Epigenetic reprogramming in mouse germ cellsPetra Hajkova1.1.1.1.2.1.3.1.4.1.5.1.6.1.6.1.1.6.2.1.6.3.1.6.4.1.6.5.665#12401.6.6.1.6.7.1.6.8.1.7.1.7.1.1.7.2.1.7.3.1.8.2.2.1.2.1.1.2.1.2.2.1.3.2.1.4.2.1.5.2.1.5.1.2.1.5.2.2.1.5.3.2.1.5.4.2.1.5.5.2.1.5.6.2.1.5.7.2.1.5.8.2.1.6.2.1.6.1.2.1.6.2.2.1.6.3.2.1.6.4.2.2.2.2.1.2.2.1.1.2.2.1.2.2.2.2.2.2.3.2.2.4.2.2.5.2.2.5.1.2.2.5.2.2.2.6.2.2.7.2.2.8.2.2.9.2.2.10.2.2.11.2.2.12.3.3.1.3.2.3.3.3.3.1.3.3.2.3.3.3.3.3.4.3.3.5.3.3.5.1.3.3.5.2.3.4.3.4.1.3.4.2.3.5.3.6.3.7.3.8.3.9.4.4.1.4.2.4.2.1.4.2.2.4.2.3.4.2.4.4.3.4.4.4.5.4.6.4.7.4.8.4.9.4.10.4.11.4.12.ErklärungAcknowledgmentReferencesCURRICULUM VITAEenTable of contentsHelpIntroduction
Brief history of DNA methylation

The presence of the so-called “fifth base” in the DNA of eukaryotes (5- methylcytosine, 5-mC) was revealed already before the final proof that DNA constitutes the real carrier of genetic information. In 1948, while trying to detect amino acid contamination in nucleic acid samples, Hotchkiss found 5-mC by paper chromatographic method (Hotchkiss (1948)). The result was shortly afterwards confirmed by Wyatt including the quantification of 5-mC contribution to the genome (Wyatt (1951)). In 1959 Kornberg suggested that 5-mC may be added onto DNA by post-replicative mechanism, implying for the first time the use of this base as a potential carrier of epigenetic information (Kornberg, et al. (1959)). This hypothesis took; however, another nine years to be demonstrated experimentally (Billen (1968); Lark (1968)).

Meanwhile, the first prokaryotic methyltransferases and restriction endonucleases had been identified and the role of DNA modification (5-mC and 6-mA) connected to the concept of bacterial restriction/modification genome defense system.

The investigation of the eukaryotic DNA methylation proved to be far more difficult. The 5-mC content of DNA had been measured by different chromatographic methods (Sneider (1972); Singer, et al. (1977); Culp, et al. (1970); Silber, et al. (1966)) and mass spectroscopy (Gautier, et al. (1977)). However, only further progress in recombinant DNA technology together with characterization of methylation sensitive restriction endonucleases permitted not only quantitative analysis of 5-mC content but also more importantly, analysis of its spatial distribution. Bird and Southern (Bird and Southern (1978)) were the first to recognize the potential of these enzymes in combination with the Southern blotting technique to assess the methylation status of defined sites within specific gene regions. In the following decade, number of experiments employed this technique to connect the gene methylation with transcriptional silencing (reviewed in Ehrlich and Wang (1981); Razin and Riggs (1980); Felsenfeld and McGhee (1982)). Based on those results the two main models connecting methylation with gene expression have been postulated. The model of (a) direct transcriptional inhibition is based on the existence of transcription factors, which are sensitive to the presence of methylated cytosine(s) in their binding sites (such as AP-2 (Comb and Goodman (1990)), E2F (Kovesdi, et al. (1987)), NF-κB (Bednarik, et al. (1991)). Methylation within the regulatory sequences of a gene may thus prevent initiation of transcription. The model of (b) indirect inhibitionis connected to proteins with the binding specificity for methylated DNA (for example MBD1 – methylated DNA binding protein (Zhang, et al. (1989)), MeCP2 – methyl-CpG binding protein (Meehan, et al. (1989)). Binding of those proteins not only makes the DNA inaccessible to transcription machinery, but as it was described rather recently, the proteins can directly interact with the histone deacetylase complexes (Feng and Zhang (2001); Rountree, et al. (2000); Robertson, et al. (2000); Ng, et al. (1999), Nan, et al. (1998)), thus bringing the chromatin into the inactive shape. The discovery of such protein factors and their interplay brought about focus on the machinery that “reads” and interprets the methylation mark inside the cell.

The growing knowledge concerning the biological significance of DNA methylation intensified the search for enzymatic activities responsible for the epigenetic marking of DNA. In late 1980s, at the time when the prokaryotic methyltransferases have been more or less thoroughly characterized (concerning both the protein structure and enzymatic functions – for review see (Noyer-Weidner and Trautner (1993)), only very little was still known about their eukaryotic counterparts. Although the existence of two distinct methyltransferase activities (the concept of maintenance and de novo methyltransferase activity) had been predicted already in 1975 (Holliday and Pugh (1975); Riggs (1975)); the first eukaryotic DNA methyltransferase (Dnmt1) was cloned from murine cells not earlier than in 1988 (Bestor (1988,Bestor, et al. (1988)). This enzyme has a 5- to 30-fold preference for hemimethylated DNA (Yoder, et al. (1997)) and has therefore been assigned a role limited to the maintenance of methylation patterns (Lyko, et al. (1999)). The residual level of DNA methylation found in Dnmt1 knock-out mouse embryos (Li, et al. (1992)) confirmed the prediction, that this enzyme is not the only factor responsible for DNA methylation. It took, however, more than 10 years to clone methyltransferases (Dnmt3a, Dnmt3b – Okano, et al. (1999)) with de novo methylation functions.

