Mosaic genome-wide maternal isodiploidy: an extreme form of imprinting disorder presenting as prenatal diagnostic challenge

Uniparental disomy (UPD) refers to the constellation of two identical (isodisomy) or homologous (heterodisomy) chromosomes inherited from only one parent [1]. Known mechanisms resulting in UPD are gametic complementation, monosomy or trisomy rescue, compensatory UPD, and post-fertilization errors including mitotic recombination [2]. Non-disjunction at meiosis I can result in heterodisomy and isodisomy while non-disjunction at meiosis II leads to isodisomy of those parts of the chromosome set not involved in homologous recombination [1, 3]. Clinical consequences of isodisomy include unmasking of recessive diseases by transmitting two affected gene copies from one heterozygous parent carrier [2]. Moreover, UPD can cause imbalance of imprinted gene expression when the involved regions are subject to genomic imprinting, i.e., showing a parent-of-origin specific gene expression [2, 4]. In these cases, UPD results in the inheritance of either two active or two repressed copies of a gene, dependent on the chromosomal region involved and the sex of the transmitting parent [4]. Indeed, UPD is a well-known mechanism underlying several imprinting disorders including Prader-Willi syndrome [PWS, UPD(15)mat], Angelman syndrome [AS, UPD(15)pat], Beckwith-Wiedemann syndrome [BWS, UPD(11p15.5)pat], transient neonatal diabetes mellitus [TNDM, UPD(6)pat], Silver-Russell syndrome [SRS, UPD(7)mat, UPD(11p15.5)mat], Temple syndrome [TS14, UPD(14)mat], Kagami-Ogata syndrome [KOS14, UPD(14)pat], and pseudo-hypoparathyroidism type IB [PHP-1B, UPD(20)pat]. The frequency of UPD in the aforementioned imprinting syndromes varies and ranges from 1 to 2% (AS) [5] up to about 40% (TNDM) [6]. Apart from UPD, imprinting disorders can result from primary epimutation [7]. The latter term in this context refers to a change in DNA methylation at the differentially methylated region (DMR) regulating the parent-of-origin specific gene expression without evidence for a genomic mutation in cis [7]. In these instances, the DNA methylation disturbance can affect more than one locus and is then referred to as “multilocus imprinting disturbance” (MLID). To date, MLID has been reported for most of the phenotypes associated with the classical imprinting disorders except KOS14 and PWS [8–10]. However, the amount and combination of loci affected is highly heterogeneous and generally does not extend to all known imprinted regions.

The most extreme form of uniparental inheritance is uniparental diploidy, i.e., UPD of all chromosomes, leading to genome-wide DNA methylation disturbances at virtually all imprinting loci in a single patient. This global imbalance of imprinting seems to be not compatible with life, since the constellation is known to be lethal in mammals [11]. However, in a mosaic state, genome-wide UPD can lead to live-born children [4, 12]. So far, only about 18 cases of genome-wide UPD have been reported in the literature and only two of them show mosaic genome-wide UPD of maternal origin.

We here add one further case to the short list of patients with this extremely rare genetic disorder that we recently diagnosed by prenatal genetic testing. We provide a detailed description of the diagnostic work-up and phenotypic outcome. Our investigations include prenatal analyses of DNA methylation at major regulatory sites associated with the classical imprinting syndromes.

Based on the obtained results of all analyses, we came to the conclusion that the underlying genetic condition in the fetus resulted from a mosaicism for a diploid cell line with genome-wide maternal isodiploidy in combination with a diploid biparental cell line with an extra chromosome X derived from paternal isodisomy X. In culture, predominantly, the latter clone expanded and most likely acquired a trisomy 10 during culturing since this finding was neither confirmed in uncultured amniotic cells nor in an independent culture. Reanalyzing the array raw data with the knowledge of the size of the two detected cell lines from FISH analyses on uncultured amniotic cells, we determined the expected B allele frequency (BAF) based on all informative parental homozygous calls for the autosomes and the X chromosome, respectively. For the autosomes in the uncultured amniotic cells, we expected the normal BAF band of 0.5 to split into two BAF bands at 0.71 and 0.29. Similarly, for the X chromosomes of the cultured amniotic cells, we expected the normal BAF of 0.5 to split at 0.64 and 0.36. These theoretically calculated values were in excellent agreement with the actually observed BAFs of the respective samples (Fig. 2).

We here report on a rare genetic disorder leading to the most extended form of multilocus imprinting disturbance. To the best of our knowledge, up to date, 18 cases of genome-wide parental UPD have been reported in the literature. Sixteen of these refer to patients with mosaic genome-wide UPD of paternal origin [12, 15–20]. Only two patients were reported with clear mosaic maternal genome-wide UPD [12, 21, 22]. In 1995, Strain et al. published the first report of genome-wide maternal UPD in a boy with aggressive behavior, hemifacial hypoplasia, and normal birth weight. They found a cell line with the karyotype 46,XX and genome-wide maternal UPD in nearly all peripheral blood cells and a cell line 46,XY in skin fibroblasts of the patient [21]. The second patient was identified in 2010 in a screening study of patients with SRS-like phenotype. This female patient had a mosaicism of one cell line with genome-wide maternal UPD (46,XX) and a second cell line lacking the second sex chromosome (45,X) [22]. Kotzot et al. list another two reports as mosaic genome-wide maternal UPD [12]. Both studies refer to a mosaic 46,XX/47,XY constellation observed prenatally with the three X chromosomes deriving from the same maternal homologue [23, 24]. However, in both reports, this observation does not extend to the whole chromosome set, and, thus, genome-wide UPD cannot be proven.

