Characterization of Shiga toxin-producing Escherichia coli isolated from aquatic environments


  • Edited by C.W. Penn

*Corresponding author. Tel.: +34 934039044; fax: +34 934039047, E-mail address:


This study reports the phenotypic and genotypic characterization of 144 Shiga toxin-producing Escherichia coli (STEC) strains isolated from urban sewage and animal wastewaters using a Shiga toxin 2 gene variant (stx2)-specific DNA colony hybridization method. All the strains were classified as E. coli and belonged to 34 different serotypes, some of which had not been previously reported to carry the stx2 genes (O8:H31, O89:H19, O166:H21 and O181:H20). Five stx2 subtypes (stx2, stx2c, stx2d, stx2e and stx2g) were detected. The stx2, stx2c, stx2d and stx2e subtypes were present in urban sewage and stx2e was the only stx2 subtype found in pig wastewater samples. The stx2c and stx2g were more associated with cattle wastewater. One strain was positive for the intimin gene (eae) and five strains of serotypes were positive for the adhesin encoded by the saa gene. A total of 41 different seropathotypes were found. On the basis of occurrence of virulence genes, most non-O157 STEC strains are assumed to be low-virulence serotypes.


Shiga toxin-producing Escherichia coli (STEC) strains are an important cause of severe disease in humans, resulting in haemorrhagic colitis (HC) and haemolytic uraemic-syndrome (HUS) [1,2]. Domestic ruminants seem to be the principal reservoir of infectious STEC [3,4]. Transmission occurs through consumption of undercooked meat, unpasteurized dairy products and vegetables or water contaminated by faeces from carriers [2].

One of the most important pathogenicity factors produced by STEC strains is Shiga toxin (Stx). It contains two major groups called Stx1, similar to Stx of Shigella dysenteriae type 1, and Stx2[5]. Whilst Stx1 is highly conserved, Stx2 encompasses 11 distinct variants [6]. Adherence to the intestinal epithelium and colonization of the gut by STEC is another important component of the pathogenic process. This property is encoded on a pathogenicity island termed the locus for enterocyte effacement (LEE) [7]. Besides chromosomal markers, most STEC strains isolated from humans (both LEE positive and LEE negative) also carry large (>90 Kb) plasmids encoding proteins such as the enterohaemorrhagic E. coli enterohaemolysin (EHEC-HlyA) [8].

STEC strains belong to diverse serotypes, O26:H11, O103:H2, O111:H8 and O157:H7 being the most common, and present a wide range of phenotypic characteristics [9]. For instance, the ability to ferment sorbitol or β-d-glucuronidase activity, negative in O157 but positive in several non-O157 strains [9]. Usually, the information reported and the methods developed for isolation of STEC are based on the study of strains isolated from human or animal faeces using selective methods. It is assumed that the clinical strains are representative of what could be found in the natural environment of the same geographical area. This hypothesis should nevertheless be verified with environmental studies to confirm that no differential selection has occurred and that the strains isolated in the clinic correspond to those present in the environment. However, the isolation of STEC in the environment is difficult, due to the low proportion of pathogens compared with the generally high microbial concentration; consequently, confirmation of the presumptive causative agent involved in an outbreak is not always achieved [10].

A suitable method for detection and isolation of potential STEC strains from the environment has been developed [11]. This method permits the isolation of STEC from coliform bacteria grown on Chromocult? coliform agar (Merck, Darmstadt, Germany) using a specific probe against stx2A. Using this method, environmental strains could be detected and isolated without any selection associated with their serotype and phenotypic or molecular characteristics. In this study, we have characterized STEC strains isolated from urban sewage and animal faecal wastes from three different animal abattoirs (cattle, swine and a mixed animal slaughterhouse). Complete characterization was performed, including phenotypic, serological and molecular characterization, providing new information about STEC strains occurring in the environment.

