Epigenetic Aspects of Fertilization and Preimplantation Development in Mammals: Lessons from the Mouse

Abstract

During gametogenesis, the parental genomes are separated and are epigenetically marked by modifications that will direct the expression profile of genes necessary for meiosis as well as for the formation of the oocyte and sperm cell. Immediately after sperm-egg fusion, the parental haploid genomes show great epigenetic asymmetry with differences in the levels of DNA methylation and histone tail modifications. The epigenetic program acquired during oogenesis and spermatogenesis must be reset for the zygote to successfully proceed through preimplantation development and this occurs as the two genomes approach each other in preparation for karyogamy. During preimplantation development, the embryo is vested with the responsibility of maintaining the primary imprints. In addition, female embryos must silence one of the X-chromosomes in order to transcribe equal levels of X-linked genes as their male counterparts. This review is intended as a survey of the epigenetic modifications and mechanisms present in zygotes and preimplantation mouse embryos, namely DNA methylation, histone modifications, dosage compensation, genomic imprinting, and regulation by non-coding RNAs.

KEYWORDS:

• DNA methylation
• embryos
• epigenetics
• gametes
• histones

Abbreviations
ncRNA:	=	non-coding RNA
PN:	=	pronuclei
HDAC1:	=	histone deacetylase 1
PRC1:	=	polycomb repressive complex
Dnmts:	=	DNA methyltransferases
MBD:	=	methyl-CpG binding domain
RNAi:	=	RNA interference
dsRNA:	=	double-stranded RNA
nt:	=	nucleotides
RISC:	=	RNA-induced silencing complex
miRNA:	=	microRNAs
Ago1 – 4:	=	Argonaute proteins 1 – 4
ICR:	=	imprinting control region
HP1:	=	heterochromatin protein 1
XCI:	=	X-chromosome inactivation
DMR:	=	differentially methylated region
Xist:	=	X-chromosome inactive specific transcript
Tsix:	=	X-chromosome active
Xp:	=	paternal X chromosome
Xm:	=	maternal X chromosome
MSCI:	=	meiotic sex chromosome inactivation
TE:	=	trophectoderm
ICM:	=	Inner cell mass

INTRODUCTION

The 31st edition of the Dorland's Illustrated Medical Dictionary [Dorland 2007] defines GENETICS as the study of genes, their molecular structure and function, and their chemical and physical nature. The same source defines EPI as a prefix of Greek origin meaning above. Therefore, a literal meaning for EPIGENETICS is ‘above genetics.’ As a field, epigenetics deals with the study of the layers of heritable and resettable information to DNA and chromatin which alter the ‘readability’ of genes and result in variant cellular phenotypes. The mechanisms involved in epigenetic regulation include: DNA methylation, post-translational modifications of histone proteins, regulation of DNA by non-histone proteins, X-chromosome inactivation, genomic imprinting, and regulation by non-coding RNAs (ncRNA).

The male genome undergoes several epigenetic changes during spermatogenesis. This topic is beyond the scope of this review and the reader is invited to refer to the following reviews to attain further information [Lewis et al. 2003; Churikov et al. 2004; Govin et al. 2004; Rousseaux et al. 2005; 2008]. Likewise, epigenetic modifications change during oocyte growth and the acquisition of developmental competence. Finally, when the two recently transcriptionally silent haploid genomes converge at syngamy, a new set of epigenetic modifications needs to be in place for the new diploid organism to acquire a permissive transcriptional state that will coordinate development. This review is intended as a survey of the abovementioned epigenetic modifications and mechanisms and how these change during fertilization and preimplantation development in the mouse. Throughout this review, the modifications found at each pronuclear stage of 1-cell embryos will be referred to by position of the pronuclei (PN) in the embryo. PN0 occurs immediately after sperm entry into the oocyte while PN5 is when the two pronuclei are next to each other prior to karyogamy. Briefly, during PN1 and PN2 zygotes are in G₁ phase, at PN3 and PN4 they are in the synthesis phase, while in PN5 they are mostly at G₂ phase and are preparing for karyogamy. For pictures and drawings of these stages the reader is referred to the following articles Adenot et al. [1997] and Santos et al. [2002].

Histone Proteins

The nucleosome in eukaryotes is composed of 146 bp of DNA wrapped around an octamer of core histone proteins (reviewed in [McGhee and Felsenfeld 1980]). Histone proteins are rich in positively-charged amino acids which have an affinity for the negatively charged DNA. The octamer is composed of two units each, of the core histones 2A (H2A), H2B, H3, and H4. A linker histone (H1 or H5) then associates with the nucleosomal core particle by engaging with the entry and exit points of linker DNA. This association establishes the spacing between nucleosomes and is involved in chromatin compaction [Izzo et al. 2008; Routh et al. 2008]. It is recognized that post-translational modifications of histones (acetylation (ac), methylation (me), ubiquitination (ub), phosphorylation (p), sumoylation, and ADP ribosylation) change the structure and function of chromatin. It is the sum of these modifications and the amino acids they modify that orchestrates a permissive or repressive chromatin conformation [Bhaumik et al. 2007]. Two of the best described and understood histone modifications are acetylation and methylation. Acetylation is typically associated with gene transcription while methylation may be associated with active or repressive transcriptional states. In addition, histone tail lysine residues can be monomethylated (me1), dimethylated (me2), or trimethylated (me3), while arginine residues have been observed in mono- and di-methylated states. Some of the most common activating marks (those associated with a permissive transcriptional state) are: acetylation on lysine 9 of H3 (H3K9ac), H3K14ac, H4K5ac, H4K12ac, H3K4me, H3K36me, H3S10p, and K2BK123ub. Examples of the histone modifications which confer repressive activity are: H3K9me, H3K27me, H2AK119ub, H4K12me, and H4K20me.

During spermiogenesis, acetylated core histone proteins are exchanged by transition proteins [Hazzouri et al. 2000] before being replaced by protamines. Protamines are low molecular weight (<8000 kDa) highly basic nuclear proteins that coil DNA into toroidal structures. The overall positive charge of protamines is the result of their high arginine content (approximately 50%; [Balhorn et al. 1977]; for a review see [Oliva 2006]). In addition, they form disulphide bonds between cysteines in their sequence which results in the strong stabilization of the nucleoprotamine complex. This tight packing of DNA results in inactivation of the entire genome. Interestingly, approximately one percent of the sperm cell's DNA remains associated with histone proteins [Pittoggi et al. 1999; van der Heijden et al. 2005] and it is thought that their function is to shield imprinted loci.

