Human Reproduction Update, Vol.10, No.1 pp.3-18, 2004
© European Society of Human Reproduction and Embryology 2004; all rights reserved
Potential significance of genomic imprinting defects for reproduction and assisted reproductive technology
1 McGill University–Montreal Children’s Hospital Research Institute and Departments of Paediatrics, Human Genetics, and Pharmacology & Therapeutics, McGill University, Montreal, Quebec, H3H 1P3, Canada and 2 University of Pittsburgh, Department of Molecular Genetics and Biochemistry, Pittsburgh, PA 15213, USA 3 To whom correspondence should be addressed at: McGill University-Montreal Children’s Hospital Research Institute, 2300 Tupper Street, Montreal, Quebec. Canada H3H 1P3. e-mail: jacquetta.trasler{at}mcgill.ca
 |
Abstract |
|---|
 TOP Abstract IntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
Recent studies suggest a possible link between human assisted reproductive technology and genomic imprinting disorders. Assisted reproductive technology includes the isolation, handling and culture of gametes and early embryos at times when imprinted genes are likely to be particularly vulnerable to external influences. Evidence of sex-specific differences in imprint acquisition suggests that male and female germ cells may be susceptible to perturbations in imprinted genes at specific prenatal and postnatal stages. Imprints acquired first during gametogenesis must be maintained during preimplantation development when reprogramming of the overall genome occurs. In this review, we will discuss both new developments in our understanding of genomic imprinting including the mechanisms and timing of imprint erasure, acquisition and maintenance during germ cell development and early embryogenesis as well as the implications of this research for future epigenetic studies in reproduction and assisted reproductive technology.
Key words: Angelman syndrome/assisted reproductive technology/Beckwith–Wiedemann syndrome/genomic imprinting/reproduction
|
Introduction |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
The functional asymmetry of mammalian parental genomes was discovered in the mid-1980s through elegant nuclear transplantation experiments, which demonstrated the non-viability of uniparental embryo development (McGrath and Solter, 1984
; Barton et al., 1984; Surani et al., 1984). A subset of our genes, 
60 to date, is known to be subject to genomic imprinting, the allele-specific expression of a gene where the allele that is expressed depends on whether it is maternal or paternal in origin. The monoallelic expression of imprinted genes results from the two parental alleles maintaining different epigenetic profiles. ‘Epigenetics’ refers to a process that regulates gene activity without affecting the genetic (DNA) code and is heritable through cell division. Germ cell development and early embryogenesis are crucial windows in the erasure, acquisition and maintenance of genomic imprints. Moreover, a number of genes regulated by imprinting have been shown to be essential to fetal growth and placental function. Increasing attention has recently focused on potential epigenetic disturbances resulting from embryo culture, somatic cell nuclear cloning and assisted reproductive technology (De Rycke et al., 2002; Gosden et al., 2003), indicating that a better understanding of genomic imprinting or parent-of-origin effects on gene expression is highly significant to the current study of reproduction and development.
In particular, publications over the last year have seeded concern about the possibility of an increased incidence of rare genomic imprinting diseases in children born of assisted reproductive technology. Cases of Angelman syndrome (AS) (Cox et al., 2002; Orstavik et al., 2003) and Beckwith–Wiedemann syndrome (BWS) (De Baun et al., 2003; Maher et al., 2003; Gicquel et al., 2003) with methylation imprint defects have been documented in children conceived via IVF and/or ICSI. It is unclear at present, but worth considering in the light of the BWS and AS cases, whether less rare imprinting abnormalities, such as those associated with placental dysfunction or postnatal cancer, are associated with assisted reproductive technology.
This overview will discuss the relevance of genomic imprinting to reproduction and highlight how aberrant imprinting can have deleterious consequences for the developing embryo. Specific points that will be addressed include the importance of proper establishment and maintenance of genomic imprints to the development of the embryo as well as to the function of extraembryonic tissues. We will also evaluate the existing evidence suggesting that imprinted genes may be particularly vulnerable to gamete and embryo manipulations in assisted reproductive technology and suggest directions for future research in this area.
|
Genomic imprinting |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
The functional and sex-specific non-equivalence of imprinted alleles explains the developmental failure of uniparental embryos and confirms the requirement of both parental genomes for normal development (Barton et al., 1984
; McGrath and Solter, 1984; Surani et al., 1984). Mouse androgenotes created from the activation of a zygote with two male pronuclei devoid of any maternal genome exhibit proliferation of extraembryonic tissues and poor embryonic development. In contrast, parthenogenetically activated oocytes give rise to relatively normal embryos that survive to an early somite stage with incomplete formation of extraembryonic structures. These observations suggest that genes expressed by the paternal genome are directed towards the development of extraembryonic tissues essential to support the growth of the embryo, while the maternal genome appears to be geared towards expressing genes that contribute to proper embryo development. The opposing tendencies of the male and female genomes as well as the elucidation from mouse studies that Igf2 and Igf2r are imprinted genes with conflicting functions led to the development of the most widely recognized theory of imprinting, the ‘parental conflict’ hypothesis (Haig and Graham, 1991; Moore and Haig, 1991). This theory proposes that the paternal genome has evolved to express genes that favour the extensive use of maternal resources and lead to optimal fetal development and growth, thus ensuring transmission of the father’s genes to the next generation. On the other hand, genes expressed by the maternal genome serve to counteract the effort made by paternally expressed genes, and limit investments in embryo development and growth in favour of salvaging resources for future pregnancies. The ‘conflict theory’ is relevant, not only to mothers and their embryos/fetuses during growth, but has been extended to postnatal effects as well, including effects on maternal behaviour (e.g. Peg3, Peg1). Mothers homozygous for targeted mutations in the paternally expressed Peg3 and Peg1 genes have defects in nurturing behaviour (Lefebvre et al., 1998; Li et al., 1999). Alternate and additional theories have also been proposed (e.g. Varmuza and Mann, 1994; McGowan and Martin, 1997; Pardo-Manuel de Villena et al., 2000; Beaudet and Jiang, 2002; Bestor, 2003; Kaneko-Ishino et al., 2003; reviewed in Wilkins and Haig, 2003).