The phenotype of genome-wide uniparental diploidies results primarily from functional imbalance of virtually all parentally imprinted loci. Thus, reported symptoms overlap significantly with the characteristic features of imprinting syndromes. Leaving placental-specific imprinted regions aside, ubiquitous imprinted regions in the human were identified on chromosomes 1, 2, 4, 6, 7, 8, 10, 11, 13, 14, 15, 16, 19, 21, and 22 [25]. As mentioned above, some of these chromosomes are related to known imprinting syndromes, but not all of them are associated with specific symptoms when dysregulation affect only single loci. Clinical consequences of maternal UPD for chromosomes 6, 16, and 20 have been investigated recently [26]. Maternal UPD(20) has been proposed as a new imprinting disorder related to prenatal and postnatal growth retardation (IUGR and PNGR), severe feeding difficulties but without characteristic dysmorphisms [27]. Clinical consequences of maternal UPD(6) and UPD(16) are still under debate but most likely do not correspond to a specific phenotype [26]. From these considerations, the clinical phenotype of the patient is expected to consist of mixed features of Silver-Russell syndrome, Temple syndrome, Prader-Willi syndrome, and UPD(20)mat syndrome. At this stage of prenatal development, this would result primarily in intrauterine growth retardation. Indeed, this was one of the leading symptoms observed by prenatal ultrasound. Interestingly, the imprinted region in 11p15.5, harboring the H19/IGF2 IG-DMR and the KCNQ1OT1 TSS-DMR, is known to have a predominant influence on phenotype in patients with MLID [9, 19, 28]. In agreement with this observation, we identified next to intrauterine growth retardation a relative macrocephaly with a ratio of head to abdominal circumference of 1.61 at 18 + 1 weeks of gestation. Moreover, left diaphragmatic hernia (Bochdaleck hernia) and pseudodextrocardia were noted prenatally. These are neither typical symptoms for Silver-Russell syndrome nor for Trisomy X or any of the abovementioned maternal UPDs. In the light of the underlying fundamental genetic disorder, it is likely that the observed malformations are associated with maternal isodiploidy. They could either be associated with disturbed expression of certain imprinted genes or result from unmasking of maternally transmitted recessive mutation(s) in the genome-wide maternal UPD cell line.

Interestingly, some cases with mosaic genome-wide parental UPD contain a mosaicism for a further unbalanced chromosome disorder [18, 22, 29]. This may indicate that cases are missed and interpreted as normal when no second cell line with aberrant copy number of chromosomes is involved. Alternatively, the underlying genetic cause of genome-wide UPD might promote additional chromosome aberrations.

The presence of only one maternal allele in both cell lines narrows the time frame of occurrence down from after the first meiotic division to the first cleavage steps. Based on the existing models in similar constellation [16, 22, 29], we propose the possibilities depicted in Fig. 4 for the generation of the isodiploid cell line in combination with the biparental cell line with trisomy X. The scenario depictured in Fig. 4a assumes pathogenetic activation of the maternal pronucleus as primary event with one pronucleus giving rise to the cell line with genome-wide maternal isodiploidy by endoreplication and mating of the second maternal pronucleus with a sperm bearing two X chromosomes. Alternatively, the second polar body may have been restrained and after endoreplication represent the origin of the isodiploid cell line (Fig. 4b). Furthermore, a failure of the paternal pronucleus to duplicate could result in one biparental cell line and a second maternal haploid chromosome set that after endoreplication again could be the origin of the maternal isodiploid cell line (Fig. 4c). For the latter model, the failure of paternal genome duplication needs to be recognized by the cell organism, and as consequence, one maternal haploid chromosome set must get spatially separated from the other chromosomes. While this is possible, it seems to be the least likely mechanism. Whether the extra X chromosome was part of the genetic information in the sperm or was gained later in development of the biparental cell line is impossible to review and, thus, is displayed randomly as being part of the sperm.

Gold standard for investigation of chromosome aberration in prenatal diagnostics remains the conventional karyotype, even though noninvasive tests on free fetal DNA in the maternal blood gain more and more attention. To overcome the need for culturing, rapid investigations for aneuploidies associated with live-born children is routinely performed either by QF-PCR of microsatellite markers or FISH analyses. Choice of methods seems to be distributed rather randomly based on the methodical focus of the individual laboratory with a slight advantage seen in QF-PCR because it can be performed on fewer cells, can detect maternal contamination, and analyses can be automated with many samples processed at the same time [30]. Moreover, QF-PCR is informative regarding parental inheritance while FISH is not. Thus, a genome-wide parental isodiploidy is not detectable by FISH. If we would have performed initial testing for aneuploidies by FISH, we would have reported a mosaicism for a trisomy X, a genetic condition of minor clinical significance often observed as coincidental finding in asymptomatic females. However, QF-PCR does not reflect results on a single cell level, leading to the false interpretation of a triploid chromosome constellation in the presented case. Thus, we would like to alert colleagues based on our experience with this case to the of course well-known but nevertheless challenging limitations of the different methods applied in routine prenatal genetic testing.

We here report the third case of mosaic genome-wide maternal UPD leading to the most extended form of multilocus imprinting disturbance. Our findings highlight the technical challenges to come to the correct diagnosis in a routine prenatal setting, which is only possible by a combination of analysis of copy number per cell and parental inheritance to unmask genome-wide UPD. In line with previous studies, we observed mosaicism for a second cell line with an unbalanced chromosome disorder, namely a cell line of biparental inheritance with trisomy X and paternal isodisomy X. This case adds more data on clinical outcome of genome-wide maternal UPD, but further reports are needed to draw clear genotype-phenotype correlations.