Colibactin-positive
Escherichia coli induce a procarcinogenic
immune environment leading to immunotherapy resistance
in colorectal cancer
Amélie Lopès1,2, Elisabeth Billard1, Al Hassan Casse3, Romain Villéger1, Julie Veziant1,4, Gwenaëlle Roche1,
Guillaume Carrier1,4, Pierre Sauvanet1,4, Arnaud Briat5, Franck Pagès6,7,8,9, Souad Naimi3, Denis Pezet1,4, Nicolas Barnich1, Bruno Dumas2and Mathilde Bonnet 1
1UMR1071 Inserm/Université Clermont Auvergne; USC-INRA 2018, Microbes, Intestin, Inflammation et Susceptibilité de l’Hôte (M2iSH), Clermont-Ferrand, France
2Biologics Research, Sanofi R&D, Vitry-Sur-Seine, France
3Histopathology and Bio-Imaging Group, Sanofi R&D, Vitry-Sur-Seine, France
4Service de Chirurgie Digestive, CHU Clermont-Ferrand, INSERM, Université Clermont Auvergne, Clermont-Ferrand, France 5UMR1240 Inserm/Université Clermont Auvergne, Imagerie Moléculaire et Stratégies Théranostiques, Clermont-Ferrand, France
6Immunomonitoring Platform, Laboratory of Immunology, AP-HP, Assistance Publique-Hopitaux de Paris, Georges Pompidou European Hospital, Paris, France
7Inserm U872, Laboratory of Integrative Cancer Immunology, Paris, France 8Université Paris Descartes, Paris, France
9Centre de Recherche des Cordeliers, Université Pierre et Marie Curie, Sorbonne Universités, Paris, France
Colibactin-producingE. coli (CoPEC) are frequently detected in colorectal cancer (CRC) and exhibit procarcinogenic properties. Because increasing evidence show the role of immune environment and especially of antitumor T-cells in CRC development, we investigated the impact of CoPEC on these cells in human CRC and in the APCMin/+mice colon. T-cell density was evaluated by immunohistochemistry in human tumors known for their CoPEC status. APCmin/+mice were chronically infected with a CoPEC strain (11G5). Immune cells (neutrophils and T-cell populations) were then quantified by immunofluorescent staining of the colon. The quantification of lymphoid populations was also performed in the mesenteric lymph nodes (MLNs). Here, we show that the colonization of CRC patients by CoPEC is associated with a decrease of tumor-infiltrating T lymphocytes (CD3+T-cells). Similarly, we demonstrated, in mice, that CoPEC chronic infection decreases CD3+and CD8+T-cells and increases colonic inflammation. In addition, we noticed a significant decrease in antitumor T-cells in the MLNs of CoPEC-infected mice compared to that of controls. Moreover, we show that CoPEC infection decreases the antimouse PD-1 immunotherapy efficacy in MC38 tumor model. Ourfindings suggest that CoPEC could promote a procarcinogenic immune environment through impairment of antitumor T-cell response, leading to tumoral resistance to immunotherapy. CoPEC could thus be a new biomarker predicting the anti-PD-1 response in CRC.
Introduction
Colorectal cancer (CRC) is the third most commonly diag-nosed cancer worldwide.1 Tumoral infiltration by immune cells heavily impacts on clinical outcome in human CRC.2
Indeed, there is now accumulative evidence showing a positive association between the density of intratumoral lymphocyte infiltrates (TILs) in solid tumors and increased patient sur-vival.3 The immune-based assay named Immunoscore basing
Guillaume Carrier current address is: Surgical Oncology Department, Institut du Cancer de Montpellier (ICM), Univ Montpellier, Montpellier, France
Additional Supporting Informationmay be found in the online version of this article.
Key words:colorectal cancer, colibactin, E. coli, immune microenvironment, T-cell
Abbreviations:AOM: azoxymethane; APC: adenomatosis polyposis coli; BLI: bioluminescence imaging; CAC: colitis-associated-colorectal can-cer; CoPEC: colibactin-producing E. coli; CPEC: cyclomodulin-producing E. coli; CRC: colorectal cancan-cer; CTLA-4: cytotoxic T-lymphocyte anti-gen 4; DAPI: 40,6-diamidino-2-phenylindole; E. coli: Escherichia coli; FAP: familial adenomatous polyposis; Foxp3: Forkhead box P3; IF: immunofluorescence; IL: interleukin; Ly6G: lymphocyte antigen 6 complex locus G6D; MLN: mesenteric lymph nodes; MPO: myeloperoxidase; p.i.: postinfection; PD-1: programmed cell death-1; pks: polyketide synthase; ROI: region of interest; SEM: standard error of the mean; TGF-β: tumor growth factor-β; TILs: tumor-infiltrating lymphocytes T-cells; TNM: tumor nodes metastasis; Treg: T regulatory cell; WT: wild-type
DOI:10.1002/ijc.32920
History:Received 13 May 2019; Accepted 4 Feb 2020; Online 9 Feb 2020
Correspondence to:Mathilde Bonnet, E-mail: mathilde.bonnet@uca.fr
Infectious
Causes
of
on the density of tumor-infiltrating CD3+
and CD8+ T-cell effectors, appears to be a strong prognostic factor in associa-tion with the TNM classificaassocia-tion for CRC.4
In line with this predictive value, CD8+ T-cells have been widely described as key effectors that are capable of recognizing and destroying tumoral cells via their cytotoxic functions.5–9 These specific T-cells are capable to recognize and destroy aberrant and tumor cells by secreting granzyme B, perforin, TNF-α and IFN-γ in tumor environment.10,11
In parallel, T-cell dysfunc-tions have also been described during colorectal carcinogene-sis, and these cells fail to effectively eliminate cancer cells.12 Programmed cell death-1 (PD-1), which is a cell surface receptor molecule on T cells, is a marker for this dysfunction of T lymphocytes.13CRC development and outcome are thus largely influenced by the immune response and its microenvi-ronment. In addition, a large amount of microorganisms are also present in this colonic tumor microenvironment and could play a role in carcinogenesis.14,15
Colorectal cancer patients often exhibit a distinct micro-biota composition (dysbiosis) compared to that of a healthy population.14,16,17 Recently, Wong et al. showed that gavage with stools from colorectal cancer patients promotes intestinal carcinogenesis in germ-free and azoxymethane-(AOM)-carcinogen mice models.18 Among the bacteria isolated from CRC tissues, some bacterial species, such as Bacteroides fragilis, Fusobacterium nucleatum and Escherichia coli, play a well-established role in CRC development.19–22 However, in vivo mechanisms by which these bacteria promote colorectal carci-nogenesis are not yet well elucidated. Interestingly, indepen-dent studies have reported that B. fragilis and F. nucleatum inhibit antitumor T-cell response and increase inflammation in CRC mouse model.19–22In addition, association between the high amount of F. nucleatum and a lower CD3+T-cell density has been described in CRC tumor patients.23
