Abstract
ATM loss-of-function variants are significantly associated with increased breast cancer risk. ATM exon 7 skipping (△(E7)) is a naturally-occurring alternative splicing event, which introduces a premature termination codon and represents a form of gene-expression regulation via unproductive splicing. Disruption of splicing regulatory elements (SRE) by variants can lead to mis-splicing, potentially contributing to disease susceptibility. To study the regulatory mechanisms of △(E7) and the impact of exonic variants on splicing, a combined in silico/minigene approach was employed, using the construct mgATM_4-9 (exons 4-9) that recapitulates this splicing event. HEXplorer analysis of ATM exon 7 predicted two splicing enhancer-rich regions (c.665-681 and c.867-898). Deletions of these intervals in mgATM_4-9 significantly increased △(E7) (57-96%), revealing their critical role for exon 7 inclusion. Forty-eight candidate variants (HEXplorer, △HZEI score < -40) within these SRE-rich segments were functionally assayed, 34 of which (71%) impaired exon 7 recognition. Nineteen variants presented strong impacts with high expression of △(E7)-transcripts (69-96%), of which c.668A > T, c.680C > A and c.680C > T exhibited particularly strong effects (4-13% full-length transcripts). DeepCLIP analysis suggested that SR proteins SRSF7 and SRSF10 play a positive regulatory role in exon 7 inclusion. Eight variants were classified as likely pathogenic according to ACMG/AMP-based guidelines. Furthermore, nine missense and two synonymous variants with strong impacts (16-29% full-length transcripts) might represent intermediate risk alleles. This work collectively demonstrates the intricate nature of ATM exon 7 recognition, regulated by cis-acting SREs, emphasizing the value of in silico predictions for initial variant filtering and minigene assays for dissecting splicing regulation and clinical interpretation of variants. KEY MESSAGES: Frameshift ATM exon 7 skipping is an unproductive alternative splicing event that may represent a form of gene-expression regulation. Two SRE-rich intervals are involved in exon 7 recognition. Thirty-four out of 48 tested variants located at these intervals disrupted exon 7 inclusion. Nineteen variants showed strong impacts on splicing. All variant types can impair splicing: 23 missense, 7 nonsense, and 4 synonymous variants upregulate exon 7 skipping. Eight variants were classified as likely pathogenic.