The human sperm epigenome and its potential role in embryonic development

Abstract

Along with many of the genome-wide transitions in chromatin composition throughout spermatogenesis, epigenetic modifications on histone tails and DNA are continuously modified to ensure stage specific gene expression in the maturing spermatid. Recent findings have suggested that the repertoire of epigenetic modifications in the mature sperm may have a potential role in the developing embryo and alterations in the epigenetic profile have been associated with infertility. These changes include DNA demethylation and the retention of modified histones at important developmental, signaling and micro-RNA genes, which resemble the epigenetic state of an embryonic stem cell. This review assesses the significance of epigenetic changes during spermatogenesis, and provides insight on recent associations made between altered epigenetic profiles in the mature sperm and its relationship to infertility.

methylation, histone, in vitro fertilization, embryogenesis, epigenetic

Introduction

In vitro fertilization (IVF) and other assisted reproductive technologies (ART) have accounted for ∼3 million births since the world's first IVF baby was born in 1978 (Cohen, 1978). Reports examining the long-term health consequences of these babies are limited; however, follow-up studies have reported increased intrauterine growth restriction and lower birth weights in singletons conceived by IVF compared with natural conceptions (Steel and Sutcliffe, 2009). More recently, increased perinatal mortality, congenital anomalies and epigenetic abnormalities have been reported to be associated with IVF (Seif et al., 2006; Allen et al., 2008; Kalra and Molinaro, 2008; Reefhuis et al., 2009; Steel and Sutcliffe, 2009).

The underlying causes of increased anomalies in IVF offspring are unknown, but alterations in the normal epigenetic state of gametes of severely infertile patients undergoing IVF has been proposed as one potential contributor (Cutfield et al., 2007; Lim et al., 2009; Manipalviratn et al., 2009). Epigenetic modifications are covalent modifications present on either the DNA itself or to the proteins that are closely associated with DNA (histones in somatic cells and histones and protamines in sperm), both of which are important in modifying gene expression without changing the genetic code itself. These modifications comprise what is commonly referred to as the epigenome, which in somatic cells regulates cellular fate and function (Bernstein et al., 2002, 2005, 2006; Li, 2002; Jones and Baylin, 2007). It is now well understood that the epigenome can become disrupted or altered, which may contribute significantly to the onset of epigenetic changes observed in many diseases and may be causative of some diseases (Feinberg, 2007; Jones and Baylin, 2007).

Recent studies have demonstrated that sperm have unique and potentially important epigenetic modifications. This brief review describes chromatin and epigenetic changes throughout spermatogenesis, their potential role in normal embryonic development, and their implications in male infertility.

Histone Modifications During Spermatogenesis

Male germ cells undergo unique and extensive chromatin and epigenetic remodeling soon after their specification (determination to become a spermatocyte) and during the differentiation process to become a mature spermatozoon (Seki et al., 2005). Although the mechanisms regulating and orchestrating specification and spermiogenesis remain poorly understood, some progress has been made in elucidating the changes associated with the complex cellular changes. During mitosis and meiosis, male germ cell DNA is packaged in nucleosomes, comprised of histone 2A (H2A), histone 2B (H2B), histone 3 (H3) and histone 4 (H4), all of which are susceptible to covalent modifications, such as methylation, acetylation, ubiquitination and phosphorylation. Each of these chemical modifications to histones works alone or in concert to influence gene repression and/or activation (Fig. 1).

Figure 1

Chromatin modifications determine gene state. Histone modifications promote either gene activation or repression; however, in embryonic stem cells and sperm a subset of genes are commonly associated with both active and inactive marks. ac = acetylation, me = methylation, ub = ubiquitination.

