Recent studies suggest that colonization of colonic mucosa by pathogenic Escherichia coli could be involved in the development of colorectal cancer (CRC), especially through the production of genotoxins such as colibactin and/or by interfering with the DNA mismatch repair (MMR) pathway that leads to microsatellite instability (MSI). The present study, performed on 88 CRC patients, revealed a significant increase in E. coli colonization in the MSI CRC phenotype. In the same way, E. coli persistence and internalization were increased in vitro in MMR-deficient cells. Moreover, we demonstrated that colibactin-producing E. coli induce inhibition of the mutL homologue 1 (MLH1) MMR proteins, which could lead to genomic instability. However, colibactin-producing E. coli were more frequently identified in microsatellite stable (MSS) CRC. The present study suggests differences in the involvement of colibactin-producing E. coli in colorectal carcinogenesis according to the CRC phenotype. Further host–pathogen interactions studies should take into account CRC phenotypes.

Clinical perspectives

  • The study presents the first clinical evidence for an interaction between pathogenic colibactin-producing E. coli and DNA MMR deficiency in CRC.

  • Moreover, this work revealed differences in the involvement of these strains in colorectal carcinogenesis according to the CRC phenotype.

  • The ability of colibactin-producing E. coli to participate in colorectal carcinogenesis is now well supported, and further host–pathogen interactions studies should take into account tumour CRC phenotypes.

Introduction

Colorectal cancer (CRC) is associated with well-known genetic lesions that are described in the adenoma to carcinoma progression sequence [1,2]. Indeed, tumorigenesis proceeds through a series of genetic alterations involving oncogenes (particularly RAS and RAF), tumour suppressor genes [particularly adenomatous polyposis coli (APC) and P53] and/or DNA mismatch repair (MMR) genes [13]. The DNA MMR system mostly involves two dimers [e.g. mutL homologue 1 (MLH1)/postmeiotic segregation 2 (PMS2) and mutS homologue 2 (MSH2)/mutS homologue 6 (MSH6)] and corrects mismatches generated during DNA replication, such as base–base mismatches and shorter insertion–deletion loops, and mismatches that have escaped proofreading by DNA polymerase, particularly in microsatellites. All of these anomalies could lead to the creation of DNA double-strand breaks, and the MMR system is thus a DNA-repair pathway that participates to maintain genetic stability [2]. Mutation or inactivation of MMR genes, predominantly on MSH2 and MLH1 genes, is observed in 15% of sporadic CRC and in 90% of hereditary non-polyposis CRC (HNPCC) [2]. These alterations typically result in a phenotype named microsatellite instability (MSI), the opposite of a microsatellite stable (MSS) phenotype.

The CRC microenvironment is a complex association of non-neoplastic/tumoural cells and a large amount of microorganisms that constitute the microbiota. An increasing number of studies have demonstrated that there is a shift in the intestinal microbiota composition (dysbiosis) during colorectal carcinogenesis [412]. However, it is unknown if this shift occurs before the onset of the disease. Different species (Fusobacterium nucleatum, Streptococcus gallolyticus, Enterococcus faecalis, Bacteroides fragilis and Escherichia coli) were described for their pro-tumoural activities, including the induction of chronic inflammation and/or the production of carcinogenic metabolites and toxins [1320]. Among these bacteria, mucosa-associated E. coli are more frequently identified in colon tissue from patients with CRC compared with controls [16,2126]. Even if E. coli is a commensal bacterium, some pathogenic strains have acquired some virulence factors including toxins, such as cyclomodulins, which interfere with the eukaryotic cell cycle [2729]. Especially, the colibactin toxin, synthetised from the pks island, was preferentially detected in strains isolated from CRC patients [10,16,2426]. Colibactin induces in vitro DNA double-strand breaks, a transient DNA damage response leading to genomic instability, and could be involved in colorectal carcinogenesis [27,28]. Some other pro-carcinogenic effects of these colibactin-producing E. coli have been reported in vivo in several mice models. Indeed, these bacteria could enhance tumour growth in both xenograft and chemically induced colitis-associated CRC (CAC) models, sustained by cellular senescence that was accompanied by the production of growth factors, such as the hepatocyte growth factor [30]. In addition, experiments using CAC mice models (IL10−/−+AOM) showed that colibactin-producing E. coli were able to highly persist in the gut, and to induce epithelial damage and cell proliferation. This bacterial cancer-promoting activity was dependent on inflammation [19,31]. Moreover, enteropathogenic E. coli (EPEC), which are the most commonly identified attaching and effacing bacteria in humans and are responsible for potentially fatal acute diarrhoea [32], have been detected in CRC. In vitro, EPEC depletes the DNA MMR system, which is altered in the MSI CRC phenotype, through the post-transcriptional effect of the bacterial-secreted effector protein EspF, and through the EspF-independent production of host reactive oxygen species (ROS) [21,33]. Then, this bacterial-induced DNA MMR system depletion was significantly associated with an increase in the spontaneous mutations frequency in host cells, thus promoting colorectal carcinogenesis [33]. These in vitro results described the relationship between the microbiota and MMR pathways alterations. In this way, a study on murine models strongly suggests an interaction between the intestinal microbiota, the diet and the MMR deficiency in CRC [34]. The aim of the present study was to investigate the association between MMR pathways and colibactin-producing E. coli in both cultured cells and human samples. We studied in vivo the human gut colonization by E. coli in relation to the MSI phenotype in CRC patients. We showed a significant increase in the colonic E. coli colonization in the MSI CRC phenotype. Furthermore, in vitro E. coli persistence and internalization were both increased in MMR-deficient cells. Moreover, we demonstrated that colibactin-producing E. coli inhibited MLH1 MMR protein expression, which increases bacterial persistence in colonic epithelial cells and could lead to genomic instability. This bacterial effect would be dependent on ROS induction and oxidative stress. Finally, colibactin-producing E. coli were more frequently identified in MSS tumour tissue, suggesting differences in the involvement of these strains in colorectal carcinogenesis according to the CRC phenotype.

Materials and methods

Ethics statement

Ethical approval for the present study was granted by the research ethics committee at the University Hospital of Clermont-Ferrand (France). The biological samples were collected from colon resections, which were required for the treatment of patients. The investigators explained the study verbally to the potential subjects and provided all pertinent information such as the purpose of the study, all study procedures and the putative risks of the study. Following this verbal explanation, the potential subject was provided with a study information sheet. After allowing the potential subject time to read the study information sheet, the investigator answered any additional questions the potential subject may have had. A signed consent to participate in the research was obtained for all patients included in the study. The dates of signed consent were tracked in a non-identifiable manner.

Patients and resection specimens

Patients underwent surgery for resectable colon cancer (n=88) in the Digestive and Hepatobiliary Surgery Department of the University Hospital of Clermont-Ferrand, France, between March 2007 and March 2014. All of the patients were prospectively enrolled in the study. All of the patients were adult volunteers and signed informed consent prior to being included in the study. The exclusion criteria for the study included clinically suspected hereditary CRC using the revised Bethesda criteria [35], neo-adjuvant chemotherapy, history of previous colonic resection, emergency surgery and use of antibiotics within 4 weeks prior to surgery. The patient's medical history was collected. Surgery was performed by either laparotomy or laparoscopy. None of the patients received mechanical bowel preparation before surgery. All of the patients received cefoxitin (2 g intravenously), an antibiotic prophylaxis, at the time of the incision. A non-necrotic fragment from the peripheral areas of the tumour was collected from the colon cancer patients. In parallel, macroscopically normal mucosa samples adjacent to the tumour (10 cm from the tumour) were collected.

Pathological and immunohistochemical analyses

After colonic resection, fresh specimens were transported to the Pathology Department laboratory, fixed in buffered 4% paraformaldehyde, embedded in paraffin, cut into 5 μm slices and stained with haematoxylin–eosin–safranin. All tumours were histologically determined to be adenocarcinomas.

Immunohistochemical staining of the human MMR nuclear proteins MLH1, MSH2, MSH6 and PMS2 was performed on the colonic mucosa sections from the colon cancer patients (n=88). Five micrometre sections of paraffin-embedded, tumour mucosa were labelled with the anti-MLH1 M1 antibody (Ventana Medical Systems®), the anti-MSH2 G219-1129 antibody (Ventana®), the anti-MSH6 44 antibody (Ventana®) or the anti-PMS2 EPR3947 antibody (Ventana®) and Ultra View Detection kit (Ventana®) on Benchmark XT stainer (Ventana®). Sections were counterstained with Harris's haematoxylin (Ventana®). For each patient, two pathologists from the Pathology Department of the University Hospital of Clermont-Ferrand (France) performed qualitative analysis of immunohistochemical staining of nuclear MMR proteins in the tumours. The absence of staining in the epithelial cells was validated only when positively stained areas were observed in the stroma of the tumours or in the normal mucosa.