To date, genomic imprinting is thought to be restricted to eutherian mammals although studies have shown that imprinting is conserved in some marsupials (the South American opossum, the tammar wallaby) (Toder et al., 1996; Killian et al., 2000; O’Neill et al., 2000; Ding et al., 2003). In flowering plants (angiosperms), parental imprinting governs development of the endosperm (non-embryonic, placenta-like tissue); increasing dosage of paternally derived genomes promotes growth of the endosperm, while increasing maternal dosage has the opposite effect (Spielman et al., 2001). Amphibians, insects and unicellular fungi show no evidence of imprinted genes in their genomes. However, similar epigenetically regulated phenomena of restricted gene expression have been described in species within these groups. For example, DNA methylation is used by Neurospora crassa to silence transposons and for X chromosome inactivation, and chromatin remodelling proteins are found in both Drosophila melanogaster and C. elegans. These observations suggest that gene control mechanisms used in lower organisms gained additional functions in higher organisms, including the control of genomic imprinting.
Imprinting is an epigenetically controlled phenomenon because something other than DNA sequence must distinguish the parental alleles and determine sex-specific gene expression. The role of DNA methylation in genomic imprinting has been extensively investigated. Although evidence suggests that it is not the only epigenetic mark used to distinguish the parental alleles of imprinted genes (Davis et al., 2000), numerous studies have confirmed its crucial role in the process. Mice completely deficient in the DNA methyltransferase (DNMT), Dnmt1, and subsequently in DNA methylation exhibit biallelic expression of several imprinted genes (Li et al., 1993). Mice lacking two additional DNMT, Dnmt1o and Dnmt3L, also display aberrant expression of a subset of imprinted genes (Bourc’his et al., 2001a; Howell et al., 2001; Hata et al., 2002). Secondly, many characterized imprinted genes have been shown to have sequences that are differentially methylated in the gametes (Tycko and Morison, 2002; Spahn and Barlow, 2003). In general, the two parental alleles have different levels of DNA methylation, and in many cases the methylation is concentrated in a single area, called a differentially methylated domain (DMD), within or near the imprinted gene. DNA methylation is a heritable yet reversible epigenetic mark that can be stably propagated after DNA replication and influence gene expression, properties that make it a particularly attractive imprinting mechanism.
In addition to allele-specific methylation, imprinting is associated with histone modifications, antisense transcripts and non-coding RNA including microRNA (Spahn and Barlow, 2003). For example, several imprinted loci, including Igf2r (Hu et al., 2000), the Igf2-H19 domain (Grandjean et al., 2001), Snrpn (Gregory et al., 2001) and U2af1-rs1 (Gregory et al., 2001) display acetylation of particular histone amino acid residues exclusively on the expressed allele. These observations, as well as the fact that DNA methyltransferase complexes associate with histone deacetylases (Fuks et al., 2000), suggest co-operation between DNA methylation, histone modifications, and overall chromatin state in the regulation of imprinted gene allele-specific expression.
As more detailed information about imprinted genes was unveiled, it became obvious that this subset of genes shares common characteristic features such as genomic clustering, regulation by antisense transcripts, and the presence of repeat elements near or within their DMD. Two imprinting clusters have been extensively described: a cluster on human chromosome 11p15 is linked to the pathogenesis of BWS, and a cluster on 15q11–13 is linked to the AS/Prader–Willi syndromes (AS/PWS) (Figure 1A and B). The mouse loci for these regions are located on chromosome 7.
|
The clustering of several imprinted genes in distinct areas of the genome is postulated to allow coordinated regulation of genes in a given chromosomal region. Supporting this concept, imprinting centres (IC) or imprint control elements have been found in some clusters, and microdeletions within the IC can affect the imprinting status of genes kilobases away (Spahn and Barlow, 2003
). Genes within both the BWS and the AS/PWS clusters also have antisense transcripts which are often non-coding and may serve to regulate the imprinted expression of their sense genes (e.g. KCNQ1/KCNQ10T1, UBE3A/UBE3A-AS). It has also been shown that the short tandem repeat elements within the DMD of several imprinted genes are important for the maintenance of their imprinting status (Reinhart et al., 2002). This is the case for the Igf2r DMD2 and is also suspected in the other genes such as Snrpn because of the repeat-rich nature of their DMD. |
Imprint dynamics and timing during gametogenesis and early embryogenesis |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
Mouse studies
Experimental evidence in mice suggests that genomic imprints in the parental gametes are erased upon every reproductive cycle, re-established in the immature germ cells of the developing embryo according to their fate as either male or female gametes, and maintained through both the preimplantation period (when the rest of the genome is demethylated) as well as postimplantation development (Figure 2). As such, imprints are dynamically changing during both germ cell and embryo development. Gametogenesis and the zygotic stage of embryogenesis are the only times when the maternal and paternal genomes are separate. As imprinted alleles are differentially marked to allow for their sex-specific expression, this period during which they are uniquely separate must be when the marking event occurs. Prior to the establishment of these sex-specific marks in the germ line, imprints are erased. In the mouse, there is evidence that this erasure occurs around the time that primordial germ cells enter the gonad, at approximately days 10.5–12.5 of gestation (Szabo and Mann, 1995b; Kato et al., 1999; Hajkova et al., 2002; Lee et al., 2002; Szabo et al., 2002).
|
Subsequent to this erasure, the timing of acquisition of genomic imprints is significantly different between the two germ lines. Studies investigating the erasure and acquisition of genomic imprints have generally used evidence of DNA methylation status, as well as evidence of mono- or biallelic expression of the genes of interest, as an indication of imprint acquisition. Other approaches have been nuclear transplantation and cloning experiments using germ cell nuclei at different stages (Obata et al., 1998
; Bao et al., 2000; Lee et al., 2002; Obata and Kono, 2002; Yamazaki et al., 2003). In these nuclear transplantation experiments, evidence of normal imprint acquisition (in the nuclear donor cells from gametes at different stages of their development) is tested by assessing postimplantation embryos for monoallelic expression of imprinted genes.