Whereas DNA methylation in prokaryotes is mainly involved in protecting the genome against the degrading nucleases (restriction/modification systems), and thus playing role in the host defense; the eukaryotic DNA methylation has evolutionary gained more complex function. Findings of the last decade show that in many eukaryotes cytosine methylation plays a pivotal role in the control of gene expression and in inactivation of transposable and repetitive elements (this genome protection function resembles the role of DNA methylation in prokaryotes) (reviewed in Yoder, et al. (1997)). Additionally, this epigenetic modification is crucial for embryonic development of mammals regulating genomic imprinting, X inactivation and cell differentiation (Reik, et al. (2001,Reik and Walter (2001); Mlynarczyk and Panning (2000)).

Genomic imprinting and its connection to DNA methylation

The term genomic imprinting mentioned throughout this thesis refers to a differential parent-of-origin dependent monoallelic expression of some genes.

The first indication that the two parental genomes contributing to the zygote are not functionally equivalent came in the early 1980s. The pronuclear transfer experiments performed by Surani and Solter demonstrated that both parental genomes are essential for normal embryonic development (McGrath and Solter (1984); Surani, et al. (1984)). Embryos that contained two paternal genomes (androgenetic embryos) showed very poor embryonic development, whereas gynogenetic embryos (containing two maternal genomes) were deficient in developing extraembryonic tissues. In both instances, lethality occurred by mid-gestation. These results demonstrated that parental genomes play obviously complementary roles, involving differential (monoallelic) expression of essential genes in embryonic development.

In 1985 Cattanach and Kirk published their studies on mouse embryos containing uniparental duplications (uniparental disomies - UPDs) of sub-chromosomal regions (Cattanach and Kirk (1985)). The thorough study showed that some of the duplications resulted in embryonic lethality. Based on these results a chromosomal map was produced showing the regions, of which both parental copies are necessary for normal embryonic development. Ten such domains have been identified, located on mouse chromosomes 2, 6, 7, 11, 12 and 17. The map has been refined since, its current form is shown in Fig. 1.

Since then number of mouse and human genes that are differentially expressed depending on their parental origin (imprinted genes) have been identified (see Table 1). It is noteworthy that all the genes mapped so far are located within the regions depicted by Cattanach and Kirk. It seems to be an important feature of imprinted genes that they appear in clusters sharing probably the main regulatory elements (Paulsen, et al. (1998); Paulsen, et al. (2000); Engemann, et al. (2000)).

Imprinted Loci

Chr.

Chr.

Region

Repressed parental allele;

maternal / paternal

Name

Nnat

2

distal 2

M

neuronatin

Gnas

2

distal 2

P

guanine nucleotide binding protein, alpha stimulating

Gnasxl

2

distal 2

M

guanine nucleotide binding protein, alpha stimulating, extra large

Nesp

2

distal 2

P

neuroendocrine secretory protein

Nespas

2

distal 2

M

neuroendocrine secretory protein antisense

Sgce

6

centromere to
T77H (A3.2)

M

sarcoglycan, epsilon

Peg1/Mest

6

proximal 6
(distal to A3.2)

M

mesoderm specific transcript

Copg2

6

proximal 6
(distal to A3.2)

P

coatomer protein complex subunit gamma

Copg2as

6

proximal 6
(distal to A3.2)

M

antisense to Copg2

Mit1/lb9

6

proximal 6
(distal to A3.2)

M

mest linked imprinted transcript 1

Zim1

7

proximal 7

P

imprinted zinc-finger gene 1

Peg3/Pw1

7

proximal 7

M

paternally expressed gene 3

Usp29

7

proximal 7

M

ubiquitin specific processing protease 29

Zim3

7

proximal 7

P

Zinc Finger Gene 3 from Imprinted domain

Zpf264

7

proximal 7

M

Zinc Finger gene 264

Snrpn

7

central 7

M

small nuclear ribonucleoprotein polypeptide N

Snurf

7

central 7

M

Snrpn upstream reading frame

Pwcr1

7

central 7

M

Prader-Willi chromosome region 1

Magel2

7

central 7

M

Magel2

Ndn

7

central 7

M

necdin

Zfp127/Mkrn3

7

central 7

M

ring zinc-finger encoding gene

Zfp127as/Mkrn3as

7

central 7

M

ring zinc-finger encoding gene antisense

Frat3

7

central 7

M

Frequently rearranged in advanced T-cell lymphomas.