Histone methylation on lysine (K) residues of H3 or H4 can promote gene activation and/or repression (Lachner and Jenuwein, 2002; Suganuma and Workman, 2008). Monomethylation, dimethylation and trimethylation modifications of H3K4, H3K9 or H3K27 display tightly controlled temporal expression and ensure proper progression through spermatogenesis (Khalil et al., 2004; Godmann et al., 2007; Payne and Braun, 2006). The level of H3K4 methylation peaks in the spermatogonial stem cell stage (Fig. 2), and a targeted loss of H3K4 methylation, caused by reduction of Mll2 activity (an H3K4 methyl transferase), results in a dramatic reduction in the number of spermatocytes (Table I), suggesting that H3K4 methylation is essential for the exit from the stem cell stage and commitment to become a spermatocyte (Glaser et al., 2009). In contrast, H3K9 methylation and H3K27methylation are low in the stem cell and increase during meiosis (Fig. 2), persisting long after meiosis is complete, presumably to ensure gene-silencing (Payne and Braun, 2006). The role ascribed to each of these modifications has been primarily characterized by immunofluorescence data with no gene specific localization. Methylation on lysine 9 of histone H3 (H3K9me) is associated with the sex chromosomes, euchromatin and heterochromatin in the late pachytene stage, however, the levels of H3K9 methylation drop upon completion of meiosis. This reduction in H3K9me is concurrently associated with an increase in H3K4me levels (Fig. 2) (Glaser et al., 2009).

Figure 2

Hypothetical expression profiles of histone modifications and DNA methyltransferases during spermatogenesis. In the top panel, solid lines indicate investigated histone level. In the bottom panel dashed lines indicate low expression. Ac = acetylation, me = methylation, TPs = transition proteins, Ax = phosphorylation, MSC1 = meiotic sex chromosome inactivation, DNMT = DNA methyl transferase, RS = round spermatid, ES = elongated spermatid, M = mature spermatid.

Table I

A Summary of chromatin modifiers and their associated functions

Chromatin modifier	Function
DNMT1	Maintenance DNA methyltransferase
DNMT3a	de-novo DNA methyltransferase
DNMT3b	de-novo DNA methyltransferase
DNMT3L	de-novo DNA methyltransferase (no catalytic activity)
Mll	H3K4 methyl transferase
JHDM2A	H3K9 demethylase
HAT	Histone acetyl transferase
HDACs	Histone deacetylase
LSD1/KDM	H3K4 demethylase

The timing of establishment and removal of methylation marks is critical to normal spermatogenesis, as demonstrated by numerous transgenic animal models. Loss of LSD1/KDM1 (H3K4me demethylase) during mid to late meiosis in Caenorhabditis elegans results in germ cell apoptosis and progressive sterility that is maintained through many generations (Shi et al., 2004; Lee et al., 2005; Katz et al., 2009) (Table I). Similarly, removal of H3K9me at the end of meiosis is essential for the completion of spermatogenesis (Okada et al., 2007). Targeted disruption of the H3K9 demethylase JHDM2A (JmjC domain containing histone demethylase 2A, also known as JMJD1A) (Table I) results in complete loss of protamine 1 (PRM1) and transition protein 1 (TNP1) expression, defective chromatin condensation, and infertility (Okada et al., 2007). These studies show that methylation acts through various mechanisms to guide spermatogenesis (Table I).

Histone acetylation of lysine residues is dynamically regulated by histone acetyltransferases (HATs) and histone deacetylases (HDACs), both of which are shown to be essential for spermatogenesis (Christensen et al., 1984; Grimes and Henderson, 1984; Hazzouri et al., 2000; Lahn et al., 2002; Sonnack et al., 2002; Fenic et al., 2004, 2008; An, 2007, Kurtz et al., 2007). Histone acetylation relaxes chromatin and promotes polymerase II (Pol II) gene transcription, whereas deacetylation is associated with gene silencing (Jenuwein and Allis, 2001). Acetylation levels on both H3 and H4 are high in the stem cell phase and are completely removed during meiosis (Fig. 2). Re-acetylation of H4 happens in the elongating spermatid and is known to be a prerequisite for the histone-to-protamine exchange process (Fig. 2) (Hazzouri et al., 2000). However, recent studies show that both H3 and H4 acetylation occur simultaneously in the elongating stage, possibly extending to H2A and H2B acetylation as well (Nair et al., 2008). These data raise the question whether all histones may need to be acetylated to ensure proper histone to protamine exchange in maturing sperm. Enzymes involved in H4 hyperacetylation in the round spermatid are unknown, however, two candidates have emerged, testis specific chromodomain protein (CDY) and HAT (monocytic leukemia) 4 (MYST4). Both of these acetyltransferases are expressed during the maturing spermatid stage, localize to the nucleus, and have been shown to have potent H4 acetylase activity (Lahn et al., 2002; McGraw et al., 2007). H3 acetylation in the elongating spermatid was shown to be Pygopus 2 (Pygo 2) dependent. Pygo 2 has an evolutionarily conserved plant homeodomain (PHD) finger domain that binds trimethylated H3K4 and facilitates H3 acetylation (Nair et al., 2008). Evidence for H2B acetylation has been recently described by mass spectrometery studies, however, very little is known about its function during spermatogenesis or the enzymes required for its acetylation (Lu et al., 2009).