DNA extraction and MSI status molecular analysis

Because of the tumour's heterogeneity, in order to obtain a sufficient number of tumour cells, the maximum tumour-rich area was selected for each formalin-fixed, paraffin-embedded (FFPE) tumour block after reviewing the corresponding H&E-stained slides. DNA was extracted from a 3–5 mm3 sample taken from the tumour-rich area using the QIAamp DNA FFPE Tissue kit (Qiagen®), according to the manufacturer's instructions. The MSI Analysis System version 1.2 (Promega®) was used to amplify 25–75 ng of DNA on a SureCycler 8800 Thermal Cycler (Agilent Technologies®), according to the manufacturer's instructions. PCR products were resolved on a 3130xl sequencer and analysed with GeneMapper software (Life Technologies®).

Microbiological analysis

The mucosa-associated and mucosa-internalized E. coli levels on normal mucosal specimens were immediately analysed as previously described by Buc et al. [25]. Based on the level of mucosa-associated E.coli, patients were divided into high and low colonized subgroups with the median abundance obtained for all samples [3.22×104 colony-forming units (CFU)/g of tissue].

The prevalence of E. coli harbouring colibactin-encoding pks island was investigated using PCR tests on all CRC samples. Total DNA was extracted from a 3–5 mm3 mucosal sample taken from the tumour-rich area using the NucleoSpin Tissue kit (Macherey-Nagel®), according to the manufacturer's instructions. DNA amplification using the Platinum Taq DNA Polymerase (ThermoFischer Scientific®) was then performed on the T-100 Thermal Cycler (Bio-Rad Laboratories®) with 10 μM primers located in the clbN gene of the pks island (CLBN-F: GTTTTGCTCGCCAGATAGTCATTC, CLBN-R: CAGTTCGGGTATGTGTGGAAGG; Eurogentec®) [36].

The five representative CRC-colibactin-producing E. coli strains (11G5, 10D12, 18H5, 16C1 and 14H4) used for cell infection experiments, were isolated from the colonic mucosa of study participants as previously described [37]. The presence of pks island-encoded genes in these E. coli strains was verified using PCR, and colibactin expression was determined by assessing a specific cytopathic effect on cultured epithelial HeLa cells [37]. The non-pathogenic K12-C600 E. coli strain was used as a control commensal bacteria strain. The two representative colibactin-negative E.coli strains (9F1 and 1D5) were previously isolated from diverticulosis mucosa [25]. The 11G5clbQ isogenic mutant was previously generated using the method described by Datsenko and Wanner [30,38], which was modified by Chaveroche et al. [39].

Cell culture and infection

Intestinal epithelial cells (T-84 and HCT-116) were grown following the ATCC's guidelines, in an atmosphere containing 5% CO2 at 37°C in appropriate medium. The cells were seeded in 24-well plates at a density of 1×105 cells/well for all the experiments and in six-well plates at a density of 5×105 cells/well for Western blot analyses.

The bacteria were grown overnight at 37°C in LB medium. Bacterial inoculums were assessed at an absorbance of 620 nm using a UV-1800 spectrophotometer (Shimadzu®) and controlled by bacterial spreading on LB agar plates. Overnight E. coli cultures were inoculated into the cell culture medium at a multiplicity of infection (MOI) of 100 bacteria per cell. Infected cells were centrifuged at 900 g for 10 minutes at room temperature (RT) and placed at 37°C.

siRNA treatment

For MLH1 silencing, siMLH1 (ThermoFischer®) and control siRNA (ThermoFischer®) were diluted in OptiMEM at a final concentration of 100 nM and placed on cultured T-84 cells for 24 h. The cells were then infected as described above.

Antioxidant treatment

To study the effect of antioxidant treatment, T-84 cells were cultured for 24 h in appropriate medium plus α-lipoic acid (LA) (Sigma–Aldrich®) diluted at a final concentration of 50 μg/ml. The cells were then infected as described above.

Adhesion, internalization and persistence assays

Adhesion, internalization and 48 h post-infection persistence assays were performed on T-84, HCT-116 and MLH1-silenced T-84 infected cells as previously described [40]. For adhesion assays, the cells were washed three times in PBS after 3 h of incubation at 37°C. To determine the number of intracellular bacteria (internalization assay), cell culture medium containing gentamicin at a concentration of 200 μg/ml was added for 1 h to kill extracellular bacteria. The epithelial cells were then lysed with 1% Triton X-100 in deionized water. This concentration of Triton X-100 had no effect on bacterial viability for at least 30 min. The samples were diluted and plated on to LB agar plates to determine the number of CFU. For persistence assays, the infected cells were treated with gentamicin after 3 h of incubation as previously mentioned, incubated for 48 h post-infection and lysed with 1% Triton X-100 in deionized water. The lysates were then diluted and plated on to LB agar plates to determine the number of CFU.

Transcriptional analysis of the MMR protein expressions

Cells were harvested 3 h after infection, washed twice with PBS and detached with the RP1 buffer of the NucleoSpin® RNA/Protein kit (Macherey-Nagel®) and placed at −80°C until total RNA extraction was performed.

Total RNA was extracted with the NucleoSpin® RNA/Protein kit (Macherey-Nagel®), according to the manufacturer's recommended protocol. The integrity and the quality of the extracted total RNA were evaluated by agarose gel electrophoresis and spectrophotometer measurements (260 and 280 nm).

RNA samples were subjected to reverse transcription using the High-Capacity cDNA Reverse Transcription Kit® and non-specific random hexamer primers (Applied Biosystems®) at 37°C for 120 min, according to the manufacturer's recommended protocol. Pre-designed TaqMan® probe and primer sets for four target genes were chosen from an online catalogue (Applied Biosystems®). Gene expression values were calculated by the ΔΔCt method. Expression levels (RQ) of target genes were normalized to those of two housekeeping genes (HPRT and GAPDH). For each gene, expression on infected cells was compared with that of uninfected cells.

Immunoblot analysis for MLH1 expression

After 3 h of infection, proteins were extracted from T-84 cells using lysis buffer (25 mM Tris pH 7.5, 1 mM EDTA, 5 mM MgCl2, 1% NP-40, 10% glycerol, 150 mM NaCl, 10 mg/ml sodium orthovanadate and 1 mM PMSF). Equal amounts of whole cell protein extracts (20 μg) were separated on an SDS/12% PAGE gel and transferred to nitrocellulose membranes (GE Healthcare). After treatment for 1 h at RT with blocking buffer (TBS, 0.05% Tween-20 and 5% BSA), the membranes were blotted with anti-MLH1 (Cell Signaling Technology®) at 1:500 or anti-GAPDH (Cell Signaling Technology®) at 1:1000 primary antibodies overnight at 4°C. Then, the membranes were washed and incubated with the appropriate HRP-conjugated antibodies for 1 h at RT. The proteins were detected using the ECL Plus® chemiluminescence detection system (ThermoFischer®) and quantified using Image Lab™ software (Bio-Rad Laboratories®).

Analysis of reactive oxygen species induction

After 3 h of infection, infected cells were washed with FBS-free cell culture medium and incubated with the 2′,7′-dichlorodihydrofluorescein diacetate (H2DCF-DA) probe (Sigma–Aldrich®) at a final concentration of 150 μM for 45 min at 37°C. The 522 nm fluorescence was then measured in nine areas of each well using the Fluoroskan fluorescent microplate reader (Labsystems®) after an excitation at 485 nm. Uninfected cells were considered as controls.

Immunofluorescence analysis of the γH2AX

After 3 h of infection, infected cells on plates were fixed in buffered 4% paraformaldehyde for 10 min at RT, and permeabilization was performed using 0.1% Triton X-100 for 10 min at RT. Unspecific sites were blocked using 5% FBS for 30 min at RT. Cells were then incubated with anti-phospho-histone H2AX (γH2AX) (Ser139) (Cell Signaling Technology®; clone 20E3) at 1:200 primary antibody for 2 h at RT. Then, the cells were washed and incubated with the appropriate FITC secondary antibody at 1:200 in PBS supplemented with 5% FBS, 0.3% Triton and 1 μl/10 ml Hoechst for 1 h at RT. Slides were mounted using Mowiol and were observed by fluorescent microscopy. The number of cells presenting nuclear γH2AX staining was evaluated on at least 100 cells per each well. All conditions were performed in triplicate. Uninfected cells were considered as controls.

