Extraintestinal pathogenic Escherichia coli are associated with intestinal inflammation in patients with ulcerative colitis

E. coli of the phylogenetic group B2 harbouring Extra intestinal Pathogenic Escherichia coli (ExPEC) genes are frequently seen as colonizers of the intestine in patients with active ulcerative colitis (UC). In this study, we describe the influence of E. coli Nissle (EcN) B2 as add-on treatment to conventional therapies in patients with active UC. For this study one hundred active UC patients were randomized to ciprofloxacin or placebo for 1 week followed by EcN or placebo for 7 weeks. Stool samples were collected at weeks 0, 1, 8, 12, where E. coli were characterized and fecal calprotectin was measured. We showed that in the active UC patient group receiving Placebo/EcN, fewer patients reached remission, in comparison to the patient group receiving Placebo/placebo (p < 0.05). Active UC patients initially colonized with E. coli B2 had increased fecal calprotectin values and Colitis Activity Index scores in comparison to patients colonized with E. coli A and D (p < 0.05*). In conclusion, treatment of UC patients with E. coli Nissle (B2) does not promote clinical remission and active UC patients colonized with E. coli B2 have an increased intestinal inflammation.

Scientific RepoRts | 6:31152 | DOI: 10.1038/srep31152 but lacks α -hemolysin 12,13 . EcN was reported to maintain remission in UC patients with active disease and to prevent colitis in different murine colitis models 10,11,[14][15][16] . In the randomized double-blinded study of EcN given as add-on treatment to patients with active UC, we showed that significantly fewer patients treated with EcN reached symptomatic remission and that significantly more patients treated with EcN withdrew from the study 17 . Previous studies show that EcN is very closely linked to ExPEC isolates isolated from urinary tract infections, such as CFT073 and 536, even though certain virulence factors are not transcribed in EcN [18][19][20] . Fecal calprotectin is a non-invasive marker of intestinal inflammation used to distinguish between functional and organic bowel diseases and to evaluate disease activity in UC patients 21 . Clinical colitis activity indices correlate significantly with microscopic and macroscopic endoscopic scores, using clinical symptoms and laboratory findings 22 . However, fecal calprotectin predicts endoscopic disease activity far more reliably than the Colitis Activity Index (CAI) score 23 . In this study, the CAI score was performed as described by Rachmilewitz, including laboratory findings, CRP and hemoglobin 24 . Studies showed that UC patients between relapses have a calprotectin value between 123-213 mg/kg 25 . Therefore, in this follow-up study a calprotectin value of > 200 mg/kg was used as the criterion for active UC. Our aim was to evaluate the effects on intestinal inflammation of ciprofloxacin (Cipro) and orally administered EcN as add-on to conventional therapies in patients with active UC in correlation to fecal calprotectin values. Furthermore, the association between intestinal inflammation and primary colonization with E. coli B2 in active UC patients was assessed.

Results
Patient characteristics. One hundred patients with active UC, defined by a CAI score ≥ 6, were included in the study and randomized to four treatment groups: 1 week of ciprofloxacin or placebo followed by 7 weeks of EcN or placebo, with 25 patients in each group 17 (Fig. 1). We designated the groups as Cipro/EcN (A), Cipro/ placebo (B), Placebo/EcN (C) and Placebo/placebo (D).
Baseline characteristics for fecal calprotectin values were analysed using the t-test (Two-tailed), where three groups (A, B, C) were compared to the Placebo/placebo group. We found no significant differences in the median values for fecal calprotectin between Placebo/placebo (D) and Placebo/EcN (C) and Cipro/EcN (A) at week 0. However, there was a difference in the median fecal calprotectin value between the Placebo/placebo (D) and Cipro/placebo (B) groups (p < 0.05) at week 0, with a higher median fecal calprotectin in the Cipro/placebo group. Yet this had absolutely no effect over time and at week 12 where the fecal calprotectin was low and patients from the Cipro/placebo group reached remission at the same rate as the Placebo/placebo group (Table 1 and

