Methylation and expression analyses of Pallister-Killian syndrome reveal partial dosage compensation of tetrasomy 12p and hypomethylation of gene-poor regions on 12p

ABSTRACT

To ascertain the epigenomic features, i.e., the methylation, non-coding RNA, and gene expression patterns, associated with gain of i(12p) in Pallister-Killian syndrome (PKS), we investigated single cell clones, harboring either disomy 12 or tetrasomy 12p, from a patient with PKS. The i(12p)-positive cells displayed a characteristic expression and methylation signature. Of all the genes on 12p, 13% were overexpressed, including the ATN1, COPS7A, and NECAP1 genes in 12p13.31, a region previously implicated in PKS. However, the median expression fold change (1.3) on 12p was lower than expected by tetrasomy 12p. Thus, partial dosage compensation occurs in cells with i(12p). The majority (89%) of the significantly deregulated genes were not situated on 12p, indicating that global perturbation of gene expression is a key pathogenetic event in PKS. Three genes—ATP6V1G1 in 9q32, GMPS in 3q25.31, and TBX5 in 12q24.21—exhibited concomitant hypermethylation and decreased expression. The i(12p)-positive cells displayed global hypomethylation of gene-poor regions on 12p, a footprint previously associated with constitutional and acquired gains of whole chromosomes as well as with X-chromosome inactivation in females. We hypothesize that this non-genic hypomethylation is associated with chromatin processing that facilitates cellular adaptation to excess genetic material.

Introduction

Pallister-Killian syndrome (PKS; MIM 601803) is a rare chromosomal disorder, first reported in the late 1970s and subsequently independently confirmed in the early 1980s,^1,2 that is cytogenetically characterized by tetrasomy 12p mosaicism through a supernumerary isochromosome 12p—i(12p) —, the presence of which is tissue-limited.³ Clinical characteristics include seizures, marked mental retardation, and dysmorphic facial features, such as high forehead with temporo-frontal alopecia, upslanting palpebral fissures, hypertelorism, flat and broad nasal bridge, short nose with upturned nares, a large mouth with downturned corners, and a prominent upper lip. Hypo- as well as hyper-pigmentation, rhizomelic limb shortness, and small hands and feet are also common, and so is a wide range of other congenital malformations, such as diaphragmatic hernia, ventricular septal heart defects, renal and anal malformations, deafness, and eye abnormalities.⁴

It has been demonstrated that the level of mosaicism for i(12p)-positive cells decreases during in vitro culturing and in peripheral blood with increasing age.^3,5 This differential growth between cells with disomy 12 and i(12p) may partly be due to a higher rate of apoptosis of the latter cells.⁶ However, it may also be caused by a decreased proliferation rate considering that whole chromosome aneuploidy has been shown to be associated with diminished cellular fitness in murine and yeast experimental systems;⁷ it is possible that this also holds true for structural chromosome aberrations resulting in large-scale genomic imbalances, such as isochromosomes, but that remains to be elucidated.

How aneuploid cells can tolerate as well as regulate extensive gene dosage imbalances in the context of DNA methylation and higher-order chromatin modifications is an intriguing question that has received increasing attention in recent years and to which the answer would have implications for both constitutional and acquired genetic changes. We and others have previously reported that widespread DNA hypomethylation of gene-poor regions on gained chromosomes, such as chromosomes 7 and 14 in colon cancer, trisomic/tetrasomic chromosomes in pediatric high hyperdiploid acute lymphoblastic leukemia, and chromosome 8 in constitutional trisomy 8 mosaicism, is an epigenetic hallmark of aneuploidy.^8-10 Whether this is a methylation feature also of partial chromosome gains, for example i(12p), is presently unknown.

The incentive for the present study was to investigate, for the first time, the methylome of i(12p)-positive fibroblasts in PKS and to compare it with the one in cells with disomy 12. Because only a few studies have investigated the gene and microRNA expression patterns in PKS,^11,12 we also analyzed the global gene expression, including non-coding RNA (ncRNA), in cells with and without i(12p). The methylation and expression analyses were performed using single cell cloning of fibroblasts from a patient with PKS, resulting in cultures with either disomy 12 or with non-mosaic i(12p); this approach was previously used to investigate epigenomic patterns associated with constitutional trisomy 8 mosaicism.¹⁰

Results

Single nucleotide polymorphism (SNP) array analysis: Origin of i(12p)

Based on the patterns of the B allele frequencies, 4 recombination sites were detected on the i(12p), as seen by the presence of 5 different genotype calls on 12p in 2 regions (Fig. 1A-C). This strongly suggests that the i(12p) occurred between meiosis II, after crossing-over had taken place in meiosis I, and fertilization.⁵ The distal recombination site at 12p (Fig. 1B) mapped in the vicinity of a previously reported PKS-associated cluster of recombination at 16 Mb.⁵

Figure 1.