Statistical analysis

All analyses were performed using GraphPad Prism 5 STATA (StataCorp). P-values were two-tailed, and those <0.05 were considered statistically significant. Comparisons of patient's characteristics between groups were carried out using the Chi-squared test for categorical variables (Table 1). The bacterial colony counts (in vivo and in vitro assays) were compared using the Mann–Whitney non-parametric test. We calculated univariable odds ratio (OR) to evaluate the association between MSI phenotype and high E.coli abundance and presence of pks-positive E.coli strains. The variations of MLH1 expression (Western blot and transcriptomic analysis), γH2AX or ROS induction were analysed by ANOVA followed by Tukey's post-test. In all the figures, S.D. of measures was reported.

Table 1
Clinical and microbial characteristics of CRC patients according to their tumour's microsatellite stability status
MSS (n=75)MSI (n=13)
CharacteristicsN%N%P value
Median age (years) 72 [35–89] – 77 [37–95] – 0.604 
Sex     0.072 
 Female 36 48 10 77  
 Male 39 52 23  
CRC stage     0.667 
 I–II 47 63 69  
 III–IV 28 37 31  
Presence of mucosa-associated E. coli     0.480 
 Yes 66 88 12 92  
 No 12  
Presence of mucosa-internalized E. coli     0.396 
 Yes 35 47 54  
 No 40 53 46  
Mucosa-associated E. coli abundance     0.001 
 High 29 39 69  
 Low 46 61 31  
Presence of colibactin-producing E. coli (in E.coli-positive-mucosa samples)     0.02 
 Yes 33 50 33  
 No 33 50 67  
MSS (n=75)MSI (n=13)
CharacteristicsN%N%P value
Median age (years) 72 [35–89] – 77 [37–95] – 0.604 
Sex     0.072 
 Female 36 48 10 77  
 Male 39 52 23  
CRC stage     0.667 
 I–II 47 63 69  
 III–IV 28 37 31  
Presence of mucosa-associated E. coli     0.480 
 Yes 66 88 12 92  
 No 12  
Presence of mucosa-internalized E. coli     0.396 
 Yes 35 47 54  
 No 40 53 46  
Mucosa-associated E. coli abundance     0.001 
 High 29 39 69  
 Low 46 61 31  
Presence of colibactin-producing E. coli (in E.coli-positive-mucosa samples)     0.02 
 Yes 33 50 33  
 No 33 50 67  

Results

Eighty-eight patients underwent surgery for CRC in the Department of Digestive and Hepatobiliary Surgery at the University Hospital of Clermont-Ferrand, France, between March 2007 and March 2014, and were eligible for the present study. Prospectively recorded clinicopathological features of enrolled patients are summarized in Table 1. MMR deficiency was assessed by immunohistochemical staining of tumour sections from these patients (Figure 1A). Loss of expression in one or more MMR proteins in neoplastic cells was noted in 13 patients (14.7%). Indeed, loss of MLH1, PMS2, MSH2 or MSH6 protein expression was observed in eight, seven, two or one patients, respectively. In four tumours, loss of two MMR protein expressions (MLH1 and PMS2) was observed. Therefore, all of these 13 corresponding tumours (14.7%) exhibited MSI phenotype, confirmed by multiplex PCR, whereas the other 75 tumours (85.3%) exhibited MSS phenotype with no loss of MMR protein expression (Supplementary Figure S1). Then, we evaluated the prevalence and the level of mucosa-associated and internalized E. coli in mucosa samples from this cohort. E. coli were isolated in similar proportions from both MSI and MSS CRC specimens (Table 1). However, there was a significant increase in mucosa-associated (P=0.03) and mucosa-internalized (P=0.03) E. coli colonization levels in patients exhibiting the MSI CRC phenotype compared with MSS CRC phenotype patients (Figures 1B and 1C). Based on the level of mucosa-associated E. coli, patients were divided into “high” and “low” colonized subgroups. As shown in Table 1, high abundance of mucosa-associated E. coli was significantly correlated with the MSI CRC phenotype (P<0.01; OR=3.481). Moreover, we also noted a significant increase in mucosa-associated (P=0.02) and mucosa-internalized E. coli (P=0.03) colonization levels in patients with MLH1-negative compared with MLH1-positive colonic tumours (results not shown). Thereafter, the prevalence of E. coli harbouring the colibactin-encoding pks island (colibactin-producing E. coli) was investigated using PCR tests on human CRC tissues. Colibactin-producing E. coli were more frequently identified in the MSS CRC phenotype (50%) compared with the MSI CRC phenotype (33%) (P=0.02; OR=2.03) (Table 1).

Significant association observed between the MSI tumour phenotype and the colonic mucosa colonization by E. coli in CRC

Figure 1
Significant association observed between the MSI tumour phenotype and the colonic mucosa colonization by E. coli in CRC

(A) Representative immunohistochemical staining of the human MMR nuclear proteins MLH1, PMS2, MSH2 and MSH6 in colon tumour sections. A significant increase in (B) mucosa-associated (Median 129650 range [806–1.4×106] compared with 26550 range [31–2×106] CFU/g of tissue P=0.03) and (C) mucosa-internalized (Median 2230 range [89–32200] compared with 722 range [48–27900] CFU/g of tissue; P=0.03) E. coli were observed in patients exhibiting the MSI tumour phenotype compared with those with the MSS tumour phenotype; *P<0.05.

Figure 1
Significant association observed between the MSI tumour phenotype and the colonic mucosa colonization by E. coli in CRC

(A) Representative immunohistochemical staining of the human MMR nuclear proteins MLH1, PMS2, MSH2 and MSH6 in colon tumour sections. A significant increase in (B) mucosa-associated (Median 129650 range [806–1.4×106] compared with 26550 range [31–2×106] CFU/g of tissue P=0.03) and (C) mucosa-internalized (Median 2230 range [89–32200] compared with 722 range [48–27900] CFU/g of tissue; P=0.03) E. coli were observed in patients exhibiting the MSI tumour phenotype compared with those with the MSS tumour phenotype; *P<0.05.

The difference of colonization between the two phenotypes was studied in vitro using the T-84 cell line as MMR system-competent cells and the HCT-116 cell line as MMR system-deficient cells [41]. Indeed, we performed adhesion, internalization and persistence assays for five representative colibactin-producing E. coli strains isolated from the colonic mucosa of CRC patients (CRC–E. coli). Adhesion capacities to both cell lines were similar for all the tested strains (Supplementary Figure S2), whereas a higher proportion of E. coli was able to invade (P=0.01) and persist (P=0.01) in MMR-deficient compared with MMR-competent cells (Figures 2A and 2B). The same results were obtained using two representative colibactin-negative-E.coli strains (Supplementary Figure S3). In order to confirm that these differences in terms of internalization and persistence were specifically related to the depletion of the DNA MMR system but not to other characteristics of these cell lines, these experiments were repeated on T-84 cells silenced for the MLH1 MMR protein, using siRNA treatment. We observed a significant reduction (−30%) in the MLH1 protein expression in T-84 cells treated with the MLH1 siRNA compared with scrambled siRNA-treated and -untreated cells respectively (Supplementary Figure S4). This inhibition of the MLH1 protein expression was associated with significantly higher internalization and persistence capacities of CRC–E. coli strains in siRNA MLH1-silenced T-84 cells compared with scrambled siRNA-treated and -untreated cells (Figures 2C and 2D).

The interaction between CRC–E. coli infection and epithelial cells related to their MMR status

Figure 2
The interaction between CRC–E. coli infection and epithelial cells related to their MMR status

(A) Internalization and (B) persistence assays of five CRC–E. coli strains in T84 (MMR system-competent) and HCT-116 (MMR system-deficient) intestinal epithelial cell lines were performed. Data are from at least three experiments performed twice. The internalization (A) and persistence (B) of CRC–E. coli strains were significantly higher in HCT-116 cells compared with T-84 cells. (C) Internalization and (D) persistence assays of two representative CRC–E. coli strains in T84 (MMR system-competent) and siRNA MLH1-silenced T-84 (MLH1 silencing expression) cell lines were performed. Data are from two independent experiments performed twice. The internalization (C) and the persistence (D) of CRC–E. coli strains were significantly higher in siRNA MLH1-silenced T-84 cells compared with scrambled siRNA-treated and -untreated cells. (E) Immunoblots of T84 cells infected with CRC–E. coli strains revealed a significant reduction in MLH1 protein expression after 3 h of infection compared with uninfected cells (NI) and cells infected with the non-pathogenic K12-C600 E. coli strain. Data are from at least three independent experiments. (F) MLH1 protein immunoblots on T-84 cells infected with the 11G5 colibactin-producing E. coli, its isogenic mutant (11G5Δpks) which is unable to produce colibactin, and its trans-complemented mutant (11G5Δpks + pks), are shown. No variations in MLH1 protein expression were observed in the 11G5Δpks mutant compared with uninfected cells (NI). Data are from two independent experiments performed twice; *P<0.05.