Although acetylation is broad, studies using HDAC inhibitors have demonstrated that the acetylation process is also necessarily specific. The effect of HDAC inhibitors on spermatogenesis is poorly defined, but a few studies have shown that although treating mice with HDAC inhibitors did not result in hyperacetylation, it did cause severe infertility (Fenic et al., 2004). Trichostatin A (TSA) treated animals had no evidence of H4 hypercetylation in the round spermatid, but the number of spermatids was significantly reduced (Fenic et al., 2004, 2008). The inability to detect the hyperacetylation following TSA treatment maybe due to an increase in apoptosis in cells with abnormal acetylation levels, or due to a compensatory mechanism involving alternative HDACs that are insensitive to TSA (Pivot-Pajot et al., 2003).

Histone phosphorylation occurs at serine residues of all core histones and is generally associated with gene activation (Berger, 2002). However, H2Ax phosphorylation (also known as γH2Ax) in germ cells confers the formation of X/Y sex body during spermatogenesis and is a marker for telomere clustering and double stranded breaks (Fernandez-Capetillo et al., 2003a, b). H2Ax phosphorylation is dependent on the ataxia telangiectasia DNA repair and Rad3 related protein ATR, and on the tumor suppressor BRCA1. Together γH2Ax, ATR and BRCA1 initiate meiotic sex chromosome inactivation (MSCI), but to maintain MSCI throughout the pachytene stage there are many other epigenetic modifications including ubiquinated H2A that are localized to the XY body, however, the exact function each performs are unknown (Hoyer-Fender, 2003).

The effect of ubiquination varies depending on the core histone modified: ubiquination of H2A associates with transcriptional repression, whereas, mono-ubiquination of H2B is linked to transcriptional activation in sperm (Baarends et al., 2005; Zhu et al., 2005). In male germ cells, recruitment of ubiquinated H2A to the sex body and telomeres occurs long after γH2Ax incorporation (Fig. 2), which indicates that H2A-ubiquination may be involved in maintaining silencing in the inactive chromatin, but not establishing MSCI.

These brief descriptions of histone modifications during spermatogenesis (data summarized in Fig. 2) demonstrate the varied ways in which epigenetic modification regulate spermatogenesis. Although most histones are replaced with protamines during the elongating spermatid stage, some of the modified nucleosomes escape the histone to protamine transition and as a result are retained in mature sperm, suggesting that these retained nucleosomes may also play a role in the paternal contribution to the embryo.

The Role of Paternal Histones in the Epigenetic Control of Embryogenesis

A hallmark of spermiogenesis is the widespread changes in chromatin structure during spermiogenesis, including the exchange of most canonical histones for protamines (Ward and Coffey, 1991). Protamines are small basic proteins that bind DNA to form toroids; tightly packed structures which compact the genome beyond what is attainable by nucleosomes. The high-level of compaction is an essential attribute for genome transport in the mature sperm head (Balhorn et al., 2000). The histone to protamine exchange process is incomplete, with a small percentage (5–15%) of the genome bound to nucleosomes (Tanphaichitr et al., 1978; Wykes and Krawetz, 2003). The replacement of somatic histones by protamines is important for nuclear chromatin compaction, sperm maturation and fertility (Gatewood et al., 1987, 1990; Balhorn et al., 1988; de Yebra et al., 1998; Corzett et al., 2002; Aoki et al., 2004, 2005, 2006a, b, c; Hammoud et al., 2009a, b). In humans the relative proportion of protamine-1 (P1) to protamine-2 (P2) is strictly regulated at approximately a 1:1 ratio and alterations in the P1/P2 ratio are very rare in fertile men and relatively common in infertile men (Chevaillier et al., 1987; Balhorn et al., 1988, 1999; Belokopytova et al., 1993; de Yebra et al., 1998; Carrell and Liu, 2001; Corzett et al., 2002; Aoki et al., 2005, Oliva, 2006; Carrell et al., 2007). Reports from many labs have shown that changes in the P1/P2 ratio are not only associated with altered sperm quality, but also associated with decreased embryo quality and IVF outcome compared with infertile patients with a normal P1/P2 ratio (Aoki et al., 2006a, b, c; Depa-Martynow et al., 2007). These preliminary associations suggest that protamines and histones may have a greater role during the preimplantation embryo development than previously expected.

