Association of Escherichia coli O157:H7 tir polymorphisms with human infection

Background Emerging molecular, animal model and epidemiologic evidence suggests that Shiga-toxigenic Escherichia coli O157:H7 (STEC O157) isolates vary in their capacity to cause human infection and disease. The translocated intimin receptor (tir) and intimin (eae) are virulence factors and bacterial receptor-ligand proteins responsible for tight STEC O157 adherence to intestinal epithelial cells. They represent logical genomic targets to investigate the role of sequence variation in STEC O157 pathogenesis and molecular epidemiology. The purposes of this study were (1) to identify tir and eae polymorphisms in diverse STEC O157 isolates derived from clinically ill humans and healthy cattle (the dominant zoonotic reservoir) and (2) to test any observed tir and eae polymorphisms for association with human (vs bovine) isolate source. Results Five polymorphisms were identified in a 1,627-bp segment of tir. Alleles of two tir polymorphisms, tir 255 T>A and repeat region 1-repeat unit 3 (RR1-RU3, presence or absence) had dissimilar distributions among human and bovine isolates. More than 99% of 108 human isolates possessed the tir 255 T>A T allele and lacked RR1-RU3. In contrast, the tir 255 T>A T allele and RR1-RU3 absence were found in 55% and 57%, respectively, of 77 bovine isolates. Both polymorphisms associated strongly with isolate source (p < 0.0001), but not by pulsed field gel electrophoresis type or by stx1 and stx2 status (as determined by PCR). Two eae polymorphisms were identified in a 2,755-bp segment of 44 human and bovine isolates; 42 isolates had identical eae sequences. The eae polymorphisms did not associate with isolate source. Conclusion Polymorphisms in tir but not eae predict the propensity of STEC O157 isolates to cause human clinical disease. The over-representation of the tir 255 T>A T allele in human-derived isolates vs the tir 255 T>A A allele suggests that these isolates have a higher propensity to cause disease. The high frequency of bovine isolates with the A allele suggests a possible bovine ecological niche for this STEC O157 subset.


Background
Shiga-toxigenic Escherichia coli O157:H7 (STEC O157) is the major STEC serotype associated with human infection in the U.S. [1]. Cattle are the predominant North American reservoir of this zoonotic pathogen [2,3] and contact with infected livestock and ingestion of contaminated meat are frequent routes of human infection [4][5][6][7]. Other sources of STEC O157 infection are contaminated fruits, vegetables and water [8][9][10] and person-to-person contact [11,12]. From 1982From -2002, there were 350 STEC O157 outbreaks reported in the U.S., resulting in 8,500 clinical cases, 1,500 hospitalizations and 40 deaths [13]. Human STEC O157 infections cause mild self-limiting diarrhea to severe disease including hemorrhagic colitis and hemolytic uremic syndrome (HUS) [14,15]. HUS, due to STEC O157 infection, is the leading cause of renal failure for children under the age of five years [1].
As with many infectious disease agents, STEC O157 strains appear to vary in their capacity to cause human infection and disease. For example, in the gnotobiotic pig challenge model, STEC O157 strains differ in both the clinical course they provoke and the histopathological lesions they induce [16]. Epidemiologic surveillance data in the U.S. also supports the idea of inter-strain variation in STEC O157 virulence. The annual U.S. incidence of clinical STEC O157 infections is estimated at 1.1 per 100,000 persons [1]. However, pooled data from five North American serological surveys found 11% of 2,251 healthy children and adults (11,000 per 100,000 persons) with serologic evidence of E. coli O157 exposure and/or subclinical infection [17][18][19][20][21]. On a smaller scale, the investigation of a recent STEC O157 outbreak linked with visiting an agricultural fair suggested that all STEC O157 are not equivalent in terms of their public health risk [22]. At least 25 people out of over 170,000 fair visitors who attended over a two-week period became ill with an STEC O157 isolate that shared the same pulse-field gel electrophoresis (PFGE) pattern. The outbreak investigation revealed that the fairground environment was heavily contaminated with multiple STEC O157 isolates with eight different PFGE patterns, including the outbreak strain. The presumed high human STEC O157 exposure but low human clinical disease incidence, suggested by both the surveillance and outbreak data, could be partially explained if only a subset of STEC O157 isolates present in the bovine (or other) zoonotic reservoirs were pathogenic to humans. Identifying markers for virulent strains as well as understanding the mechanisms responsible for disparities in virulence may provide new insights into the epidemiology and control of STEC O157 infections in both human and animal reservoirs.
An important step in the pathogenesis of human infection with STEC O157 is colonization of the lower gastrointes-tinal (GI) tract. STEC O157 have a number of virulence and putative virulence factors which aid in this colonization including the locus of enterocyte effacement (LEE), production of Shiga toxin 2, flagellin, OmpA, Lpf and ToxB [14,[23][24][25][26][27][28]. The interaction of two LEE-encoded genes, tir and eae, is responsible for the tight bacterial adherence to host epithelial cells characteristic of STEC O157 infections. The eae-encoded ligand protein intimin is located on the bacterial outer membrane. The intimin receptor protein Tir is translocated into the epithelial cell by type III secretion and integrated in the host cell membrane [29,30]. Given the role of intimin and Tir in STEC O157 pathogenesis and the well documented role of cattle as a zoonotic reservoir, the purpose of this study was to characterize sequence variation in STEC O157 eae and tir genes and to evaluate whether it associates with human or bovine host origin.

