Postzygotic single-nucleotide mosaicisms in whole-genome sequences of clinically unremarkable individuals

Abstract

Postzygotic single-nucleotide mutations (pSNMs) have been studied in cancer and a few other overgrowth human disorders at whole-genome scale and found to play critical roles. However, in clinically unremarkable individuals, pSNMs have never been identified at whole-genome scale largely due to technical difficulties and lack of matched control tissue samples, and thus the genome-wide characteristics of pSNMs remain unknown. We developed a new Bayesian-based mosaic genotyper and a series of effective error filters, using which we were able to identify 17 SNM sites from ~80× whole-genome sequencing of peripheral blood DNAs from three clinically unremarkable adults. The pSNMs were thoroughly validated using pyrosequencing, Sanger sequencing of individual cloned fragments, and multiplex ligation-dependent probe amplification. The mutant allele fraction ranged from 5%-31%. We found that C→T and C→A were the predominant types of postzygotic mutations, similar to the somatic mutation profile in tumor tissues. Simulation data showed that the overall mutation rate was an order of magnitude lower than that in cancer. We detected varied allele fractions of the pSNMs among multiple samples obtained from the same individuals, including blood, saliva, hair follicle, buccal mucosa, urine, and semen samples, indicating that pSNMs could affect multiple sources of somatic cells as well as germ cells. Two of the adults have children who were diagnosed with Dravet syndrome. We identified two non-synonymous pSNMs in SCN1A, a causal gene for Dravet syndrome, from these two unrelated adults and found that the mutant alleles were transmitted to their children, highlighting the clinical importance of detecting pSNMs in genetic counseling.

Figure 1

A new computational pipeline for genome-wide identification of pSNMs without matched control tissue samples. (A) Overall framework of the pipeline including read pre-processing, genotyping and filtering. The processes of mosaic identification and filtering were implemented in our scripts. (B) The Bayesian-based genotyper demonstrated as a probabilistic graphical model. Four genotypes were defined: ref-hom for “homozygous for the reference allele”, het for “heterozygous”, alt-hom for “homozygous for the alternative allele”, and mosaic for “mosaic”. The posterior probabilities were inferred from prior and conditional probabilities that were calculated or simulated from known population genetics data and next-generation sequencing data (see Materials and Methods). (C) Simulated behavior of the Bayesian genotyper when the sequencing depth and base quality varied. The X axis denotes the alternative allele fraction. The Y axis denotes the posterior probability of the four genotypes. Columns 1 to 4 represent sequencing depths of 20, 40, 80, and 160, respectively, and rows 1 to 3 represent base qualities of 10, 20, and 30, respectively. It showed that increasing sequencing depth could improve the power to distinguish between mosaic and heterozygous sites, whereas increasing base quality could be helpful in distinguishing between mosaic and homozygous sites. (D) The power to distinguish mosaic sites from the simulated ∼20 000 homozygous and ∼20 000 heterozygous sites by sequentially applying the Bayesian genotyper and each of the ten error filters. This result demonstrates the high specificity of our pipeline in excluding germline sites and the relative contribution of the genotyper and filters.

Figure 2

Specificity and precision of identifying pSNMs using our pipeline, Varscan 2, and muTect. (A-B) False positive rate for true reference sites (A) and true non-reference sites (B). Error bars: 95% confidence intervals. (C) Proportion of true pSNMs among all identified sites. The postzygotic mutation rate was set to 4.4 × 10⁻⁷ per base, based on estimates in this study. The X axis denotes the minor allele fraction of the pSNM sites.

Figure 3

Identification and validation of pSNMs in the whole-genome sequences from peripheral blood samples of three individuals. (A-C) Pedigree structures of the three participating families. Red arrows point to the three individuals selected for whole-genome sequencing. (D-F) Alternative allele fractions and sequencing depth of the pSNMs identified in the individuals ACC1-II-1 (D), DS1-II-2 (E), and DS2-I-1 (F) using our pipeline. The candidate pSNM sites are shown in red along with the germline sites shown in gray. The sites with extreme depth or allele fraction are not shown. Different shades of red represent mosaic posterior probabilities. (G) Validation by pyrosequencing. The X axis shows the pSNMs identified and validated in the three individuals. pSNM site IDs correspond to Table 2. The Y axis shows the alternative allele fractions of the pSNM sites in the case, unrelated control, and parents, detected by pyrosequencing. The dashed line represents allele fraction of 0.05, which is the detection threshold of pyrosequencing. (H) Copy number abnormalities are ruled out for all but two of the pSNM sites using MLPA. Seventeen sites showed normal copy numbers with normalized signal ratios between 0.7 and 1.3. Two sites with extra DNA copies are marked by asterisks. Error bars represent the SD of three replications of MLPA. (I) Correlation of the minor allele fractions estimated by whole-genome sequencing and pyrosequencing of the validated sites. The sequencing depth is represented by the size of the dots.

Figure 4

Characteristics of the validated pSNMs. (A) The mutation spectrum of validated pSNMs. The C→T and C→A mutations accounted for half of the mutations identified at the mosaic sites. (B) Allele fractions of the pSNM sites in different samples within the same individuals. Three pSNMs were detected in both somatic and semen samples. (C) Similarity of the allele fractions of the pSNMs between different samples. The blood and saliva samples showed the highest similarities in the six samples.

Figure 5

pSNMs detected in DS1-II-2 and DS2-I-1 in the gene SCN1A and transmitted to their respective child with Dravet syndrome as a heterozygous mutation. (A) The two non-synonymous mutations are highlighted by red arrows on transmembrane structure of the sodium channel encoded by SCN1A. These mutations alter residues located at the ends of the loop structures in domains III and IV, adjacent to previously known pathogenic mutations in Dravet syndrome which are shown here as small circles with different colors representing different mutation type. (B) The parent-to-offspring transmission model is illustrated for c.5003C→G pSNM. In the mother, the mutant allele generated by postzygotic mutations is present in a proportion of the cell population and identified by our pipeline as a pSNM. The mosaicism apparently affected germ cells, and thus the offspring had a chance to inherit the mutation during gametogenesis and fertilization, leading to the heterozygous genotype.