|Home | About | Journals | Submit | Contact Us | Français|
Characterization of an Escherichia coli O157 strain collection (n = 42) derived from healthy Hungarian cattle revealed the existence of diverse pathotypes. Enteropathogenic E. coli (EPEC; eae positive) appeared to be the most frequent pathotype (n = 22 strains), 11 O157 strains were typical enterohemorrhagic E. coli (EHEC; stx and eae positive), and 9 O157 strains were atypical, with none of the key stx and eae virulence genes detected. EHEC and EPEC O157 strains all carried eae-gamma, tir-gamma, tccP, and paa. Other virulence genes located on the pO157 virulence plasmid and different O islands (O island 43 [OI-43] and OI-122), as well as espJ and espM, also characterized the EPEC and EHEC O157 strains with similar frequencies. However, none of these virulence genes were detected by PCR in atypical O157 strains. Interestingly, five of nine atypical O157 strains produced cytolethal distending toxin V (CDT-V) and carried genes encoding long polar fimbriae. Macro-restriction fragment enzyme analysis (pulsed-field gel electrophoresis) revealed that these E. coli O157 strains belong to four main clusters. Multilocus sequence typing analysis revealed that five housekeeping genes were identical in EHEC and EPEC O157 strains but were different in the atypical O157 strains. These results suggest that the Hungarian bovine E. coli O157 strains represent at least two main clones: EHEC/EPEC O157:H7/NM (nonmotile) and atypical CDT-V-producing O157 strains with H antigens different from H7. The CDT-V-producing O157 strains represent a novel genogroup. The pathogenic potential of these strains remains to be elucidated.
Escherichia coli O157:H7 is a food- and waterborne zoonotic pathogen with serious effects on public health. E. coli O157:H7 causes diseases in humans ranging from uncomplicated diarrhea to hemorrhagic colitis and hemolytic-uremic syndrome (HUS) (30). Typically, enterohemorrhagic E. coli (EHEC) strains express two groups of important virulence factors: one or more Shiga toxins (Stx; also called verotoxins), encoded by lambda-like bacteriophages, and a pathogenicity island called the locus of enterocyte effacement (LEE) encoding all the proteins necessary for attaching and effacing lesions of epithelial cells (41). Comparative genomic studies of E. coli O157:H7 strains revealed extensive genomic diversity related to the structures, positions, and genetic contents of bacteriophages and the variability of putative virulence genes encoding non-LEE effector proteins (29, 43).
Ruminants and, in particular, healthy cattle are the major reservoir of E. coli O157:H7, although the prevalence of O157:H7 strains in cattle may vary widely, as reviewed by Caprioli et al. (12). E. coli O157:H7 has been found to persist and remain infective in the environment for a long time, e.g., for at least 6 months in water trough sediments, which may be an important environmental niche.
In Hungary, infections with E. coli O157 and other Shiga toxin-producing E. coli (STEC) strains in humans in cases of “enteritidis infectiosa” have been notifiable since 1998 on a case report basis. Up to now, the disease has been sporadic, and fewer than 100 (n = 83) cases of STEC infection among 2,700 suspect cases have been reported since 2001. However, until the present study, no systematic, representative survey of possible animal sources had been performed.
In this study, our aim was to investigate healthy cattle in Hungary for the presence of strains of E. coli O157 and the genes encoding Shiga toxins (stx1 and stx2) and intimin (eae) and a wide range of putative virulence genes found in these strains. In addition, the phage type (PT) was determined, and pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST) were used to further compare the strains at the molecular level. Shiga toxin and cytolethal distending toxin (CDT) production was also examined, and phage induction experiments were conducted. The high incidence of enteropathogenic E. coli (EPEC; eae-positive) O157:H7 strains and atypical (eae- and stx-negative) O157 strains indicates that cattle are a major reservoir of not only EHEC O157 but also EPEC O157 and atypical E. coli O157 strains. These atypical, non-sorbitol-fermenting O157 strains frequently produced CDT-V and may represent a novel O157 clade as demonstrated by MLST and PFGE.
Altogether, 756 samples—colon specimens (n = 428), feces and rectal scrapings (n = 214), and milk samples (n = 114)—from 542 healthy cattle were collected. The 114 milk and 174 fecal samples were obtained from a large dairy farm (Enying), and all the other samples were taken from five slaughterhouses representing different regions of Hungary in the years 2002 and 2003. Samples were transferred to the laboratory on ice and investigated immediately after arrival or kept at −70°C until processing.