Figure 2
The interaction between CRC–E. coli infection and epithelial cells related to their MMR status

(A) Internalization and (B) persistence assays of five CRC–E. coli strains in T84 (MMR system-competent) and HCT-116 (MMR system-deficient) intestinal epithelial cell lines were performed. Data are from at least three experiments performed twice. The internalization (A) and persistence (B) of CRC–E. coli strains were significantly higher in HCT-116 cells compared with T-84 cells. (C) Internalization and (D) persistence assays of two representative CRC–E. coli strains in T84 (MMR system-competent) and siRNA MLH1-silenced T-84 (MLH1 silencing expression) cell lines were performed. Data are from two independent experiments performed twice. The internalization (C) and the persistence (D) of CRC–E. coli strains were significantly higher in siRNA MLH1-silenced T-84 cells compared with scrambled siRNA-treated and -untreated cells. (E) Immunoblots of T84 cells infected with CRC–E. coli strains revealed a significant reduction in MLH1 protein expression after 3 h of infection compared with uninfected cells (NI) and cells infected with the non-pathogenic K12-C600 E. coli strain. Data are from at least three independent experiments. (F) MLH1 protein immunoblots on T-84 cells infected with the 11G5 colibactin-producing E. coli, its isogenic mutant (11G5Δpks) which is unable to produce colibactin, and its trans-complemented mutant (11G5Δpks + pks), are shown. No variations in MLH1 protein expression were observed in the 11G5Δpks mutant compared with uninfected cells (NI). Data are from two independent experiments performed twice; *P<0.05.

We then studied the effect of the infection with colibactin-producing E. coli strains on the MLH1 MMR protein expression in T-84 cells. Immunoblots revealed a significant reduction in the MLH1 protein expression in cells after 3 h of infection compared with uninfected cells (P<0.01) or cells infected with a non-pathogenic K12-C600 E. coli strain (P<0.01) (Figure 2E). However, this effect was not transcriptional (Table 2). Thereafter, the MLH1 protein expression modulation by the colibactin toxin was assessed on T-84 cells infected with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks which is unable to produce the colibactin, and its trans-complemented mutant (11G5Δpks + pks). The inhibition of the MLH1 protein expression was dependent on the presence of the pks island in CRC strains (Figure 2F). No significant variations in the MLH1 protein expression were observed after infection with the 11G5Δpks mutant in comparison with uninfected cells (P=0.47).

Table 2
Transcriptional analysis of the four MMR genes in T-84 cells after infection

Normalized target gene expression level (ΔΔCt method) using HPRT gene as housekeeping gene control and a non-infected cell as the control sample. Data are from three experiments and means ± S.D. were reported for each conditions of infection and each MMR system genes; NI: not infected.

Gene symbolFull nameAccession numberNIK12-C60011G5
MLH1 mutL homologue 1 NM_000249 1.05±0.08 1.19±0.49 1.21±0.02 
MSH2 mutS homologue 2 NM_000251 0.99±0.01 0.96±0.22 1.30±0.11 
MSH6 mutS homologue 6 NM_000179 1.10±0.10 1.01±0.38 1.62±0.18 
PMS2 postmeiotic segregation increased 2 NM_000535 1.12±0.11 1.23±0.71 0.91±0.14 
Gene symbolFull nameAccession numberNIK12-C60011G5
MLH1 mutL homologue 1 NM_000249 1.05±0.08 1.19±0.49 1.21±0.02 
MSH2 mutS homologue 2 NM_000251 0.99±0.01 0.96±0.22 1.30±0.11 
MSH6 mutS homologue 6 NM_000179 1.10±0.10 1.01±0.38 1.62±0.18 
PMS2 postmeiotic segregation increased 2 NM_000535 1.12±0.11 1.23±0.71 0.91±0.14 

Regarding the fact that the oxidative stress was suspected to play a role in the modulation of the DNA MMR system [33,42], we studied the induction of ROS in T-84 cells infected with CRC–E. coli strains for 3 h using the H2DCF-DA fluorescent probe. The induction of ROS in infected cells was significantly increased in cells infected with CRC–E. coli strains compared with uninfected cells or cells infected with a non-pathogenic K12-C600 E. coli strain (Figure 3A). Moreover, after the infection of T-84 cells with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks and its trans-complemented mutant (11G5Δpks + pks), we showed that the ROS induction was dependent on the presence of the pks island in CRC strains (P=0.02) (Figure 3B). Beyond the induction of ROS, we studied the histone H2AX subunit phosphorylation (γH2AX), which is a marker of DNA double-strand breaks, using immunofluorescence on T-84 cells infected with CRC–E. coli strains. We observed a marked nuclear staining in cells infected with CRC–E. coli strains compared with uninfected cells or cells infected with a non-pathogenic K12-C600 E. coli strain (Figure 3C, Supplementary Figure S5). The number of cells presenting a nuclear γH2AX staining was significantly increased in cells infected with CRC–E. coli strains compared with uninfected cells (P<0.001) or cells infected with a non-pathogenic K12-C600 E. coli strain (P<0.001) (Figure 3D). Moreover, after the infection of T-84 cells with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks and its trans-complemented mutant (11G5Δpks + pks), we showed that this increase in the γH2AX nuclear staining was dependent on the presence of the pks island in CRC strains (P<0.001) (Figure 3E).

Induction of oxidative stress and DNA damage after infection by pathogenic CRC–E. coli strains

Figure 3
Induction of oxidative stress and DNA damage after infection by pathogenic CRC–E. coli strains

(A) Using the H2DCF-DA fluorescent probe we showed that the induction of ROS in infected cells was significantly increased in cells infected with CRC–E. coli strains compared with uninfected cells (NI) (P<0.05) or cells infected with a non-pathogenic K12-C600 E. coli strain (P<0.05). For each condition, data are from the mean of the nine measures of each well performed twice. Three independent experiments were performed. (B) After infection of T84 cells infected with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks and its trans-complemented mutant 11G5Δpks + pks, we showed that ROS induction was dependent on the presence of the pks island in CRC strains (P=0.02 for 11G5 compared with 11G5Δpks and P=0.01 for 11G5Δpks compared with 11G5Δpks + pks). For each condition, data are from the mean of the nine measures of each well performed twice. Three independent experiments were performed. (CE) Histone H2AX subunit phosphorylation (γH2AX) immunostaining on T-84 cells infected with CRC–E. coli strains. (C) Representative γH2AX immunostaining showed a marked nuclear staining in cells infected with 11G5 CRC–E. coli strain compared with uninfected cells (NI) or cells infected with a non-pathogenic K12-C600 E. coli strain. (D) The number of cells presenting γH2AX staining was significantly increased in cells infected with CRC–E. coli strains compared with uninfected cells (NI) or cells infected with a non-pathogenic K12-C600 E. coli strain (P<0.001). Data are from two independent experiments performed in triplicate. (E) After infection of T84 cells infected with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks and its trans-complemented mutant 11G5Δpks + pks, we showed that this increase in γH2AX nuclear staining was dependent on the presence of the pks island in CRC strains (P<0.001). Data are from two independent experiments performed in triplicate; *P<0.05.

Figure 3
Induction of oxidative stress and DNA damage after infection by pathogenic CRC–E. coli strains

(A) Using the H2DCF-DA fluorescent probe we showed that the induction of ROS in infected cells was significantly increased in cells infected with CRC–E. coli strains compared with uninfected cells (NI) (P<0.05) or cells infected with a non-pathogenic K12-C600 E. coli strain (P<0.05). For each condition, data are from the mean of the nine measures of each well performed twice. Three independent experiments were performed. (B) After infection of T84 cells infected with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks and its trans-complemented mutant 11G5Δpks + pks, we showed that ROS induction was dependent on the presence of the pks island in CRC strains (P=0.02 for 11G5 compared with 11G5Δpks and P=0.01 for 11G5Δpks compared with 11G5Δpks + pks). For each condition, data are from the mean of the nine measures of each well performed twice. Three independent experiments were performed. (CE) Histone H2AX subunit phosphorylation (γH2AX) immunostaining on T-84 cells infected with CRC–E. coli strains. (C) Representative γH2AX immunostaining showed a marked nuclear staining in cells infected with 11G5 CRC–E. coli strain compared with uninfected cells (NI) or cells infected with a non-pathogenic K12-C600 E. coli strain. (D) The number of cells presenting γH2AX staining was significantly increased in cells infected with CRC–E. coli strains compared with uninfected cells (NI) or cells infected with a non-pathogenic K12-C600 E. coli strain (P<0.001). Data are from two independent experiments performed in triplicate. (E) After infection of T84 cells infected with the colibactin-producing E. coli 11G5 strain, its isogenic mutant 11G5Δpks and its trans-complemented mutant 11G5Δpks + pks, we showed that this increase in γH2AX nuclear staining was dependent on the presence of the pks island in CRC strains (P<0.001). Data are from two independent experiments performed in triplicate; *P<0.05.