Fecal calprotectin levels in patients treated with ciprofloxacin and/or EcN compared to placebo.
Fecal calprotectin values from weeks 0, 1, 8, and 12 of all 100 UC patients (regardless of the calprotectin value of the initial stool sample) were evaluated using the Two-way ANOVA test. There were no significant differences in the distribution of fecal calprotectin values over time, when comparing the four treatment groups, Cipro/ EcN (A), Cipro/placebo (B), Placebo/EcN (C) and Placebo/placebo (D), p = 0.067 ( Table 1). The table shows the number of fecal samples sent to the laboratory during the study, and the median fecal calprotectin values, by study groups. However, in order to see whether EcN has an overall effect, groups treated with EcN (AC) were compared to groups not treated with EcN (BD). When comparing the experimental groups receiving EcN from weeks 0, 1, 8, and 12 (Cipro/EcN and Placebo/EcN (AC)) with groups not receiving EcN (Cipro/placebo and Placebo/placebo (BD)), a non-significant trend of higher fecal calprotectin levels was found in patients treated with EcN, p = 0.053.

The effect of antibiotic treatment on intestinal inflammation measured by fecal calprotectin.
Patients with an initial fecal calprotectin level > 200 are included. Of patients treated with antibiotics (Cipro/EcN and Cipro/placebo) as add-on treatment, 52% (19/38) reached a fecal calprotectin value < 200 mg/kg, thus under remission; and 40% (14/35) of patients who were not treated with antibiotics (Placebo/EcN (C) and Placebo/ placebo (D) reached a fecal calprotectin value < 200 mg/kg. This difference was, however, not significant (Fig. 4). in the E. coli phylogenetic groups colonising the intestine. This suggests that E. coli Nissle does not colonise the intestine by competitive exclusion of other E. coli ( Table 2).

E. coli colonisation of UC
The effect of antibiotic treatment on E. coli B2 colonization. When evaluating the effect of antibiotics on E. coli B2 colonization, all 100 UC patients were included regardless of the fecal calprotectin concentration in the initial stool sample. Patient group B (treated with ciprofloxacin only) was initially colonized with 50% (12/24) E. coli B2 at week 0, prior to ciprofloxacin treatment. After antibiotic treatment, the number of the patients colonized with E. coli B2 was reduced to 5% (1/21), 37% (7/19) and 40% (6/15) at weeks 1, 8, and 12, respectively ( Table 2). These results demonstrate that one week's treatment with ciprofloxacin reduced the number of active UC patients colonized with E. coli B2 immediately after the treatment, but no long-term effect on E. coli B2 colonization could be demonstrated.

Colonization with E. coli B2 and intestinal inflammation in active UC patients at inclusion. UC
patients initially colonized with E. coli B2 had significantly increased fecal calprotectin values at week 0 in comparison to UC patients colonized with E. coli of phylogenetic groups A and D p < 0.05 (Fig. 5). However, the differences were not significant when comparing fecal calprotectin values of patients initially colonized with E. coli B2 with patients colonized with E. coli B1, p > 0.05 (Fig. 5).
Likewise, significantly increased CAI scores were found at week 0 in patients colonized with E. coli B2 compared to patients colonized with E. coli A and D, p < 0.05 (Fig. 6). The differences were not significant when comparing CAI scores of patients initially colonized with E. coli B2 with those of patients colonized with E. coli B1, p > 0.05.

