Editing of the gut microbiota reduces carcinogenesis in mouse models of colitis-associated colorectal cancer

Abstract

Chronic inflammation and gut microbiota dysbiosis, in particular the bloom of genotoxin-producing E. coli strains, are risk factors for the development of colorectal cancer. Here, we sought to determine whether precision editing of gut microbiota metabolism and composition could decrease the risk for tumor development in mouse models of colitis-associated colorectal cancer (CAC). Expansion of experimentally introduced E. coli strains in the azoxymethane/dextran sulfate sodium colitis model was driven by molybdoenzyme-dependent metabolic pathways. Oral administration of sodium tungstate inhibited E. coli molybdoenzymes and selectively decreased gut colonization with genotoxin-producing E. coli and other Enterobacteriaceae. Restricting the bloom of Enterobacteriaceae decreased intestinal inflammation and reduced the incidence of colonic tumors in two models of CAC, the azoxymethane/dextran sulfate sodium colitis model and azoxymethane-treated, Il10-deficient mice. We conclude that metabolic targeting of protumoral Enterobacteriaceae during chronic inflammation is a suitable strategy to prevent the development of malignancies arising from gut microbiota dysbiosis.

Figure 1.

Impact of tungstate treatment on gut microbial communities during chronic inflammation. Groups of C57BL/6 mice were treated i.p. with 20 mg/kg AOM. After 7 d, animals received streptomycin (Strep; 2 mg/ml in the drinking water) and were intragastrically inoculated with E. coli MP13 2 d later. Mice were treated for 1 wk with 2% DSS (n = 15), 2% DSS supplemented with 0.2% sodium tungstate (W, n = 18), or filtered drinking water (mock, n = 5). Animals were allowed to recover for 2 wk, with the tungsten-treated group receiving 0.2% sodium tungstate throughout the experiment. Injury was repeated two more times. One group of animals received tungstate exclusively during DSS treatment (W_(Tx), n = 9). Another group received DSS but was not colonized by MP13 (n = 6). At predetermined time points, feces were collected, and tissues were harvested 73 d after AOM treatment. Two independent experiments were performed and are shown in B; C–E show results from one of these experiments. (A) Schematic representation of the experimental design. (B) Burden of E. coli MP13 in the feces and colon content at the indicated time points determined by plating on selective media. LOD, limit of detection. Bars represent the geometric mean ± 95% confidence interval. P values were calculated by two-way ANOVA and a post hoc Tukey’s multiple comparison test on log-transformed data. (C) Principal coordinate analysis (PCoA) of the gut microbiota composition, as determined by 16S rDNA amplicon sequencing. DSS-treated, red circles; DSS+W–treated, blue circles. Data from each mouse are linked by a colored line. (D) Bacterial phyla that were significantly enriched in indicated groups identified by linear discriminant analysis (LDA) effect size measurements. (E) Comparison of gut microbiota composition at the class level in DSS-treated and DSS+W–treated mice. Bars represent the mean ± 95% confidence interval. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not statistically significant.

Figure 2.

Analysis of tumorigenesis in the AOM/DSS colitis model. Tissue was obtained from the animals described in Fig. 1 (73 d after AOM treatment). Animals were initially treated with AOM and colonized with the E. coli MP13 wild-type strain, as indicated. Groups received filtered drinking water (mock; white bars, n = 5), 2% DSS (black bars, n = 15), or 2% DSS + 0.2% sodium tungstate (gray bars, n = 18). Subsets of the animals received tungstate exclusively during DSS-induced flares (W_(Tx), n = 9; gray dash pattern) or received DSS but were not colonized by MP13 (black dash pattern, n = 6). Data from two independent experiments are shown. (A) Colon length was normalized to the total body weight of each animal. White dots represent data from individual animals. (B) Tumor incidence in the large intestine. (C) Gross pathology of the distal colon. Scale bars represent 1 mm. (D) Representative images of H&E-stained colonic sections. Scale bars represent 250 µm. (E) Tumor area as determined by a morphometric analysis. White dots represent data from individual animals. Bars represent the geometric mean ± 95% confidence interval. P values were calculated by unpaired, two-tailed Student’s t test on log-transformed data. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 3.

Contribution of the native gut microbiota to tumorigenesis in the AOM/DSS colitis model. Mice harboring endogenous Enterobacteriaceae were treated with AOM and DSS as described in Fig. 1, except that the antibiotic treatment and colonization with E. coli was omitted. Tissue was obtained 73 d after AOM treatment. Groups received either 2% DSS (n = 10, black bars) or 2% DSS + 0.2% sodium tungstate (n = 11, gray bars). Data from one experiment are shown. (A) The burden of Enterobacteriaceae in the feces and in the colon content at the indicated time points was determined by plating on MacConkey agar plates. Bars represent the geometric mean ± 95% confidence interval. P values were calculated by two-way ANOVA and a post hoc Tukey’s multiple comparison test on log-transformed data. (B) Tumor incidence in the colon. Bars represent the median ± interquartile range. P values were calculated by two-tailed Mann–Whitney U test. (C) Representative images of the distal colon. Scale bars represent 1 mm. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 4.