To confirm the role of the oxidative stress in bacterial-induced MMR system modulations, we then studied the effect of an antioxidant pretreatment of the T-84 cells before the infection with bacteria. Immunoblots revealed that the MLH1 protein expression was restored in infected cells after an antioxidant pretreatment using α-LA, at similar levels to those of uninfected cells (Figure 4A). Moreover, we observed a significant increase in the MLH1 protein expression in infected cells after the antioxidant pretreatment compared with untreated infected cells (P=0.03) (Figures 4A and 4B). This antioxidant pretreatment using α-LA was also associated with a significant decrease in the ROS induction (P=0.03) and in the number of cells presenting with a nuclear γH2AX staining (P=0.003) compared with untreated infected cells (Figure 4C–4E).

Effect of an antioxidant pretreatment with α-LA on ROS induction, DNA damages and MLH1 expression in cells infected with pathogenic colibactin-producing-E.coli strains

Figure 4
Effect of an antioxidant pretreatment with α-LA on ROS induction, DNA damages and MLH1 expression in cells infected with pathogenic colibactin-producing-E.coli strains

(A) Immunoblots of T84 cells infected with colibactin-producing-E. coli strains (11G5) revealed that MLH1 protein expression was similar in uninfected cells (NI) and in infected cells pretreated with α-LA (50 μg/ml). (B) A significant increase in MLH1 protein expression was observed in infected cells after the antioxidant α-LA pretreatment compared with untreated infected cells (P=0.03). Data are from immunoblots performed in two independent experiments. (C) Using the H2DCF-DA fluorescent probe, we showed that ROS induction was significantly reduced in infected cells after α-LA treatment (P=0.03). For each condition, data are from the mean of the nine measures of each well performed in triplicate. Two independent experiments were performed. (D) Representative γH2AX immunostaining after 11G5 strains infection with or without α-LA pretreatment. (E) The number of cells presenting γH2AX nuclear staining was significantly decreased in cells infected with CRC–E. coli strains after α-LA pretreatment, compared with untreated infected cells (P=0.003). Data are from three experiments; *P<0.05.

Figure 4
Effect of an antioxidant pretreatment with α-LA on ROS induction, DNA damages and MLH1 expression in cells infected with pathogenic colibactin-producing-E.coli strains

(A) Immunoblots of T84 cells infected with colibactin-producing-E. coli strains (11G5) revealed that MLH1 protein expression was similar in uninfected cells (NI) and in infected cells pretreated with α-LA (50 μg/ml). (B) A significant increase in MLH1 protein expression was observed in infected cells after the antioxidant α-LA pretreatment compared with untreated infected cells (P=0.03). Data are from immunoblots performed in two independent experiments. (C) Using the H2DCF-DA fluorescent probe, we showed that ROS induction was significantly reduced in infected cells after α-LA treatment (P=0.03). For each condition, data are from the mean of the nine measures of each well performed in triplicate. Two independent experiments were performed. (D) Representative γH2AX immunostaining after 11G5 strains infection with or without α-LA pretreatment. (E) The number of cells presenting γH2AX nuclear staining was significantly decreased in cells infected with CRC–E. coli strains after α-LA pretreatment, compared with untreated infected cells (P=0.003). Data are from three experiments; *P<0.05.

Discussion

A possible role for gut microbiota in colorectal carcinogenesis is increasingly obvious [411,20]. The study presents the first clinical evidence for an interaction between E. coli and MMR deficiency in CRC, which was previously described in vitro [21,33]. In the present study, we observed a rapid down-regulation of the MLH1 protein expression in T-84 cells infected with colibactin-producing-E. coli strains, which was dependent on the colibactin expression (Figure 2). We found that this regulation resulted from a post-transcriptional effect and seemed to be dependent on the oxidative stress induction. These results are consistent with previously published studies [42,43]. Indeed, ROS-induced oxidative stress is known to inhibit the DNA MMR repair system involved in colorectal carcinogenesis [42]. Then, when altered in an MMR-deficient CRC mice model, the DNA MMR system is unable to suppress the tumorigenesis induced by the oxidative stress [43]. In addition, a down-regulation of MMR pathways was also reported after infection with other E. coli strains, such as EPEC, through the post-transcriptional effect of the bacterial-secreted effector protein EspF and by ROS induction [21,33]. In the present study, we demonstrated that colibactin-producing E. coli strains induced ROS, at levels consistent with those reported by Maddocks et al. [33], and thus DNA double-strand breaks that probably strongly participate in the genetic instability reported in CRC. In addition, this ROS induction could be involved in the inhibition of the MLH1 protein expression observed after bacterial infection. Furthermore, this bacterial-induced MLH1 protein depletion was significantly reduced after an antioxidant pretreatment.

Even if further in vivo studies are needed to better explore the mechanisms of the interactions between bacteria and MMR system alterations, our results showed a new pro-carcinogenic property of colibactin-producing E. coli strains by interfering with the MMR pathways. In addition to the well-known genotoxicity of the colibactin [27,28], colibactin-producing E. coli could be a “driver” bacteria of colorectal carcinogenesis through MLH1 protein silencing, leading to the deregulation of the MMR pathways, the promotion of an incomplete DNA repair and thus to a genomic instability involved in CRC. Moreover, we observed in vitro that inhibition of MLH1 induces an increase in the bacterial persistence in colonic epithelial cells (cf. Figures 2C and 2D) which could amplify the genomic instability. However, in human samples, we observed a lower prevalence of pathogenic colibactin-producing E. coli in MSI than in MSS phenotype CRC tissues. Even if these genotoxic bacteria were less prevalent at the later stages of the disease, colibactin-producing E. coli strains could interfere with the MLH1 protein expression and thus play a role in early steps of carcinogenesis in the MSI CRC phenotype (Figure 5). Indeed, it has been suggested that MLH1 protein silencing is likely to be a relatively early and major event in carcinogenesis [44], and colibactin-producing E. coli strains were recently detected in adenomas [10], suggesting that these bacteria could act early in colorectal carcinogenesis. Then, this MMR system deregulation could select a tumour microenvironment that could favour a high level of colonization by other E. coli strains. This hypothesis is supported by our observations of an in vitro increased persistence of E.coli strains in MMR-deficient cell lines, and of a higher in vivo colonic E. coli colonization in the MSI CRC phenotype. These bacteria could thus be “passenger” bacteria of carcinogenesis, which are involved in tumour promotion and progression, especially throughout inflammatory processes such as cyclooxygenase 2 (COX2) induction, which was described as a colibactin-independent mechanism (Figure 5) [15,45]. In the same way, independent studies reported a high level of colonization by other “passenger” bacteria in the MSI CRC phenotype, such as F. nucleatum [11,18,46,47]. Mima et al. [46] described an association between the mucosal colonization by F. nucleatum and both the local tumour environment (T-cell density) and the MMR system deficiency, suggesting a potential involvement of this “passenger” bacterium in the promotion steps of CRC [48]. Moreover, Wei et al. [47] found that F. nucleatum and B. fragilis were associated with a loss of MLH1 protein expression. Therefore, there is now evidence that dysbiosis is probably strongly different between MSS and MSI CRC phenotypes, also involving E. coli species, as we described. Contrary to the MSI CRC phenotype in which a large number of genetic alterations have ever been accumulated [49], the carcinogenic pressure leading to genetic instability should be maintained in MSS tumour carcinogenesis, possibly through the effect of “driver-bacteria” and its genotoxic agent, such as the pro-carcinogenic colibactin toxin. Indeed, despite the observation of lower levels of E. coli in the gut mucosa of MSS CRC phenotype patients, we noted the persistence of colibactin-producing E. coli. Given the pro-carcinogenic properties of these E. coli genotoxic strains, a lower level of colonization appears sufficient to trigger carcinogenesis [30]. In addition, in the first step of MSS phenotype carcinogenesis, colibactin-producing E.coli could also deplete MMR pathways in mucosa. This inhibition could be transient in vivo and could participate to carcinogenesis. Indeed, the unrepair of mismatched DNA base pairs in MMR-deficient enterocytes can result in somatic mutations leading to genetic alterations [50]. Moreover, Li et al. [50] recently suggested that depletion of MLH1 or MSH2 protein expression could contribute to CRC progression through reduced rates of apoptosis. In addition, colibactin-producing E.coli could enhance tumour growth by the induction of the emergence of senescent epithelial cells secreting growth factor [30], and/or by the secretion of pro-tumoural molecules by infiltrating cells [13,51].