For the isolation of E. coli O157 and STEC, the following procedures were applied. First, polymyxin B extraction was used to detect Stx with the Seiken verotoxigenic E. coli screening kit. In the case of Shiga toxin detection, the samples were plated onto cefixime-tellurite-sorbitol-MacConkey agar and the presence of stx genes was investigated by stx universal and stx1- and stx2-specific PCR analyses of up to 50 colonies.
In addition to being utilized for the isolation of E. coli O157 strains, the samples were processed according to the International Organization for Standardization reference method (ISO 16654) using an O157-specific immunomagnetic separation (IMS) kit (Dynal, Oslo, Norway). For enrichment, novobiocin-containing modified tryptic soy broth was applied, and the suspensions were plated onto cefixime-tellurite-sorbitol-MacConkey agar, CHROMagar, and bromothymol blue agar plates. Five to 10 non-sorbitol-fermenting coliform colonies were agglutinated with O157 latex (Oxoid Ltd.) particles and with O26- and O111-specific immune sera. The agglutinating colonies were subcultured and confirmed biochemically to be E. coli. Only one E. coli O157 colony per sample was selected and analyzed further. One-third of samples were assayed by using O26 and O111 IMS kits. The isolated colonies were tested with O26 and O111 antibodies.
Serotyping of O (lipopolysaccharide) and H (flagellar) antigens was performed with O-specific and H-specific rabbit antisera prepared at the Federal Institute for Risk Assessment (Berlin, Germany) according to standard methods (44). Antigens O1 to O181 and H1 to H56 were investigated, and nonmotile (NM) E. coli O157 strains were evaluated by PCR and restriction fragment length polymorphism analyses of RsaI-digested fliC PCR products (16).
Phage typing was performed by the method of Khakria et al. (33) at the National Center for Epidemiology (Budapest, Hungary). By using the 16 typing phages all together, 90 PTs could be differentiated.
Colicin production was tested as described previously (1) using an E. coli K-12 strain sensitive to a wide range of colicins.
The resistance of E. coli strains to the following antimicrobials was investigated by the disk diffusion method using antibiotic disks (Oxoid Ltd.) on Mueller-Hinton agar: ampicillin, cefotaxime, chloramphenicol, ciprofloxacin, gentamicin, kanamycin, nalidixic acid, streptomycin, sulfonamide, sulfamethoxazole-trimethoprim, and tetracycline. The zones of growth inhibition were evaluated according to the 2005 recommendations of the Clinical and Laboratory Standards Institute.
Vero cell assays were used to detect Shiga toxin production by E. coli O157 strains as described previously (49). Briefly, doubling dilutions of E. coli O157 culture supernatants were examined for the presence of Stx on Vero cell monolayers (104 cells per well). The degree of cytotoxicity was determined after 2 days of incubation at 37°C in 5% CO2 by microscopic examination. The wells showing at least 50% cytotoxicity were considered positive.
CDT production was investigated with bacterial lysates as described previously (54). The morphological changes in HeLa cells characteristic of the presence of CDT were investigated after staining of the cells with Giemsa stain. CDT-V-producing E. coli strain 493/89 (27) was used as a positive control, and E. coli C600 was used as a negative control.
For phage induction, mitomycin C at a final concentration of 0.5 μg/ml and a subinhibitory concentration of norfloxacin (34) were used, and plaques were isolated as described previously using E. coli K-12 derivative C600 as the indicator strain (55). Single plaques were picked up and investigated by stx2- and stx1-specific PCR analyses.