Bacterial strains
For sequence discovery, 22 diverse STEC O157 isolates were assembled that varied by source, either human clinical (n = 9) or bovine (n = 13) ( Table 1). A further 101 epidemiologically unrelated human clinical and 64 bovine isolates were included to estimate tir polymorphism allele frequencies. Each isolate was characterized by ELISA using anti-O157 and H7 monoclonal antibodies and multiplex PCR for stx1, stx2, eae, hlyA, rfb O157 and fliC H7 [31][32][33][34]. For the purpose of this study, isolates were defined as STEC O157 if they were E. coli O157 antigen positive by ELISA, rfbE O157 and fliC H7 -positive by PCR, and stx1 and/or stx2 positive by PCR.

PCR amplification, DNA sequencing and analysis
A 2,755-kb segment of the eae gene and 1,627-kb segment of the tir gene were amplified and sequenced using primers listed in Table 2. The amplification reactions contained 0.5 ng of DNA, 0.75 uM of each primer, 200 uM of each dNTP, 1.5 mM MgCl 2 and 1 U of Platinum Taq DNA polymerase (Invitrogen Corporation, Carlsbad, CA) in a 55 ul reaction. PCR amplifications were performed using a PTC-200 (MJ Research, Waterton, MA) at the following conditions: 1 min at 95°C followed by 30 sec at 96°C, 30 sec at 52°C and 2 min at 72°C for 35 cycles and finally 72°C for 7 min for 1 cycle.
PCR products were purified and concentrated using the QIAquick PCR Purification Kit (Qiagen Inc., Valencia, CA). DNA sequencing reactions were prepared using the ABI PRISM BigDye terminator cycle sequencing ready reaction kit (PE Applied Biosystems, Foster City, CA) with slight modifications of the manufacturer's protocol to reduce the final volume to 10 ul. The sequencing reactions were cycled with a PTC-200 (MJ Research) at the following conditions: 1 min at 96°C followed by 30 sec at 96°C, 1 min at 50°C and 4 min at 60°C for 30 cycles. DNA sequences were determined with either an ABI PRISM 3700 DNA analyzer or an ABI PRISM 377 DNA sequencer (PE Applied Biosystems).
Nucleotide sequences were analyzed using SeqMan and alignments were constructed using Clustal X, both from the Lasergene software package (DNASTAR, Inc., Madison, WI). A consensus parsimony tree was generated in PHYLIP (version 3.65) from tir DNA sequences using the program PARS [35] and viewed in TreeView (version 1.6.6) [36].
Pulse field gel electrophoresis of isolates used for genotyping PFGE was performed on all human and bovine derived E. coli O157:H7 isolates by using the PulseNet protocol and the restriction endonuclease XbaI [37]. Restriction frag-ment patterns were analyzed using Bionumerics version 4 (Applied Maths, Belgium).

Statistical analysis of tir variation and host association
The frequencies of each identified tir nucleotide or repeat polymorphism and tir genotype were compared between STEC O157 isolates of human and bovine origin. The data were analyzed as an unmatched case-control study by exact logistic regression using the LOGISTIC procedure of SAS 9.1 (SAS Institute, Inc., Cary, NC). The binary response variable (outcome) of interest was the probability of the strain being of bovine origin (case) vs human origin (control). Each tir polymorphism and genotype was converted into a categorical explanatory (predictor) variable, where each possible variant within a given polymorphism was coded separately. Genotype 10 and the most common variant for each polymorphism were used as reference conditions. The association of each tir poly- morphism variant with the likelihood of being a case STEC O157 strain was examined by generating univariate exact odds ratios (OR) with exact 95% confidence intervals (CI) and corresponding p values. Stx profiles of isolates defined by PCR were also examined for association with human or bovine strain origin.  Figure 1A). In one isolate, a chimeric repeat within RR1 was identified consisting of approximately the 5' half of RU2 and the 3' half of RU4 ( Figure 1B).

Polymorphisms in STEC O157 tir and eae
The five polymorphic tir loci defined ten unique tir genotypes ( Figure 1B). Two genotypes, 4 and 7, accounted for 83% (n = 185) of the isolates sequenced, while four genotypes were observed in only one isolate each ( Figure 1B). A consensus parsimony tree generated from these genotypes defined two major clades ( Figure 1B). Alleles of tir 255 T>A and RR1-RU3 were responsible for discrimination between these clades ( Figure 1B). Alleles of these two polymorphisms are strongly correlated. . This isolate also contained tir genotype 7, the most common tir genotype.

