Parental germline mosaicism in genome-wide phased de novo variants: recurrence risk assessment and implications for precision genetic counselling

doi:10.21203/rs.3.rs-4874550/v1

Download PDF

Research Article

Parental germline mosaicism in genome-wide phased de novo variants: recurrence risk assessment and implications for precision genetic counselling

https://doi.org/10.21203/rs.3.rs-4874550/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Background:

De novo mutations (DNMs) significantly impact health, particularly through developmental disorders. DNMs occur in both paternal and maternal germlines via diverse mechanisms including parental early embryonic mosaicism, which increases recurrence risk for future pregnancies through germline mosaicism. Embryonic mosaicism is divided based on primordial germ cell specification (PGCS): pre-PGCS events may affect both germline and somatic tissues, while post-PGCS events are only found in the germline. The specific contribution of germline mosaicism to DNMs across the genome is not well defined. We aimed at categorizing DNMs and their recurrence risk by detecting a large set of DNMs followed by systematic deep sequencing of parental blood and sperm DNA.

Methods:

We performed trio-based short-read genome sequencing for initial DNM detection and long-read genome sequencing for phasing, followed by high-depth targeted sequencing of parental blood and paternal sperm to detect germline mosaicism.

Results:

We detected a total of 428 DNMs (on average 85.6 per trio, n = 5 trios), with an expected paternal bias of 80%. Targeted resequencing of parental blood and sperm (depth > 5000x) unveiled 20/334 parental germline mosaics (2–5 per trio) with variant allele fractions (VAFs) ranging from 0.24–14.7%, including 7 that were detected in paternal sperm exclusively (1–2 per trio). We estimate that individual genomes harbour about 2 paternal and 2 maternal pre-PGCS DNMs and 2 paternal post-PGCS DNMs (detectable in sperm only). Due to paternal bias, maternally phased variants appear 3.4x more likely to be mosaic in blood. By using average VAFs in sperm as a direct indicator, we estimate recurrence risk of genome -wide paternally phased de novo variants to be 0.3%, prior to any sperm sequencing assessment. This estimate is an average between a majority of variants with a null recurrence risk and a handful of variants with a high recurrence risk.

Conclusions:

Genetic counselling of DNM may not rely anymore on empirical estimates of recurrence risk. Sperm sequencing may be an effective method to reliably specify the recurrence risk of most individual DNMs. Long-read sequencing, allowing the phasing of DNMs, may also become critical in this process.

De novo mutations

DNMs

germline mosaicism

long-read genome sequencing

sperm sequencing

smmip

recurrence risk

De novo mutations (DNMs) represent a pool of 50–120 sequence variants that are present in a child but absent in their parents' genomes. This natural and evolutionary constrained phenomenon is however a major source of genetic diseases and it is estimated that about one birth in 300 is subject to a severe developmental disorder caused by a DNM in the coding sequence [1]. Trio-based genome sequencing studies have shown a high paternal bias as on average 75–80% of DNMs occur on the paternal haplotype, highlighting significant differences in mutability between female and male germlines [2–4]. DNMs also harbour a strong paternal age effect such as paternal age at conception is a major determinant in the count of DNMs [2, 5]. It is well recognized that DNMs constitute a composite assembly of distinct types of mutational events regarding the timing and the cells in which they appear along the germline, from the zygote to the germ cells in both sexes [6, 7]. The magnitude of paternal bias and paternal age effect implies that mutational events occurring in spermatogonia in adult men spermatogenesis are a common cause of DNMs. Subsequent to this kind of mutational event, 50% of sperm cells produced by a mutated spermatogonia are expected to harbour the variant. However, since sperm is produced from millions of spermatogonia, the probability that the same mutation recurs in multiple children (i.e. originating from the same cell) is considered negligible [6]. In contrast, DNMs can result from events occurring in early embryonic cells in one parent. In these situations, mutations may be present in a significant proportion of germ cells (i.e. quiescent oocytes or spermatogonia), and therefore be at high risk of recurrence for future pregnancies.

These two types of mutational events in spermatogonia and early embryonic cells exemplify the heterogeneity within DNM events, regarding both mechanism and risk of recurrence in siblings. This latter property has major implications for genetic counseling in DNM-mediated genetic diseases. The phenomenon of germline mosaicism has long been recognized and led to the widely accepted understanding that de novo variations carry a recurrence risk of approximately 1% for subsequent pregnancies [8]. Many families in which a child carries a severe genetic disease caused by a DNM do worry about a possible recurrence in subsequent pregnancies and frequently resort to invasive fetal genotyping procedures [9]. However, this 1% estimate represents an average between a majority of families with negligible or absence of recurrence risk, notably following spermatogonia events (or more broadly “one-off” events, [7]), and families at high risk of recurrence in case of germline mosaicism. Given the considerable impact of DNMs in certain pathologies and their increased detectability thanks to sequencing advances, a finer stratification of DNMs according to mutational event type is needed for clinical care.