E. coli O157 strains were first examined for O157:H7 rfbE (rfbEO157:H7) by diagnostic PCR analysis as described by Paton and Paton (47) and for the fliC gene as described by Fields et al. (16). The presence of stx genes was evaluated with Lin up/Lin down stx universal primers (36), and stx genes were typed using stx1-specific B54/B55 and stx2-specific B56/B57 PCR primers (13). B52/B53 primers were used for eae (13). EAF1/EAF2 PCR primers were used for eaf (17) and BFP1/BFP2 were used for bfp (22) according to the published protocols. The eae genes were typed as described by Oswald et al. (45), and tir genes were typed as described by Ogura et al. (42). Ehly1/Ehly5 primers were used for ehxA (57). The presence of efa1 5′ and efa1 3′ regions was investigated with primers Efa1-upper/Efa1-lower (R. La Ragione, unpublished data) and Efa1 3′fwd/Efa1 3′rev (39), respectively. The tccP-F1/tccP-R1 primers (21) were used for tccP genes, urea-F/urea-R primers (18) were used for ureA, and terB1/terB2 primers (52) were used for terB. Z4321-a/Z4321-b, Z4326-a/Z4326b, Z4332-a/Z4332-b, and Z4333-a/Z4333-b PCR primers were used for detecting O island 122 (OI-122)-specific marker genes pagC, sen, and efa1 as described by Karmali et al. (32). Detection and typing of cdt genes were performed with previously described PCR primers (54), as well as cdt-VA-specific (c338f/c2135r) and cdt-VC-specific (P105/c2767r) primers (8). Q antiterminator genes were typed with primers 595, Q 933, and Q 21 as described by Lejeune et al. (35). The presence of fuyA was detected with FyuA-F/FyuA-R primers (51), and lpfA-F/lpfA-R primers (15) were used for detecting lpfAO113. The sodC-specific O157-2For/O157-2Rev primers were used as described previously (14), and icf-specific primers were kindly provided by R. La Ragione (Veterinary Laboratories Agency, Weybridge, United Kingdom). Modified versions of PCR primers described previously were used for detecting the non-LEE effector genes espJ, espK, espM1, espM2, and espT (Table (Table11).
In this study, the sequences of the housekeeping genes adk, arcA, fumC, mdh, and mtlD were determined and compared as described previously (7). Briefly, fresh single colonies, which were grown overnight on Luria-Bertani agar plates, were suspended in 50 μl of a 0.9% NaCl solution by using sterile toothpicks and subsequently diluted 1:6 in 0.9% NaCl solution. PCR samples were prepared in a total volume of 50 μl containing 5 μl of this bacterial suspension, 5 μl of 10-fold-concentrated polymerase reaction buffer with 15 mM MgCl2 (Promega, Mannheim, Germany), 200 μM (each) deoxynucleoside triphosphates, 30 pmol of each primer, and 1.5 U of Taq DNA polymerase (Promega). PCR was performed in a GeneAmp PCR system 2700 (Applied Biosystems Applera, Weiterstadt, Germany). Samples of 8 μl of the PCR products were analyzed for purity on 0.7 to 1% Tris-borate-EDTA agarose gels using a maximum voltage of 5 V/cm. The primers used are listed in Table Table22.
For the removal of PCR primers, 5-μl samples of the PCR products were combined with 2 μl of a reaction mixture containing 10 U of exonuclease I (New England Biolabs) and 2 U of shrimp alkaline phosphatase (USB Biochemicals). Exonuclease I was diluted to 1 U/μl in 50 mM Tris-HCl, pH 7.5, and incubated at 37°C for 30 min and subsequently at 80°C for 15 min to inhibit further enzymatic activity.
The treated PCR products were sequenced with a CEQTM 8000 genetic analysis system (Beckmann Coulter, Krefeld, Germany). The applied DNA concentration was 50 fmol. Precipitation of DNA with ethanol and sequence analysis were conducted with the GenomeLab dye terminator cycle sequencing with quick start kit according to the instructions of the manufacturer (Beckmann Coulter). The program for amplification consisted of 30 cycles of 96°C for 20 s, 50°C for 20 s, and 60°C for 360 s. In general, the reading lengths were between 400 and 650 bp. The single fragments were assembled with BioEdit sequence alignment software (version 18.104.22.168.; Ibis Biosciences, Carlsbad, CA).
Following double-strand sequencing, the sequences were edited and aligned. DNA fragments between 410 and 725 bp long from each sequence were used for sequence analysis (Table (Table2).2). Raw DNA sequence data were analyzed with ABI 377 software. DNA sequences were edited and aligned with BioEdit, version 4.8.10 (http://www.mbio.ncsu.edu/BioEdit/bioedit.html) (23), converted into FASTA files, and loaded into S.T.A.R.T. (Sequence Type Analysis and Recombinational Tests; http://outbreak.ceid.ox.ac.uk/software.htm). The dendrograms were compiled with S.T.A.R.T. using the unweighted-pair group method with arithmetic mean (UPGMA). This software constructs a phylogenetic tree on the basis of allele numbers. Split decomposition analysis (5, 25) was performed with SplitsTree, version 2.0 (http://bibiserv.techfak.uni-bielefeld.de/splits/).
