Somatic mutation profiles of MSI and MSS colorectal cancer identified by whole exome next generation sequencing and bioinformatics analysis

Abstract

Background: Colorectal cancer (CRC) is with approximately 1 million cases the third most common cancer worldwide. Extensive research is ongoing to decipher the underlying genetic patterns with the hope to improve early cancer diagnosis and treatment. In this direction, the recent progress in next generation sequencing technologies has revolutionized the field of cancer genomics. However, one caveat of these studies remains the large amount of genetic variations identified and their interpretation.

Methodology/principal findings: Here we present the first work on whole exome NGS of primary colon cancers. We performed 454 whole exome pyrosequencing of tumor as well as adjacent not affected normal colonic tissue from microsatellite stable (MSS) and microsatellite instable (MSI) colon cancer patients and identified more than 50,000 small nucleotide variations for each tissue. According to predictions based on MSS and MSI pathomechanisms we identified eight times more somatic non-synonymous variations in MSI cancers than in MSS and we were able to reproduce the result in four additional CRCs. Our bioinformatics filtering approach narrowed down the rate of most significant mutations to 359 for MSI and 45 for MSS CRCs with predicted altered protein functions. In both CRCs, MSI and MSS, we found somatic mutations in the intracellular kinase domain of bone morphogenetic protein receptor 1A, BMPR1A, a gene where so far germline mutations are associated with juvenile polyposis syndrome, and show that the mutations functionally impair the protein function.

Conclusions/significance: We conclude that with deep sequencing of tumor exomes one may be able to predict the microsatellite status of CRC and in addition identify potentially clinically relevant mutations.

Figure 1. Qualities of the targeted whole exome sequencing approach.

(A) Venn diagram of captured exons of normal and tumor samples. Captured exons with at least one read were counted. (B) Representative normalized coverage-distribution plot. The fraction of bait-covered exons in the genome achieving coverages equal or lower than the normalized coverage is indicated on the x-axis. The mean coverage per exon was divided by the mean coverage of all exons.

Figure 2. Identification process of somatic relevant SNVs.

(A) Schematic of the bioinformatics SNV detection workflow. (B) Extraction of functionally relevant somatic mutations for MSI and MSS colorectal cancers. Variants were detected with the GS Reference Mapper before they were filtered for their localization, annotation in dbSNP130 or the 1000genomes, somatic and functionally impairment. From dbSNP130 or the 1000genomes variants with frequencies above 1% were used. For MSI CRC 359 variants and for MSS CRC 45 with predicted altered protein functions were identified.

Figure 3. Characterization of primary identified SNVs.

(A) Proportional Venn diagram. Fractions of called SNVs identical to the Genomes Project data and dbSNP130. Only data for which the minor allele frequency or the average heterozygosity was known and below 1% were used for comparison. (B) Distribution of synonymous, missense, nonsense and mutations affecting the start or stop codon are shown in relation to all somatic mutations. (C) BMPR1A mutations p.W487R and p.E502G are located at the protein kinase domain of BMPR1A. Reference amino acids are in green, the mutated forms are shown in red. The net structure at the left lower side indicates the ATP binding domain. (D) BMPR1A mutations show decreased signaling acitivity. Activity of wt mBMPR1A, mBMPR1A E502G and mBMPR1A W487R was determined in C2C12 cells using a SMAD-responsive Luciferase reporter gene assay. Induced Luciferase activity was normalized to Renilla acitivty. The activity of untransfected cells was set to 0% and the activity of wt mBmpr1a was set to 100%. Significant differences were calculated with a two-tailed t-test and marked as: * p≤0.05, ** p≤0.01, *** p≤0.001.