One key step in the biology of germline development is the individualisation of the germline from the soma. This phenomenon, called primordial germ cell specification (PGCs) occurs early during human embryogenesis at about embryonic day 17 [10], and leads to the specification of 20 to 40 cells [11] after about 10–15 mitotic divisions. Variation occurring before this stage may be present in both the germline and the soma, in the form of “mixed somatic and germline mosaicism” [12] detectable in somatic tissues, while variants occurring after PGCS can only be clonal in the germline (“confined germline mosaicism”). Many studies have aimed to assess the recurrence risk of specific pathogenic variants by detecting these two kinds of germline mosaicism using deep sequencing in somatic or sperm samples (Supplementary Fig. 1, Supplementary Table 1). In contrast, few studies systematically analysed genome-wide DNMs for parental mosaicism, and the prevalence of low-level confined germline mosaicism is not well studied.

In this study, we aimed at categorizing a set of genome-wide DNMs, by (i) detecting DNMs and systematically phasing them using long-read genome sequencing, (ii) performing targeted deep sequencing of parental blood samples and (iii) targeted deep sequencing of paternal sperm samples.

Patients and samples

Five trios consisting in one child and both parents were included, as previously described ([13]). EDTA blood samples were collected for each individual, as well as sperm samples for the five fathers. DNA was extracted from blood using standard procedures for short-read based sequencing techniques. Longer fragments were also extracted from peripheral blood mononuclear cells (PBMCs) using Revolugen kit for 4 trios, and from frozen blood using Circulomics kit for one. Sperm samples were collected into a sterile container (Clinisperm®, CML, Nemours, France) directly in the Rouen University Hospital Reproductive Biology Laboratory-CECOS after sexual abstinence for 3 to 5 days according to WHO quality guidelines. A liquefaction time of 20–30 min was allowed before sperm freezing in straws (Spermfreeze, dilution ½, JCD International Laboratory, Lyon, France). A one-layer gradient was prepared using the 90% fractions of Puresperm® (JCD International Laboratory, Lyon, France) diluted in IVF medium® (Origio, CooperSurgical, Måløv, Denmark) and centrifuged at 150 x g for 20 min. Then, the 90% fraction was washed with IVF medium® by centrifugation at 350 × g for 10 min. DNA was extracted from sperm pellet using the TCEP-based method from Wu et al. [14].

This study was approved by the CPP Ouest V (20/043 − 2) ethics committee. Informed written consent was obtained for all families.

Genome sequencing

Short-read genome sequencing was performed at the Centre National de Recherche en Génomique Humaine (CNRGH, Institut de Biologie François Jacob, CEA, Evry, France), using paired-end 150bp reads on NovaSeq6000 and targeting an average sequencing depth of ~ 40x. Long-read genome sequencing was performed by CNRGH on Oxford Nanopore Promethion using R9 chemistry after a preparation using SQK-LSK109 or SQK-LSK110 ligation kits. Further details on short and long-read sequencing procedures for these five trios are available elsewhere [13].

De novo variant identification and phasing

De novo single nucleotide variant (SNV) and short insertion/deletion (indel) candidates were identified from 43x short-read genome data. Reads were aligned on GRCh38 using BWA and short variants were called using Deepvariant V1.5 using default parameters for Illumina WGS. A 15 samples multi-vcf was produced using Glnexus. A two-step workflow was applied to isolate high quality de novo variants. First, de novo SNV and indel candidates were detected by applying simple filtration steps using a bcftools-based custom python script (DeNovoMiner.py, [15]). These filters included genotype (GT = alt in child and ref in both parents), depth (DP > 20 in all three individuals), genotype quality (GQ > 29 in all three individuals), variant allele fraction (VAF > 0.25 in child), an exclusion of multi allelic loci (AD1 + AD2 > 0.7 x DP), and a shift of VAF between each parent and the child of at least 4x. This last requirement was used to avoid using strict alt read counts or VAF in parents and allow for the detection of cases of parental mosaicism. The second step consisted in a manual reviewing of de novo candidate calls using an IGV based classifier interface ([16]). Substitution based signatures were extracted using the Signal software [17].

Variant phasing (i.e. identification of the parental haplotype on which the variant occurred) was achieved using short and long-read data. Long read genomes were aligned on GRCh38 using Minimap2. SNVs and indels detected from short-read WGS were phased in trios using long-read information by WhatsHap phase. Because the WhatsHap version used did not allow for phasing of de novo variants directly, we used a manual method based on manual inspection of long-read haplotypes. Phased VCF was used to add the phase to individual Nanopore reads using WhatsHap haplotag, and a manual reviewing of the alignments was applied for a definite parental haplotype attribution for each de novo variant. Variants were also phased using short reads only, using the Unphazed software [18].

