Genetic diversity of Shiga toxin-producing Escherichia coli O157 : H7 recovered from human and food sources

Microbiology (2015), 161, 112 119 DOI 10.1099/mic.0.083063-0 Genetic diversity of Shiga toxin-producing Escherichia coli O157 : H7 recovered from human and food sources Mohamed Elhadidy, 1 Walid F. Elkhatib, 2,3 Eman A. Abo Elfadl, 4 Karen Verstraete, 5 Sarah Denayer, 6 Elodie Barbau-Piednoir, 6 Lieven De Zutter, 7 Bavo Verhaegen, 5,7 Klara De Rauw, 8 Denis Piérard, 8 Koen De Reu 5 and Marc Heyndrickx 5,9 Correspondence Mohamed Elhadidy mm_elhadidy@mans.edu.eg 1 Department of Bacteriology, Mycology and Immunology, Faculty of Veterinary Medicine, Mansoura University, Mansoura 35516, Egypt 2 Department of Microbiology & Immunology, Faculty of Pharmacy, Ain Shams University, African Union Organization St Abbassia, Cairo 11566, Egypt 3 Department of Pharmacy Practice, School of Pharmacy, Hampton University, Kittrell Hall Hampton, VA 23668, USA 4 Department of Animal Husbandry and Development of Animal Wealth, Faculty of Veterinary Medicine, Mansoura University, Mansoura 35516, Egypt 5 Technology and Food Science Unit, Institute for Agricultural and Fisheries Research (ILVO), Brusselsesteenweg 370, Melle 9090, Belgium 6 Foodborne Pathogens, Scientific Institute of Public Health, Juliette Wytsmanstraat 14, 1050 Brussels, Belgium 7 Department of Veterinary Public Health and Food Safety, Ghent University, Salisburylaan 133, Merelbeke 9820, Belgium 8 UZ Brussels, Department of Microbiology, Belgian VTEC Reference Lab, Laarbeeklaan 101-1090 Brussels, Belgium 9 Department of Pathology, Bacteriology and Poultry Diseases, Ghent University, Salisburylaan 133, Merelbeke 9820, Belgium Received 5 August 2014 Accepted 14 November 2014 The aim of this study was to identify an epidemiological association between Shiga toxinproducing Escherichia coli O157 : H7 strains associated with human infection and with food sources. Frequency distributions of different genetic markers of E. coli O157 : H7 strains recovered from human and food sources were compared using molecular assays to identify E. coli O157 : H7 genotypes associated with variation in pathogenic potential and host specificity. Genotypic characterization included: lineage-specific polymorphism assay (LSPA-6), clade typing, tir (A255T) polymorphism, Shiga toxin-encoding bacteriophage insertion site analysis and variant analysis of Shiga toxin 2 gene (stx 2a and stx 2c ) and antiterminator Q genes (Q 933 and Q 21 ). The intermediate lineage (LI/II) dominated among both food and human strains. Compared to other clades, clades 7 and 8 were more frequent among food and human strains, respectively. The tir (255T) polymorphism occurred more frequently among human strains than food strains. Q 21 and Q 933+Q21 were found at significantly higher frequencies among food and human strains, respectively. Moreover, stx 2a and stx 2a+c were detected at significantly higher frequencies among human strains compared to food strains. Bivariate analysis revealed significant concordance (P,0.05) between the LSPA-6 assay and the other typing methods. Multivariable regression Abbreviations: HP, hairpin; HUS, haemolytic uraemic syndrome; LSPA-6, lineage-specific polymorphism assay; SBI, Shiga toxin bacteriophage insertion; STEC, Shiga toxin-producing Escherichia coli; stx 1, Shiga toxin 1; stx 2, Shiga toxin 2. Five supplementary tables are available with the online Supplementary Material. 112 083063 G 2015 The Authors Printed in Great Britain

STEC O157 : H7 genotypes from human and food sources analysis suggested that tir (255T) was the most distinctive genotype that can be used to detect bacterial clones with potential risk for human illness from food sources. This study supported previous reports of the existence of diversity in genetic markers among different isolation sources by including E. coli O157 : H7 strains from both food and human sources. This might enable tracking genotypes with potential risk for human illness from food sources. INTRODUCTION Shiga toxin-producing Escherichia coli (STEC) is a common food-borne zoonotic pathogen of global significance (Nataro & Kaper, 1998). E. coli O157 : H7 is an important E. coli serotype isolated from patients with foodborne illness that represent a considerable public health problem worldwide (Lee et al., 2011). Human infection is usually transmitted through direct contact with the ruminant reservoir or indirectly through ingestion of contaminated food or water (Mainil & Daube, 2005). E. coli O157 : H7 causes symptoms ranging from asymptomatic carriage to progressive sequelae extending from bloody diarrhoea, haemolytic colitis, to haemolytic uraemic syndrome (HUS), giving rise to the designation of this pathogen as enterohaemorrhagic E. coli (EHEC) (Rangel et al., 2005). To assess the public health risks associated with E. coli O157 : H7, it is crucial to understand the phenotypic and genotypic differences among strains from clinical illness, as well as from their sources of infection, that might enable better prediction of clinical and epidemiological outcomes (Stanton et al., 2014). Molecular typing and microbial genomics have facilitated the characterization and comparison of E. coli O157 : H7 strains recovered from different isolation sources demonstrating non-random distribution of genotypes among clinical and