Meiotic errors followed by two parallel postzygotic trisomy rescue events are a frequent cause of constitutional segmental mosaicism

For decades, knowledge about copy number variation (CNV) in the human genome was limited to microscopically visible changes. Advances in technology have led to the discovery of submicroscopic CNVs, ranging from kilobases to megabases in size and covering up to 13% of the human genome [1, 2]. These CNVs can cause recurrent genomic disorders and sporadic disease, or they can represent benign changes found in the healthy population [3, 4]. Recent studies have revealed that CNVs are not only polymorphic between unrelated individuals, but also form a frequent source of somatic variation [5, 6].

Chromosomal mosaicism is defined as the coexistence of two or more chromosomally different cell lines in an organism which developed from a single zygote. The majority of those mosaicisms are aneuploidies. Several studies investigating in vitro fertilized embryos at the preimplantation stage demonstrated a very high number of chromosomal mosaicisms in early human embryos [7–9]. While many of these embryos will not reach the stage of implantation, some do continue to develop leading to fetal mosaicisms, confined placental mosaicism or mosaic infants. Postnatally, mosaicism is detected in 0.4-1% of patients referred for genetic diagnostic screening [10–12]. A recent study revealed that mosaic aberrations are present in about 0.8% of phenotypically normal adults [13]. In addition, mosaicism appears to be variable amongst different tissues: chromosomal aneuploidies were detected in approximately 10% of normal human brain cells [14].

Segmental aneuploidies make up a significant part of mosaic chromosome anomalies. Analysis of several large series of prenatal samples by karyotyping has shown that, of the 0.25-2% mosaic cases that are detected, up to a third comprise segmental imbalances [15, 16]. In postnatal clinical diagnosis of patients with developmental anomalies this increases to about half of the mosaic cases [10, 17]. The majority of mosaic segmental imbalances are marker chromosomes [16]. A smaller number of cases have been reported to consist of mosaic segmental deletions and/or duplications, ring chromosomes and translocations that have a 46,abnormal/46,normal karyotype. In recent years various case reports have been published [18–36]. The introduction of genome wide aneuploidy detection tools with a higher resolution such as array comparative genomic hybridisation (array CGH) or SNP arrays and the collection of large patient groups have also increased the detection rate of mosaic segmental abnormalities [10, 12, 17].

For mosaic aberrations, it is intuitively assumed that the rearrangement arose postzygotically. During embryogenesis, a mitotic rearrangement in an otherwise normal diploid embryo results in a second cell line carrying a chromosomal rearrangement. Nevertheless, evidence is mounting that such rearrangements can originate during meiosis. If so, a trisomic zygote carrying the abnormal chromosome has to be rescued twice, in parallel: once loosing the normal and once loosing the abnormal chromosome. Considering that the mitotic error rate is extremely high during early embryogenesis [7–9] and that most cases of mosaic aneuploidy detected in a large cohort of patients were of mitotic origin [11], we reasoned that the latter mechanism might be an important mechanism by which such mosaics arise. In this study we collected nine cases with mosaic structural imbalances to determine their origin and to ascertain whether a meiotic or postzygotic origin might be more prevalent.

All nine samples contained two different cell lines, each with 46 chromosomes: one normal diploid cell line and a second cell line carrying a chromosomal rearrangement. In five cases the mosaicism was detected in blood lymphocytes, in three cases (case 4, 8 and 9) in amniocytes and in one case (case 6) in chorionic villi. In addition to the mosaicism present in the white blood cells, the mosaicism was confirmed in buccal cells in cases 7 and 9. Conventional karyotyping was carried out in five cases and detected the mosaicism in four out of five. Microarrays were performed on DNA from 8/9 cases to identify or confirm the segmental aneuploidy, determine its size and rule out any additional chromosomal imbalances (Figures 1 and 2). The array intensity ratios indicated duplications in approximately 66, 55, 53, 72 and 15% of the cells in patients 1, 2, 3, 4 and 5 respectively. The array intensity ratios indicated the presence of segmental deletions in approximately 20 and 30% of the cells in samples 8 and 9. In two samples, 6 and 7, both deletions and duplications were identified by karyotyping (case 6) or array CGH (case 7) in respectively 50 and 20% of the cells. Conventional karyotyping showed those imbalances to be unbalanced translocations. FISH confirmed the mosaic segmental imbalances in six samples. FISH also determined that the duplication in cases 1 and 4 was not translocated to another chromosome and that cases 5 and 7 carry unbalanced translocations. The karyotypes are presented in Table 1. The degree of mosaicism was estimated based on the intensity ratios of the targets located within the segmental aneuploidy or on data from FISH analysis. The segment sizes and estimated relative presence of the segmental aneuploidy range between 1 and 124 Mb and an overview is shown in Table 1. Chromosome analysis of the parents, including FISH for submicroscopic aberrations, was normal for all.

