RESULTS AND DISCUSSION

E. coli O157:H7 was detected in 7·6% of the food samples analysed overall with frequencies of 8·4% (18/215) from meat products and 6·9% (15/215) from dairy products. This finding contrasts with the lower rate (3·1% in meat and 3·6% in dairy products) found in a recent large-scale survey of E. coli O157:H7 from food samples (800 meat and 800 dairy products) in different cities and villages in Egypt [Reference Ahmed and Shimamoto21]. This difference between the surveys might be attributed, in part, to differences in study methodology such as sampling strategy, type of samples, enrichment procedures, immunomagnetic separation and culture media [Reference Islam22]. In meat products, E. coli O157:H7 was detected at a higher rate in minced beef samples (12%), followed by hamburgers (7·1%), and fresh beef meat (5·7%), whereas in dairy samples, detection rates were highest in raw milk cheese samples (7·8%), followed by raw milk samples (6%).

Earlier studies have reported the importance of the molecular assays used in this study in identifying E. coli O157:H7 genotypes with variation in pathogenic potential and host specificity [Reference Ahmad and Zurek3–Reference Yang8]. Furthermore, such assays were proposed as valuable indicators for countries with high rates of outbreak investigations with this organism [Reference Mellor23]. The LSPA-6 assay detects differences in six loci representing conserved regions of the O157 backbone that differentiate E. coli O157:H7 strains into three genotypes referred to as ‘lineages’ (LI, LI/II, LII) [Reference Ziebell11, Reference Liu, Knabel and Dudley24]. Table 1 shows that of the 33 E. coli O157:H7 strains investigated here 25 (76%) were of lineage I/II, seven (21%) strains were lineage II and only one strain was lineage I. Previous studies indicate that the predominant lineage I/II shares characteristics of both lineage I and lineage II, but in contrast to lineage II, it is associated with human illness at frequencies similar to those of lineage I strain, and includes hypervirulent strains recovered from a multistate ‘spinach’ outbreak in Canada [Reference Zhang10, Reference Ziebell11]. Moreover, lineage I/II strains have been correlated with severe human disease outcomes such as HUS [Reference Abu-Ali25]. The current study provides further evidence of the epidemiological relevance of lineage I/II as a possible emerging risk genotype in many countries and possibly Egypt. On the other hand, a different LSPA-6 distribution profile (lineage II: 48·3%, lineage I/II: 37·9%, lineage I: 13·8%) was reported in E. coli O157:H7 from food sources in The Netherlands and such frequencies exhibited an intermediate distribution of lineage I/II between bovine and human clinical strains, suggesting a possible role of food in the selection of potentially virulent genotypes in humans [Reference Franz26]. This divergence in LSPA-6 distribution may be attributed to geographical differences and disease incidence in different countries, as well as the environmental load originating from super-shedding of E. coli O157:H7 by cattle [Reference Franz26, Reference Chase-Topping27].

Table 1.

Distribution of different genotypes in E. coli O157:H7 isolates recovered from food sources in Egypt

LSPA-6, Lineage-specific polymorphism; SBI, Shiga-toxin-encoding bacteriophage insertion site.

Differentiation of SBI genotypes relies on amplification of the toxin genes Stx ₁ and Stx ₂ and the insertion site junctions of their encoding bacteriophages. This assay has been used to categorize E. coli O157:H7 isolates based on their propensity to cause disease in humans and some SBI genotypes (1, 2, 3) were found to be clinically biased genotypes [Reference Besser4, Reference Shaikh and Tarr17]. In the strains studied here five different SBI genotypes were identified (1, 3, 5, 6, 21) (Table 1) and genotype 1 (60·6%) was significantly more prevalent (P < 0·05) which may provide initial evidence of its epidemiological significance as an indicator of a potential infection risk from food sources in Egypt. This finding is consistent with former studies that described SBI genotype 1 as a potential risk genotype in the USA [Reference Besser4] and The Netherlands [Reference Franz26]. SBI 5 and 6 accounted for fewer strains (18·2% and 15·2%, respectively) while SBI 3 and 21 were infrequent (3%) suggesting the possible limited relevance of these genotypes in Egypt.

Another subtyping scheme based on 32 single nucleotide polymorphisms was recently developed that could distinguish E. coli O157:H7 strains into nine distinct evolutionary clades, of which clade 8 was more frequently associated with severe disease outcomes such as HUS than other clades [Reference Manning6]. The majority (69·7%) of the strains here were typed as clade 8 (Table 1).