non-clinical strains (Besser et al., 2008; Franz et al., 2012; Hartzell et al., 2011; Lee et al., 2011; Whitworth et al., 2010; Yang et al., 2004; Yokoyama et al., 2011; Zhang et al., 2010; Ziebell et al., 2008). In addition, several epidemiological studies supported the growing evidence of substantial variability in the virulence of E. coli O157 : H7 genotypes implicated in different outbreaks and variation in patient symptoms (Besser et al., 2008; Grant et al., 2008; Manning et al., 2008). The lineage-specific polymorphism assay (LSPA-6) uses six loci derived from octamer-based genome scanning that differentiate E. coli O157 : H7 into three lineages (LI, LI/II and LII) that display apparent differences based on virulence potential as well as isolation sources (Lee et al., 2011; Zhang et al., 2010; Ziebell et al., 2008). Another subtyping scheme based on 32 SNPs has been recently developed to distinguish E. coli O157 : H7 strains into nine distinct evolutionary clades, with clade 8 exhibiting more virulence and more association with severe disease outcome than other clades (Manning et al., 2008). The tir (A255T) polymorphism for E. coli O157 : H7 is the SNP in the tir gene [tir (255T) and tir (255A)]. This gene encodes a translocated intimin receptor that mediates adhesion to mammalian cells and the formation of attaching and effacing (A/E) lesions through binding to intimin (Kaper et al., 2004). Some investigators previously suggested that tir (255T) harbouring strains are more virulent in humans than tir (255A) harbouring strains (Bono et al., 2007; Franz et al., 2012; Mellor et al., 2013). Shiga toxin bacteriophage insertion (SBI) site analysis relies on amplification of the stx toxin genes (stx 1 and stx 2 ) and the insertion site junctions of their encoding bacteriophages that can be used as a valuable genotyping technique to distinguish E. coli O157 : H7 strains, based on their distribution, gene expression and virulence potential (Besser et al., 2008; Shaikh & Tarr, 2003). Moreover, Shiga toxin 2 variants (stx 2a and stx 2c ) and stx 2 -Q antiterminator gene variants are clinically relevant genetic markers among E. coli O157 : H7 (Ahmad & Zurek, 2006; Persson et al., 2007). The aim of this study was to determine the genetic diversity and variability in frequencies of distribution of different genetic markers among E. coli O157 : H7 strains recovered from food and human clinical sources, using a combination of molecular subtyping methods. Consequently, from this characterization it was assessed if human clinical strains were more represented by specific genotypes of E. coli O157 : H7 and at which frequency these genotypes or markers were present in food strains. Furthermore, the correlation between various genetic markers was assessed to facilitate the prediction of pathogenic clones through food-borne transmission. METHODS Bacterial strains. A total of 170 E. coli O157 : H7 strains isolated from Belgium during the period 2000-2013 were used in this study (food, n570; human, n5100). Human clinical strains were collected by the Belgian national VTEC reference laboratory (UZ Brussels). The majority of the strains originated from humans suffering from diarrhoea, bloody diarrhoea, HUS or asymptomatic infection. Food strains were obtained from food samples of cattle origin including swabs from cattle carcasses, beef, minced beef and raw-milk cheese. Food strains were obtained by the Belgian national VTEC reference laboratory, the Scientific Institute of Public Health (WIV Brussels), the Department of Veterinary Public Health and Food Safety (Ghent University) and the Belgian Food Safety Authority (FAVV/AFSCA Brussels). All strains possessed either the gene encoding Shiga toxin 1 (stx 1 ), Shiga toxin 2 (stx 2 ) or both, identified by PCR analysis. Only unique strains per isolation source verified by PFGE were included in this study after selection from a total of 235 strains. Bacterial strains were stored at 280 uc using Pro-Lab Microbank cryovials (Pro-Lab) according to the manufacturer s instructions. Strains were cultured on Tryptone Soy Agar (TSA; Oxoid) and incubated aerobically at 37 uc for 24 h. Genomic DNA was extracted from tested strains as described (Flamm et al., 1984). The concentration and purity of the http://mic.sgmjournals.org 113

M. Elhadidy and others purified DNA were determined by measuring l 260 nm and the A260/ 280 ratio, using the Nanodrop ND-1000 UV-VIS Spectrophotometer (Nanodrop Technologies). Pulsed-field gel electrophoresis. PFGE was performed to determine non-identical strains for genotypic characterization. PFGE for tested E. coli O157 : H7 strains was performed using the restriction endonuclease XbaI, according to the Pulse Net protocol from the Centers for Disease Control and Prevention (Ribot et al., 2006). Images were analysed in BioNumerics version 6.5 (Applied Maths). The difference of at least one band was used as the criterion to distinguish one pulsotype from another. Lineage-specific polymorphism assay (LSPA-6). The LSPA-6 assay was performed in two multiplex PCRs using primer sequences and cycling conditions previously described (Ziebell et al., 2008) (Table S1, available in the online Supplementary Material). Amplification was performed using an Applied Biosystems GeneAmp 9700 thermal cycler. After amplification, equal volumes of PCR products from both