Analysis of the Clonal Relationship of Serotype O26:H11 Enterohemorrhagic Escherichia coli Isolates from Cattle

Shiga toxin (Stx)-producing Escherichia coli (STEC) strains comprise a group of zoonotic enteric pathogens (42). In humans, infections with some STEC serotypes result in hemorrhagic or nonhemorrhagic diarrhea, which can be complicated by hemolytic-uremic syndrome (HUS) (49). These STEC strains are also designated “enterohemorrhagic E. coli” (EHEC). Consequently, EHEC strains represent a subgroup of STEC with a high pathogenic potential for humans. Strains of the E. coli serogroup O26 were originally classified as enteropathogenic E. coli due to their association with outbreaks of infantile diarrhea in the 1940s. In 1977, Konowalchuk et al. (37) recognized that these bacteria produced Stx, and 10 years later, the Stx-producing E. coli O26:H11/H− strains were classified as EHEC. EHEC O26 strains constitute the most common non-O157 EHEC group associated with diarrhea and HUS in Europe (12, 21, 23, 24, 26, 27, 55, 60). Reports on an association between EHEC O26 and HUS or diarrhea from North America including the United States (15, 30, 33), South America (51, 57), Australia (22), and Asia (31, 32) provide further evidence for the worldwide spread of these organisms. Studies in Germany and Austria (26, 27) on sporadic HUS cases between 1996 and 2003 found that EHEC O26 accounted for 14% of all EHEC strains and for ∼40% of non-O157 EHEC strains obtained from these patients. A proportion of 11% EHEC O26 strains was detected in a case-control study in Germany (59) between 2001 and 2003. In the age group <3 years, the number of EHEC O26 cases was nearly equal to that of EHEC O157 cases, although the incidence of EHEC O26-associated disease is probably underestimated because of diagnostic limitations in comparison to the diagnosis of O157:H7/H− (18, 34). Moreover, EHEC O26 has spread globally (35). Beutin (6) described EHEC O26:H11/H−, among O103:H2, O111:H, O145:H28/H−, and O157:H7/H−, as the well-known pathogenic “gang of five,” and Bettelheim (5) warned that we ignore the non-O157 STEC strains at our peril.

EHEC O26 strains produce Stx1, Stx2, or both (15, 63). Moreover, these strains contain the intimin-encoding eae gene (11, 63), a characteristic feature of EHEC (44). In addition, EHEC strains possess other markers associated with virulence, such as a large plasmid that carries further potential virulence genes, e.g., genes coding for EHEC hemolysin (EHEC-hlyA), a catalase-peroxidase (katP), and an extracellular serine protease (espP) (17, 52). The efa1 (E. coli factor for adherence 1) gene was identified as an intestinal colonization factor in EHEC (43). EHEC O26 represents a highly dynamic group of organisms that rapidly generate new pathogenic clones (7, 8, 63).

Ruminants, especially cattle, are considered the primary reservoir for human infections with EHEC. Therefore, the aim of this study was the molecular characterization of bovine E. coli field isolates of serogroup O26 using a panel of typical virulence markers. The epidemiological situation in the beef herds from which the isolates were obtained and the spatial and temporal behavior of the clonal distribution of E. coli serogroup O26 were analyzed during the observation period. The potential risk of the isolates inducing disease in humans was assessed.

In our study, 56 bovine E. coli O26:H11 isolates and one bovine O26:H32 isolate were analyzed for EHEC virulence-associated factors. The isolates had been obtained from three different beef farms during a long-term study. They were detected in eight different cattle in farm A over a period of 15 months (detected on 10 sampling days), in 3 different animals in farm C over a period of 8 months (detected on 3 sampling days), and in one cow on one sampling day in farm D (Table 1) (28).

