Escape from crossover interference increases with maternal age

Abstract

Recombination plays a fundamental role in meiosis, ensuring the proper segregation of chromosomes and contributing to genetic diversity by generating novel combinations of alleles. Here, we use data derived from direct-to-consumer genetic testing to investigate patterns of recombination in over 4,200 families. Our analysis reveals a number of sex differences in the distribution of recombination. We find the fraction of male events occurring within hotspots to be 4.6% higher than for females. We confirm that the recombination rate increases with maternal age, while hotspot usage decreases, with no such effects observed in males. Finally, we show that the placement of female recombination events appears to become increasingly deregulated with maternal age, with an increasing fraction of events observed within closer proximity to each other than would be expected under simple models of crossover interference.

Figure 1. Properties of recombination partitioned by sex and age.

(a) The number of events per meiosis for females (red, n=9152) and males (blue, n=9150), with median values indicated by a vertical line. For phase-unknown individuals, the average number of events per meiosis was used. (b) Squared Pearson correlation between the 23andMe map, the deCODE map and the HapMap map, as a function of scale. (c) The number of recombination events as a function of parental age for females (red, n=9152) and males (blue, n=9150), relative to parents of between 20 and 25 years of age. Parents were grouped into 5-year age bins, and the mean number of events estimated. Error bars show a 95% confidence interval for each group.

Figure 2. Sex differences in recombination hotspot usage.

(a) Hotspot usage for female (red, n=9152) and male (blue, n=9150) meioses. Median values for each sex are shown by vertical lines. (b) Mean hotspot usage, subdivided by parental population. Females are shown in red, males in blue and a combined estimate in black. Error bars indicate a 95% confidence interval.

Figure 3. Estimation of crossover interference parameters.

(a) Fit of three models of interference to the inter-crossover distances observed on chromosome 1, derived from phase-known mothers (red, n=2184) and fathers (blue, n=2092). The interference-free model is shown as a dotted line, the gamma simple interference model is shown as a dashed line and the Housworth–Stahl interference escape model is shown as a solid line. (b) Per-chromosome estimates of the interference parameter as estimated from the Housworth–Stahl interference escape model. Error bars indicate a 95% confidence interval. Note that chr21 in males is excluded due to an extremely high estimate. (c) Per-chromosome estimates of the proportion of events escaping interference. Error bars indicate a 95% confidence interval.

Figure 4. Departures from simple crossover interference.

(a) Inferred escape parameter as a function of maternal age. Mothers were divided into 10 approximately equal-sized deciles on the basis of age, and the Housworth–Stahl interference escape model was fitted for each group separately. The inset shows the estimates of the escape parameter when considering phase-known (blue, n=2184) and phase-unknown (green, n=6968) individuals separately. Estimates for ν show no correlation with age (Supplementary Fig. 9). Error bars indicate 95% confidence intervals. (b) Distribution of inter-crossover distances for young and old mothers, where the boundary between young and old is taken as median maternal age (30 years). Error bars represent a 95% confidence interval assessed via 1000 bootstrap samples, and the arrow highlights a significant difference between the young and old groups for tightly clustered events. The inset shows the cumulative distribution function (CDF) up to 5 cM. (c) Distribution of inter-crossover distances for young and old fathers, where the boundary between young and old is taken as median paternal age (32 years).