Significant differences were revealed in the diversity and virulence-gene carriage of Enterobacteriaceae and E. coli clones isolated from faeces of active and non-active coeliac patients and healthy controls, suggesting that this bacterial group could be primarily or secondarily involved in coeliac disease.
E. coli was the most common member of the Enterobacteriaceae family isolated from human faecal samples of healthy and coeliac disease children. This result is consistent with a previous study of 49 coeliac and healthy children, in which the main Enterobacteriaceae species detected was E. coli. In the aforementioned study, the ratio of E. coli to Enterobacteriaceae was greater in active coeliac children than in healthy controls . Furthermore, biopsy specimens taken from paediatric patients with inflammatory bowel disease also revealed that E. coli was the most common Enterobacteriaceae species detected in MacConkey agar . Thus, a reduction in the relative abundance of non-E. coli species is associated with coeliac disease and could favour the predominance of harmful clones in the gut ecosystem.
Although several Enterobacteriaceae colonies were often selected from a sample, because they differed morphology and were suspected to represent different strains, they often turned out to be identical regarding their phylogenetic classification and virulence-associated gene patterns, usually some of them represented the same clone. This can be explained by the fact that only one clone usually dominates human gastrointestinal gut so that the other clones are not detectable .
The phylogenetic groups A, B1, B2 and D were found in different proportions in healthy, active coeliac and non-active coeliac patients. This fact also suggests links between lower diversity of intestinal E. coli population and coeliac disease. In healthy children, isolates of each of the four groups were found in equal proportions. By contrast, isolates belonging to groups A and B2, followed by those of group D dominated the composition of E. coli microbiota in active coeliac children, while integrants of group B1 were not isolated; group D was the most common in non-active coeliac disease, followed by groups A, B2, and B1. These results reflect imbalances in the composition of the four E. coli phylogenetic groups in coeliac disease children, revealing the most remarkable differences between non-active coeliac patients and healthy controls. In patients with IBD, such as Crohn's disease and ulcerative colitis, B2 and D phylogenetic groups were more prevalent than in the control group. Moreover, a relationship was established between the presence of serine protease autotransporter proteins (SPATE) or adhesins in B2+D-positive E. coli isolates . In active coeliac patients, the dominance of group B2 virulent clones could provide pathogenic inflammatory factors either causative or consequence of the inflammatory status associated with gluten intake in the active phase of this disorder creating a vicious circle. Nowrouzian et al.  concluded that group D strains preferentially colonize infants with less complex intestinal microbiota, since this group is more abundant than others in one-year-old infants. Both genetic make-up and gluten-free diet could contribute to changes in the gut milieu of non-active coeliac patients, leading to a predominant colonization of the virulent D group in detriment to the others. In fact, lower total bacterial populations have been detected in duodenal samples taken from this population group when compared with active coeliac patients and controls .
Overall, E coli clones of the virulent phylogenetic groups (B2+D) isolated from active and non-active coeliac patients, carried a higher number virulent genes than those from healthy controls. A higher prevalence of P fimbriae gene (papC) was found in coeliac patients as compared to healthy children, regardless of the phase of this disorder; however, the association of type-1 fimbriae (fimA) and S fimbriae (sfaD/E) gene carriage with coeliac disease was not so evident. Significant differences in the carriage of type-1 fimbriae (fimA) gene were not detected in the group of children under study but E. coli clones from active and non-active coeliac disease isolates tended to carry fewer type-1 fimbriae (fimA) genes than those from healthy children. Previous studies have indicated that the proportion of type-1 fimbriated E. coli is lower in IgA-deficient subjects than in control individuals [20, 21] and this may also be the case for coeliac disease patients since this disease is associated with IgA deficiency. E. coli adhesins, including P fimbriae, confer mannose-resistant (MR) adherence to intestinal epithelial cells. MR adhesions are well-know virulence factors in urinary-tract infection, septicaemia and meningitis. Strains that persist in the human intestinal microbiota (resident strains) are more often P fimbriated, whereas S fimbriated E. coli are not associated with long-term persistence in the gut of healthy individuals [20, 22]. Thus, the significantly higher prevalence of P fimbriae E. coli clones in treated and untreated coeliac patients constitutes a novel link between gut microbiota and coeliac disease, and revealed that gut health may be compromised in these subjects, even when subjected to a gluten-free diet.
A higher prevalence of capsular K5 gene (kfiC) of E. coli clones belonging to the virulent groups was associated with active coeliac disease, while a similar but less remarkable trend was found in non-active coeliac patients, as compared to controls. Capsular polysaccharides are known to render bacterial surfaces hydrophilic and negatively charged, making the bacterium resistant to entrapment in mucus. In addition, capsules contribute to virulence by protecting bacteria from phagocytosis and possibly from serum killing, in part by blocking activation of the alternative complement pathway [23, 15].
Moreover, a higher prevalence of the haemolysin (hlyA) gene in E coli clones of the virulent groups was associated with non-active coeliac patients in particular. α-Haemolysin is the most common cytolytic protein secreted by haemolytic E. coli strains. Haemolysin activity might contribute to persistence by attacking enterocytes and releasing nutrients for the bacteria. In fact, E. coli appears to use membrane lipids as its main nutrient source in the large intestine [23, 24].
It has also been confirmed that E. coli clones of the virulent groups (B2 + D) carried more virulent-associated genes than those of commensal groups (A + B1), with the exception of the aerobactin gene (iutA). Remarkably, genes for S (sfaD/E) and P fimbriae (papC), K5 capsule (kfiC) and haemolysin (hlyA) were significantly more common in virulent E. coli clones than in commensal strains of our population groups. In agreement, genes coding for virulence-associated genes have been found more frequently in pathogenic strains than in commensal strains [8, 23]. In addition, virulent E. coli isolates (B2 + D) have previously been shown to carry virulence-associated genes more commonly than commensal clones [10, 25, 26]. Virulence-associated genes are usually encoded on pathogenicity islands (PAIs) providing a mechanism for coordinated horizontal transfer of virulence genes, thus favouring dissemination of pathogenic determinants, as could be the case in the gut ecosystem of coeliac patients .