2Materials and methods

2.1Strains and media

The E. coli O157:H7 strain ATCC 43889, which produces Stx2, and E. coli O157:H7 strain ATCC 43888, which does not produce either Stx1 or Stx2 and does not possess the genes for these toxins, were used as positive and negative controls, respectively, in the hybridization protocol. Bacteria were grown in Tryptic Soy Broth (TSB) and Tryptic Soy Agar (TSA) at 37 °C for 24 h. Chromocult? coliform agar (Merck, Darmstadt, Germany) was used for recovery of E. coli (EC) and total coliforms (TC) at 37 °C. The positive and negative control strains used in the PCR studies are shown in Table 1.

Table 1.  Characteristics of the control strains used
StrainSerotypeOriginVirulence genesReference
E. coli C600 (933W)Laboratory strainstx2[39]
E. coli DH5αGibco-BRL collection[40]
E. coli B2F1O91:H21HUSstx2c (stx2vha, stx2vhb)[41]
E. coli OX3:H21O174:H21Sudden infant deathstx2d[41]
E. coli FAC9ONTPorcine edema diseasestx2e[41]
E. coli ED431ONTPigeonsstx2f[36]
E. coli ATCC 43895O157:H7ATCCstx1,stx2, eae, ehxA[42]
E. coli E14b887O113:H21Laboratory isolatestx2, saa, ehxA[43]

2.2Isolation of stx2-carrying E. coli from environmental samples

Raw sewage samples of urban origin, mostly contaminated by human faecal wastes, and wastewater samples from three different abattoirs (cattle, pig, and a mixed cattle, lamb, goat and poultry slaughterhouse) were used for the isolation of stx2-strains by colony hybridization. Samples were collected aseptically and transferred into sterile containers according to standard procedures [12]. The samples were then placed in coolers, transported to the laboratory, and kept at 4 °C. Analysis was performed within 6 h of sampling. The characteristics of the samples and the sampling sites have been previously described [13]. The stx2 gene-carrying bacteria, E. coli organisms, faecal coliforms (FC) and total coliforms (TC) presented similar values through all the period of the study, showing only minor variations. TC, FC, and E. coli were present in quantities up to 106, 105, and 105 CFU ml−1, respectively. Tenfold dilutions of these samples were performed with Ringer 1/4 solution (Oxoid). Aliquots of 250 μl were spread on 140-mm diameter Chromocult? agar plates. Incubation was performed at 37 °C for 24 h. Those plates from dilutions presenting heavy but non-confluent colony growth were selected for colony hybridization [11].

2.3Preparation of digoxigenin-labeled stx2A-specific gene probes

A 378 bp DNA fragment of the stx2-A gene generated by amplification with primers UP 378 and LP 378 (Table 2) was labeled by incorporating digoxigenin-11-deoxy-uridine-triphosphate (Roche Diagnostics, Barcelona, Spain) during PCR, as described elsewhere [14], and used as probe.

Table 2.  Oligonucleotides used for PCR analyses in this study
PrimerNucleotide sequenceTarget sequenceReference
UP 378GCGTTTTGACCATCTTCGTstx2 (generic)[44]
27-FAGAGTTTGATCCTGGCTCAG16S rRNA of enterobacteria[49]

2.4Colony hybridization

The colony hybridization conditions used to isolate stx2-carrying E. coli have been described elsewhere [11]. Colonies showing positive hybridization were counted and isolated on LB from the original Chromocult? coliform agar plate. The presence of the stx2 gene was confirmed by PCR, as described below.

2.5PCR studies

PCRs were performed with a GeneAmp PCR system 2400 (Perkin–Elmer, PE Applied Biosystems, Barcelona, Spain). DNA template was prepared directly from two colonies of each strain suspended in 50 μl of double-distilled water and heated to 96 °C for 10 min prior to addition of the reaction mixture. Purified bacterial DNA was diluted 1:20 in double-distilled water. Primer oligonucleotides used in this study are described in Table 2. Five μl of each PCR product was analyzed by agarose (1.5%) gel electrophoresis and bands were visualized by ethidium bromide staining. When necessary, PCR products were purified using a PCR Purification Kit (Qiagen Inc., Valencia, USA). DNA fragments were purified from agarose gels using a Gel Extraction Kit (Qiagen Inc., Valencia, USA).

