Genetic susceptibility to earlier ovarian ageing increases de novo mutation rate in offspring

Abstract

Human genetic studies have provided substantial insight into the biological mechanisms governing ovarian ageing, yet previous approaches have been largely restricted to assessing common genetic variation. Here we report analyses of rare (MAF<0.1%) protein-coding variants in the exomes of 106,973 women from the UK Biobank study, implicating novel genes with effect sizes up to ~5 times larger than previously discovered in analyses of common variants. These include protein truncating variants in ZNF518A, which shorten reproductive lifespan by promoting both earlier age at natural menopause (ANM, 5.61 years [4.04-7.18], P=2*10-12) and later puberty timing in girls (age at menarche, 0.56 years [0.15-0.97], P=9.2*10-3). By integrating ChIP-Seq data, we demonstrate that common variants associated with ANM and menarche are enriched in the binding sites of ZNF518A. We also identify further links between ovarian ageing and cancer susceptibility, highlighting damaging germline variants in SAMHD1 that delay ANM and increase all-cause cancer risk in both males (OR=2.1 [1.7-2.6], P=4.7*10-13) and females (OR=1.61 [1.31-1.96], P=4*10-6). Finally, we demonstrate that genetic susceptibility to earlier ovarian ageing in women increases de novo mutation rate in their offspring. This provides direct evidence that female mutation rate is heritable and highlights an example of a mechanism for the maternal genome influencing child health.

Characterising the contribution of rare protein-coding germline variants to prostate cancer risk and severity in 37,184 cases

“…We identified rare protein-truncating variants in the DDR genes BRCA2 (OR = 3.23 [2.65-3.90], P = 7.5×10 -29 ) and ATM (OR = 2.92 [2.34-3.63], P = 1.17×10 -19 ), and additionally rare damaging variants in SAMHD1 (OR = 2.02 [1.65-2.45], P = 2.36×10 -11 ) as significantly associated with increased prostate cancer risk (Figures 1 & 2, Table 2 and Supplementary Table 3). Germline variants in SAMHD1 have recently been reported as associated with the risk of prostate cancer in an analysis of the UK Biobank only 24 , and additionally, there is evidence of association with breast cancer 25 and primary cancers collectively 26,27 . Here, we independently validate the SAMHD1 association with prostate cancer in the UK Biobank and replicate the finding in additional cohorts (Supplementary Table 3).…”

Monogenic causes of Premature Ovarian Insufficiency are rare and mostly recessive

“…This includes two green Panel App genes -NOBOX and POLG -where 137/139 and 52/55 of the identified PTV alleles, respectively, were found in controls. Our previous work demonstrated that heterozygous LOF of ZNF518A has the largest effect in the protein-coding genome on menopause timing 27 , yet carriers report menopause only 6 years earlier than noncarriers, with only 12% experiencing POI. Taken together, our observations suggest that fully, or even largely, penetrant autosomal dominant effects are likely to cause very few cases of POI.…”

“…For 100 of the 105 POI genes, we did not find an association with ANM (P<1.6*10 -4 ; P=0.05/(3 tests × 105 genes) (Supplementary Table 5). For two AR genes, we have previously reported an effect on ANM: BRCA2 (P=2.6*10 -8 ; beta:1.32 years earlier ANM [95% CI: -1.79, -0.85]) and HROB (P=4.7*10 -7 ; beta: 2.69 years earlier ANM [95% CI: -3.73, -1.65]) 27 . There were novel associations with earlier ANM for a further two AD and one AR genes, with at least one of the variant categories passing our threshold for multiple testing (P<1.6*10 -4 , Figure 2).…”

“…All manipulations were conducted in R (v4.1.2) on the UK Biobank RAP. Rare variant burden tests of functional variant categories (defined as for main analyses) were performed using a custom implementation of BOLT-LMM v2.3.6 50 for the UK Biobank RAP, as described in Stankovic et al (2022) 27 . Analyses used a winsorised ANM phenotype, with everyone reporting ANM at younger than 34 years given a value of 34.…”

“…A second analysis team (Cambridge) independently performed analyses of WES data in UK Biobank. The ANM phenotype was derived as described in Stankovic et al (2022) 27 . Briefly, a different approach was used to generate the phenotype by handling data from multiple visits and missing data differently to the main method of generating the phenotype.…”