Variants not meeting PVS1
We found 6 missense variants with discordant classifications of pathogenicity in ClinVar (Table 1) in 13 unrelated patients. In addition, one of the healthy (non-cancer) controls carried theCHEK2 c.349A>G, p.(Arg117Gly) variant. To better interpret missense variants, a comprehensive review of previous functional studies was done, main results are summarized in Table S3 and Table S4. In a further effort to improve variant classification, after classical ACMG we also followed the allele risk criteria reported recently (Senol-Cosar et al., 2019). For this, we searched for association studies of our CHEK2 variants (Table S2).
CHEK2c.190G>A, p.(Glu64Lys) is located in a weakly conserved amino acid in the SQ/TQ cluster domain (SCD). It is predicted deleterious by in silico analysis. It shows a partially reduced phosphorylation by ATM at the Thr68 residue, as well as partially reduced auto-phosphorylation and Cdc25C phosphorylation. It affects KAP1 phosphorylation and has discrepant results about DNA damage response (Table S3). Furthermore, there are no high-quality case-control studies. Therefore, this variant only meets PP3 criterion, remaining as VUS (Table 1). Variant c.349A>G, p.(Arg117Gly) affects a highly conserved amino acid (class C65 according to GVGD) in the FHA domain. It is predicted deleterious by in silico analysis. It does not affect phosphorylation by ATM nor oligomerization, but affects all the rest of the studied protein functions (Table S3). This variant accomplished PS3, PS4_moderate, PM1 and PP3 criteria, being classified as LP by all frameworks. It has been studied in a large high-quality case-control study, reporting a BC OR of 2.26 (95% CI, 1.29-3.95) (Table 1), therefore it could be considered as LRA within the Senol-Cosar framework (Table S2). Variant c.433C>T, p.(Arg145Trp) is located in a moderately conserved amino acid of the FHA domain. It is predicted deleterious by in silico . It reduces CHK2 expression and stability. In functional assays, it has been consistently reported to impair kinase and DNA repair activity. Evidences for classification includes PS3, PM1 and PP3, being classified as LP by all frameworks. Variant c.470T>C, p.(Ile157Thr) lies in a weakly conserved amino acid of the FHA domain. It is predicted deleterious by in silico analysis. It has been widely studied, nevertheless the functional assays reported to date show discordant results (Table S3). The reported OR in the biggest CHEK2meta-analysis was 1.58 (95% CI, 1.42 - 1.75), therefore PS4_moderate was applied, application of PP3 was not enough to classify this variant as LP/P. However, following recommendations from Senol-Cosar et al, it would be an ERA due to the existence of multiple case-control studies (Senol-Cosar et al., 2019). Variant c.499G>A, p.(Gly167Arg) is located in a highly conserved amino acid of the FHA domain. It is predicted deleterious by in silico analysis. Although there are only 2 functional studies, they both reported an impaired DNA repair activity in yeast assays (Table S3). PS3, PM1, PM2_supporting and PP3 were assigned, being classified as LP by all frameworks. Variant c.1427C>T, p.(Thr476Met) lies in a moderately conserved amino acid of the kinase domain. It is predicted deleterious by in silico analysis. Functional assessment of KAP1 phosphorylation results deleterious in vitro and good enough in vivo . Furthermore, SOX phosphorylation was reported equal to that of the pathogenic c.1100delC variant. Assays on DNA repair activity have found it damaging or with intermediate activity (Table S3). Due to these discordant functional assay results PS3 was not weighted. Classification remained as VUS since c.1427C>T only accomplished PP3.