Mosaic maternal uniparental disomy of chromosome 15 in Prader–Willi syndrome: Utility of genome-wide SNP array

Abstract

Prader–Willi syndrome is caused by the loss of paternal gene expression on 15q11.2–q13.2, and one of the mechanisms resulting in Prader–Willi syndrome phenotype is maternal uniparental disomy of chromosome 15. Various mechanisms including trisomy rescue, monosomy rescue, and post fertilization errors can lead to uniparental disomy, and its mechanism can be inferred from the pattern of uniparental hetero and isodisomy. Detection of a mosaic cell line provides a unique opportunity to understand the mechanism of uniparental disomy; however, mosaic uniparental disomy is a rare finding in patients with Prader–Willi syndrome. We report on two infants with Prader–Willi syndrome caused by mosaic maternal uniparental disomy 15. Patient 1 has mosaic uniparental isodisomy of the entire chromosome 15, and Patient 2 has mosaic uniparental mixed iso/heterodisomy 15. Genome-wide single-nucleotide polymorphism array was able to demonstrate the presence of chromosomally normal cell line in the Patient 1 and trisomic cell line in Patient 2, and provide the evidence that post-fertilization error and trisomy rescue as a mechanism of uniparental disomy in each case, respectively. Given its ability of detecting small percent mosaicism as well as its capability of identifying the loss of heterozygosity of chromosomal regions, genome-wide single-nucleotide polymorphism array should be utilized as an adjunct to the standard methylation analysis in the evaluation of Prader–Willi syndrome. © 2012 Wiley Periodicals, Inc.

INTRODUCTION

Prader–Willi syndrome (PWS) is characterized by obesity, hyperphagia, developmental delay, hypotonia, feeding problems, and hypogonadism [Cassidy et al., 2012]. PWS is caused by the loss of paternal gene expression at 15q11.2–q13.2, and one of the mechanisms resulting in the PWS phenotype is maternal uniparental disomy (UPD) of chromosome 15 (matUPD15) [Nicholls et al., 1989]. The majority of matUPD15 cases involve the mixture of iso/heterodisomic segments across the chromosome 15, and maternal isodisomy of the entire chromosome 15 is seldomly seen in individuals with PWS-associated UPD15 [Robinson, 2000; Altug-Teber et al., 2005]. There are three main mechanisms leading to UPD—trisomy rescue, monosomy rescue, and post-fertilization mitotic error [Liehr, 2010]. Any of these 3 mechanisms can present as a tissue-limited mosaicism (Fig. 1). For example, trisomic cells can be observed in cases of UPD resulting from trisomy rescue. Similarly, monosomic cells and normal diploid cells can be observed in cases of monosomy rescue and post-fertilization errors, respectively. Therefore, the presence of a mosaic cell line provides clues for the underlying mechanism of UPD.

Figure 1. Schematic illustration of the mechanisms leading to mosaic uniparental disomy including trisomy rescue, monosomy rescue, and post-fertilization mitotic error.

Various cytogenetic techniques, including traditional chromosome analysis using G-banding and fluorescence in situ hybridization (FISH), have long been used for the detection of such mosaic cell lines. However, because these studies require actual visualization of individual cells, they are not optimal in the identification of low-level mosaicism. Therefore, low-level mosaicism coexisting with UPD may have gone unrecognized in the past. The more recent development of single nucleotide polymorphism (SNP)-based microarray (SNP array) analysis can overcome this problem of detecting low-level mosaicism, because analysis of SNP allele patterns provides sufficient sensitivity for the detection of mosaicism than routine clinical cytogenetic testing [Conlin et al., 2010]. Furthermore, SNP arrays can detect long contiguous stretches of homozygosity suggestive of uniparental isodisomy, making SNP arrays a very useful tool for the detection of UPD events [Conlin et al., 2010; Kearney et al., 2011; Papenhausen et al., 2011].

Here, we report on two infants with PWS caused by mosaic, maternal uniparental disomy 15, as detected by SNP array analysis. The purpose of this report is to demonstrate the utility of SNP arrays for the evaluation of an imprinting disorder and to discuss potential mechanisms resulting in the unique mosaic maternal UPD15.

