Inflammatory bowel disease (IBD), comprises of Crohn'sdisease (CD) and ulcerative colitis (UC), and is a chronic inflammation of the digestive tract of children and adults.1, 2 It is accepted that microorganisms play a central role in IBD; however, specific microbial etiologies remain elusive.3 For instance, Helicobacter pylori, Listeria monocytogenes, Yersinia pseudotuberculosis, Mycoplasma species, Chlamydia species, and even measles virus have been implicated in the pathogenesis of IBD.4, 5Mycobacterium avium spp paratuberculosis has been a popular topic of study for a long time because of its association with Johne's disease in cattle.2, 6–8
More recently, Escherichia coli have been strongly implicated in IBD.1, 6, 8–10 This has led to the description of a new pathovar of E. coli called adherent and invasive E. coli (AIEC).10–12 AIEC are able to adhere to intestinal epithelium and colonize the mucosa. These strains invade epithelial cells and macrophages and are able to replicate intracellularly.6, 11 Potential virulence factors identified for AIEC include serine protease autotransporters (SPATE), Ag43, and AIDA,1 as well as various adherence structures, particularly type I pilus.13, 14 In our laboratory,1 we isolated 3–4 logs more E. coli from IBD biopsies than healthy individuals, which is in agreement with Rhodes.10 These IBD isolated E. coli were predominantly of B2 and D phylotypes, which tend to be more pathogenic as described by Clermont et al.15
To further our understanding of the association of putative AIEC strains and IBD tissue we conducted multi-locus sequence typing (MLST) and fimH sequencing of E. coli isolated from IBD biopsies. MLST is based on the sequencing of conserved genes and can be used to evaluate the evolutionary relationships between closely related strains.16–19 FimH, the terminal subunit of the E. coli type I pilus, may be highly specific for IBD tissues.13, 14, 20 Carvalho et al21 demonstrated that when flagellin was deleted from the AIEC strain LF82, the experimental colitis in a rodent model was truncated.
- Top of page
- MATERIALS AND METHODS
A total of 36 E. coli isolates from IBD patients and healthy individuals was analyzed.1 Two strains of E. coli newly isolated from cattle feces, 2 strains from swine feces, the probiotic E. coli Nissle 1917,22 the nonpathogenic E. coli K12, and pathogenic enterohemorrhagic E. coli O157:H7 (EHEC), uro-pathogenic E. coli CFT073 (UPEC), and avian-pathogenic E. coli O1:K1:H7 (APEC) were included.
Identification of all E. coli strains were confirmed using API 20 E strips (Table 1). All E. coli isolates fermented glucose (GLU) and mannitol (MAN) but not inositol (INO). All strains were positive for ortho-nitrophenyl-β-D-galactopyranosidase (ONPG), and negative for urease (URE), tryptophane deaminase (TDA), and gelatinase (GEL). No isolate produced H2S or acetone (VP). Differences were detected between isolates in arginine dehydration (ADH), lysine decarboxylation (LDC), ornithine decarboxylation (ODC), citrate utilization (CIT), indole production (IND), and fermentation of sorbitol (SOR), rhamnose (RHA), sucrose (SAC), melibiose (MEL), amygdalin (AMY), and arabinose (ARA). No significant statistical association was found between any of biochemical characteristics of E. coli and the disease state (data not shown).
The allelic profiles of all isolates are shown in Table 3. The fumC and gyrB exhibited the greatest allelic diversity (17 alleles) and recA the least (12 alleles). Of 100 alleles obtained from this study, 9 were newly identified and added to the MLST.net database. All of the new alleles were associated with IBD. A total of 26 sequence types (ST) were identified and 11 of them were new STs (Table 3).
Table 4 shows all the STs, the number of isolates related to each ST (FREQ), the number of STs in the group that differ at a single locus (SLVs), double loci (DLVs), and those that are more distantly related (SATs). Isolates were placed into 4 different groups with only 1 isolate designated a singleton. All E. coli strains fell into separate clonal groups based on their phylotype except 1 strain (strain 128B). Group 1 primarily encompassed A and B1 phylotypes while Groups 2 and 4 only included B2. Group 3 was only made up of phylotype D. Chi-square analysis indicated that disease state (IBD) was significantly related to clonal groups (P < 0.05).
Concatenated sequences (total of 3423 bp) of housekeeping genes were used to construct a radiation tree that consisted of 3 major clusters, one B2 cluster, one D cluster, and one hybrid cluster of AB (Fig. 1). The clusters encompassed virtually the same clonal groups as determined by eBURST allelic analysis (Table 4).
When amino acid substitutions were compared the level of amino acid replacement was much higher with the fimH gene than for adk, fumC, gyrB, mdh genes (Table 5). No substitutions were detected for icd, purA, and recA genes (Table 5). However, amino acid substitutions of valine for alanine at 48 and 140, serine for asparagine at 91, and asparagine for serine at 99 residues of FimH were significantly associated with IBD (P < 0.05). Phylogenetic analysis based on fimH sequences further resolved the relationships between the isolates, and a cluster consisting of B2 isolates, including the adherent invasive strain LF82, uro-pathogenic E. coli CFT073 and the avian-pathogenic E. coli O1:H1:K7 (Fig. 2) was formed. Discriminant analysis using STs as the independent variable and disease state as the dependent variable associated the new STs to IBD (Fig. 3).