reaction mixtures were mixed together and a 10 ml aliquot of this combined mixture was loaded on 3 % Metaphor agarose gel (Lonza) and visualized by staining with ethidium bromide. Images were analysed in BioNumerics version 6.5 (Applied Maths). Alleles shared with the lineage I control strain (93-001) were designated 1, and those shared with the lineage II control strain (FRIK 1999) were designated 2. Unique alleles were designated 3, and a zero character was assigned if no band was detected (Ziebell et al., 2008). All different genotypes were generated and assigned in the order previously described (Zhang et al., 2010; Ziebell et al., 2008). LSPA-6 control strains (93-001 and FRIK 1999) were both obtained from Dr Edward Dudley at Penn State University and Dr Victor Gannon at the Laboratory of Food-Borne Zoonosis, Public Health Agency of Canada. Clade typing. Clade typing was performed to differentiate E. coli O157 : H7 strains to nine different clades using hairpin (HP) primers targeting eight specific SNPs located in eight open reading frames (ORFs) as previously described (Manning et al., 2008). For each SNP, three primers were designed: two HP primers, each of which was specific for either the wild-type (reference) SNP or the mutant (diagnostic) SNP, and a conserved non-hp primer (Table S1). The assay was performed under the cycling conditions previously described (Riordan et al., 2008) and results were analysed with LightCycler 480 (Roche Diagnostics) using the LightCycler 480 software. Difference in critical threshold (CT) values between reference and diagnostic primers was used to interpret different clade types as published (Riordan et al., 2008). tir (A255T) polymorphism assay. E. coli O157 : H7 strains were genotyped for the tir 255 T.A allele by real-time PCR genotyping using primers, probes and cycling conditions as described (Bono et al., 2007) (Table S1). SBI genotyping. SBI genotypes of strains were determined as published (Besser et al., 2008; Shaikh & Tarr, 2003) with minor modification. Amplification of yehv or wrba was performed in two multiplex PCRs for detection of bacteriophage integration and amplification of stx 1 and stx 2 genes was performed in two uniplex reactions as described (Botteldoorn et al., 2003). Multiplex no. 1 included the right wrba-bacteriophage junction and the left yehvbacteriophage junction. Multiplex no. 2 included the left wrbabacteriophage junction and the right yehv-bacteriophage junction. Amplification was performed using Applied Biosystems GeneAmp 9700 thermal cycler with primer sequences and cycling conditions previously described (Shaikh & Tarr, 2003) (Table S1). PCR amplicons were coded for six characters (0 for absence and 1 for presence) in the following order: stx 1, stx 2, yehv-left, yehv-right, wrba-left and wrba-right and SBI genotypes were assigned as published (Whitworth et al., 2008). Detection of stx 1, stx 2a and stx 2c variants. Detection of stx 1 and stx 2 genes was performed by PCR using primer sequences and cycling conditions previously described (Botteldoorn et al., 2003) (Table S1). Isolates that showed a positive result for stx 2 were further tested for the presence of stx 2a and stx 2c by their respective PCR (Wang et al., 2002). Detection of Q 933 and Q 21 alleles. The bacteriophage antiterminator gene alleles (Q 933 and Q 21 ) were detected by PCR using the Applied Biosystems GeneAmp 9700 thermal cycler with the primer sequences and cycling conditions described by Ahmad & Zurek (2006) (Table S1). Statistical analysis. Statistical analysis was carried out using SPSS version 18.0 (SPSS). Data output of analyses with P-values less than 0.05 were considered statistically significant. Differences in frequencies of genetic markers among E. coli O157 : H7 strains from different isolation sources (food and human strains) were statistically evaluated using the chi-squared test in order to identify the variation in distribution of genetic markers among different isolation sources. Multivariate binary logistic regression analysis was performed to elucidate the most distinctive genetic marker that can be used to trace clinical-biased strains from food sources. Non-parametric correlation tests including Spearman s rho rank correlation, Pearson s correlation and Kendall s tau-b at two-tailed significance level were performed to determine the relationships between various genotyping methods. RESULTS Lineage-specific polymorphism assay (LSPA-6) LSPA-6 typing of E. coli O157 : H7 strains from human and food sources revealed that 70.6 % of tested strains belonged to lineage I/II (LSPA-6 211111). A total of 33 (19.4 %) strains belonged to lineage II exhibiting 16 different LSPA-6 designations (Table S2). Lineage I (LSPA-6 111111) was the least frequent lineage recovered from tested strains (10 %) (Table 1). Lineage II was detected in significantly higher frequency among food strains compared to human strains [x 2 59.72, degree of freedom (df)51, P,0.002]. On the other hand, no significant differences in distribution frequencies of lineages I and I/II among food and human strains were detected. Clade typing Clade typing assay identified seven different clades among E. coli O157 : H7 strains (clades 2, 3, 5, 6, 7, 8 and 9). Clades 5, 7 and 8 collectively represented 82.3 % (140/170) of E. coli O157 : H7 strains analysed. Clade 8 was found in a significantly higher frequency among human strains compared to food strains (x 2 55.882, df51, P,0.02) and the opposite was true for clade 7, which was detected in a significantly higher frequency among food strains compared to human strains (x 2 512.5, df51, P,0.001) (Table 1). tir (A255T) polymorphism analysis In total, 81.2 % (138/170) of E. coli O157 : H7 strains analysed possessed the tir (255T) allele. The tir (255A) occurred more 114 Microbiology 161

