Epigenetics in Male Germ Cells

Abstract

We provide an overview of epigenetic mechanisms comprising DNA methylation, histone modification, and RNA-associated silencing of gene expression by noncoding RNAs and subsequently focus attention on germ cell—specific epigenetic processes including histone to protamine exchange in haploid spermatids, genomic imprinting, and reprogramming of epigenetic information after fertilization. Finally, aberrant epigenetic reprogramming and its association with testicular cancer and possible risks for assisted reproductive technology (ART) techniques are discussed. The total of the genetic information that is maintained in the nucleotide sequence of DNA in an organism is known as the genome. Although each cell of an organism contains exactly the same genome, only a specific subset of genes will be active at a certain time point in a particular cell. Whether a specific gene is active or silent is regulated by the epigenome, representing the total of the chromatin (DNA and histones) modifications that index the genetic information. The study of processes that are involved in establishing these chromatin modifications is called epigenetics. Epigenetic processes include 1) methylation of DNA at deoxycytosine residues within CpG dinucleotides located within gene promoters (CpG islands); 2) modification of core histones, such as acetylation, methylation, phosphorylation, and ubiquitinylation at specific amino acid residues within the amino terminus; and 3) RNA-associated silencing of gene expression by noncoding RNAs. All of these processes participate in the regulation of gene activity without affecting the genetic code. Although these processes are heritable, they are potentially reversible and therefore open up opportunities for epigenetic therapy (Figure 1). . Genetic regulation of gene activity occurs at the transcriptional level through binding of transcription factors to the promoter and at the translational level through binding of protein repressors to the 3′ untranslated region and the poly-A tail of the transcript. Alternative splicing of the transcript results in the generation of various isoforms with different functions. Mutations within the genetic code are heritable and nonreversible. Epigenetic regulation of gene activity involves DNA methylation, core histone modification, and RNA interference. Although epigenetic effects are heritable, they do not affect the genetic code and are potentially reversible. This article provides an overview of the above mentioned epigenetic processes and subsequently focuses attention on germ cell—specific epigenetic processes including 1) histone to protamine exchange in haploid spermatids that is known to be associated with core histone acetylation, 2) genomic imprinting causing specific genes to be expressed monoallelically in a parent of origin—dependent manner, and 3) epigenetic reprogramming after fertilization. Finally, aberrant epigenetic reprogramming and its association with testicular cancer and possible risks for ART techniques will be discussed. Although × chromosome inactivation represents one of the most dramatic examples of long-term gene silencing and monoallelic gene expression, it is beyond the scope of this review (see a recent review by Heard [2005]). Methylation of deoxycytosine residues in CpG dinucleotides requires the activity of DNA methyltransferases (DNMTs), namely the maintenance DNMT1 (Bestor et al, 1988) or de novo DNMTs 3a and 3b (Okano et al, 1998), which catalyze the transfer of a methyl group from the S-adenosyl-methionine to the 5′ position of cytosine in DNA. DNA methylation, in general, correlates with gene activity in that hypomethylation is associated with transcriptional activation, whereas hypermethylation is associated with transcriptional repression. While hypermethylation has been suggested to be responsible not for initiating but for maintaining gene silencing similar to the role of CpG island hypermethylation on the inactive × chromosome (Clark and Melki, 2002), its repressive effect is believed to result from selective recognition of 5-methyl CpG dinucleotides by 5-methyl- CpG—binding domain proteins that are known to be involved in reading of DNA methylation marks (Wade, 2001). Specific information on DNA methylation in the testis. Recently analysis of 2600 gene loci revealed that the methylation status of testicular DNA is highly distinct from somatic tissues, exhibiting 8-fold the number of hypomethylated regions located within nonrepetitive sequences (Oakes et al, 2007). Most studies on DNA methylation in male germ cells, however, focus on ejaculated spermatozoa (see “Epigenetic Risks in ART”), tumor cells (see “The Formation of Testicular Cancer”), and expression of DNMTs that are responsible for DNA methylation. Detailed expression patterns are available for DNMTs 1, 3a, 3b, and 3-like (3L) (La Salle et al, 2004; Galetzka et al, 2007). In mice, expression of DNMT1 is down-regulated between