The retained nucleosomes are comprised of canonical histones (H2A, H2B, H3, H4) as well as a testes-specific histone variant (tH2B) (Gatewood et al., 1990; Kimmins and Sassone-Corsi, 2005). Until recently, the role for the retained nucleosomes was unknown, but it was speculated that the retained nucleosomes were either remnants of incomplete histone to protamine replacement, or that they may have a biologically significant role during early embryogenesis. In recent work from our laboratory, we have shown that nucleosomes retained in sperm are not simply randomly distributed remnants of inefficient protamine replacement, but are instead significantly enriched at many loci important for embryo development, including genes of key embryonic transcription factors and signaling pathway proteins. Histones were also significantly enriched at the promoters of miRNAs and imprinted genes (addressed in more detail below) (Hammoud et al., 2009a, b). These findings challenge the widely assumed notion that the paternal genome provides little in epigenetic contributions beyond a small set of paternally imprinted genes and a modest repertoire of packaged RNA, due to the repackaging of the vast majority of the genome by protamine.

The identification of retained nucleosomes at key developmental genes was striking, but to have any potential paternal contribution to the developing embryo secondary modifications on the retained nucleosomes (such as the modification discussed above) may be key to differentiate the paternally poised genes from all other genes that have acquired acetylated histones following protamine displacement soon after fertilization. To better understand the chromatin landscape at histone-associated developmental, signaling and miRNA genes three chromatin attributes were tested: histone variants, histone modifications and DNA methylation. We hypothesized that consistent, orderly and biologically relevant patterns in histone modifications, variants and/or DNA methylation could imply a programmatic marking, such as ‘poising’ genes for activation during early embryogenesis, as opposed to a random and non-biologically relevant inefficiency in protamine replacement and epigenetic marking.

The first logical candidate tested for paternal genome poising was the testis specific H2B variant (tH2B) which is incorporated late in spermatogenesis and comprises a large percentage of retained histones (Hammoud et al., 2009a, b). Analysis of tH2B distribution throughout the sperm genome revealed a very significant enrichment of this histone variant at genes for ion channels and genes involved in spermiogenesis, but not at promoters for developmental genes. A second variant in spermatogenesis that has been recently implicated in gene poising in other cell types, such as embryonic stem (ES) cells, is the histone variant H2Az. H2Az in ES cells were shown to be a key regulator of chromatin function and associated with targets of the Polycomb complex at genes essential for ES cell differentiation (Creyghton et al., 2008). In contrast to its role in ES cells, H2Az in sperm was enriched at pericentric heterochromatin, which is consistent with prior immunostaining studies (Rangasamy et al., 2003). Data from these two variants suggested that canonical histones might be an alternative at developmental genes.

Chromatin immunoprecipitation (ChIP) followed by either microarray or deep sequencing analysis clearly shows that modified canonical nucleosomes reveal attributes of a remnant spermatogenesis program (primarily enriched with H3K4me3), as well as a future developmental program (enriched with H3K27me3, H3K4me2 and H3K4me3) (Hammoud et al., 2009a, b). GO term analyses for H3K4me3 yielded genes important for changing nuclear architecture, RNA metabolism, spermatogenesis and a selected number of transcription factors important for embryonic development. However, the majority of the developmental and signaling transcription factors were significantly enriched with H3K27me3 and H3K4me2. Interestingly, many of the developmental promoters harboring an activation mark such as H3K4me3 also retained a silencing mark H3K27me3 (Bernstein et al., 2002, 2006), resembling the bivalently marked (H3K4me3 and H3K27me3) promoters commonly seen in ES cells that are typically silenced prior to ES cell differentiation, but necessary for embryonic differentiation (Fig. 1). Furthermore, modified canonical nucleosomes extended beyond known gene promoters to promoters of non-coding RNAs, miRNAs and imprinted loci. No clear pattern was seen at many of the miRNA and non-coding RNA promoters, but this is primarily due to the limited knowledge on the functional role of these miRNAs in development. Taken together, these data reveal extensive histone modification patterns, and significant similarities to patterns observed in ES cells that may indicate a significant role for sperm epigenetic marking in the establishment of embryonic totipotency.