2.6Biotyping and phenotypic characterization of stx2 strains

Strains carrying the stx2 gene were phenotypically characterized using PhP-RE microplates of the PhenePlate? system, according to the manufacturer's instructions (PhP-Plate Microplates Technique AB, Stockholm, Sweden). Then, the obtained biochemical profiles were used for statistical analyses. Diversity index (Di) and Similarity population index (Sp) were calculated and clustering studies were performed by grouping isolates with correlation coefficient higher than 0.975 [15].

Isolates were also characterized with the commercial gallery API 20E. Additionally, the gene coding for the 16S rRNA of some strains that could not be identified by API 20E was amplified by PCR and sequenced as described below. Primers used for sequencing are described in Table 2.

The ability to ferment sorbitol within 24 h was confirmed by inoculating Purple Bromocresol Broth tubes containing 1%d-sorbitol (Sigma, St Louis MO, USA) with the strain to be tested and incubating at 37 °C for 24 h. Additionally, β-d-glucuronidase activity was assessed. The strain was inoculated in 250 μl of PBS and then a β-d-glucuronidase tablet (Diatabs, Rosko, Denmark) was added. The solution was incubated for 4 h or overnight at 37 °C, according to the manufacturer's instructions. E. coli DH5α and E. coli O157:H7 ATCC 43889 were used as positive and negative controls respectively, in both experiments.

2.7Lipopolysaccharide (LPS) profile

Extraction of LPS was performed using the proteinase K method [16]. Then, 10 μl of the extracted LPS was electrophoresed on a denaturing polyacrylamide gel [17] using a Mini-Protean II Dual Slab Cell (Bio-Rad, Richmond, CA, USA). Later, the gel was silver stained [18] to study LPS patterns.


Determination of O and H antigens was carried out using the method described by [19], employing all available O (O1–O181) and H (H1–H56) antisera. All antisera were adsorbed with the corresponding cross-reacting antigens to remove non-specific agglutinins. The O antisera were produced in the LREC (Lugo, Spain, and the H antisera were obtained from the Statens Serum Institute (Copenhagen, Denmark).

2.9Phage typing

Phage typing was performed using the method of [20] with phages provided by the National Laboratory for Enteric Pathogens, Laboratory for Disease Control, Ottawa, Ontario (Canada). The sixteen different phages used were able to identify 88 phage types.

2.10Toxin protein production

The production of Shiga toxins in the 65 representative isolates, both Stx1 and Stx2, was tested using the commercial Duopath VT? Detection Kit (Merck, Darmstadt, Germany), according to the manufacturer's instructions. The efficiency of this method has been demonstrated previously [21]. The bacterial controls used are described in Table 1.

2.11Isolation of bacterial DNA and detection of stx2

Chromosomal DNA was isolated from 40 ml cultures of each strain by lysozyme treatment and phenol–chloroform extraction, as described elsewhere [22], and suspended in a final volume of 200 μl with double-distilled water. One hundred μl of DNA was treated with RNAse (10 mg ml−1) and incubated at 37 °C for 1 h.

Five μl of purified DNA was digested with EcoRI and ClaI restriction endonuclease (Promega Inc., Madison, USA) at 37 °C for 3 h. Restriction fragments were analyzed by separation on 0.8% agarose gels in 1X Tris borate EDTA buffer and stained with ethidium bromide. After electrophoresis, DNA was transferred to nylon membranes (Hybond N+, Amersham Pharmacia Biotech, Spain) by capillary blotting [23]. The membranes were hybridized at 65 °C with a digoxigenin-labeled stx2-A probe, as described above.