Association of tir 255 T>A and RR1-RU3 alleles with host origin of STEC O157 isolates
Unmatched case-control analysis showed that only tir 255 T>A and RR1-RU3 (presence or absence) alleles were significantly associated with host origin. Specifically, isolates with tir 255 T>A A allele were 34.0 times more likely (5.7 to 1381.9 95% CI, p < 0.0001) to be of bovine than human origin. Similarly, isolates with RR1-RU3 present were 32.0 times more likely (5.3 to 1302.9 95% CI, p < 0.0001) to be of bovine than human origin. Because tir 255 T>A and RR1-RU3 alleles discriminate between genotypes 4 and 7, these genotypes also associate with host origin. Specifically, isolates with genotype 4 were 37.0 times more likely to be of bovine than human origin (6.6 to ∞ 95% CI, p < 0.0001), while those with genotype 7 were 2.9 times more likely to be of human than bovine origin    [38]. Qgene allelic variation (upstream of the prophage stx region), Shiga-toxin 2 production differences and Shiga toxin-encoding bacteriophage insertion site-defined genotypes also had biased distributions of isolates from bovine and human origin [39][40][41]. However, none of these previously described methods provided as clear a discrimination between human and bovine isolates as those described in this study. Furthermore, the presence of one or both stx1 and stx2 genes (as determined by PCR) was statistically independent of an isolate's tir 255 T>A allele or RU3 presence or absence. The high degree of discrimination provided by tir 255 T>A and the central role of Tir in human infection points towards a possible functional role, rather than solely as a marker, for this polymorphism.
Previous studies indicate a paucity of nucleotide polymorphisms in most STEC O157 genes [42][43][44]. The presence of five polymorphic loci with high minor allele frequency within tir, therefore, appears to be atypical in STEC O157. Furthermore, all five polymorphisms are non-synonymous. In contrast, only one low frequency synonymous polymorphism was found in eae, suggesting that these two loci are under different selective pressures. The association of tir 255 T>A T allele with human infection argues that host factors may impose some selection pressure on tir.
The fact that the tir 255 T>A A minor allele frequency is over 30% in bovine isolates, where the frequency of minor alleles for most SNPs in STEC O157 is considerably less than that, also argues for some selection on this allele or another locus tightly linked to 255 T>A [43,44].
Limited information exists on complete tir gene sequence from STEC O157. Examination of the two published STEC O157 genomic sequences, EDL 933 and Sakai [45,46], showed that their tir genes were both genotype 7. Our sequencing of the tir gene from these two isolates confirmed this finding (data not shown for EDL 933). RR2, RR3 and RR4 together were previously used as a marker for high-resolution molecular typing of E. coli O157:H7 [47]. In the present study, the sequencing of a broad population of STEC O157 strains that included both human clinical and bovine reservoir isolates revealed additional informative tir polymorphisms, particularly the 255 T>A A allele and the presence of RR1-RU3.
Two polymorphic tir loci, tir 255 T>A and RR1-RU3, appear to have epidemiologic significance by their clear and strong association with isolate host source. Both loci are located near the amino terminus of Tir, a portion of the molecule that is normally located in the host cytosol during Tir-Intimin binding in host cell-bacterial adherence [48], in a region where no function has been described. However, these loci may have functional significance based on the biased distribution of their alleles in human derived isolates. One explanation for this could be variation in avidity, kinetics or tropism of adherence to epithelial cells, a major function of the Tir protein, from isolates with the tir 255 T>A A allele compared to isolates with the tir 255 T>A T allele. However, more investigation will be necessary to delineate the structure-function relationships of these tir polymorphisms.

Conclusion
Many host, bacterial and environmental factors impact whether or not infection results from human exposure to STEC O157. This study demonstrates that genomic polymorphisms in tir but not eae predict the likelihood that STEC O157 strains can cause human disease. Intriguing but unexplained findings include the host bias in tir allele frequency between human and bovine hosts. The overrepresentation of the tir 255 T>A T allele in humanderived isolates -vs the A allele proves its merit as a marker for virulence in humans. Also of interest is the high degree of tir sequence variation relative to that found in eae, even though the two proteins encoded by these genes interact together as receptor and ligand during adherence to host intestinal epithelial cells. Further research will be necessary to determine if the tir 255 T>A and RR1-RU3 polymorphisms are simply markers of strain virulence or are functional components of hostpathogen interactions.
genes and drafted the manuscript. JEK provided isolates for the study, carried out the statistical analysis and helped edit the manuscript. MLC carried out the phylogenic analysis of the tir DNA sequence. LMD participated in the PFGE analysis of the isolates and helped edit the manuscript. MPH participated in DNA sequence analysis. WWL co-conceived the study and experimental design, carried out the initial sequencing of the tir and eae genes and helped draft and edit the manuscript. All authors read and approved the final manuscript.