Ipw

7

central 7

M

imprinted in Prader-Willi Syndrome

Ube3a

7

central 7

P

E6-AP ubiquitin protein ligase 3A

Ube3aas

7

central 7

M

Ube3a antisense

Nap1l4

7

central 7

P

H19

7

distal 7

P

H19 fetal liver mRNA

Igf2

7

distal 7

M

insulin-like growth factor 2

Igf2as

7

distal 7

M

insulin-like growth factor 2, antisense

Ins2

7

distal 7

M

insulin 2

Mash2

7

distal 7

P

mammalian achaete-scute homologue 2

Kvlqt1

7

distal 7

P

-

Kvlqt1-as

7

distal 7

M

Kvlqt1 antisense

Tapa1/Cd81

7

distal 7

P

Cd 81 antigen

p57 KIP2 / Cdkn1c

7

distal 7

P

cyclin-dependent kinase inhibitor 1C

Msuit

7

distal 7

P

mouse specific ubiquitously expressed imprinted transcript 1

Slc221l
Note Slc221l was formally known as Impt1, Itm, and Orctl2.

7

distal 7

P

Solute carrier family 22 (organic cation transporter memter-1 like).

Ipl/Tssc3

7

distal 7

P

imprinted in placenta and liver (Tdag51?)

Tssc4

7

distal 7

P

Obph1

7

distal 7

P

oxysterol-binding protein 1

Rasgrf1

9

9

M

Ras protein specific guanine nucleotide-releasing factor 1

Zac1

10

10

M

Zinc finger DNA binding protein

Meg1/Grb10

11

proximal 11 (A1-A4)

P

growth factor receptor bound protein 10

U2af1- rs1

11

proximal 11 (A3.2-4)

M

U2 small nuclear ribonucleoprotein auxiliary factor (U2AF), 35kDa, related sequence 1

Dlk

12

distal 12 (E-F)

M

delta like

Meg3/Gtl2

12

distal 12 (E-F)

P

gene trap locus 2

Htr2a

14

distal 14

P

5-hydroxytryptamine (serotonin) receptor 2 A

Slc22a2

17

proximal 17

P

Membrane spanning transporter protein

Slc22a3

17

proximal 17

P

Membrane spanning transporter protein

Igf2r

17

proximal 17

P

insulin-like growth factor 2 receptor

Igf2ras/Air

17

proximal 17

M

insulin-like growth factor 2 receptor antisense

Impact

18

proximal 18 (A2-B2)

M

Homology with yeast & bacterial protein family YCR59c/yigZ

Ins1

19

19

M

insulin 1
Table 1: Imprinted genes

The following gene has now been shown not to be imprinted.

Mas

17

proximal 17

M

Mas proto-oncogene

Fig. 1: Physical maps of imprinted domains in the mouse chromosomes(according to www.mgu.har.mrc.ac.uk/imprinting)

For a paternal imprint to be established there must be a mechanism (marking) which distinguishes the DNA inherited from mother and father. Such a mark is presumably established in the germ line, when the two parental genomes are separated, is propagated by further post-fertilisation events and has to be erased and re-established again in the germ line (see Fig. 2).Possible candidates for such a mark include DNA methylation, or differences in chromatin structure, which may influence the accessibility of the region to imprinting factors.

Before any endogenous imprinted genes were identified, a number of transgenes were observed to be active only after passage through the germ line of one sex (reviewed in Reik, et al. (1990)). Some of these were studied in detail and were found to carry different methylation patterns depending on their parental origin (Reik, et al. (1987); Sapienza, et al. (1987)). These findings turned out to be real break-through to the field. Although cytosine methylation had long been proposed to fulfil all requirements for the postulated imprint (DNA methylation affects gene expression, is heritable and is reversible), only these discoveries of Reik and Sapienza finally unified the two fields in practice.

The later analysis revealed that all imprinted genes so far identified in mouse and human show regions that are differentially methylated in an allele-specific manner (differentially methylated regions - DMRs). These imprinting control regions are often complex with multiple functions acting to repress genes when methylated, or serving as boundary elements when unmethylated (Bell and Felsenfeld (2000); Hark, et al. (2000)). Some DMRs also function as silencer elements when unmethylated (Constancia, et al. (2000)), a function which is abolished when the DMR is methylated. In other cases, a DMR is associated with the expression of an antisense transcript whose expression in turn ensures repression of the upstream gene (Lyle, et al. (2000)). In all the cases allele-specific methylation of DMRs ensures the monoallelic expression of imprinted genes.

The “life cycle” of imprinting in the mammalian development

The process of parental imprinting involves 4 distinct biological stages (steps) (see Fig. 2). The imprints are established during gametogenesis (establishment), they are maintained throughout the embryogenesis during the time when the rest of the genome undergoes de-methylation (maintenance) and are finally read in the somatic tissues of embryo and adult (reading). In the embryonic germ line, however, the imprints must be erased (erasure) and re-established according to the sex of individual (establishment), thus closing the “life circle” of imprint.

Fig. 2: “Life cycle” of imprinting in mammalian developmentThe figure shows the requirements genomic imprinting has to fulfill: parental specific marks present in gametes (red - maternal marks, blue - paternal marks) are combined in zygote and maintained through the waves of demethylation and de novo methylation to be finally read in the form of monoallelic expression in embryo. During the establishment of the germ line the imprinting marks have to be erased and re-established according to the sex of developing individual (according to Reik and Walter, 2001)

Imprinting and methylation changes during early embryogenesis

One of the basic criteria for the imprinting mark is that it is present in gametes. The differentially methylated regions (DMRs) of most of the imprinted genes carry the parental specific methylation in oocytes and sperms (Olek and Walter (1997); Tremblay, et al. (1997); El-Maarri, et al. (2001); Stoger, et al. (1993)). The methylation mark might be not the only signal involved in recognising the paternal origin, though, as the exceptions to the rule have been identified. For example, the promoter of maternally methylated Snrpn gene was found to be methylated in mouse oocytes, but demethylated in human oocytes, despite of showing maternal methylation and imprinted monoallelic expression in embryonic as well as in somatic tissues of both species (El-Maarri, et al. (2001)).