Effect of oral sodium tungstate administration on genotoxin-producing E. coli populations and tumorigenesis in the AOM/DSS colitis model. Groups of C57BL/6 mice were treated i.p. with AOM. After 7 d, streptomycin-pretreated animals were intragastrically inoculated with the E. coli NC101 wild-type or an isogenic Δpks mutant (n = 8). Groups were then treated for 1 wk with 2% DSS (n = 29), 2% DSS supplemented with 0.2% sodium tungstate (W, n = 32), or filtered drinking water (mock, n = 10) as indicated. Animals were allowed to recover for 2 wk, with the tungsten-treated group receiving 0.2% sodium tungstate throughout the experiment. Injury was repeated two more times. At predetermined time points, feces were collected, and tissues were harvested 73 d after AOM treatment. See Fig. 1 A for a schematic representation. Data from five independent experiments are shown. (A) Burden of NC101 in the feces and the colon content at various time points determined by a culture-dependent method. The dotted line indicates the limit of detection (LOD). Bars represent the geometric mean ± 95% confidence interval. P values were calculated by two-way ANOVA and a post-hoc Tukey’s multiple comparison test on log-transformed data. ns, not significant. (B) Representative images of gross pathology of the distal large intestine. Scale bars represent 1 mm. (C) Tumor incidence in the large intestine. Each dot represents of one animal. Bars indicate the median and the interquartile range. P values were calculated by two-tailed Mann–Whitney U test. (D) Correlation between the NC101 population size (mean of three flares) and the tumor incidence at the end of the experiment. Each dot represents one animal; the color of the dot indicates the treatment. The Spearman’s rank correlation coefficient is indicated, and a linear trendline is shown (P < 0.001). *, P < 0.05; **, P < 0.01, ***, P < 0.001; ns, not statistically significant.

Figure 5.

Analysis of DNA damage and intestinal inflammation. (A) Groups of C57BL/6 mice were treated i.p. with 20 mg/kg AOM. After 7 d, animals received streptomycin (2 mg/ml in the drinking water) and were intragastrically inoculated with the colibactin-producing E. coli NC101 2 d later. Mice were treated for 1 wk with either 2% DSS (black bars) or 2% DSS supplemented with 0.2% sodium tungstate (W; gray bars). Animals were allowed to recover for 2 wk, with the tungsten-treated group receiving 0.2% sodium tungstate throughout the experiment. After 38 d, samples of the colon were obtained. Data from two independent experiments are shown. (B–D) H&E-stained sections were scored by a veterinary pathologist for epithelial damage, polymorphonuclear neutrophil infiltration, submucosal edema, and exudate in the lumen. DSS group: n = 7; DSS+W group: n = 8. (B) Combined histopathology score. Bars indicate the median and the interquartile range. P values were calculated by two-tailed Mann–Whitney U test. (C and D) Representative images of H&E-stained sections of the proximal (C) and distal (D) colon. Scale bars represent 100 µm. (E–G) Transverse sections of colonic tissues were stained for E. coli (red), DNA damage using γ-H2AX foci as marker (green), and cell nuclei (blue). Individual image tiles were assembled to quantify E. coli burden and number of γ-H2AX–positive cells and total mammalian cells. Both groups: n = 4. (E) Representative images. Scale bars represent 500 µm. (F) Burden of E. coli normalized to the number of mammalian cells. (G) Abundance of γ-H2AX–positive cells normalized to the number of all mammalian cells. For F and G, P values were calculated by unpaired, two-tailed Student’s t test on log-transformed data. Bars represent the geometric mean ± 95% confidence interval. *, P < 0.05; **, P < 0.01.

Figure 6.

Impact of oral sodium tungstate treatment on cell cycle progression and DNA damage response in the AOM/DSS model. Groups of C57BL/6 mice were treated i.p. with 20 mg/kg AOM. After 7 d, animals received streptomycin (2 mg/ml in the drinking water) and were intragastrically inoculated with the colibactin-producing E. coli NC101 2 d later. Mice were treated for 1 wk with 2% DSS (n = 14), 2% DSS supplemented with 0.2% sodium tungstate (DSS+W, n = 13), or filtered drinking water (mock, n = 5). Animals were allowed to recover for 4 d. Colonocytes were isolated, total RNA was extracted, and relative mRNA levels of marker genes were determined by quantitative RT-PCR. Data from two independent experiments are shown. (A) Schematic representation of the experiment. (B) mRNA levels of Ercc1 and Mlh1, whose gene products are involved in DNA damage response. (C) mRNA levels of Trp53, encoding the tumor suppressor p53. (D) mRNA levels of Cdk4, Cdk6, and Ccnd1, whose products are involved in cell cycle progression. Bars represent the geometric mean ± 95% confidence interval. P values were calculated by unpaired, two-tailed Student’s t test on log-transformed data. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

Figure 7.

Impact of oral sodium tungstate treatment on E. coli populations and tumorigenesis in AOM-treated Il10-deficient mice. Groups of Il10-deficient mice on the C57BL/6 background were injected i.p. with AOM, treated with streptomycin (Strep), and then colonized with the E. coli NC101 wild-type strain. Inflammation was induced by administration of piroxicam in the rodent diet for 4 wk. One group was treated with 0.02% sodium tungstate (Piroxicam+W, n = 10) in the drinking water while the other received filtered drinking water (Piroxicam, n = 10). Samples were obtained 38 d after AOM injection. Data from two independent experiments are shown. (A) Schematic representation of the experimental design. (B) Burden of E. coli NC101 in the colon content. Bars represent the geometric mean ± 95% confidence interval. P values were calculated by unpaired, two-tailed Student’s t test on log-transformed data. (C) The taxonomic composition of the microbial community in colonic content determined by 16S rDNA amplicon sequencing. (D) Representative images of the colon. Scale bars represent 1 mm. (E) Tumor incidence in the large intestine. Bars indicate the median and the interquartile range. P values were calculated by two-tailed Mann–Whitney U test. *, P < 0.05; ***, P < 0.001.