Genomic characterization of i(12p) reveals that it occurred after crossing-over had taken place in Meiosis I. (A) Fluorescence in situ hybridization (FISH) and single nucleotide polymorphism (SNP) array analyses show 2 copies of 12p in the cultures with disomy 12. (B) FISH and SNP array analyses show 4 copies of 12p in the cultures with i(12p). The arrows indicate recombination sites. (C) Genotyping using the B allele frequency revealed 3 haplotypes (BB, BA, and AA) on 12p in cells with disomy 12 (left), 4 haplotypes (BBB, BBA, BAA, and AAA) on 12p not affected by crossing-over in cells with i(12p) (middle), and 5 haplotypes (BBBB, BBBA, BBAA, BAAA, and AAAA) on 12p involved in recombination in cells with i(12p) (right); the latter demonstrates that the i(12p) arose after recombination in meiosis I, between meiosis II and fertilization.

Expression array: General expression patterns of genes on chromosome 12

When comparing the gene expression signatures among the non-mosaic i(12p), disomy 12, and the 2 reference cultures combined, there was a significant overexpression of 37 of the 285 (13%) genes on 12p (Fig. 2A; ANOVA and Tukey's honestly significant difference tests; P ≤ 0.0001); in contrast, there was no chromosome-wide expression difference of genes on 12q between i(12p)-positive and disomic cells (Fig. 2B). The median expression fold changes were 1.3 for genes on 12p and 0.99 for those on 12q.

Figure 2.

Significant overexpression of genes on 12p in i(12p)-positive but not in disomy 12 and reference cultures. (A) Median centered gene expression (left) of 12p demonstrates that 13% of the genes on this chromosome arm are upregulated in cells with i(12p), as illustrated by order 4 polynomial trendlines. Independent and weighted ANOVA and Tukey's honestly significant difference tests (right) revealed an average expression fold change of 1.3 for genes on 12p in the i(12p)-positive cells. (B) Independent and weighted ANOVA and Tukey's honestly significant difference tests revealed no significant expression differences of genes on 12q between i(12p)-positive and disomic cultures.

Expression array: Unsupervised hierarchical clustering and principal component analyses (HCA, PCA) of gene and ncRNA expression

A standard unsupervised HCA, after using a variance filter with a cut-off value of 87% of the highest standard deviating genes to limit the number of probes, and the Pearson's correlation test for complete linkage clustering were used to investigate similarities and differences in global gene and ncRNA expression patterns among the i(12p), disomy 12, and the 2 reference cultures. The unsupervised HCA clustered the disomy 12 and reference samples together, whereas the i(12p)-positive cultures comprised a separate branch (Fig. 3A), indicating that the cells with i(12p) had a distinct expression signature. PCA of the 3 culture groups yielded similar results, with the i(12p)-positive cells being significantly distanced from the disomy 12 and reference cultures, which were in shortest Euclidean distance of each other (Fig. 3B). Analyzing only the expression of genes mapping to chromosome 12, unsupervised HCA of genes on 12p, but not of those on 12q, clustered the cells with i(12p) separately from the disomy 12 and reference cultures (Fig. 3C). When repeating the same analyses on the global ncRNA dataset only, no significant differences between the cells with i(12p) and disomy 12 could be demonstrated (data not shown).

Figure 3.

Gene expression profiling of i(12p) reveals that it is associated with a distinct expression pattern. (A) Unsupervised hierarchical clustering analysis (HCA) of global gene and ncRNA expression levels (“whole genome”) clustered the disomy 12 and reference samples together, whereas the i(12p)-positive cultures comprised a separate branch. (B) Principal component analysis (PCA) of global gene and ncRNA expression levels identified 480 genes that displayed significant expression differences among the i(12p), disomy 12, and reference cultures. (C) Unsupervised HCA of genes mapping to 12p clustered the i(12p)-positive cells separately from the disomy 12 and reference cells (top), whereas unsupervised HCA of genes on 12q clustered the cultures with i(12p) and disomy 12 in the same branch (bottom).