The paternal methylation imprint on H19, a maternally expressed gene, has been extensively characterized (Bartolomei et al., 1991, 1993; Ferguson-Smith et al., 1993; Tremblay et al., 1995; Davis et al., 1999; Ueda et al., 2000). A DMD was shown to span a 2 kb region roughly 2–4 kb from the transcription start site of the gene. This region is methylated in mature sperm and unmethylated in oocytes (Olek and Walter, 1997; Lucifero et al., 2002). The methylation imprint at this locus was shown to be acquired in prospermatogonia and complete by the pachytene stage of meiosis (Davis et al., 1999, 2000; Ueda et al., 2000). Additional experiments on H19 showed that paternally inherited alleles in male germ cells may have some mechanism for ‘remembering’ their origin and thus acquire their methylation imprint before maternally inherited alleles (Davis et al., 1999, 2000). A paternally inherited methylation imprint has also been described for two additional imprinted genes in the mouse, Gtl2 (Takada et al., 2002) and Rasgrf1 (Yoon et al., 2002).
Further proof that paternal imprints are complete by the haploid phase of spermatogenesis comes from round spermatid microinjection experiments. Shamanski et al. (1999) examined the imprinting status of three maternally expressed and three paternally expressed genes in embryos generated by injecting Mus castaneus spermatids into Mus musculus oocytes. Expression of imprinted genes was similar in embryos derived from round spermatids as compared to control embryos (from normal matings or derived by mature sperm microinjection).
In the female germ line, a number of studies have suggested that imprint acquisition occurs in the postnatal growth phase while oocytes are arrested at the diplotene stage of prophase I (Chaillet et al., 1991; Ueda et al., 1992; Brandeis et al., 1993; Stoger et al., 1993; Kono et al., 1996; Obata et al., 1998, 2002; Bao et al., 2000; Lucifero et al., 2002; Obata and Kono, 2002). We have previously shown that the methylation imprint on the paternally expressed gene Snrpn is acquired during this stage and that the establishment of the maternal methylation imprint on Igf2r, Peg1 and Peg3 is complete in metaphase II oocytes (Lucifero et al., 2002). Nuclear transplantation experiments using postnatal oocytes at various stages of maturation point to this same window of oocyte development as the time when functional imprints are acquired (Bao et al., 2000; Obata and Kono, 2002). Interestingly, and of potential relevance for the practice of human in vitro oocyte culture, one of the imprinted genes examined by Obata and Kono (2002), Impact, was only imprinted late in follicular development in the antral follicle stage.
Imprints established in the gametes must be faithfully maintained during preimplantation development while the methylation status of non-imprinted genes undergoes dynamic changes (Figure 2). In the mouse, within 4 h of fertilization the male pronucleus undergoes rapid demethylation by a presumably active mechanism (Mayer et al., 2000; Oswald et al., 2000; Santos et al., 2002). The maternal genome remains protected from this process and is therefore likely to undergo passive demethylation (Monk et al., 1987; Rougier et al., 1998). The overall methylation status of non-imprinted genes reaches a minimum at the blastocyst stage of development after which de novo methylation begins. During this wave of genome-wide methylation loss in the preimplantation embryo, imprinted genes maintain the marks inherited from the gametes, which finally translate into monoallelic sex-specific gene expression (Szabo and Mann, 1995a). The dynamics of methylation imprints at H19 (Olek and Walter, 1997), Igf2r (Stoger et al., 1993), U2afbp-rs (Shibata et al., 1997) and Ndn (Hanel and Wevrick, 2001) have been described during various stages of preimplantation development.
Human studies
In humans, limited information is available on the methylation status of imprinted genes during gametogenesis and embryogenesis, although the data available suggest some conservation of the methylation imprint acquisition and maintenance dynamics described in the mouse. The paternally inherited methylation imprint on H19 has been shown to be conserved in human sperm (Kerjean et al., 2000; Hamatani et al., 2001). In fetal gonocytes H19 is unmethylated, while in adult spermatogonia and subsequent stages of male germ cell development the H19 methylation imprint is acquired and maintained (Kerjean et al., 2000). Jinno et al. (1996) have shown that the paternal imprint on H19 is conserved in the preimplantation embryo.
An initial study on the methylation status of the SNURF-SNRPN locus within 15q11–13 in a small number of human oocytes suggested that the methylation imprint in fully grown human oocytes is absent and that this imprint must be acquired during or following fertilization in humans (El Maarri et al., 2001). However, more in keeping with the mouse data, a recent study on a larger number of oocytes at different developmental stages indicated that SNRPN is methylated in human oocytes (Geuns et al., 2003). From the latter study, human methylation imprints on SNRPN were present by the germinal vesicle stage and maintained in metaphase I and metaphase II oocytes. Certainly, from other data, a stable distinguishing feature must be established before the 4-cell stage as monoallelic (parental allele-specific) expression of SNRPN is initiated in 4-cell human embryos (Huntriss et al., 1998).