MATERIALS AND METHODS

Clinical Report

Patient 1

Patient 1 is a newborn baby with history of intrauterine growth retardation and hypotonia. She was born to a 21-year-old G₁P_0–1 mother after 37 weeks of pregnancy. Family history was unremarkable. Pregnancy was complicated by intrauterine growth retardation and oligohydramnios. She was delivered by cesarean due to breech presentation at the time of rupture of the membrane. Birth weight was 2,013 g (less than 3rd centile, 50th centile for 34 weeks of gestation). Poor suck, weak cry, and hypotonia were noted after birth. Brain MRI was unremarkable. Because of poor suck, she required gavage feeding. At 1-week-old, her weight was 1,975 g (less than 3rd centile, 50th centile for 34 weeks of gestation), length was 45.6 cm (15th centile), head circumference was 31 cm (less than 3rd centile, 50th centile for 33 weeks of gestation). Her physical examination revealed narrow forehead, upslanted palpebral fissure, bilateral epicanthus, underfolded left ear helix, slightly high palate, micrognathia, wide spaced nipples, and very long slender fingers. Neurological exam showed significant hypotonia. There were no hemihypertrophy or irregular hyperpigmentation spots, suggestive of tissue-limited mosaicism. Clinically, her physical features were suggestive of PWS.

She was re-evaluated by geneticists at the age of 6 months. Gavage feeding was discontinued, and she was able to be fed by mouth. Developmentally, she was able to roll from prone to supine position, she could track objects, and demonstrated smiling. Her weight was 5.57 kg (5th centile), height was 57.3 cm (below 5th centile, 50th centile for 2 months of age), and head circumference was 40 cm (25th centile). The physical exam demonstrated almond-shaped eyes with epicanthus, underfolded left ear helix, and wide spaced nipple (Fig. 2). Her hand length was below the 5th centile. On neurological exam, hypotonia remained.

Figure 2. Facial appearances of Patient 1 at the age of 6 months. Note narrow forehead, almond-shaped eyes, bilateral epicanthus, and underfolded left ear helix.

Patient 2

Patient 2 is a baby girl born to a 41-year-old G₄P_3–4 mother delivered by cesarean due to breech presentation at 36 weeks of gestation. The pregnancy was complicated by intrauterine growth retardation identified during the third trimester. Her birth weight was 1,946 g (less than 10th centile; 50th centile for 34 weeks of gestation), length was also less than 10th centile; 50th centile for 32 weeks of gestation, and head circumference was 25th centile. Upon birth, she was noted to have mild facial dysmorphisms and hypotonia as well as feeding difficulty. Her physical examination revealed mild upslanted palpebral fissures, bilateral mild adduction of thumbs, mild hypotonia, and was awake and alert, but demonstrated diminished spontaneous movements. Her palm length was 60th centile and middle finger length was 25th centile. There were no hemihypertrophy or skin pigmentation anomalies.

Cytogenetic and Molecular Evaluations

Methylation Sensitive PCR (MS-PCR) was performed on bisulfite converted genomic DNA using primer sets covering the SNRPN loci. Briefly, DNA was treated with sodium bisulfite to modify unmethylated cytosine residues using the EZ DNA Methylation-Gold Kit (Zymo Research, Irvine, CA) according to manufacturer's instructions. MS-PCR was performed using primers specific for the methylated (maternal) and unmethylated (paternal) alleles of the SNRPN gene as previously described [Kubota et al., 1997; Zeschnigk et al., 1997]. Amplification products were electrophoresed on an agarose gel along with appropriate positive and negative controls (Supplementary eFig. 1 see Supporting Information online). In unaffected individuals, two PCR products of 174 and 100 bp are present. In patients with Prader–Willi syndrome, the 100 bp product from the paternal allele is absent and only the 174 bp product is observed. In patients with Angelman syndrome, the 174 bp product from the maternal allele is absent and only the 100 bp product is observed. Primer sequences and PCR conditions are available upon request. Methylation-sensitive multiplex ligation-dependent probe amplification (MS-MLPA) analysis was performed using a commercially available kit (MRC-Holland, Amsterdam, Netherlands). Normal and positive controls were processed along with patient samples. Data analysis for methylation defects was performed following manufacturer's instructions (MRC Holland, Inc., Amsterdam, Netherlands and Softgenetics, Inc., State Collage, PA). FISH studies were performed using standard protocols with a Spectrum Red™ (Abbott Molecular, Des Plaines, IL) labeled fosmid probe mapping to 15q11.2 (SNRPN, G248P86612A8). SNP array was performed using the Illumina HumanQuad610 BeadChip on DNA prepared directly from peripheral blood and buccal swab of Patient 1, and SNP array was performed using the Illumina Omni1-Quad on a peripheral blood DNA sample of Patient 2. All the molecular analysis was performed from the peripheral blood samples. G-banded chromosome analysis was performed on PHA-stimulated peripheral blood lymphocytes of Patient 2 using standard methods.