Upon fertilization, the male chromatin undergoes a rapid sequence of remodeling events [Rodman et al. 1981]. Protamines are actively removed from the male genome within the first 5–10 min following sperm entry into the oocyte [Gao et al. 2004]. The protamines are removed as the chromatin expands and this can be visualized as immunoreactive bulb-like spheres which disperse and disappear as decondensation proceeds. The removal of protamines must be an enzyme directed process since an in vitro study calculated that the rate of passive protamine dissociation from 1.5×10⁹ DNA bases (the number of bases in the sperm genome) would take at least 6 years [Brewer et al. 1999]. A rapid decondensation of the paternal chromatin occurs by 30 min after penetration and it is followed by a condensation event approximately 20 min later. Finally, the paternal chromatin undergoes one more decondensation to form the pronucleus [van der Heijden et al. 2005]. The removal of protamines is coincident with the gain of nucleosomes, an activity that extends over a period of approximately 4 h. It should be noted that the histone transfer activity of oocytes is acquired during oocyte growth and achieved following entry into metaphase and requires calcium [McLay et al. 2002].

Synthesis of histone proteins occurs early during preimplantation development. H3 and H4 are synthesized during the 1-cell stage (from maternally stored mRNA) while synthesis of H2A, H2B, and H1 start during the late 1-cell to 2-cell stage [Wiekowski et al. 1997]. H3.3 is observed as a punctuate pattern at the 2-cell, 4-cell, and blastocyst stages which coincide with the major waves of gene activation observed during preimplantation development [Hamatani et al. 2004; Torres-Padilla et al. 2006]. H3.3 is a replication-independent variant of H3 which correlates with an active transcriptional state, and can be deposited onto nucleosomes at times other that the S-phase of the cycle [Ahmad and Henikoff 2002]. Covalent modifications of histone proteins on the chromatin of early embryos is very dynamic and most repressing and activating modifications to histone tails have been detected by immunofluorescence [Erhardt et al. 2003; Akiyama et al. 2004; Liu et al. 2004; Yeo et al. 2005; Huang et al. 2007; Torres-Padilla et al. 2007]. These modifications are likely involved in the initial establishment of the transcriptional repressive state that is observed concurrently with genome activation (reviewed by [Schultz 2002]). Whereas expression of the linker somatic protein H1 does not correlate with this repressive state [Stein and Schultz 2000], expression of the histone deacetylase 1 does (HDAC1; [Ma and Schultz 2008]). In mouse eggs and embryos, H4K5ac (a mark of histone hyperacetylation) has an inverse immunofluorescent localization with HDAC1. Other HDACs (i.e., HDAC2 and HDAC3) do not have opposing staining with H4K5ac, making HDAC1the likely candidate for removal of acetyl groups from the hyperacetylated H4. This observation was corroborated by a HDAC1 dsRNA injection which gave the expected result of increased H4K5ac in embryos. Western blot demonstrated that HDAC1 is maintained at about the same level from the MI to the 1-cell stage but then increases consistently until the blastocyst stage. HDAC 2 and 3 actually decrease as development proceeds.

MacroH2A, an H2A variant, is usually found in association with heterochromatin in its inherent transcriptionally silent state. In the zygote, macroH2A is not present in the paternal pronucleus and it is found early on in the maternal pronucleus but by 7 h post fertilization, coincident with synthesis (PN3-PN4), it becomes undetectable [Erhardt et al. 2003]. In addition, though it is present in developing and mature oocytes, where it resides in pericentric heterochromatin, no localization is found in the pronuclear stage zygotes through the 8-cell stage. MacroH2A appears again in blastomeres of morula stage embryos and by the blastocyst stage it is present in all nuclei. H1foo is an oocyte specific linker histone protein variant [Tanaka et al. 2001; 2003]. It is expressed in the oocyte and its mRNA progressively decreases from the zygote to the 8-cell stage where it is no longer detected [Tanaka et al. 2001; Zeng and Schultz 2003]. However, a methylation analysis of the 5' CpG island of the H1foo gene indicated that it is unmethylated in oocytes and also in blastocyst stage embryos [Maeda et al. 2008]; this is interesting given the lack of H1foo mRNA in embryos at that developmental stage. It is tempting to speculate that silencing of this gene is the result of histone modifications since methylation is lacking from the promoter region until after implantation. Immunolocalization places this linker histone in the nucleus of the growing oocyte with the highest amount at the fully grown stage [Gao et al. 2004]. Depletion by morpholino oligonucleotide against H1foo impairs first polar body extrusion and causes the oocytes to arrest at metaphase I [Furuya et al. 2007]. Immediately after fertilization, H1foo localizes to both parental pronuclei and a faint immunoreactive signal can be observed in the paternal pronucleus as soon as five min after intracytoplasmic sperm injection [Gao et al. 2004]. H1foo is replaced by H1 in the early embryo. H1's immunoreactive signal is first visible at the 2-cell stage, augments by the 4-cell stage, and inversely correlates with the amount of H1foo [Gao et al. 2004].

Epigenetic asymmetry between the parental pronuclei is evident at the 1-cell stage which seems to be largely the result of the protamine to nucleosome exchange. At the time of decondensation, the paternal chromatin associates with H3.3 and Hira (histone 3.3′s chaperone; [Tagami et al. 2004; van der Heijden et al. 2005; Torres-Padilla et al. 2006]. This is in accordance with the higher transcriptional activity observed in the paternal pronucleus [Aoki et al. 1997]. The core histone H3.1 immunolocalizes in all maternal pronuclear stages but is absent in the paternal pronucleus until DNA synthesis commences approximately 6–7 h post-fertilization. During this time, the paternal DNA also incorporates H2B [Polanski et al. 2008] and acetylated histones (i.e., H4K5ac and H4K12ac; [Adenot et al. 1997; Wu et al. 2008; Suo et al. 2009] are also transferred onto the male chromatin [Santos et al. 2002]. As expected from a higher transcriptional state of the paternal PN, acetylated histone content is increased in the male PN0 to PN2. However, by pronuclear stages 3 and 4 the two pronuclei stain with similar intensities and this occurs without a dependency on DNA replication. Further, during decondensation, the sperm PN is positive for H3K9me1 but not H3K9me2 or H3K9me3 while the maternal pronucleus intensely stains for all three modifications [Kim et al. 2003; Lepikhov and Walter 2004; Santos et al. 2005; Yeo et al. 2005; Wu et al. 2008]. This is not due to a competition between the parental pronuclei since parthenogenetic embryos stained brightly for H3K9me2 while androgenotes embryos did not. Interestingly, work from Liu et al. [2004] indicates that the H3K9 methylase might be deactivated after fertilization since transfer of paternal pronucleus to an enucleated oocyte promotes H3K9me2 on the paternal DNA. A very recent study determined the importance of the H3K9me3 demethylase JMJD2C during preimplantation development [Wang et al. 2010]. They found that the transcript for this protein is expressed at high levels from the 2- to the 8-cell stages of development, decreasing thereafter. RNAi-mediated depletion at MII showed the requirement of the demethylation activity of this protein for proper development as the majority of the JmJ2c depleted embryos were unable to cleave past the 8-cell stage. Additionally, a decrease in the transcript level of Jmj2c by double stranded technology caused a 60–90% reduction in the expression of the pluripotent genes Pou5f1, Nanog, and Sox2 as well as downregulation of the cell proliferation associated genes Myc and Klf4. Another histone modification has been correlated with pluripotency in early mouse embryos [Torres-Padilla et al. 2006]. In this study, they demonstrated that levels of H3R26me are different in the various blastomeres of a 4-cell embryo. They go further to show that those blastomeres with the higher amount of H3R26me are most often destined to localize to the inner cell mass. In addition, overexpression of the H3K26′s methyltransferase CARM1 induces upregulation of Nanog and Sox2. These observations point at the importance of the epigenetic machinery for the timely activation of pluripotency-associated genes.