PFGE was performed by following the PulseNet protocol for EHEC with XbaI (Fermentas) as the restriction enzyme. The same plugs were also digested with 20 U/150 μl NotI (Fermentas) in the buffer provided by the manufacturer. Fingerprints were analyzed using the software Fingerprinting II (Bio-Rad). Similarity was assessed using the Dice coefficient (optimization, 1%; position tolerance, 1 to 1.5%), and clustering was performed using UPGMA. The threshold for relatedness was chosen at 90%.
The nucleotide sequences obtained by sequencing of the PCR products from all alleles of five E. coli housekeeping genes have been entered into the EMBL nucleotide database under continuous accession numbers from FN257307 to FN257337 for adk, FN257338 to FN257368 for arcA, FN257369 to FN257399 for fumC, FN257400 to FN257430 for mdh, and FN257431 to FN257461 for mtlD.
Using the verotoxigenic E. coli screening kit, the E. coli O157-specific IMS technique, and a latex agglutination kit, we were able to isolate a total of 42 E. coli O157 and 13 non-O157 STEC strains. In these isolation trials, 756 samples originating from 542 healthy cattle were processed. The detailed investigation was conducted with strains that proved to be E. coli O157.
E. coli O157 isolates were subjected to numerous PCR assays, but first, the presence of stx and eae genes was established and the strains were grouped. We defined an E. coli O157 strain as EHEC if it carried both eae and stx genes; EPEC was identified if the strain carried only the eae gene. Finally, in the absence of these key virulence genes, the E. coli O157 strains were designated atypical O157. Of the 42 E. coli O157 strains, 11 proved to be EHEC, 22 proved to be EPEC, and 9 strains were classified as atypical O157. Ten EHEC strains carried stx1 and stx2 genes, and in one strain, only stx2 was detected (Table (Table33).
The EPEC and EHEC O157 strains carried eae-gamma and tir-gamma genes and the tccP gene. The other investigated virulence genes also characterized the EPEC and EHEC O157 strains at similar frequencies (Fig. (Fig.1).1). Like the EHEC EDL933 and Sakai strains, none of our E. coli O157 strains carried the complete efa (EHEC factor for adherence) gene. Nine EHEC and 13 EPEC O157 strains carried the 5′ fragment of the efa gene. Interestingly, in one EPEC O157 strain the terminal fragment of the efa1 gene was detected, but this strain did not carry the initial fragment of efa1. The genes icf (synonymous with paa) and sodC were present in all EHEC strains and in 21 and 20 EPEC strains, respectively. Five EHEC and 11 EPEC O157 strains harbored the ureA gene. The tenB gene was detected in 8 EHEC and 19 EPEC strains. Two EHEC and two EPEC O157 strains carried fyuA. The lpfO113 gene occurred in one EHEC and six EPEC O157 strains. The OI-122-specific genes were detected in all the EHEC O157 strains, while sen was detected in 18, pagC was detected in 21, efa1 open reading frame (ORF) Z4332 was found in 21, and efa1 ORF Z4333 occurred in 21 EPEC O157 strains (Fig. (Fig.11).
In addition to tccP, the strains were tested for other non-LEE effector protein genes, including espJ, espK, espM1, espM2, and espT. The Sakai-specific espJ gene was present in 10 EHEC O157 strains (90.9%) and in 19 EPEC O157 strains (86.4%). The espK gene was found in four EHEC O157 strains (36.4%). The espM1 gene was present in 8 EHEC O157 strains (72.7%) and in 15 EPEC O157 strains (68.2%), and espM2 was found in 10 EHEC O157 strains (90.9%) and in 17 EPEC O157 strains (77.3%). The espT gene was absent from all O157 strains. None these effector genes were detected in the atypical O157 strains (Fig. (Fig.11).