To determine whether the initial rearrangement occurred during meiosis or during postzygotic mitosis, polymorphic marker analysis or SNP array analysis was performed. For patients 1 to 6, carrying segmental duplications, markers were selected within and flanking the duplicated region. If the duplication arose during meiosis, three alleles might be observed. If only two alleles are detected, there are two possible explanations: either the duplication arose postzygotically or the duplication arose during meiosis and the segmental duplication is one of both sister chromatids present during meiosis I or II. Results of the short tandem repeat (STR) analyses are shown in Figures 3, 4 and 5 and a summary is presented in Table 2.

For two samples, three different alleles were detected within the duplicated region indicating a meiotic origin. Polymorphic marker analysis of sample 1 revealed the presence of three different alleles for marker D13S128 (Figure 3A). Two of the alleles were inherited from the father demonstrating the paternal origin of the segmental aneusomy. At marker D13S129, still within the duplication, two alleles with a 2:1 intensity ratio were detected, including only one of the two paternal alleles (Figure 3B). The remaining markers within the duplication were not informative. In case 6, carrier of an unbalanced translocation, two markers (D1S1653 and D1S1171) spread throughout the duplicated 1q region showed three different alleles (Figure 4). The other markers on 1q had either one or two alleles. Marker D20S117 in the deleted 20p region confirmed the deletion, with a 1:2 reduced peak intensity for one of both alleles.

For the other cases, only two different alleles were identified within the duplication region. Genotyping of cases 2 to 5 within the region of duplication showed in each case only two alleles. One of both alleles clearly had a higher intensity with about a 2:1 ratio (Figure 5). For sample 5, in which array CGH revealed that the rearrangement is only present in 10 to 20% (log₂ value of 0.0888) of all cells, these peak intensity differences were still present, but to a lesser extent.

For patients 7 to 9 with mosaic segmental deletions, single nucleotide polymorphisms were analyzed by 250 K Affymetrix SNP arrays. To determine the origin of the mosaic unbalanced translocation in case 7, genotyped by Affymetrix SNP arrays, the following strategy was followed: the A and B allele ratios of individual SNPs within the segmental mosaic duplication and deletion were determined for those SNPs that were homozygous but with a different allele in both parents. Analysis of the A and B allele ratios transmitted from parental homozygous AA and BB SNPs showed 77 SNPs to fulfill this criteria in the duplication and 51 SNPs in the deletion and demonstrated a paternal origin for both the deleted and the duplicated segments of the unbalanced translocation. Subsequently, the A and B allele ratios of SNPs in the duplicated region that were paternal heterozygous but maternal homozygous were interrogated. Analysis of the SNPs did not show transmission of both paternal alleles. Only a single duplicated allele from the father was detected (Figure 6A).

To determine the origin of the deletion in cases 8 and 9, again for the SNPs that were homozygous but with a different allele in both parents in the deletion, the A and B allele ratios were determined. The analysis showed that the signal of the paternal allele was reduced in case 8 and that a maternal allele was removed in case 9. To search for evidence of a potential meiotic event, the rest of the chromosome was screened for the presence of two paternal (case 8) or two maternal alleles (case 9). Hereto, the B-allele frequencies of SNPs on the affected chromosome, but outside the somatic CNV region, were evaluated for additional haplotypes. If the deletion resulted from a meiotic event, altered B-allele frequencies would be observed, with additional values between 0 and 0.5 as well as between 0.5 and 1. No such haplotypes were detected in the regions surrounding the deletion (Figure 6B).