Targeted deep sequencing

Deep sequencing at DNM positions was performed in child and parental blood samples as well as paternal sperm samples by smMIP (single molecule molecular inversion probes)-based sequencing, similar to previously described [19]. One smMIP was designed around each DNM position using MIPGEN using arms_length_sums = 38 and varying capture_size from 90 to 110. Ten nucleotides of unique molecular identifiers were used (2 x 5nt) to allow a maximum of 4¹⁰ (1048576) combinations. Count of occurrences of extension and ligation probes in the reference genome provided by MIPGEN were used to exclude smMIPs, if either one of the two arms had a sequence occurring > 20 times or both arms had multiple occurrences, and in silico PCR (UCSC, default parameters) led to more than one result. Final design of 346 oligos (Supplementary Table 3) was produced by IDT DNA technologies. Individual smMIPS were pooled and phosphorylated. An amount of 300ng of input DNA was used for smMIP capture with a 1:4000 ratio (1 genome copy for 4000 smMIP molecules). Capture product was then amplified and indexed by a 16-cycle PCR. Libraries were pooled and sequenced on three High 2x75 flowcells on an Illumina NextSeq500 sequencer. Deep sequencing reads were aligned on reference genome and duplicates were removed using UMI-tools. Variant allele fraction (VAF) and sequencing depth were assessed for all variants in all samples using Samtools mpileup, launched via a python script ([20]).

Mosaic variant identification and statistics

For each variant, VAF and sequencing depth were established from father’s blood, sperm, mother’s blood and controls from the sequencing pool. VAF was defined as the proportion alt_read_count / (ref_read_count + alt_read_count). Controls consisted in child and three parental samples for the 4 other trios (16 samples total). To detect candidate mosaic variants, VAFs in father’s blood, sperm or mother’s blood were compared to the VAF in merged controls. To account for extremely low allelic ratios among controls, we adopted a one-sided Poisson test. Thanks to phasing, not all three mosaicisms had to be tested at every position. When the child’s variant could be phased to the paternal haplotype, potential mosaicism was searched within father’s blood and sperm only. When the variant was of maternal origin, potential mosaicism was obviously searched within maternal blood only. When phasing was not possible, father’s blood and sperm, as well as mother’s blood were investigated for parental mosaicism. As a result, a Bonferroni correction was applied to account for a total of 637 haplotype-coherent tests, with the requirement of an overall type-I error threshold of 0.05/637 = 7.8x10-5 for experiment-wide significance.

Candidate mosaic variants were confirmed on the parental samples only using a deeper sequencing assay on independent smMIP captures using a restricted pool of 40 smMIP, sequenced on a 2x75 flowcell on an Illumina NextSeq500 sequencer.

Establishment of a set of high quality phased de novo mutations

We used short read genome sequencing to call a set of 428 high-confidence DNMs in five families (Supplementary Table 2), ranging from 56 to 119 per individual, with a mean of 77 SNVs and 9 indels (Fig. 1). Targeted smMIP sequencing on 349 variations accessible to a MIP design showed a very low false positive rate, with only 1 variant that appeared inherited and 348 true de novo variants. However, it is likely that false positive rates would be higher in more complex genomic regions where a design is not possible. By using long read genome sequencing data, we successfully phased 90.5% of the DNM, 80% of which were assigned to the paternal haplotype (ranging from 70 to 85%). Restricting to short-read data, only 34% variants could have been phased, however retrieving the same paternal bias (79%). The paternal age effect was visible for all DNMs (Fig. 1C) and for phased DNMs (Supplementary Fig. 2). Single base substitutions analysis identified the two standard “clock-like” mutational signatures, in expected rates: SBS5 (67%) and SBS1 (24%) (Supplementary Fig. 3). Three percent of de novo variants (13/428) were located in mutational clusters (i.e. variants distant from less than 20kb), and analysis of these clusters also revealed expected properties, including variant counts, genomic distribution and a biased Ti/Tv ratio (Supplementary Fig. 4). In summary, we reliably detected de novo variants in these genomes, which recapitulated known properties of de novo variants.