Expression array: Genes and ncRNAs that contribute to the PKS phenotype

Two-class unpaired significance analysis of microarrays (SAM) of the global gene and ncRNA expression in the i(12p)-positive cultures vs. the disomy 12/reference cultures combined revealed 172 significantly overexpressed and 155 significantly underexpressed protein-coding genes in the i(12p)-positive cells (Supplementary Table 1); there were no significantly differentially expressed ncRNAs. Among the 172 overexpressed genes, 37 (22%) were located on 12p; none of the 155 underexpressed genes mapped to this chromosome arm. In total, 290 (89%) of the 327 deregulated genes were not situated on 12p.

Quantitative real-time digital droplet PCR (ddPCR) was used to validate the expression patterns (identified by the array analysis) of 3 upregulated genes (ATN1, COPS7A, and NECAP1) and 3 downregulated genes (ATP6V1G1, GMPS, and TBX5) (Fig. S1A). The ddPCR gene expression fold change levels correlated well with the array data; the Pearson's correlation coefficient was r = 0.9776 (P = 0.0007; Fig. S1B).

Protein-protein interaction analysis (http://string-db.org)¹³ of the 327 differentially expressed genes, with an overlaying GeneMania analysis (http://www.genemania.org)¹⁴ to assess functional associations of the identified interactions, revealed several deregulated pathways in PKS, such as apoptosis, muscle contraction, and axon guidance (Fig. 4A). This was further validated on the same data set using the MetaCore data-mining analytical tool (Thomson Reuters, New York, NY), which identified overlapping and additional signaling pathways, including cell differentiation, protein degradation, inflammatory response, and transcription regulation (Fig. 4B). Furthermore, when performing the MetaCore “disease by biomarkers” analysis of the 327 genes, some phenotypic characteristics of PKS were top hits, such as alopecia and joint disease (Fig. 4C).^4,15

Figure 4.

Functional association analyses reveal key pathways that are dysregulated in i(12p)-positive cells in Pallister-Killian syndrome (PKS). (A) Protein-protein interaction and functional association analyses using String (illustration) and GeneMania (red text) of the 327 differentially expressed genes identified several key pathways that may be deregulated and hence implicated in the PKS phenotype. (B) Using the MetaCore data-mining analytical tool to identify dysregulated gene signaling, both pathways overlapping with those shown in (A) as well as additional ones were identified. (C) When performing the MetaCore “disease by biomarkers” analysis of the 327 differentially expressed genes, some phenotypic characteristics of PKS were top hits, such as alopecia and joint disease. (D) Two-class unpaired significance analysis of microarrays of methylation data revealed that the underexpressed gene TBX5 was hypermethylated at 2 independent CpG islands, which were located at the transcription start site and inside the gene body, respectively. The figure is based on the UCSC Genome Browser. (E) GeneMania functional association analysis revealed that TBX5 is not only involved in cardiac chamber morphogenesis but that it also interacts with important regulators of epithelial and mesoderm development.

Methylation array: Differentially methylated genes in i(12p)-positive cells

Two-class unpaired SAM on the methylation data of i(12p)-positive cultures vs. disomy 12/reference cultures combined revealed that 3 of the 155 underexpressed genes in the cells with i(12p) displayed promoter CpG hypermethylation with a concomitant, highly significant decreased expression: ATP6V1G1 (ATPase, H⁺ transporting, lysosomal 13kDa, V1 subunit G1) in 9q32, GMPS (guanine monphosphate synthetase) in 3q25.31, and TBX5 (T-box 5) in 12q24.21. The TBX5 gene was hypermethylated at 2 independent CpG islands, which were located at the transcription start site and inside the gene body, respectively (Fig. 4D). The hypermethylation of the CpG island located in the TBX5 promoter region was confirmed by bisulfite sequencing (Fig. S2). GeneMania functional association analysis of TBX5 revealed that it is involved in cardiac chamber morphogenesis and that it also interacts with important regulators of epithelial and mesoderm development (Fig. 4E). A similar analysis of ATP6V1G1 and GMPS did not reveal any relevant interactions or functions.