Proposed model for the contribution of colibactin-producing bacteria in CRC according to microsatellite status tumour phenotype

Figure 5
Proposed model for the contribution of colibactin-producing bacteria in CRC according to microsatellite status tumour phenotype

(A) In the MSS CRC phenotype, colibactin-producing E. coli could be a “driver” bacteria of the carcinogenesis that exerts its pro-carcinogenic properties from the early steps of carcinogenesis to late stage CRC where these strains persist. Colibactin could play a role in genomic instability of the mucosa via direct induction of DNA double-strand breaks [27] but also by a transient depletion of the MMR pathways leading to the accumulation of mismatched DNA base pairs, resulting in somatic mutations. Then, these pro-carcinogenic strains persist during all carcinogenesis steps probably to maintain genomic instability. (B) In the MSI CRC phenotype, colibactin-producing E. coli could be a “driver” bacteria at the early stages of the carcinogenesis process through MMR defect in addition to the well-known colibactin-induced genotoxicity. This bacterial-induced MSI persists afterwards at the latest stages of the disease and could favour colonization by E. coli strains that do not harbour the pks island but that could be a “passenger” bacteria during the carcinogenesis process. These “passenger” strains could then exert colibactin-independent pro-carcinogenic effects involved in tumour promotion and progression, especially through colibactin-independent E. coli pro-carcinogenic effects such as inflammatory processes with COX2 induction and local inflammatory cell infiltration [45]. Other “passenger” bacteria such as Fusobacterium nucleatum could also be implicated in these late steps [46,47,48].

Figure 5
Proposed model for the contribution of colibactin-producing bacteria in CRC according to microsatellite status tumour phenotype

(A) In the MSS CRC phenotype, colibactin-producing E. coli could be a “driver” bacteria of the carcinogenesis that exerts its pro-carcinogenic properties from the early steps of carcinogenesis to late stage CRC where these strains persist. Colibactin could play a role in genomic instability of the mucosa via direct induction of DNA double-strand breaks [27] but also by a transient depletion of the MMR pathways leading to the accumulation of mismatched DNA base pairs, resulting in somatic mutations. Then, these pro-carcinogenic strains persist during all carcinogenesis steps probably to maintain genomic instability. (B) In the MSI CRC phenotype, colibactin-producing E. coli could be a “driver” bacteria at the early stages of the carcinogenesis process through MMR defect in addition to the well-known colibactin-induced genotoxicity. This bacterial-induced MSI persists afterwards at the latest stages of the disease and could favour colonization by E. coli strains that do not harbour the pks island but that could be a “passenger” bacteria during the carcinogenesis process. These “passenger” strains could then exert colibactin-independent pro-carcinogenic effects involved in tumour promotion and progression, especially through colibactin-independent E. coli pro-carcinogenic effects such as inflammatory processes with COX2 induction and local inflammatory cell infiltration [45]. Other “passenger” bacteria such as Fusobacterium nucleatum could also be implicated in these late steps [46,47,48].

In conclusion, the ability of CRC–pathogenic E. coli strains to trigger carcinogenesis is now well supported. Considering the fact that dysbiosis is probably different according to the microsatellite stability status, further host–pathogen interactions studies in CRC should take into account tumour phenotypes.

Author contribution

Johan Gagnière and Mathilde Bonnet were involved in data acquisition, statistical analysis, interpretation and writing of the manuscript. Virginie Bonnin, Emilie Cardamone, Anne-Sophie Jarrousse, Allison Agus, Nancy Uhrhammer, Pierre Sauvanet, Pierre Déchelotte and Richard Bonnet were involved in data acquisition. Nicolas Barnich and Richard Bonnet critically revised the manuscript. Johan Gagniére and Denis Pezet were involved in clinical data acquisition. Mathilde Bonnet designed the study. Mathilde Bonnet and Denis Pezet supervised the study.

We thank the CICS platforms from the Université d'Auvergne and Emmanuel Bourgeois from the CHU of Clermont-Ferrand for their technical assistance.

Funding

This work was supported by the Ministère de la Recherche et de la Technologie, Inserm and Université d'Auvergne [grant number UMR1071]; INRA [grant number USC-2018]; the Conseil Régional d'Auvergne and the « Robert Debré pour la recherche médicale » association; “Nuovo Soldati Foundation for Cancer Research”; and an “Inserm Région” [grant number 358013 (to A.A.)].

