A epigenética aborda o controle da expressão gênica através de diversos fatores que agem sob a cromatina, os melhor estudados são a metilação do DNA e a acetilação em histonas, relacionadas à repressão e ativação gênica, respectivamente. Em mamíferos, existem dois fenômenos epigenéticos interessantes: a inativação do cromossomo X (ICX) em fêmeas, que garante o equilíbrio transcricional gênico entre os sexos, e o imprinting genômico, caracterizado pela expressão monoalélica dependente da origem parental. No presente estudo, propusemos verificar a manutenção do controle epigenético em células humanas normais e transformadas em condições semelhantes de hipometilação do DNA e hiperacetilação em histonas (após uso das drogas 5-aza-2-\'deoxicitidina (5-aza-dC) e ácido valproico, respectivamente), através do monitoramento da expressão alelo-específica pelo uso de polimorfismos de única base presentes em regiões codificadoras. Em células normais houve manutenção da ICX e do imprinting genômico, enquanto que em células transformadas hipometiladas foram observadas indução de XIST, e perda de imprinting dos genes IGF2, H19 e PEG10. Observamos que ambas as drogas podem diminuir a expressão de DNMT1, e 5-aza-dC altera o equilíbrio entre acetilação e desacetilação da histona H4. Ainda, a ordem de adição dos reagentes ocasionou diferenças no nível de acetilação da histona H4 e na expressão gênica de XIST e PEG10. Nossos dados sugerem que: células humanas normais apresentam maior estabilidade do controle epigenético comparadas às células humanas transformadas, genes submetidos à ICX e \"imprintados\" não apresentam diferenças na rigidez do controle de expressão, e a cascata de reação seguida após perturbação de marcas epigenéticas pode ser alterada dependendo da modificação inicial. / Epigenetics refers to mechanisms related to gene activity through conformational modifications in DNA without changes in the nucleotide sequence. Key players in the epigenetic control are DNA methylation and histone acetylation, which are related to gene activation and repression, respectively. Two striking epigenetic phenomena in mammalians are X chromosome inactivation (XCI) and genomic imprinting. XCI triggers the transcriptional silencing of all but one X chromosome in each female cell, while genomic imprinting is a process that leads to mono-allelic gene expression based on parental origin. In the present study, we intended to verify the maintenance of epigenetic control in normal and transformed human cells under the same conditions of epigenetic disturbance. For this purpose, 5-aza-2\'-deoxycytidine (5-aza-dC) and valproic acid (VPA) were used to cause DNA hypomethylation and histone hyperacetylation, respectively. By monitoring allelic-specific expression using single nucleotide polymorphisms present in coding regions, we were able to check the effects of the modifications in the expression pattern of imprinted or subjected to XCI genes. While in female normal cells XCI and genomic imprinting were not affected by VPA or 5-aza-dC treatments, transformed male cells showed XIST activation and loss of imprinting of PEG10, IGF2 and H19 genes in the hypomethylation scenario. In addition, both drugs can decrease the expression of DNMT1, and 5-aza-dC alters the balance between acetylation and deacethylation of histone H4. Furthermore, we could see different degrees of histone H4 acetylation levels and of XIST and PEG10 expression, depending on which of the drugs was added first. Our data suggest that the epigenetic control in normal human cells is more stable when compared to transformed human cells. In addition, both XCI and genomic imprinting are epigenetic features equally hard to disturb. Finally, depending on the initial epigenetic modification (global demethylation or acethylation), it will induce different epigenetic control networks, with consequence to the final status of gene expression.