Significance of DNA methylation in the paternal germline

Packaging and transcriptional control of DNA in eukaryotes is by in large governed by the highly conserved role of histones, however, in higher organisms DNA methylation has been shown to have an essential role in normal embryonic development, regulating gene expression, X chromosome inactivation, genomic imprinting and silencing of endogenous retroviruses (Jaenisch and Jahner, 1984; Surani, 1998; Ng and Bird, 1999). DNA methylation occurs primarily at cytosine residues in a CpG context and is catalyzed by two important classes of DNA methyltransferases (DNMTs): maintenance DNA methyltaransferase (Dnmt1) and de novo DNA methyltransferases (Dnmt3a, Dnmt3b, Dnmt3l) (Fig. 2, Table I) (Eden and Cedar, 1994).

Dnmt1 is the major methyltransferase in somatic cells. It has a preference for hemimethylated DNA and is critical for the maintenance of DNA methylation patterns in replicating cells (Bestor, 1992; Lei et al., 1996) (Table I). Mice homozygous for a targeted partial deletion of Dnmt1 or complete loss of function of Dnmt1 have retarded growth and die by mid-gestation (Li et al., 1992, 1996). Dnmt1 deficient embryos have less than 5% of the normal levels of cytosine methylation, regain biallelic expression at imprinted loci and have ectopic expression of Xist and retrotransposons (Li et al., 1993; Panning and Jaenisch, 1996; Walsh et al., 1998; Goll and Bestor, 2005).

The second class of DNMTs are de novo methyltransferases (Dnmt3a, 3b, 3l) and are essential for establishing new DNA methylation patterns during development (Table I). Embryos lacking Dnmt3a or 3b lost all de novo methylation capabilities in ES cells and early embryos (resulted in embryonic lethality), but had no effect on the maintenance of imprinted loci (Hsieh, 1999; Lyko et al., 1999; Okano et al., 1999). Dnmt3a and b were required for methylating centromeric loci and imprinted genes. Dnmt3l is closely related in sequence to DNMT3A and B, but lacks the catalytic domain. Dnmt3l mediates de novo methylation by stimulating the catalytic activity of DNMT3A2, an isoform variant of DNMT3A (Hata et al., 2002; Suetake et al., 2004). Dnmt3A2 along with its cofactor Dnmt3l establishes locus-specific DNA methylation of paternal imprints prior to meiosis in spermatogenesis (Bourc'his et al., 2001; Kaneda et al., 2004). Males’ haploinsufficient for Dnmt3l are phenotypically normal and fertile, but have subtle changes in methylation and chromatin state of the genome in pre-meiotic spermatogonia (Bourc'his and Bestor, 2004; Webster et al., 2005). The phenotype for DNMT3L null male germ cells is significantly different in male versus female germ cells. Male germ cells fail to methylate LINE-1 (long interspersed elements) and IAP (intracisternal A particles) classes of retrotransposons, have severe asynapsis at meiotic prophase, and undergo apoptosis of all germ cells before pachytene (Bourc'his and Bestor, 2004). Whereas, methylation patterns at the small number of paternally methylated DMRs are almost normal, suggesting some functional redundancy between DNMTs (Bourc'his and Bestor, 2004; Webster et al., 2005). Strikingly, in the oocyte DNMT3L deficient germ cells carried out normal meiosis and methylation at repeat sequences, but the obvious methylation defect was limited to maternally imprinted loci. The differences observed between male and female knockouts are intriguing and raise many questions regarding targeting mechanism and differential regulation between sexes.

In addition to the DNMT gene knockout or happloinsufficiency studies used to assess the functional significance of DNA methylation in the germ line and early embryo, animals treated with DNMT inhibitors such as 5-aza-2'-deoxyCytidine showed altered gene expression patterns and loss of methylation in the germline. The severity of the phenotype observed in the treatment group was heavily dependent on the duration of treatment: short-term exposure in mice and rats decreased fertility (Seifertova et al., 1976; Raman and Narayan, 1995),whereas, prolonged treatment of 5-azacytidine (11weeks) in male rats resulted in a dose-dependent reductions in testis, epididymal weights and sperm counts, increase in germ cell apoptosis, and a significant increase in preimplantation loss (Doerksen and Trasler, 1996; Doerksen et al., 2000). These findings suggest that sperm DNA methylation plays a critical role in the differentiation of spermatogonia and early embryo viability, however, this appears counterintuitive since DNA methylation patterns are erased and reestablished immediately after fertilization and once again when primordial germ cells (PGCs) reach the genital ridge (Reik et al., 2001; Hajkova et al., 2002, Seki et al., 2005, 2007).

