OUP user menu

Genetic differences between Escherichia coli O26 strains isolated in Brazil and in other countries

Jean C.C. Peixoto, Silvia Y. Bando, Juana A.G. Ordoñez, Beatriz A. Botelho, Luiz R. Trabulsi, Carlos A. Moreira-Filho
DOI: http://dx.doi.org/10.1111/j.1574-6968.2001.tb10571.x 239-244 First published online: 1 March 2001


Genomic diversity among 34 strains of Escherichia coli belonging to different serotypes of the O26 serogroup – encompassing strains from different geographical origins and Shiga toxin-negative Brazilian strains – was evaluated through random amplified polymorphic DNA (RAPD) analysis. Our results indicate that Brazilian and non-Brazilian O26 strains fall under distinct but closely related differentiation clusters. RFLP-PCR analysis of the fliC gene sequence was done in order to identify the H serotypes and served to confirm the clustering pattern obtained in the dendrogram generated from RAPD data. The epidemiological significance of these data is discussed.

  • Shiga toxin
  • Enteropathogenic Escherichia coli
  • Shiga toxin-producing Escherichia coli
  • Random amplified polymorphic DNA
  • Restriction fragment length polymorphism polymerase chain reaction
  • Escherichia coli O26

1 Introduction

Escherichia coli O26 strains were first described by Orskov [1] in 1951, who associated them with infantile diarrhea and white scours. Since then, the O26 serogroup has become one of the most important groups among the enteropathogenic Escherichia coli (EPEC) O serogroups [2]. Like other EPEC O serogroups, O26 includes several O:H types that differ in virulence properties [3]. The most common O26 strains belong to serotypes O26:H11 and O26:H32, or are non-motile (O26:H). The eae (EPEC attaching and effacing) gene is frequently present in O26:H11 and H strains, while O26:H32 strains do not possess this gene or any other enteropathogenicity marker [4].

Unlike other EPEC serotypes, some O26:H11 and O26:H strains produce Shiga toxin (Stx). These strains have been isolated in several countries from patients suffering from hemolytic uremic syndrome (HUS) [5,6]. Therefore, the O26 serogroup is rather heterogeneous since it includes the non-enteropathogenic serotype H32, the Shiga toxin-producing E. coli (STEC) strains (eae+ and Stx+), and the EPEC strains (eae+ and Stx). The virulence characteristics of O26 strains vary in accordance with their geographical distribution; the H and H11 strains isolated in Brazil have consistently proven to be Stx-negative. Although these serotypes are frequently isolated in children with diarrhea, no HUS cases have been reported in Brazil.

In this study we examined a collection of O26 strains isolated in Brazil, North America, and some European countries for the presence of eae and stx genes and for genomic diversity. The stx and eae markers were confirmed by polymerase chain reaction (PCR) for all strains. Genetic differences among strains were estimated through the analysis of polymorphisms obtained by the RAPD (random amplified polymorphic DNA) technique [13,14]. RFLP-PCR (restriction fragment length polymorphism polymerase chain reaction) analysis of the fliC gene sequence was performed for the H strains in order to get a molecular typing of these non-motile bacteria. The RFLP patterns were compared with the dendrogram clustering generated from RAPD data.

2 Materials and methods

2.1 Bacterial strains

Bacterial strains were obtained from culture collections kept by one of us (L.R.T.) at Instituto Butantan, São Paulo, Brazil. We studied 34 E. coli strains of serogroup O26. A Salmonella typhimurium strain (506/97) was used as an outgroup. All these strains are listed in Table 1.

View this table:
Table 1

Description of the strains analyzed in the present study

Strain no.Origin no.SerotypeOriginYear of isolationstxeae

2.2 DNA extraction

Genomic DNA was extracted from 30 ml cell cultures grown in Luria–Bertani broth (LB) medium and incubated aerobically at 37°C in a shaker for 18 h. The cells were harvested by centrifugation (20 min at 6800×g) and DNA was extracted using CTAB according to the procedures described by Bando et al. [7]. DNA yield and integrity were verified through electrophoresis in a 1% agarose gel and staining with ethidium bromide. Afterwards, DNA samples were analyzed in a spectrophotometer at 260/280 nm to determine their purity.