La différenciation et le développement requièrent une régulation fine de l’expression desgènes, médiée en partie par les modifications épigénétiques. Parmi les modificationsd’histones, la chromatine bivalente, signature chromatinienne atypique associant lesmarques permissive H3K4me2/3 et répressive H3K27me3, est de par sa plasticité, pressentiepour jouer un rôle décisionnel dans l’acquisition d’une identité cellulaire. Pour étudier le rôlede la chromatine bivalente au cours du développement, nous avons choisi d’utiliserl’empreinte parentale. Ce cadre développemental bien caractérisé, conduit à l’expression decertains gènes à partir d’un seul des deux allèles selon son origine parentale. La méthylationdifférentielle de l’ADN d’une région clé, appelée ICR (Imprinting Control Region), bienqu’absolument requise pour l’expression mono-allélique de ces gènes, n’est pas suffisantepour rendre compte de la complexité du profil d’expression de ces gènes suggérantl’implication d’autres mécanismes. Sur 15 ICR méthylés sur l’allèle maternel, nous avonsprécisément mis en évidence que la chromatine bivalente est présente par défaut sur l’allèlenon-méthylé lorsque celui-ci est transcriptionnellement inactif, quel que soit le stadedéveloppemental ou le tissu étudié, participant ainsi à la régulation fine de l’expressiontissu-spécifique à partir de ces régions. Dans leur ensemble, nos données révèlent que lachromatine bivalente joue un rôle moins dynamique que pressentie. Ainsi, au niveau del’empreinte parentale, sa fonction principale serait de protéger l’allèle non-méthylé des ICRcontre l’acquisition de méthylation tout en aidant à le maintenir réprimé dans certainstissus. Nous proposons que la chromatine bivalente joue un rôle similaire sur l’ensemble desîlots CpG du génome, contribuant ainsi à la protection de l’identité cellulaire. Afin decompléter cette première étude, j’ai étudié la régulation de l’expression d’un candidat de larégulation de la dynamique de la chromatine bivalente, l’histone déméthylase pourH3K27me3, JMJD3. Les résultats obtenus suggèrent que l’induction d’expression observéeau cours de la différenciation neurale s’appuie sur une dynamique de la structuretridimensionnelle de la chromatine qui pourrait elle-même être régulée par la transcriptiond’un eARN (enhancer ARN) et l’hydroxyméthylation. Ce modèle souligne un mode derégulation complexe de ce nouvel acteur épigénétique, impliquant des régionsintragéniques, et pourrait notamment permettre de comprendre les mécanismes impliquésdans sa dérégulation dans les cancers. / Fine-tuned regulation of gene expression is required for cell fate determination anddevelopment. Epigenetics modifications are well documented to be instrumental in thisprocess. Among them, bivalent chromatin, an unusual chromatin signature, which associatesthe permissive mark H3K4me2/3 and the repressive mark H3K27me3, is believed to arbitrategene expression during cell commitment. To study its precise role in development, we haveundertaken to study bivalency in the context of genomic imprinting. This well-defineddevelopmental frame is a process restricting expression of some genes to one parental alleleonly. The constitutive differential DNA methylation at the key region called ICR (ImprintingControl Region), is absolutely required but not sufficient to explain the complexity of themono-allelic expression pattern of imprinted genes, indicating that other mechanisms couldbe involved. Specifically, on 15 maternally methylated ICR, we showed that bivalentchromatin is acquired by default on the unmethylated allele of ICR when it istranscriptionally inactive whatever the developmental stage or the tissue studied and thuscontribute to tissue-specific expression from these regions. Altogether, our results revealthat chromatin bivalency is much less dynamic than proposed. In the context of genomicimprinting, it seems to plays more a safeguard function at ICR by protecting theunmethylated allele against DNA methylation acquisition while keeping it silent in a subsetof tissues. To complete this study, I studied the regulation of JMJD3, a histone demethylasefor H3K27me3, candidate to regulate bivalency dynamic. Our results suggest that theinduction of Jmjd3 expression observed during neural differentiation rely on the dynamic ofthe tridimensional architecture at the locus which could be regulated by the transcription ofan eRNA (enhancer RNA) and by hydroxymethylation. This model highlight a complex way ofregulation for this new epigenetics actor, involving intragenic regions and could help tounderstand how Jmjd3 expression is deregulated in a pathological context such as in cancer.

