Gut microbiota composition in colorectal cancer patients is genetically regulated

Abstract

The risk of colorectal cancer (CRC) depends on environmental and genetic factors. Among environmental factors, an imbalance in the gut microbiota can increase CRC risk. Also, microbiota is influenced by host genetics. However, it is not known if germline variants influence CRC development by modulating microbiota composition. We investigated germline variants associated with the abundance of bacterial populations in the normal (non-involved) colorectal mucosa of 93 CRC patients and evaluated their possible role in disease. Using a multivariable linear regression, we assessed the association between germline variants identified by genome wide genotyping and bacteria abundances determined by 16S rRNA gene sequencing. We identified 37 germline variants associated with the abundance of the genera Bacteroides, Ruminococcus, Akkermansia, Faecalibacterium and Gemmiger and with alpha diversity. These variants are correlated with the expression of 58 genes involved in inflammatory responses, cell adhesion, apoptosis and barrier integrity. Genes and bacteria appear to be involved in the same processes. In fact, expression of the pro-inflammatory genes GAL, GSDMD and LY6H was correlated with the abundance of Bacteroides, which has pro-inflammatory properties; abundance of the anti-inflammatory genus Faecalibacterium correlated with expression of KAZN, with barrier-enhancing functions. Both the microbiota composition and local inflammation are regulated, at least partially, by the same germline variants. These variants may regulate the microenvironment in which bacteria grow and predispose to the development of cancer. Identification of these variants is the first step to identifying higher-risk individuals and proposing tailored preventive treatments that increase beneficial bacterial populations.

Figure 1

Microbiota diversity in resected colorectal mucosa from patients with colorectal cancer. Data grouped by tumor site, according to the V4-V5-V6 dataset. (A) Chao1 estimator, P = 0.049, Kruskal–Wallis rank-sum test with continuity correction. (B) Number of observed operational taxonomic units (OTUs), P = 0.049. The box marks the interquartile range, and the central horizontal line marks the median. Outliers (extreme values, > 1.5 times the interquartile range) are shown as circles.

Figure 2

Manhattan plots of Shannon index-specific mbQTLs. (A) V1-V2-V3 dataset. (B) V4-V5-V6 dataset. Red horizontal line, threshold for genome-wide significance (P = 5.0 × 10^–8). Blue horizontal line, FDR = 0.10.

Figure 3

Manhattan plots of OTU-specific mbQTLs. (A) Bacteroides V1-V2-V3, (B) Bacteroides V4-V5-V6, (C) Ruminococcus V4-V5-V6, (D) Akkermansia, V4-V5-V6 (E) Faecalibacterium and (F) Gemmiger V4-V5-V6. Red horizontal line, threshold for genome-wide significance (P = 5.0 × 10^–8). Blue horizontal line, FDR = 0.10.