days 14.5 and 15.5 of gestation and up-regulated around birth, suggesting a role in the maintenance of methylation patterns in proliferating spermatogonia. By contrast, expression of DNMTs 3a and 3b peaks around week 21 of gestation, suggesting a role in de novo methylation in mitotically arrested spermatogonia. This is consistent with the observation that Dnmt1−/– mice exhibit abnormal expression of imprinted genes (Li et al, 1992), whereas inactivation of DNMTs 3a and 3b blocks de novo methylation in early embryos (Okano et al, 1999). In addition, it has been suggested that DNMT3a generates a mark that is important for entrance of male germ cells into meiosis (Yaman and Grandjean, 2006). In the adult testis, a 5.2-kb Dnmt1 mRNA and its corresponding protein were demonstrated in spermatogonia, early spermatocytes, and round spermatids, whereas a 6.2-kb Dnmt1 mRNA was present in pachytene spermatocytes (Benoit and Trasler, 1994; Numata et al, 1994). Subsequently it was demonstrated that the Dnmt1 gene is regulated through alternative splicing, resulting in the production of 2 protein products, a somatic isoform that is also present in spermatogonia, early spermatocytes, and round spermatids and an oocyte-specific isoform (Mertineit et al, 1998). The pachytene spermatocyte—specific transcript is expressed at high levels in the testis but is not translated. Recently this expression pattern was confirmed in man. Interestingly, in infertile men revealing round spermatid maturation arrest, only pachytene spermatocytes exhibited positive signals for the DNMT1 protein, while the formerly positive round spermatids remained immunonegative (Omisanjo et al, 2007). Loss of DNMT3L, known to be present in precursors of spermatogonial stem cells, results in meiotic failure in spermatocytes, although these cells do not express DNMT3L (Bourc'his and Bestor, 2004). DNMT3L therefore represents an absolute requirement for male fertility; as in Dnmt3L−/– mice, pachytene spermatocytes exhibit homologous chromosomes that fail to align and form synaptonemal complexes, resulting in apoptosis of these cells (Webster et al, 2005). DNMT3L is thought to represent a methyltransferase, as it lacks enzymatic activity but is required for de novo methylation of imprinted genes in female (Bourc'his et al, 2001) and male germ cells (Bourc'his and Bestor, 2004). It has been demonstrated that expression of DNMT3L is regulated by 3 sex-specific promoters (Shovlin et al, 2007). Prospermatogonia encode a full-length protein that is involved in de novo methylation. Late pachytene spermatocytes activate a promoter located in intron 9 of the Dnmt3L gene. In haploid male germ cells, 3 truncated and noncoding transcripts can be observed. In contrast, in the ovary, an oocyte-specific promoter produces only 1 full-length transcript that is absent in Dnmt3L−/– female mice. Practical approaches for analysis of DNA methylation. Practical approaches for analysis of DNA methylation can be divided roughly into 2 groups: those analyzing global methylation within the genome and those analyzing specific methylation within the promoter region of specific genes, namely the CpG islands, short regions of DNA (approximately 500 bp) with high CG content (>50%) that are present in 1% to 2% of gene promoters within the genome (Gardiner-Garden and Frommer, 1987). For global methylation analysis, the overall level of methyl cytosines within the genome is determined by applying either chromatographic methods or methyl-accepting capacity assays. For gene-specific methylation analysis, various methods are available. While former studies used methylation-sensitive restriction enzymes for DNA digestion followed by Southern detection or polymerase chain reaction (PCR) (Rein et al, 1998), at present the most common approach for analysis of DNA methylation is bisulfite genomic sequencing in combination with methylation-specific PCR. In brief, genomic DNA is subjected to sodium bisulfite treatment, which converts unmethylated cytosine residues into uracil residues but leaves methylated cytosines unaltered. Subsequent PCR amplification with specific primer pairs and sequencing provides the methylation status of every CpG dinucleotide in the amplified region (Figure 2). Semiquantitative data can be obtained with methylation-sensitive single-strand conformation analysis (SSCA), in which sodium bisulfite treatment of DNA is followed by PCR using primer pairs without a CpG site to avoid selective amplification of either methylated or unmethylated DNA and SSCA. Recently CpG island microarrays became available to identify unknown methylation hot spots or methylated CpG islands within the genome. In addition, there are numerous approaches to study the activity of DNMTs (Woo et al, 2005). For further information, refer to Fraga and Esteller (2002), Tollefsbol (2004), and Mager and Bartolomei (2005). Detailed protocols are available at http:www.epigenome-noe.net and http:www.protocol-online.org. . Schematic