The serotyping of the O26 isolates was confirmed by the results of the fliC PCR-restriction fragment length polymorphism (RFLP) analysis performed according to Fields et al. (25), with slight modifications described by Zhang et al. (62). All O26:H11 isolates showed the H11 pattern described by Zhang et al. (62). In contrast, the O26:H32 isolate demonstrated a different fliC RFLP pattern that was identical to the H32 pattern described by the same authors. It has been demonstrated that EHEC O26:H11 strains belong to at least four different sequence types (STs) in the common clone complex 29 (39). In the multilocus sequence typing analysis for E. coli (61), the tested five EHEC O26:H11 isolates (WH-01/02/003-1, WH-01/20/005-3, WH-01/27/009-9, WH-03/14/004-8, and WH-04/22/001-1) of different farms and clusters were characterized as two sequence types (ST 21 and ST 396). The isolates from farms A and C belong to ST 21, the most frequent ST of EHEC O26:H11 isolates found in humans and animals (39), but the single isolate from farm D was characterized as ST 396.

Typing and subtyping of genes (stx₁ and/or stx₂, eae, tir, espA, espB, EHEC-hlyA, katP, and espP) associated with EHEC were performed with LightCycler fluorescence PCR (48) and different block-cycler PCRs. To identify the subtypes of the stx₂ genes and of the locus of enterocyte effacement-encoding genes eae, tir, espA, and espB, the PCR products were digested by different restriction endonucleases (19, 26, 46). The complete pattern of virulence markers was detected in most bovine isolates examined in our study. An stx₁ gene was present in all O26 isolates. In addition, an stx₂ gene was found in nine O26:H11 isolates in farm A and in three isolates of the same type in farm C, as well as in the O26:H32 isolate. Both Stx1 and Stx2 were closely related to families of Stx1 and Stx2 variants or alleles. EHEC isolates with stx₂ genes are significantly more often associated with HUS and other severe disease manifestations than isolates with an stx₁ gene, which are more frequently associated with uncomplicated diarrhea and healthy individuals (13). In contrast to STEC strains harboring stx₂ gene variants, however, STEC strains of the stx₂ genotype were statistically significantly associated with HUS (26). The stx₂ genotype was found in all O26 isolates with an stx₂ gene, while the GK3/GK4 amplification products after digestion with HaeIII and FokI restriction enzymes showed the typical pattern for this genotype described by Friedrich et al. (26). The nucleotide sequences of the A and B subunits of the stx₂ gene of the selected bovine O26:H11 isolate WH-01/27/017-1 (GenBank accession no. EU700491) were identical to the stx₂ genes of different sorbitol-fermenting EHEC O157:H− strains associated with human HUS cases and other EHEC infections in Germany (10) and 99.3% identical in their DNA sequences to the stx₂ gene of the EHEC type strain EDL933, a typical O157:H7 isolate from an HUS patient. A characteristic stx₁ genotype was present in all O26 isolates. The nucleotide sequences of the A and B subunits of the stx₁ gene of the tested bovine O26:H11 isolate WH-01/27/017-1 (GenBank accession no. EU700490) were nearly identical to those of the stx₁ genes of the EHEC O26:H11 reference type strains H19 and DEC10B, which had been associated with human disease outbreaks in Canada and Australia. Nucleotide exchanges typical for stx_1c and stx_1d subtypes as described by Kuczius et al. (38) were not found. All bovine O26:H11 strains produced an Stx1 with high cytotoxicity for Vero cells tested by Stx enzyme-linked immunosorbent assay and Vero cell neutralization assay (53). The Stx2 cytotoxicity for Vero cells was also very high in the O26:H11 isolates.