|
Mechanisms of genomic imprinting |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
Although a number of different enzymes have been postulated to be involved in either erasing imprints or ‘marking’ imprinted genes for parental allele-specific expression including DNA demethylases, histone modification enzymes and DNA methyltransferases (DNMT), with the exception of the DNMT, little is known. DNMT are the family of de novo and maintenance enzymes responsible for the addition of a methyl group to the 5-position of cytosine within CpG dinucleotides. Investigations into the dynamics of DNMT expression during germ cell and preimplantation development as well as studies of mice and humans with mutations in these enzymes have provided insight into the mechanisms responsible for imprint acquisition and propagation. DNMT1 is the predominant mammalian DNA methylating enzyme; methylation of the genome is only
5% of normal levels in mice homozygous for a targeted deletion in Dnmt1 (Lei et al., 1996). As DNMT1 has a preference for hemimethylated DNA, its role is postulated to be most important for the maintenance and propagation of methylation patterns. More recently, DNMT3A, DNMT3B and DNMT3L have been characterized (Okano et al., 1998; Aapola et al., 2000). DNMT3A and DNMT3B are thought to perform an important function as de novo methyltransferases, whereas DNMT3L plays a role in DNA methylation but does not appear to have DNA methyltransferase activity (Okano et al., 1998). Erasure of imprints
Recent evidence from mouse studies indicates that erasure may take place over a very short time, in as little as 24 h, at about the time when the germ cells initially enter the gonad. This observation suggests an active erasure process, although the identity of the enzymes or the molecular complex that is responsible for this demethylation is unknown (Hajkova et al., 2002).
Acquisition of imprints
The DNMT involved in the acquisition of methylation imprints in the male germ line are currently unknown; however, DNMT3A and DNMT3L are postulated to be involved since male mice with targeted knockouts of the genes encoding these enzymes have abnormalities in spermatogenesis (Bourc’his et al., 2001a; Hata et al., 2002). The fact that methylation imprints are normal in mouse oocytes deficient in the only form of DNMT1, i.e. DNMT1o, that is present in growing oocytes, indicates that DNMT1 is not required for the acquisition of imprints in female germ cells (Howell et al., 2001). In contrast, mouse knockout studies of Dnmt3L suggest a critical role for this protein in the acquisition of maternal methylation imprints in the oocyte (Bourc’his et al., 2001a; Hata et al., 2002). Interestingly, DNMT3L does not share any of the conserved DNMT catalytic motifs responsible for enzymatic activity. The involvement of DNMT3A and DNMT3B in the acquisition of imprints is suggested by experiments done by Hata et al. (2002). The authors transplanted the ovaries from Dnmt3a–/–, Dnmt3b+/– mice (such mice do not normally survive to adulthood) into normal recipient mice, derived embryos from the oocytes from these ovaries and showed that Igf2r, Peg1, Peg3 and Snrpn have completely unmethylated DMD in the embryos (Hata et al., 2002). Thus, DNMT3L is postulated to be a regulator of maternal imprint establishment, which may interact with known (e.g. DNMT3A or DNMT3B) (Hata et al., 2002) or yet-to-be-discovered DNA methyltransferases.
Maintenance of imprints
DNMT1o is produced in mouse oocytes during their postnatal growth phase and is present throughout preimplantation development when genome epigenetic reprogramming occurs. Gene-targeting experiments have shown that DNMT1o is essential for the maintenance of methylation on imprinted genes at the 8-cell stage, the only time-point when this normally cytoplasmic isoform translocates to the nucleus; all embryos developing in the absence of DNMT1o during the preimplantation period implant successfully but then die, prior to birth (Dean and Ferguson-Smith, 2001; Howell et al., 2001). That this DNMT1 variant is present to faithfully propagate imprints at this single stage of preimplantation development is remarkable, and suggests that maintaining imprints is a crucial and highly regulated process. Although other DNMT are postulated to maintain gametic methylation imprints at stages of preimplantation development other than the 8-cell stage, the identity of such enzymes is currently unknown. Gene-targeting studies indicate that DNMT1 is required for the maintenance of DNA methylation patterns on imprinted and non-imprinted genes in the postimplantation period (Li et al., 1992, 1993); it is possible that other DNMT are also involved.
Errors in erasure, acquisition or maintenance of imprints
Genomic imprinting defects may occur at any of the described stages of imprint erasure, acquisition or maintenance. A complete defect of imprint erasure would result in 50% of the gametes maintaining an inappropriate imprint and carrying the opposite sex’s epigenotype at certain imprinted loci. Establishment defects could result in absence of an imprint at a specific locus and again lead to the gametes harbouring the opposite sex’s imprint epigenotype. Defects in imprint maintenance could occur at any stage of pre- or postimplantation embryo development. Maintenance defects in post-zygotic embryos could lead to mosaicism with a subset of cells affected by the maintenance defect and the remainder of cells unaffected (i.e. with normal imprints).
Defects at any of these stages may arise because of problems with the machinery (enzymes) responsible for erasing, setting down, or maintaining imprints. Alternatively, epigenetic insults may cause changes in the methylation status or chromatin conformation within imprinted genes, leading to abnormal (i.e. other than monoallelic) expression patterns.
Few studies have directly investigated imprinting defects in normal gametes and embryos. Croteau et al. (2001) have shown that spontaneous imprinting defects resulting in biallelic H19 and Igf2 expression occur in 1.6 and 0.5% of day 7.5 or 8.5 mouse post-implantation embryos respectively. The incidence of biallelic expression of either gene was dependent on the genetic background of the embryo, and thus presumably on trans acting factors. These observations suggest that imprinting defects may occur sporadically in normal embryos and that the processes of imprint erasure, establishment and maintenance are vulnerable to errors.