Concerning E. coli, particular strains which have acquired virulence factors, including cyclomodulins, were isolated from CRC.24,25 These strains named Cyclomodulin-Producing E. coli (CPEC) have been more frequently detected in colorec-tal adenocarcinoma and adenoma in comparison to non-neoplastic samples.16,20,26,27 Most CPEC strains harbor colibactin-encoding polyketide synthase (pks) pathogenicity island (CoPEC, Colibactin-producing E. coli).28These CoPEC strains were more prevalent in most aggressive CRC tumors (TNM stage).27,29 Colibactin is a genotoxin that induces oxi-dative stress and DNA double-strand breaks in epithelial cells, leading to genomic instability.24,30–32In addition to this geno-toxic effect, CoPEC strains induce cellular senescence associ-ated with the production of growth factors, such as hepatocyte growth factor, leading to an increase of tumor growth in chemically induced colitis-associated CRC (CAC) models.33In addition, experiments using other CAC mouse models (IL-10−/−/AOM and APCMin/+IL10−/− models) showed that CoPEC were able to strongly persist in the gut and to induce epithelial damage and cell proliferation.26,34,35 This bacterial
cancer-promoting activity was dependent on inflammation. However, even though inflammation seems to play a crucial role, the histological score of tumoral inflammation was gener-ally unchanged in infected animals, as described in the APCMin/+/IL-10−/− or APCMin/+ models by Tomkovich et al. and Bonnet et al., respectively. In conclusion, the mechanisms by which CoPEC promote colorectal carcinogenesis are diverse and somewhat specific to the animal models and the microbial status of the animals (germ-free or specific-pathogen-free). Thus, the aim of this work was to evaluate the impact of chronic CoPEC infection on immune cells, in a spontaneous intestinal cancer context which represents the majority of CRC. Because T-cell modulation seems to be central for CRC carcinogenesis,4 this study was focused on T-cell population. Wefirst evaluated the T-cell density in a collection of human sporadic CRC samples that were previously characterized for their pks island status. To confirm the human data, we used a CoPEC chronic infection model in APCMin/+ mice. Finally, since the CoPEC infection had a strong effect on T-cell density, we evaluated the impact of chronic infection on the anti-tumoral response after anti-PD-1 immunotherapy treatment using the MC38 syngeneic CRC tumor model.
Materials and Methods
Immune contexture determination on human CRC biopsies
The immune contexture (CD3 and CD8 densities) was deter-mined for 40 CRC samples from the MiPaCor collection sam-ples (DC-2017-2972). Patients underwent surgery for resectable colon cancer at the Digestive and Hepatobiliary Sur-gery Department of the University Hospital of Clermont-Fer-rand. All the samples were previously tested for the pks status of colonic tissues by PCR.32 We selected 21 pks-negative and 19 pks-positive samples. After colonic resection, fresh speci-mens were transported to the Pathology Department labora-tory of the University Hospital of Clermont-Ferrand,fixed in buffered 4% paraformaldehyde, embedded in paraffin and cut into 5 μm slices. Two tissue paraffin sections of 4 μm were processed for immunohistochemistry by the reference center (Immunomonitoring Platform, Hôpital Européen Georges Pompidou AP-HP, Inserm, Paris, France) as previously described.4 Digital images of the stained tissue sections were obtained at 20× magnification and 0.45 μm/pixel resolution. The densities of CD3+ and CD8+T-cells in colon tumor and invasive margin regions were determined as previously described.4
Bacterial strains
The representative CoPEC strain named 11G5 was isolated from the colonic tissue of a patient with colon cancer as previ-ously described.28Isogenic pks-negative mutant (11G5ΔClbQ) and trans complemented isogenic mutant (11G5 ΔClbQ: ClbQ) were previously described by Cougnoux et al.33 The commensal E. coli strain K-12 MG1655 was used as the non-pathogenic control. For in vivo imaging, the bioluminescent
Infectious
Causes
of
11G5 strain was obtained by transformation with pAT881 plasmids.36
Animals
All studies were approved by the local ethical committee (No. CE-2912) and the French Ethical Animal Use Committee (Apafis#5401 and Apafis#13812). Studies were performed using 6- to 7-week-old C57BL/6J-APCMin/+ (Min mice; The Jackson Laboratory, Bar Harbor, ME) or wild-type (WT) C57BL/6J mice (Charles River Laboratories, L’Abresle, France). All mice were housed in conventional conditions at the animal care facil-ity of Université Clermont Auvergne (Clermont-Ferrand, France) and had unlimited access to food and water.
Chronic infection of APCMin/+mice model
Bacterial infection of C57BL/6J-APCMin/+ females was per-formed as previously described.29 To enhance E. coli strain colonization, we administered streptomycin (2.5 g/l) for 3 days prior to oral inoculation with bacteria (≈ 1 × 108
bac-teria in PBS) or PBS (noninfected controls). Three E. coli strains were tested: the pks+ 11G5 strain, its isogenic mutant 11G5ΔClbQ and the K-12 MG1655 commensal strain. At 2, 7, 14, 21 and 49 days postinfection, we quantified bacterial load in stools as previously described.37 According to the parameters studied, at least six animals per group were tested, and the experiments were repeated at least twice. Fifty days after inoculation, the mice were sacrificed. The colons were removed, and the polyps were counted. Tissues were prepared for bacterial colonization, immunostaining and molecular biology experiments. Mesenteric lymph nodes (MLNs) were harvested and homogenized, bacterial colonization was evalu-ated by selective culture, and immune cells were analyzed by flow cytometry.
For bioluminescence imaging (BLI) of bacteria, the same protocol of chronic infection was performed using the lumi-nescent 11G5 strain in 10 C57BL/6J-APCMin/+mice. Biolumi-nescence imaging was performed using an IVIS Spectrum. After 24 hr, the mice were sacrificed. The colons were removed and imaged was performed using the IVIS spectrum.
Inflammation monitoring in APCMin/+mice
To monitor inflammation in noninfected and infected APCMin/ +
mice at 3, 14 and 49 days postinfection, we performed an intraperitoneal (i.p.) injection of 150μl of the XenoLight Rediject Inflammation probe (Perkin Elmer, Waltham, MA) per mouse. Mice were then imaged 10 min after the i.p. injection of the probe (exposure time of 5 min). Prior to imaging, animals were anesthetized with 2–3% isoflurane in an induction chamber; then, 2% isoflurane in air/O2was
continu-ously delivered via a nose cone system in the dark box of the imaging system. Quantitative analysis was performed using Living Image®Software (Caliper Life Science, Hopkinton, MA; see Supporting Information for full detailed protocol). In addi-tion, myeloperoxidase (MPO) and cytokines in colonic tissue
were evaluated by ELISA or qRT-PCR (see Supporting Infor-mation for full detailed protocol).