RESULTS

Patient 1

MS-PCR of SNRPN gene demonstrated a strong signal from the maternal allele; however, a faint band from the paternal allele was also identified in Patient 1. This faint band suggested the presence of mosaic methylation defect, which could be either due to deletion, uniparental disomy or imprinting defects (Supplementary eFig. 1 see Supporting Information online). The result was confirmed by using a second primer set covering the SNRPN loci. Sequence analysis of the genomic region complimentary to the methylation specific PCR primers failed to demonstrate any polymorphic base changes that could explain the poor amplification of the paternal allele. MS-MLPA analysis was performed, which revealed a pattern consistent with the diagnosis of a mosaic methylation defect that would lead to PWS. FISH with a fosmid probe, G248P86612A8, revealed a normal pattern (i.e., no deletions or duplications) in 50 metaphase cells on the cultured peripheral blood (data not shown). SNP array revealed mosaic loss of heterozygosity of entire chromosome 15 (Fig. 3). The mosaic loss of heterozygosity is estimated to be at the level of approximately 85–90% in peripheral lymphocytes [Conlin et al., 2010]. No trisomy or monosomy was suggested by the SNP array, therefore Patient 1 was presumed to be mosaic for uniparental isodisomy 15 and a normal biparental cell line. In Patient 1, SNP array was also performed with the cheek swab sample, which revealed the similar mosaic loss of heterozygosity for all of chromosome 15 with the same level of mosaicism (data not shown).

Figure 3. Result of single nucleotide polymorphism (SNP) genome-wide array analysis. A: SNP array analysis revealed mosaic loss of heterozygosity for all of chromosome 15 in Patient 1. B: Mosaic trisomy and a genotyping pattern demonstrating a region of isodisomy of chromosome 15 in Patient 2. Top: Log R ratio demonstrating two copies of chromosome 15. Bottom: B-allele frequency showing the presence of a second genotype at a low percentage across the entire chromosome.

Patient 2

MS-PCR of SNRPN gene demonstrated a strong signal from the maternal allele as well as a faint band from the paternal allele in Patient 2. SNP array demonstrated mosaicism for trisomy 15 at approximately 5–10% and a genotyping pattern consistent with mixed uniparental hetero and isodisomy 15 in the majority of cells. G-banded chromosome analysis on Patient 2 demonstrated a single trisomy 15 cell in 40 cells examined (Supplementary eFig. 2 see Supporting Information online).

In summary, Patient 1 had two cell lines composed of uniparental isodisomy 15 cells and normal biparental cells, and Patient 2 had two cell lines composed of uniparental hetero/isodisomy 15 cells and trisomy 15 cells.

DISCUSSION

In this study, we describe infants with PWS and intrauterine growth retardation who were found to have mosaic maternal UPD15. Phenotypically, neither case showed any atypical features for neonatal PWS [Oiglane-Shlik et al., 2006]. Furthermore, there were no clinical signs suggestive of mosaicism, such as skin pigmentation abnormalities, or hemihypertrophy. Therefore, on a clinical basis, the mosaic condition was not considered in the differential diagnosis. Our study verified the utility of SNP array analysis for the detection of low-level mosaicism [Conlin et al., 2010].

The findings in Patient 1 are very unique; this patient showed uniparental isodisomy of the entire chromosome 15. Because of meiotic recombination, even UPD events involving an entire chromosome are usually not purely isodisomic or heterodisomic, but instead often have a mixture of isodisomic and heterodisomic segments [Liehr, 2010]. Therefore, maternal uniparental isodisomy of the entire chromosome 15 is seldom seen in patients with PWS caused by matUPD15 [Robinson, 2000]. In the absence of a mosaic cell line such as monosomic cells, trisomic cells, or heterodisomic cells, it is difficult to determine whether monosomy rescue, trisomy rescue, or post-fertilization mitotic error is the cause of uniparental isodisomy (Fig. 1). Therefore, in the previously reported cases with maternal uniparental isodisomy 15, the mechanism leading to UPD cannot be determined. Patient 1 represents the first reported case of PWS caused by mosaic maternal uniparental isodisomy 15 without trisomy. The presence of mosaic uniparental isodisomy provides a unique opportunity to determine the mechanism of UPD. The presence of chromosomally normal cell lines and isodisomy in Patient 1 suggests that post-fertilization mitotic error is the cause of UPD (Fig. 1). On the basis of molecular analyses demonstrating the presence of cells containing biparental chromosome 15 without trisomic cells, we conclude that Patient 1 is the first case to demonstrate a post-fertilization mitotic error that led to maternal uniparental isodisomy 15.