Abbreviations

    Abbreviations
     
  • APC

    adenomatous polyposis coli

  •  
  • AOM

    Azoxymethane

  •  
  • CAC

    chemically-induced colitis-associated colorectal cancer

  •  
  • CFU

    colony-forming unit

  •  
  • COX2

    cyclooxygenase 2

  •  
  • CRC

    colorectal cancer

  •  
  • EspF

    E. coli secreted-protein F

  •  
  • EPEC

    enteropathogenic Escherichia coli

  •  
  • FFPE

    formalin-fixed, paraffin-embedded

  •  
  • GAPDH

    glyceraldehyde-3-phosphate dehydrogenase

  •  
  • H2DCF-DA

    2′,7′-dichlorodihydrofluorescein diacetate

  •  
  • HNPCC

    hereditary non-polyposis CRC

  •  
  • HPRT

    hypoxanthine-guanine phosphoribosyl-transferase

  •  
  • α-LA

    α-lipoic acid

  •  
  • MLH1

    mutL homologue 1

  •  
  • MMR

    mismatch repair

  •  
  • MOI

    multiplicity of infection

  •  
  • MSH2

    mutS homologue 2

  •  
  • MSH6

    mutS homologue 6

  •  
  • MSI

    microsatellite instability

  •  
  • MSS

    microsatellite stable

  •  
  • OR

    odds ratio

  •  
  • pks

    polyketide synthase

  •  
  • PMS2

    postmeiotic segregation 2

  •  
  • ROS

    reactive oxygen species

  •  
  • RT

    room temperature

References

References
1
Fearon
,
E.R.
,
Vogelstein
,
B.
(
1990
)
A genetic model for colorectal tumorigenesis
.
Cell
61
,
759
767
2
Pritchard
,
C.C.
,
Grady
,
W.M.
(
2011
)
Colorectal cancer molecular biology moves into clinical practice
.
Gut
60
,
116
129
3
Sohn
,
K.J.
,
Choi
,
M.
,
Song
,
J.
,
Chan
,
S.
,
Medline
,
A.
,
Gallinger
,
S.
,
Kim
,
Y.I.
(
2003
)
Msh2 deficiency enhances somatic Apc and p53 mutations in Apc+/-Msh2-/- mice
.
Carcinogenesis
24
,
217
224
4
Ahn
,
J.
,
Sinha
,
R.
,
Pei
,
Z.
,
Dominianni
,
C.
,
Wu
,
J.
,
Shi
,
J.
,
Goedert
,
J.J.
,
Hayes
,
R.B.
,
Yang
,
L.
(
2013
)
Human gut microbiome and risk for colorectal cancer
.
J. Natl. Cancer Inst.
105
,
1907
1911
5
Chen
,
W.
,
Liu
,
F.
,
Ling
,
Z.
,
Tong
,
X.
,
Xiang
,
C.
(
2012
)
Human intestinal lumen and mucosa-associated microbiota in patients with colorectal cancer
.
PloS One
7
,
e39743
6
Sanapareddy
,
N.
,
Legge
,
R.M.
,
Jovov
,
B.
,
McCoy
,
A.
,
Burcal
,
L.
,
Araujo-Perez
,
F.
,
Randall
,
T.A.
,
Galanko
,
J.
,
Benson
,
A.
,
Sandler
,
R.S.
et al (
2012
)
Increased rectal microbial richness is associated with the presence of colorectal adenomas in humans
.
ISME J.
6
,
1858
1868
7
Sobhani
,
I.
,
Tap
,
J.
,
Roudot-Thoraval
,
F.
,
Roperch
,
J.P.
,
Letulle
,
S.
,
Langella
,
P.
,
Corthier
,
G.
,
Tran Van Nhieu
,
J.
,
Furet
,
J.P.
(
2011
)
Microbial dysbiosis in colorectal cancer (CRC) patients
.
PLoS One
6
,
e16393
8
Wang
,
T.
,
Cai
,
G.
,
Qiu
,
Y.
,
Fei
,
N.
,
Zhang
,
M.
,
Pang
,
X.
,
Jia
,
W.
,
Cai
,
S.
,
Zhao
,
L.
(
2012
)
Structural segregation of gut microbiota between colorectal cancer patients and healthy volunteers
.
ISME J.
6
,
320
329
9
Wu
,
N.
,
Yang
,
X.
,
Zhang
,
R.
,
Li
,
J.
,
Xiao
,
X.
,
Hu
,
Y.
,
Chen
,
Y.
,
Yang
,
F.
,
Lu
,
N.
,
Wang
,
Z.
et al (
2013
)
Dysbiosis signature of fecal microbiota in colorectal cancer patients
.
Microb. Ecol.
66
,
462
470
10
Nakatsu
,
G.
,
Li
,
X.
,
Zhou
,
H.
,
Sheng
,
J.
,
Wong
,
S.H.
,
Wu
,
W.K.
,
Ng
,
S.C.
,
Tsoi
,
H.
,
Dong
,
Y.
,
Zhang
,
N.
et al (
2015
)
Gut mucosal microbiome across stages of colorectal carcinogenesis
.
Nat. Commun.
6
,
8727
11
Viljoen
,
K.S.
,
Dakshinamurthy
,
A.
,
Goldberg
,
P.
,
Blackburn
,
J.M.
(
2015
)
Quantitative profiling of colorectal cancer-associated bacteria reveals associations between Fusobacterium spp., Enterotoxigenic Bacteroides fragilis (ETBF) and clinicopathological features of colorectal cancer
.
PloS One
10
,
e0119462
12
Goedert
,
J.J.
,
Gong
,
Y.
,
Hua
,
X.
,
Zhong
,
H.
,
He
,
Y.
,
Peng
,
P.
,
Yu
,
G.
,
Wang
,
W.
,
Ravel
,
J.
,
Shi
,
J.
,
Zheng
,
Y.
(
2015
)
Fecal microbiota characteristics of patients with colorectal adenoma detected by screening: A population-based study
.
EBioMedicine
2
,
597
603
13
Schwabe
,
R.F.
,
Jobin
,
C.
(
2013
)
The microbiome and cancer
.
Nat. Rev. Cancer
13
,
800
812
14
Marchesi
,
J.R.
,
Adams
,
D.H.
,
Fava
,
F.
,
Hermes
,
G.D.
,
Hirschfield
,
G.M.
,
Hold
,
G.
,
Quraishi
,
M.N.
,
Kinross
,
J.
,
Smidt
,
H.
,
Tuohy
,
K.M.
et al (
2015
)
The gut microbiota and host health: a new clinical frontier
.
Gut
65
,
330
339
15
Tjalsma
,
H.
,
Boleij
,
A.
,
Marchesi
,
J.R.
,
Dutilh
,
B.E.
(
2012
)
A bacterial driver-passenger model for colorectal cancer: beyond the usual suspects
.
Nat. Rev. Microbiol.
10
,
575
582
16
Arthur
,
J.C.
,
Perez-Chanona
,
E.
,
Muhlbauer
,
M.
,
Tomkovich
,
S.
,
Uronis
,
J.M.
,
Fan
,
T.J.
,
Campbell
,
B.J.
,
Abujamel
,
T.
,
Dogan
,
B.
,
Rogers
,
A.B.
et al (
2012
)
Intestinal inflammation targets cancer-inducing activity of the microbiota
.
Science
338
,
120
123
17
Wu
,
S.
,
Rhee
,
K.J.
,
Albesiano
,
E.
,
Rabizadeh
,
S.
,
Wu
,
X.
,
Yen
,
H.R.
,
Huso
,
D.L.
,
Brancati
,
F.L.
,
Wick
,
E.
,
McAllister
,
F.
et al (
2009
)
A human colonic commensal promotes colon tumorigenesis via activation of T helper type 17 T cell responses
.
Nat. Med.
15
,
1016
1022
18
Mima
,
K.
,
Nishihara
,
R.
,
Qian
,
Z.R.
,
Cao
,
Y.
,
Sukawa
,
Y.
,
Nowak
,
J.A.
,
Yang
,
J.
,
Dou
,
R.
,
Masugi
,
Y.
,
Song
,
M.
et al (
2015
)
Fusobacterium nucleatum in colorectal carcinoma tissue and patient prognosis
.
Gut
in the press
19
Arthur
,
J.C.
,
Gharaibeh
,
R.Z.
,
Muhlbauer
,
M.
,
Perez-Chanona
,
E.
,
Uronis
,
J.M.
,
McCafferty
,
J.
,
Fodor
,
A.A.
,
Jobin
,
C.
(
2014
)
Microbial genomic analysis reveals the essential role of inflammation in bacteria-induced colorectal cancer
.
Nat. Commun.
5
,
4724
20
Gagniere
,
J.
,
Raisch
,
J.
,
Veziant
,
J.
,
Barnich
,
N.
,
Bonnet
,
R.
,
Buc
,
E.
,
Bringer
,
M.A.
,
Pezet
,
D.
,
Bonnet
,
M.
(
2016
)
Gut microbiota imbalance and colorectal cancer
.
World J. Gastroenterol.
22
,
501
518
21
Maddocks
,
O.D.
,
Short
,
A.J.
,
Donnenberg
,
M.S.
,
Bader
,
S.
,
Harrison
,
D.J.
(
2009
)
Attaching and effacing Escherichia coli downregulate DNA mismatch repair protein in vitro and are associated with colorectal adenocarcinomas in humans
.
PLoS One
4
,
e5517
22
Martins
,
S.A.
,
Prazeres
,
D.M.
,
Cabral
,
J.M.
,
Monteiro
,
G.A.
(
2003
)
Comparison of real-time polymerase chain reaction and hybridization assays for the detection of Escherichia coli genomic DNA in process samples and pharmaceutical-grade plasmid DNA products
.
Anal. Biochem.
322
,
127
129
23
Swidsinski
,
A.
,
Khilkin
,
M.
,
Kerjaschki
,
D.
,
Schreiber
,
S.
,
Ortner
,
M.
,
Weber
,
J.
,
Lochs
,
H.
(
1998
)
Association between intraepithelial Escherichia coli and colorectal cancer
.
Gastroenterology
115
,
281
286
24
Bonnet
,
M.