Not only factors influencing the basic and inducible Stx production are important in STEC pathogenesis. It has been suggested that the eae and EHEC-hlyA genes are likely contributors to STEC pathogenicity (2, 3, 13, 50). Ritchie et al. (50) found both genes in all analyzed HUS-associated STEC isolates. In all O26:H11 isolates we obtained, stx genes were present in combination with eae genes. Only the O26:H32 isolate lacked an eae gene. To date, 10 distinct variants of eae have been described (1, 19, 36, 45, 47). Some serotypes were closely associated with a particular intimin variant: the O157 serogroup was linked to γ-eae, the O26 serogroup to β-eae, and the O103 serogroup to ε-eae (4, 19, 20, 58). Our study confirms these associations. All bovine O26:H11 isolates were also typed as members of the β-eae subgroup. A translocated intimin receptor gene (tir gene) and the type III secreted proteins encoded by the espA and espB genes were found in all 56 O26:H11 isolates but not in the O26:H32 isolate. These other tested locus of enterocyte effacement-associated genes belonged to the β-subgroups. These results are in accord with the results of China et al. (19), who detected the pathotypes β-eae, β-tir, β-espA, and β-espB in all investigated human O26 strains. Like the eae gene, the EHEC-hlyA gene was found in association with severe clinical disease in humans (52). Aldick et al. (2) showed that EHEC hemolysin is toxic (cytolytic) to human microvascular endothelial cells and may thus contribute to the pathogenesis of HUS. In our study, the EHEC-hlyA gene was detected in 50 of the 56 bovine E. coli O26:H11 isolates which harbored virulence-associated plasmids of different sizes (Table 1). The presence of virulence-associated plasmids corresponded to the occurrence of additional virulence markers such as the espP and katP genes (17). The katP gene and the espP gene were detected in 49 and 50 of the 56 O26:H11 isolates, respectively. The espP gene was missing in six of the seven bovine O26:H11 isolates in which the katP genes were also absent. Both genes were not found in the O26:H32 isolate (Table 1). Although we found large plasmids of the same size in O26:H11 isolates, they lacked one or more of the plasmid-associated virulence factors (Table 1). Two DNA probes were used to detect the efa1 genes by colony hybridization. (DNA probes were labeled with digoxigenin [DIG] with lifA1-lifA2 and lifA3-lifA4 primers [14] using the PCR DIG probe synthesis kit [Roche Diagnostics, Mannheim, Germany]; DIG Easy Hyb solution [Roche] was used for prehybridization and hybridization.) Positive results with both DNA probes were obtained for 52 of 56 E. coli O26:H11 isolates. A positive signal was only found in three isolates with the lifA1-lifA2 DNA probe and in one isolate with the lifA3-lifA4 probe. An efa1 gene was not detected in the O26:H32 isolate (Table 1).