STEC O157 : H7 genotypes from human and food sources Table 1. Distribution of different genetic markers among E. coli O157 : H7 strains from food and human clinical sources Assay Genotypes No. (%) of food strains (n570) No. (%) of human strains (n5100) Total (n5170) LSPA-6 LI 6 (8.6 %) 11 (11 %) 17 (10 %) LI/II 42 (60 %) 78 (78 %) 120 (70.6 %) LII* 22 (31.4 %) 11 (11 %) 33 (19.4 %) SBI 1 44 (62.9 %) 55 (55 %) 99 (58.2 %) 3 6 (8.6 %) 12 (12 %) 18 (10.6 %) 5 10 (14.3 %) 5 (5 %) 15 (8.8 %) 6 5 (7.1 %) 5 (5 %) 10 (5.9 %) 11 1 (1.4 %) 3 (3 %) 4 (2.4 %) 16 0 (0 %) 2 (2 %) 2 (1.2 %) 19 1 (1.4 %) 0 (0 %) 1 (0.6 %) 21 3 (4.3 %) 17 (17 %) 20 (11.7 %) NT 0 (0 %) 1 (1 %) 1 (0.6 %) Clade typing 2 0 (0 %) 1 (1 %) 1 (0.6 %) 3 6 (8.6 %) 8 (8 %) 14 (8.2 %) 5 19 (27.1 %) 24 (24 %) 43 (25.3 %) 6 2 (2.8 %) 12 (12 %) 14 (8.2 %) 7 30 (42.9 %) 17 (17 %) 47 (27.7 %) 8 13 (18.6 %) 37 (37 %) 50 (29.4 %) 9 0 (0 %) 1 (1 %) 1 (0.6 %) tir (A255T) tir (255T) 46 (65.7 %) 92 (92 %) 138 (81.2 %) tir (255A) 24 (34.3 %) 8 (8 %) 32 (18.8 %) Q 21 Q 933 assay Q 21 32 (45.7 %) 20 (20 %) 52 (30.6 %) Q 933 34 (48.6 %) 61 (61 %) 95 (55.9 %) Q 933+ Q 21 4 (5.7 %) 19 (19 %) 23 (13.5 %) stx 2a /stx 2c assay stx 2a 17 (24.3 %) 44 (44 %) 61 (35.9 %) stx 2c 49 (70 %) 27 (27 %) 76 (44.7 %) stx2 a+c 3 (4.3 %) 26 (26 %) 29 (17 %) Non-stx 2a or stx 2c 1 (1.4 %) 3 (3 %) 4 (2.4 %) *Significant dissimilar distribution frequencies between human and food strains are indicated in boldface. LII, clade 7, tir (255A), Q 21 and stx 2c were detected in a significantly higher frequency among food strains than human strains. On the other hand, tir (255T), clade 8, Q 933+ Q 21, stx 2a and stx 2a+c were detected in significantly higher frequencies among human strains than food strains, suggesting the epidemiological importance of the latter genotypes as potential risk markers for human illness from food sources. NT, Non-typed. frequently in food strains compared to human strains (34.3 % and 8 %, respectively). On the other hand, tir (255T) was significantly more frequent in human clinical strains (92 %) compared to food strains (65.7 %) (Table 1). SBI genotyping Characterization of the Shiga toxin-encoding bacteriophage insertion sites identified eight unique SBI genotypes (Table 1). No significant association between SBI genotypes and isolation source was detected (x 2 514.85, df58, P.0.05). SBI genotype 1 was the most frequent SBI genotype representing 58.2 % (99/170) of E. coli O157 : H7 strains analysed (Table 1). Detection of Q 21 /Q 933 and stx 2a /stx 2c gene variants Characterization of Q 933 and Q 21 genes revealed that 52 (30.6 %) strains were positive for Q 21 alone, 95 (55.6 %) were positive for Q 933 alone and 23 (13.5 %) strains carried both Q alleles (Q 933+ Q 21 ) (Table 1). Q 21 was found at significantly higher frequencies in food strains compared to clinical strains (x 2 511.46 df51, P,0.001) and the combined presence of both alleles was detected at significantly higher frequency in human strains compared to food strains (x 2 55.128, df51, P,0.03). On the other hand, no significant variation in the distribution of Q 933 among food and clinical strains was observed (P.0.1). Screening strains for stx 2 variants (stx 2a and stx 2c ) identified 61 (35.9 %) strains carrying stx 2a, 76 (44.7 %) strains carrying stx 2c and 29 (17 %) carrying both variants (stx 2a +stx 2c ) (Table 1). Four strains (including three human strains) carried neither of the two variants. While stx 2a was detected at significantly higher frequency in human strains compared to food strains (x 2 56.125, df51, P,0.02) and the same for stx 2a+c (x 2 57.30, df51, P,0.03), stx 2c was detected at higher frequency in food strains compared to human strains (x 2 529.08, df51, P,0.001). http://mic.sgmjournals.org 115

M. Elhadidy and others Bivariate associations between LSPA-6 genotyping and other genotyping methods Pearson s correlation, Kendall s tau-b and Spearman s rho tests revealed statistically significant correlations (P,0.01) between LSPA and clade typing, tir (A255T), as well as Q gene variants. It is worth noting that statistical confidence levels differ between Kendall s tau-b and Spearman s rho correlation (two-tailed confidence levels of 95 %) and Pearson s correlation (two-tailed confidence level of 99 %). Accordingly, the high statistically significant (P,0.001) correlations were observed between LSPA-6 and clade typing at 99 % confidence level (based on Pearson s correlation) as well as between LSPA-6 and tir (A255T) at 95 % confidence levels (based on Kendall s tau-b and Spearman s rho correlation tests) (Table 2). All clade 8 strains belonged to lineage I/II and the remaining lineage I/ II strains belonged to clades 5, 6 and 7. All clade 3 strains belonged to lineage I, and all clade 6 strains belonged to lineage I/II. Clade 7 was mostly distributed between lineages II (59.6 %) and lineages I/II (40.4 %) and