,
Buc
,
E.
,
Sauvanet
,
P.
,
Darcha
,
C.
,
Dubois
,
D.
,
Pereira
,
B.
,
Déchelotte
,
P.
,
Bonnet
,
R.
,
Pezet
,
D.
,
Darfeuille-Michaud
,
A.
(
2014
)
Colonization of the human gut by E. coli and colorectal cancer risk
.
Clin. Cancer Res.
20
,
859
867
25
Buc
,
E.
,
Dubois
,
D.
,
Sauvanet
,
P.
,
Raisch
,
J.
,
Delmas
,
J.
,
Darfeuille-Michaud
,
A.
,
Pezet
,
D.
,
Bonnet
,
R.
(
2013
)
High prevalence of mucosa-associated E. coli producing cyclomodulin and genotoxin in colon cancer
.
PloS One
8
,
e56964
26
Prorok-Hamon
,
M.
,
Friswell
,
M.K.
,
Alswied
,
A.
,
Roberts
,
C.L.
,
Song
,
F.
,
Flanagan
,
P.K.
,
Knight
,
P.
,
Codling
,
C.
,
Marchesi
,
J.R.
,
Winstanley
,
C.
et al (
2014
)
Colonic mucosa-associated diffusely adherent afaC+ Escherichia coli expressing lpfA and pks are increased in inflammatory bowel disease and colon cancer
.
Gut
63
,
761
770
27
Cuevas-Ramos
,
G.
,
Petit
,
C.R.
,
Marcq
,
I.
,
Boury
,
M.
,
Oswald
,
E.
,
Nougayrede
,
J.P.
(
2010
)
Escherichia coli induces DNA damage in vivo and triggers genomic instability in mammalian cells
.
Proc. Natl. Acad. Sci. U.S.A.
107
,
11537
11542
28
Nougayrede
,
J.P.
,
Homburg
,
S.
,
Taieb
,
F.
,
Boury
,
M.
,
Brzuszkiewicz
,
E.
,
Gottschalk
,
G.
,
Buchrieser
,
C.
,
Hacker
,
J.
,
Dobrindt
,
U.
,
Oswald
,
E.
(
2006
)
Escherichia coli induces DNA double-strand breaks in eukaryotic cells
.
Science
313
,
848
851
29
Vizcaino
,
M.I.
,
Crawford
,
J.M.
(
2015
)
The colibactin warhead crosslinks DNA
.
Nat. Chem.
7
,
411
417
30
Cougnoux
,
A.
,
Dalmasso
,
G.
,
Martinez
,
R.
,
Buc
,
E.
,
Delmas
,
J.
,
Gibold
,
L.
,
Sauvanet
,
P.
,
Darcha
,
C.
,
Déchelotte
,
P.
,
Bonnet
,
M.
et al (
2014
)
Bacterial genotoxin colibactin promotes colon tumour growth by inducing a senescence-associated secretory phenotype
.
Gut
63
,
1932
1942
31
Raisch
,
J.
,
Buc
,
E.
,
Bonnet
,
M.
,
Sauvanet
,
P.
,
Vazeille
,
E.
,
de Vallee
,
A.
,
Déchelotte
,
P.
,
Darcha
,
C.
,
Pezet
,
D.
,
Bonnet
,
R.
et al (
2014
)
Colon cancer-associated B2 Escherichia coli colonize gut mucosa and promote cell proliferation
.
World J. Gastroenterol.
20
,
6560
6572
32
Levine
,
M.M.
,
Edelman
,
R.
(
1984
)
Enteropathogenic Escherichia coli of classic serotypes associated with infant diarrhea: epidemiology and pathogenesis
.
Epidemiol. Rev.
6
,
31
51
33
Maddocks
,
O.D.
,
Scanlon
,
K.M.
,
Donnenberg
,
M.S.
(
2013
)
An Escherichia coli effector protein promotes host mutation via depletion of DNA mismatch repair proteins
.
mBio
4
,
e00152
e00153
34
Belcheva
,
A.
,
Irrazabal
,
T.
,
Robertson
,
S.J.
,
Streutker
,
C.
,
Maughan
,
H.
,
Rubino
,
S.
,
Moriyama
,
E.H.
,
Copeland
,
J.K.
,
Kumar
,
S.
,
Green
,
B.
et al (
2014
)
Gut microbial metabolism drives transformation of msh2-deficient colon epithelial cells
.
Cell
158
,
288
299
35
Umar
,
A.
,
Boland
,
C.R.
,
Terdiman
,
J.P.
,
Syngal
,
S.
,
de la Chapelle
,
A.
,
Ruschoff
,
J.
,
Fishel
,
R.
,
Lindor
,
N.M.
,
Burgart
,
L.J.
,
Hamelin
,
R.
et al (
2004
)
Revised Bethesda Guidelines for hereditary nonpolyposis colorectal cancer (Lynch syndrome) and microsatellite instability
.
J. Natl. Cancer Inst.
96
,
261
268
36
Johnson
,
J.R.
,
Johnston
,
B.
,
Kuskowski
,
M.A.
,
Nougayrede
,
J.P.
,
Oswald
,
E.
(
2008
)
Molecular epidemiology and phylogenetic distribution of the Escherichia coli pks genomic island
.
J. Clin. Microbiol.
46
,
3906
3911
37
Dubois
,
D.
,
Delmas
,
J.
,
Cady
,
A.
,
Robin
,
F.
,
Sivignon
,
A.
,
Oswald
,
E.
,
Bonnet
,
R.
(
2010
)
Cyclomodulins in urosepsis strains of Escherichia coli
.
J. Clin. Microbiol.
48
,
2122
2129
38
Datsenko
,
K.A.
,
Wanner
,
B.L.
(
2000
)
One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products
.
Proc. Natl. Acad. Sci. U.S.A.
97
,
6640
6645
39
Chaveroche
,
M.K.
,
Ghigo
,
J.M.
,
d'Enfert
,
C.
(
2000
)
A rapid method for efficient gene replacement in the filamentous fungus Aspergillus nidulans
.
Nucleic Acids Res.
28
,
E97
40
Darfeuille-Michaud
,
A.
,
Boudeau
,
J.
,
Bulois
,
P.
,
Neut
,
C.
,
Glasser
,
A.L.
,
Barnich
,
N.
,
Bringer
,
M.A.
,
Swidsinski
,
A.
,
Beaugerie
,
L.
,
Colombel
,
J.F.
(
2004
)
High prevalence of adherent-invasive Escherichia coli associated with ileal mucosa in Crohn's disease
.
Gastroenterology
127
,
412
421
41
Knutsen
,
T.
,
Padilla-Nash
,
H.M.
,
Wangsa
,
D.
,
Barenboim-Stapleton
,
L.
,
Camps
,
J.
,
McNeil
,
N.
,
Difilippantonio
,
M.J.
,
Ried
,
T.
(
2010
)
Definitive molecular cytogenetic characterization of 15 colorectal cancer cell lines
.
Genes Chromosomes Cancer
49
,
204
223
42
Chang
,
C.L.
,
Marra
,
G.
,
Chauhan
,
D.P.
,
Ha
,
H.T.
,
Chang
,
D.K.
,
Ricciardiello
,
L.
,
Randolph
,
A.
,
Carethers
,
J.M.
,
Boland
,
C.R.
(
2002
)
Oxidative stress inactivates the human DNA mismatch repair system
.
Am. J. Physiol. Cell Physiol.
283
,
C148
C154
43
Piao
,
J.
,
Nakatsu
,
Y.
,
Ohno
,
M.
,
Taguchi
,
K.
,
Tsuzuki
,
T.
(
2013
)
Mismatch repair deficient mice show susceptibility to oxidative stress-induced intestinal carcinogenesis
.
Int. J. Biol. Sci.
10
,
73
79
44
Giovannucci
,
E.
,
Ogino
,
S.
(
2005
)
DNA methylation, field effects, and colorectal cancer
.
J. Natl. Cancer Inst.
97
,
1317
1319
45
Raisch
,
J.
,
Rolhion
,
N.
,
Dubois
,
A.
,
Darfeuille-Michaud
,
A.
,
Bringer
,
M.A.
(
2015
)
Intracellular colon cancer-associated Escherichia coli promote protumoral activities of human macrophages by inducing sustained COX-2 expression
.
Lab. Investig.
95
,
296
307
46
Mima
,
K.
,
Sukawa
,
Y.
,
Nishihara
,
R.
,
Qian
,
Z.R.
,
Yamauchi
,
M.
,
Inamura
,
K.
,
Kim
,
S.A.
,
Masuda
,
A.
,
Nowak
,
J.A.
,
Nosho
,
K.
et al (
2015
)
Fusobacterium nucleatum and T Cells in Colorectal Carcinoma
.
JAMA Oncol.
1
,
653
661
47
Wei
,
Z.
,
Cao
,
S.
,
Liu
,
S.
,
Yao
,
Z.
,
Sun
,
T.
,
Li
,
Y.
,
Li
,
J.
,
Zhang
,
D.
,
Zhou
,
Y.
(
2016
)
Could gut microbiota serve as prognostic biomarker associated with colorectal cancer patients' survival? A pilot study on relevant mechanism
.
Oncotarget
19
,
46158
46172
48
Park
,
J.H.
,
Powell
,
A.G.
,
Roxburgh
,
C.S.
,
Horgan
,
P.G.
,
McMillan
,
D.C.
,
Edwards
,
J.
(
2016
)
Mismatch repair status in patients with primary operable colorectal cancer: associations with the local and systemic tumour environment
.
Br. J. Cancer
114
,
562
570
49
Vogelstein
,
B.
,
Papadopoulos
,
N.
,
Velculescu
,
V.E.
,
Zhou
,
S.
,
Diaz
,
L.A.
Jr
,
Kinzler
,
K.W.
(
2013
)
Cancer genome landscapes
.
Science
339
,
1546
1558
50
Li
,
S.K.
,
Martin
,
A.
(
2016
)
Mismatch repair and colon cancer: mechanisms and therapies explored
.
Trends Mol. Med.
22
,
274
289
51
Arthur
,
J.C.
,
Jobin
,
C.
(
2013
)
The complex interplay between inflammation, the microbiota and colorectal cancer
.
Gut Microbes
4
,
253
258

Supplementary data