Immunofluorescent staining and quantification of immune cells in APCMin/+mice samples
At the end of the experiment, the colons were swiss-rolled and fixed for 24 hr in 10% formalin (Sigma, Tokyo, Japan) at room temperature. Tissues were embedded in paraffin and cut into 5 μm sections. Then, the tissue sections were prepared for hematoxylin–eosin–safranin staining or immunofluorescence staining to analyze colonic immune cells in three regions of interest: the mucosa, lymphoid follicle and tumor. Immuno-staining was performed on four serial sections to analyze four immune cell populations: leukocytes (CD45+staining); neutro-phils (Ly6G+); CD4 T-cells (CD3+CD4+); and CD8 T-cells (CD3+CD8+). All antibodies used in our study are described in Supporting Information Table S1. All IF staining was per-formed using an automated stainer, a Discovery XT processor (Roche, Bâle, Swiss), and the tyramide signal amplification (TSA)-conjugated fluorochrome method was used on whole colon slides (see Supporting Information for all full detailed protocol). Cells quantification was performed on the whole colonic roll. Each slide was automatically analyzed by a specific DIA algorithm developed for our study that was based on Definiens Cognition Network Technology®. All steps and
set-tings have been previously detailed (see Supporting Informa-tion for all full detailed protocol).38
APCMin/+mice mesenteric lymph node (MLN) immune cell analysis byflow cytometry
All antibodies used in our study are described in Supporting Information Table S1. MLNs were immediately harvested and crushed in ice-cold PBS with 2% FCS through nylon-mesh cell strainers. MLN-derived cells were washed in PBS and stained withfixable Viability Dye eFluor450 (eBioscience, San Diego, CA) according to the manufacturer’s instructions. The cells were then washed and incubated at 4C with anti-CD16/ CD32 before the surface staining of the CD3, CD4, CD8, CD25 and TCRβ immune receptors. For intracellular staining of the Foxp3 transcription factor, cells werefixed and perme-abilized with a Transcription Factor Staining buffer set (eBioscience) according to the manufacturer’s instructions. Data were acquired on a BD LSR IIflow cytometer (Becton-Dickinson, San Jose, CA), and analysis was performed using FlowJo™ software (TreeStar).
Impact of11G5 on intratumoral immune cells and anti-PD-1 mAb efficacy in mice
The transplantable MC38 CRC cells derived from C57BL/6J mice were kindly provided by Dr. G. Azar (Sanofi). Mycoplasma-free MC38 cells were maintained as monolayers using culture medium consisting of DMEM (Invitrogen, Cergy Pontoise, France) supplemented with 10% FCS, 1% glutamine, 1% HEPES, 1 mM nonessential amino acids and 1 mM sodium
Infectious
Causes
of
pyruvate at 37C in a humidified incubator containing 5% CO2.
Experiments were performed in 6- to 8-week-old WT C57BL6/J mice with 10 animals per group. To enhance E. coli strain colo-nization, we administered streptomycin (2.5 g/l) for 3 days before the oral inoculation of bacteria (≈1 × 109
bacteria) or PBS (noninfected controls). Nine days after infection, female or male mice were anesthetized by isoflurane inhalation and were inoculated with 1 × 106 MC38 cells by dorsal subcutaneous injection at Day 0 of the experiment. To monitor tumoral growth, the tumor volume in mm3was calculated twice a week from the measurement of two perpendicular diameters using a caliper according to the formula L× S2/2, where L and S are the largest and smallest diameters in mm, respectively. The mice were sacrificed on Day 30 of the experiment. Tumors were removed and weighed. Intratumoral immune cells (CD3+and CD8+CD3+ T-cells, and CD11b+Ly6G+Ly6Cint neutrophils) were analyzed byflow cytometry (see Supporting Information for detailed protocol). All antibodies used in our study are described in Supporting Information (Table S1).
The same protocols were used in the experiment of anti-PD-1 mAb therapy. Male mice harboring a 50–100 mm3
tumor graft were selected. At 8, 11, 14, 18, 20 and 22 days after tumor cell injection, noninfected or 11G5-infected mice were injected intraperitoneally (i.p.) with a PD-1 specific blocking antibody (mu anti-PD-1; mIgG1—a chimeric version of the RPM1-14 rat monoclonal antibody with a mouse IgG1 Fc domain) or an IgG1 isotype control antibody (mIgG1, clone 1B711; 10 μg/g of mice). The antibodies used were obtained from Sanofi R&D (Vitry-Sur-Seine, France). To eval-uate potential toxicity, body weight was measured twice a week.
Statistical analyses
The bacterial colony count in tissues and fecal samples was compared using Kruskal-Wallis nonparametric analysis. Fish-er’s exact test of χ2
analysis was used for the comparison of the proportion between groups. For all other parameters, one-way ANOVA followed by Tukey’s posttest or unpaired Student’s t-test were used. Correlations were determined by Spearman’s test. All tests were performed using GraphPad Prism 7 (StataCorp, College Station, TX).
Ethical statement for human studies
Samples from CRC patients came from the MiPaCor collec-tion (Ethical approval for human study No. DC-2017-2972). All patients were adult volunteers and underwent surgery for resectable CRC in the Digestive and Hepatobiliary Surgery Department of the University Hospital of Clermont-Ferrand. A study information sheet and a verbal explanation were pro-vided to the patients.
Data availability
The data that support the findings of our study are available from the corresponding author upon reasonable request.
Results
Association between colibactin-positive-E. coli status and tumor-infiltrating T lymphocytes density at the invasive margin of human CRC samples
Tumor-infiltrating T lymphocytes (TILs) density was investi-gated in a collection of human CRC colonic tissues that were previously characterized for their pks island status by PCR determining the presence of CoPEC colonization. CD3 and CD8 staining were performed on 40 CRC tumor samples at the Immunomonitoring Platform of the Hôpital Européen Georges Pompidou (Paris). The densities of CD3+and CD8+T-cells in colon tumor and invasive margin regions were determined and correlated with the total E. coli colonization and the presence of CoPEC. No significant correlation was observed between the colonization of the tissue by E. coli and the CD3+or CD8+cell populations (data not shown). Figure 1 shows a significant decrease in CD3+ cells in the invasive margin, in the pks island-positive samples. The proportion of patients with low levels of CD3+cell density (<600 cells/mm2) was increased in the group having CoPEC colonization (11/19; 57.8%) com-pared to pks island-negative samples (5/21; 23.8%; p = 0.05). CD8+cell density tended to be low in the pks island-positive tumors in the invasive margin (Supporting Information Fig. S1). The proportion of patients with low levels of CD8+ cell density (<100 cells/mm2) was increased in the group hav-ing CoPEC colonization (7/19; 36.8%) compared to pks island-negative samples (1/21; 4.8%; p < 0.05).