The gene tir encodes the translocated intimin receptor that mediates adhesion of E. coli to mammalian cells and the formation of attaching and effacing (A/E) lesions through binding to intimin [Reference Kaper, Nataro and Mobley28]. The Tir (A255 T) polymorphism assay identifies a base change (A or T allele) at position 255 in the sequence of the gene and the variant Tir (255 T) has been significantly more associated with human disease than Tir (255A) [Reference Bono5]. Furthermore, most (62·1%) of E. coli O157:H7 isolates recovered from food sources in The Netherlands demonstrated the Tir (255 T) allele [Reference Franz26]. Congruent to this finding, our results revealed that Tir (255 T) was significantly more frequent (P < 0·05) in the tested E. coli O157:H7 isolates with 24 isolates (72·7%) exhibiting the T allele (Table 1). This observation may suggest a possible link of Tir (255 T) with increased risk of human infection with E. coli O157:H7 strains from food sources in Egypt.

The final characterization of the E. coli O157:H7 isolates included variants of the Stx ₂ -Q anti-terminator junction alleles (Q ₉₃₃ and Q ₂₁) [Reference LeJeune29] as well as Stx ₂ gene variants [Reference Beutin30–Reference Friedrich32]. The Q ₉₃₃ variant which is located upstream of the Stx ₂ gene and results in relatively high expression of the latter, was initially reported as a possible human infection risk indicator [Reference Lowe33]. We found an almost equal distribution of Q ₉₃₃ and Q ₂₁ gene variants (45·5% and 48·5%, respectively) in Egyptian strains and two strains carried both gene variants (Table 1). Analysis of Stx ₂ variants (Stx _2a and Stx _2c) based on phage chromosomal integration site, gene content and lineage placement [Reference Eppinger34] revealed that 11 (33·3%) strains carried Stx _2a, 15 (45·5%) strains carried Stx _2c and the remaining seven strains carried both variants (Stx _2a+c) (Table 1). Stx _2c variants have been shown to elicit milder symptoms and be less clinically significant, as well exhibiting less virulence in human kidney cell lines and mouse infection models than Stx _2a variants [Reference Persson7, Reference Beutin30, Reference Andersson35, Reference Fuller36].

Further analysis of correlations between the various genotyping methods showed that all clade 8 isolates and the majority (95·8%) of Tir (255 T)-carrying strains were represented by lineage I/II. By contrast, 70% of non-clade 8 and Tir (255A) (77·8%) carrying strains were grouped in lineage II. However, Tir (255 T) and Tir (255A) variants were more frequent in clade 8 (87·5%) and non-clade 8 (66·7%) strains, respectively. Last, most Q ₉₃₃ variants (93·3%) were associated with SBI 1, but Q ₂₁ variants grouped in different SBI genotypes [SBI 5 (37·5%), SBI 6 (31·25%), SBI 4 (25%), SBI 1 (6·25%)]. Multivariate analysis of different genotypes revealed evidence of a significant association between LSPA-6 and clade types as well as Tir (A255 T) genotypes at a confidence interval of 95% and a significance level of P < 0·001 (Table 2). This suggested that clade types and Tir (A255 T) polymorphism might be used not only as useful indicators for microbial risk potential by E. coli O157:H7 from food sources but also as possible surrogate markers of predictive of LSPA-6 genotypes. This strong association between clade 8 and lineage I/II is consistent with similar results from previous studies [Reference Liu, Knabel and Dudley24, Reference Hartzell37, Reference Laing38]. Consequently, such findings suggest that within lineage I/II, clade 8 strains exhibit higher virulence and develop more severe clinical diseases compared to other clades [Reference Manning6]. Conversely, no specific association between clade 8 and LSPA-6 genotypes was observed in E. coli O157:H7 strains recovered from bovine, food, and human sources in The Netherlands [Reference Franz26]. Nevertheless, a strong correlation between LSPA-6 lineage and Tir (A255 T) was previously reported in E. coli O157:H7 strains recovered from super-shedding cattle in USA [Reference Arthur39] as well as in bovine and human E. coli O157:H7 strains isolated in the United States and Australia [Reference Mellor23].

Table 2.

Multivariate analysis of genotypes in E. coli O157:H7 isolated from food sources in Egypt

LSPA-6, Lineage-specific polymorphism.

Significant values appear in bold.

In conclusion, characterization of E. coli O157:H7 strains recovered from different food sources in Egypt revealed evidence of a higher frequency of distribution of lineage I/II, SBI 1, clade 8, Tir (255 T), and Stx _2c compared to other genotypes. Consequently, these genotypes might be utilized as potential valuable indicators of human infection risk from food sources. While our study was limited to a relatively small number of E. coli O157:H7 isolates, the findings may contribute to the international monitoring of infection risk genotypes that represent high virulence potential. To the best of our knowledge, there are no published data on the frequency of E. coli O157:H7 from human infections in Egypt and as a consequence such studies are urgently required to extend our analysis to determine the clinical incidence of such infections in Egypt and to determine the relationship between the described genetic markers, disease incidence, and clinical outcomes.