De novo variants resulting from parental germinal mosaicism are detectable in every genome

Parental mosaicism was assessed for 334 variants with high quality pileup data. Mean smMIP sequencing depth (after deduplication, one x per good quality read pair) was 5.557x, 8.314x, and 5.755x for child blood, parental blood and paternal sperm samples, respectively. Candidate parental mosaicism was called if VAF differed significantly from sequencing noise in controls, and subsequently confirmed by an independent smMIP experiment. In total, 20/334 variants displayed evidence for parental mosaicism (6.0%)(Fig. 2, Supplementary Table 4). We found that every child carried at least one DNM that was detectable in parental blood (1–4, average 2.6), with VAFs ranging from 0.35–14.7%. Parental blood mosaicism indicates early, pre-PGCs mutational events occurring before the sexual differentiation of the germline, and are therefore likely to be as common in paternal and maternal germline. In line with this, we found similar counts of paternal and maternal mosaics (7 and 6, respectively). However maternally-phased variants were 3.4x more likely to display blood mosaicism than paternally phased variants (6/62 = 9.7% versus 7/244 = 2.9%, respectively, Fisher test p = 0.0289), in line with the “dilution” of paternally phased de novo variants by events occurring during spermatogenesis [21]. In accordance with the mandatory transmission of mosaic variants to the children in this study, all paternal mosaic variants detected in blood were also detectable in sperm. For these shared mosaic variants, VAFs were higher in sperm than in blood (median difference 2.0%, paired Wilcoxon signed-rank exact test, p = 0.1094, Supplementary Fig. 5). This trend, though non statistically significant, is consistent with previous studies [7], and can be attributed to a selection bias, as the variants included have all been transmitted to one child.

Germline mosaicism can also occur after PGCs and therefore be only detectable in the germline. Deep sequencing analysis of paternal sperm samples also identified this kind of events in every trio (1 to 2 events per trio, average 1.4). Consistent with a later occurrence in paternal embryonic development, the point estimate of VAF of such DNMs identified in sperm and not in blood was lower than sperm VAF of variants also detectable in paternal blood samples (median 2.9% versus 4.2% respectively), recapitulating previous observations [22], although the difference was not significant due to limited sample size (Mann-Whitney U-test p = 0.3176). VAF of mosaicism confined to sperm ranged from 0.24–9.5%.

Detection of post-zygotic mutations using a combined approach

Another type of DNM with lowest recurrence risk are the post-zygotic variants in the child. Detecting those may be as important as detecting germline mosaics for genetic counselling. Using concordant calls from smMIP deep sequencing in children, in which we looked for variants with VAF deviating from 0.5, and long-read genome sequencing data, we could detect three instances of high confidence post-zygotic mosaicism in probands (Fig. 2A, Supplementary Table 4). Even though both sequencing depth and error rates were suboptimal to detect mosaicism in theory from long read sequencing data, we exploited the phasing information by focusing on haplotype-specific VAF, corresponding to the VAF of the variant within the mutated haplotype. While this is supposed to be 100% in non-mosaic variants, deviation from 100% indicates mosaicism with a very high probability. Although this approach was only possible for a subset of variants, we detected 3/163 (1.85%) post-zygotic variants (Supplementary Fig. 6), which would translate to an estimated number of 7.9 post-zygotic variants in our dataset.

Recurrence risk assessment

To assess the recurrence risk in future pregnancies, we hypothesized that the VAF in sperm cells of paternally phased variants reflects the actual recurrence risk for this subset of variants. This assumption entails that (i) the variant does not affect the likelihood of embryo development and (ii) the proportion of mutated sperm cells, as indicated by the VAF, remains constant over time. Among 244 assessed paternally phased variants, we found 14 instances of sperm mosaicism, 13 of which had a VAF above the empirical 1% recurrence risk. By contrast, 230 variants did not show evidence for sperm mosaicism, leading to a very low recurrence risk, below the mosaic detection rate of our approach. Attributing a null VAF to variants that did not reach statistical evidence for an enrichment over sequencing noise in our analysis to qualify as germline mosaics, the average VAF and therefore recurrence risk for paternally phased de novo variants was 0.27%. Including the raw detected VAF for non-mosaic variants still led to a similar mean VAF of 0.32% indicating both (i) a low magnitude of sequencing noise and (ii) a limited impact on recurrence risk assessment of true mosaics that we would have failed to distinguish from background noise. Altogether, our results based on sperm mosaic detection show a low overall risk of recurrence for paternally phased de novo variants, stratified into ~ 5% of variants at high risk and ~ 95% of variants with null or very low risk.