In humans, the relationship between bulk methylation levels with respect to IVF outcome was recently examined in one study. No significant correlation was made between bulk DNA methylation levels and the fertilization rate or embryo quality, but a lower 5-methyl cytosine signal (<555 AU) intensity correlated with a lower pregnancy (8.3 versus 33.3%) rates (Benchaib et al., 2005). These preliminary associations in humans are interesting, but interpretation and implications of measuring bulk methylation levels are limited and provide very little understanding of causality or the programs perturbed (activation of retrotransposons, changes at imprinted genes, etc.), and more genome-wide approaches are needed.

Sperm DNA Methylation Profiles and its Role in the Paternal Germline

Germ cells undergo extensive epigenetic reprogramming during proliferation and migration to the genital ridge (Seki et al., 2005, 2007). DNA methylation is erased and re-established in a sex and sequence specific manner during gametogenesis (Reik et al., 2001; Hajkova et al., 2002). The timing for methylation reestablishment differs between sexes and is continual for certain gene classes (Trasler, 2006). In males, de novo genomic methylation begins prenatally (prospermatogonia) at imprinted loci and repetitive elements with a general consensus being, that methylation patterns are completed by the end of pachytene stage of meiosis (Oakes et al., 2007a, b). However, exceptions have been reported at a few gene promoters that are expressed early in spermatogenesis (Pgk-2, ApoA1 and Oct-3/4) but are silenced in the maturing spermatid by gradually acquiring methylation in postmeiotic spermatocytes (Ariel et al., 1991, 1994).

Recent genome-wide methylation studies have indicated that the sperm epigenome differs markedly from that of somatic cells, but is very similar to ES cells and embryonic germ cells (EG) (Eckhardt et al., 2006, Oakes et al., 2007a, b; Weber et al., 2007; Farthing et al., 2008). Examining sperm DNA methylation of chromosomes 6, 20 and 22 using restriction landmark genomic studies showed that many loci were differentially methylated between sperm and somatic cells. Extending methylation analysis genome-wide to all promoters of the human genome showed that the promoters that were differentially methylated in sperm and somatic tissue (or acquire methylation upon differentiation) were promoters with ‘weak’ CpG islands. More recently, we have shown in our gene ontology analysis that the hypomethylated promoters in the mature sperm are the promoters of developmental transcription and signaling factors. Interestingly, the DNA hypomethylated promoters in mature sperm greatly overlapped with the developmental promoters bound by the self renewal network transcription factors in human ES cells (e.g. OCT4, SOX2, NANOG, KLF4 and FOXD3 proteins) (Boyer et al., 2005). In ES cells, these pluripotency proteins promote self renewal and also work with repressive polycomb complexes (PRC2) to help repress a large set of developmental regulators (including HOX genes) to prevent differentiation (Cao et al., 2002; Bernstein et al., 2006; Lee et al., 2006; Muller and Kassis, 2006; Schwartz et al., 2006; Takahashi et al., 2007; Tanay et al., 2007; Wernig et al., 2007; Kopp et al., 2008). This overlap suggested that pluripotency or polycomb complex factors might be involved in the establishment and maintenance of sperm DNA methylation poising in sperm. Unfortunately, pluripotency and polycomb factors were not detected in the mature sperm, in fact many of the pluripotency promoters including several key members of the self-renewal network (OCT4, NANOG, FOXD3) themselves acquire methylation throughout spermatogenesis, whereas their developmental target genes remain hypomethylated, consistent with recent studies in mice (Down et al., 2008; Farthing et al., 2008; Illingworth et al., 2008; Mohn et al., 2008). These findings show that genes encoding early developmental transcription factors as well as signaling proteins are DNA hypomethylated and histone bound. Furthermore, developmental promoters are selectively methylated during development, which may help commit cells to differentiation decisions. Histone retention and DNA demethylation contribute to a poised state that ensures transcriptional competence and activation of developmental regulators in the early embryo.