That post-translational modifications of histones associate differentially with the chromatin of each parental pronucleus was also demonstrated in the work of Puschendorf and coworkers [2008] who found that localization H3K27me3 was exclusive to the maternal pronucleus from PN 0–3. At PN5, however, intense staining was observed at major satellite heterochromatic regions of the paternal pronucleus. H3K27me3 is a modification that is catalyzed by one of the members of the Polycomb initiation complex (Polycomb repressive complex 2; PRC2), namely EZH2. The trimethylated state of H3K27 serves as a ‘docking’ site for the maintenance Polycomb repressive complex (PRC1) through its chromodomain protein CBX2/8. Once on site, the catalytic subunit of the PRC1 complex, the ubiquitin ligase RNF2, modifies H2A at its 119 amino acid (i.e., lysine) thus inducing a heterochromatic state that can be remembered during subsequent cell divisions. In the zygote, maternally inherited PRC1 components localize to the paternal pronucleus at PN5 an activity that is required to silence major satellite transcription [Puschendorf et al. 2008] from the paternally inherited DNA. This was clearly demonstrated in 2-cell embryos that formed from RNF2 deficient oocytes which had a three-fold increase in major satellite paternal transcript level when compared to wild-type embryos.

DNA METHYLATION

In mammals, DNA methylation is a post-replicative, reversible, and heritable covalent chemical modification that occurs primarily at the C5 position of the cytosine pyrimidine ring when in a CpG context (reviewed in [Prokhortchouk and Defossez 2008]. DNA methyltransferases (Dnmts) recognize palindromic CpG sequences as substrate and associate covalently with the C6 position of the cytosine to catalyze a methyl group transfer to the C5 using a methyl group donated by S-adenosyl-methionine. DNA methylation is generally associated with the control of gene expression as methylated sequences often undergo transcriptional repression [Chomet 1991]. Methylated cytosines account for one percent of all bases, varying slightly in different tissue types, and play a role in long-term silencing of transcription and heterochromatin formation.

The methyl groups on modified cytosines are accessible in the major groove of DNA and cause changes in DNA conformation that can be inhibitory [Watt and Molloy 1988] or permissive [Meehan et al. 1989; Lewis et al. 1992] to protein binding. There are several methyltransferases that can methylate DNA. These enzymes are divided in two groups: maintenance DNA methyltransferase (Dnmt1) and de novo Dnmts (3a and 3b). Dnmt1 forms a complex with the processivity factor PCNA [Chuang et al. 1997; Sharif et al. 2007] during the S-phase to copy the methylation profile of the parent DNA strand onto the daughter strand as replication is underway. The de novo methyltransferases have a preference for unmethylated substrates [Yokochi and Robertson 2002]. Dnmt3a and 3b require an association with Dnmt3l which, by binding to their catalytic domain, increases their catalytic activities [Gowher et al. 2005]. Once a sequence of DNA is marked by methylation, another set of proteins is involved in recognizing the methylated cytosines and interacting with chromatin remodelers in order to silence transcription (reviewed in [Klose and Bird 2006]). These are the DNA methyl-CpG binding domain (MBD) family of proteins. The MBD family is comprised of at least five proteins namely; MeCP2, MBD1, MBD2, MBD3, and MBD4. MBD4 is mainly implicated in DNA repair mechanisms and act as a DNA glycosylase involved in fixing T:G mismatches [Zhu 2009]. It appears that there is not a requirement for MBD2 during the acquisition of methylation which occurs concomitant with oocyte growth because the maternal pronucleus of MBD2 null zygotes have comparable levels of global DNA methylation than that of wild-type embryos [Santos et al. 2002]. In the preimplantation embryo, expression of MBD2 (an MECP1 complex protein) is not present before the 16-cell stage [Kantor et al. 2003] but immunoreactivity is detected by the blastocyst stage of development. In addition, other MECP1 complex proteins, namely, MBD3, HDAC1, HDAC2, and Mi2 colocalize with MBD2 at the blastocyst stage. Knock out studies show that the transcriptional repressor MBD2 is not necessary for normal pre- and post-implantation development, however, MBD3 null embryos fail to develop after implantation [Hendrich et al. 2001].