The colibactin-producingE. coli 11G5 stimulates colon tumor development in a colibactin-dependent manner
Significant increases in the number and size of tumors were observed in the 11G5-colonized APCMin/+mice in comparison to those of the noninfected animals (PBS) or 11G5ΔClbQ- or K12-E. coli-infected animals (Figs. 2a and 2b) without body weight loss after 11G5 infection (Supporting Information Fig. S2). Polyps were distributed from the proximal colon to the distal colon without specific localization patterns (Supporting Information Fig. S3). Histological analysis indi-cated that all polyps were adenocarcinomas (data not shown). The determination of bacterial colonization in the stools of mice showed that 11G5 and 11G5ΔClbQ-infected mice were chronically infected with similar levels of colonization (Fig. 2c). In contrast, significantly reduced bacterial levels were detected in stools of mice infected with E. coli K-12 (Fig. 2c). Moreover, high levels of 11G5 or 11G5ΔClbQ bacteria in the colonic tissues of infected mice were confirmed at 50 days p.i., as well as the total clearance of the K12 E. coli bacteria (Fig. 2d). The location of the 11G5 strain along the digestive tract was assessed by optical imaging using bioluminescent technology. Figure 2e shows that the 11G5 bacteria were pref-erentially localized at the caecum and the proximal colonic area. For all the mice, inflammation was localized preferen-tially in the abdomen area, as visualized by optical imaging
Infectious
Causes
of
using a XenoLight Rediject Inflammation probe (Fig. 2f). The quantitation of the BLI signals showed a significant increase of inflammation only for the 11G5-infected animals at a late stage of the infection (Fig. 2g). To complete this analysis, the levels of a proinflammatory enzyme (MPO) and
anti-inflammatory cytokines (TGF-β and IL-10) were quantified. A weak but significant increase of MPO was detected in both stools and colonic tissues of 11G5-infected mice and a decrease of the two anti-inflammatory cytokines IL-10 and TGF-β was observed compared to those of controls (Supporting Information Fig. S4). In conclusion, these data suggest that 11G5 bacteria promote tumor development and increase gut inflammation in a colibactin-dependent manner.
The colibactin-producingE. coli 11G5 strain increases lymphoid follicle numbers and decreases mucosal leukocyte density in the colon of min mice in a colibactin-dependent manner
Given the critical functions of tumor-infiltrating leukocytes in inflammation and tumor immunosuppression, we then evaluated the effect of 11G5 chronic infection on the colonic leukocyte population. Immunostaining for the CD45 leukocyte-specific protein was performed and quantified on colonic sections from APCMin/+ mice (the noninfected, 11G5-infected and 11G5ΔClbQ-infected groups) in three colonic areas: the lymphoid follicles, mucosa and tumors. Rep-resentative images are shown for these three areas (Figs. 3a and 3f, and Supporting Information Fig. S5, respectively). The lymphoid follicle number and total size per colon were signi fi-cantly increased in 11G5-infected animals in comparison to those of both control groups (Figs. 3b and 3c). The lymphoid follicle size significantly correlated with tumoral development only in the 11G5-infected group (p = 0.02) (Fig. 3d). No dif-ference was observed between the animal groups regarding the leukocyte density in the lymphoid follicles (Fig. 3e). In con-trast, a decrease in the CD45-positive cell density was observed in the 11G5-infected mucosa (Fig. 3g), and this decrease was spread along the entire colon. In addition, a decrease in CD45+cell density was also observed in the tumor areas of the 11G5-infected group (Supporting Information Fig. S5b).
Neutrophil infiltration was then assessed by staining for the Ly6G neutrophil-specific surface protein performed on colon sections at 50 days p.i. The percentage of neutrophils
Figure1.Legend on next coloumn. Figure1.Legend on next page.
Figure1.Association between pks+Escherichia coli colonization and the decrease of tumor-infiltrating lymphocyte T-cells (TILs) at the invasive margin of colorectal cancer (CRC) patients (n =40). (a) Immunohistochemistry picture of TIL detection in CRC patients harboring pks+E. coli strains (top box) and in patients harboring pks−E. coli strains (bottom box). (b) Densities of TILs in tumors and at the invasive margin of patients colonized by pks+or pks−E. coli were assessed by digital image analysis after CD3 surface protein immunostaining. The presence of E. coli harboring colibactin-encoding pks island was investigated using polymerase chain reaction (PCR) tests on all CRC samples. Data were compared by nonparametric Mann–Whitney test.*p = 0.05. [Color figure can be viewed at wileyonlinelibrary.com]
Infectious
Causes
of
Figure2.The colibactin-producing E. coli11G5 strain stimulates colon tumor development in APCMin/+mice in a pks-dependent manner. (a) Number of visible colonic polyps in noninfected mice (PBS) and in mice infected with the pks+11G5, pks−11G5ΔClbQ and
K12-commensal E. coli strains at 50 days postinfection (p.i.). (b) Volume of polyps for each group of mice at 50 days p.i. was assessed from the Caliper measurement. (c) Bacterial colonization in the stools from2 to 50 days p.i. was evaluated by selective culture. (d) Bacterial colonization in colon at50 days p.i. was evaluated by selective culture. (e) Location of the 11G5 strain along the mouse digestive tract was assessed using bioluminescence at6 and 24 hr p.i. C, colon; SI, small intestine; Ca, caecum; P, proximal; D, distal. (f) and (g) Monitoring of intradigestive inflammation in infected mice by optical imaging using the XenoLight Rediject Inflammation probe. (f) Representative images obtained at49 days p.i. (g) Bioluminescent signals were quantified from ROIs that were drawn manually at 3, 14 and 49 days p.i. The results are expressed as the ratio between the digestive signal ROI (average radiance in p/s/cm2/sr) and the muscle ROI, which was used as the signal background.*p < 0.05, **p < 0.01, ***p < 0.005. Abbreviation: ROI, region of interest. [Color figure can be viewed at
wileyonlinelibrary.com]
Infectious
Causes
of
was determined as the ratio of the Ly6G+ cell density to CD45+leukocyte density as measured by digital image analy-sis. Representative images are shown for lymphoid follicles (Supporting Information Fig. S6a), mucosa (Supporting Information Fig. S6c) and tumors (Supporting Information Fig. S6e). No difference in neutrophil percentages was observed in lymphoid follicles (Supporting Information Fig. S6b) or in tumors (Supporting Information Fig. S6f ) between the 11G5-infected animal group and the control group. In the mucosa, a significant increase in the neutrophil percentage was observed in the 11G5-infected mice com-pared to that of the control group (Supporting Information Fig. S6d).