A single case of mosaic uniparental disomy 15 without trisomic cell lines has been described by Horsthemke et al. [2003]. Horsthemke et al. described a case of PWS who was found to have mosaicism for normal cells and cells with maternal UPD 15. Microsatellite analysis of several loci within chromosome 15 revealed three haplotypes, suggesting the trisomy rescue as a mechanism of mosaic UPD15 in their case, although trisomic cells were not demonstrated. From a mechanistic standpoint, the case described by Horsthemke et al. is distinctively different from that of Patient 1, and is similar to that of Patient 2. Both the case of Horsthemke et al. and Patient 2 exhibited maternal uniparental heterodisomy resulting from trisomy rescue. In the medical literature, there are several reports of mosaic uniparental disomy resulting from trisomy 15 rescue manifested with PWS phenotype [Milunsky et al., 1996; Devriendt et al., 1997; Olander et al., 2000]. However, even in the cases of prenatally identified mosaic trisomy 15 cases, the postnatal detection of trisomy 15 cells in peripheral blood cells is a rare phenomenon, and to demonstrate the presence of trisomic cells, it often requires skin biopsy [Cassidy et al., 1992; Purvis-Smith et al., 1992; Milunsky et al., 1996; Olander et al., 2000]. In fact, the proportion of trisomic cells was very low in Patient 2. Therefore, the ability of SNP array analysis of detecting low-percent mosaicism is well demonstrated in the detection of such a lower percentage trisomic cell line of Patient 2. Discerning between mat UPD15 associated with and without trisomy 15 mosaicism has significant prognostic implications, because trisomy 15 mosaicism cases tend to have more severe systemic phenotype including congenital heart defects [Olander et al., 2000].

In the report of Horsthemke et al., MS-PCR showed a faint band from the paternal allele in addition to a strong band corresponding to the maternal allele, which was similar to the result observed in Patients 1 and 2 (Supplementary eFig. 1 see Supporting Information online) [Horsthemke et al., 2003]. Such a faint band, in conjunction with a strong paternal band in MS-PCR, should raise suspicion of mosaic methylation defects.

The findings in our patients demonstrate the utility of SNP array analysis for the evaluation of imprinting disorders such as PWS. First, SNP arrays can detect low-level mosaicism, particularly copy-number-neutral mosaicism, which is unlikely to be detected by array-based comparative genomic hybridization or standard chromosome analysis. The presence of mosaicism often suggests the mechanism of chromosomal abnormality in the case. Second, SNP arrays can detect the presence of homozygous regions, which could be associated with UPD and could manifest as imprinting disorders. These features make SNP array analysis an ideal tool for the evaluation of imprinting disorders. Identification of isodisomy is clinically important because it can unmask recessive mutations. Such an example has been demonstrated in a case of Angelman syndrome caused by paternal uniparental isodisomy 15 and presenting with tyrosinemia type 1, which is caused by homozygous mutation in the FAH gene located at 15q25.1 [Ferrer-Bolufer et al., 2009]. We recommend the use of SNP arrays as an adjunct to methylation analysis for the evaluation of imprinting disorders.

The reason for the rarity of mosaic matUPD15 remains unknown. We hypothesize that the rarity of mosaicism in PWS might represent the fact that for the development of PWS, a high epigenetic mutation load is required in the brain and medium- to low-level mosaicism may not result in an apparent phenotype. The finding in Patient 1 that matUPD15 is observed in 85–90% of cells supports this hypothesis. Another possibility is that mosaic matUPD15 is underdiagnozed because of the technical difficulty associated with its assessment (i.e., methylation-sensitive PCR may not be sufficiently sensitive for detecting very low-level mosaicism). In general, monosomy/trisomy is not tolerated and will be actively selected against these abnormal cells, as demonstrated in the peripheral blood samples of Pallister–Killian syndrome [Tang and Wenger, 2005; Conlin et al., Manuscript Submitted]. These facts probably explain the reason of very small percentage of cells were abnormal trisomic/monisomic cells, complicating the identification of mosaic status. Further studies on PWS patient samples by using SNP arrays would clarify whether mosaic UPD is truly rare or is merely underdiagnozed in individuals with PWS.

In summary, we report on two infants with PWS caused by mosaic maternal UPD15. One of the cases showed mosaicism for uniparental isodisomy of the entire chromosome 15 and normal biparental cell line, supporting the hypothesis that post-fertilization mitotic error is the mechanism underlying PWS. This report further highlights the utility of SNP arrays for the evaluation of imprinting disorders. We recommend that SNP arrays be used for the evaluation of PWS as an adjunct to methylation analysis of the SNRPN locus.

Acknowledgements

We thank the members of the Clinical CytoGenomics Laboratory and the Molecular Genetics Laboratory at The Children's Hospital of Philadelphia for patient sample analyses.