2.3 Amplification of stx1, stx2 and eae gene sequences by PCR

The presence of the stx1 and stx2 genes, responsible for Stx production, and of the eae gene, which encodes intimin, was verified by PCR analysis [810]. Table 2 shows the primer sequences, their reaction conditions, and respective amplification products. DNA samples for these analyses were obtained from cell lysates (the bacterial cells were boiled at 100°C for 10 min). Amplification reactions were performed as follows. One μl of bacterial lysate was added to a reaction mixture of 50 μl containing: 25 pmol of each primer (Table 2), 200 μM of dNTPs, 1.5 mM MgCl2 and 1 U Taq DNA polymerase. After amplification products were electrophoresed in 1% agarose gels, stained with ethidium bromide and visualized using UV light. DNA samples from E. coli strain O157:H7 and from the reference strain E2348/69 (O127:H6, EPEC) were used as positive controls for stx1/2 and eae respectively. Strain HB101 (E. coli K12) was used as a negative control.

View this table:
Table 2

PCR and RAPD primers, and their amplification conditions

Primer designationNucleotide sequence (5′-3′)TargetPCR conditionsProduct length (bp)
stx-BTCAACTGCTAATAGTTCT1 min1 min1 min
stx-2BCCGGAGCCTGATTCACAGG1 min1 min1 min
45 s45 s1 min 20 s
45 s45 s1 min 20 s
45 s45 s1 min 20 s
45 s45 s1 min 20 s
45 s45 s1 min 20 s
  • aThirty cycles.

  • bThese primers were used only for retyping strains 17, 18, 27, and 31–34.

  • cThirty-five cycles; initial denaturation step of 4 min at 95°C and final extension step of 7 min at 72°C.

  • dForty cycles; initial denaturation step of 4 min at 94°C and final extension step of 7 min at 72°C.

2.4 RAPD analysis

The PCR was performed in a final volume of 20 μl containing 20 mM Tris–HCl pH 8.4, 50 mM KCl, 2.5 mM MgCl2, 200 μM of each dNTP (dATP, dCTP, dGTP, and dTTP) (Gibco BRL), 0.3 μM of random primer (Operon Technologies, Alameda, CA, USA), 10 ng of genomic DNA, 1.5 U Taq DNA polymerase (Gibco BRL). All primers used in the RAPD experiments are listed in Table 2. Reactions were performed in a Lab-Line model 210 thermal cycler. The reaction products were stored at 4°C prior to analysis.

After PCR, 15-μl aliquots of the amplification products were electrophoresed in 1.4% agarose gels in 1×TBE (0.1 M Tris, 0.09 M boric acid and 1 mM EDTA, pH 8.3) buffer. The gels were stained with ethidium bromide and DNA bands were visualized and photographed using UV light. The 1-kb and 123-bp DNA ladders (Gibco BRL) were used as molecular size markers in all gels.

2.5 fliC RFLP-PCR analysis

The protocol described by Fields et al. [11] was followed with minor modifications. The amplification reactions were performed in a final volume of 30 μl containing 20 mM Tris–HCl pH 8.4, 50 mM KCl, 2 mM MgCl2, 50 μM of each dNTP (dATP, dCTP, dGTP, and dTTP) (Gibco BRL), 0.3 μM of each primer (F-FLIC1 and R-FLIC2, Table 2), 10 ng of genomic DNA, 2 U Taq DNA polymerase (Gibco BRL). The reactions were amplified in a PTC100 MJ thermal cycler.

Restriction analyses with RsaI enzyme (Gibco BRL) were performed in a final volume of 15 μl containing approximately 400 ng of PCR amplification product and 7 U of the endonuclease. The reactions were incubated at 37°C for 2 h. Digested DNA samples were electrophoresed in 1.5% agarose gels in 1×TBE buffer. The gels were stained with ethidium bromide and DNA bands were visualized and photographed using UV light.

2.6 Analysis of RAPD data

Statistical analyses of the data were performed using the NTSYS-pc [12] program (Numerical Taxonomy and Multivariate Analysis System), 1.7 version (Exeter Software, Setauked, NY, USA).