Just shortly after fertilisation, in the zygote, dramatic epigenetic changes occur. Studies at the level of the whole genome as well as the methylation analysis of unique genes showed that prior to fusion of pronuclei, paternal genome undergoes overall (and most presumably active) demethylation (Oswald, et al. (2000); Mayer, et al. (2000), see Fig. 3). The demodification is probably connected to the remodelling of the paternal pronucleus accompanied by exchange of protamines for histones and commences before the onset of replication. It was well documented that some of the paternal methylation imprints (for example DMR2 of Igf2 gene) do not withstand this demodification event (Oswald, et al. (2000)), whereas others, as for example the upstream DMR of H19 (Warnecke, et al. (1998)) or Ras Grf1 (Shibata, et al. (1998)) seem to resist. Which mechanism keeps the memory of the erased paternal imprints in order to re-establish them later on during embryogenesis has still to be elucidated. Perhaps, only the imprints of secondary importance are erased, whereas the marks at the real imprinting centres of the whole cluster stay.

During the cleavage, following the early zygotic demethylation, the whole genome undergoes passive demethylation (see Fig. 3). The exceptions to the rule are DMRs of the imprinted genes, which are documented to keep their differential methylation (Brandeis, et al. (1993); Olek and Walter (1997)). The pre-implantation demethylation is probably due to the exclusion of the maintenance methyltransferase (Dnmt1) from the nuclei of the early dividing embryo (Mertineit, et al. (1998); Howell, et al. (2001)). However, in mouse eight-cell stage embryos, Dnmt1 is relocated back to the nucleus for just one replication cycle. The presence of this oocyte and early embryo form of Dnmt1 (Dnmt1o – Mertineit, et al. (1998)) in the nucleus at that stage is apparently crucial for the maintenance of imprinted methylation - in a knockout of this Dnmt1 version the methylated allele of imprinted genes loses exactly 50% methylation (Howell, et al. (2001)). It is plausible that other methyltransferases (perhaps Dnmt3a and Dnmt3b) are responsible for methylation of imprinted genes before and after the eight-cell stage when Dnmt1o is excluded from the nucleus into the cytoplasm.

Whereas the majority of CpG sites within the mammalian genome become demethylated prior to implantation, distinct sites connected to the imprinted genes are documented to undergo allele-specific de novo methylation (Oswald, et al. (2000); Brandeis, et al. (1993)). Denovo methylation thus reconstitutes the methylation marks lost by zygotic paternal demethylation (as in the case of Igf2 DMR2 – Oswald, et al. (2000)) or establishes allele specific differential methylation (El-Maarri, et al. (2001)) in regions which possibly carry gametic imprints of other type than methylation.

Around the time of implantation (blastocyst stage) a wave of global de novo methylation occurs (see Fig. 3). Non-imprinted genes as well as repetitive elements were found to gain methylation between 5.5 and 6.5 dpc (Kafri, et al. (1992); Monk, et al. (1987)). The same process is probably responsible also for de novo methylation of retroviruses integrated into the DNA of pre-implantation mouse embryos (Jahner, et al. (1982)). The wave of de novo methylation occurs mainly in the inner cell mass (ICM) cells of the expanded blastocyst and is proposed to be connected with the process of lineage decision and cell differentiation (Monk (1995); Reik, et al. (2001)).

Fig. 3: Global changes of genomic methylation during early embryogenesisShortly after fertilisation the paternal genome (green) undergoes rapid and most probably active demethylation. To the contrary, the maternal genome ( r e d ) is demethylated by a slow passive mechanism during the first cleavages. Around the time of implantation (blastocyst stage) the wave of de novo methylation occurs (to a different extent in embryonic (EM) and extraembryonic (EX) tissues). (Adapted from Reik et al., 2001).

Germ line and its key role in the epigenetic “life cycle”

Whereas the cells of the future soma undergo described de novo methylation and differentiation, at the time when the work of this thesis was started it was still unclear what is the epigenetic origin of the germ lineage.

The germ cell lineage plays certainly a key role in the imprinting “cycle” (see Fig. 2). The germ cells are responsible for epigenetic resetting which not only deletes and re-establishes the imprinting marks according to the sex of the developing individual, but also is crucial to prevent the existing epimutations to be passed onto the next generation.

The origin and the epigenetic status of the germ cells have been extensively discussed over the years. Based on the limited number of experiments showing that the DNA of primordial germ cells (embryonic precursors of the gametes) present in the developing genital ridges of the embryo is hypomethylated (Brandeis, et al. (1993); Kafri, et al. (1992)), two main developmental models have been proposed (see Fig. 4) : (1) The germ cell lineage is derived early in the blastocyst prior to the wave of de novo methylation – i.e. the germ cells keep their undermethylated status to be re-programmed according to the sex later during the gametogenesis (see also Jaenisch (1997)). (2) The germ cells are derived in the blastocyst but undergo the de novo methylation process, the methylation is erased later on by germline specific mechanism and new methylation marks re-established subsequently in the gametogenesis.