was absent in lineage I (Table S3). On the other hand, tir (255A) was absent in lineage I strains, was present at lower frequency in lineage I/II (18.7 %) and was dominant in lineage II (81.3 %). Furthermore, tir (255T) carrying strains were dominant in lineage I/II (82.6 %), followed by lineage I (12.4 %) and lineage II (5 %) (Table S4). Further significant association between LSPA-6 and Q 933 /Q 21 was observed (P,0.001) where the Q 21 allele was present at high frequency among lineage I/II strains (78.9 %) and all strains carrying the Q 933+ Q 21 alleles belonged to the same lineage. On the other hand, the Q933 allele was mostly distributed between lineages II and I/II (55.8 % and 42.3 %, respectively), but lineage I accounted for only 1.9 % of strains with this allele (Table S4). In addition, a significant association between stx 2 variants and LSPA-6 genotypes revealed that most stx 2a, stx 2c and stx 2a+c carrying strains belonged to lineage I/II (72.1 %, 60.5 % and 96.6 %, respectively) and both stx 2c and stx 2a+c were totally absent in lineage I that was only associated with stx 2a. Moreover, stx 2a was absent in lineage II (Table S4). Finally, most SBI genotypes 1, 3 and 5 belonged to lineage I/II, lineage I and lineage II, respectively (82.2 %, 94.4 % and 80 %, respectively) (Table S5). Multivariate binary regression analysis Multivariate binary regression analysis using the dataset of the genetic markers performed in this study was applied to elucidate the most distinctive genetic markers that can be used to predict clinical-biased strains from food sources. Multivariate analysis revealed that tir (255T) was the genetic marker that showed significant difference between human and food strains (P,0.05) (Table 3). DISCUSSION The current study was addressed to compare the frequencies of distribution of different genetic markers among E. coli O157 : H7 strains recovered from human and food sources. This analysis has been suggested to be crucial in monitoring different genotypes that might confer a high risk for virulence and/or transmission potential to humans (Mellor et al., 2013). Characterization of lineage types among E. coli O157 : H7 strains in Belgium revealed higher prevalence of lineage I/II among E. coli O157 : H7 strains analysed (70.6 %). This finding is consistent with previous studies that reported the higher frequency of distribution of this lineage among tested E. coli O157 : H7 strains from the Netherlands (55.1 %), Argentina (90 %) and Australia (88 %) (Franz et al., 2012; Mellor et al., 2012). This lineage has been shown to share characteristics of both lineage I and lineage II, but has been presented to cause human illness at frequencies similar to those of lineage I strain and includes a hyper-virulent isolate of which multistate spinach outbreak strains are representative (Zhang et al., 2010; Ziebell et al., 2008). Clade 8 strains were observed to be more significantly overrepresented among human clinical strains than food strains. To the best of our knowledge, this is the first study that reported the identification of clade 8 strains in Belgium. Moreover, a strong association between clade 8 and lineages I/II was observed where all clade 8 strains were represented by lineage I/II and were absent in the other lineages. A similar observation was reported by other studies (Laing et al., 2008; Liu et al., 2009; Hartzell et al., 2011) supporting the previous postulation that within lineage I/II, certain clades are more virulent and are Table 2. Correlation tests between LSPA-6 versus other subtyping methods used for genotyping of different E. coli O157 : H7 strains isolated from human and food sources Bivariate* Kendall s tau-b Pearson s correlation Spearman s rho correlation Covariance LSPA Q 933 /Q 21 0.338 (P,0.001) 0.271 (P,0.001) 0.373 (P,0.001) 0.116 LSPA stx 2a /stx 2c 0.149 (P50.037) 0.246 (P50.001) 0.150 (P50.052) 0.114 LSPA SBI 0.139 (P50.045) 0.129 (P50.095) 0.172 (P50.026) 0.568 LSPA clade types 0.319 (P,0.001) 0.526 (P,0.001) 0.359 (P,0.001) 0.551 LSPA tir (A255T) 0.427 (P,0.001) 0.292 (P,0.001) 0.441 (P,0.001) 0.077 *Correlations are significant at two-tailed confidence levels of 0.05, 0.01 and 0.05 for Kendall s tau-b, Pearson s correlation and Spearman s rho correlation tests, respectively. 116 Microbiology 161

STEC O157 : H7 genotypes from human and food sources Table 3. Results of uni- and multivariate logistic regression analysis differentiating human and food strains of E. coli O157 : H7 Genotypes Multivariable analysis OR (95 % CI)* P-value LI 2.12 1.000 LI/II ND ND LII 1.101 (0.273 4.43) 0.893 SBI 1 2.67610 8 1.000 SBI 3 0.390 (0.045 3.40) 0.394 SBI 5 1.72610 8 1.000 SBI 6 1.16 (0.096 19.67) 0.917 SBI 11 0.220 (0.013 3.77) 0.296 SBI 16 1.39 (0.036 54.46) 0.859 SBI 19 2.45610 9 0.999 SBI 21 2.85610 210 1.000 Q 21 0.690 (0.109 4.37) 0.694 Q 933 0.524 (0.099 2.78) 0.448 Q 933 +Q 21 ND ND tir (255T) 10.25 (1.2 88.05) 0.034 tir (255A) ND ND Clade 2 1.89610 28 1.000 Clade 3 1.62610 217 0.999 Clade 5 1.89610 28 1.000 Clade 6 2.51610 28 1.000 Clade 7 2.05610 28 1.000 Clade 8 2.11610 28 1.000 Clade 9 ND ND stx 2a 0.220 (0.008 6.32) 0.378 stx 2c 0.108 (0.006 2.03) 0.137 stx 2a+c 0.327 (0.011 9.44) 0.514 *95 % CI, 95 % confidence interval; OR, odds ratio. Odds ratios above 1 indicate that the presence of the genetic markers is positively associated with