INTRODUCTION
Shiga toxin-producing Escherichia coli (STEC) are recognized as emergent pathogens that have been implicated in numerous foodborne outbreaks and enteric infections around the world [Reference Nataro and Kaper1]. These microorganisms colonize the gut and can cause watery diarrhoea, haemorrhagic colitis (HC) and haemolytic–uraemic syndrome (HUS). The most prevalent serotype associated with severe infections and HUS is E. coli O157:H7 [Reference Griffin and Tauxe2]. Although the virulence factor best characterized in this serotype is the production of Shiga toxin (STX), adherence of strains to the gastrointestinal epithelium also plays a key role during infection [Reference Paton and Paton3]. The more virulent serotypes of STEC harbour a pathogenicity island termed locus of enterocyte effacement (LEE), which is associated with intimate adherence to epithelial cells, the initiation of host signal transduction pathways and with the formation of the typical attaching and effacing lesions (A/E) [Reference Jerse, Gicquelais and Kaper4, Reference McDaniel and Kaper5]. The protein intimin is encoded by the eae gene which is located in the LEE locus, and this protein is the only adherence factor so far proven to be associated with intestinal colonization in vivo [Reference Donnenberg6, Reference McKee7]. The eae gene is present in the most virulent strains, but the isolation of disease-associated strains lacking this gene suggests the existence of other adherence factors, a number of which have been described by several investigators: Iha (IrgA homologue adhesin), associated with adherence to HeLa cells in a non-fimbriated strain [Reference Tarr8]; Efa1 (EHEC factor for adherence), required for bacterial adherence to Chinese hamster ovary (CHO) cultured cells [Reference Nicholls, Grant and Robins-Browne9]; ToxB, required for total adherence in E. coli O157:H7 Sakai strain [Reference Tatsuno10]; Saa (STEC autoagglutinating adhesin), described in STEC O113 LEE negative strains [Reference Paton11]; Sfp (sorbitol-fermenting plasmid-encoded fimbriae), fimbriae present in sorbitol-fermenting STEC O157:H− [Reference Brunder12] and Lpf (long polar fimbriae), fimbriae of E. coli O157:H7 [Reference Torres13]. Other reports have implicated STEC proteins (Iha, Cah and OmpA), in addition to Efa1, Saa and Lpf, as mediators of adherence but their role in pathogenesis has yet to be defined [Reference Torres, Zhou and Kaper14]. We describe here the detection and distribution of these three putative adherence factors, among a collection of human and food strains of STEC.
MATERIALS AND METHODS
Bacterial strains
The STEC strains studied were isolated from 2000 to 2003 from human patients with different clinical syndromes [acute diarrhoea (57), HC (7) and HUS (26)] and from food samples [hamburger (20), bovine meat (10), chicken (1), sausage (2)]. The strains were collected by public health services from different regions of Chile, mainly from Santiago, and food strains were recovered from different meat products. Eighty-three human strains were from unique patients and seven were isolated from two patients in two separate outbreaks.
Stool samples were plated on McConkey agar and incubated at 37°C for 24 h. For food samples, 12·5 g was mixed with 125 ml trypticase soy broth, blended in a stomacher, and 100 μl were plated on McConkey agar and incubated at 37°C for 24 h. E. coli strains were identified by standard biochemical tests and serotyped by agglutination using commercial available antisera (O26, O55, O86, O111, O119, O114, O125, O126, O127, O128, O142, O157, O158) (PROBAC, Sao Paulo, Brazil); three strains of serogroups O91 and O174, were kindly serotyped at the Laboratory for Foodborne Zoonoses, Canada.
Detection of virulence genes
The presence of eae, stx 1 and stx 2 genes was determined by the multiplex PCR of Vidal et al. [Reference Vidal15]. Strains were grown on McConkey agar overnight and five lactose-positive colonies were suspended in 150 μl of 1% Triton X-100, boiled for 10 min, and centrifuged for 5 min at 8000 rpm; 3 μl of the supernatant was used as template for the PCR reaction. STEC strains EDL 933 (eae, lpfA O157 and efa1) and 472-1 (saa) were used as positive controls in the PCR. Amplification of eae, stx 1, stx 2, efa1, lpfA O157 and saa genes was performed with the oligonucleotides described in Table 1. The primers for efa1 and lpfA O157 were designed from sequences available in the GenBank database using OMIGA 2.0 software (Oxford Molecular Ltd, Madison, WI, USA) for alignment and the Primer 3 program (Whithead Institute for Biomedical Research, Cambridge, MA, USA) for primer design. Each PCR reaction (efa1, lpfA O157 and saa) was carried out independently and performed in 50 μl reaction mixture containing 1× reaction buffer (10 mm Tris–HCl, 50 mm KCl), 1·5 mm MgCl2, 1 mm dNTPs, 10 pmol each primer, 0·25 U Taq DNA polymerase (Biotools, Madrid, Spain) and 3 μl template DNA. Samples were amplified for 35 cycles, with each cycle consisting of 1·5 min at 94°C for denaturing, 1·5 min at 60°C for primer annealing and 1·5 min at 72°C for strand elongation. PCR products were visualized following electrophoresis in 1·5% agarose gels and staining with ethidium bromide; amplicons were identified by reference to molecular size markers.