The colibactin-producingE. coli 11G5 strain decreases CD3+ total T-cell and CD8+cytotoxic T-cell in colon of min mice in a colibactin-dependent manner
T-cell populations were investigated by analyzing the total T-cell population by CD3 staining and CD8 or CD4 subpopu-lation double staining. The representative CD3 staining of the lymphoid follicles, mucosa and tumors are shown in Fig-ures 4a, 4c and 4e, respectively. No variation in CD3+ T-cell density was observed in the lymphoid follicles (Fig. 4b). A sig-nificant decrease in CD3+ T-cell density was observed throughout the mucosa of 11G5-infected animals in compari-son to that of the control animals (Fig. 4d). Similarly, a signi fi-cant decrease in CD3+ cells was observed in the tumors
Figure3.The colibactin-producing E. coli11G5 strain modulates the global immune environment in colon by inducing an increase of the lymphoid follicle numbers, a decrease of leukocytes and an increase of neutrophils in the mucosa of APCMin/+mice. Immunofluorescence stainings (IF) for the CD45 leukocyte-specific surface protein (green) and DAPI (Blue) was performed on colon sections at 50 days p.i. to assess the lymphoid follicle number and size and leukocyte density. (a) Representative IF images of colonic mucosa-associated lymphoid follicles for one mouse in each group. Blue:40,6-diaminodino-2-phenylindole (DAPI); Green: CD45 staining. (b) Means + SEM of the number of lymphoid follicles per colon for each group of mice. (c) The lymphoid follicle size per colon was assessed by digital image analysis. (d) Correlation between lymphoid follicle size and total colonic polyp volume in11G5-infected mice was analyzed (Spearman’s test; n = 6). (e) Density of leukocytes in the colonic lymphoid follicles was assessed by DIA for each mouse. Density of leukocytes was obtained by analyzing the number of CD45+cells per lymphoid follicle surface. (f ) Representative IF images of CD45 staining in the colonic mucosa of one mouse from each group. ( g) Density of leukocytes in the colonic mucosa of each mouse was assessed by DIA for the total colonic mucosal section for each mouse. Density of leukocytes is expressed as the number of CD45+cells per tissue surface. Data were compared by one-way ANOVA analysis followed by Tukey’s posttest. *p < 0.05, **p < 0.01. [Color figure can be viewed at wileyonlinelibrary.com]
Infectious
Causes
of
compared to that of controls (Fig. 4f ). As shown in Figure 4e, the intratumoral distribution of CD3+ cells was heteroge-neous, and the decrease was particularly observed in the
margin of the polyps (see arrows in Fig. 4e). Focusing on the CD4 and CD8 T-cell subpopulations, we observed a variation in the CD8+ population after 11G5 chronic infection.
Figure4.The colibactin-producing E. coli11G5 strain modulates colonic CD3+and CD
8+T-cells of APCMin/+mice. CD
3+and CD
8+IF staining (IF) were performed on colon sections at50 days p.i. Cell density and percentage were obtained by the DIA of each stained section. (a) Representative IF image of CD3+T-cells in the lymphoid follicle of one mouse. Blue: DAPI; Orange: CD
3 staining. (b) Density of CD3+ T-cells (Means + SEM) in the lymphoid follicles. (c) Representative IF image of CD3+T-cells in the colonic mucosa. (d) Density of CD
3+T-cells (Means + SEM) in the mucosa. (e) Representative IF image showing the distribution of tumor-infiltrating lymphocytes (TILs), M: mucosa, T: tumor area, Yellow arrows: base of polyps. (f ) Density of TILs (Means + SEM) for each group of mice. ( g) CD8+T-cell density (Means + SEM) in lymphoid follicles, mucosa and tumors. Data were compared by one-way ANOVA analysis followed by Tukey’s posttest. *p < 0.05, **p < 0.01, ***p < 0.001. [Color figure can be viewed at wileyonlinelibrary.com]
Infectious
Causes
of
Figure 4g and Supporting Information Figure S7 report the results of the double CD8/CD3 staining. A significant decrease in the CD8+population was observed throughout the mucosa of 11G5-infected mice compared to that of controls, while no variation was measured in lymphoid follicles. Similarly to our observations in human CRC samples reported above, no vari-ation of tumor-infiltrating CD8+
T-cells was measured in APCMin/+mice tumors (Fig. 4g). No difference was observed for the CD4 T-cell population (Supporting Information Fig. S8).
The colibactin-producingE. coli 11G5 strain decreases CD4+CD
25+and CD 8+CD
25+T-cell populations in the mesenteric lymph nodes (MLNs) of min mice in a colibactin-dependent manner
We completed our analyses of 11G5-infected APCMin/+mice by flow cytometry analysis of T-cells from the mesenteric lymph nodes (MLNs), which are an important site for T-cell activation for the colon.39 We detected the 11G5 bacteria in the MLNs of 6/8 animals (75%), whereas the 11G5ΔClbQ bac-teria was detected in only 1/7 animals (14%) (Fig. 5a). The
Figure5.Colibactin-producing-E. coli11G5 strain translocates to mesenteric lymph nodes (MLNs) and modulates immune cell subpopulations. MLNs from mice were analyzed at50 days p.i. (a) Bacterial translocation in MLNs was evaluated by selective culture. (b–g) Percentages of immune cells were determined by flow cytometry analysis. (b) Percentages of CD3+T-cells (Means + SEM) among live cells. (c) Percentages of CD4+and CD
8+T-cells among T-cells (Means + SEM). (d) Percentages of CD 25+CD
4+T-cells (Means + SEM). (e) Correlation analysis between CD25+CD
4+T-cells and MLN colonization by
11G5 E. coli (n = 8; Spearman’s test). (f) CD25+CD 8+ percentages among CD8+T-cells (Means + SEM). ( g) Analysis of the correlation between CD
25+CD
8+T-cells and MLN colonization by E. coli 11G5 (n = 8; Spearman’s test). *p < 0.05, **p < 0.01, ****p < 0.001. [Color figure can be viewed at wileyonlinelibrary.com]
Infectious
Causes
of
11G5 strain translocation was only observed in APCMin/+mice since no translocation was detected in WT mice (data not shown). No significant variation was observed between the three groups for the MLN cellularity and cell viability (Supporting Information Fig. S9). Flow cytometry analysis showed similar MLN composition for CD3+, CD3+CD4+and
CD3+CD8+populations in the three groups of mice (Figs. 5b and 5c). However, significant decreases in CD4+
CD25+ and CD8+CD25+cells were observed in the 11G5-infected animals compared to controls (Figs. 5d and 5f ). Interestingly, these decreases were correlated with the level of bacterial transloca-tion in the MLNs (Figs. 5e and 5g). In additransloca-tion, a negative
Figure6.Legend on next page.
Infectious
Causes
of
correlation was observed between CD4+FoxP3+ Treg cells in the MLNs and the 11G5-bacterial translocation level (Supporting Information Fig. S9).