We also analyzed the spatial and temporal behavior of the O26:H11/H32 isolates in the beef herds by cluster analysis (conducted in PAUP* for Windows version 4.0, 2008 [http://paup.csit.fsu.edu/about.html]). This was performed with distance matrices using the neighbor-joining algorithm, an agglomerative cluster method which generates a phylogenetic tree. The distance matrices were calculated by pairwise comparisons of the fragmentation patterns produced by genomic typing through pulsed-field gel electrophoresis analysis with four restriction endonucleases (XbaI, NotI, BlnI, and SpeI) and the presence or absence of potential virulence markers (Fig. 1 and Table 1). To this end, the total character difference was used, which counts the pairwise differences between two given patterns. During a monitoring program of 3 years in four cattle farms (29), different O26:H11 cluster groups and one O26:H32 isolate were detected in three different farms. The genetic distance of the O26:H32 isolate was very high relative to the O26:H11 isolates. Therefore, the O26:H32 isolate was outgrouped. The O26:H11 isolates of each farm represented independent cluster groups. The single isolate from farm D fitted better to the isolates from farm C than to those from farm A. This finding is in accord with the geographical distance between the farms. The fact that the farms were located in neighboring villages may suggest that direct or indirect connections between the farms were possible (e.g., by person contacts or animal trade). However, the isolates from farm C and farm D belonged to different sequence types (ST 21 and ST 396), which may argue against a direct connection. Interestingly, O26:H11 isolates with and without stx₂ genes were detected in the same clusters. This phenomenon was observed in both farm A and farm C. In farm A, the isolates with additional stx₂ genes were found in animal 27 and were grouped in clusters 8 and 9 (day 218). An stx₂ gene was repeatedly found (four isolates) in the same animal (animal 27). The isolates grouped in cluster 8 on a later day of sampling (day 478). All other O26:H11 isolates grouped in the same clusters and obtained from the same animals (27 and 29) on different sampling days lacked an stx₂ gene. Also, the isolates obtained from animal 27 on previous sampling days, which grouped in clusters 3 and 5, exhibited no stx₂ genes. In farm C, the three isolates with additional stx₂ genes obtained from animal 7 grouped in clusters 11 and 12. An stx₂ gene was absent from all other O26:H11 isolates grouped in the same cluster 12 on later sampling days, and no other isolates of cluster 11 were found later on. However, we detected members of many clusters over relatively long periods (clusters 5, 8, and 9 in farm A and cluster 12 in farm C), but members of other clusters were only found on single occasions. This patchy temporal pattern is apparently not a unique property of O26:H11, as we found similar results for cluster groups of other EHEC serotypes of bovine origin (28). The isolates grouped in the dominant cluster 8 were found on 5 of 9 sampling days over a period of 10 months. In contrast, we found the members of clusters 4, 5, 9, and 12 only on two nonconsecutive sampling days. The period during which isolates of these groups were not detected was particularly long for cluster 4 (231 days). We also observed the coexistence of different clusters over long periods in the same farm and in the same cattle (clusters 8 and 9), while one of the clusters dominated. Transmission of clusters between cattle was also observed. These results suggest that some of the EHEC O26:H11 strains had the potential for a longer persistence in the host population, while others had not. The reasons for this difference are not yet clear. Perhaps the incomplete efa1 gene found in isolates of clusters which were only detected once might explain why some strains disappeared rapidly. Efa1 has been discussed as a potential E. coli colonization factor for the bovine intestine used by non-O157 STEC, including O26 (54, 56). The O165:H25 cluster detected during a longer period in farm B may have disappeared after it had lost its efa1 gene (28). The precise biological activity of Efa1 in EHEC O26 is not yet known, but it has been demonstrated that the molecule is a non-Stx virulence determinant which can increase the virulence of EHEC O26 in humans (8).

We distinguished 12 different clusters, but complete genetic identity was only found in two isolates. The variations in the O26:H11 clusters may be due to increasing competition between the bacterial populations of the various subtypes in the bovine intestine or to potential interactions between EHEC O26:H11 and the host.

The ephemeral occurrence of additional stx₂ genes in different clusters and farms may be the result of recombination events due to horizontal gene transfer (16). The loss of stx genes may occur rapidly in the course of an infection, but the reincorporation by induction of an stx-carrying bacteriophage into the O26:H11 strains is possible at any time (9, 40). Nevertheless, an additional stx₂ gene may increase the dangerousness of the respective EHEC O26:H11 strains. While all patients involved in an outbreak caused by an EHEC O26:H11 strain harboring the gene encoding Stx2 developed HUS (41), the persons affected by another outbreak caused by an EHEC O26:H11 strain that produced exclusively Stx1 had only uncomplicated diarrhea (60).

In conclusion, our results showed that bovine O26:H11 isolates can carry virulence factors of EHEC that are strongly associated with EHEC-related disease in humans, particularly with severe clinical manifestations such as hemorrhagic colitis and HUS. Therefore, strains of bovine origin may represent a considerable risk for human infection. Moreover, some clusters of EHEC O26:H11 persisted in cattle and farms over longer periods, which may increase the risk of transmission to other animals and humans even further.

• Open in new tab
• Download powerpoint

FIG. 1.

Neighbor-joining tree of bovine E. coli O26:H11/H32 strains based on the restriction pattern obtained after digestion with XbaI, NotI, BlnI, and SpeI.

• American Society for Microbiology