There appears to be little experimental evidence that defects in genomic imprinting can be repaired. Due to the parental allele-specific nature of imprints, it is difficult to envisage a mechanism that would allow damaged imprints to be repaired post-zygotically in the embryo.
|
Roles of imprinted genes in fetal development, placental function and human disease |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
Mouse gene targeting experiments and studies of mutations in imprinted genes associated with human imprinting disorders have revealed a number of important physiological functions for imprinted genes. Since these findings have been the subject of a number of recent reviews (e.g. Reik and Walter, 2001
a; Tycko and Morison, 2002), only a few examples of the functions of imprinted genes are illustrated here in order to underline the variety of phenotypes that would be expected following perturbations in imprinting in the germ line or embryo. Fetal development
It is estimated that the total number of imprinted genes in the mouse and human genomes may range between 100 and 200. Of those that have been identified to date, a significant number appear to have important roles in fetal development (Table I). IGF2 is among these genes. Knockout mice deficient in Igf2 were found to have an absence of phenotype when the targeted allele was inherited through the female germ line, yet when transmitted through the male germ line, the offspring were shown to be severely growth-retarded (De Chiara et al., 1991). It was subsequently concluded that Igf2 was both an imprinted gene and a potent fetal-specific growth factor whose signalling was required for normal embryo development. Other imprinted genes that have since been shown also to be important for fetal growth include H19, IGF2R, PEG3 as well as others listed in Table I. Their role in fetal development was revealed as a result of gene targeting strategies similar to that described for Igf2. These genes belong to various families of signalling molecules or transcription factors. In humans, IGF2R is not imprinted in most individuals, although cases of polymorphic imprinting have been reported (Tycko and Morison, 2002).
|
Placental function
Many of the imprinted genes that were first identified in the mouse were shown to be expressed in the placenta. It has been argued that imprinted genes play essential roles in controlling the placental supply of maternal nutrients to the fetus, by regulating the growth of the placenta and/or the activity of transplacental transport systems (Reik et al., 2003). ASCL2 is a maternally expressed placental-specific transcription factor and is expressed in the spongiotrophoblast and labyrinthine layers. Ascl2 knockout mice die within 19 days of embryonic development because of placental failure (Guillemot et al., 1994). Human studies have mapped ASCL2 to the same chromosomal imprinting cluster where IGF2 and H19 reside. In human extraembryonic tissue, ASCL2 expression is confined to extravillous trophoblast cells. Recent studies suggest that ASCL2 is either not imprinted or escapes imprinting in humans (Miyamoto et al., 2002).
Tssc3 is a maternally expressed gene whose expression is limited to the labyrinth. Targeted deletion of Tssc3 results in global hyperplasia of placental tissues with expansion of spongiotrophoblast (Frank et al., 2002). Two other imprinted genes that are implicated in placental function are Peg1 and Peg3. Peg1 has been shown to be expressed in murine extraembryonic mesoderm derivatives and in human trophoblast. Peg3 has recently been shown to be expressed in the extraembryonic lineages of the mouse and in human placental tissue (Hiby et al., 2001).
Igf2, in addition to being an important regulator of fetal growth, also has profound effects on placental development. Recently a targeted deletion of the labyrinthine trophoblast-specific isoform of Igf2, Igf2P0, resulted in impaired placental growth and reduced placental transfer functions with subsequent fetal growth retardation (Constancia et al., 2002). It was proposed from this study that imprinted genes expressed in extraembryonic tissues may serve to regulate the placental supply of maternal resources and nutrients to the developing fetus (Constancia et al., 2002). A number of additional imprinted genes are specific transporters or channels involved in nutrient exchange (Reik et al., 2003). For example, a knockout of the maternally expressed extraneuronal monoamine transporter, Slc22a3, affects placental transfer functions (Zwart et al., 2001).
The fact that many imprinted genes are essential for placental function in the mouse and that they have human counterparts, which also appear to be imprinted and expressed in similar extra-embryonic tissues, make these an important factor that may explain cases of placental dysfunction in humans. There is increasing evidence that imprinted genes play important roles in the placenta to control the balance between supply and demand for nutrients, suggesting that defects in imprinted genes expressed in the placenta may be associated with clinical syndromes such as intrauterine growth retardation (IUGR).
Human disease
Loss of function of a number of imprinted genes has been found to be linked to human genetic diseases, the progression of certain cancers, and has also been implicated in a number of neurological disorders. Two different syndromes that were initially described in the 1950s and 1960s, namely AS and PWS, were found over 25 years later both to be associated with a deletion at the same chromosomal location, 15q11-q13. This locus carries a cluster of imprinted genes, which includes SNRPN, UBE3A, ZNF127, IPW and NDN, and is syntenic to the group of imprinted genes found centrally on mouse chromosome 7 (Figure 1A).
Angelman syndrome is caused by a loss of function of the maternal allele or duplication of the paternal allele within a region that spans UBE3A. Ube3a/UBE3A shows tissue-specific imprinting, being biallelically expressed in most tissues with the paternal allele silenced selectively in hippocampal and cerebellar neurons (Tycko and Morison, 2002). AS is characterized by ataxia, hypotonia, severe mental and motor retardation, epilepsy and absence of speech. The methylation defect associated with AS involves loss of methylation within the SNRPN imprinting centre (IC) which subsequently assumes a paternal (undermethylated) profile. Only a small percentage (5%) of AS patients have disrupted SNRPN IC methylation.
Prader–Willi syndrome is associated with a loss of function of the paternal allele or maternal duplication at the SNRPN locus, which is harboured within the same 15q11-q13 region. Patients with PWS are generally obese, mentally retarded, of short stature, suffer from muscular hypotonia, hypogonadotropic hypogonadism and have characteristic reduced fetal activity in the womb. In a subset of PWS patients, a methylation defect within the SNRPN IC has been described. This defect results in a gain of methylation at the IC resulting in the paternal allele assuming a maternal (methylated) profile and loss of function of the paternally expressed genes within this region (ZNF127, NDN, MAGEL2, SNRPN and IPW).