Oocytes enclosed within primordial follicles have negligible global DNA methylation [Almamun and Rivera unpublished]. As the oocytes increase in size there is a concomitant increase in DNA methylation [Kageyama et al. 2007; Almamun and Rivera unpublished]. For example, while repetitive elements are undermethylated in the dictyate stage oocyte they are completely methylated in the ovulated egg. The increase in methylation correlates with an increase in the expression of the de novo Dnmts, which attain their highest level at approximately 60 μm diameter [Lucifero et al. 2007]. Dnmt1 has several isoforms with the somatic (Dnmt1s) form being 190 kDa. In oocytes, transcription of a sex-specific alternatively spliced isoform results in a 175 kDa protein lacking 118 amino acids in its N-terminus [Carlson et al. 1992; Mertineit et al. 1998]. This oocyte-specific isoform, Dnmt1o, accumulates in the oocyte during the growth phase and localizes in the cytoplasm of the fully grown oocyte. In the zygote, at PN0, Dnmt1o is associated with the maternal genome [Kurihara et al. 2007]. From PN1 to PN4 it also associates with the paternal DNA, albeit at lower intensities, and it is interesting to note that Dnmt1o localizes to the paternal DNA even when global demethylation is taking place. At PN5, the maternal pronucleus Dnmt1o staining decreases relative to that of the paternal pronucleus, suggesting that Dnmt1o has been exported out of the pronucleus by this stage. During the preimplantation stages, Dnmt1o localizes in the cytoplasm except during the 8-cell stage where it has a nuclear localization [Carlson et al. 1992; Mertineit et al. 1998; Doherty et al. 2002]. This nuclear translocation has been shown to be necessary to maintain the primary methylation imprints in embryos as Dnmt1o deficient embryos experience loss of methylation from the 8-cell stage onwards when compared to wild type controls [Cirio et al. 2008]. The nuclear movement of Dnmt1o accounts for maintenance of methylation imprints at the fourth cell cycle but does not explain how these are maintained in earlier stages as well as in the stages immediately following the 8-cell stage. Recently, two independent laboratories [Kurihara et al. 2007; Cirio et al. 2008] demonstrated that mouse embryos have embryo-derived Dnmt1s activity throughout the preimplantation stages. This isoform localizes to the nucleus in all stages of preimplantation development where it has been shown to be involved in maintaining methylation of repetitive elements and imprinted loci [Cirio et al. 2008]. Lastly, no function has been attributed to the de novo methyltransferases in maintaining methylation state of preimplantation embryos [Hirasawa et al. 2008].

It is well established, and widely accepted, that an active mechanism of global demethylation exists in the murine 1-cell embryo soon after fertilization [Oswald et al. 2000; Barton et al. 2001; Santos et al. 2002]. The elusive demethylase specifically acts on the paternal pronucleus (immunostaining with a 5-methyl cytosine antibody revealed a 70% reduction in methylation at ∼6 h post-fertilization) and this activity is not the result of replication [Oswald et al. 2000]. No concomitant decrease in global DNA methylation is observed in the female pronucleus. As the blastomeres divide, the maternal DNA passively demethylates, reaching its lowest methylation levels in all cells of the morula-stage embryo as well as in the trophectoderm (TE) cells of the blastocyst [Santos et al. 2002]. The passive demethylation of the maternal chromosomes is a result of cell division as it can be almost completely prevented by inhibiting replication. Interestingly, the parental chromatin seem to be differentially marked because the pronuclei of parthenotes do not undergo active demethylation [Barton et al. 2001]. To this effect, it has been recently demonstrated [Nakamura et al. 2007] that the maternal factor PGC7/Stella protects the maternal pronucleus from demethylation immediately after fertilization. In another recent report, the requirement for an active demethylation of the paternal DNA has been questioned [Polanski et al. 2008] as no differences were observed in development and survival to birth of embryos produced by round spermatid injection or mature sperm injection. DNA of round spermatids does not undergo demethylation upon injection into eggs.

Control of DNA methylation during preimplantation development is crucial for proper embryonic development and gene expression. This has been demonstrated in studies in which the demethylating agent (5-AZA-2′-deoxycytidine; 5-AZA-Cdr) was applied to pronuclear, 2-, and 4-cell stage embryos. 5-AZA-Cdr prevented development past the 7-cell stage [Yu et al. 2009]. This was associated with a decrease in the transcript level of several of the connexins (gap junction protein family members) as well as proteins involved in tight junction formation. This is intriguing given that mouse embryos are naturally undergoing global DNA demethylation [Oswald et al. 2000; Barton et al. 2001; Santos et al. 2002]. Further, we recently made the observation [Huffman and Rivera unpublished] that blastocyst stage embryos that formed from oocytes that had been hormonally stimulated to ovulate (i.e., superovulation; SO) had increased transcript level of several genes when compared to their naturally ovulated (NO) counterparts. We reasoned that the increased gene expression of embryos that formed from superovulated eggs was the result of improper acquisition of DNA methylation during oocyte growth. In order to test this hypothesis we analyzed global DNA methylation in the maternal pronucleus of zygotes using an antibody that recognizes methylated cytosines. The fluorescence intensity of the maternal pronucleus of SO embryos was ∼50% that of NO embryos. Taken together these observations would suggest that proper acquisition of methylation during oocyte growth and maintenance at certain loci, even during a state of global demethylation, is required for normal gene expression during the preimplantation development.

SMALL RNA-DIRECTED GENE SILENCING

A recently identified regulatory mechanism is that which is directed by small RNAs. These RNAs range in size from 18 to 30 nucleotides (nt) in length, play a central role in regulating gene expression, and can be classified by size and ribonuclease III dependency. The more prevalent and better studied small RNAs are microRNA (miRNA; 21–23 nt), small interfering RNA (siRNA; 20–24 nt), and Piwi-interacting RNA (piRNA; 24–30 nt). An association has been made with these small RNAs and the establishment of DNA methylation [Kuramochi-Miyagawa et al. 2008; Sinkkonen et al. 2008].

RNA interference (RNAi), initially referred to as ‘co-suppression’, was first identified in plants by Napoli et al. [1990] who, in an attempt to intensify the color of petunias, introduced a transgene to overexpress the rate limiting enzyme involved in flower coloration. This resulted in a remarkable reduction in the expression of a homologous gene. The mechanisms of this silencing came to be identified by the seminal work of Andrew Fire and Craig Mello [Fire et al. 1998] who, using C. elegans, identified that double-stranded RNA (dsRNA) induced gene silencing as evidenced by known phenotypes and decreased expression of GFP. Work followed in other species and it was soon demonstrated that this mode of gene silencing was present and active in mammals [Svoboda et al. 2000; Wianny and Zernicka-Goetz 2000]. RNAi is believed to be a primitive form of cellular defense mechanism involved in protecting the cell from foreign nucleic acids [Naqvi et al. 2009]. In addition, endogenous long dsRNA exist which are the product of transposable elements, long stem-loop structures, and sense-antisense pairs [Kim et al. 2009]. These dsRNA are processed in the cytoplasm by the ribonuclease III family member Dicer and then incorporated into the RNA-induced silencing complex (RISC with its catalytic component Argonaute 2; Ago2) to produce an ∼21nt siRNA that induces post-transcriptional silencing of the perfectly base-paired target [Carthew and Sontheimer 2009].