Genomic imprinting is responsible for monoallelic gene expression that depends on the sex of the parent from which the alleles (one active, one silent) were inherited. X-chromosome inactivation is also a form of monoallelic gene expression. One of the two X chromosomes is transcriptionally silenced in the somatic cells of females, effectively equalising gene dosage with males who have only one X chromosome that is not complemented by a gene poor Y chromosome. X chromosome inactivation is random in eutherian mammals, but imprinted in marsupials, and in the extraembryonic membranes of some placentals. Imprinting and X inactivation have been studied in great detail in placental mammals (particularly humans and mice), and appear to occur also in marsupial mammals. However, both phenomena appear to have evolved specifically in mammals, since there is no evidence of imprinting or X inactivation in non-mammalian vertebrates, which do not show parent of origin effects and possess different sex chromosomes and dosage compensation mechanisms to mammals.¶ In order to understand how imprinting and X inactivation evolved, I have focused on the mammals most distantly related to human and mouse. I compared the sequence, location and expression of genes from major imprinted domains, and genes that regulate genomic imprinting and X-chromosome inactivation in the three extant mammalian groups and other vertebrates. Specifically, I studied the evolution of an autosomal region that is imprinted in humans and mouse, the evolution of the X-linked region thought to control X inactivation, and the evolution of the genes thought to establish and control differential expression of various imprinted loci. This thesis is presented as a collection of research papers that examines each of these topics, and a review and discussion that synthesizes my findings.¶ The first paper reports a study of the imprinted locus responsible for the human Prader-Willi and Angelman syndromes (PWS and AS). A search for kangaroo and platypus orthologues of PWS-AS genes identified only the putative AS gene UBE3A, and showed it was in a completely different genomic context to that of humans and mice. The only PWS gene found in marsupials (SNRPN) was located in tandem with its ancient paralogue SNRPB, on a different chromosome to UBE3A. Monotremes apparently have no orthologue of SNRPN. The several intronless genes of the PWS-AS domain also have no orthologues in marsupials or monotremes or non-mammal vertebrates, but all have close paralogues scattered about the genome from which they evidently retrotransposed. UBE3A in marsupials and monotremes, and SNRPN in marsupials were found to be expressed from both alleles, so are not imprinted. Thus, the PWA-AS imprinted domain was assembled from many non-imprinted components relatively recently, demonstrating that the evolution of imprinting has been an ongoing process during mammalian radiation.¶ In the second paper, I examine the evolution of the X-inactivation centre, the key regulatory region responsible for X-chromosome inactivation in humans and mice, which is imprinted in mouse extraembryonic membranes. By sequencing and aligning flanking regions across the three mammal groups and non-mammal vertebrates, I discovered that the region homologous to the X-inactivation centre, though intact in birds and frogs, was disrupted independently in marsupial and monotreme mammals. I showed that the key regulatory RNA of this locus (X-inactive specific transcript or XIST) is absent, explaining why a decade-long search for marsupial XIST was unsuccessful. Thus, XIST is eutherian-specific and is therefore not a basic requirement for X-chromosome inactivation in all mammals.¶ The broader significance of the findings reported in these two papers is explored with respect to other current work regarding the evolution and construction of imprinted loci in mammals in the form of a review. This comparison enabled me to conclude that like the PWS-AS domain and the X-inactivation centre, many domains show unexpected construction from disparate genomic elements that correlate with their acquisition of imprinting.¶ The fourth and last paper examines the evolution of CCCTC-binding Factor (CTCF) and its parologue Brother Of Regulator of Imprinted Sites (BORIS) which contribute to the establishment and interpretation of genomic imprinting at the Insulin-Like Growth Factor 2/H19 locus. In this paper I show that the duplication of CTCF giving rise to BORIS occurred much earlier than previously recognised, and demonstrate that a major change in BORIS expression (restriction to the germline) occurred in concert with the evolution of genomic imprinting. The papers that form the bulk of this thesis show that the evolution of epigenetic traits such as genomic imprinting and X-chromosome inactivation is labile and has apparently responded rapidly to different selective pressures during the independent evolution of the three mammal groups. I have introduced these papers, and discussed them generally in terms of current theories of how and why these forms of monoallelic expression have evolved in mammals.