The combination of a normal and an aberrant cell line in these cases could, aside from mosaicism, also result from chimerism: the fusion of two cell lineages from separate fertilization events. In samples 1 to 6, the presence of chimerism was evaluated by testing four STR markers spread over different chromosomes not involved in the aberrant region. in case of a mosaic all autosomal genotypes outside the duplicated region are expected to only contain two alleles (1 maternal & 1 paternal). For chimeras, some loci would have three or four alleles, or a skewed dosage of two alleles [37]. None of the samples showed more than two alleles on any of the markers, confirming the mosaic status of the aberration (data not shown). In cases 7 to 9, the two different genotypes present in case of chimerism would be seen as aberrant B-allele frequencies in all autosomes, with an altered ratio between the two haplotypes. No alterations in the B-allele frequency were seen in the unaffected chromosomes, thereby ruling out the possibility of chimerism.

Somatic chromosomal mosaicisms are thought to arise from postzygotic somatic mutations. Here, we show that in two out of nine cases (>15%) the rearrangements occurred preconception. Cases 1 and 6 showed markers with three different alleles in the duplicated region. The most parsimonious explanation is that both aberrations originated from a meiotic segregation error resulting in a trisomic conception, followed by two parallel trisomy rescue events during the successive mitotic divisions (Figure 7). In the other seven cases (pt 2-5 & pt 7-9), the duplication was derived from the chromosome with the same parental allele as the transmitted chromosome in the normal cell line or the deletion was on the chromosome with the same parental allele as the normal chromosome in the normal cell line. In these cases the copy variation likely occurred during postzygotic cell divisions. Nevertheless, those rearrangements could also have occurred during meiosis II. Hence, the percentage of mosaic segmental imbalances generated during meiosis by two separate postzygotic trisomy rescue events could be higher than 15%. The occurrence of multiple postzygotic trisomy rescue events may appear to be unexpected. However, a high frequency of mosaic trisomies and monosomies is reported to occur during cleavage stage in in vitro fertilized embryos [7–9]. The observation that two independent trisomy rescue events can underlie those mosaicisms is further testimony of this high mutational burden during early embryogenesis. Postzygotic trisomy rescue events are also frequently observed in patients with imprinting diseases, including Prader-Willi syndrome and Silver-Russell syndrome. Such patients may be fully disomic with complete uniparental disomy (UPD) or be mosaic with a trisomic and a disomic (UPD) cell line [11, 38–40].

While the origin of mosaic segmental CNVs has only rarely been investigated, two case reports support our finding that pre-zygotic rearrangements followed by multiple postzygotic trisomy rescue events can underlie segmental mosaicisms. Blouin et al. detected a mosaic de novo direct tandem duplication of 21q11.2q22.2 that resulted from a meiosis I crossover followed by an unequal sister chromatid exchange and two trisomy rescue events, similar to our case 1. The mosaic de novo unbalanced translocation with partial trisomy 16p and maternal UPD 16 reported by Schinzel et al. likely originated from a maternal trisomy 16 (MI), a subsequent postzygotic translocation of the paternal 16p segment and finally loss of the paternal chromosomes 16 in the translocated and normal cell lines [18, 32]. Even after combining our results with previous mosaic segmental cases the total study population remains small, but future studies may be able to provide additional evidence for the high frequency of multiple independent trisomy rescue events.

In conclusion, we demonstrated that in 2/9 cases, the detected mosaic segmental aberrations resulted from meiotic errors followed by multiple parallel trisomy rescue events and not from postzygotic mitotic changes. In addition, we show that the use of sensitive array technology is especially suited for the detection of mosaicism, which is clinically relevant to a patient's diagnosis.