To evaluate the stx2 variants harbored in the chromosome of the strains, stx2, stx2c, stx2d, stx2e and stx2f variants were analyzed by PCR. The primers used are listed in Table 2. Some strains showed positive PCR amplification for the generic stx2 primers, although they failed to amplify with the specific primers for the stx2 variants. In these cases, the toxin was sequenced to identify the toxin gene.

2.12Detection of other virulence factors

The genes encoding Stx1 (stx1), EHEC-haemolysin (ehxA), intimin (eae) and autoagglutinating adhesin (saa) were analyzed by PCR, as described above. The primers used are listed in Table 2.


Sequencing was performed in duplicate with the ABI PRISM Big Dye III Terminator cycle Sequencing Ready reaction Kit (Perkin–Elmer, Applied Biosystems, Spain) in an ABI PRISM 3700 DNA Analyzer (Perkin–Elmer, Applied Biosystems, Spain) according to the manufacturer's instructions. Nucleotide sequence analysis with homologous DNA sequences from the EMBL and Genbank databases was performed using the Wisconsin Package version 10.2 (Genetics Computer Group, Madison, Wisc). BLAST analyses were performed using available web-based tools: Multiple sequence alignment was performed using the Multalin software version 5.4.1 [24].


3.1Isolation of stx2-carrying E. coli

A total of 144 strains carrying the stx2 gene were isolated from the different samples (66 from urban sewage, 55 from cattle faecal waste, six from pig faecal waste and 17 from mixed animal faecal waste, which will be referred to as human, cattle, pig, and mixed isolates respectively, in the text). These strains were confirmed by PCR with the generic primers for stx2 and all of them presented the corresponding 378 bp amplification product. However, five strains lost the gene after subcultivation; they were nevertheless included in the study since stx2 was detected in the primary culture.

3.2Phenotypic characterization

The isolates were clustered according to the biochemical fingerprinting obtained with the PhenePlate? system and the LPS pattern. A total of 48 different phenotypes were observed by pairwise comparison (data not shown). No cluster was exclusively associated with a specific sample origin. A high diversity index was found for those isolates (0.96). A representative isolate was chosen among the isolates belonging to the same cluster and coming from the same sample. The representative isolate was the one that presented the highest mean similarity within the cluster and the lowest correlation coefficient with other cluster. Those isolates belonging to the same biochemical cluster but showing different LPS patterns were also included as representatives. The characterization was then performed on 65 strains selected as representatives.

Most of the isolates presented the typical pattern of E. coli with the API 20E gallery. However, 6 isolates could not be identified by the API 20E gallery because of atypical results in some of the biochemical tests. These atypical results included a positive result for urease and a negative result for indole. The gene coding for the 16S rRNA of the unidentified strains was sequenced, showing the highest homology (98%) with the sequence of E. coli 16S rRNA (Accession Nos. AF233451 and Z83205) after comparison with the sequences in GenBank.

3.3Serotypes and phage types

Thirty-four different serotypes were obtained (Table 3). The origin of the samples marked the distribution of serotypes. Only the serotype O171:H2 was detected in both human sewage and cattle wastewater samples. However, the O8 somatic antigen was present in all of the samples, but was combined with different flagellar antigens (H9 in human sewage, H31 in cattle and no flagellum detected in swine samples). The O2 somatic antigen was also detected in cattle and swine wastewater samples (in combination with the H25 and H21 flagellar antigens, respectively). Only one strain belonged to serotype O157:[H7]. This strain possessed O157 rfbE and fliCH7 genes and was identified as phage type 8. Some serotypes have been detected for the first time in this study (Table 3).

Table 3.  Distribution of STEC serotypes in the different water samples analyzed
Urban sewageCattle faecal wastePig faecal wasteMixed animal faecal waste
  1. aSerotypes previously found in human STEC.