Non-coding antisense RNAs have been implicated in the epigenetic silencing of individual gene as well as chromosomal domains. While silencing of the overlapping gene by antisense RNAs has been well investigated, their functional role in silencing of chromosomal domains remains enigmatic. To elucidate mechanisms underlying the non-coding RNA mediated epigenetic silencing of chromosomal domains, we have chosen an antisense non-coding RNA, Kcnq1ot1, as a model system. Previously, a functional role of Kcnq1ot1 RNA and/or its transcriptional process has been implicated in silencing of multiple genes in the Kcnq1 imprinted cluster. However, these studies could not rule out the mechanisms involving other than Kcnq1ot1 RNA. Furthermore, it was also unclear how the Kcnq1ot1 promoter escapes silencing when its encoded RNA is capable of silencing flanking genes in cis. We have shown that NF-Y transcription factor plays a central role in the Kcnq1ot1 promoter activity, and that mutation of the NF-Y binding sites not only resulted in loss of silencing of flanking genes but also the ability of the Kcnq1ot1 promoter to protect against repressive chromatin marks, indicating that NF-Y maintains transcription-competent chromatin at the promoter through resisting the strong silencing effects of Kcnq1ot1 RNA. The Kcnq1ot1 RNA is an RNA Polymerase II encoded 91 kb long moderately stable nuclear transcript. We have demonstrated that it is the RNA not the act of transcription responsible for silencing and that the degree of silencing was proportional to the length of Kcnq1ot1 RNA. The kinetics of heterochromatin formation in relation to Kcnq1ot1 transcription revealed that overlapping gene was silenced initially by occlusion of basal transcription machinery and heterochromatin formation, whereas nonoverlapping gene was silenced subsequently by Kcnq1ot1-mediated heterochromatin spreading. This transcriptional silencing by Kcnq1ot1 RNA is mediated by an 890 bp region through promoting its interaction with the chromatin. Interestingly, we show that Kcnq1ot1 RNA establishes heterochromatin structures in a lineage-specific fashion by interacting with chromatin and chromatin remodelling complexes such as G9a and PRC2 complexes. More importantly, one of the parental chromosomes comprising Kcnq1 domain always found in the vicinity of perinucleolar region. Based on these data we proposed a mechanism whereby Kcnq1ot1 RNA establishes transcriptional silencing through recruitment of chromatin remodelling machinery and the maintenance of silencing achieved via targeting to the perinucleolar region.

In mammals, the most frequently observed hybrid dysgenesis effects are growth disturbances and male sterility. Profound defects in placental development have been described and our work on hybrids in genus Mus has demonstrated putative hybrid dysgenesis effects that lead to defects in lipid homeostasis and maternal behavior. Interestingly, mammalian interspecies hybrids exhibit strong parent-of-origin effects in that offspring of reciprocal matings, even though genetically identical, frequently exhibit reciprocal phenotypes. Recent studies have provided strong link between epigenetic regulation and growth, behavior and placental development. Widespread disruption of genomic imprinting has been described in hybrids between closely related species of the genus Peromyscus. The studies presented in this thesis aim to investigate the effects of disrupted epigenetics states on altered growth, female infanticide and placental dysplasia observed in Mus hybrids. We showed that loss-of-imprinting (LOI) of a paternally expressed gene, Peg1, was correlated with increased body weight of F1 hybrids. Furthermore, we investigated whether LOI of Peg1 in F1 females would interfere with maternal behavior. A subset of F1 females indeed exhibited highly abnormal maternal behavior in that they rapidly attacked and killed the pups. By microarray hybridization, a large number of differentially expressed genes in the infanticidal females as compared to normally behaving females were identified. In addtion to Peg1 LOI, we studied allelic expression of numerous imprinted genes in adult Mus interspecies hybrids. In contrast to the study from Peromyscus, patterns of LOI were not consistent with a direct influence of altered expression levels of imprinted genes on growth. Finally, we investigated the allelic interaction between an X-linked locus and a paternally expressed gene, Peg3, in placental defects in Mus hybrids. This study further strengthened the notion that divergent genetic and epigenetic mechanisms may be involved in hybrid dysgenesis in diverse groups of mammals.

Genomic imprinting is an epigenetic phenomenon by which a subset of genes is expressed in a parent of origin specific manner. Most of the imprinted genes are located in clusters. Genetic evidences suggest that genes in imprinted clusters are regulated by Imprinting Control Regions (ICRs). To elucidate the mechanisms by which the imprinting is maintained in clusters, we have chosen a well characterized cluster at the distal end of mouse chromosome 7. This cluster contains 15 imprinted genes and they have been shown to be regulated by H19 and Kcnq1 ICRs. The mouse H19 ICR, which is shown to have a chromatin insulator function, is implicated in the regulation of H19 and Igf2 genes by interacting with the CTCF protein. It has been documented that CTCF is also involved in the maintenance of differential methylation at the ICR. In this investigation we demonstrated that CTCF maintained differential methylation is lost when we subjected the ICR containing episomal plasmids to de novo methylation machinery of the human choriocarcinoma cell line, JEG3, suggesting that the H19 ICR looses its methylation privilege property under neoplastic conditions. The Kcnq1 ICR has been implicated in the regulation of 11 imprinted genes. The Kcnq1 ICR is methylated on the active maternal allele but unmethylated on the inactive paternal allele and overlaps an oppositely oriented and paternally expressed gene known as Kcnq1ot1. In this investigation, we documented that the Kcnq1 ICR controls the imprinting of neighboring genes by behaving as a bidirectional silencer and that this function is regulated by antisense RNA Kcnq1ot1. Furthermore, we have documented that duration of antisense transcription plays a critical role in the antisense RNA- mediated silencing. In conclusion, this thesis provides more insights into the complex mechanistic aspects by which ICRs, control imprinting of genes in clusters during development and neoplasia.