- Top of page
- MATERIALS AND METHODS
Although AIEC has been associated with IBD on a consistent basis,10–12 its phylogenetic position in relation to other pathogenic and nonpathogenic E. coli has not been explored. AIEC strains have been isolated from IBD tissue but it is unclear what their relationship is to E. coli that resides in the gut but causes extraintestinal disease. By extension it is not known whether AIEC forms a separate phylogenetic lineage within the E. coli species. MLST analysis of 7 housekeeping genes identified 26 different STs (new and old) among 45 E. coli isolates (Table 3) and even though 11 of the 26 ST were novel, this did not place them into a separate lineage (Fig. 1). In fact, most were placed into the same cluster as the urinary tract pathogen CFT073 and the avian pathogen APEC 01:K1:H7. However, the eBURST analysis of the allelic profiles grouped IBD isolates into separate clonal clusters (P = 0.02) (Table 4), suggesting that these isolates may have taken advantage of a specific “IBD microenvironment.” Actual phenotype differences are not obvious as biochemical profiling did not show any separation (Table 1).
We demonstrated that MLST analysis refined E. coli groups as previously defined by ABD typing (Fig. 1). E. coli from A and B1 group are less pathogenic, and B2 and D strains are mostly pathogens.23, 27 We previously demonstrated that IBD isolates tended to be B2 or D.1 For about 2 decades it was hypothesized that ABD lineages were the ancestral source of all polymorphism within E. coli.23, 28 However, a more detailed phylogenetic study by Wirth et al23 using MLST on a wide range of E. coli strains, revealed the presence of hybrid groups (AB, ABD) which hold ancestry from multiple phylotypes. They demonstrated that the hybrid groups accounted for one-third of E. coli strains and tended to contain E. coli that had undergone frequent recombinations. These hybrid groups were also rich in pathogens, indicating a possible link between virulence and recombination. According to their model, any commensal E. coli has the ability to acquire virulence by horizontal gene transfer, and as a result of this pathogenic lifestyle being more exposed to the host's immune system and consequently undergo higher rates of mutation and recombination.23 Based on MLST analysis, we showed that hybrid groups exist in IBD isolated E. coli (Fig. 1). Taken together, the MLST analysis and ABD typing support the conclusions of Wirth et al23 and confirm our previous finding that putative AIEC strains isolated from IBD tissue tend to belong to more pathogenic B2 or D phylotypes.1
Although commensal E. coli are normal inhabitants of the human gut, pathogenic E. coli can cause serious intestinal or extraintestinal infections.20 Type 1 fimbriae, which is responsible for bacterial adherence to the host epithelial surfaces, is a virulence factor of E. coli.13, 20 FimH, the adhesive subunit of type 1 pili, is located at the tip of the fimbriae and thought to mediate E. coli colonization.13, 29 It has been shown that amino acid substitutions in FimH can change E. coli tropism toward epithelial cells.29 We sequenced and aligned fimH sequences of the 45 E. coli isolates and determined that 4 amino acid substitutions were significantly associated with IBD (P < 0.05): fimH-V48A, N91S, S99N, and V140A. Some of these amino acid substitutions have also been reported previously as important in AIEC,13 and may facilitate the interaction of AIEC with CEACAM 6 (carcinoembryonic antigen-related cellular adhesin molecule) in IBD patients.20 The role of FimH amino acid substitutions in the pathogenesis of E. coli has been confirmed previously with mutation analysis.13
Phylogenetic analysis of fimH sequences delineated a tight cluster containing LF82, the best-described AIEC, as well as our own isolates that we had previously demonstrated to have a number of virulence genes (Fig. 2).1 However, strains CFT073 and APEC O1:K1:H7 were also in this cluster, and neither of these pathogenic strains was associated with IBD. This raises the possibility that IBD isolated E. coli are members of a general pool of extraintestinal pathogenic E. coli that reside in the gut and have evolved specific potentialities based on the microenvironment within which they find themselves. Bacteria that reside in human gut have the capacity to activate protective mucosal immune responses (tolerance, clearance, etc.) in normal hosts. In at least a subset of IBD patients, the mucosal immunity is dysfunctional and is unable to clear microbial pathogens. In these genetically predisposed hosts, the mucosal immune response to the putative enteric pathogens could be detrimental and result in the induction of chronic intestinal inflammation.30 An AIEC strain would then by extension be an organism that has accumulated genes and mutations to take advantage of this specific microniche. This hypothesis is supported by the observation that considerable intraspecies genomic size variation of E. coli is found in different strains.31 Addition or deletion of genome segments likely represent the accumulation, or loss, of genes encode for an adapted phenotype. Future research should focus on virulence factors that are unique for AIEC so that a better understanding of the mechanism of action of this group of bacteria can be achieved.