  2. bSerotypes previously associated with human STEC that cause haemolytic uraemic syndrome (HUS).

  3. cNew serotypes not previously found in STEC.

  4. dH antigen is indicated in brackets because the strain was phenotypically non-motile (H) and the H type was detected by PCR. The strain possessed O157 rfbE and fliCH7 genes.

O166:H21c1/1O91:H21b1/1  ONT:H21a1/1
ONT:H9b1/1 O113:H21b1/1   

3.4Sorbitol fermentation and β-glucuronidase activity

The majority of the isolates showed a positive reaction with the sorbitol fermentation and β-glucuronidase tests. Only six isolates belonging to O2:H21, O157:[H7], O159:H O171:H2, ONT:H21 and ONT:H serotypes failed to ferment sorbitol, and 7 belonging to O22:H8, O89:H19, O157:[H7], O171:H2, O181:H49 and ONT:H serotypes were negative for the β-glucuronidase test. However, only two isolates were negative for both tests, one of which was the unique isolate of the O157:[H7] serotype. The other one belonged to the O171:H2 serotype.

3.5Toxin protein production

Only 7% (one of 14) of the human isolates exhibited toxin protein production compared with a much higher percentage of cattle isolates (75%; 27 of 36). In contrast, no signal for toxin protein production was detected in any of the six pig isolates. Toxin protein production was correlated with stx2 subtype in most cases. Thus, in the majority of stx2e subtypes there was no toxin production, whilst the other subtypes were more commonly associated with toxin production (Table 4). However, 75% of the human isolates that carried the stx1 gene showed Stx1 toxin protein production, while 100% of the cattle isolates did produce the protein.

Table 4.  Prevalence of virulence genes in STEC isolated from different sources
Origin Of the isolatesN of strainsstx1 geneStx1 protein productionstx2 genestx2c genestx2d genestx2e genestx2f genestx2g geneStx2 protein productionehxA geneeae genesaa gene
  1. aOne strain did not amplify with the primers used, but was classified as stx2e variant after sequencing.

Urban sewage14433185001100
Cattle faecal waste3655824000827915
Pig faecal waste6000006a000000
Mixed animal faecal waste9101016001100

3.6stx2 Variants

The genetic characterization of the virulence factors of the 65 isolates is shown in Table 4. Five of the six main stx2 subtypes were detected (stx2, stx2c, stx2d, stx2e and stx2g). The stx2f variant was not detected in any of the strains studied. Amplification of the stx2 gene with the generic stx2 primers was observed in 10 isolates, but amplification with the primers for the main subtypes failed. Sequencing of the A subunit of the stx2 gene of these isolates revealed 99.8% similarity with the recently described stx2g gene (GenBank Accession Nos. AY286000) in eight of the isolates, 99.9% similarity with the stx2c gene in one isolate (GenBank Accession Nos. M59432) and 98.8% similarity with the stx2e gene in another isolate (GenBank Accession Nos. M36727).

Samples from urban sewage contained a high percentage of isolates carrying the stx2d variant, whilst cattle and swine faecal waste predominantly contained the stx2c and stx2e variants, respectively. However, all four variants were detected in urban sewage samples. In cattle faecal waste, besides stx2c, the stx2 subtype was detected in eight out of 36 isolates. In swine faecal waste only the stx2e variant was detected. The stx2g gene was only detected in isolates from cattle faecal waste. The stx1 gene was detected in human and cattle samples, in 28% and 11% of the isolates, respectively.

Whilst 59 isolates showed positive PCR amplification for only one stx2 subtype, six isolates presented more than one stx2 subtype. This observation was later confirmed by Southern blotting of EcoRI/ClaI-digested chromosomal DNA using the stx2 probe. All possible combinations of stx2 genes observed in the strains are shown in Fig. 1. Most of the strains that showed a PCR amplification product for a single stx2 subtype displayed only one band in the hybridization with the stx2 generic probe, suggesting that there was only one copy of the gene. Some variants were detected showing only a single copy or not combined with any other variant (stx2g), as shown in Fig. 1A.