In one atypical E. coli O157 strain, the ehxA gene was present, while three strains carried the ureA gene. However, in five atypical O157 strains, cdt-VABC genes were detected, and CDT production was demonstrated in tissue cultures (data not shown). Interestingly, these CDT-V-producing strains also carried the lpfAO113 gene, encoding the major fimbrial subunit of long polar fimbriae (Fig. (Fig.11).
Phage induction experiments were performed with 11 EHEC O157 strains. Lytic phages induced from eight strains and the purified plaques were lysed and investigated by PCR for the presence of stx genes. The results revealed that all eight functioning lytic phages carried stx2 genes (Table (Table4).4). Interestingly, in the three strains with noninducible stx2 phages, the Q21-like antiterminator gene sequences were identified, while in seven of eight inducible stx2 phages, the EDL99-like Q antiterminator gene was detected. Stx production by EHEC strains was proven in Vero tissue cultures (data not shown).
E. coli O157 strains all carried the rfbE gene (specific for O157 antigen), detected by PCR. Four strains positive for O157 antigen by PCR and latex agglutination proved to be O rough by O tube agglutination. Three E. coli O157 EHEC strains did not produce detectable H antigen. However, PCR-restriction fragment length polymorphism analysis revealed that the RsaI-digested fliC PCR products from these strains yielded the same patterns as those from H7-producing E. coli O157 strains (data not shown). Three atypical O157 strains produced H43, one strain produced H12, and one was NM. The O rough but PCR-positive O157 strains carried flagella associated with H12 (n = 2), H9 (n = 1), and H37 (n = 1) (Table (Table33).
None of the 42 E. coli O157 strains fermented sorbitol, produced colicin, or showed multidrug resistance. Only some strains were resistant to one or two antibiotics (data not shown).
The EHEC and EPEC O157 strains proved to belong to PTs typical for O157 EHEC, including PT8, PT21, and PT33. The atypical O157 strains could not be classified by the typing phages used: all these strains either proved to be phage resistant or belonged to noncharacteristic PTs. Results of serotyping and phage typing are shown in Table Table33.
The virulence gene profiles and the subtyping of eae and tir genes clearly indicate a strong relationship between the EHEC and EPEC O157 strains and substantial diversity between these strains and the E. coli O157 atypical strains. To explore further the phylogenetic relationship among O157 strains, MLST and PFGE were conducted.
Altogether, 31 E. coli O157 strains, including 11 EHEC, 11 EPEC, and 9 atypical O157 strains, were compared by investigating the sequences of five housekeeping genes. The genes chosen for MLST were distributed evenly throughout the chromosome. DNA sequences of corresponding housekeeping genes from the E. coli K-12 strain MG1655 (accession no. NC_000913) and the O157:H7 strains EDL933 (accession no. NC_002655) and Sakai (accession no. NC_002695) were compiled from the respective published genome sequences.
DNA sequences were obtained from all 31 E. coli O157 bovine strains studied, and the sequences of all five housekeeping genes could be aligned without gaps. The corresponding DNA sequences from E. coli K-12 strains (MG1655 and C600) and O157:H7 strains (EDL933 and Sakai) were included in our scheme. The alleles of the five genes analyzed were numbered in ascending order. Numbers and types of base substitutions were not taken into account for allele designation (7). The allele frequencies and combinations of allele numbers for all isolates are shown in Table Table5.5. Each unique combination of allele numbers represents one sequence type (profile). As a result, seven MLST profiles (I to VII) were established; six MLST profiles (I to VI) were attributed to the E. coli O157 strains, while MLST profile VII characterized the K-12 strains. The 11 EPEC and 11 EHEC O157 strains showed MLST profile I, and the 9 atypical strains showed five different MLST profiles: II to VI (Table (Table6).6). One of the five dendrograms is shown in Fig. Fig.22 as an example.
PFGE was performed on 42 bovine O157 isolates, 1 human (C81) O157:NM strain (53), and 2 EHEC reference strains: O157:H7 EDL933 (48) and Sakai (24). Macro-restriction fragment analysis using the enzyme XbaI yielded 45 patterns (Fig. (Fig.3).3). Two human EHEC type strains (Sakai and EDL933) and two epidemiologically unrelated Hungarian EHEC strains (a bovine isolate, 4979, and a human isolate, C81) were also tested for comparison.
