OR03: Functional assays and bioinformatics predictions reveal a high contribution of splicing mutations in the most frequent forms of hereditary cancer

Pascaline Gaildrat1,2, Omar Soukarieh1,2, Gaia Castelain1,2, Hélène Tubeuf1,2, Sophie Krieger1,2,3, Stéphanie Baert-Desurmont1,2,4, Daniela Di Giacomo1,2, Mohamad Hamieh1,2, Sandrine Caputo5, Julie Abdat1,2, Audrey Killian1,2, Jean-Christophe Thery1,2, Isabelle Tournier1,2, Céline Bonnet1,2, Samira Aklil1, Grégoire Davy1,2,3, Aurélie Drouet1,2, Myriam Vezain1,2, Etienne Rouleau5, Claude Houdayer5,6, Thierry Frebourg1,2,4, Mario Tosi1,2, Alexandra Martins1,2 in collaboration with the French Oncogenetics Network

1 – Inserm U1079-IRIB, Rouen University, France.2 – Normandy Centre for Medical Genomics and Personalized Medicine, France. 3 – Department of Clinical Biology and Oncology, François Baclesse Cancer Centre, Caen, France. 4 – Department of Genetics, University Hospital, Rouen, France. 5 – Genetics Department, Curie Institute, Paris, France. 6 – University Paris Descartes, Paris, France.

Aim

The identification of a causal mutation is essential for molecular diagnosis and clinical management of hereditary cancers. Even if DNA-seq has greatly improved the detection of nucleotide changes, the biological interpretation of most variants remains challenging.

Method

We performed minigene assays to evaluate the impact on RNA splicing of more than 600 variants identified in genes implicated in Lynch syndrome or in hereditary breast and ovarian cancer syndrome, notably MLH1/MSH2(MMR), BRCA1/BRCA2(BRCA). Patient RNA was also analyzed, when available. Experimental results were compared to bioinformatics predictions generated by using both conventional and newly developed algorithms.

Results

We found that more than 25% of variants of unknown significance in the MMR/BRCA genes have an impact on RNA splicing. Furthermore, our targeted studies on “model-exons”, including MLH1 exon 10 and BRCA2 exon 7, revealed an unexpected large number of variants altering potential exonic splicing regulatory elements (ESR), an effect that could be predicted by two newly developed ESR-dedicated in silico tools, but not by commonly used bioinformatics approaches.

Conclusion

This work revealed the important contribution of splicing mutations in hereditary cancer, contributed to the clinical classification of several variants, and pinpointed the potential of new prediction methods as filtering tools for prioritizing variants for functional analyses.

Top