Another disease that exhibits parent-of-origin effects in its inheritance is BWS. This disorder is linked to a loss of function of the maternal allele at 11p15 where the imprinting cluster that includes H19, IGF2, CDKN1C, KCNQ1 and KCNQ1OT1 resides (Figure 1B). BWS is an overgrowth disorder and its main features are exomphalos, macroglossia, visceromegaly, neonatal hypoglycaemia, umbilical and abdominal wall abnormalities, as well as characteristic indentations of the ear. Children with BWS are predisposed to developing embryonic and childhood cancers.
Two methylation defects have been described in BWS patients. A BWS imprinting centre 1 (BWSIC1) defect results in a gain of methylation within the H19 DMD on the maternal allele such that it assumes a paternal (methylated) profile. As a result the normally expressed H19 maternal allele is silenced and IGF2 is biallelicly expressed (Figure 1Ci). The second methylation defect occurs at BWSIC2 and involves loss of maternal methylation at KCNQ1OT1, the KCNQ1 antisense transcript. Expression of KCNQ1OT1 becomes biallelic while both CDKN1C and KCNQ1 are silenced (Figure 1Cii). Wilms’ tumours are not seen in patients with BWSIC2 effects. Five to 10% of sporadic BWS cases have a BWSIC1 defect, while 40% manifest a BWSIC2 defect (Maher and Reik, 2000).
Monozygotic twinning occurs more frequently among BWS patients than it does in the general population (Weksberg et al., 2002). Many of these twins are discordant for BWS, with the affected twin showing a loss of methylation and subsequent biallelic expression at KCNQ1OT1 (Weksberg et al., 2002). The authors propose that this discordance in KCNQ1OT1 imprinting may arise via two possible mechanisms; (i) during twinning an unequal splitting of the inner cell mass may result in mosaicism where only one twin carries the defect resulting in BWS; or (ii) an imprinting defect during preimplantation development could lead to a problem in the imprint maintenance at KCNQ1OT1, suggesting that this gene may be particularly susceptible to imprinting insults.
Silver–Russell syndrome is a disorder characterized by low birthweight, dwarfism and lateral asymmetry and has been linked to the loss of function of a gene(s) within a less well described imprinted cluster (Preece, 2002). About 7–10% of patients with Silver–Russell syndrome show maternal uniparental disomy for a region on chromosome 7, while patients with paternal uniparental disomy of the same region are unaffected; these findings implicate an imprinted gene(s) in the aetiology of the disease in a subset of patients. Although PEG1 was proposed to be a likely candidate, the gene affected in this disease remains unclear (Nishita et al., 1996; Riesewijk et al., 1998).
In addition to the diseases described above, loss of imprinting is also involved in the progression of certain cancers (Tycko and Morison, 2002). Tumours that show imprinting effects include Wilms’ tumour where, in a subset of cases, loss of imprinting occurs at chromosomal region 11p15, which is in close proximity to the region involved in the pathogenesis of BWS. Loss of function of the maternal allele leading to the suppression of H19 and biallelic expression of IGF2 appears to be involved. Evidence suggests that other cancers show loss of imprinting, including hepatoblastoma, neuroblastoma, sporadic osteosarcoma, rhabdomyosarcoma and choriocarcinoma. Some imprinted genes act as tumour suppressor genes, the best characterized being IGF2R and WT1. Compared to non-imprinted tumour suppressor genes where loss of function requires mutations on both alleles, loss of function of both IGF2R (IGF2R imprinting is a polymorphic trait in humans) and WT1 require inactivation of only one allele.
Much like the observation that a number of imprinted genes are localized in extraembryonic tissues, it has also been noted that many are also expressed in the central nervous system. Studies of Peg1, Peg3, Ube3a, Grf1 and Gabrb3 knockout mice, as well as mice carrying a uniparental disomy at chromosome 2, suggest a functional role of imprinted genes in cognition and behaviour. Mouse knockouts of some imprinted genes show significant neurological defects ranging from abnormal maternal behaviour (Peg3, Peg1) and impaired memory (Grf1, Gabrb3) to motor dysfunction with seizures (Ube3a). In addition, several of the human imprinting diseases described have a neurodevelopmental impairment component to them. Other neurological disorders also appear to be inherited in a parent-of-origin-dependent manner. Some examples include bipolar affective disorder, autism, epilepsy, schizophrenia, Tourette syndrome, Turner’s syndrome and late onset Alzheimer’s disease (reviewed in Isles and Wilkinson, 2000; Nicholls, 2000).
|
Imprinting defects in uniparental and molar pregnancies |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
Spontaneous uniparental development has been well documented in humans and occurs at a significant frequency (Mutter, 1997
). Ovarian teratomas are the product of gynogenetic development derived from the parthenogenetic activation of an unfertilized oocyte within the ovary. These benign tumours are characterized by the presence of several types of differentiated but disorganized tissues including teeth, hair and adipose tissue, with no evidence of extraembryonically derived tissues. Teratomas constitute roughly 20% of all ovarian tumours.
Consistent with observations made in mice, human androgenetic conceptuses exhibit hyperplasia of extraembryonic trophoblastic tissues with a lack of embryo development. These hydatidiform moles, as they are referred to, originate from the fertilization of an enucleated egg or develop after the loss of maternal chromosomes in the zygote. Hydatidiform moles account for 1 in 1000 pregnancies and lead to gestational loss (Grimes, 1984). The incidence of molar pregnancies varies in different regions, with Japan, Eastern Asia, Indonesia and Iran having a 5–15-fold higher rate of occurrence (Lindor et al., 1992). The initial development of hydatidiform moles also results in a subsequent increased risk for future incidence of trophoblastic tumours.