Genomic imprinting is an epigenetic phenomenon that causes a subset of mammalian genes to be expressed from only one allele in a parent-of-origin manner. The defects in the imprinting regulation result in disorders that affect development, growth and metabolism. We have used the Kcnq1 imprinted cluster as a model to understand the mechanism of imprinted gene regulation. The imprinting at the Kcnq1 locus is regulated by a long noncoding RNA, Kcnq1ot1, whose transcription on the paternal chromosome is associated with the silencing of at least eight neighboring genes. By destabilizing Kcnq1ot1 in an episomal system, we have conclusively shown that it is the RNA and not the process of transcription that is required for the gene silencing in cis. Kcnq1ot1 RNA interacts with the chromatin modifying enzymes such as G9a and Ezh2 and recruits them to imprinted genes to establish repressive chromatin compartment and gene silencing. Using the episomal system, we have identified an 890 bp silencing domain (SD) at the 5’ end of Kcnq1ot1 RNA, which is required for silencing of neighboring reporter genes. The deletion of the SD in the mouse resulted in the relaxation of imprinting of ubiquitously imprinted genes (Cdkn1c, Kcnq1, Slc22a18, and Phlda2) as well as reduced DNA methylation over the somatic DMRs associated with the ubiquitously imprinted genes. Moreover, Kcnq1ot1 RNA interacts with Dnmt1 and recruits to the somatic DMRs and this recruitment was significantly affected in the SD mutant mice. By using a transgenic mouse, we have conditionally deleted Kcnq1ot1 promoter at different developmental stages and demonstrated that Kcnq1ot1 maintains imprinting of the ubiquitously imprinted genes by regulating DNA methylation over the somatic DMRs. Kcnq1ot1 is dispensable for the maintenance of repressive histone marks and the imprinting of placental-specific imprinted genes (Tssc4 and Osbpl5). In conclusion, we have described the mechanisms by which Kcnq1ot1 RNA establishes and maintains expression of multiple imprinted genes in cis.

Genomic imprinting refers to the parent-of-origin specific monoallelic expression of a gene. Imprinted genes are often clustered in the genome and their expression is regulated by an imprinting centre (IC). ICs are regions of DNA that propagate the parental specific regulation of gene expression, which are usually characterized by differential DNA methylation, histone marks and the presence of non-coding RNAs. Beckwith-Wiedemann syndrome (BWS) is an overgrowth syndrome associated with the dysregulation of imprinted gene expression on human chromosome band 11p15.5. The 11p15.5 imprinted region has two imprinting centres, IC1 and IC2. IC1 is telomeric and regulates the imprinted expression of the genes H19 and IGF2. IC2 is ~700kb centromeric and is associated with a cluster of nine imprinted genes including CDKN1C, KCNQ1 and an imprinted non-coding RNA associated with IC2, KCNQ1OT1. Loss of differential DNA methylation at IC2 is seen in 50% of patients with BWS with loss of imprint of the non-coding RNA KCNQ1OT1 and associated with a decreased expression of the putative tumour suppressor CDKN1C. Patients with BWS also have a thousand-fold increased risk of pediatric cancer. The focus of this thesis involves investigation of dysregulation of imprinting in three groups of BWS patients. Firstly, I show that BWS patients with alveolar rhabdomyosarcoma have constitutional loss of methylation at IC2 and biallelic expression of KCNQ1OT1. Secondly, loss of methylation at IC2 has been previously associated with female monozygotic twins discordant for BWS. In male monozygotic twins with BWS, however, the molecular lesions reflect the molecular heterogeneity seen in BWS singletons. Thirdly, BWS patients associated with translocations and inversions that have breakpoints within the KCNQ1 gene near IC2 show regional gain of DNA methylation around the breakpoint and decreased expression of CDKN1C. Therefore, using a rare collection of BWS patients, I have attempted to determine the various roles of the imprinting centres IC1 and IC2 and their involvement in tumourigenesis, monozygotic twinning and structural chromosomal rearrangements causing BWS.