representation of epigenetic protocols for the study of DNA methylation via bisulfite genomic sequencing and histone modification via chromatin immunoprecipitation (ChIP) and ChIP on chip. Methylated cytosines are underlined. The N-terminal region of core histones represents a hot spot for posttranslational modifications, such as acetylation, methylation, phosphorylation, ubiquitinylation, sumoylation, and ribosylation (Bradbury, 1992). The chemical modifications of core histones are referred to as the histone code (Jenuwein and Allis, 2001), which considerably extends the information potential of the genetic code (Figure 3). Although methylation of lysine 4 of histone H3 (H3K4) can be found in transcriptionally active chromatin regions (Lachner and Jenuwein, 2002; Santos-Rosa et al, 2002), in general, histone methylation is primarily attributed to transcriptionally silenced regions, whereas histone acetylation correlates with transcriptional activity (Rice and Allis, 2001). A hallmark of silent DNA is methylation of H3K9, which is globally distributed throughout heterochromatin regions and silenced promoters and can also inhibit acetylation of the H3 tail at several lysines (Fischle et al, 2003). Histone modifications are recognized by bromodomain-containing proteins that specifically bind to acetylated lysines and chromodomains that contain methylated H3K9 (Bannister et al, 2001). . Sites of acetylation (A) and methylation (M) within the N-terminal region of core histones H2A, H2B, H3, and H4 (Spencer and Davie, 1999). There are 2 classes of enzymes that are involved in determining the state of histone acetylation: histone acetyl transferases (HATs) and histone deacetylases (HDACs) (Davie, 1998). Substrates for these enzymes include amino groups of lysine residues located in the amino terminal tails of core histones. Both HATs and HDACs can be regulated through their availability to interact with cofactors, activity, or concentration (Legube and Trouche, 2003; Sengupta and Seto, 2004). While HAT-mediated lysine acetylation activates binding, neighboring residues determine bromodomain specificity (Bottomley, 2004). Bromodomains adopt a left-handed, twisted 4-helical bundle with 2 loops at one end of the bundle forming a hydrophobic lysine-binding pocket that selects acetyl-lysine rather than the charged unmodified lysine. Compared with acetylation, signaling by methylation is more complex, as lysines can be mono-, di-, or trimethylated, arginines can be mono- or dimethylated, and both site specificity and the number of methyl groups modulate the epigenetic signal. Methylation of lysine residues is mediated by Su(var)3–9, enhancer-of-zeste, trithorax protein (SET) domains that can recruit other proteins through their chromodomains. Chromodomains themselves are linked to transcriptional repression through the formation of heterochromatin or heterochromatin-like complexes and might ultimately signal epigenetic gene expression (Bottomley, 2004). Histone demethylation is performed by either Jumonji C domain—containing histone demethylase that is specific for H3K36 (Klose et al, 2006; Tsukada et al, 2006) or lysine-specific demethylase that is specific for mono- and dimethylated H3K4 (Forneris et al, 2005, 2006). Processes of histone modification and DNA methylation often involve dynamic interactions that can either reinforce or inhibit epigenetic changes. It has been demonstrated that the N-terminal domain of DNMT1 is able to bind to HDACs and can suppress gene transcription through facilitation of histone deacetylase activity (Roundtree et al, 2000). Furthermore, there is evidence that cooperation between DNA methylation and histone modification is involved in the regulation of imprinted gene allele—specific expression: 1) several imprinting loci, such as the insulin-like growth factor 2 (IGF2)-H19 domain (Grandjean et al, 2001) and the SNRPN domain (Gregory et al, 2001), display acetylation of particular histone amino acid residues exclusively on the expressed allele and 2) DNMT complexes associate with HDACs (Fuks et al, 2000, 2005). Recently silencing of the melanoma antigen—encoding (MAGE) gene family was reported to require both promoter demethylation and histone deacetylation (Wischnewski et al, 2006). Both epigenetic silencers and activators for gene transcription are polycomb group genes representing part of multitask protein complexes including HDAC and histone methyltransferase activities (Valk-Lingbeek et al, 2004; Grimaud et al, 2006). Specific information on histone modifications in the testis. Hyperacetylation of histone H4 has been reported to be associated with histone to protamine exchange in haploid spermatids (Meistrich et al, 1992; Hazzouri et al, 2000; Sonnack et al, 2002). Recently it was demonstrated that the double bromodomain-containing protein BRDT, which has been localized in haploid spermatids, binds hyperacetylated histone H4 before accumulating in condensed chromatin (Govin