PFGE typing was very useful for the identification of genetically closely related stx-positive and stx-negative MLST profile I strains. Four main clusters (A, B, C, and D) were outlined by enzyme XbaI (Table (Table5).5). XbaI cluster A included 17 EPEC strains from different slaughterhouses in Hungary; cluster B comprised 7 EHEC isolates, interestingly including the single human Hungarian O157:NM EHEC isolate; cluster C comprised four EHEC strains isolated from cattle on the same farm; and cluster D consisted of four atypical CDT-V-producing strains, all derived from milk samples from that dairy farm (Fig. (Fig.3.3. and Table Table55).
The 22 MLST profile I E. coli O157 bovine strains were subdivided by PFGE into three XbaI clusters, A to C. Interestingly, EDL933 and Sakai strains gave very similar XbaI patterns and belonged to none of the four clusters.
In the present study, 42 E. coli O157 strains isolated from dairy and slaughterhouse cattle in Hungary were characterized genetically and phenotypically. This process entailed determining the PT, serotype, and antibiotic resistance profile; detecting the presence of key virulence genes and a wide range of putative virulence genes; and identifying genomic DNA macro-restriction fragment patterns and MLST profiles. Stx production by EHEC O157:H7 strains and CDT production by atypical O157 strains were also examined, and phage induction experiments were conducted.
Our results show that E. coli O157 strains are relatively common in the Hungarian cattle population: indeed, E. coli O157 strains were isolated from more than 7% of the cattle investigated.
In our non-sorbitol-fermenting EPEC and EHEC O157 strains, as in the sequenced prototype EHEC O157:H7 strains EDL933 and Sakai, only initial or terminal fragments of the efa1 gene (also called lifA, for lymphocyte inhibitory factor) were detected (24, 48). These results are in accord with the data reported in the literature since the whole efa1 gene has never been found in non-sorbitol-fermenting E. coli strains. However, the large, 9,669-bp gene was detected previously in sorbitol-fermenting EHEC O157:NM strains (28). The efa1 gene in these strains showed 99.9% sequence homology to efa1 from EHEC O111:NM and lifA from EPEC strain E2348/69 of the O127:H6 serotype (28).
In harmony with the findings described in the literature, non-LEE effector genes tccP and espJ (21), as well as espM (3), were detected frequently in EHEC and EPEC O157 strains, while espT was absent from all O157 strains (3). In contrast, none of these effector genes were present in atypical O157 strains.
The urease gene cluster is located within OI-43 and OI-48 (48) in EDL933. Urease genes were detected in the major EHEC groups O26, O111, and O157, as well as in sorbitol-fermenting EHEC O157:NM strains, but were absent from diarrheagenic E. coli strains of several other pathogroups, including enteroaggregative E. coli, enteroinvasive E. coli, and EPEC strains (18, 40). In enterotoxigenic E. coli, there is a homologue of the O157 urease-encoding OI-48 (46). These investigations indicate the virulence potential of urease in EHEC. Not only do the tellurite resistance (52) and the periplasmic or membrane-associated superoxide dismutase (6) characterize the prototype EHEC strains, but these virulence factors also were identified in EHEC clinical isolates. The fact that the marker genes for urease (ureA), potassium tellurite resistance (terB), and superoxide dismutase (sodC) frequently characterized our EPEC and EHEC O157 bovine strains is suggestive of their pathogenic and zoonotic potential.
MLST demonstrates that our non-sorbitol-fermenting O157 EHEC and EPEC strains represent a single clone complex that may belong to the non-sorbitol-fermenting O157:H7 worldwide clone and that the atypical E. coli O157 strains are independent. Our genotyping results indicate that stx genes may be part of the gain and loss of Stx-converting bacteriophages among E. coli O157 strains isolated from healthy bovines. Evidence in support of either possibility exists. First, Schmidt et al. (50) lysogenized laboratory strains as well as a broad range of enteric E. coli strains, including two EPEC strains, with a derivative of a Shiga toxin 2-encoding phage originating from an E. coli O157 strain. James et al. (26) also lysogenized several wild-type E. coli and Shigella strains in vitro by using an aph3 gene-labeled Stx2-encoding phage. Acheson et al. (2) were able to transduce a laboratory strain in the murine gastrointestinal tract with a derivative of phage H-19B encoding Stx1, while Tóth et al. (55) were able to lysogenize a porcine EPEC O45 strain with a derivative of an Stx2-encoding phage in porcine ligated ileal loops. The results of these transduction experiments are consistent with the idea that EHEC emerged from EPEC by the acquisition of stx genes. Recently, Mellmann et al. (38) reported the loss of Stx phages from some EHEC strains in HUS patients within a short interval. Comparative molecular analyses of the E. coli strains isolated from original and sequential stool samples from 210 HUS patients revealed that of the 137 stx- and eae-positive E. coli strains originally isolated, 6 strains lost their Stx phages. Five of these six strains were serotype O26:H11, and one was O157:NM. Bielaszewska et al. (10) also verified the gain and loss of the stx gene in E. coli O26 in vitro. We observed a similar spontaneous loss of the stx2 gene during the storage of one of our EHEC O157 strains (data not shown).