Figure 1.

Illustration of the combination of stx2 variants observed by southern hybridization of EcoRI-digested chromosomal DNA with a specific stx2A probe in some of the studied strains. (A) One toxin gene; (B) two toxin genes.

Some of the isolates displayed two bands. In these isolates, variants appeared combined or showed two copies of the same gene, e.g., stx2c (Fig. 1B). Six isolates that displayed two bands had previously displayed positive PCR amplification for two distinct variants. However, four isolates that carried only one stx2 subtype according to PCR amplification displayed two bands in the Southern blot. These isolates were considered to carry two copies of the same gene. The isolates presenting more than one copy of the stx2 gene corresponded to multiple serotypes (O26:H, O90:H, O91:H21, O127:H, O162:H7, O171:H2 [three strains], O181:H20, and ONT:H).

3.7Other virulence markers

The prevalence of the ehxA gene was very low in the non-O157 strains, being detected in only 10 out of 64 non-O157 isolates. The eae gene was only detected in the O157:[H7] strain, which was positive for γ intimin. The saa gene was only detected in five strains isolated from cattle wastewaters, but not coincident with the one carrying the eae gene, as expected.

3.8Pathotypes and seropathotypes

The combination of the different virulence factors is shown in Table 5. There was a high variability of pathotypes (18), independent of the origin of the sample. For example, the pathotype stx1stx2d was present in four different isolates (three human and one cattle) represented by the O70:H, O90:H (two strains) and O166:H21 serotypes. However, the more prevalent pathotypes were the strains that carried stx2c (which was mainly present in cattle isolates) or stx2e (which was present in pig and human isolates) genes alone, which were represented by the O98:HNT, O171:H2, O181:H20, ONT:H and ONT:HNT, and the O2:H21, O8:H9, O8:H, O26:H, O54:H21, O100:H, O159:H, O177:H, ONT:H9, ONT:H21 and ONT:H serotypes respectively. The O157:[H7] strain, which was isolated from cattle wastewater, was the only one that presented the stx1stx2cehxA eae-γ1 pathotype. In total, 41 different seropathotypes (associations between serotypes and virulence genes) were found.

Table 5.  Seropathotypes (serotypes and virulence genes) of STEC strains analyzed
  1. Number of strains belonging to the same serotype is shown in brackets. N, number of isolates with the same pathotype.

  2. aH antigen is indicated in brackets because the strain was phenotypically non-motile (H) and the H type was detected by PCR. The strain carried O157 rfbE and fliCH7 genes.

stx23O89:H19 (1)stx1stx2d4O70:H (1)
  O146:H (1)  O90:H (2)
  O171:H2 (1)  O166:H21 (1)
stx2c17O98:HNT (1)stx2ehxA1O113:H21 (1)
  O171:H2 (12)   
  O181:H20 (2)   
  ONT:H (1)   
  ONT:HNT (1)   
stx2d2O90:H (2)stx1stx2cehxA1O22:H8 (1)
stx2e15O2:H21 (1)stx1stx2dehxA1O146:H21 (1)
  O8:H9 (3)   
  O8:H (1)   
  O26:H (1)   
  O54:H21 (1)   
  O100:H (1)   
  O159:H (1)   
  O177:H (1)   
  ONT:H9 (1)   
  ONT:H21 (1)   
  ONT:H (3)   
stx2g8O2:H25 (6)stx1stx2eehxA1O176:H16 (1)
  O8:H31 (1)   
  O136:H1 (1)   
stx2stx2c2O162:H7 (1)stx2stx2cehxA1O91:H21 (1)
  O171:H2 (1)   
stx2stx2d1O127:H (1)stx2ehxA saa3O1:H20 (2)
     O181:H49 (1)
stx2stx2e1O26:H (1)stx1stx2cehxA saa2O22:H8 (1)
     O76:H2 (1)
stx2cstx2d1O171:H2 (1)stx1stx2cehxA eae-γ11O157:[H7]a (1)
     Phage type 8