et al, 2006). Therefore, it has been suggested that BRDT is involved in organizing the spermatozoon's genome by mediating a general histone acetylation—induced chromatin compaction and maintaining a differential histone acetylation of specific regions. Misregulation of histone H4 acetylation is associated with impaired spermatogenesis. In infertile men exhibiting Sertoli cell only (SCO) syndrome, Sertoli cell nuclei displayed an increase in histone H4 acetylation (Faure et al, 2003). In mice, treatment with the HDAC inhibitor trichostatin A (TSA) resulted in reversible infertility due to meiotic arrest at the level of pachytene spermatocytes (Fenic et al, 2004). Practical approaches for analysis of histone modifications. Chromatin immunoprecipitation (ChIP) has emerged as a powerful tool for the study of interactions between DNA and chromatin-associated proteins, such as transcription factors and chemically modified core histones, across a defined DNA domain (Das et al, 2004). Two approaches can be distinguished by their methods of chromatin preparation: nChIP using native chromatin prepared by micrococcal nuclease digestion of nuclei (Hebbes et al, 1988) and xChIP using cross-linked chromatin prepared by adding formaldehyde to cells and exposing cells to ultraviolet irradiation (Solomon et al, 1988). Although ChIP at present is most accurate when there is an abundance of material (>107 cells) and only 1 cell type in the sample, it facilitates a precise and complex analysis of particular loci or even of whole genomes when combined with microarray technology (ChIP on chip) (Buck and Lieb, 2004) (Figure 2). Recently a protocol was reported that allows the analysis of as few as 100 cells using carrier chromatin ChIP in combination with PCR (O'Neill et al, 2006; Turner et al, 2006). For further information, refer to Tollefsbol (2004) and Mager and Bartolomei (2005). Detailed protocols are available at http:www.epigenome-noe.net. A popular approach to intervene directly in the chromatin remodeling process and to monitor changes in gene activity is to inhibit HDACs by HDAC inhibitors such as TSA, leading to increasing acetylation of the gene promoter and changes in gene activity, if the gene is regulated by histone (de)acetylation (Liu et al, 2006). Recently Heltweg et al (2005) presented a nonradioactive HDAC assay in which substrates with known and unknown HDAC activity are treated with an HDAC inhibitor. The HDAC activity is calculated relative to the conversion value. Since analysis of the eukaryotic transcriptome revealed a large number of noncoding RNAs (ncRNAs) containing a high density of stop codons and lacking any extensive open reading frames, it has become clear that RNA plays a profound and complex role in the regulation of gene expression (Costa, 2005). ncRNAs are 20 to 22 nucleotides in length and include micro RNAs that are generated from hairpin-shaped precursors and small interfering RNAs (siRNAs), which are generated from double-stranded RNAs (dsRNAs) (Figure 4). Regulation of gene expression occurs at either the transcriptional level (eg, via control of chromatin organization) or the posttranscriptional level (eg, via control of mRNA stability and protein synthesis). It happens through RNA-associated silencing that requires the presence of Argonaute family proteins and is operable through RNA interference (RNAi) (Kawasaki et al, 2005). RNAi constitutes a powerful natural and valuable tool for the dynamic silencing of specific gene expression in basic research with therapeutic potential (Barik, 2005; Cejka et al, 2006). . Schematic representation of the RNA interference pathway. Double-stranded RNAs are processed into approximately 21- to 23-nucleotide small interfering RNAs (siRNAs) by an RNase III—like enzyme called dicer. However, siRNAs can also be synthesized outside the cell and then be introduced into the cell. Subsequently siRNAs assemble into endoribonuclease-containing complexes called RNA-induced silencing complexes (RISCs). Finally, siRNAs guide RISCs to complementary RNA molecules, where they cleave and destroy the cognate RNA. Specific information on RNA-associated silencing in the testis. Recently a novel class of small RNAs, 26 to 30 nucleotides in length, was demonstrated to be associated with MIWI, a murine male germline—specific PIWI subfamily member of the Argonaute protein family, and therefore named piwi-interacting RNAs (piRNAs) (Girard et al, 2006; Kim, 2006). While MIWI was reported to be present in spermatocytes and spermatids, representing a prerequisite for the initiation of spermiogenesis, piRNA was localized to spermatids and found associated with polysomes, suggesting a role in translational regulation (Grivna et al, 2006a,b). Practical approaches for analysis of RNA-associated silencing. During the last decades loss of function studies were traditionally performed on knockout mice to obtain mutant embryonic cells. While this remains a laborious method, RNAi has evolved