Epigenetic alterations and male infertility

The incidence of infertility has been rising, currently affecting one in every seven couples in the western population. Male infertility is responsible for roughly half of the cases of infertility. The underlying cause of male infertility is unexplained in ∼50% infertile men, and genetic causes have been proposed to be likely (Carrell, 2008a, b; Matzuk and Lamb, 2008). Several studies have explored possible genetic causes using mouse models (O'Bryan and de Kretser, 2006), candidate gene sequencing (Miyamoto et al., 2003; Aoki et al., 2006a, b, c; Hammoud et al., 2007, 2009a, b), and recently the first genome-wide association study in oligozoospermic and azoospermic men has been completed (Aston and Carrell, 2009). However, these studies have revealed that single gene polymorphisms are not likely to be the cause of most cases of male infertility, but male infertility is likely to be a multifactorial disease. Similar to other complex diseases, such as cancer, epigenetic alterations may be a component contributing to infertility.

As described above, recent studies have shown that epigenetic modifications in sperm (both histone modifications and DNA methylation) appear to poise the paternal genome to participate in early embryogenesis. Additionally, several studies indicate that DNA methylation is altered, in at least imprinted genes, oligozoospermic men and men with improper histone to protamine replacement (Marques et al., 2004, 2008; Bowdin et al., 2007; Doornbos et al., 2007; Hammoud, in press; Kobayashi et al., 2007, 2009). These observations beg the question of whether methylation defects, of both imprinted and non-imprinted genes, as well as other epigenetic defects (such as histone localization or modifications in the mature sperm), may play an important role in the development and growth of ART offspring (Manipalviratn et al., 2009). If epigenetic profiles of the mature sperm are critical, then alterations in epigenetic patterns in infertile males can provide a logic for the increased risk for preterm birth, low birthweight, congenital anomalies, perinatal mortality, and several other pregnancy-related complications seen at a higher frequency in babies conceived by IVF (Hansen et al., 2002; Kalra and Molinaro, 2008). This growing field of epigenetics in early gametes and embryos may be of benefit in understanding such observations (Carrell, 2008a, b). Current studies in our laboratory are focusing on genome-wide changes in histone localization as well as DNA methylation (discussed below) in male partners of recurrent pregnancy loss patients, repeated failed IVF patients and infertile males with an altered histone to protamine ratio.

Current findings showing that epigenetics patterns in germline are extensive and of potential significance, only strengthens further the previous associations that showed abnormal methylation of imprinted genes (genes expressed in a parent-of-origin manner), in the gametes of some infertile men or babies conceived by IVF (DeBaun et al., 2003; Gosden et al., 2003; Maher, 2005). A gain or loss of expression of imprinted genes has been implicated in many diseases (Jaenisch and Bird, 2003; Seitz et al., 2004; Morgan et al., 2005; Royo et al., 2006) including Beckwith–Weidemann syndrome (BWS) and Angelman's Syndrome (AS), both of which have been significantly correlated with IVF babies (DeBaun et al., 2003; Gosden et al., 2003; Maher, 2005).

Whether the increased incidence of imprinting abnormalities in IVF babies arises from in vitro manipulations of embryos or gametes, are due to ovulation induction medications, or are inherited from the gametes of infertile patients is unclear (Cummins and Jequier, 1994; de Kretser, 1995; Edwards and Ludwig, 2003; Marques et al., 2004, 2008; Bowdin et al., 2007, Doornbos et al., 2007, Kobayashi et al., 2007, 2009; Laprise, 2009). Support for all former hypotheses have been provided but this review will focus on one aspect, preexistent methylation alterations in the gametes of infertility patients (Cummins and Jequier, 1994; de Kretser, 1995; Edwards and Ludwig, 2003; Marques et al., 2004, 2008; Bowdin et al., 2007, Doornbos et al., 2007, Kobayashi et al., 2007, 2009; Laprise, 2009). This finding first surfaced a few years ago and showed that DNA methylation patterns at paternally imprinted loci are altered in the sperm of severely oligozoospermic patients (Marques et al., 2004, 2008). Subsequently in 2007, Houshdaran et al. reported that a broader alteration in DNA methylation in sperm is seen at a handful of imprinted loci, CpG islands upstream of gene promoters, and a few repetitive elements in infertile patients with poor semen parameters (Houshdaran et al., 2007). Furthermore, our lab has shown that methylation alterations extend beyond severely oligozoospermic patients to patients with relatively normal sperm counts but with abnormal chromatin packaging, defined by an altered P1/P2 ratio (Hammoud, in press). Interestingly, in the small number of patients tested with either oligozoospermia or abnormal protamine expression we observed that methylation alterations varied between the two different etiologies of infertility. For instance oligozoospermic patients were hypermethylated at MEST, an imprinted gene associated with Silver Russell Syndrome (SRS), whereas abnormal protamine patients had significant changes at LIT1 and small nuclear ribonucleoprotein polypeptide N (SNRPN), genes that may be associated with cases of transient neonatal diabetes milletus (TNDM) and alternative splicing (AS). These findings suggest that the risk of transmitting epigenetic alterations may vary with the classification of infertility; however, it is important to note that not all patients or alleles were affected to a similar extent. The differences in the degree of methylation within some genes or alleles compared with others raises an important questions for future studies: whether there is a variable risk to the different CpGs and whether abnormal methylation has a threshold level for conferring disease risk in the embryo or is a gradual continuum.