Human sewage and animal wastewater have been shown to be potential reservoirs of stx2 gene-carrying bacteria [11,13,25]. In this study, we analyzed 144 strains carrying stx2 isolated from human sewage and faecal wastes from different animal sources. The biochemical phenotyping carried out with the PhenePlate? system revealed a relatively high diversity of isolates, with 48 different clusters. With the exception of 12, all of the strains presented the typical pattern for E. coli with the API 20E gallery. However, those that could not be identified with the API 20E gallery were classified as atypical E. coli after 16S rRNA sequencing. Although previous studies reported the isolation of Citrobacter freundii and Enterobacter cloacae carrying the stx2 gene [26], only E. coli was detected in this study. This may be due to the lower proportion of these strains in comparison with E. coli in sewage [11].

STEC strains that cause human infections belong to a large number of different O:H serotypes (a total of 472 serotypes are listed in the authors’ website, http://www.lugo.usc/ecoli). Most outbreaks of HC and HUS have been attributed to strains of the enterohaemorrhagic serotype O157:H7. However, as non-O157 STEC are more prevalent in animals and as contaminants in foodstuffs, humans are likely to be more exposed to these strains. STEC strains belonging to serotypes O26:H11, O103:H2, O111:H8, O145:H28 and O157:H7 are recognized as classical enterohaemorrhagic E. coli (EHEC) types, which occur in various countries worldwide [27–29]. The 65 STEC strains analyzed in the present study belonged to 25 different O serogroups, 12 H types, and 34 O:H serotypes. However, 42% of strains belonged to only four serotypes: O2:H25, O8:H9, O90:H and O171:H2. In total, four new O:H serotypes not previously reported in STEC strains were found in this study (O8:H31, O89:H19, O166:H21 and O181:H20). The majority (46 strains, 71%) of STEC strains isolated in the present work belonged to serotypes previously found among human STEC and 18% (12 strains) to serotypes associated with STEC isolated from patients with haemolytic uraemic syndrome (HUS) (O22:H8, O26:H, O91:H21, O113:H21, O157:H7, ONT:H9, ONT:H). However, only one representative isolate of four strains was identified as E. coli O157:[H7]. This strain was phage type 8, predominantly found in human and bovine STEC O157:H7 strains in Spain, as well as in many other European countries [30,31].

The presence of the stx2 gene in the infecting strain was previously reported to correlate with severe disease in humans [28,32] and the administration of purified Stx2, but not of Stx1, was shown to cause HUS in experimentally treated primates [33]. Six genetic variants of stx2 have been detected by nucleotide sequence analysis: stx2, stx2c, stx2d, stx2e, stx2f and stx2g. Some of these stx2 variants were previously found to be associated with STEC from sheep (stx2d) [34], pigs (stx2e) [35], pigeons (stx2f) [36], or cattle (stx2g) [37]. Some genetic variants (stx2d and stx2e) are not present in eae-positive strains belonging to classical EHEC serotypes, but are found in eae-negative STEC strains from patients with uncomplicated diarrhoea or asymptomatic infections. Other variants, such as stx2f and stx2g, have not so far been associated with STEC from humans. In contrast, stx2 and stx2c genes are associated with highly virulent EHEC serotypes that cause severe diseases such as haemorrhagic colitis and HUS. Two virulence profiles are significantly more frequently associated with HUS: (i) stx2 and (ii) stx2+stx2c[28]. In the present study, the most common stx2 variants (stx2, stx2c, stx2d and stx2e) and the newly described variant stx2g[37] were found among the different isolates, stx2c, stx2d and stx2e being more prevalent in cattle, human and pig isolates, respectively. Human samples exhibited a higher variability of stx2 subtypes. In cattle isolates, stx2c variant prevalence was followed by the stx2 subtype, which is in agreement with previous studies [6]. The stx2e variant was the only stx2 subtype identified in pig isolates. This variant is characteristic of porcine STEC strains, which cause oedema disease in pigs [35], although it was also found in some human samples. None of the stx2e strains from our study belonged to typical porcine enteropathogenic serotypes. Recently, a new stx2 variant designated stx2g was described [37] and detected in serotypes O2:H25, O2:H45 and ONT:H. In our study, eight stx2g strains of serotypes O2:H25 (six strains), O8:H31 and O136:H1 were identified in cattle wastewater.