Methylation array: Chromosome-wide methylation patterns

As a proof-of-principle, we initially investigated the methylation pattern of the X chromosome in XX (female reference) and XY cultures (male reference and the patient). As expected, unlike its active counterpart, the inactivated X chromosome (Xi) in the XX cells exhibited widespread hypermethylation of gene promoters (Fig. 5A; Mann-Whitney U; P ≤ 0.0001) as well as hypomethylation of gene-poor regions (Fig. 5B; Mann-Whitney U; P ≤ 0.01), both considered epigenetic hallmarks of Xi.^8,16 No significant expression difference of X-linked genes between male and female cultures could be detected (Fig. 5C; Mann-Whitney U; P > 0.05; the median expression fold change was 0.999), indicating a virtually complete sex chromosome dosage compensation.

Figure 5.

Global hypomethylation of gene-poor regions on 12p in i(12p)-positive but not in disomy 12 and reference cultures. (A) Median centered promoter methylation (left) of the inactivated X chromosome demonstrates that the majority of genes on this chromosome are hypermethylated in the female reference culture compared with the male cultures, as illustrated by order 4 polynomial trendlines and verified using the Mann-Whitney U test (right). (B) Median centered non-genic methylation (left) of the inactivated X chromosome demonstrates that the majority of gene-poor regions on this chromosome are hypomethylated in the female reference culture compared with the male cultures, as illustrated by order 4 polynomial trendlines and verified using the Mann-Whitney U test (right). (C) The expression levels of genes on the X chromosome did not differ significantly (Mann-Whitney U test) between female and male cultures; the median expression fold change was 0.999. (D) Mean methylation levels of promoter-associated regions on 12p were not significantly different (Mann-Whitney U test) between cells with i(12p) or disomy 12. (E) Median centered non-genic methylation (left) of 12p demonstrates that the majority of gene-poor regions on this chromosome arm are hypomethylated in cells with i(12p) compared with cells with disomy 12, as illustrated by order 4 polynomial trendlines and verified using the Mann-Whitney U test (right).

No significant differential methylation patterns of promoter-associated regions between the cells with i(12p) and disomy 12 cells were detected on 12p or 12q (Figs. 5D and S3A; Mann-Whitney U; P > 0.05). In contrast, there was a significant (P < 0.05) hypomethylation of gene-poor regions on 12p (Fig. 5E), but not on 12q (Fig. S3B), in the i(12p)-positive cultures compared with those with disomy 12. The methylation levels in gene-associated as well as non-genic CpG islands in the PKS minimal critical region on 12p13.31¹⁸ did not differ from those in other regions of 12p in the i(12p)-positive cells (Fig. S4).

Discussion

Several mechanisms for the formation of i(12p) in PKS and of other constitutional, mosaic supernumerary isochromosomes, as cataloged in the small supernumerary marker chromosome database (http://ssmc-tl.com/chromosome-12.html#iso), have been suggested, including 1) centromeric misdivision during a germinal premeiotic mitosis followed by nondisjunction in meiosis I, 2) centromeric misdivision and nondisjunction in meiosis I, 3) nondisjunction in meiosis I followed by a postzygotic centromeric misdivision, and 4) pre-meiotic nondisjunction followed by centromeric misdivision either during meiosis II or postzygotically.¹⁷ Conlin et al.⁵ identified, by the use of SNP arrays, several cases with recombination sites on 12p and concluded that the i(12p) most frequently arises in meiosis II, after crossing-over has taken place in meiosis I. Our SNP array findings, which revealed 4 recombination sites (Fig. 1B-C), agree very well with that conclusion. Interestingly, the distal recombination site on 12p in our proband was located the same region (16 Mb) as a previously reported PKS-associated recombination cluster,⁵ suggesting that this chromosome segment may be associated with the formation of i(12p) in PKS.