Pre-existent methylation alterations in the gametes of infertile patients pose a risk for transgenerational epigenetic inheritance. Evidence for transgenerational epigenetic inheritance remains controversial in humans (Oswald et al., 2000; Reik et al., 2003; Morgan et al., 2005), however, considerable evidence for transgenerational epigenetic inheritance in mice has become apparent at intracisternal A particles (IAPs) and imprinted genes(Bultman et al., 1992; Morgan et al., 1999; Rakyan et al., 2003). In humans, evidence for germline epigenetic inheritance has come almost exclusively from epidemiological studies. The strongest evidence for germline epigenetic inheritance comes from the work of Horsthemke and colleagues (Buiting et al., 2003) where they have shown that the presence of epimutations and not genetic mutations, at the SNRPN–SNURF upstream reading frame locus was inherited from the paternal grandmother (Buiting et al., 2003). Furthermore, Kagami et al. (2007) showed that defective methylation at the DMR of MEST in sperm may have been inherited by an ART born baby with SRS. More recently, Kobayashi et al. showed abnormal DNA methylation at many imprinted loci in 17 of 78 assisted reproductive technology (ART) embryos (21.8%) tested. Although some of the imprinting errors identified may have risen during the ART process, however, in seven cases hypomethylation at H19 and GTL2 was present both in the sperm and in the embryo, suggesting that abnormal hypomethylation may be paternally inherited (Kobayashi et al., 2009). In summary, human transgenerational epigenetic inheritance is uncertain, but is possible if altered DNA methylation is inefficiently cleared between generations or if methylation pattern are not properly reestablished due to refractory elements such as retrotransposons or certain histone modifications in the embryo and PGCs.

Structural and in vitro data show that certain modified nucleosmes such as H3K4 methylation can deter DNA methylation in mice (Ooi et al., 2007). This interdependency or cooperativity relationship can have potential implications in reprogramming, especially at imprinted loci. Recent studies have shown that maternally and paternally imprinted alleles retain differential histone modifications (methylation and acetylation) to promote either allele activation or repression in somatic cells (Fournier et al., 2002; Delaval and Feil, 2004; Delaval et al., 2007). Consistent with the findings in somatic cells, imprinted genes (imprints are established in the gametes) in human sperm used similar poising mechanisms: H3K4me3 associated with many of the paternally expressed DMRs (Hammoud et al., 2009a, b), whereas, maternally imprinted (paternally repressed) loci lacked H3K4me3 and had moderate levels of H3K9me3, a repressive chromatin signature, residing at a few tested loci by qPCR. The presence of modified nucleosomes in the germline may serve as an epigenetic cellular memory to help reestablish and maintain parent of origin identity. However, in the cases of male infertility with altered histone retention this may be problematic at the time of reprogramming if the retained nucleosomes in the mature sperm are improperly placed or modified. Whether this is one of the underlying factors that contributes to the poor embryo outcome in patients with an abnormal histone to protamine ratio is unknown.

Conclusion and future directions

Sperm chromatin state is highly dynamic and retains important chromatin attributes that help facilitate the proper progression of spermatogenesis as well as being a potential contributor to early developmental processes. This continual epigenetic remodeling state may make sperm cells susceptible to impediments of environmental factors, aging process, or diseases such as infertility, but the ramifications of the altered chromatin states in the germ-line are not entirely known. Future studies are needed to establish perdurance of paternally retained modified nucleosomes in the early embryo, and their potential effects if abnormally retained.