With the aim of exploring the timing of mutational events in human germline, we used a three-step method to detect genome-wide DNMs, attribute a parental haplotype and assess parental blood and sperm mosaicism, in five individuals. We found that parental embryonic mosaicism was a common source of DNMs, detectable in every genome. Pre-PGC events (detected in blood) appear equally distributed on paternal and maternal haplotype, which account respectively for n = 1.9 and 1.8 events per child after adjustment on detectability. Post-PGC events in male germline, as variants detected in sperm only, were detected at a similar rate than pre-PGC events, at about 2.0 events per child. Therefore, sequencing sperm samples appear twice as sensitive to identify the risk of recurrence as sequencing blood. Altogether, our approach based on mosaicism detection found an average risk of recurrence of paternally phased variants of 0.27%. We compared this estimate with a model based on the actual recurrence rate of variants within an Icelandic population [21], and found that our paternally phased variants would be 0.55% of risk of recurrence (Supplementary Information). A possible explanation could be that our set of de novo variants does not capture all variants that are at risk of recurrence. Indeed, the work from Decode, besides detecting de novo variants by trio approach as we did, the authors further detected variants by a haplotype-based method in large families, allowing for the detection of variants with high VAF in parents (high mosaicism) that would be considered as inherited variants by trio-based methods. Indeed, the authors estimated that the trio-based method would miss about half of the variants that actually recurred [21]. Therefore, our result represent an estimation of recurrence risk for trio-accessible DNMs only, which is yet a valuable information since clinically relevant DNMs are often detected by this approach.

The risk of recurrence of maternally derived DNMs is more difficult to assess from clonal VAF detection due to the inaccessibility of germline cells which harbour post-PGC variants. However, post-PGC events detectable in bulk analysis of germ cells are supposed to occur as an very early embryonic event (“peri-PGC”, [23]), in primordial germ cells, prior to sexual differentiation. Therefore, this shared biology argue that the absolute count and VAF of oocyte mosaicisms should be similar to those of sperm cells. This assumption would mean that the risk of recurrence for maternally derived variants equals RR_pat x α where RR_pat is the risk of recurrence for paternally phased variants and α is the ratio of paternal/maternal counts (Supplementary Information). With this approach and the value of α = 4 in our cohort, we estimate the maternal recurrence risk to be 1.08%, and the overall risk of recurrence to be 0.43%. This appears lower than the commonly accepted risk of recurrence of 1% for DNMs [8]. Once again, the detection method should be considered and our estimate concerns DNMs detected by stringent trio-based rules.

Recent studies on the genome-wide mutability of somatic tissues showed that variants accumulate at a relatively constant rate throughout life in many tissues, with a high correlation with time and minor impact of cell division rate [24]. However, in the germline, mutation rate does not appear constant overtime. Indeed, we found that 6.9% of the assessed de novo variants exhibited evidence for an early embryonic mutational event, which is a significant proportion considering the short period of time in which these variant are considered to happen (days or weeks post fertilization) in relation to the duration of a generation in which DNMs can occur. This observation can be linked to a very significant hypermutability of the first few cell divisions after the zygote, which has recently been detected by multiple approaches [25]. This hypermutability coincides with rapid cellular divisions termed “cleavages” without G1 and G2 phases and a suppression of cell cycle checkpoint. This special cellular state may be much prone to mutations, explaining this critically enriched short period of time.

In this study, we applied a sensitive deep sequencing method to detect parental mosaicism. Using these results as a gold standard, we can compare the performances to that of parental WGS VAF alone to detect parental blood mosaicism. Considering only variants with at least 1 alternate read in parental WGS, we would have had surprisingly good performances, with 77% recall and 67% precision (Supplementary Fig. 7). Notably, this would have captured all variants with a VAF of > 1%. On a technical note, the alternate count was assessed using samtools mpileup because DeepVariant did not report any alt count reads in its output VCF. This limitation prevents the use of DeepVariant for VAF quantification in mosaics. This suggests that, in absence of deep sequencing data in parental samples, looking back at the raw data from WGS may still be very useful to assess recurrence risk. However, in case of a single alternate read, a confirmation in an independent assay remains mandatory.

One major originality of our study was to assess the recurrence risk of genome-wide variants, whatever their effect on biology or disease risk. Most previous studies have focused on the assessment of recurrence risk of individual pathogenic de novo variants [7, 26–28]. In a remarkable example on 59 de novo variants, the authors applied a general framework consisting in (i) phasing the variants using targeted long-read sequencing and (ii) sequencing multiple parental tissues [7]. In our study, we used long read genome sequencing only to phase the DNMs called from short-read data, due to the low performances of v9 chemistry of Nanopore in small variant calling. However, recent advancements in long-read sequencing technologies have significantly improved this quality. These improvements enable highly accurate and efficient identification of de novo variants [29]. Therefore, it is likely that the transition from short-read to long-read genome sequencing in future years will enable much more systematic phasing of DNMs and therefore benefit to the genetic counselling of DNM associated diseases. With long-read based DNM identification, the pipeline for risk recurrence assessment could be restricted to a deep sequencing analysis on paternal sperm for paternal variants. Such a viable approach would lead to a precise estimation of recurrence risk for 80% of DNMs and avoid unnecessary invasive prenatal testing procedures in most of these cases. Evidence shows little variation in the VAF of sperm mosaicism over time [22, 30], which could corroborate this approach of using VAF as a proxy for risk of recurrence of paternally phased variants. While techniques of prenatal diagnosis improve and non-invasive techniques (NIPT) becomes accessible for de novo variants [31], the anticipation of the recurrence risk by sperm analysis before any pregnancy could better suit some families, and would present the advantage to be performed only once versus one NIPT at each pregnancy.