MicroRNAs (miRNA) occur in clusters (reviewed by [Zhang et al. 2009]) and are synthesized as long primary transcripts which form double stranded stem-loops as a result of base complementarity of the transcript. In animals, miRNA genes are firstly transcribed by RNA polymerase II or RNA polymerase III to generate primary miRNA transcripts [Lee et al. 2004; Borchert et al. 2006]. They are then processed by Drosha (an RNase III family member; [Tomari and Zamore 2004]) and Dgcr8 (its dsRNA binding partner) to produce the hairpin precursors of 70—90 nt. These pre-miRNAs are then exported into the cytoplasm where Dicer and the RISC complex generate a single stranded ∼22nt RNA (mature miRNA). These microRNAs bind to the 3′ UTR of their target RNA and due to their imperfect complementarity (usually one nucleotide mismatch) they do not induce post-transcriptional silencing but rather translational repression [Chu and Rana 2006].

piRNAs are a more recently identified group of small RNAs [Aravin et al. 2006] with germ cell-specific expression. They are derived from only one strand of DNA as a result of transcription of transposable elements and are processed as long primary transcripts. Once in the cytoplasm they are processed by a Dicer-independent mechanism involving the PIWI proteins (Argonaute subfamily) and, through a ‘ping-pong’ cleavage and amplification cascade, they protect the germline from transposon expression ([Aravin et al. 2007]; reviewed in [Kim et al. 2009]). The ping-pong mechanism results from recognition and binding of sense piRNA by Mili which directs the cleavage of antisense piRNA [Aravin et al. 2007; Tushir et al. 2009]. The processed antisense piRNA is bound to Miwi2 and this is used to cleave sense transcripts in a revolving amplification process. Mili and Miwi2 are Piwi subfamily genes in the mouse genome.

One of the first reports showing that the RNAi machinery was active in mammals came from the work of Wianny and Zernicka-Goetz [2000] who eliminated specific gene expression by injecting dsRNA into oocytes and embryos. Using a transgenic GFP line of mice they demonstrated that injection of dsRNA for GFP at the 1-cell stage abolished GFP expression at the 4-cell, morula-, or blastocyst-stages of development. They went further to demonstrate that the interference of GFP gene expression was specific as injection of other dsRNA, namely E-cadherin, had no effect on green fluorescence but showed phenotypes known to occur in E-cadherin-null mutant mice. E-cadherin dsRNA injection into zygotes disrupted development past the compacted morula stage and inhibited cavitation to form blastocysts. A similar observation was made by Svoboda et al. [2000]. These authors reported a knockdown effect of dsRNA directed towards cMos (a maternal transcript that encodes a cytostatic factor) and Plat mRNAs in mouse oocytes which was time and concentration dependent. Targeting the cMos mRNA inhibited the appearance of MAP kinase activity and promoted parthenogenetic activation of the oocytes while injecting Plat mRNA inhibited production of tissue plasminogen activator activity.

With the identification of an effective RNAi mechanism in mammalian oocytes and embryos came some concerns. These concerns included the possibility of off-target effects of RNAi, as well as the engagement of the non-specific degradation of RNA by an antiviral response as some reports had demonstrated the ability of dsRNA to induce the interferon response [Samuel 2001]. In order to shed light into this question Stein et al. [2005] performed studies aimed at determining if off-targeted responses existed in transgenic oocytes expressing long dsRNA. By comparing global gene expression between wild type and transgenic knock-down oocytes they showed that long dsRNA did not induce interferon-associated genes nor off-targeting effects in mouse oocytes. Further, quantitative PCR revealed that genes involved in the interferon response pathway are not expressed in mouse oocytes.

Endogenous retroelement-derived siRNAs have been identified in oocytes [Watanabe et al. 2006]. Two independent laboratories [Tam et al. 2008; Watanabe et al. 2008] demonstrated that gene expression in oocytes is regulated by pseudogene derived siRNA. Even though the biogenesis and function of endogenous dsRNA is not known, what appears to be clear is the oocyte's need to use this pathway to attain proper gene regulation. In oocytes with no Dicer and Ago2 activity an increase in retrotransposons and protein-coding transcripts is observed, indicating a function of dsRNA in gene regulation [Watanabe et al. 2008].

Several laboratories have confirmed the presence and roles of miRNAs and their processing machinery in oocytes and preimplantation embryos [Amanai et al. 2006; Watanabe et al. 2006; Cui et al. 2007; Morita et al. 2007; Murchison et al. 2007; Tang et al. 2007; Lykke-Andersen et al. 2008; Mtango et al. 2009]. miRNA are expressed in a stage-dependent manner [Yang et al. 2008] with a decrease from the 1- to 2-cell stage possibly followed by de novo synthesis from the 2- to the 4-cell stage and this continues as development progresses [Tang et al. 2007]. Further, there are no differences in miRNA expression between blastomeres at the 2- and 4-cell stage. Synthesis of maternal miRNA requires Dicer as a conditional knockout showed that most of the miRNAs were essentially lost from these oocytes, resulting in loss of female fertility [Tang et al. 2007].

As it is well established that miRNA regulate gene expression, and based on the fact that mature sperm cells contain RNAs, including miRNAs (reviewed in [Dadoune 2009]), Amanai et al. [2006] performed investigations to elucidate possible silencing of oocyte mRNA by sperm-borne miRNA during fertilization and preimplantation development. miRNA analysis of detergent extracted sperm detected a core of 54 miRNAs in all samples analyzed which can be expected to enter the oocyte along with the paternal nuclear and/or perinuclear compartment during fertilization. Comparison of the amount of miRNA between unfertilized MII eggs and newly fertilized eggs (2 h) failed to show differences between the two. Lastly, chemically inhibiting several of the sperm miRNA had no effect on egg activation or their cleavage rates. In conclusion, these studies failed to attribute a contribution of sperm-borne miRNA at fertilization.

The amount of Dicer transcript decreases from the GV to the blastocyst stage but knockdown of Dicer from GV oocytes and zygotes does not affect oocyte maturation or blastocyst development [Cui et al. 2007]. However, Dicer may directly or indirectly regulate fate of the blastocysts; removing this protein by siRNA caused down-regulation of the totipotency-related genes POU5f1, Nanog, and Sox2. In contrast to the apparent dispensability for resumption of meiosis in oocytes and preimplantation development, females with Dicer deficient oocytes are not fertile and this phenotype was not the result of oocyte growth or hormonal response [Murchison et al. 2007]. Dicer deficient oocytes were able to undergo germinal vesicle breakdown but unable to extrude the first polar body during in vitro culture which indicated that Dicer is required for efficient completion of meiosis I. Closer examination of these oocytes by confocal microscopy demonstrated abnormal meiotic spindle at meiosis I and II as well as multiple spindles, which reflects a requirement of Dicer for proper spindle formation, chromosome attachment, and progression through meiotic maturation [Murchison et al. 2007; Tang et al. 2007]. Furthermore, Dicer seems to be intimately involved in the maternal transcript turnover during meiotic degradation and removal of Dicer from the oocytes resulted in an increase of 173 targets found in a study by Su et al. [2007] to be degraded during meiotic maturation.