Both unsupervised HCA and PCA of the gene expression patterns clearly revealed that the i(12p)-positive fibroblasts in our proband displayed a characteristic signature that was distinct from the one observed in the cells with disomy 12, both with regard to global expression profiles and expression of genes located on 12p (Fig. 3). This, at least to some extent, differs from the findings reported by Kaur et al.,¹² who performed genome-wide gene expression analyses of fibroblasts from 17 PKS probands and 9 unrelated controls. Although they found more than 350 differentially expressed genes, unsupervised clustering analysis did not clearly separate the PKS and control groups. The reasons why the cultures with disomy 12 and i(12p) in the present study clustered in different branches are most likely 2-fold. First, by single cell cloning we obtained fibroblasts with i(12p) in a non-mosaic form, thus avoiding the problem when analyzing cases with different levels of mosaicism. Second, by comparing the expression patterns between cultures with i(12p) and disomy 12 from the same individual, confounding factors such as gender and inter-individual variations could be excluded.

Based on a PKS patient with 2 small interstitial duplications of 12p and a review of 26 individuals with variably sized 12p duplications, Izumi et al.¹⁸ concluded that the minimal critical region for PKS mapped to 12p13.31, comprising 26 potential candidate genes for the PKS phenotype. The same group subsequently reported that the most significantly differentially expressed genes on 12p in PKS were also located in this particular sub-band.¹² It is hence worthy of note that 17 (46%) of the 37 overexpressed genes in our study mapped to this chromosome segment (online supplementary Table 1); this provides further support for this region being a key player in PKS. Indeed, among the genes in 12p13.31 previously reported to be overexpressed in PKS (ATN1, COPS7A, FOXJ2, ING4, LOC374443, NECAP1, RIMKLB, and TAPBPL),¹² 3 (ATN1, COPS7A, and NECAP1) were also significantly upregulated in the i(12p)-positive cells in our proband (online supplementary Table 1). Interestingly, ATN1, which when mutated causes dentatorubral pallidoluysian atrophy,¹⁹ interacts with several of the proteins encoded by the differentially expressed genes identified herein, such as ENO2, GAPDH, and VAMP1 in 12p13.31, UBC in 12q24.31, and EEF1A1 in 6q13 (online supplementary Table 1). COPS7A, which is a subunit of the COP9 signalosome complex that regulates protein ubiquination and subsequent degradation, has previously not been implicated in human disease; however, it may be noted that protein degradation was among the dysregulated pathways in PKS identified herein (Fig. 4). The pathogenetic effect, if any, of NECAP1 overexpression is presently unknown. Apart from the 3 above-mentioned overexpressed genes in 12p13.31 identified by us and Kaur et al.,¹² an additional 11 genes on 12p were found to be overexpressed in both studies, namely DNM1L, EPS8, FKBP4, ITFG2, JARID1A, KLHDC5, NDUFA9, NOP2, TM7SF3, USP5, and WBP11. Furthermore, deregulated expression of 4 genes not mapping to 12p was also seen in both studies, namely CCDC80 (in 3q13.2), DNAJC15 (13q14.11), FARP1 (13q32.2), and SH2B3 (12q24.12). Aberrant expression of these 15 genes hence also seems to be associated with the PKS phenotype.

Promoter CpG hypermethylation and concomitant decreased expression was not more frequently observed in the i(12p)-positive cells compared with the disomy 12 and reference cells, indicating that methylation does not play a major role in the gene deregulation seen in PKS. However, we identified ATP6V1G1, GMPS, and TBX5 as epigenetically deregulated candidate genes in PKS. The ATP6V1G1 protein is a component of a vacuolar ATPase associated with autosomal recessive distal renal tubular acidosis.²⁰ GMPS catalyzes the amination of xanthine monophosphate to guanine monophosphate in the biosynthesis of purine nucleotides²¹ and defects in this pathway have been linked to DNA and RNA expression errors though incorporation of deaminated nucleobases.²² TBX5 is involved in cardiac morphogenesis and also interacts with important regulators of epithelial and mesoderm development. Inactivating mutations of the gene encoding this transcription factor cause the Holt-Oram syndrome [MIM 142900], a disorder characterized by cardiac septal defects and skeletal abnormalities in the upper limbs.²³ However, further studies are needed to clarity the impact of these 3 genes in the pathogenesis of PKS.