A high percentage of cattle isolates showed Stx2 toxin production (75%) with the toxin detection test used in this study [21]. However, only 7% of the human representative isolates were positive for toxin production and the toxin protein was not detected in any of the pig isolates. A possible explanation for these observations could be that those representative isolates that are unable to produce the toxin, or at least produce the toxin at levels below the limit of detection (around one colony) should be regarded as a reservoir for the stx2 gene instead of being pathogenic bacteria. Therefore, the gene could theoretically be transmissible to new bacteria and, depending on several unknown factors (the insertion site, host, etc.), the toxin gene could be expressed. However, the majority of the strains carrying the stx1 gene showed Stx1 toxin protein production, independent of the origin of the sample. These data question whether these immunological methods would be suitable for the detection of slightly different Stx2 variants that might be present in the environment. Nevertheless, all of the control strains carrying the different Stx2 subtypes found in this study (c, d, e and g) presented positive results for the Duopath? VT detection kit, though showing different intensities.

The eae gene was only found in the E. coli O157:[H7] isolate. Several studies have reported an association between STEC infection and the presence of the eae gene [38]. The low prevalence observed in the isolates analyzed in this study could be explained by the fact that these isolates do not represent clinical isolates. The other gene associated with adhesion to the host cell, the saa gene, was found in only five strains. These genes were only present in strains isolated from cattle wastewater. However, the ehxA gene, which has also been linked to human disease [8] was present in isolates from both human and animal wastewater samples, but in a different proportion, 7% and 20%, respectively.

The observed combination of the different virulence factors shows a high diversity among the different STEC serotypes. The fact that in some cases the same serotype shows different combinations of genes suggests that these virulence factors effectively move through the bacterial population. They could represent a part of the existing pool of STEC present in the environment that are not necessarily human pathogens, but may play an important role in the emergence of new pathogenic STEC strains. In fact, the majority of these serotypes have not been linked to severe human disease, and show combinations of virulence genes not frequently associated with haemorrhagic colitis or HUS. Thus, according to the virulence genes and serotypes, the majority of the non-O157 STEC strains isolated in this study are assumed to be low-virulence variants. However, the surveillance of these genes in the bacterial population in the environment should be regarded as providing new information about STEC and as a useful tool for the prevention of epidemiological risks in the population.


This work was supported by grants from the Generalitat de Catalunya (2001SGR00099), the Centre de Referència en Biotecnologia (CeRBa), the Ministerio de Ciencia y Tecnología (BMC 2000-0549), the Fondo de Investigación Sanitaria (FIS G03-025-COLIRED-O157) and the Xunta de Galicia (PGIDIT02BTF26101PR). Maite Muniesa is a researcher of the ’Ramon y Cajal’ programme of the Ministerio de Ciencia y Tecnología.

We gratefully acknowledge Aurora Echeita (Centro Nacional de Microbiología, Madrid) for the phage typing and Yolande Bertin (Centre de Recherche INRA, Clermont-Ferrand, France), Alfredo Caprioli (Istituto Superiore di Sanità, Rome, Italy), Claire Jenkins (Laboratory of Enteric Pathogens, Central Public Health Laboratory, London, United Kingdom) and IPRAVE for the provision of control strains.