to be a powerful novel technique for highly specific post-translational gene silencing in which siRNAs work as sequence-specific RNAi mediators. Gene function was eliminated by applying RNAi approaches in mouse embryonic stem cells to direct cell differentiation (Lykke-Andersen, 2006). In the RNAi pathway, silencing of gene expression occurs in response to the introduction of dsRNAs into the cell of interest (Figure 4). For further information, refer to Duxbury and Whang (2004), Sandy et al (2005), and Lykke-Andersen (2006). Detailed protocols are available at http:www.protocol-online.org. During spermatogenesis, restructuring of the nuclear chromatin is accompanied by a gradual replacement of part of the somatic histones by testis-specific subtypes, such as H1t, H1t2, HILS1, H2A.X, TH2B, H3.3A, and H3.3B (Kimmins and Sassone-Corsi, 2005). Subsequently both somatic and testis-specific histones are replaced by protamines. Protamine-DNA interactions are followed by chromatin condensation, resulting in a total stop of gene expression in haploid spermatids (Steger, 1999, 2001). Protamines also undergo several chemical alterations including phosphorylation, dephosphorylation (Marushige and Marushige, 1997), and disulfide bond formation (Calvin and Bedford, 1971) (Figure 5). In man, histone to protamine exchange is only about 85% complete (Tanphaichitr et al, 1978). This dual nucleohistone/nucleoprotamine structure may correspond to a specific differential arrangement of chromatin regions according to potential functions (Gardiner-Garden et al, 1998; Wykes and Krawetz, 2003). The presence of genomic islands that maintain a somatic-like chromatin structure despite the global transition into the nucleoprotamine structure may correspond to an essential mark for the establishment of adequate epigenetic information in the offspring. Gatewood et al (1987, 1990) demonstrated sequence-specific packaging of DNA in nucleohistone and nucleoprotamine components, suggesting that DNA linked to nucleohistone may represent sites of active transcription. There is now evidence to suggest that the paternal genome may be involved in starting early gene expression in the zygote. . Histone to protamine exchange during spermatogenesis. Spg indicates spermatogonia; ScyI, primary spermatocyte; ScyII, secondary spermatocyte; rSpd, round spermatid; eSpd, elongated spermatid; AC, acetylated; Deac, deacetylated; H, core histone; Ph, phosphorylated; P, protamine; and S=S, disulfide bond. Histone to protamine exchange is associated with core histone acetylation, as acetyl groups turn the basic state of histones into a neutral one which, as a consequence, decreases the affinity of histones for DNA and allows protamines to interact with DNA. In rat (Meistrich et al, 1992), mouse (Hazzouri et al, 2000), and man (Sonnack et al, 2002), hyperacetylated histones have been demonstrated in mitotically active spermatogonia and in elongating spermatids, where histone to protamine exchange takes place. Interestingly, impaired spermatogenesis has been reported to be associated with both precocious (in pachytene spermatocytes) and reduced (in round spermatids) hyperacetylation of histone H4, possibly followed by incorrect histone to protamine exchange resulting in round spermatid maturation arrest (Sonnack et al, 2002). Whether this represents a cause or a consequence of impaired spermatogenesis remains to be elucidated. Following fertilization, male and female pronuclei exhibit functional differences, although they reside within the same zygotic cytoplasm. In mice, the male pronucleus has been demonstrated to exhibit a volume 1.65-fold greater and transcriptional activity twofold higher than that of the female pronucleus. Both volume and transcriptional activity were approximately equal after the first mitosis of the embryo (Liu et al, 2005). In addition, sperm constitutive heterochromatin was found to be enriched for nucleosomes that carry specific histone modifications that are transmitted to the oocyte. This suggests an epigenetic mechanism for inheritance of chromosomal architecture (van der Heijden et al, 2006). Authors evaluated the dynamics of sperm chromatin remodeling in the zygote directly after gamete fusion and found histone H4 acetylated at lysines 8 and 12 prior to full decondensation of the sperm nucleus. This suggests that these marks are transmitted by the spermatozoon. Genomic imprinting describes an epigenetic process by which sex-specific marks, so-called imprints, on certain chromosomal regions, so-called imprinted genes, are established in a parent of origin—dependent manner and thus expressed monoallelically. This functional asymmetry of parental genomes was discovered through nuclear transplantation experiments demonstrating that parthenogenetic mouse embryos die by day 10 of gestation (Barton et al, 1984; McGrath and Solter, 1984). Data from Kono et al (2004) suggest that paternal imprinting is obligatory for descendants to prevent