human isolates. The P-values shown represent the P-values of the ORs. Significant values are indicated in boldface. ND, Not determined. responsible for more frequent and severe diseases than others (Manning et al., 2008; Vanaja et al., 2010). Furthermore, these findings provided more evidence that clade 8 might be used as a useful surrogate genotype for the highly virulent lineage I/II strains. Our results revealed that the remaining lineage I/II strains were representatives of clades 5, 6 and 7. Clade 6 and 7 strains have been previously noted among US O157 strains that were designated to be represented by lineage I/II (Liu et al., 2009). To the best of our knowledge, this study provides the first evidence of clade 5 association with lineage I/II. Clade 7 was less frequent among human isolates than food strains, supporting the previous postulation that clade 7 is considered as of limited clinical relevance compared to clade 8 (Vanaja et al., 2010). While tir (255T) was overrepresented among human clinical strains, tir (255A) was found more frequently among food strains. Comparisons of lineage and tir SNP data revealed that tir (255A) was more associated with lineage II, while tir (255T) was associated at higher frequency with lineage I/II. A similar correlation between LSPA-6 and tir (A255T) has been reported in other studies (Arthur et al., 2013; Mellor et al., 2013), suggesting that tir (A255T) might be used as a useful predictor of lineage and as a promising indicator for potential pathogenic strains. Future experiments should be directed to analyse strains harbouring different alleles at position 255 of the tir gene for human intestinal colonization. Although SBI sites have been previously utilized to identify variations between clinical and non-clinical strains of E. coli O157 : H7 (Besser et al., 2008; Shaikh & Tarr, 2003), non-significant variation in the distribution of SBI genotypes among food and human strains was observed. SBI genotype 1 was the most frequent genotype present among tested strains, highlighting the potential epidemiological importance of this SBI genotype in Belgium. In contrast to the higher prevalence of SBI genotype 3 among human clinical E. coli O157 : H7 strains in the USA (Besser et al., 2008), the results of the current study revealed a lower frequency of SBI genotype 3 among human clinical strains (8.6 %), suggesting the possible limited epidemiological significance of this genotype in Belgium. Additional genetic markers exist at different relative frequencies among clinical and non-clinical isolates, including variants of the stx 2 -Q antiterminator junction alleles (Q 933 and Q 21 ) (LeJeune et al., 2004) as well as stx 2 gene variants (stx 2a and stx 2c ) (Beutin et al., 2007; Eklund et al., 2001; Friedrich et al., 2002). The Q 933 variant of the antiterminator gene Q is located upstream of the stx 2 gene, resulting in relatively high expression of the latter (Wagner et al., 2001), and was initially reported as a useful human risk indicator (Lowe et al., 2009). In Belgium, no significant difference was observed in the distribution of Q 933 among food strains and human clinical strains, suggesting that Q 933 might not be considered as a useful marker in detecting clinical-biased strains in Belgium. In addition, stx 2a and stx 2c were detected at significantly higher frequencies among human clinical and food strains, respectively. These results supported recent epidemiological studies that demonstrated higher association between stx 2a and clinical outcome, as compared to stx 2c (Andersson et al., 2011; Beutin et al., 2007; Persson et al., 2007). In the present study, a strong correlation between stx 2a /stx 2c genotypes and LSPA-6 genotypes was observed, supporting a recent postulation that both lineage types, together with stx genotype, may be informative surrogate markers for epidemiological and evolutionary investigations (Stanton et al., 2014). Multivariable logistic regression analysis was used to incorporate the confounding effects affecting variation in genotypes analysed. Our results revealed that tir (255T)was the only genetic marker that exhibited significant association with the isolation source (P,0.05), suggesting that http://mic.sgmjournals.org 117

M. Elhadidy and others tir (255T) is the most distinct genotype that can be used to track human-relevant strains from food sources. In conclusion, our analyses of Belgian E. coli O157 : H7 strains demonstrated a significant difference in distribution frequencies of genotypes between human and food strains. Our findings supported the growing epidemiological evidence of the existence of genetic markers that are overrepresented among E. coli O157 : H7 clinical strains and that are important determinants of human infection risk. Therefore, our results may have major implications in the development of effective control measures for foodborne STEC O157 : H7, which should be specifically directed against pathogenic bacterial clones responsible for human infections. Our results might also contribute to international monitoring of E. coli O157 : H7 genotypes associated with high virulence and increased infectivity potential. Consequently, our future studies will be directed to verify the obtained results using animal models to delineate the function of these genetic markers with the clinical outcome and disease