Fig. 4: “Epigenetic origin” of primordial germ cells At the onset of this thesis there were two different hypothesis concerning the origin of the hypomethylated state of primordial germ cells : The germ cells could be derived from the blastocyst before the wave of de novo methylation or the germ lineage undergoes epigenetic changes similar to somatic (ev. extraembryonic) tissues, methylation is erased later by germ line specific mechanism.

Biological studies of primordial germ cells – origin, characteristics, developmentOrigin of primordial germ cells

Primordial germ cells (PGCs) are the earliest recognised precursors of gametes. During the early embryogenesis, the primordial germ cells are first detectable by their high level of tissue non-specific alkaline phosphatase activity midway through gastrulation at 7.2dpc. At that time they form a cluster of about 40-50 cells in the extraembryonic mesoderm at the base of the allantois (Ginsburg, et al. (1990); see Fig. 5). Initial experiments with mouse chimeras showed that PGCs are derived from the epiblast (Gardner (1985)). Later on, by following the fate of single epiblast cells injected with a lineage marker, Lawson and Hage (Lawson and Hage (1994)) were able to locate the ancestral population as being among the most proximal epiblast cells, adjacent to the extraembryonic ectoderm. The xenotypic transplantation of epiblast cells revealed that even distal epiblast cells, which normally give rise to neuroectoderm, have the capacity to form PGCs when transplanted to the proximal region (Tam and Zhou (1996)). Additional experiments based on ex vivo cultivation of dissected epiblast cells confirmed the necessity of interaction with the extraembryonic tissues for the PGC formation. In the experimental set up, only the epiblast cells co-cultivated with adjacent extraembryonic tissues were able to form PGCs (Yoshimizu, et al. (2001)). Thus, the location of the cells and their vicinity to the signal produced in the extraembryonic ectoderm is involved in the germ cell determination rather than any segregation of preformed cytoplasmatic determinants (as for example in D. melanogaster). The process of PGC determination is probably connected to the expression of bone morphogenetic protein-4 (Bmp4). The expression of this protein is confined to the extraembryonic extoderm prior to gastrulation; and moreover the Bmp4 homozygous null embryos fail to form PGCs (Lawson, et al. (1999)).

Fig. 5: Origin of primordial germ cellsPGCs are derived form the proximal epiblast; around 7.2 dpc PGCs are first detectable due to their high level of alkaline phosphatase activity.

The acquisition of germ cell status is accompanied by a marked reduction in proliferation rate (Lawson and Hage (1994)). From an original doubling time of 7 hours the proliferation slows down to a doubling time of 16-17 hours, a rate which is maintained steadily for the next 6 days (Tam and Snow (1981)). At about the same time the Oct4 gene (responsible for repression of differentiation genes (Pesce, et al. (1998)) switches to a transcript regulated by germ cell-specific distal enhancer (Scholer, et al. (1990); Yeom, et al. (1996)).

Germ cells migration and colonisation of the embryonic genital ridges

Towards the end of gastrulation (around 8.0 dpc) the posterior visceral endoderm moves in to form the hindgut, carrying with it the germ cells from the cluster and distributing them along the length of the hindgut. It has been observed that PGCs do not have pseudopodia before they occur in hind gut endoderm (Tam and Snow (1981)), suggesting that earlier PGCs do not undergo active migration but instead passively move with a morphogenetic expansion of embryonic tissues. Subsequently, around 9.5 dpc the PGCs emigrate from hind gut and move actively along the dorsal mesentery until they reach the genital ridge anlage (10.5 – 11.5 dpc) (see Fig. 6). The germ cells seem to emigrate independently, but soon afterwards they start to form extensive processes (up to 40 μm – Gomperts, et al. (1994)) by which they are linked up to each other to form an extensive network. During migration PGCs express on their surface the Stage Specific Embryonic Antigen 1 (SSEA1, Fox, et al. (1981)). Expression of this antigen, which is first evident at 9.5 dpc and is down regulated at about 12.5dpc, has been used as a PGC marker in several studies (Gomperts, et al. (1994); Garcia-Castro, et al. (1997)).

The cellular and molecular basis of route finding during germ cell migration is poorly understood. The proliferation of migrating cells is dependent on the c-kit/stem cell factor signal transduction pathway, as the embryos homozygous for mutations in genes coding for either the receptor (W) of the ligand (Steel) are deficient in germ cells. It has been shown that PGCs interact with diverse extracellular matrix proteins on their way to the genital ridge. Among others, the interaction with laminin (Garcia-Castro, et al. (1997)) and B1 integrins (Anderson, et al. (1999)) seems to be the most important for migration and colonisation of the genital ridges.

Fig. 6: PGC migration during the development of mouse embryoMicroscopic images show the migrating PGCs in 9.25 dpc (A), 9.5 dpc (B) and 10.5 dpc (C) mouse embryos. The migrating PGCs were visualised using the expression of lacZ under the control of germ line-specific Oct4 promoter. D shows the dark field image of 10.5 dpc embryo. The arrows indicate the PGCs moving into the forming genital ridges. (Yeom et al., 1996)

Germ cells in the mouse enter the genital ridge area between 10.5 and 11.5 dpc, at the time when the ridge is forming. By 11.5dpc a clear demarcation exists between genital ridge and mesonefros, making any subsequent PGC entry unlikely. Although the PGCs maintain their steady proliferation rate on entering the genital ridge, their morphology changes dramatically (Donovan, et al. (1986)). The PGCs lose their ability to elongate, become round, less motile, and their ability to spread on the substrate declines. Within the genital ridges the dividing germ cells form clonal clumps of up to 32 cells that tend to go through mitosis synchronously (Pepling and Spradling (1998)).