parthenogenesis. Researchers created the first viable parthenogenetic mouse using a reconstructed oocyte with 2 haploid sets of the maternal genome derived from a fully grown oocyte and a nongrowing oocyte. The mouse developed to adulthood and was able to produce offspring. This development was possible by appropriate expression of the IGF2 and H19 genes, as the genome from the nongrowing oocyte was obtained from mutant mice carrying a deletion within the H19 gene that results in the absence of maternal-specific methylation. Only a few imprinted genes have been found to acquire a methylation mark in the male germ line, and only 3 of these genes have been studied in detail, namely IGF2/H19 (Davis et al, 1999, 2000; Ueda et al, 2000), Rasgrf1 (Li et al, 2004), and Gtl2 (Li et al, 2004). IGF2/H19 is a reciprocally imprinted locus exhibiting paternal IGF2 and maternal H19 expression. Therefore, the noncoding RNA gene H19 exhibits a paternal imprinting pattern with hypermethylation on the silent paternal allele and hypomethylation on the expressed maternal allele (Figure 6). At present, there is no evidence that H19 has a function other than acting as a regulatory decoy for IGF2. A 2-kb differentially methylated domain (DMD) located −2 to −4 kb relative to the H19 transcriptional start site on the paternal allele acts as the imprinting center, the deletion of which results in loss of imprinting of both the IGF2 and H19 genes (Arney, 2003; Smith et al, 2004). . Genomic imprinting at the insulin-like growth factor 2 (IGF2)/H19 locus. While the IGF2 gene exhibits maternal imprinting resulting in the expression of the paternal allele, the noncoding RNA gene H19 reveals paternal imprinting that is followed by the expression of the maternal allele. A differentially methylated domain on the paternal allele approximately 2 kb upstream of the H19 promoter represents the imprinting center, the deletion of which results in loss of imprinting of both genes. On the unmethylated maternal allele, the imprinting center provides multiple binding sites for the enhancer blocking protein CCCTC-binding factor (CTCF), which assembles a boundary element to prevent access of downstream enhancers to the IGF2 promoters. On the paternal allele, methylation inhibits CTCF binding, allowing the enhancers to act exclusively on the IGF2 promoters (Bell and Felsenfeld, 2000; Hark et al, 2000). Repression of the paternal H19 promoter may require additional action of a local silencer (Banerjee et al, 2001). For further information, see Arney (2003) and Smith et al (2004). Arrows indicate transcription. In humans, 2 imprinting clusters have been extensively described: one on human chromosome 11p15 is linked to the pathogenesis of Beckwith-Wiedemann syndrome (BWS) and another on 15q11–13 is linked to Angelman syndrome (AS)/Prader-Willi syndrome (PWS) (Falls et al, 1999) (Figure 7). See further information in “Epigenetic Risks in ART.” . Genomic organization of imprinting clusters 11p15 (A) and 15q11-13 (B). Defects within imprinting cluster 11p15 are associated with Beckwith-Wiedemann-Syndrome (BWS). In BWS-1 patients, the maternal alleles acquire aberrant methylation within imprinting center (IC) 1, while IC2 remains normally methylated, resulting in silencing of H19 and biallelic expression of insulin-like growth factor 2. BWS-2 patients show loss of methylation within IC2, while IC1 remains unmethylated as normal, resulting in silencing of CDKN1C and biallelic expression of KCNQ1OT1. Defects within imprinting cluster 15q11-13 are associated with Prader-Willi syndrome if the deletion concerns the paternal chromosome—the candidate gene is the maternally imprinted SNRPN (not shown)—or Angelman syndrome if the deletion concerns the maternal chromosome—the candidate gene is the paternally imprinted UBE3A. While the majority of patients show a de novo deletion of the 15q11-13 region or a uniparental disomy, a few patients reveal aberrant imprinting and gene silencing. For further information, see Nicholls and Knepper (2001), Buiting et al (2003), Lucifero et al (2004a), Horsthemke and Ludwig (2005), and Kantor et al (2006). DMD indicates differentially methylated domain. Arrows indicate transcription. Imprinted genes share common characteristic features such as genomic clustering, suggesting that the primary control of imprinting occurs not at the single gene level but at the chromosome domain level. Parental copies of imprinted regions differ with respect to DNA methylation, histone modification, and consequently gene expression. A key component of imprinted regions is DNA methylation at CpG islands. Despite the frequent association, direct repeats do not seem to provide a common sequence motif that acts as a methylation signal. It is likely that DNA methylation is not acting in isolation and that chromatin modifications are an important factor in determining DNA methylation patterns (Reik and Walter, 2001; Verona et al, 2003; Delaval and Feil, 2004). Insulator pr