Genomic imprinting refers to the parent-of-origin specific monoallelic expression of a gene. Imprinted genes are often clustered in the genome and their expression is regulated by an imprinting centre (IC). ICs are regions of DNA that propagate the parental specific regulation of gene expression, which are usually characterized by differential DNA methylation, histone marks and the presence of non-coding RNAs. Beckwith-Wiedemann syndrome (BWS) is an overgrowth syndrome associated with the dysregulation of imprinted gene expression on human chromosome band 11p15.5. The 11p15.5 imprinted region has two imprinting centres, IC1 and IC2. IC1 is telomeric and regulates the imprinted expression of the genes H19 and IGF2. IC2 is ~700kb centromeric and is associated with a cluster of nine imprinted genes including CDKN1C, KCNQ1 and an imprinted non-coding RNA associated with IC2, KCNQ1OT1. Loss of differential DNA methylation at IC2 is seen in 50% of patients with BWS with loss of imprint of the non-coding RNA KCNQ1OT1 and associated with a decreased expression of the putative tumour suppressor CDKN1C. Patients with BWS also have a thousand-fold increased risk of pediatric cancer. The focus of this thesis involves investigation of dysregulation of imprinting in three groups of BWS patients. Firstly, I show that BWS patients with alveolar rhabdomyosarcoma have constitutional loss of methylation at IC2 and biallelic expression of KCNQ1OT1. Secondly, loss of methylation at IC2 has been previously associated with female monozygotic twins discordant for BWS. In male monozygotic twins with BWS, however, the molecular lesions reflect the molecular heterogeneity seen in BWS singletons. Thirdly, BWS patients associated with translocations and inversions that have breakpoints within the KCNQ1 gene near IC2 show regional gain of DNA methylation around the breakpoint and decreased expression of CDKN1C. Therefore, using a rare collection of BWS patients, I have attempted to determine the various roles of the imprinting centres IC1 and IC2 and their involvement in tumourigenesis, monozygotic twinning and structural chromosomal rearrangements causing BWS.

Epigenetic differences between maternally inherited and paternally inherited chromosomes, such as CpG methylation, render the maternal and paternal genome functionally inequivalent, a phenomenon called genomic imprinting. This functional inequivalence is exemplified with imprinted genes, whose expression is parent-of-origin specific. The dosage of imprinted gene expression is disrupted in cells with uniparental disomy (UPD), which is an unequal parental contribution to the genome. I have derived mouse embryonic stem (ES) cell sub-lines with maternal UPD (mUPD) for mouse chromosome 6 (MMU6) to characterize regulation and maintenance of imprinted gene expression. The main finding from this study is that maintenance of imprinting in mitotic UPD is extremely variable. Imprint maintenance was shown to vary from gene to gene, and to vary between ES cell lines depending on the mechanism of loss of heterozygosity (LOH) in that cell line. Certain genes analyzed, such as Peg10, Sgce, Peg1, and Mit1 showed abnormal expression in ES cell lines for which they were mUPD. These abnormal expression levels are similar to that observed in ES cells with meiotically-derived full genome mUPD (parthenogenetic ES cells). Imprinted CpG methylation at the Peg1 promoter was found to be abnormal in all sub-lines with mUPD for Peg1. Two cell sub-lines which incurred LOH through mitotic recombination showed hypermethylation of Peg1, consistent with the presence of two maternal alleles. Surprisingly, a cell sub-line which incurred LOH through full chromosome duplication/loss showed hypomethylation of Peg1. The levels of methylation observed in these sub-lines correlates with expression, as the first two sub-lines showed a near-consistent reduction of Peg1, while the latter showed Peg1 levels close to wild-type. Altogether these results suggest that certain imprinted genes, like Peg1 and Peg10, have stricter imprinting maintenance, and as a result show abnormal expression in UPD. This strict imprint maintenance is disrupted, however, in UPD incurred through full chromosome duplication/loss, possibly because of the trisomic intermediate stage which occurs in this mechanism.