Gametogenesis inside the developing gonads

The sex specific differences appear first at 12.5dpc when the genital ridges become morphologically distinct – differentiated Sertoli cells appear in male genital ridges, whereas in female embryos the supporting cells differentiate as granulosa cells (for review see McLaren (2000)). One day later (at 13.5dpc) the germ cells in male undergo mitotic arrest, whereas the female PGCs enter meiotic prophase and pass through leptotene, zygotene and pachytene before arresting in diplotene around the time of birth (Peters (1970), see Fig. 7)

Fig. 7: Timing of germ cell development in the mouse embryo (for details see the text).

Brief summary of spermatogenesis

The male PGCs are mitotically arrested around 13.5dpc as T-prospermatogonia, in the G1 (G0) stage of the cell cycle (McLaren (1984)), not resuming mitosis until a week later, just after birth. The spermatogonia then proliferate rapidly, generating some progeny that retain the capacity to continue dividing indefinitely (as stem cell spermatogonia) and other progeny (maturing spermatogonia) that will, after a limited number of further normal division cycles, enter meiosis. The first spermatogenic stages in mouse do not enter meiosis until at least a week after birth (McLaren and Southee (1997)). After completion of second meiotic division the haploid spermatids are formed that differentiate into mature sperm (spermatozoa) (see Fig. 8).

Brief summary of oogenesis

Around 13.5dpc PGCs in female mice enter meiosis and arrest in diplotene of the first meiotic prophase around the time of birth (Constantini, et al. (1994)). The arrested oocytes (15-20 μm in diameter) undergo a progressive growth – the volume increases about 200-fold as the diameter reaches 75-80 μm. At the time when the diameter reaches 60 μm, the oocytes become capable of re-entering the cell cycle but are maintained in meiotic arrest by the surrounding follicular cells. The final period of oogenesis does not proceed before the sexual maturity. In this hormonally controlled phase the oocytes are stimulated to resume the first meiotic cell cycle and undergo the first meiotic division before arresting at metaphase of meiosis II (see Fig. 8). The second meiotic division is then completed only after fertilisation.

Fig. 8: Comparison of oogenesis and spermatogenesis.according to Constantini et al., 1994

Germ cell development is connected with epigenetic reprogramming

The epigenetic changes occurring during the germ cell development have been an issue of great interest since the postulation of imprinting. The progress in the field has, however, been hindered for a long time by technical difficulties caused by the laborious germ cell isolation and disputable sample purity on one hand, on the other hand, new highly sensitive methods had to be developed for the analysis of very limited cell samples.

The first attempts to characterise the methylation status of early germ cells have not been published earlier than in late 1980s. Based on their previous observation of X chromosome re-activation in the female foetal germ cells (Monk and McLaren (1981)), Monk and colleagues predicted the presence of general epigenetic changes involving possibly also changes in DNA methylation. Later, by checking the general methylation status in PGCs isolated from genital ridges of different embryonic stages, the authors confirmed that the genome of 12.5 and 14.5 dpc primordial germ cells is hypomethylated in comparison with the embryonic somatic tissues (Monk, et al. (1987)). The technique used in the study was, however, of a questionable sensitivity.

The observation of Monk et al. was further confirmed in early 1990s using more sophisticated methylation sensitive PCR assays (see chapter 1.7 Molecular techniques used for DNA methylation studies (Kafri, et al. (1992); Brandeis, et al. (1993)). Investigation of the methylation status of number of restriction sites within well-characterised genes (imprinted as well as non-imprinted) revealed complete absence of methylation at all the tested sites in 12.5dpc and 13.5dpc (the earliest stages tested) primordial germ cells of either sex. The lost of imprinting in early PGCs was documented also at the level of transcription (Szabo and Mann (1995)). In postmigratory PGCs purified from sexually still indifferent genital ridges, selected imprinted genes were shown to be expressed biallelically.

It is noteworthy that up to now all the studies describe PGCs at the stage when they are already free of imprints. Thus, it stays unclear whether the earlier stages of PGCs do posses established imprints, or whether they are derived earlier and escape the wave of de novo methylation.

More information about the epigenetic properties of primordial germ cells was brought about by the introduction of PGC-derived cell lines (embryonic germ (EG) cell lines). The first studies showed that EG cells have a similar epigenotype to PGCs from which they are derived (Resnick, et al. (1992); Matsui, et al. (1992)). In culture the EG cells keep their undifferentiated character, morphologically resembling embryonic stem cells (ES cells) and embryonal carcinoma cells. EG cells can contribute to most if not all the somatic tissues, as well as the germ line, and to this extent they are developmentally totipotent (Labosky, et al. (1994); Stewart, et al. (1994); Tada, et al. (1998)). Both male and female 11.5-12.5 dpc EG cells undergo comparable epigenetic changes (Tada, et al. (1998)) characterised by loss of imprinting of majority of the tested imprinted genes. The unique reprogramming activity of EG cells has been also shown in the EG-somatic cell fusion experiment (Tada, et al. (1997)). It has still to be elucidated, however, whether the processes occurring in PGCs in vivo are indentical to those described for EG cells.