Infection with the colibactin-producingE. coli 11G5 strain induces a decrease of intratumoral CD3+and CD3+CD8+ T cells and a resistance to antimouse PD-1 immunotherapy
To confirm the impact of CoPEC on TILs observed in 11G5-infected APCMin/+mice, we investigated the effect of CoPEC chronic infection on MC38 murine graft model usingflow cyto-metry. A significant decrease in CD3+ TILs was confirmed in
MC38 tumors of 11G5-infected animals in comparison to control animals in both male (Figs. 6a and 6b) and female animals (Supporting Information Fig. S10). A global decrease of CD8+TILs was observed only in 11G5-infected mice group (Figs. 6c and 6d). Similarly to our results in APCMin/+mice, no effect was noticed for the CD4+TILs (data not shown). In addition, a significant increase of neutrophils population was measured in 11G5-infected samples (Figs. 6e and 6f; Supporting Information Fig. S10). All these data were in line with our observations in APCMin/+mice. In addition, same results were obtained in mice infected with the 11G5 strain and trans complemented isogenic mutant (11G5ΔClbQ: ClbQ) showing that complementation restored the colibactin biological effect and confirming the effect of this genotoxin on immune cells (Supporting Information Fig. S10).
Because chronic infection of the gut with 11G5 strains induces a decrease in antitumoral immunity, we investigated the impact of this infection on antimouse PD-1 therapy effi-cacy, targeted immunotherapy based on restoration of anti-tumor CD8+T-cells activity. For this purpose, MC38 murine tumor graft model was chosen because this model is known to be sensitive to immunotherapy.40Figures 6g and 6h show that anti-PD-1 treatment did not impact the bacterial colonization of the gut. As shown in Figure 6i, we observed a significant response to anti-PD-1 treatment in noninfected animals. Indeed, tumoral growth was significantly slowed, demonstrat-ing the antitumoral efficacy of the anti-PD-1 treatment in our experimental conditions. In contrast, no effect on tumoral growth was observed in animals infected with the 11G5 bacte-ria that were subjected to the same treatment (Fig. 6i).
Discussion
Increasing evidence suggests that colorectal carcinogenesis is not only governed by the genetic changes inherent to cancer cells but also by tumor environment such as immune cells and microbiota.41In the gut microbiota, different species have emerged as important procarcinogenic factors such as CoPEC which are more prevalent in the gut of patients with advanced TNM stage CRC.27,29In this work, we providedfirst evidence that the chronic infection with CoPEC bacteria impacts the T-cell colonic population which could lead to colorectal carci-nogenesis and resistance to immunotherapy.
A number of studies have described the procarcinogenic properties of CoPEC using various animal models that mostly mimic colitis-associated CRC (CAC), which represent less than 1% of sporadic CRC.26,33,35Here, we investigated these proper-ties in the APCMin/+mouse model bearing an alteration of the Apc gene which is thefirst genomic alteration in the adenoma-cancer sequence in about 80% of sporadic CRC.29We observed that CoPEC bacteria were preferentially detected in the caecum and proximal areas of the APCMin/+ mice colon, while their procarcinogenic properties affected the whole colon of infected mice. Our results were in agreement with previous studies per-formed in CAC-related models showing that CoPEC oncogenic properties are not restricted to their persistence site.33 In the first publication describing the in vitro colibactin effect, it was suggested that direct contact between bacteria and target-cells was required to induce double-strand DNA breaks and geno-mic instability.30,42 However, other studies have already described a bystander effect of CoPEC through the induction of cellular senescence in infected cells and the production of growth factors that could promote the proliferation of uni-nfected cells and, subsequently, tumor growth.33,43Ourfindings also support an alteration of immunosurveillance after CoPEC-chronic infection which could be implicated on carcinogenesis.
The decrease of CD3+ T-cell populations in the invasive margin of CoPEC-positive human CRC tumors was shown. In the same way, a similar decrease of CD3+T-cell density com-pared to controls, affecting particularly the CD8+subpopulation was observed in the colon of APCMin/+mice chronically infected by CoPEC as well as in MC38 tumors of CoPEC-infected mice.
Figure6.Antitumoral effect of PD-1-based immunotherapy on the growth rate of MC38 murine colonic tumors and TILs level in colibactin-producing-E. coli preinfected mice. (a) Representativeflow-cytometry plots to assess percentages of CD3+T-cells among live CD
45+ leucocytes in grafted MC38 tumors for each group of male mice. (b) Significant decrease of CD3+percentages in tumors of
11G5-infected mice compared to control mice (Means + SEM). (c) Representativeflow-cytometry plots to assess percentages of CD8+CD
3+T-cells among live CD45+leucocytes in grafted MC
38 tumors for each group of mice. (d) Significant decrease of Percentages of CD8+CD
3+T-cells among live CD45+leucocytes in tumors of
11G5-infected mice (Means + SEM) compared to control mice (e) Representative flow-cytometry plots to assess percentages of neutrophils (CD11b+Ly
6G+Ly
6C−cells) among live CD45+leucocytes in grafted MC
38 tumors for each group of mice. (f ) Percentages of neutrophils (Means + SEM) significantly increased in tumors of 11G5-infected mice compared to control mice (Means + SEM). ( g–i) Impact of chronic 11G5 infection on anti-PD1 mAb therapy efficacy. (g, h) The level of 11G5-bacterial colonization in the feces ( g) or in the colonic tissue (h) was evaluated by selective culture after injection of isotype control or anti-PD1. (i) MC38 tumor growth in uninfected animals was assessed from the Caliper measurement and showed a significant response to PD-1-based immunotherapy. No antitumoral effect of PD-1-based immunotherapy was observed in mice chronically infected with the 11G5 CoPEC strain. Results shown here are representative of two independent experiments (n =6–9 mice according to the group in each experiment). Data were compared by one-way ANOVA analysis followed by Tukey’s posttest. *p < 0.05, ***p < 0.005. [Color figure can be viewed at wileyonlinelibrary.com]
Infectious
Causes
of
Even though the antitumor function of T-cells was not directly investigated in our study, the significant decrease of TILs and the resistance of MC38 tumors to anti-PD-1 therapy in 11G5-infected mice supported the depletion of the tumor-reactive T-cell fraction in these mice. Indeed, anti-PD-1 is sup-posed to counterbalance T-cell exhaustion and restore the antitumor effector properties of T-cells; thus, the inefficiency of anti-PD-1 in 11G5-infected mice strongly suggests that tumor-specific T-cells are decreased in these mice.44
In the same way, it was described that induction of IFNγ+
/CD8+ T-cell accumulation by gut microbiota manipulation was asso-ciated with an increase of anti-PD-1 therapy efficacy in the MC38 graft-model.45 In addition, the decrease of the CD4+CD25+and CD8+CD25+T-cells in the MLNs after infec-tion and translocainfec-tion of the 11G5 E. coli were consistent with the alteration of antitumoral immunity in the colon. Indeed, these cells have been considered regulatory T-cells in the liter-ature, and it has been described that adoptive transfer of CD4+CD25+ Tregs can induce tumor regression in APCMin/+ mice.46 The modulation of immune cells by CoPEC infection was observed in non-neoplastic colonic mucosa and could thus reflect the early events of carcinogenesis. The impairment of T-cell-mediated immunosurveillance in CoPEC-colonized mucosa could prevent elimination of aberrant epithelial cells, leading to carcinogenesis.
