfA, the latter being widespread in various intestinal E. coli pathotypes (Zhou et al., 2019). Genes encoding the iron siderophore enterobactin (Raymond et al., 2003) were universally present, and the prevalence of the iss gene encoding an increased serum survival (Johnson et al., 2008) was also high.

Based on its virulence gene array containing S fimbria (Dobrindt et al., 2001) and on being a member of ST405, strain ML2/9C was categorized as ExPEC, with strains of this ST being reported as emerging uroseptic pathogens (Chowdhury et al., 2019). Strains of this pathotype are rarely reported from cattle (Bélanger et al., 2015; Schmidt et al., 2015).

The high number of prophages harbored by the sequenced genomes support the notion that these MGEs play a significant role in the evolution and genomic variability of intestinal E. coli pathotypes (Ogura et al., 2009; Shaaban et al., 2016). It must be noted however, that as we worked with draft genomes, probably not all prophage regions were identified in the genomes. This might be due to the fact that some phages are highly homologous in certain regions and assembling programs cannot distinguish between these regions but assemble these different regions into one single contig.

The distribution of antimicrobial resistance genes among the strains underlines the notion that these determinants are in circulation among bovine zoonotic E. coli, (Kennedy et al., 2017; Um et al., 2018) which can probably be attributed to the pressure of antimicrobials applied against animal pathogens (Hammerum and Heuer, 2009). The number of strains harboring antibiotic resistance genes was comparably low (5/26 of sequenced strains), probably owing to the banning of their use as growth-promoters in 2006 by the European Union. Only the ExPEC strain of bovine origin, ML2/9C showed a multidrug-resistance genotype, harboring genes encoding resistance toward eight different classes of antibiotics, including the ESBL gene bla_CTX-M-15.

MLST analysis showed that all O157:H7 strains, regardless of pathotype, belonged to ST11, which is in line with earlier findings (González-Escalona and Kase, 2017; Galarce et al., 2021), while the EHEC O26:H11 strains represented ST21, which is also typical for strains of these sero- and pathotype (Abdalhamid et al., 2019; Galarce et al., 2021). A remarkable finding was that one bovine STEC, as well as two bovine and one human aEPEC strain belonged to ST10, which contains strains of various sero- and pathotypes, and is frequently isolated from human and animal sources (Manges et al., 2015).

We also attempted to place our strains on a global phylogenetic relationship of STEC and aEPEC strains, using a core-genome-based approach and the sequence set published by Arimizu et al. (2019). We found that the placement of bovine strains on the tree was generally aligned with their isolation source, and strains isolated from the same farm (i.e., the “KP” or “ML6” strains) tended to cluster together (Figure 1). The strains of human origin, however, were placed on clades with predominantly bovine isolates. This notion can be indicative of their close geographic origin, the genomic variability of the pathotype in general, as well as the zoonotic risk posed by the bovine strains isolated in our study.

In conclusion, we determined the WGS of 22 E. coli strains representing the STEC, EHEC, aEPEC and ExPEC pathotypes, both of human and bovine origin from Hungary, together with four commensal E. coli from bovine sources. Our results underline the pathogenic potential of bovine strains to humans, as EHEC and aEPEC strains of human origin as well as several bovine strains of the same pathotype harbored very similar virulence gene sets. The phylogenetic position of several human strains within the “bovine-like” lineages corroborates this notion. Several strains harbored Stx prophages with unusual combinations of integration sites. Comparative investigation of stx-negative, LEE-negative strains revealed the carriage of a small number of virulence genes, as well as a potential to carry virulence-related MGEs. The prophage content of the strains showed various profiles, with several widespread prophages with functions yet to be identified. Investigating the inducibility of prophages could be important toward understanding their role in the dissemination of virulence and fitness genes, as well their role in determining architecture of various STEC and aEPEC lineages within cattle populations.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih.gov/, PRJNA764596, SAMN21509413-SAMN21509438, and SRR15970245-SRR15970246.

IT and TC conceived and outlined the study. DS performed strain isolation and preliminary genetic characterization and participated in the genomic analysis as well. TM provided strains and performed serotyping. LF performed sequencing and bioinformatical analysis. DS, LF, and IT wrote the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by the National Research, Development and Innovation Office (NKFIH), grant number K 124335 and the Bundesministerium fuer Bildung und Forschung (BMBF, Germany) within the German Center for Infection Research (DZIF; grant nos. 8032808818 and 8032808820 to TC) and the State Ministry of Higher Education, Research and Arts of the state Hessen (HMWK) through the HuKKH project (Hessisches universitaeres Kompetenzzentrum Krankenhaushygiene). DS was supported by the János Bolyai Research Scholarship of the Hungarian Academy of Sciences.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.