Re-establishment of genomic imprints

Whereas at the onset of the work of this thesis it was still unclear whether the initial reprogramming events in early developing gonads differ in male and female embryos, the establishment of new epigenetic modifications is undoubtedly sex-specific. Generally, more detailed knowledge is available on the establishment of new imprints during spermatogenesis. The fact is probably due to the easier preparation procedure and relative abundance of sperm samples. The immunostaining of developing mouse testis with anti mC antibodies showed that the euchromatic regions of germ pass from a demethylated to a strongly methylated status between 16 and 17 dpc (i.e. prenatally, before the onset of meiosis) (Coffigny, et al. (1999)). Additional support for the idea, that the functional paternal imprint is established prior to meiosis came from the nuclear transfer experiments published independently by two groups in 1995 and 1998 (Kimura and Yanagimachi (1995); Ogura, et al. (1998)). The mouse oocytes receiving nuclei from primary spermatocytes (see Fig. 8) developed normally, albeit at a low success rate. The pre-meiotic de novo methylation has been confirmed also at the level of a single gene: the bisulphite analysis of H19 upstream DMR documented de novo methylation processes starting from 15.5 dpc on (Davis, et al. (2000) ; Ueda, et al. (2000)). These experiments furthermore suggest that if methylation is indeed the imprinting mark, the paternal specific methylation pattern might be established by the somatic form of Dnmt1, known to be present during this stage of spermatogenesis (see below).

The major step towards the understanding of imprint establishment during the oogenesis has been done by the work of Kono and colleagues (Kono, et al. (1996); Obata, et al. (1998)). By generating parthenogenetic embryos with nuclei from non-growing and fully-grown oocytes, the authors determined that the development of these embryos was extended by 3 days compared to parthenogenetic embryos derived only from the fully-grown oocytes. This improved developmental potential was proposed to be caused by epigenetic inequivalence of the oocyte genomes of the two developmental stages. The expression of imprinted genes revealed that the non-growing oocyte had apparently not acquired the maternal identity yet, thus allowing expression of some of paternally expressed genes, which are normally maternally repressed (for example Peg3, Peg1/Mest and Snrpn – Obata, et al. (1998)). Further experiments narrowed down the time window during the oocyte growth when the major epigenetic changes occur. The maternal genome is first competent to support development to term during the latter half of oocyte growth, at the time when oocyte becomes competent to enter metaphase of the first meiotic division (Bao, et al. (2000)).

Factors possibly involved in the establishment of gametic imprinting

Dnmt1 was the first candidate suggested to be involved in the establishment of gametic imprints. The oocyte-specific Dnmt1 isoform (Dnmt1o) has been found to be present during the oocyte growth – ie. around the predicted time of imprint establishment. A spermatocyte-specific 5’exon, to the contrary, interferes with translation and prevents production of Dnmt1 during the crossing-over stage of male meiosis, thus protecting a preferred methylation target from aberrant modification (Mertineit, et al. (1998)). The hypothesis was, however, disproved by the recent knockout of the oocyte-specific Dnmt1 isoform (Howell, et al. (2001)). The normal establishment of imprints in oocytes deprived of Dnmt1o suggests presence of other DNA methyltransferases that are responsible for de novo methylation during oogenesis. Although Dnmt3a and Dnmt3b methyltransferases with the described de novo methylation activities (Okano, et al. (1999)) might be good candidates, further experiments are needed to elucidate their possible role in gametogenesis. The newest findings revealed another candidate likely to be involved in the establishment of imprinting methylation marks (Bourc'his, et al. (2001); Hata, et al. (2002)). Dnmt3L is expressed at the key stages of gametogenesis, the disruption of the gene, moreover, causes infertility in homozygous males and aberrant imprint establishment in oocytes of homozygous females. As Dnmt3L lacks the key catalytic domains characteristic for DNA cytosine methyltransferases, the protein is more likely to act as a regulator of imprint establishment rather than as a DNA methyltransferase.

Molecular techniques used for DNA methylation studies

Since the Hotchkiss’es discovery of 5mC number of techniques have been developed in order to perform DNA methylation analysis. In general, the techniques can be divided into two main groups: i) techniques for non-specific methylation analysis and ii) techniques for sequence-specific methylation analysis.

Non-specific methylation analysis

Large-scale genome-wide changes in cytosine methylation levels are probably best monitored by reverse-phase HPLC. This procedure is in principle the oldest available for methylation analysis (Kuo, et al. (1980); Christman (1982); Gomes and Chang (1983)). It relies on the quantitative hydrolysis of DNA using DNase I and nuclease P1 (Kuo, et al. (1980)) or snake venom phosphodiesterase (Gomes and Chang (1983)), followed by alkaline phosphatase treatment. The amount of material required for the analysis is relatively high (several μg of DNA); the technique is moreover rather demanding concerning technical optimisation (assuring the complete DNA degradation) and the sample purity.