Hydatidiform moles belong to a group of gestational trophoblastic diseases which also include choriocarcinoma and placental site trophoblastic tumour (reviewed in Li et al., 2002). These molar pregnancies can be characterized as partial or complete moles depending on whether they are triploid (partial hydatidiform moles or PHM with two paternal contributions and one maternal contribution) or androgenetic (complete hydatidiform moles or CHM with only two paternal contributions) in addition to histopathological features. However, it is recognized that a number of reported CHM are biparental in origin (Helwani et al., 1999; Moglabey et al., 1999; Fisher et al., 2000; Judson et al., 2002). These biparental moles are proposed to result from normal fertilization (i.e. maternal and paternal contributions) where the maternal alleles carry paternal imprints (or the paternal epigenotype). Biparental CHM appear to be recurrent with some being familial in origin in contrast to PHM, which are exclusively sporadic. Paternal contribution to the development of these biparental CHM is unlikely as one patient was reported to have had three biparental CHM by two different partners (Fisher et al., 2000).
Judson et al. postulated that familial and repetitive biparental CHM may in fact result from the global disruption of maternal imprinting (likely disrupted during oocyte development when imprints are acquired in the female) thus explaining their uniparental phenotype. They examined the methylation status of a number of genes (KCNQ1OT1, SNRPN, PEG1, PEG3) which are normally maternally methylated and confirmed that in biparental moles, these genes assumed a paternal epigenotype (Judson et al., 2002). In contrast, H19, which is methylated in the paternal germ line, remained unaffected. These results suggest that biparental CHM are caused by a recessive maternal effect mutation which leads to the failure of establishment of maternal imprints and leads the maternal genome to assume a paternal epigenetic appearance. It was proposed by Judson et al. that a mutation in DNMT3L, the enzyme shown to be essential to maternal methylation imprint establishment, may explain this phenotype. Although they looked for a mutation in this enzyme, none was identified (Judson et al., 2002). In addition, the same group did not find evidence of mutations in DNMT1, DNMT1o, DNMT2, DNMT3A or DNMT3B, leading them to suggest that factors other than the known DNMT are involved in imprint establishment in the female germ line (Hayward et al., 2003). Another group, Moglabey et al. (1999), mapped a maternal locus responsible for biparental CHM to 19q13.4. In members of the same family, El Maarri et al. (2003) went on to show that NESP55, SNRPN and H19 were abnormally methylated on the maternal alleles in the biparental CHM and suggested that abnormal methylation in biparental CHM as a result of a defective 19q13.4 locus is due to an error in establishment of maternal methylation during oogenesis or its maintenance during preimplantation development.
A case of confined placental mosaicism for complete hydatidiform mole with a normal female infant in a singleton pregnancy was recently reported (Makrydimas et al., 2002). This child was a chromosomally and phenotypically normal female infant. The mechanism resulting in this confined placental mosaicism must have taken place post-zygotically and resulted in selective loss of the maternal genome in cells that would give rise to extraembryonic tissues. Alternatively, a global loss of maternal imprints could have affected a subset of cells (destined to contribute to extraembryonic tissue) in the early embryo.
BWS is sometimes misdiagnosed in utero as partial hydatidiform mole. The characteristics of a BWS placenta include gross umbilical edema, scattered and enlarged villi, placentomegaly, and placental chorangioma (Genest, 2001). Placental abnormalities in BWS patients may be linked to the disregulation of ASCL2 or IGF2, both located within the 11p15 imprinting cluster.
|
Evidence of imprinting defects associated with assisted reproductive technology procedures |
|---|
|
TOPAbstractIntroductionGenomic imprintingImprint dynamics and timing...Mechanisms of genomic imprintingRoles of imprinted genes...Imprinting defects in...Evidence of imprinting defects...Imprinting mechanisms and...Perspectives and outlookAcknowledgementsReferences |
|---|
Much debate has recently surrounded the issue of possible epigenetic alterations brought about by embryo culture and manipulation in assisted reproductive technology (De Rycke et al., 2002
; Thompson et al., 2002; Gosden et al., 2003). While it is generally accepted that somatic cell nuclear transfer technology bypasses essential reprogramming steps in gametogenesis and early embryogenesis and that an epigenetic component is disregulated (reviewed in Wilmut et al., 2002), controversy surrounds the idea that assisted reproductive technologies may be vulnerable to similar yet admittedly less pronounced effects. The expression of imprinted genes as well as overall methylation status has been shown to be abnormal in a subset of cloned animals (Bourc’his et al., 2001b; Dean et al., 2001; Humpherys et al., 2001; Kang et al., 2001; Ohgane et al., 2001; Humpherys et al., 2002; Kang et al., 2002; Mann et al., 2003). Animal studies
The observation that in vitro culture may affect embryo outcome was initially made in ruminants. A connection to imprinting was proposed, as some of the lambs and calves born after embryo culture exhibited overgrowth abnormalities, which are now collectively referred to as ‘large offspring syndrome’. This proposed link was confirmed when it was found that sheep with ‘large offspring syndrome’ showed both lack of expression and aberrant methylation of Igf2r (Young et al., 2001).
Studies using mouse models have also established a relationship between embryo culture and the disrupted expression and methylation of imprinted genes. Khosla et al. have shown that the addition of fetal calf serum to M16 medium can disrupt imprinted gene function. They allowed the preimplantation development of 1- and 2-cell mouse embryos to proceed in a chemically defined M16 culture medium in the presence or absence of 5 and 10% fetal calf serum. Cultured blastocysts were then transferred to recipients to determine effects on postimplantation development. Analysis of the fetuses was carried out at day 14 of gestation and it was observed that M16 + FCS fetuses were lower in weight than control fetuses and exhibited decreased H19 and Igf2 expression. The decreased expression of H19 was correlated with an increase in methylation of its DMD. Grb7 and Grb10, two additional imprinted genes, also showed altered expression. In contrast, Peg1 did not show an altered expression profile, suggesting that some genes may be more resistant to epigenetic insults (Khosla et al., 2001).