The vast majority (89%) of the significantly deregulated genes identified was not located on 12p (Supplementary Table 1). This is in line with previous data showing that only a minority of the differentially expressed genes in PKS map to 12p.¹² Thus, the global gene expression patterns in PKS are analogous to what has been reported in Down syndrome, trisomy 13, constitutional trisomy 8 mosaicism, and Emanuel syndrome [+der(22)t(11;22)(q23;q11)] in the sense that deregulated genes are often located in chromosome regions that are not genomically gained.^10,24,25 Furthermore, of all the genes on 12p, only 13% were overexpressed in the i(12p)-positive cells, with a fold-change of 1.3 that is lower than expected considering the presence of 2 additional copies of 12p. Taken together, this strongly indicates a partial dosage compensation of tetrasomy 12p in PKS. The underlying mechanism for this is unknown; it apparently does not involve promoter hypermethylation. It may perhaps be caused by loss of histone modifications associated with active chromatin, such as H3K9 acetylation and H3K4 methylation, as has been implicated in X chromosome inactivation.²⁶ However, this remains to be elucidated.

Analysis of the chromosome-wide methylome revealed no significant differential methylation patterns of promoter-associated regions on chromosome arms 12p and 12q in the i(12p)-positive cells compared with the controls. However, a significant hypomethylation of gene-poor regions on 12p, but not on 12q, was detected. Noncoding DNA hypomethylation of acquired and constitutional chromosomal gains has previously been described in neoplasia and genetic syndromes.^9,10 The exact role of this epigenetic signature is unknown. Besides influencing gene expression, DNA methylation is involved in higher-order chromatin structure, regulating nucleosome dynamics.²⁷ In fact, it has been demonstrated that hypomethylation increases nuclear clustering of pericentric heterochromatin, thus directly influencing chromatin organization.²⁸ Furthermore, large hypomethylated blocks with extreme gene expression variability is a universal, defining epigenetic alteration in human solid tumors.²⁷ These blocks correspond to large organized chromatin K9 modifications and lamina-associated domains,^29,30 regions that expand during differentiation³¹ and are associated with transcriptional repression.^32,33 One may hence speculate that the observed hypomethylation of noncoding regions in i(12p) reflects chromatin remodeling and nuclear positioning that may enable cellular adaptation to aneuploidy. However, this remains to be tested experimentally.

Material and methods

Patient history and single cell cloning

The PKS patient was born prematurely in week 27 and weighed 1190 g. On physical examination, short long bones, small feet, hypertelorism, and hypotonia were observed. Gain of the short arm of chromosome 12 was identified by SNP array analysis. Subsequent cytogenetic analyses of fibroblasts and peripheral blood revealed 47,XY,+i(12)(p10)[23]/46,XY[2] and 47,XY,+i(12)(p10)[5]/46,XY[20], respectively. After the cytogenetic analysis, the fibroblasts were viably frozen in liquid nitrogen.

For the present study, the fibroblasts were thawed and cultured in standard conditions with RPMI 1640 medium (Life Technologies, Carlsbad, CA) containing 10% FBS (Sigma-Aldrich, St Louis, MI). The culture was then single cell cloned as previously described.¹⁰ Interphase fluorescence in situ hybridization analysis, using the Cy3 CEP-12 and Cy5 TelVysion 12p probes (Abbott Molecular, Des Plaines, IL), was applied to ascertain i(12p)-positive and -negative cells. A minimum of 285 (95%) of 300 normal or abnormal nuclei was used as a cut-off to designate a culture as being either 46,XY [“disomy 12”] or non-mosaic 47,XY,+i(12)(p10) [“i(12p)”]. Three cultures with disomy 12 (cultures 1–3) and 3 with i(12p) (cultures 4–6) were then passaged to cell culture flasks. The six cell cultures were harvested simultaneously, after which RNA and DNA were extracted; the total time in culture was 7 weeks with a total of 5 passages from thawing of the cells. Dermal fibroblasts from a male neonate and a female adult, obtained from the American Type Culture Collection (Manassas, VA), were used as additional controls (cultures 7–8). The regional ethics board at Lund University approved the study, which was conducted according to the Declaration of Helsinki.

SNP, expression, and methylation arrays

Total DNA, mRNA, and ncRNA were extracted from the patient's fibroblasts with either disomy 12 or i(12p) and from the reference fibroblasts using the DNeasy Blood and Tissue kit (Qiagen, Venlo, The Netherlands) and the mirVana kit (Life Technologies). The concentrations were measured with a NanoDrop 1000 spectrophotometer (Thermo Scientific, Waltham, MA) and the RNA quality was assessed with a 2100 Bioanalyzer (Agilent Technologies, Santa Clara, CA).

