Interaction between Host MicroRNAs and the Gut Microbiota in Colorectal Cancer

Abstract

Although variation in gut microbiome composition has been linked with colorectal cancer (CRC), the factors that mediate the interactions between CRC tumors and the microbiome are poorly understood. MicroRNAs (miRNAs) are known to regulate CRC progression and are associated with patient survival outcomes. In addition, recent studies suggested that host miRNAs can also regulate bacterial growth and influence the composition of the gut microbiome. Here, we investigated the association between miRNA expression and microbiome composition in human CRC tumor and normal tissues. We identified 76 miRNAs as differentially expressed (DE) in tissue from CRC tumors and normal tissue, including the known oncogenic miRNAs miR-182, miR-503, and mir-17~92 cluster. These DE miRNAs were correlated with the relative abundances of several bacterial taxa, including Firmicutes, Bacteroidetes, and Proteobacteria. Bacteria correlated with DE miRNAs were enriched with distinct predicted metabolic categories. Additionally, we found that miRNAs that correlated with CRC-associated bacteria are predicted to regulate targets that are relevant for host-microbiome interactions and highlight a possible role for miRNA-driven glycan production in the recruitment of pathogenic microbial taxa. Our work characterized a global relationship between microbial community composition and miRNA expression in human CRC tissues. IMPORTANCE Recent studies have found an association between colorectal cancer (CRC) and the gut microbiota. One potential mechanism by which the microbiota can influence host physiology is through affecting gene expression in host cells. MicroRNAs (miRNAs) are small noncoding RNA molecules that can regulate gene expression and have important roles in cancer development. Here, we investigated the link between the gut microbiota and the expression of miRNA in CRC. We found that dozens of miRNAs are differentially regulated in CRC tumors and adjacent normal colon and that these miRNAs are correlated with the abundance of microbes in the tumor microenvironment. Moreover, we found that microbes that have been previously associated with CRC are correlated with miRNAs that regulate genes related to interactions with microbes. Notably, these miRNAs likely regulate glycan production, which is important for the recruitment of pathogenic microbial taxa to the tumor. This work provides a first systems-level map of the association between microbes and host miRNAs in the context of CRC and provides targets for further experimental validation and potential interventions.

Keywords: colorectal cancer; gene regulation; microRNA; microbiome; tumor microenvironment.

FIG 1

Small RNA sequencing data quality. Principal-component analysis showing principal component 1 (PC1) on the x axis and PC2 on the y axis. Each dot is colored according to its normal/tumor status (A), tumor location (B), patient gender (C), patient age (D), raw read count (E), and mature miRNA mapped read count (F). (G) Bar plot of the numbers of mature miRNAs identified in each sample, with coverages over 1 read (gray) and over 5 reads (blue).

FIG 2

Differentially expressed miRNAs between matched normal and tumor samples. Box plot and dot plot showing differentially expressed miRNAs. Each panel represents a single miRNA with a normalized expression level on the y axis. Lines connect a normal and a tumor sample from the same individual, with red lines indicating a higher expression level in tumor tissues and green lines indicating a higher expression level in normal tissues. miR-17, -18a, -20a, -92a, -182, and -503 were found to have significantly higher expression levels in tumor tissues.

FIG 3

Bacteria significantly correlated with DE miRNAs. (A) Heatmap showing bacterial genera (in columns) that were significantly correlated with the DE miRNAs (in rows). Red indicates negative correlations, and green indicates positive correlations. (B) Interaction network showing the nine most significantly DE miRNAs and their correlated bacteria (showing bacteria with a relative abundance of >0.1% and a correlation pseudo-P value of ≤0.05). Edge thickness represents the magnitude of the correlation, with blue indicating negative correlation and with red indicating positive correlation. (C) Heatmap showing the correlations displayed in panel B, with bacterial taxa in columns and miRNAs in rows. Red indicates negative correlations, and green indicates positive correlations.

FIG 4

miRNA target pathways correlated with CRC-associated bacteria. The heatmap shows the predicted pathways of miRNAs (rows) correlated with CRC-associated bacteria (columns) with a q value of <0.01 (modified Fisher exact test; FDR corrected). Positive correlations are shown in blue, and negative correlations are shown in red. The color intensity is shown in a negative log₁₀ scale of FDR-corrected P values from a modified Fisher exact test generated by mirPath, with a darker color indicating a lower q value. CoA, coenzyme A; ECM, extracellular matrix.