Targeted deep-intronic sequencing in a cohort of unexplained cases of suspected Lynch syndrome

Abstract

Lynch syndrome (LS) is caused by germline defects in DNA mismatch repair (MMR) pathway, resulting in microsatellite instability (MSI-H) and loss of immunohistochemical staining (IHC) of the respective protein in tumor tissue. However, not in all clinically suspected LS patients with MSI-H tumors and IHC-loss, causative germline alterations in the MMR genes can be detected. Here, we investigated 128 of these patients to possibly define new pathomechanisms. A search for large genomic rearrangements and deep-intronic regulatory variants was performed via targeted next-generation sequencing (NGS) of exonic, intronic, and chromosomal regions upstream and downstream of MLH1, MSH2, MSH6, PMS2, MLH3, MSH3, PMS1, and EPCAM. Within this cohort, two different large rearrangements causative for LS were detected in three cases, belonging to two families (2.3%). The sensitivity to detect large rearrangements or copy number variations (CNV) was evaluated to be 50%. In 9 of the 128 patients (7%), previously overlooked pathogenic single-nucleotide variants (SNV) and two variants of uncertain significance (VUS) were identified in MLH1, MSH2, and MSH6. Pathogenic aberrations were not found in MLH3, MSH3, and PMS1. A potential effect on regulation was exerted for 19% of deep-intronic SNVs, predominantly located in chromosomal regions where the modification of histone proteins suggests an enhancer function. In conclusion, conventional variation analysis of coding regions is missing rare genomic rearrangements, nevertheless they should be analyzed. Assessment of deep-intronic SNVs is so far non-conclusive for medical questioning.

Fig. 1. Target region of the deep-intronic sequencing kit.

The six tracks show the complete chromosomal region that is covered by the custom-designed kit. The four genes associated with LS (a–c) are highlighted in red and the four genes involved in the MMR pathway (d–f) are highlighted in green. The target region includes the following chromosomal positions: chr2:47,500,000–48,070,000; chr2:190,625,643–190,768,067; chr3:36,964,000–37,115,000; chr5:79,927,190–80,260,701; chr7:5,996,249–6,078,750; chr14:75,469,099–75,525,752 (color figure online).

Fig. 2. Overview of annotations for all coding and deep-intronic SNV calls detected in the whole-patient cohort.

Functional annotations and linkage disequilibrium information for all detected SNVs were received with the SNiPA tool. The majority of variants are located in intronic regions with a putative effect on transcript.

Fig. 3. Distribution of non-coding variants in putative regulatory regions.

Ratios of observed versus expected number of variants per regulatory region track are shown in a heatmap (a) and as a QQ-plot (b). It was assumed that the variants are evenly distributed in the target region and consequently, the expected number of variants was defined by the size of the respective regulatory region track. 5ʹUTR variants occur more frequently in DNase-hypersensitivity regions and transcription factor-binding sites (b).

Fig. 4. Results of rearrangement detection.

a Schematic view of two duplication events in the control cohort. In these two cases, rearrangement analysis of the deep-intronic sequencing data revealed the exact breakpoints and, thereby, also the chromosomal locus of the duplicated parts. The green parts are duplicated. b In one sample of the patient cohort, we detected a complex rearrangement on chromosome 2 that causes a paracentric inversion between the DCLK3 gene and MLH1. The same structural variation was identified in two additional family members. c In another patient, we identified a gross insertion (2.2 Kb) of a 5ʹ truncated SVA element in PMS2 intron 7 (color figure online).