Discussion
Ulcerative colitis patients are conventionally treated with anti-inflammatory medication. Since bacterial dysbiosis is suggested to cause disease relapses in UC, antibiotics and probiotics have also been used as treatment strategies [26][27][28] . E. coli strain Nissle 1917 (EcN), of the B2 phylogenetic group has been reported to maintain remission in UC patients and prevent colitis in a mouse model 10,11,[14][15][16] .
In the previous study 17 , it was shown that there was no benefit of using EcN as an add-on treatment to conventional therapies for active UC. Activity was determined by the Colitis Activity Index (CAI) questionnaire. Clinical, laboratory and endoscopic evidence show high specificity and sensitivity using CAI scores 29 . The disadvantage of using the CAI score is that it is unknown whether or not a high score is caused by a UC flare, by irritable bowel syndrome-like symptoms or by adverse reactions to treatment regimes, e.g. ciprofloxacin or the  probiotic EcN. In the last decade, fecal calprotectin has been used as a surrogate marker for inflammation, as a predictor for relapses [30][31][32] , and as a marker for mucosal healing in IBD patients 33,34 . Since fecal calprotectin predicts endoscopic disease activity far more reliably than a CAI score 22 , we performed the present follow up study using calprotectin as the marker of intestinal inflammation. The potential beneficial use of EcN as a probiotic, maintaining remission in patients with UC, has previously been described 10,11 . In this extension of our previous study of add-on treatments with one week of ciprofloxacin and/or EcN for seven weeks compared to ciprofloxacin and/or placebo for seven weeks 17 , we demonstrate by the use of fecal calprotectin measurements that EcN was in fact not beneficial as add-on treatment given to patients with active UC. Treatment with EcN resulted in fewer active UC patients reaching remission (fecal calprotectin < 200 mg/kg) in comparison to active UC patients not treated with EcN. Actual fecal calprotectin values from weeks 0, 1, 8 and 12 between patients treated with EcN compared to patients not treated with EcN (including all patients as long as they participated in the study) show a non-significant trend of higher fecal calprotectin levels in patients treated with ECN, (p = 0.053) ( Table 1). The explanation why this observation did not reach statistical significance might  be that more UC patients in the groups treated with EcN as add-on treatment withdrew from the study, making it impossible to follow up on the calprotectin levels among patients with the possibly worst outcome.
Based on fecal calprotectin results from weeks 0, 1, 8 and 12 of active UC patients with an initial fecal calprotectin level > 200 mg/kg, no significant differences in active UC disease activity could be found between patients treated or not treated with ciprofloxacin. A meta-analysis performed on the efficacy of broad spectrum antibiotics  in IBD patients showed 2.3 times better clinical improvement in patients receiving antibiotics such as ciprofloxacin 35,36 . However, patients included in the ciprofloxacin treatment groups were only treated with ciprofloxacin for 7 days, while studies so far recommend up to 16 weeks of treatment 36 . A longer duration of treatment with ciprofloxacin might change the outcome and cause remission in UC patients. Additionally, 50% of the patients included in this study were treated with EcN, which might have a negative effect on the ciprofloxacin treatment outcome. When comparing the ciprofloxacin treatment effect on E. coli B2 colonization in UC patients, there were no significant differences in the number of patients colonized with E. coli B2 either before treatment (week 0) or after week 12, even though a reduction in E. coli B2 colonization was found immediately after week 1 in patients treated with ciprofloxacin (Table 2). It has previously been shown in mice colonized with IBD associated E. coli that IBD associated E. coli reappeared some days after treatment with ciprofloxacin for 7 days 37 . This study indicates that 7 days of treatment with ciprofloxacin is not efficient to eradicate IBD associated E. coli. However, more studies are needed to clarify the efficiency of antibiotics in order to eradicate IBD associated E. coli.
Recent bacteriological studies on IBD patients show a characteristic individual variability in the mucosal bacteria and a large number of E. coli species belonging to the B2 and D phylogenetic groups 3,4 . The present study shows that UC patients initially colonized with E. coli B2 had significantly increased fecal calprotectin values and CAI scores at week 0 compared to patients colonized with E. coli D and A. The phylogenetic group B2 comprises among others ExPEC strains 38 . Furthermore, it is shown by genome sequencing that EcN is very closely linked to ExPEC isolates causing urinary tract infections, such as CFT073 and 536, even though certain virulence factors are not transcribed in EcN 19 .
These results indicate that the UC-associated E. coli B2 are associated not only with active disease but also with an increased burden of inflammation in UC patients. We did not see any beneficial effect of neither ciprofloxacin nor EcN. Therefore, it is still possible that eradication of UC associated E. coli B2, using other constellations of antibiotics and/or probiotics might benefit UC patients with active disease. Future trials should, however, take the presence or absence of E. coli B2 into account when evaluating the effect of these treatments.