3 Results

3.1 Detection of stx1/2 and eae gene sequences by PCR

PCR analyses revealed that all the Brazilian O26:H11 and H strains in our sample are negative for the stx1 and stx2 gene sequences but positive for the eae gene. The stx1 gene sequence was detected in 11 out of 16 O26:H11 non-Brazilian strains while no stx2-positive strain was found. Fourteen of the 15 non-Brazilian O26 strains (13 H11 and one H) carry the eae gene sequence. PCR amplification results for stx1 and eae gene sequences appear in Table 1.

3.2 RAPD analysis

We tested 10 primers and chose five that gave the highest numbers of polymorphic markers for the 34 strains. These five primers generated 145 polymorphisms that were used to construct a binary data matrix of presence and absence of shared bands. The data from these comparisons were used to calculate the similarities between pairs of samples using the Jaccard coefficient. Afterwards, the dendrogram shown in Fig. 1A was generated. An additional primer, OPK-08, was used to check if an increase in the number of polymorphisms could modify the relative positions of the strains in the dendrogram. We obtained 23 new polymorphisms but as the dendrogram remained unchanged, these data were excluded from further analyses.

Figure 1

A: Dendrogram based on UPGMA cluster analysis of Jaccard coefficients. The S. typhimurium strain was used as an outgroup. Data on molecular typing (eae, stx), serotypes and geographical origin appear on the right. B: Dendrogram based on grouped data from: (i) O26:H32; (ii) O26:H11 Brazilian strains; (iii) O26:H11 non-Brazilian strains; (iv) O26:H Brazilian strains. S. typhimurium was used as an outgroup.

The dendrogram presented in Fig. 1A has two main groups: group A includes all non-pathogenic strains belonging to serotype O26:H32 and group B encompasses Brazilian Stx strains and the non-Brazilian O26 strains. Clusters B1 and B5 grouped all Brazilian Stx strains, whereas clusters B3 and B4 grouped all non-Brazilian strains. It is worth noting that all strains in the B3 cluster are stx-negative. Sample number 10, a O26:H Brazilian strain, occupies an isolated position between clusters B1 and B3. The clusters belonging to group B have a coefficient of similarity between 0.50 and 0.62. The S. typhimurium strain, used as an outgroup control, formed a distinct cluster (coefficient of similarity lower than 0.20).

A new dendrogram was constructed grouping the data relative to: (i) O26:H32 non-pathogenic; (ii) O26:H11 Brazilian strains; (iii) O26:H11 non-Brazilian strains; (iv) O26:H Brazilian strains. These results are presented in Fig. 1B and show that Brazilian and non-Brazilian strains belonging to the O26:H11 serotype form two distinct but closely related differentiation clusters (coefficient of similarity of 0.80).

3.3 Restriction analysis of the fliC gene by RFLP-PCR

The PCR amplification of the fliC gene produced a single band of approximately 1.4 kb in size for H11 and H serotypes, and of 1.6 kb in size for H32 (Table 1). Three different RFLP patterns (here called A, B, and C) were obtained for the strains under study. Fig. 2 displays these different restriction patterns. Pattern A was found in the H11 strains (strains 1, 3, 7, 15, 28) and in two of the H types (strains 11 and 30). The remainder of the H strains (strains 10, 12, and 13) showed a distinct pattern, designated B. Pattern C was found in H32 strains (strains 31 and 32).

Figure 2

Representative RFLP-PCR patterns for strains of E. coli O26:H11, O26:H32, and O26:H. Pattern A: lanes 2–6 (O26:H11, strains 1, 3, 15, 28, 7, respectively) and 10–11 (O26:H, strains 30 and 11, respectively). Pattern B: lanes 7–9 (O26:H, strains 10, 12, and 13, respectively). Pattern C: lanes 13–14 (O26:H32, strains 31 and 32, respectively). Non-digested fliC PCR products: lane 1, strain 1 (O26:H11); lane 12, strain 31 (O26:H32). M indicates the 1-kb ladder (Gibco BRL).