DNA fingerprinting of strains
Strains were typed by indexing variation in the conserved Enterobacterial Repetitive Intergenic Consensus (ERIC) sequences of the genome. DNA was extracted from overnight growth of pure subcultures and cells were lysed by boiling for 10 min. Genotyping was performed using the ERIC1 fingerprinting assay [Reference Liu16]. PCR amplifications were performed in 25 μl volumes containing 2·5 mm MgCl2, 2 U Taq DNA polymerase (Promega, Madison, WI, USA), 1 mm dNTPs, 1·5 μl template DNA and 100 pmol ERIC1 primer. ERIC-PCR products were visualized following electrophoresis in 3% agarose gels and stained as above. Electrophoretic patterns were entered into Treecon for Windows v. 1.3b matrix analysis software (University of Konstanz, Germany).
Statistical test
Statistical analyses were performed with the SPSS 11.0.1. package for Windows (SPSS Inc., Chicago, IL, USA). The χ2 test was used for calculations and P values of <0·05 indicated a significant difference in the distribution of genes in different serogroups.
RESULTS
For human samples, 58 (64·4%) strains belonged to the O157 serogroup and 26 (28·8%) were of the O26 serogroup. These two serogroups accounted for 25 and 26 of the 57 cases of acute diarrhoea and all of the HC and HUS cases were caused by serogroup O157. The remainder of the human strains fell into serogroups O174, O91, O125 and two strains were serologically non-typable (ONT). In contrast, 19 of the 33 food strains were ONT and, with the exception of O125, none of the serogroups identified in food strains was found in human strains (Table 2).
HC, Haemorrhagic colitis; HUS, haemolytic–uraemic syndrome.
* ONT, Non-typable with available E. coli antisera.
Table 3 shows that all E. coli serogroup O157 strains harboured only the stx 2 gene. The majority (23/26) serogroup O26 isolates were positive for the stx 1 gene while both genes were present in the two strains of serogroup O174 tested and in two other serogroups (O125 and O158) and five ONT strains. Gene stx 1 was not detected in any of the other serotypable isolates but was present in three ONT strains. On the other hand the stx 2 gene occurred in varying numbers of five other serogroups and in 14 ONT strains.
H, human; F, food.
* ONT, Non-typable with available E. coli antisera.
The amplification of saa, lpfA O157 and efa1 genes produced PCR products of 129, 273 and 456 bp respectively (Fig. 1). The eae gene was detected in 84 (93·3%) human strains compared with 2/33 of food strains. The most prevalent putative adhesin genes in human samples were efa1 (90%), and lpfA O157 (74·4%), while the saa gene was present in only six (6·6%) of these strains; efa1 gene in 79 of the eae-positive strains and lpfA O157 in 65 strains. The most prevalent gene among food strains was saa (45·5%), while efa1 was present in only three strains (9%). The lpfA O157 gene was not detected.
By ERIC-PCR profiles the 122 STEC strains were grouped into two main clusters, which could be further subdivided into two minor clusters each (Fig. 2). Cluster Ia comprised the great majority of serogroup O157 strains while Ib mainly contained food strains of serogroup O113. A small number of food strains fell in cluster IIa but IIb consisted mainly of disease-associated serogroup O26 strains and 10 strains of other serogroups including six representatives of serogroup O157 from HC and HUS cases. These groupings did not correlate with the adhesin genes profile.
DISCUSSION
We determined by specific PCR assays the distribution of three putative adhesin genes in STEC strains isolated from human and food sources, belonging to different serogroups and isolated over a period of 4 years. We found that eae was the most prevalent gene in human STEC strains (93%) and there was a close association between the eae-positive strains and the presence of efa1 (P=0·0006) and lpfA O157 (P=0·005) genes, a situation similar to that described by Toma et al. [Reference Toma17]. The majority of human strains corresponded to O157 and O26 serogroups, which have not been previously associated with the presence of the saa gene [Reference Paton11]; we only detected this gene in four (6%) human isolates belonging to O125, O174 and O91 serogroups.
The eae gene was very rare in food strains while the saa gene was the most common marker identified in these strains confirming the negative correlation of the presence the two genes previously observed by Paton et al. [Reference Paton11]. Other studies on the frequency of the saa gene indicate that it is more frequently found in bovine STEC [Reference Jenkins18, Reference Osek, Weiner and Hartland19]. Overall, these results suggest that the Saa protein may have a more important role in attachment of STEC organisms to the bovine gut than the human intestine [Reference Jenkins18].
Our data indicate that Chilean STEC strains of human origin belong to different serogroups than those common in food with the exception of serogroup O125 which was found in both groups of samples. It is noteworthy that some strains isolated only from food, e.g. serogroups O2 and O113 have been associated with human disease previously [Reference Caprioli20]. Other serogroups in food such as O114 and O158 could represent normal intestinal microbiota of animals or are serogroups not yet described as human pathogens. The most prevalent serogroups of human strains were O157 and O26 which were entirely absent from foodstuffs. In conclusion, the distribution of the adherence genes of STEC was related to the source of isolation. In human strains the eae gene predominated along with the putative adherence factors efa1 and lpfA O157 and, therefore, these gene products may be potential candidates for vaccines directed to inhibit the colonization of the human intestine.
ACKNOWLEDGMENTS
We thank Dr Alfredo Torres for critical reading of the manuscript. This work was supported by Doctoral Fellowship, Conicyt AT-403131 and Grant DI INC 03/04, Universidad de Chile.
DECLARATION OF INTEREST
None.