Materials and Methods
Study design and samples. One hundred consecutive patients with flares of UC without any known gastrointestinal infections and without use of antibiotics for the past 4 weeks were included in the study. Patients included were aged > 18 years, with a CAI score ≥ 6 39 . The study was designed as a randomized double-blind placebo-controlled study of the effect of an add-on treatment of patients with flares of UC. Patients were allocated to one of four treatment groups: 1) Ciprofloxacin (500 mg twice daily) for one week followed by EcN (1 capsule twice daily) for 7 weeks; 2) Ciprofloxacin (500 mg twice daily) for one week followed by placebo for 7 weeks; 3) Placebo for one week followed by EcN (1 capsule twice daily) for 7 weeks; and 4) Placebo for one week followed by placebo for 7 weeks 17 . Standard medical care and therapies were allowed throughout the study, however, patients requiring treatment with systemic steroids or TNF-alfa inhibitors were excluded from the study, topical steroids were allowed. Standard medical treatments were comparable between groups as described in Petersen et al. 17 . Figure 6. CAI score in experimental groups (25 patients in each group) included in the study. One way ANOVA test was performed (mean, SEM), comparing differences between the mean CAI score of the patients colonized with E. coli B2 phylogenetic group versus E. coli A, D, B1. There is a significant difference between CAI scores in the patients initially colonized with B2 E. coli and the patients colonized with E. coli A and D, p < 0.05*. However, the differences were not significant when comparing the CAI scores of patients initially colonized with E. coli B2 with patients colonized with E. coli B1.
Scientific RepoRts | 6:31152 | DOI: 10.1038/srep31152 Patients were randomized 1:1:1:1, allowing 25 patients to be included in each group. Fecal samples were collected at weeks 0, 1, 8 and 12. An overview of the study material and the samples taken is seen in Fig. 1. Ethical Statement. Permission for treatment study and the recruitment of participants was approved by the Scientific Ethic Committee for Copenhagen Regional Hospitals (Permission no. H-1-2009-110). All participants gave their informed written consent. Collection of samples and data was carried out in accordance with the relevant guidelines as previously described 17 . Isolation of E coli from fecal samples. Fecal samples were sent for analysis at Statens Serum Institut, Copenhagen, Denmark. Laboratory staff analysed E. coli from the fecal samples without knowledge of the randomisation in the placebo-controlled study. Ten μ g fresh stool sample were mixed in 2 ml phosphate buffered saline (PBS, pH 7.38) and 10 μ l were plated on SSI enteric medium agar plates (SSI, Hillerød, Denmark, product no. 724) and incubated overnight at 37 °C. Bacteria from the SSI enteric medium were harvested and plated on SSI blue agar plates (SSI, Hillerød, Denmark, product no. 694) selective for gram negative bacteria and incubated overnight at 37 °C. E. coli colonies were isolated from the SSI blue agar plate and tested for Beta-glucuronidase 40 ((PGUA), SSI, Hillerød, Denmark, product no. 1033) and Indol (Biomérieux, Denmark, product no. 56541). Isolated E. coli were inoculated in Luria broth (LB) (Sigma-Aldrich, GmbH, Germany) and incubated overnight at 37 °C. Twenty-five μ l of bacterial culture in LB were diluted in 975 μ l of sterile water, boiled at 100 °C for 15 min., centrifuged for 10 minutes at 14.000 × g, and the supernatant (bacterial DNA) was transferred to a new tube and stored at − 20 °C for PCR testing. E. coli isolates with different colony morphologies from the SSI blue agar plates were harvested, dissolved in 15% glycerol (Sigma-Aldrich, GmbH, Germany. Cat: 67757) in beef stock (SSI, Hillerød, Denmark, product no. 1056), and stored at − 80 °C. E. coli phylogenetic group determination. E. coli phylogenetic groups (A, B1, B2 and D) were determined by a simple PCR procedure based on the genes chuA, yjaA and an anonymous DNA fragment, using primers and conditions exactly as described by Clermont et al. 41 .
Determination of fecal calprotectin. CALPROLAB TM Calprotectin ELISA (ALP) (Calpro AS, Oslo, Norway) is an enzyme-linked immunoassay (ELISA) based on polyclonal antibodies to human calprotectin (S100A8/A9). According to the manufacturer's instructions, 100 mg frozen fecal samples were homogenized in 4.9 ml extraction buffer using fecal extraction tubes (Calpro AS, Product No. CAL0500). The supernatants were diluted in sample diluent solution (1:100) before testing. Fecal extracts with values above measuring range were further diluted and re-tested. Values > 50 mg/kg were regarded positive 42,43 . Statistical analysis. Kaplan-Meier curves were used to compare groups. Test of equality of survival distributions for the different clinical treatment groups was performed using the Mantel-Cox (log-rank) test. The softwares "SAS 9.4" and "GraphPad Prism 5" were used for statistical analyses. The differences between the fecal calprotectin levels in the four patient groups among patients treated/not treated with EcN, were analysed using the SAS Two-way ANOVA test. The differences between fecal calprotectin levels at week 0 in patients colonized/not colonized with B2 E. coli were analysed using the t-test. A p < 0.05 is considered significant.