SNP and methylation array analyses were performed on genomic DNA from the patient and reference cultures using the Human1M-Duo BeadChip and the HumanMethylation450 BeadChip platforms (Illumina, San Diego, CA), respectively. All hybridizations were done according to the manufacturer's instructions. Quality controls, β value calculations, and initial analyses were performed using the BeadStudio 3.1.3.0 and GenomeStudio 2003.1 software (Illumina). Probe positions were extracted from the GRCh37 genome build. For DNA bisulfite conversion, the EZ-96 DNA Methylation Kit (Zymo Research, Irvine, CA), including normal male (Promega, Madison, WI) and methylated/unmethylated (Zymo Research) DNA controls, was used.

Total RNA was hybridized to the DirectHyb HumanHT-12 v4.0 (Illumina) and GeneChip miRNA arrays (Affymetrix, Santa Clara, CA) according to the manufacturers' protocols. Quality controls, quantile normalizations, and initial analyses were performed using the BeadStudio 3.1.3.0 and GenomeStudio 2003.1 software (Illumina) and the miRNA QC Tool (Affymetrix).

Bioinformatic analyses

HCA and SAM were performed using the MultiExperiment Viewer v4.9³⁴ to identify genes and ncRNAs that were differentially expressed and methylated. The Benjamini-Hochberg method was applied for error correction by q-value calculation, with data confidence set to ≤ 5%. PCA, based on Pearson correlation matrix (mean = 0, σ = 1) and filtering variance by σ/σ_max, was used to investigate relationships between subgroup data patterns using the Qlucore Omics Explorer v2.3 (Qlucore AB, Lund, Sweden). All statistical analyses were performed with Prism 6 (GraphPad software, La Jolla, CA).

Bisulfite sequencing

To investigate the reliability of the HumanMethylation450 BeadChip methylation array, the TBX5 gene was selected for bisulfite sequencing. DNA (2 µg) from 4 samples, comprising one i(12p), one disomy 12, and the 2 reference cultures, was treated with sodium bisulfite using the Cells-to-CpG Bisulfite Treatment kit (Life Technologies). The bisulfite-treated DNA was used as a template in standard PCR amplification using 3 primer pairs directed toward the CpG island located in the promoter region of TBX5, indicated as hypermethylated in the i(12p) but not in the disomy 12 or references cultures by the HumanMethylation450 BeadChip methylation array. The primers were designed using the MethPrimer software.³⁵ The PCR products were sub-cloned using the TOPO-TA system (Life Technologies) and sequenced using standard methods (Eurofins, Ebersberg, Germany). A minimum of 8–12 clones were sequenced and analyzed using the BiQ Analyzer software.³⁶ For a gene to be considered hypermethylated, the following criteria were used: at least 80% of the investigated clones should be methylated in at least 80% of the sequenced CpG sites.

Quantitative real-time ddPCR

The QX100 ddPCR system (Bio-Rad, Hercules, CA) was used to validate the expression levels of 6 significantly dysregulated genes: the upregulated genes ATN1, COPS7A, and NECAP1 and the downregulated genes ATP6V1G1, GMPS, and TBX5. RNA expression levels from all samples were measured with commercially available FAM labeled prime-time qPCR probes (Integrated DNA Technologies, Coralville, IA), standardized to HEX labeled TBP and GAPDH prime-time qPCR probes (Integrated DNA Technologies). The ddPCR was prepared using 10 μl of Bio-Rad One-Step RT-ddPCR Supermix, 1.25 μl of each of 20x primer/probe mix (Integrated DNA Technologies), 0.8 μl 25 mM Manganese solution along with 1 μl 30 ng/μl of total RNA. The reaction mix was then emulsified with 70 μl of oil using a QX-100 droplet generator. Following PCR amplification, droplets were quantified by the QX100 droplet reader (Bio-Rad). The QuantaSoft software version 1.6.1 (Bio-Rad) was used to quantify the copies/µl of each queried target per well. All samples were run in triplicates for all genes analyzed.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors thank Dr Lao H. Saal at the Translational Oncogenomics Unit at the Medical Faculty, Lund University, Sweden, for advice on and access to ddPCR equipment.

Funding

This work was supported by the Swedish Research Council.