Messenger RNA expression patterns are different for the four Argonaute proteins (Ago1-4). While transcripts of Ago2 and Ago3 are present from the GV oocyte to the blastocyst stage, no Ago1 is detected at these stages. Ago4 had a similar pattern of expression as Ago2 and 3 early on but was down-regulated by the 4-cell stage [Lykke-Andersen et al. 2008]. Ago2 protein localizes to the P-body, consistent with what has been previously demonstrated in somatic cells [Liu et al. 2005]. In Ago2 knockdown embryos, genes are both up- and down-regulated indicative of a need of RNAi in the maintenance of correct amounts of transcript during embryonic development. The up-regulated genes included several maternal transcripts that are normally degraded during the 2-cell stage. In sum, reports are conflicting as to the requirement for Dicer and Ago2 during preimplantation development. While some reports show that both Dicer and Ago2 expression do not seem to be absolutely required during preimplantation development [Bernstein et al. 2003; Morita et al. 2007], the results from another report [Lykke-Andersen et al. 2008] argue the need for Ago2 since siRNA at the zygote stage resulted in developmental arrest at the 2-cell stage.

piRNA have also recently been detected in mouse oocytes [Tam et al. 2008] but are not present in mature spermatozoa [Girard et al. 2006]. It appears that this class of small RNA shares the task with miRNA in regulating retrotransposons and controlling pseudogene generated dsRNA [Watanabe et al. 2008; Tam et al. 2008].

A role in DNA methylation has been ascribed for some of the small RNAs in two recently published reports. In one study, Kuramochi-Miyagawa et al. [2008] described a requirement for Mili and Miwi2 for de novo DNA methylation. Their results showed a reduction in the amount of piRNA and an associated hypomethylation of several retrotransposons' (IAP and Line-1) regulatory regions in germ cells of Mili- and Miwi2-null male fetuses. In the other report, Sinkkonen et al. [2008] showed an involvement of miRNAs in DNA methylation. They demonstrate defective de novo methylation in a Dicer-null mouse embryonic stem cell model. This mechanism is thought to involve the miRNA 290 cluster which represses an inhibitor of the de novo DNA methyltranferases.

GENOMIC IMPRINTING

Genomic imprinting is an epigenetic modification in mammals which results in parent-specific gene expression. To date, approximately 80 imprinted genes have been identified in mice (http://www.har.mrc.ac.uk/research/genomic_imprinting/maps.html; [Morison et al. 2005]). Imprinting is a multi-step process that starts each reproductive cycle with the epigenetic markings of the gametes in a sex-specific manner. The stable transmission of the parent-specific mark to the offspring and the resulting monoallelic expression of imprinted genes are necessary for proper regulation of embryonic growth, placental function, and neurobehavioral processes. Imprinted genes occur in clusters throughout the genome. The correct allelic expression of the cluster's genes is achieved by epigenetic modifications (e.g., DNA methylation and/or histone modifications) of a regulatory region known as the imprinting control region (ICR; [Verona et al. 2003; Edwards and Ferguson-Smith 2007]). The ICR is often a differentially methylated region (DMR) of DNA. In addition, every cluster identified to date contains at least one non-coding RNA (ncRNA; [Royo et al. 2006; Royo and Cavaillé 2008; Ideraabdullah et al. 2008]). Although the function of some ncRNAs is unknown, others are required for maintenance of the allele-specific expression of the locus' genes [Royo and Cavaillé 2008]. Further, differences exist in the manner in which ICRs regulate allele specific expression of genes. For example, whereas the Igf2/H19 cluster of genes is regulated by binding of the insulator protein CTCF to the unmethylated maternal ICR [Verona et al. 2003; Engel et al. 2006], the main regulatory mechanisms of the neighboring KvDMR1 cluster is the transcription of the ncRNA Kcnq1ot1 [Thakur et al. 2004; Kanduri et al. 2006; Pandey et al. 2008; Terranova et al. 2008].

There are currently 20 described germline DMRs, 17 maternal and 3 paternal (for a summary list of the germline DMRs see [Chotalia et al. 2009]). The establishment of the parental primary imprints occurs at different times in male and female gametes. In the male, acquisition of methylation occurs before birth, during germ cell development, while the maternal alleles are marked after birth during oocyte growth. The deposition of DNA methylation at DMRs occurs in an asynchronous manner as the oocyte increases in diameter and accumulates the de novo methyltranferases [Lucifero et al. 2004; Hiura et al. 2006]. Even though there are several gametic DMRs, monoallelic expression of these genes does not all start during preimplantation development. On one hand, the paternally-expressed growth factor Igf2 is expressed from the 2-cell stage onwards but its expression is biallelic during preimplantation development [Latham et al. 1994; Szabó and Mann 1995]. On the other hand, two other paternally-expressed genes, namely Kcnq1ot1 and Snrpn, are expressed monoallelicaly from the earliest preimplantation stages [Szabó and Mann 1995; Terranova et al. 2008].

Once the sperm cell enters the egg and the zygote forms, the imprinting marks of each parental allele must be maintained during preimplantation development. As discussed earlier, the maintenance methyltranferases Dnmt1o and Dnmt1s are essential for maintaining the methylation patterns when global demethylation is taking place. There exist several proposed mechanisms that explain how the parental alleles are differentiated and their marks propagated while in the same nucleus. One of these is asynchronous timing of replication during the S-phase. For example, Simon et al. [1999] showed this to be the case in preimplantation embryos where Igf2's active paternal allele replicated earlier than the repressed maternal allele where replication lagged to the later part of the phase. Other mechanisms have emerged to explain the resistance of imprinted loci to demethylate during these early stages. For example, Reese et al. [2007] showed that Mbd3 preferentially localizes to the paternal allele of the H19 DMR although a similar mechanism is not needed for maintaining other DMRs (i.e., Snrpn). Lastly, the work of Nakamura et al. [2007] proposes a requirement for the maternal factor PGC7/Stella in maintaining the methylation imprints during demethylation at the 1-cell stage, as PGC7 null embryos had considerable loss of methylation at several maternal DMRs. This was not the result of inappropriate deposition of methylation imprints since methylation at the same DMRs in unfertilized null eggs was undistinguishable from controls.