incidence. Furthermore, it is worth noting that the advent of next-generation sequencing has provided a rapid increase in sequenced bacterial genomes, which can replace PCR-based typing in providing more information on the phenotypic and genotypic differences between strains that would allow better identification of increased virulence and transmissibility between strains and for tracking an ongoing outbreak that can be used in phylogenetic and evolutionary applications. ACKNOWLEDGEMENTS We would like to thank Dr Edward Dudley at Penn State University and Dr Victor Gannon at the Laboratory of Food-Borne Zoonosis, Public Health Agency of Canada, for providing control strains for LSPA-6 assay. The authors are grateful to Dr Shannon Manning at University of Michigan, Dr James Bono at Meat Animal Research Center, USA, and Dr Glen Mellor, CSIRO Animal, Food and Health Sciences, Australia, for technical advice during the experimental work as well as to the Erasmus Mundus action two mobility programme. This work was presented in part at the 19th Conference on Food Microbiology, Brussels, Belgium, 18 19 September 2014. REFERENCES Ahmad, A. & Zurek, L. (2006). Evaluation of the anti-terminator Q933 gene as a marker for Escherichia coli O157:H7 with high Shiga toxin production. Curr Microbiol 53, 324 328. Andersson, T., Nilsson, C., Kjellin, E., Toljander, J., Welinder-Olsson, C. & Lindmark, H. (2011). Modeling gene associations for virulence classification of verocytotoxin-producing E. coli (VTEC) from patients and beef. Virulence 2, 41 53. Arthur, T. M., Ahmed, R., Chase-Topping, M., Kalchayanand, N., Schmidt, J. W. & Bono, J. L. (2013). Characterization of Escherichia coli O157:H7 strains isolated from supershedding cattle. Appl Environ Microbiol 79, 4294 4303. Besser, T. E., Shaikh, N., Holt, N. J., Tarr, P. I., Konkel, M. E., Malik- Kate, P., Walsh, C. W., Whittam, T. S. & Bono, J. L. (2007). Greater diversity of Shiga toxin-encoding bacteriophage insertion sites among Escherichia coli O157: H7 isolates from cattle than in those from humans. Appl Environ Microbiol 73, 671 679. Beutin, L., Miko, A., Krause, G., Pries, K., Haby, S., Steege, K. & Albrecht, N. (2007). Identification of human-pathogenic strains of Shiga toxin-producing Escherichia coli from food by a combination of serotyping and molecular typing of Shiga toxin genes. Appl Environ Microbiol 73, 4769 4775. Bono, J. L., Keen, J. E., Clawson, M. L., Durso, L. M., Heaton, M. P. & Laegreid, W. W. (2007). Association of Escherichia coli O157:H7 tir polymorphisms with human infection. BMC Infect Dis 7, 98. Botteldoorn, N., Heyndrickx, M., Rijpens, N. & Herman, L. (2003). Detection and characterization of verotoxigenic Escherichia coli by a VTEC/EHEC multiplex PCR in porcine faeces and pig carcass swabs. Res Microbiol 154, 97 104. Eklund, M., Scheutz, F. & Siitonen, A. (2001). Clinical isolates of non- O157 Shiga toxin-producing Escherichia coli: serotypes, virulence characteristics, and molecular profiles of strains of the same serotype. J Clin Microbiol 39, 2829 2834. Flamm, R. K., Hinrichs, D. J. & Thomashow, M. F. (1984). Introduction of pam beta 1 into Listeria monocytogenes by conjugation and homology between native L. monocytogenes plasmids. Infect Immun 44, 157 161. Franz, E., van Hoek, A. H., van der Wal, F. J., de Boer, A., Zwartkruis- Nahuis, A., van der Zwaluw, K., Aarts, H. J. & Heuvelink, A. E. (2012). Genetic features differentiating bovine, food, and human isolates of shiga toxin-producing Escherichia coli O157 in The Netherlands. J Clin Microbiol 50, 772 780. Friedrich, A. W., Bielaszewska, M., Zhang, W. L., Pulz, M., Kuczius, T., Ammon, A. & Karch, H. (2002). Escherichia coli harboring Shiga toxin 2 gene variants: frequency and association with clinical symptoms. J Infect Dis 185, 74 84. Grant, J., Wendelboe, A. M., Wendel, A., Jepson, B., Torres, P., Smelser, C. & Rolfs, R. T. (2008). Spinach-associated Escherichia coli O157:H7 outbreak, Utah and New Mexico, 2006. Emerg Infect Dis 14, 1633 1636. Hartzell, A., Chen, C., Lewis, C., Liu, K. Q., Reynolds, S. & Dudley, E. G. (2011). Escherichia coli O157:H7 of genotype lineage-specific polymorphism assay 211111 and clade 8 are common clinical isolates within Pennsylvania. Foodborne Pathog Dis 8, 763 768. Kaper, J. B., Nataro, J. P. & Mobley, H. L. T. (2004). Pathogenic Escherichia coli. Nat Rev Microbiol 2, 123 140. Laing, C., Pegg, C., Yawney, D., Ziebell, K., Steele, M., Johnson, R., Thomas, J. E., Taboada, E. N., Zhang, Y. X. & Gannon, V. P. J. (2008). Rapid determination of Escherichia coli O157:H7 lineage types and molecular subtypes by using comparative genomic fingerprinting. Appl Environ Microbiol 74, 6606 6615. Lee, K., French, N. P., Hara-Kudo, Y., Iyoda, S., Kobayashi, H., Sugita-Konishi, Y., Tsubone, H. & Kumagai, S. (2011). Multivariate analyses revealed distinctive features differentiating human and cattle isolates of Shiga toxin-producing Escherichia coli O157 in Japan. J Clin Microbiol 49, 1495 1500. LeJeune, J. T., Abedon, S. T., Takemura, K., Christie, N. P. & Sreevatsan, S. (2004). Human Escherichia coli O157:H7 genetic marker in isolates of bovine origin. Emerg Infect Dis 10, 1482 1485. Liu, K., Knabel, S. J. & Dudley, E. G. (2009). rhs genes are potential markers for multilocus sequence typing of Escherichia coli O157:H7 strains. Appl Environ Microbiol 75, 5853 5862. Lowe, R. M., Baines, D., Selinger, L. B., Thomas, J. E., McAllister, T. A. & Sharma, R. (2009). Escherichia coli O157:H7 strain origin, lineage, and Shiga toxin 2 expression affect colonization of cattle. Appl Environ Microbiol 75, 5074 5081. 118 Microbiology 161