Finally, some other criteria should be taken into account when assessing the recurrence risk of DNMs. Some pathogenic variants in specific genes can lead to a developmental advantage of the wild type or mutant cell over the other [25], leading to biased recurrence risk. For instance, selfish mutations, affecting RAS/MAPK pathway occur almost systematically in paternal adult germline and even though these mutations lead to spermatogonial clonality, the overall proportion of mutated cells is very limited [32]. In line with this, epidemiological observations have revealed a low risk of recurrence for selfish mutations, questioning the necessity of prenatal diagnostic testing in subsequent pregnancies after an affected child [33]. In contrast, pathogenic mutations in other genes, such as SCN1A, appear enriched in parental mosaicism and de novo recurrence risk [34–42]. Another genomic feature that could potentially be used for recurrence risk assessment could be the presence of the variant in a mutational cluster (i.e. multiple variants within a small genomic interval, typically 20kb). Many mutation clusters are thought to derive from age-related changes in the biology of the germline, notably on oocytes [43]. Therefore, clustered variants could be indicative of low recurrence risk variants. Interestingly, none of the 11 clustered variants in which deep sequencing was performed showed evidence of parental mosaicism. Larger studies are needed to assess the correlation between risk of recurrence and occurrence in mutation clusters.

In summary, we present the proportion of genome-wide DNMs mediated by a mechanism of parental embryonic mosaicism. We estimate the average recurrence risk of de novo variations detected in WGS trio analysis to be less than 1%. For the 80% of variants mapping to the paternal haplotype, sequencing of paternal sperm samples enabled a more precise assessment of recurrence risk, with 95% of these variants classified as having negligible risk and 5% with a risk greater than 1%.

BWA

Burrows-Wheeler Aligner

CNRGH

Centre National de Recherche en Génomique Humaine

DNM

De novo mutation

indel

Insertion/deletion

PBMC

Peripheral blood mononuclear cells

PGCS

Primordial germ cell specification

SNV

Single nucleotide variant

TCEP

tris(2-carboxyéthyl)phosphine

VAF

Variant allele fraction

Ethics approval and consent to participate

The study was approved by the CPP Ouest V (20/043-2) ethics committee. Informed consent for study participation was collected for each participant. GERMETHEQUE biobank (BB-0033-00081), site of Rouen, provided 5 samples of spermatozoa and their associated data to realize this project. GERMETHEQUE obtained consent from each patient to use their sperm samples (CPP 2.15.27). The GERMETHEQUE pilotage committee approved the study design the 17/11/2020. The Biobank has a declaration DC-2021-4820 and an authorization AC-2019-3487. The request’s number made to Germethèque is the 20201117.

Consent for publication

Written consent for publication was obtained from each participant in this study.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests

The authors have no relevant conflicts to disclose.

Funding

This study was funded by a grant from GIRCI Nord-Ouest (AAP-AE_19-36). Collaboration CEA‐DRF‐Jacob‐CNRGH—CHU de Rouen. This study was co‐supported by the European Union and Région Normandie in the context of Recherche Innovation Normandie (RIN2018). This work did benefit from support of the France Génomique National infrastructure, funded as part of the “Investissements d'Avenir” program managed by the Agence Nationale pour la Recherche (contract ANR‐10‐INBS‐09).

Authors' contributions

FL and GN developed initial concept. ND, FJ, FC and FL performed the experiments. AB, RO, VM and JFD produced the short and long read genome data. FL, SC, OQ and SF performed bioinformatics data management. FL, CC and GN analysed the data. FL wrote the initial draft and FL, CC, and GN revised the manuscript for important intellectual content. All authors read and approved the final manuscript

Acknowledgements

We thank the participants and their family members. This study was supported by Germethèque Biobank (France – Site of Rouen) which provided samples.