Although our atypical E. coli O157 strains did not carry stx and/or eae genes, five atypical O157 strains carried cdt-VABC genes and produced CDT. It was reported previously that the cdt-V allele is present in the majority of sorbitol-fermenting E. coli O157:NM strains (27) and in 5% of non-O157 EHEC clinical isolates (8). CDT-V occurs frequently in EHEC serotypes O113:H21 and O91:H21, the rare eae-negative EHEC types that cause HUS (31). These data suggest that CDT may contribute to the pathogenicity of eae-negative EHEC. Previously, Friedrich et al. (19) identified cdt-V in 4.9% of “classical” non-sorbitol-fermenting EHEC O157:H7 strains. The cdt-V EHEC O157:H7 strains belonged to five different PTs, including PT2, PT4, PT8, PT14, and PT34. Here, we report CDT-V production by non-sorbitol-fermenting O157 strains. The fact that these atypical O157 strains lack not only eae but also all the known key virulence genes underlines the virulence potential of CDT-V in E. coli O157:H37/H9. Furthermore, the CDT-V-producing atypical E. coli O157 strains harbored the major fimbrial subunit gene lpfA as well. The presence of polar fimbria genes in the CDT-V-producing strains may also be an important virulence characteristic since lpfO113 was reported to function as an adhesin in LEE-negative isolates of O113:H21 EHEC (15). These CDT-producing strains may represent a potentially novel clone, as demonstrated by PFGE and MLST.
E. coli O157 strains that lack Shiga toxin genes (stx) and the EPEC adherence factor plasmid are classified as atypical EPEC and cause diarrhea worldwide, especially in children. In our case, the bovine EPEC O157 strains isolated from healthy cattle resembled EHEC O157 strains on the basis of their serotypes and virulence gene profiles, and their multilocus sequence types were identical to those of the investigated stx- and eae-positive O157 (EHEC) strains, suggesting the zoonotic potential of these bovine EPEC O157 strains. Thus, our results confirm the recently published results of Bielaszewska et al. (9), who gave epidemiological and clinical evidence for the pathogenic significance of stx-negative atypical EPEC O157 strains in human patients with bloody diarrhea. Furthermore, they also stated that these stx-negative, EPEC adherence factor-negative attaching and effacing O157 strains are most likely former EHEC strains that lost Shiga toxin genes during infection, and they termed them EHEC-LST. Similarly, Friedrich et al. (20) were able to isolate strains of stx-negative E. coli O157 from stool samples obtained from patients with uncomplicated diarrhea and from HUS patients.
The bovine E. coli O157 strains described in our study represent different pathotypes, including EHEC, EPEC O157:H7, and atypical O157, and most of them possess virulence genes with zoonotic potential. These results indicate that in addition to the typical EHEC strains, the bovine O157 EPEC and CDT-V-producing atypical O157 strains deserve attention. Based on these observations, we conclude that calves and cattle represent important reservoirs of eae-positive E. coli O157 and that CDT-producing lpfO113-positive O157 strains may represent a novel pathogroup of E. coli O157 with virulence potential for animals and humans.
This study was supported by grants from NKTH (4/040/2001), EU NoEs EPG, and MedVetNET. Short-term mission support by MedVetNet for I.T. is also acknowledged.
We thank Márta Puruczki (Budapest) and Marcus Kranz (Stuttgart) for skillful technical assistance and Roberto La Ragione for the PCR primers. We also thank Noémi Nógrády for testing antimicrobial resistance and Vic Norris for critical reading of the manuscript.
Published ahead of print on 14 August 2009.