STEC O157 : H7 genotypes from human and food sources Mainil, J. G. & Daube, G. (2005). Verotoxigenic Escherichia coli from animals, humans and foods: who s who? J Appl Microbiol 98, 1332 1344. Manning, S. D., Motiwala, A. S., Springman, A. C., Qi, W., Lacher, D. W., Ouellette, L. M., Mladonicky, J. M., Somsel, P., Rudrik, J. T. & other authors (2008). Variation in virulence among clades of Escherichia coli O157:H7 associated with disease outbreaks. Proc Natl Acad Sci U S A 105, 4868 4873. Mellor, G. E., Sim, E. M., Barlow, R. S., D Astek, B. A., Galli, L., Chinen, I., Rivas, M. & Gobius, K. S. (2012). Phylogenetically related Argentinean and Australian Escherichia coli O157 isolates are distinguished by virulence clades and alternative Shiga toxin 1 and 2 prophages. Appl Environ Microbiol 78, 4724 4731. Mellor, G. E., Besser, T. E., Davis, M. A., Beavis, B., Jung, W., Smith, H. V., Jennison, A. V., Doyle, C. J., Chandry, P. S. & other authors (2013). Multilocus genotype analysis of Escherichia coli O157 isolates from Australia and the United States provides evidence of geographic divergence. Appl Environ Microbiol 79, 5050 5058. Nataro, J. P. & Kaper, J. B. (1998). Diarrheagenic Escherichia coli. Clin Microbiol Rev 11, 142 201. Persson, S., Olsen, K. E., Ethelberg, S. & Scheutz, F. (2007). Subtyping method for Escherichia coli shiga toxin (verocytotoxin) 2 variants and correlations to clinical manifestations. J Clin Microbiol 45, 2020 2024. Rangel, J. M., Sparling, P. H., Crowe, C., Griffin, P. M. & Swerdlow, D. L. (2005). Epidemiology of Escherichia coli O157:H7 outbreaks, United States, 1982-2002. Emerg Infect Dis 11, 603 609. Ribot, E. M., Fair, M. A., Gautom, R., Cameron, D. N., Hunter, S. B., Swaminathan, B. & Barrett, T. J. (2006). Standardization of pulsedfield gel electrophoresis protocols for the subtyping of Escherichia coli O157:H7, Salmonella, and Shigella for PulseNet. Foodborne Pathog Dis 3, 59 67. Riordan, J. T., Viswanath, S. B., Manning, S. D. & Whittam, T. S. (2008). Genetic differentiation of Escherichia coli O157:H7 clades associated with human disease by real-time PCR. J Clin Microbiol 46, 2070 2073. Shaikh, N. & Tarr, P. I. (2003). Escherichia coli O157:H7 Shiga toxinencoding bacteriophages: integrations, excisions, truncations, and evolutionary implications. J Bacteriol 185, 3596 3605. Stanton, E., Park, D., Döpfer, D., Ivanek, R. & Kaspar, C. W. (2014). Phylogenetic characterization of Escherichia coli O157 : H7 based on IS629 distribution and Shiga toxin genotype. Microbiology 160, 502 513. Vanaja, S. K., Springman, A. C., Besser, T. E., Whittam, T. S. & Manning, S. D. (2010). Differential expression of virulence and stress fitness genes between Escherichia coli O157:H7 strains with clinical or bovine-biased genotypes. Appl Environ Microbiol 76, 60 68. Wagner, P. L., Neely, M. N., Zhang, X., Acheson, D. W., Waldor, M. K. & Friedman, D. I. (2001). Role for a phage promoter in Shiga toxin 2 expression from a pathogenic Escherichia coli strain. J Bacteriol 183, 2081 2085. Wang, G., Clark, C. G. & Rodgers, F. G. (2002). Detection in Escherichia coli of the genes encoding the major virulence factors, the genes defining the O157:H7 serotype, and components of the type 2 Shiga toxin family by multiplex PCR. J Clin Microbiol 40, 3613 3619. Whitworth, J. H., Fegan, N., Keller, J., Gobius, K. S., Bono, J. L., Call, D. R., Hancock, D. D. & Besser, T. E. (2008). International comparison of clinical, bovine, and environmental Escherichia coli O157 isolates on the basis of Shiga toxin-encoding bacteriophage insertion site genotypes. Appl Environ Microbiol 74, 7447 7450. Whitworth, J., Zhang, Y., Bono, J., Pleydell, E., French, N. & Besser, T. (2010). Diverse genetic markers concordantly identify bovine origin Escherichia coli O157 genotypes underrepresented in human disease. Appl Environ Microbiol 76, 361 365. Yang, Z., Kovar, J., Kim, J., Nietfeldt, J., Smith, D. R., Moxley, R. A., Olson, M. E., Fey, P. D. & Benson, A. K. (2004). Identification of common subpopulations of non-sorbitol-fermenting, beta-glucuronidase-negative Escherichia coli O157:H7 from bovine production environments and human clinical samples. Appl Environ Microbiol 70, 6846 6854. Yokoyama, E., Hashimoto, R., Etoh, Y., Ichihara, S., Horikawa, K. & Uchimura, M. (2011). Biased distribution of IS629 among strains in different lineages of enterohemorrhagic Escherichia coli serovar O157. Infect Genet Evol 11, 78 82. Zhang, Y. X., Laing, C., Zhang, Z. Z., Hallewell, J., You, C. P., Ziebell, K., Johnson, R. P., Kropinski, A. M., Thomas, J. E. & other authors (2010). Lineage and host source are both correlated with levels of Shiga toxin 2 production by Escherichia coli O157:H7 strains. Appl Environ Microbiol 76, 474 482. Ziebell, K., Steele, M., Zhang, Y., Benson, A., Taboada, E. N., Laing, C., McEwen, S., Ciebin, B., Johnson, R. & Gannon, V. (2008). Genotypic characterization and prevalence of virulence factors among Canadian Escherichia coli O157:H7 strains. Appl Environ Microbiol 74, 4314 4323. Edited by: A. van Vliet http://mic.sgmjournals.org 119