Allele-specific modifications to histones and chromatin conformations have also been investigated during early embryogenesis at imprinted loci. In a recent study Kim and Ogura [2009] were able to show the existence of several histone modifications at two ICRs (i.e., KvDMR1 and IGF2r) using a modified chromatin immunoprecipitation assay. Specifically, they showed enrichment of H3K9me2 and H3K27me3 but not of H4K12ac, on the repressed allele. The authors suggested that an association between Dnmt1 and H3K9me2 through HP1 (heterochromatin protein 1) binding may reinforce the silencing signal at these loci. In fact, HP1interacts with Dnmt1 and promotes its methyltransferase activity in a H3K9me2 dependent manner [Smallwood et al. 2007]. Furthermore, higher order chromatin conformations have also been associated with maintenance of imprinting in preimplantation embryos. The KvDMR1 is located within the 10^th intron of the maternally-expressed gene Kcnq1. The KvDMR1 is a primary imprint [Yatsuki et al. 2002] which serves a dual purpose. First, it is the ICR of a cluster of one paternally-expressed gene and several maternally-expressed genes. Second, it is the promoter region for the paternally-expressed ncRNA Kcnq1ot1 [Mancini-DiNardo et al. 2003]. Kcnq1ot1 is a long (>80 kb in mice) untranslated RNA with a bidirectional silencing function [Thakur et al. 2004; Kanduri et al. 2006]. An unmethylated KvDMR1 on the paternal chromosome allows for the expression of this gene which results in silencing of several flanking imprinted genes (e.g., Cdkn1c). On the maternal chromosome however, the KvDMR1 is methylated, which prevents Kcnq1ot1 expression and allows maternal genes to be expressed. Terranova et al. [2008] showed that transcription of Kcnq1ot1 is initiated in the zygote. The ncRNA transcript associates (or is in close proximity) with the catalytic members of the polycomb repressive complex 1 (Rnf2) and 2 (Ezh2) and their respective imparted histone modifications, H2AK119ub1 and H3K27me3. These repressive histone modifications also correlated with the contracted chromatin conformation of the paternal allele. In addition, the parental alleles were physically distant from each other and showed differences in immunoreactivity of RNA polymerase II [Terranova et al. 2008].

X-CHROMOSOME INACTIVATION

In mammals, females (homogametic sex; XX) must silence one of the X-chromosomes during embryonic development in order to transcribe equal levels of X-linked genes as that of males (heterogametic sex; XY). Dosage compensation of X-linked genes is achieved by a process known as X-chromosome inactivation (XCI). The X-chromosome has a region known as the X-chromosome inactivation center (Xic; [Rastan 1983]) which harbors two ncRNAs necessary for keeping the X-chromosome in an inactive (Xist X-inactive specific transcript) or active (Tsix) state. The 17 Kb Xist transcript coats the X chromosome in cis and this initiates a cascade of epigenetic modifications that impart a silencing configuration to this large chromosome (reviewed in [Thorvaldsen et al. 2006 and Wutz and Gribnau 2007]). The early events of XCI include counting, choice, and initiation (reviewed by [Avner and Heard 2001]). In the counting step the number of X-chromosomes is stochastically determined and all Xs except one will undergo inactivation so that the ploidy of Xs is the same between male and female cells [Rastan 1983; Monkhorst et al. 2008]. The ncRNA Tsix, is 40-kb in length and originates 15 Kb downstream of Xist and is transcribed across the Xist locus in an antisense manner. Its transcription negatively regulates Xist [Lee et al. 1999].

In the mouse, an imprinted form of XCI exists since only the paternal X chromosome (Xp) is silenced during the first few cell divisions as a result of Xist expression [Kay et al. 1994]. Xist begins to be transcribed at the G2-phase of 2-cell female embryos and this expression seems directed by a fertilization-induced molecular clock, since culturing 1-cell embryos in the cytokinesis inhibitor cytochalasin D did not prevent its expression [Zuccotti et al. 2002]. At the 2-cell stage, immunofluorescent studies show the localization of Xist on the Xp as punctuate staining [Nesterova et al. 2001; Patrat et al. 2009]. Although some controversy exists about when the Xp is first inactivated most reports agree that by the 4- to 8-cell stage most cells of an embryo have expression and accumulation of Xist [Nesterova et al. 2001; Zucotti et al. 2002; Okamoto et al. 2004]. Yet, the Xist association to Xi metaphase chromosomes is weaker during the preimplantation stage than after implantation [Matsui et al. 2001]. To that effect, there is no enrichment of H3K9 or H3K27 methylation before the 8-cell stage nor is there accumulation of the polycomb group repressive complex members Eed or Ezh2 before the 16-cell stage. Even though these could be detected on the Xp by the 16-cell stage in a portion of blastomeres, they do not seem to coexist until the blastocyst stage. Detection of H3K9me occurs at >32-cells stage and this is preceded by hypoacetylation of H3K9 and hypomethylation of H3K4 [Okamoto et al. 2004]. Further, RNA polymerase II was progressively excluded from the Xi starting at the 4-cell stage and was universally excluded from the Xp in all the blastomeres by the 32-cell stage.

A debate exists as to the mechanism involved in imprinted X chromosome inactivation [Huynh and Lee 2003; Namekawa et al. 2006; Patrat et al. 2009], and the reason why the Xp becomes inactivated while the maternal X (Xm) remains active during preimplantation development. Xp inactivation has been proposed to be the result of a previous inactive state (i.e., meiotic sex chromosome inactivation; MSCI) of the Xp which then renders this chromosome ‘vulnerable’ by carrying a repressive conformation into the egg [Huynh and Lee 2003; Namekawa et al. 2006]. The Huynh and Lee [2003] report indicates a lack of transcription in the vicinity of Xist and this prompted their conclusion that the presumed Xp is inactive at this stage. However, work from Patrat et al. [2009] argues against this hypothesis. These authors demonstrate that there is transcriptional activity of repeats on the Xp at the 2-cell stage. They also show no difference in nuclear space occupancy or the size of the maternal and paternal X chromosomes at the 2-cell stage which points to a lack of condensation of the paternal X. Patrat et al. [2009] go further to demonstrate transcriptional activity from the Xp by using RNA FISH against 18 genes located along the length of the X chromosome. Although those genes in close proximity to the Xic had a higher likelihood of being inactivated earlier this was not the case for all genes with some genes escaping inactivation all together by the blastocyst stage. Interestingly, two genes known to undergo MSCI, namely Huwe1 and Phda1, escape inactivation during the preimplantation stages. What is agreed upon is that the X-chromosome inactivation starts at the X-inactivation center and migrates towards the chromosome ends creating a gradient of gene inactivation (although leaky) from the center to the edges. This has been demonstrated by parent-specific gene expression in several studies [Huynh and Lee 2003; Mak et al. 2004; Namekawa et al. 2006; Patrat et al. 2009].