Deciphering Developmental Disorders Study. Prevalence and architecture of de novo mutations in developmental disorders. Nature. 2017;542:433–8.
Jónsson H, Sulem P, Kehr B, Kristmundsdottir S, Zink F, Hjartarson E, et al. Parental influence on human germline de novo mutations in 1,548 trios from Iceland. Nature. 2017;549:519–22.
Kaplanis J, Ide B, Sanghvi R, Neville M, Danecek P, Coorens T, et al. Genetic and chemotherapeutic influences on germline hypermutation. Nature. 2022;605:503–8.
Goldmann JM, Wong WSW, Pinelli M, Farrah T, Bodian D, Stittrich AB, et al. Parent-of-origin-specific signatures of de novo mutations. Nat Genet. 2016;48:935–9.
Kong A, Frigge ML, Masson G, Besenbacher S, Sulem P, Magnusson G, et al. Rate of de novo mutations and the importance of father’s age to disease risk. Nature. 2012;488:471–5.
Breuss MW, Yang X, Gleeson JG. Sperm mosaicism: implications for genomic diversity and disease. Trends Genet. 2021;37:890–902.
Bernkopf M, Abdullah UB, Bush SJ, Wood KA, Ghaffari S, Giannoulatou E, et al. Personalized recurrence risk assessment following the birth of a child with a pathogenic de novo mutation. Nat Commun. 2023;14:853.
Campbell IM, Stewart JR, James RA, Lupski JR, Stankiewicz P, Olofsson P, et al. Parent of origin, mosaicism, and recurrence risk: probabilistic modeling explains the broken symmetry of transmission genetics. Am J Hum Genet. 2014;95:345–59.
Eyal O, Berkenstadt M, Reznik-Wolf H, Poran H, Ziv-Baran T, Greenbaum L, et al. Prenatal diagnosis for de novo mutations: Experience from a tertiary center over a 10-year period. Mol Genet Genomic Med. 2019;7:e00573.
Tang WWC, Kobayashi T, Irie N, Dietmann S, Surani MA. Specification and epigenetic programming of the human germ line. Nat Rev Genet. 2016;17:585–600.
Canovas S, Campos R, Aguilar E, Cibelli JB. Progress towards human primordial germ cell specification in vitro. Mol Hum Reprod. 2017;23:4–15.
Wood KA, Goriely A. The impact of paternal age on new mutations and disease in the next generation. Fertil Steril. 2022;118:1001–12.
Lecoquierre F, Quenez O, Fourneaux S, Coutant S, Vezain M, Rolain M, et al. High diagnostic potential of short and long read genome sequencing with transcriptome analysis in exome-negative developmental disorders. Hum Genet. 2023;142:773–83.
Wu H, de Gannes MK, Luchetti G, Pilsner JR. Rapid method for the isolation of mammalian sperm DNA. Biotechniques. 2015;58:293–300.
DeNovoMiner.py [Internet]. Available from: https://github.com/francois-lecoquierre/de_novo_tools/blob/main/DeNovoMiner.py
classify_vcf_from_igv_v1.0.py [Internet]. Available from: https://github.com/francois-lecoquierre/genomics_shortcuts/blob/main/blob/main/classify_vcf_from_igv_v1.0.py
Degasperi A, Amarante TD, Czarnecki J, Shooter S, Zou X, Glodzik D, et al. A practical framework and online tool for mutational signature analyses show inter-tissue variation and driver dependencies. Nat Cancer. 2020;1:249–63.
Belyeu JR, Sasani TA, Pedersen BS, Quinlan AR. Unfazed: parent-of-origin detection for large and small de novo variants. Bioinformatics. 2021;37:4860–1.
Lecoquierre F, Cassinari K, Drouot N, May A, Fourneaux S, Charbonnier F, et al. Assessment of parental mosaicism rates in neurodevelopmental disorders caused by apparent de novo pathogenic variants using deep sequencing. Sci Rep. 2024;14:5289.
generate_and_analyse_pileups.py [Internet]. Available from: https://github.com/francois-lecoquierre/smmip_analysis/blob/main/generate_and_analyse_pipeups.py
Jónsson H, Sulem P, Arnadottir GA, Pálsson G, Eggertsson HP, Kristmundsdottir S, et al. Multiple transmissions of de novo mutations in families. Nat Genet. 2018;50:1674–80.
Yang X, Breuss MW, Xu X, Antaki D, James KN, Stanley V, et al. Developmental and temporal characteristics of clonal sperm mosaicism. Cell. 2021;184:4772–4783.e15.
Lindsay SJ, Rahbari R, Kaplanis J, Keane T, Hurles ME. Similarities and differences in patterns of germline mutation between mice and humans. Nat Commun. 2019;10:4053.
Abascal F, Harvey LMR, Mitchell E, Lawson ARJ, Lensing SV, Ellis P, et al. Somatic mutation landscapes at single-molecule resolution. Nature. 2021;593:405–10.
Waldvogel SM, Posey JE, Goodell MA. Human embryonic genetic mosaicism and its effects on development and disease. Nat Rev Genet. 2024;
