Accurate identification of cancer-predisposing deep intronic variants in tumour-suppressor genes with Oxford Nanopore sequencing

Detecting deep intronic variants is challenging but important

Deep intronic variants are genetic mutations located within introns. Whilst most variants of this type are thought to be benign, some rare variants can alter RNA splicing¹. This alteration in splicing can lead to exonification of intronic sequence, creating pseudoexons and causing abnormal transcriptional consequences such as premature stop codons and loss of gene function². Damaging deep intronic variants are likely to be understudied and underreported causes of disease because they are missed — most clinical genetic tests do not sequence full intronic regions and many of these variants are within complex repetitive areas of the genome that cannot be aligned with legacy short-read sequencing techniques.

Despite advances in the discovery and detection of the genes and variants responsible for inherited predisposition to cancer, a considerable portion of the hereditary risk associated with breast, ovarian, pancreatic, and metastatic prostate among other cancers remains unidentified. Recent research has pointed towards deep intronic variants as an explanation for this missing heritability. James et al. found significant enrichment for rare deep intronic variants in BRCA1, BRCA2, and PALB2 — all tumour-suppressor genes — among patients with familial breast cancer versus older women who are cancer free³. Additionally, a study by Ambry Genetics, which used short-read RNA sequencing, identified likely pathogenic deep intronic variants that were previously classified as variants of uncertain significance⁴.

Investigating cancer predisposition with targeted Oxford Nanopore sequencing

To systematically evaluate deep intronic variants, Gulsuner and AbuRayyan et al. used targeted nanopore sequencing to analyse DNA and cDNA from patient research samples⁵. These samples were taken from families severely affected with breast, ovarian, pancreatic, and/or metastatic prostate cancer, but with no causal variant identified by multiple conventional methods, including exome sequencing and short-read whole-genome sequencing. Oxford Nanopore technology generates reads of any length — from short to ultra long — meaning that, for DNA sequencing, entire intronic regions can be captured within a single read, overcoming the challenges associated with highly repetitive regions; and for cDNA sequencing, single nanopore reads can span multiple exons, revealing the precise locations of splice sites for all transcripts.

The authors carried out two nanopore sequencing workflows in tandem. First, they performed targeted DNA sequencing to identify deep intronic variants in 10 tumour-suppressor genes — BRCA1, BRCA2, PALB2, ATM, CHEK2, BARD1, BRIP1, RAD51C, RAD51D, and TP53. Loss-of-function variants in these genes are widely thought to be responsible for predisposition to breast and ovarian cancer. Then, the researchers used targeted cDNA sequencing to investigate whether the deep intronic variants identified within these genes had an impact on the transcripts.

For DNA sequencing, the authors prepared patient-derived DNA from 240 affected relatives from 120 unsolved families using the Native Barcoding Kit and sequenced the libraries on PromethION Flow Cells for 72 hours. The researchers employed adaptive sampling to enrich for and sequence 1 Mb regions around the 10 tumour-suppressor genes. Adaptive sampling is a unique on-device target enrichment methodology from Oxford Nanopore that allows specific genomic regions to be targeted without upfront wet laboratory sample enrichment (Figure 1), providing high depth of coverage for targets whilst ‘simplifying library preparation’. This simple targeted nanopore workflow revealed 92 rare deep intronic variants in 88 (73%) of the families.

Figure 1. Adaptive sampling uses real-time basecalling: DNA fragments can be accepted or rejected for further sequencing based on their initial sequence composition. A flexible list of regions to be enriched or rejected is provided as a BED file. Adaptive sampling can be implemented in advance of, or even during, a run to increase coverage of specific targets.

The team then used the in silico tools SpliceAI⁶ and Pangolin⁷ to predict which variants would have a functional consequence. This analysis prioritised seven deep intronic variants present in eight families for further evaluation — three in BRCA1, one in PALB2, and three in ATM. To perform targeted cDNA nanopore sequencing, the authors reverse transcribed patient-derived RNA with a pool of gene-specific primers, enriching the candidate genes. Individual cDNA libraries were sequenced on PromethION Flow Cells for 72 hours, resulting in >100x depth of sequencing coverage at the critical gene for each of the samples. All seven variants caused exonification and yielded transcripts with pseudoexons, introducing premature stop codons and ultimately leading to loss of gene function.

The future of deep intronic variant identification

Using targeted DNA and cDNA nanopore sequencing, Gulsuner and AbuRayyan et al. successfully identified rare deep intronic variants contributing to inherited cancer risk in eight of the 120 (6%) previously unsolved families. Nanopore sequencing effectively detected these variants that had been missed by legacy short-read sequencing methods and other conventional approaches.

Therefore, the integration of Oxford Nanopore sequencing into variant discovery workflows can better characterise previously undetected pathogenic variants. This approach has the potential to enhance genetic diagnostics and personalised medicine in the future, ensuring patients with rare deep intronic variants receive more precise risk predictions, leading to tailored therapeutic strategies, such as adjusted cancer screening schedules.