In addition to this crucial decrease of T-cell response, we observed an increase in intestinal inflammation along the digestive tract of 11G5-infected APCMin/+mice with weak but significant increases in the proinflammatory enzyme MPO, lymphoid follicle number and size, and neutrophil percentage in the colonic mucosa compared to those of the controls. His-tological analyses and leukocyte quantitation did not reveal massive immune infiltrate in the tumors of infected animals. As with other tumorigenic bacteria, CoPEC could subvert host immunity and establish persistent infections associated with low grade but chronic inflammation. In the same way, previ-ous works performed on CAC models showed that inflamma-tion is essential for the carcinogenic properties of CoPEC, while no significant increase in the inflammatory score was observed in the colon of E coli-infected mice.34,35 Our data suggest that, through the recruitment of neutrophils to the mucosa, CoPEC could generate a proinflammatory microenvi-ronment as confirmed by the increase of MPO. Since Raisch et al. previously demonstrated that the induction of inflamma-tion in CoPEC-infected macrophages is not linked to col-ibactin, macrophages were not investigated in the present study.47 However, we measured a decrease in anti-inflammatory cytokines in the proximal colon mucosa after infection with CoPEC, which could amplify the mucosal
inflammation. In conclusion, we have therefore identified a protumoral signature of immune environment after CoPEC chronic infection, with an increase of inflammation and a decrease of antitumoral T cells that could participate in the cancer-promoting activity of colibactin. This work allows a better understanding of the mechanism underlying the inter-actions between sporadic CRC tumor and its environment.
Finally, in addition to their procarcinogenic effects, the presence of CoPEC could affect the efficacy of antitumoral treatments. Because we observed an impact of the infection on immune cells and particularly on T-cells, we tested the effect of anti-PD-1 immunotherapy in a CoPEC infection context using the MC38 graft-model. The experimental therapeutic response of this tumoral model was previously shown to be sensitive to the microbiota. We showed that CoPEC infection induces the resistance of MC38 tumors to anti-PD-1 immuno-therapy. Our work thus suggests that CoPEC could be a new biomarker to predict the anti-PD-1 response of CRC patients. Indeed, it is known that CRCs that are heavily infiltrated with TILs are associated with good prognosis.4Here, we showed a significantly decreased CD3+
cells density in human tumors colonized by pks-positive E. coli compared to that of tumors colonized by pks-negative E. coli, suggesting a relationship between the presence of CoPEC and CRC prognosis. Of note, we previously showed that colonization of mucosa by CoPEC is associated to poor prognostic factors for CRC (TNM stage).29In addition, it was recently reported that E. coli spe-cies were more abundant in the stools of melanoma patients who were resistant to an anti-PD-1 antibody.48 Large multi-center prospective clinical studies are needed to evaluate the prognostic value of CoPEC for CRC management.
Acknowledgements
We thank Michael Rodrigues and Florence Marliot for their technical assistance, Elodie Jouberton and Leslie Mazuel from the In vivo Imaging Auvergne platform, the CICS platform from Université Clermont-Auvergne for assistance with transmission electron microscopy and Abdelkrim Alloui (Animal Facilities) for animal care. We kindly thank Catherine Grillot Courvalin (Institut Pasteur, Paris, France) and George Azar (Sanofi) for providing the pAT881 plasmid and MC38 cells, respec-tively. Our study was supported by the Ministère de la Recherche et de la Technologie, Inserm and Université Clermont-Auvergne (UMR1071), INRA (USC-2018) and by grants from the Conseil Régional d’Auvergne, La ligue contre le cancer 63/03 and by the French government IDEX-ISITE initiative (Grant number: 16-IDEX-0001-CAP 20-25) of the Univer-sity of Clermont Auvergne. Amélie Lopès and Romain Villéger were supported by a CIFRE grant (Grant number: 2015/622) and by a CPER EPICURE 2016, ANRT (Cifre 2015/622) grant, respectively.
Conflict of interest
The authors declare no conflict of interest.
References
1. Ferlay J, Soerjomataram I, Dikshit R, et al. Cancer incidence and mortality worldwide: sources,
methods and major patterns in GLOBOCAN 2012. Int J Cancer 2015;136:E359–86.
2. Fridman WH, Pagès F, Sautès-Fridman C, et al. The immune contexture in human tumours:
Infectious
Causes
of
impact on clinical outcome. Nat Rev Cancer 2012; 12:298–306.
3. Galon J, Costes A, Sanchez-Cabo F, et al. Type, density, and location of immune cells within human colorectal Tumors predict clinical out-come. Science 2006;313:1960–4.
4. Pagès F, Mlecnik B, Marliot F, et al. Interna-tional validation of the consensus Immunoscore for the classification of colon cancer: a prognos-tic and accuracy study. Lancet 2018;391: 2128–39.
5. Colangelo T, Polcaro G, Muccillo L, et al. Friend or foe? The tumour microenvironment dilemma in colorectal cancer. Biochim Biophys Acta 1867; 2017:1–18.
6. Dunn GP, Koebel CM, Schreiber RD. Interferons, immunity and cancer immunoediting. Nat Rev Immunol 2006;6:836–48.
7. Gajewski TF, Schreiber H, Fu Y-X. Innate and adaptive immune cells in the tumor microenvi-ronment. Nat Immunol 2013;14:1014–22. 8. Koebel CM, Vermi W, Swann JB, et al. Adaptive
immunity maintains occult cancer in an equilib-rium state. Nature 2007;450:903–7.
9. Zhang N, Bevan MJ. CD8+ T cells: foot soldiers of the immune system. Immunity 2011;35: 161–8.
10. Salama P, Phillips M, Platell C, et al. Low expres-sion of Granzyme B in colorectal cancer is associ-ated with signs of early metastastic invasion. Histopathology 2011;59:207–15.
11. Jifu E, Yan F, Kang Z, et al. CD8+CXCR5+ T cells in tumor-draining lymph nodes are highly acti-vated and predict better prognosis in colorectal cancer. Hum Immunol 2018;79:446–52. 12. Thommen DS, Schumacher TN. T cell
dysfunc-tion in cancer. Cancer Cell 2018;33:547–62. 13. Pardoll DM. The blockade of immune
check-points in cancer immunotherapy. Nat Rev Cancer 2012;12:252–64.