4 Discussion

The RAPD technique has been used commonly in the study of bacterial genomic diversity, in the characterization of strains, species and genera [1315], and in the investigation of clonal relationships in bacterial populations [1618]. In the present work, the analysis of RAPD polymorphisms revealed that O26:H11 bacterial strains are genetically similar (with a Jaccard coefficient index of 0.80) in spite of their different geographical origins (Fig. 1B). Nevertheless, the Stx Brazilian strains and the non-Brazilian strains studied here (O26 serogroup, H11, and H serotypes) fall in separate clusters in the dendrogram (Fig. 1A). The Brazilian strains fall in two major clusters, B1 and B5, well apart in the dendrogram, whereas the non-Brazilian strains are closely related (clusters B3 and B4). These results indicate that the Brazilian strains are different from strains isolated in other countries and, furthermore, that the O26 serogroup has more than a single clonal origin, as has been previously suggested in other studies [19,20].

The O26:H32 non-enteropathogenic strains forming group A in the dendrogram is well separated from the clusters encompassing the O26:H11 strains (Jaccard coefficient of 0.30), thus confirming that RAPD typing is concordant with the serological typing [13,17]. The same correspondence was observed regarding the RFLP-PCR analysis of the fliC gene sequence. Single specific patterns were observed for the O26:H32 strains and for the O26:H11 strains grouped in cluster B2. Interestingly, the O26:H strains numbered 10, 12, and 13, which formed distinct subclusters in the dendrogram (Fig. 1A), presented a RFLP fingerprint (Fig. 2) that differs from those obtained for strains O26:H11 and O26:H32. These strains probably derived from other serotypes.

PCR analyses screening for the presence or absence of stx1 and stx2 gene sequences showed that none of the Brazilian O26 strains carry those genes. These results confirm previous findings that indicate that Stx is not present as a virulence factor in Brazilian strains of the O26 serogroup [4,21,22]. On the other hand, among the non-Brazilian O26 strains there is a group of stx bacteria that form cluster B3 in the dendrogram. These strains are genetically distinct from those clustered in B4, where 11 out 13 strains are stx+. The stx gene is a bacteriophage-derived gene [23], indicating its chromosomal insertion depends on phage contact and probably on the genetic background of each bacterial strain. Although one can suppose that the Brazilian strains forming clusters B1 and B5, and the strains of cluster B3, never had contact with stx-carrying phages, it is more likely that the genetic background of all these strains does not favor phage insertion. If this is true, the receptive clone is not spread in Brazil so far. In the case of strains 20 and 29 of cluster B4, which are genetically very similar to many O26:H11 stx+ strains, it is reasonable to consider loss of a phage due to long storage periods [24].

The nomenclature of O26 strains, especially of those belonging to serotypes O26:H and O26:H11, needs to be redefined. In Brazil, the strains belonging to these two serotypes have been classified as EPEC due to the absence of Stx production and their association with sporadic cases of infantile diarrhea [25]. In Europe and North America, strains O26:H11 and H were classified as STEC on the assumption that these strains were always Stx-positive. However, it should be recognized that both serotypes comprise EPEC and STEC strains, a situation already found in other E. coli serotypes, such as in O128:H2, which also encompasses EPEC and STEC strains [26].

Finally, although the expression of some important phenotypic characteristics (e.g. virulence factors) possibly depends on specific minor genetic differences, likely also plasmid and phage acquisition [19,27], the overall genomic background plays a significant role in determining the evolution of pathogenic clonal subgroups [20].


This work was partially supported by FAPESP Grant 99/12695-4. J.C.C.P. is a fellow supported by CAPES. S.Y.B., J.A.G.O., and B.A.B. are fellows supported by FAPESP.


  1. [1].
  2. [2].
  3. [3].
  4. [4].
  5. [5].
  6. [6].
  7. [7].
  8. [8].
  9. [9].
  10. [10].
  11. [11].
  12. [12].
  13. [13].
  14. [14].
  15. [15].
  16. [16].
  17. [17].
  18. [18].
  19. [19].
  20. [20].
  21. [21].
  22. [22].
  23. [23].
  24. [24].
  25. [25].
  26. [26].
  27. [27].
View Abstract