No expression of Tsix is detected until the blastocysts stage with only the maternal allele being transcribed in the trophectoderm (TE; reviewed in [Sado et al. 2001]). By the blastocyst stage, the paternal Xist remains active in the TE while it is silenced in the ICM in preparation for random XCI [Zuccotti et al. 2002; Okamoto et al. 2004]. In the ICM, reactivation of the Xp occurs between the early and the late blastocyst stages as determined by loss of Xist coating, lack of Eed and Ezh2 enrichment, and a gradual decrease in H3K9 and K27 methylation. Mak et al. [2004] observed loss of Eed foci in the ICM of late blastocyst stage embryos while the Eed staining remained positive in Nanog negative cells (i.e., primitive endoderm lineage). Further, expression and binding of the pluripotency markers Nanog, POU5f1, and Sox2 to the first intron of Xist have been suggested to be involved in the switch from imprinted to random XCI in the ICM (indirectly studied by the use of ES cells; [Navarro et al. 2008]). Finally, the levels of Tsix were not downregulated, pointing at the pluripotent markers and not Tsix in the reversal of the imprinted XCI. Therefore, XCI is Tsix independent until around the time when the Xp reactivates and random XCI initiates.

There is indication that the maternal X chromosome is marked in a way that is recognized by the blastomeres in the early preimplantation embryo. However, the nature of these marks remains elusive. In search of this ‘imprint’ Chiba et al. [2008] conducted studies using female preimplantation mouse embryos that were produced from oocytes deficient in Dnmt3a and Dnmt3b. When females with these double null oocytes were bred to males that had a deletion for Xist the previously observed focus with H3K27me3 staining was no longer evident suggesting that the Xp is coated with this histone modification. This result also suggested that the Xm has an imprint that maintains it in an active state. In addition, in these embryos, expression of genes adjacent to the XIC were silenced while those farther away had not undergone inactivation indicating that de novo methylation is not necessary for proper maternal X-chromosome function. The ‘imprint’ which confers protection against inactivation to the Xm of female preimplantation embryos is deposited on the X chromosomes during oocyte growth, specifically after the first meiotic prophase [Tada et al. 2000]. The strength of this maternal imprint to resist inactivation was shown by Tada et al. [1993] and Goto and Takagi [1999] who generated embryos that contained two Xm and one Xp (XmXmXp). Those embryos failed to inactivate the second maternal X in the trophectoderm. A series of successive transplantations of nuclei from non-growing or fully grown oocytes showed that XCI in embryos developing from oocytes which had received fully grown nuclei had an active X-chromosome in the extraembryonic tissues [Tada et al. 2000]. This was not the case, however, in embryos that developed from oocytes receiving the nucleus of a non-growing oocyte. In this group, these Xs preferentially underwent inactivation in the trophectoderm derived tissues. Further, parthenogenetic embryos do not have Xist expression until the morula stage indicating that around this stage the maternal imprint or factor is decreasing [Kay et al. 1994; Nesterova et al. 2001]. An alternate explanation for these observations is that the embryo might be unable to count extra maternal chromosomes. That this might be the case comes from the work of Matsui et al. [2001] who using XmXmXp embryos and parthenogenetic embryos (XmXm) could not detect inactivation of the extra Xm in embryos containing <10 cells. Matsui et al. [2001] also showed that embryos lacking a maternal X (XpO) expressed Xist at the 4–8 cell stage and expression decreased as development to the blastocyst stage progressed. This finding prompted them to conclude that the Xp is positioned to express Xist and will do so during the first cleavage divisions until another signal (counting mechanism) is engaged around the blastocysts stage.

Polycomb group proteins associate with the Xi. During the early and mid-blastocyst stage Eed and Ezh2 immunolocalize to the Xi in both ICM and TE cells [Erhardt et al. 2003; Okamoto et al. 2004; Kalantry and Magnuson 2006]. However, the association is lost in the ICM of late blastocyst stage which is concurrent with the reactivation of the Xp in preparation for random X inactivation. Eed can be detected as immunoreactive foci on the Xi in the blastomeres of ∼50% of morula-stage embryos [Silva et al. 2003] and this coincides with gain of methylation of H3K27, loss of methylation of H3K4, and hypoacetylation of H3K9 [Mak et al. 2004]. As the embryo develops to the blastocyst stage, the H3K27me3 signal remains longer than the signal of its effector but disappears by the late blastocyst stage. Eed co-localizes with macroH2A in trophectoderm cells at the blastocyst stage but is lacking from most of the ICM cells [Costanzi et al. 2000; Erhardt et al. 2003]. MacroH2A accumulation begins around the 8- and 16-cell stage on the Xi [Costanzi et al. 2000].

Located in the proximal part of the Xist RNA is an A-repeat which has been shown to be required for the silencing function of this ncRNA but not its coating ability [Wutz et al. 2002]. However, recent data reveal that the repeat is necessary for both the coating and silencing mechanisms of Xist [Hoki et al. 2009]. The difference in conclusions may be the result of differences in the deletions of the A-repeat between the two laboratories. However, what remains apparent is that the repeat is necessary for Xist's silencing action. Recent elegant work from Zhao et al. [2008] identified by RNA immunoprecipitation that the A repeat interacts with Ezh2 and is necessary for its binding. Ezh2 also binds to the Tsix antisense transcript and they speculated that Tsix could block Xic by titrating away PRC2. In fact, competitive studies show this could be a potential mechanism for Tsix inhibition of Xist's coating ability. In the early embryo (8-cell to blastocyst), a mutation of the A-repeat prevents coating by Xist which results in failure to undergo XCI [Hoki et al. 2009]. Further, deletion of this repeat caused ectopic activation of Tsix on the Xp at the blastocyst stage.

CONCLUDING REMARKS

Most of the epigenetic work done to date using oocytes and preimplantation embryos has dealt with the immunoreactive localization of specific modifications. It is recognized that a crosstalk exists between DNA methylation and post-translational modifications of histones and that histone modifications combine their message to impart active or repressive signals to genes. The ability to do whole epigenome studies in mammalian oocytes and preimplantation embryos is not very feasible due to their paucity of cells. However, as newer technologies emerge and the sensitivity of procedures improves it will be exciting to learn how the epigenetic machinery as a whole interprets genetic information that directs such dramatic functional changes in these models.

ACKNOWLEDGMENTS

The author apologizes for any relevant research that was omitted. Also, the author wishes to thank Mrs. Sarah Huffman for assistance with the preparation of this manuscript.

Declaration of Interest: The author reports no conflicts of interest. The author alone is responsible for the content and writing of the paper.