Somatic Mutations in miRNA Genes in Lung Cancer-Potential Functional Consequences of Non-Coding Sequence Variants

Abstract

A growing body of evidence indicates that miRNAs may either drive or suppress oncogenesis. However, little is known about somatic mutations in miRNA genes. To determine the frequency and potential consequences of miRNA gene mutations, we analyzed whole exome sequencing datasets of 569 lung adenocarcinoma (LUAD) and 597 lung squamous cell carcinoma (LUSC) samples generated in The Cancer Genome Atlas (TCGA) project. Altogether, we identified 1091 somatic sequence variants affecting 522 different miRNA genes and showed that half of all cancers had at least one such somatic variant/mutation. These sequence variants occurred in most crucial parts of miRNA precursors, including mature miRNA and seed sequences. Due to our findings, we hypothesize that seed mutations may affect miRNA:target interactions, drastically changing the pool of predicted targets. Mutations may also affect miRNA biogenesis by changing the structure of miRNA precursors, DROSHA and DICER cleavage sites, and regulatory sequence/structure motifs. We identified 10 significantly overmutated hotspot miRNA genes, including the miR-379 gene in LUAD enriched in mutations in the mature miRNA and regulatory sequences. The occurrence of mutations in the hotspot miRNA genes was also shown experimentally. We present a comprehensive analysis of somatic variants in miRNA genes and show that some of these genes are mutational hotspots, suggesting their potential role in cancer.

Keywords: TCGA; lung cancer; miRNA; non-coding; somatic mutations.

Figure 1

General characteristics of somatic sequence variants in miRNA genes in LUAD and LUSC. (A) Table summarizing the main statistics and type of identified variants in LUAD and LUSC. (B) Venn diagram showing the overlap of miRNA genes mutated in LUAD and LUSC. (C) Distribution of miRNA gene variants over chromosomes. Gray bars show the number of miRNA genes analyzed in the study (background). Blue and pink bars show the number of mutated miRNA genes in LUAD and LUSC, respectively. Light blue and light pink bars show the number of variants in miRNA genes in LUAD and LUSC, respectively. (D) Correlation of the total number of miRNA genes (x-axis) and the number of mutated miRNA genes (y-axis) on a particular chromosome (dots). Blue and pink dots and regression lines indicate the results for LUAD and LUSC, respectively. Corresponding r² and p-values are indicated on the graph. Dashed lines represent the confidence interval with 95% probability.

Figure 2

Distribution of somatic sequence variants within miRNA precursors. (A) An overview of a primary miRNA transcript with indicated subregions considered in the study. The miRNA duplex is indicated in blue, and sequence consensus motifs recognized as enhancers of miRNA biogenesis are represented by green circles. (B) and (C) The distribution of substitutions in the subregions of miRNA precursors for LUAD and LUSC samples, respectively. miRNA duplex positions are indicated in blue, seed regions in black, and flanking regions and terminal positions of the apical loop in gray. The numbers in the lower-right corner represent the number of plotted substitutions (upper) and the number of mutated miRNA genes (lower). Analyses were also performed in narrowed groups of miRNAs that preferentially release the guide miRNA strand from the 5′ or 3′ arm (below). If present, sequence variants localized beyond position 22 in miRNA are shown cumulatively at position 22. As the size and structure of loops differ substantially among miRNA precursors, the mutation density maps do not show variants located inside the loops.

Figure 3

Consequences of selected seed mutations on a pool of predicted target genes. Venn diagrams showing the effect of representative seed mutations on target recognition. Gray and yellow circles indicate predicted targets of the wild-type and mutant seeds, respectively. If more than one sequence variant occurs in a particular seed, the effect of the second variant is shown as a blue circle. The position in the seed sequence and the nucleotide change are shown next to the corresponding circles.

Figure 4

Localization of mutations in hotspot miRNA genes. Schematic secondary structure representations (generated with mfold) of miRNA precursors are shown. Blue, pink, and orange arrowheads indicate sequence variants detected in the LUAD and LUSC datasets and in the panel of lung cancer samples analyzed experimentally in this study, respectively. Light green circles indicate nucleotide positions within sequence consensus motifs or motifs bound by regulatory proteins. Blue and black bolded fonts indicate mature miRNA and seed sequences, respectively. The first and last nucleotides of miRNA precursors annotated in miRBase are indicated in squares.

Figure 5

Characteristics of mutations in hotspot miRNA genes. (A) Co-mutation plot showing the occurrence of mutations in particular hotspot miRNA genes, all 10 hotspot miRNA genes, and all miRNA genes in cancer samples sorted by cancer type and other cancer characteristics (years smoked, tumor stage, and gender). (B) Co-occurrence of mutations in LUAD samples with mutations in known LUAD driver genes and mutations in LUAD hotspot miRNA genes (all 8 hotspot miRNA genes). Note that only 340 (of 569) LUAD samples with mutations in either known cancer drivers and/or miRNA hotspot genes are shown. The frequency of variants in particular genes is shown on the right.

Figure 6

Impact of miRNA gene mutations on the structure of miRNA precursors. (A) and (B) Representative examples of miR-664b and miR-890 gene mutations, respectively. The effect of each mutation is shown at the levels of secondary and 3D structures. The secondary structures encompass the pre-miRNA precursor sequence and the flanking 25-nt 5′ and 3′ sequences (blue and black bolded fonts indicate miRNA and seed sequences, respectively; pink circles indicate mutation positions). 3D structures encompass sequences indicated by horizontal lines over the corresponding secondary structures. 3D mutant structures (green) are aligned with the corresponding wild-type structures (black). Mutation positions are indicated in pink.