14. Tilg H, Adolph TE, Gerner RR, et al. The intesti-nal microbiota in colorectal cancer. Cancer Cell 2018;33:954–64.
15. Mendonça LABM, dos Santos FR, de Cássia Avellaneda Guimar~aes R, et al. The complex puz-zle of interactions among functional food, gut microbiota, and colorectal cancer. Front Oncol 2018;8:325.
16. Nakatsu G, Li X, Zhou H, et al. Gut mucosal microbiome across stages of colorectal carcino-genesis. Nat Commun 2015;6:8727.
17. Feng Q, Liang S, Jia H, et al. Gut microbiome development along the colorectal adenoma-carcinoma sequence. Nat Commun 2015;6:6528. 18. Wong SH, Zhao L, Zhang X, et al. Gavage of Fecal
samples from patients with colorectal cancer pro-motes intestinal carcinogenesis in germ-free and conventional mice. Gastroenterology 2017;153: 1621–1633.e6.
19. Thiele Orberg E, Fan H, Tam AJ, et al. The mye-loid immune signature of enterotoxigenic Bacte-roides fragilis-induced murine colon
tumorigenesis. Mucosal Immunol 2017;10:421–33. 20. Dejea CM, Fathi P, Craig JM, et al. Patients with
familial adenomatous polyposis harbor colonic biofilms containing tumorigenic bacteria. Science 2018;359:592–7.
21. Kostic AD, Chun E, Robertson L, et al. Fusobacterium nucleatum potentiates intestinal tumorigenesis and modulates the tumor-immune microenvironment. Cell Host Microbe 2013;14: 207–15.
22. Wu S, Rhee K-J, Albesiano E, et al. A human colonic commensal promotes colon tumorigenesis via activation of T helper type 17 T cell responses. Nat Med 2009;15:1016–22.
23. Mima K, Sukawa Y, Nishihara R, et al. Fusobacterium nucleatum and T cells in colorectal carcinoma. JAMA Oncol 2015;1:653–61. 24. Cuevas-Ramos G, Petit CR, Marcq I, et al.
Escherichia coli induces DNA damage in vivo and triggers genomic instability in mammalian cells. Proc Natl Acad Sci USA 2010;107: 11537–42.
25. Vizcaino MI, Crawford JM. The colibactin war-head crosslinks DNA. Nat Chem 2015;7:411–7. 26. Arthur JC, Perez-Chanona E, Mühlbauer M, et al.
Intestinal inflammation targets cancer-inducing activity of the microbiota. Science 2012;338:120–3. 27. Eklöf V, Löfgren-Burström A, Zingmark C, et al.
Cancer-associated fecal microbial markers in colo-rectal cancer detection: Fecal microbial markers in colorectal cancer detection. Int J Cancer 2017;141: 2528–36.
28. Buc E, Dubois D, Sauvanet P, et al. High preva-lence of mucosa-associated E coli producing cyclomodulin and genotoxin in colon cancer. PLoS One 2013;8:e56964.
29. Bonnet M, Buc E, Sauvanet P, et al. Colonization of the human gut by E. coli and colorectal cancer risk. Clin Cancer Res 2014;20:859–67. 30. Nougayrède J-P, Homburg S, Taieb F, et al.
Escherichia coli induces DNA double-strand breaks in eukaryotic cells. Science 2006;313: 848–51.
31. Bossuet-Greif N, Vignard J, Taieb F, et al. The Colibactin Genotoxin generates DNA Interstrand cross-links in infected cells. mBio 2018;9: e02393-17.
32. Gagnière J, Bonnin V, Jarrousse A-S, et al. Inter-actions between microsatellite instability and human gut colonization by Escherichia coli in colorectal cancer. Clin Sci (Lond) 2017;131: 471–85.
33. Cougnoux A, Dalmasso G, Martinez R, et al. Bac-terial genotoxin colibactin promotes colon tumour growth by inducing a senescence-associated secre-tory phenotype. Gut 2014;63:1932–42.
34. Arthur JC, Gharaibeh RZ, Mühlbauer M, et al. Microbial genomic analysis reveals the essential role of inflammation in bacteria-induced colorec-tal cancer. Nat Commun 2014;5:4724.
35. Tomkovich S, Yang Y, Winglee K, et al. Locoregional effects of microbiota in a preclinical model of colon carcinogenesis. Cancer Res 2017; 77:2620–32.
36. Foucault M-L, Thomas L, Goussard S, et al. In vivo bioluminescence imaging for the study of intestinal colonization by Escherichia coli in mice. Appl Environ Microbiol 2010;76:264–74. 37. Carvalho FA, Barnich N, Sivignon A, et al.
Crohn’s disease adherent-invasive Escherichia coli colonize and induce strong gut inflammation in transgenic mice expressing human CEACAM. J Exp Med 2009;206:2179–89.
38. Lopès A, Cassé AH, Billard E, et al. Deciphering the immune microenvironment of a tissue by digi-tal imaging and cognition network. Sci Rep 2018; 8:16692.
39. Houston SA, Cerovic V, Thomson C, et al. The lymph nodes draining the small intestine and colon are anatomically separate and immunologi-cally distinct. Mucosal Immunol 2016;9:468–78. 40. Vetizou M, Pitt JM, Daillere R, et al. Anticancer immunotherapy by CTLA-4 blockade relies on the gut microbiota. Science 2015;350:1079–84. 41. Fletcher R, Wang Y-J, Schoen RE, et al.
Colo-rectal cancer prevention: immune modulation taking the stage. Biochim Biophys Acta 1869; 2018:138–48.
42. Wilson MR, Jiang Y, Villalta PW, et al. The human gut bacterial genotoxin colibactin alkylates DNA. Science 2019;363:709.
43. Dalmasso G, Cougnoux A, Delmas J, et al. The bacterial genotoxin colibactin promotes colon tumor growth by modifying the tumor microenvi-ronment. Gut Microbes 2014;5:675–80. 44. Simon S, Labarriere N. PD-1 expression on
tumor-specific T cells: friend or foe for immunotherapy? Oncoimmunology 2017;7: e1364828.
45. Tanoue T, Morita S, Plichta DR, et al. A defined commensal consortium elicits CD8 T cells and anti-cancer immunity. Nature 2019;565:600–5. 46. Erdman SE, Sohn JJ, Rao VP, et al. CD4+ CD25+
regulatory lymphocytes induce regression of intes-tinal tumors in ApcMin/+ mice. Cancer Res 2005; 65:3998–4004.
47. Raisch J, Rolhion N, Dubois A, et al. Intracellular colon cancer-associated Escherichia coli promote protumoral activities of human macrophages by inducing sustained COX-2 expression. Lab Invest 2015;95:296–307.
48. Gopalakrishnan V, Spencer CN, Nezi L, et al. Gut microbiome modulates response to anti–PD-1 immunotherapy in melanoma patients. Science 2018;33:570–80.
