Microbial genomic analysis reveals the essential role of inflammation in bacteria-induced colorectal cancer

Abstract

Enterobacteria, especially Escherichia coli, are abundant in patients with inflammatory bowel disease or colorectal cancer (CRC). However, it is unclear whether cancer is promoted by inflammation-induced expansion of E. coli and/or changes in expression of specific microbial genes. Here we use longitudinal (2, 12 and 20 weeks) 16S rRNA sequencing of luminal microbiota from ex-germ-free mice to show that inflamed Il10(-/-) mice maintain a higher abundance of Enterobacteriaceae than healthy wild-type mice. Experiments with mono-colonized Il10(-/-) mice reveal that host inflammation is necessary for E. coli cancer-promoting activity. RNA-sequence analysis indicates significant changes in E. coli gene catalogue in Il10(-/-) mice, with changes mostly driven by adaptation to the intestinal environment. Expression of specific genes present in the tumour-promoting E. coli pks island are modulated by inflammation/CRC development. Thus, progression of inflammation in Il10(-/-) mice supports Enterobacteriaceae and alters a small subset of microbial genes important for tumour development.

Figure 1. Change in microbial community composition over time

a-c) Bray-Curtis Principal Coordinate Analysis (PCoA) at the operational taxonomic unit (OTU) level, with ANOSIM R and P-values nested on cage. Each symbol represents an individual mouse at the indicated time-point. d) Mixed linear model FDR corrected P-values for the first 10 coordinates of PCoA (explaining 58.4% of the variance) evaluating the null hypothesis that the fixed factor indicated above each plot does not impact the coordinate. Gray line represents P=0.05 significance level. e-f) Comparisons by the mixed linear model, with all comparisons and FDR corrected P values shown in (f). Il10^−/− week 2 n=17, week 12 n=16, week 20 n=15; WT week 2 n=24, week 12 n=22, week 20 n=24.

Figure 2. Abundance of Enterobacteriaceae and Escherichia/Shigella

Increased abundance of (a) Enterobacteriaceae, and (b) OTU consensus 27 (Escherichia/Shigella) in Il10^−/− mice. Box and whisker plots show the minimum, first quartile, median, third quartile and maximum relative abundance (showing the median of each cage). FDR corrected P values from the mixed linear model. Il10^−/− week 2 n=17, week 12 n=16, week 20 n=15; WT week 2 n=24, week 12 n=22, week 20 n=24.

Figure 3. RNA-seq reveals changes to the E. coli transcriptome over time in AOM/Il10^−/− mice

a) Principal Component Analysis plot constructed from the normalized E. coli gene counts from AOM/Il10^−/− mice at all time points. Each symbol indicates an individual mouse at each time-point (white=day 2, grey=week 12, black=week 18). b) Number of differentially expressed genes (FDR corrected P-value < 0.1) in the top 5 most represented KEGG pathways. Positive values on y-axis represent genes up-regulated and negative values represent genes down-regulated relative to day 2 time-point.

Figure 4. Inflammation is required for E. coli-enhanced tumorigenesis in Il10^−/− mice

Histologic scoring of (a) inflammation and (b) tumorigenesis. c) Representative H&E histology at 40X magnification, scale bars indicate 1.0mm, and neoplastic lesion indicated with arrowhead. d) Luminal E. coli load by qPCR of fecal genomic DNA. (a, b, d) Each symbol represents an individual mouse, line at mean, P-values by Kruskal-Wallis with Dunn’s test for multiple comparisons.

Figure 5. Microbial adaptation to the mammalian intestine drives E. coli transcriptional changes over time

a) Timeline of sample collection. b) Principal Component Analysis plot constructed from the normalized E. coli gene counts from all samples and time points. Each symbol indicates an individual mouse at each time-point (white=week 2, grey=week12, black=week 20). Shape indicates genotype/disease: circle=AOM/Il10^−/− n=4 (20 weeks n=3), triangle=Il10^−/− n=4, square=AOM/Il10^−/−;Rag2^−/− n=4. c-d) Venn diagrams of differential expression (FDR corrected P<0.10) in c) AOM/Il10^−/−;Rag2^−/− vs. AOM/Il10^−/− or Il10^−/− (inflammation), d) AOM/Il10^−/− vs. Il10^−/− (cancer).