Yang X, Yang X, Chen J, Li S, Zeng Q, Huang AY, et al. ATP1A3 mosaicism in families with alternating hemiplegia of childhood. Clin Genet. 2019;96:43–52.
Frisk S, Wachtmeister A, Laurell T, Lindstrand A, Jäntti N, Malmgren H, et al. Detection of germline mosaicism in fathers of children with intellectual disability syndromes caused by de novo variants. Mol Genet Genomic Med. 2022;10:e1880.
Yang X, Liu A, Xu X, Yang X, Zeng Q, Ye AY, et al. Genomic mosaicism in paternal sperm and multiple parental tissues in a Dravet syndrome cohort. Sci Rep. 2017;7:15677.
Kucuk E, van der Sanden BPGH, O’Gorman L, Kwint M, Derks R, Wenger AM, et al. Comprehensive de novo mutation discovery with HiFi long-read sequencing. Genome Med. 2023;15:34.
Breuss MW, Antaki D, George RD, Kleiber M, James KN, Ball LL, et al. Autism risk in offspring can be assessed through quantification of male sperm mosaicism. Nat Med. 2020;26:143–50.
Verebi C, Gravrand V, Pacault M, Audrezet M-P, Couque N, Vincent M-C, et al. [Towards a generalization of non-invasive prenatal diagnosis of single-gene disorders? Assesment and outlook]. Gynecol Obstet Fertil Senol. 2023;51:463–70.
Salazar R, Arbeithuber B, Ivankovic M, Heinzl M, Moura S, Hartl I, et al. Discovery of an unusually high number of de novo mutations in sperm of older men using duplex sequencing. Genome Res. 2022;32:499–511.
Wilkie AOM, Goriely A. Gonadal mosaicism and non-invasive prenatal diagnosis for “reassurance” in sporadic paternal age effect (PAE) disorders. Prenat Diagn. 2017;37:946–8.
Depienne C, Arzimanoglou A, Trouillard O, Fedirko E, Baulac S, Saint-Martin C, et al. Parental mosaicism can cause recurrent transmission of SCN1A mutations associated with severe myoclonic epilepsy of infancy. Hum Mutat. 2006;27:389.
Liu AJ, Yang XX, Xu XJ, Wu QX, Tian XJ, Yang XL, et al. [Study on mosaicism of SCN1A gene mutation in parents of children with Dravet syndrome]. Zhonghua Er Ke Za Zhi. 2017;55:818–23.
Sharkia R, Hengel H, Schöls L, Athamna M, Bauer P, Mahajnah M. Parental mosaicism in another case of Dravet syndrome caused by a novel SCN1A deletion: a case report. J Med Case Rep. 2016;10:67.
Halvorsen M, Petrovski S, Shellhaas R, Tang Y, Crandall L, Goldstein D, et al. Mosaic mutations in early-onset genetic diseases. Genet Med. 2016;18:746–9.
Guala A, Peruzzi C, Gennaro E, Pennese L, Danesino C. Maternal germinal mosaicism for SCN1A in sibs with a mild form of Dravet syndrome. Am J Med Genet A. 2015;167A:1165–7.
Selmer KK, Eriksson A-S, Brandal K, Egeland T, Tallaksen C, Undlien DE. Parental SCN1A mutation mosaicism in familial Dravet syndrome. Clin Genet. 2009;76:398–403.
Marini C, Scheffer IE, Nabbout R, Mei D, Cox K, Dibbens LM, et al. SCN1A duplications and deletions detected in Dravet syndrome: implications for molecular diagnosis. Epilepsia. 2009;50:1670–8.
Marini C, Mei D, Helen Cross J, Guerrini R. Mosaic SCN1A mutation in familial severe myoclonic epilepsy of infancy. Epilepsia. 2006;47:1737–40.
Morimoto M, Mazaki E, Nishimura A, Chiyonobu T, Sawai Y, Murakami A, et al. SCN1A mutation mosaicism in a family with severe myoclonic epilepsy in infancy. Epilepsia. 2006;47:1732–6.
Goldmann JM, Seplyarskiy VB, Wong WSW, Vilboux T, Neerincx PB, Bodian DL, et al. Germline de novo mutation clusters arise during oocyte aging in genomic regions with high double-strand-break incidence. Nat Genet. 2018;50:487–92.
Coursimault J, Cassinari K, Lecoquierre F, Quenez O, Coutant S, Derambure C, et al. Deep intronic NIPBL de novo mutations and differential diagnoses revealed by whole genome and RNA sequencing in Cornelia de Lange syndrome patients. Hum Mutat. 2022;43:1882–97.

No competing interests reported.

Download PDF

Version 1

posted

You are reading this latest preprint version

Parental germline mosaicism in genome-wide phased de novo variants: recurrence risk assessment and implications for precision genetic counselling

Status:

Version 1

Abstract

Background:

Methods:

Results:

Conclusions:

Figures

BACKGROUND

METHODS

Patients and samples

Genome sequencing

Targeted deep sequencing

Mosaic variant identification and statistics

RESULTS

Detection of post-zygotic mutations using a combined approach

Recurrence risk assessment

DISCUSSION

CONCLUSION

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1