Another study investigated the effect of different culture media on the expression of H19 and Snrpn in blastocysts that had been allowed to develop in vitro from the 2-cell stage (Doherty et al., 2000). Whitten’s medium was compared to KSOM supplemented with amino acids. Results indicated that the culture of preimplantation embryos in Whitten’s medium leads to aberrant expression of the normally silent paternal allele of H19. This abnormal expression correlated with a lack of methylation on the paternal allele. In contrast, embryos cultured in KSOM with amino acids showed no sign of abnormal H19 expression or methylation. Much like Peg1 in the above study (Khosla et al., 2001), blastocyst Snrpn expression was shown to be unaffected by either culture medium.
Human studies
In light of the concern caused by these animal studies, investigations into the occurrence of genomic imprinting diseases among children conceived with assisted reproductive technology have been conducted. The first one published was carried out to determine whether ICSI manipulations induced abnormal methylation patterns at 15q11-13, the locus linked to the pathogenesis of both AS and PWS (Manning et al., 2000). No evidence of abnormal methylation was found in samples from the 92 children examined, none of whom showed any clinical symptoms of either syndrome.
More recently, five studies have presented cases of ICSI- and IVF-conceived children afflicted with imprinting diseases (reviewed in Gosden et al., 2003). Cox et al. (2002) describe two children and Orstavik et al. (2003) a third child conceived using ICSI who were born with AS. Analysis within the 15q11-13 locus revealed an abnormal methylation pattern within SNRPN in samples from all three children. In each case, a loss of methylation on the maternal allele was found. Parents of the children were shown to have normal methylation profiles, and it was also confirmed that the childrens’ phenotype was not caused by a chromosomal deletion at this locus. AS normally occurs in 1 in 15 000 children, and when looking at the basis for the disorder, abnormal imprinting accounts for <5% of cases, deletion being the main aberration observed in patients. The authors therefore suggest that a causal relationship between ICSI and increased incidence of imprinting disorders is likely (Cox et al., 2002; Orstavik et al., 2003).
Three studies point to an association between IVF and ICSI and the development of another imprinting disease, BWS. De Baun et al. (2003) initiated a BWS registry in the USA in an effort to follow-up the incidence of cancer in these high-risk patients. In analysing information that was collected from these BWS children, they observed that several among them had been conceived via assisted reproductive technology. They proceeded to conduct a prospective study of BWS assisted reproductive technology-born children and identified a total of seven IVF- and/or ICSI-conceived children afflicted with BWS (four from their initial registry and three from their prospective study). ICSI was implicated in the conception of five of these children (four with ejaculated sperm, one with testicular sperm), one child was conceived via IVF, and it was unknown whether another child was conceived via IVF and/or ICSI. To determine the genetic defect responsible for the disease they performed methylation analysis on samples from these patients. Hypomethylation of KCNQ1OT1, corresponding to a BWSIC2 defect, was observed in five of the six children for whom samples were available. The methylation profile of H19 was also found to be abnormal in one of these children (BWSIC1 defect). One BWS-afflicted child showed normal methylation patterns of both H19 and KCNQ1OT1 (De Baun et al., 2003). At about the same time as the results from the USA were published, concern for an association between BWS and assisted reproductive technologies was raised independently in separate studies in two other countries. In a British BWS registry, six of 149 cases were in children conceived with assisted reproductive technologies (Maher et al., 2003) and in a French registry the same numbers, six of 149, were reported (Gicquel et al., 2003); eight out of the 12 patients (two from the British study and all six from the French study) had epigenetic BWSIC2 defects. In a previous report, we evaluated whether there was a significant increase in the risk of BWS associated with assisted reproductive technologies (Gosden et al., 2003). Based on the studies by De Baun (2003), Maher (2003) and Gicquel (2003), if we assume that assisted reproductive technologies make up 1% of all births, there is a significant 4.2-fold increase in the risk of BWS for children born using assisted reproductive technologies; however, if the frequency of assisted reproductive technology births is >1.9%, then the increase in BWS is no longer significant. Together, the results suggest the need for further studies to monitor children conceived with assisted reproductive technologies for epigenetic disorders such as BWS.
In response to these observations, we were interested in examining whether previous studies looking into the outcome of children conceived via assisted reproductive technology noted any evidence of children born with imprinting diseases. We performed independent full-text literature or abstract searches of studies looking at the outcome of assisted reproductive technology manipulations using the words Beckwith, Angelman, Prader–Willi and Wilms’ tumour. Our findings, in addition to the studies described above, are summarized in Table II. We were surprised to find a number of studies where reports of major congenital malformations within the IVF or ICSI group included children afflicted with BWS (Sutcliffe et al., 1995a; Koudstaal et al., 2000; Olivennes et al., 2001; Bonduelle et al., 2002; Boerrigter et al., 2002). However, as we have argued previously, these studies were small and it is likely that internationally coordinated, much larger, studies in the range of 100 000 births, will be needed to detect a doubling in the risk of BWS associated with assisted reproductive technologies (Gosden et al., 2003). Nevertheless, the appearance of BWS among assisted reproductive technology-conceived children may be due to a unique vulnerability of the imprinting locus on chromosome 11p15. Observations made in the study describing discordant KCNQ1OT1 methylation and expression in monozygotic twins discordant for BWS may support this suggestion (Weksberg et al., 2002). Also, in the mouse studies looking at the outcome of imprinted gene expression and methylation in embryos after culture, H19 has been shown to be particularly vulnerable (Doherty et al., 2000). The fact that H19 resides in this cluster and is linked to the pathogenesis of BWS may also predispose assisted reproductive technology children to increased risk of BWS particularly among imprinting diseases.
|
In addition, we noted a case of Silver–Russell syndrome among the IVF children identified in the study of the MRC Working Party on Children Conceived by IVF in Great Britain from 1978 to 1987 (MRC Working Party, 1990
). Among the malformations listed by Hansen et al. (2002, Appendix 2), a case of PWS is included. However, because of confidentiality reasons cited by the authors, child