Thin layer chromatography represents an alternative procedure for studying genome-wide methylation levels (Bestor, et al. (1984); Schmitt, et al. (1997)). The DNA is initially cleaved with MspI restriction endonuclease (a recognition site CCGG is cleaved regardless of the methylation status; the method is based on the assumption that most of the vertebrate methylation occurs within CpG dinucleotides) and the internal cytosine labelled using [γ-32P]ATP and polynucleotide kinase. The DNA is then hydrolysed to mononucleotides using nuclease P1, and separated on cellulose thin-layer chromatography plates. The relative intensity of the C to 5mC spots will show the proportion of MspI sites that are methylated in the genome. In theory, only C and 5mC should give spots in this experimental set-up; nevertheless additional signals corresponding to A, G and T are often observed probably due to random nicks in the DNA.

Also methyl accepting assay and a chloracetaldehyde reaction belong among the applicable though rather rarely used methods for methylation analysis. In the methyl accepting assay, SssI prokaryotic methyltransferase transfers the 3H labelled methyl groups of a methyl group donor (SAM – S-adenosyl-methionin) onto the isolated DNA (Schmitt, et al. (1997)). The choracetaldehyde reaction couples the bisulphite modification (see below) with the subsequent chemical reaction of unconverted cytosines with chloracetaldehyde that yields an intensely fluorescent product (Oakeley, et al. (1999)).

Last but not least, immunological methods have been applied for methylation studies. The detection of methylation using anti-mC antibodies has been described in many publications (Piyathilake, et al. (2000); Mayer, et al. (2000)). The results (though very spectacular) are only qualitative, giving mainly the first impression about the overall methylation level.

Sequence-specific methylation analysis

Original methods to detect sequence specific genomic methylation were based on the digestion of DNA by methylation-sensitive restriction enzymes and subsequent Southern blot hybridization (Southern (1975)). Despite rather high amount of DNA needed for such experiments (> 5 μg) and the possibility to investigate just the limited numbers of CpGs situated within suitable restriction sites, the method is still useful as the first indication of methylation in a specific region. To improve the sensitivity, the method was combined with PCR amplification (Singer-Sam, et al. (1990)) and subsequent quantification of PCR products (Brandeis, et al. (1993,Kafri, et al. (1992)). Although the use of PCR decreased the amount of template DNA necessary for the analysis, the whole procedure is highly demanding in terms of strictly standardized conditions of DNA preparation and PCR, since quantification is only possible within the exponential phase of amplification. Additionally, incomplete digestion of chromosomal DNA might be a frequent source of artifacts. Another disadvantage of such methods is that they provide data only about the average level of methylation; it is neither possible to discriminate between mosaic and even methylation patterns or to address hemi-methylation, which remains in general undetected.

The first information about the methylation of cytosine residues irrespective of their sequence context was obtained using a genomic sequencing protocol (Maxam and Gilbert (1980)). This method identifies a position of 5-methylcytosine (5-MeC) in the genomic DNA as a site that is not cleaved by any of the Maxam and Gilbert sequencing reactions (Church and Gilbert (1984)) and thus appears as a gap in a sequencing ladder. Although a detailed distribution of methylation in a given sequence can be analyzed by this method, it still requires relatively large amounts of genomic DNA and a certain level of experience in interpreting the sequencing results as bands of varying intensity and shadow bands may occur. An elegant combination of the chemical cleavage method with ligation mediated PCR (Pfeifer, et al. (1989)) increases the sensitivity, but this modification makes the whole procedure rather laborious and technically challenging.

Bisulphite genomic sequencing

With a bisulphite genomic sequencing method (Clark, et al. (1994,Frommer, et al. (1992)) a qualitatively and quantitatively new approach to methylation analysis has appeared. The bisulphite reaction leads to the conversion of cytosines into uracil residues, which are recognized as thymines in subsequent PCR amplification and sequencing, whereas the modified cytosines do not react and are therefore detected as cytosines. Thus the method allows direct and positive determination of methylation sites in the genomic DNA, as only methylated cytosines are detected as cytosines. Products of PCR-amplified bisulphite-treated DNA can be used directly for sequencing (detection of average methylation status) or cloned and sequenced individually, when the information about the methylation pattern of single molecules is desired. Not only the methylation status of each single molecule but also the pattern of each DNA strand can be investigated, as the strands are no longer complementary following the bisulphite treatment and are amplified and sequenced separately.

Several modifications of the original bisulphite sequencing protocol improving the sensitivity and quality of the results have been published (Feil, et al. (1994,Olek and Walter (1997,Paulin, et al. (1998,Raizis, et al. (1995)). In some cases a direct sequence analysis of the PCR products obtained may be desirable to estimate the average methylation at specific sites. For such direct quantification Gonzalgo and Jones (Gonzalgo and Jones (1997)), proposed an elegant and simple procedure (Ms-SNuPE). A more sophisticated protocol for direct quantification of sequencing results is described by Paul et al. (Paul and Clark (1996)).

The attributes of high sensitivity, the ability to detect single molecule methylation patterns as well as the possibility of addressing non-symmetrical methylation make bisulphite-based genomic sequencing the method of choice for a variety of applications.

The aim of the thesis

Epigenetic reprogramming in germ line has been one of the key questions in the field, since the discovery of imprinting (Surani, et al. (1984); McGrath and Solter (1984)). However, at the time when the work on this thesis was being initiated only limited knowledge was still available about the underlying processes occurring in primordial germ cells. Whereas the majority of previous publications had concerned the onset of new, sex specific, imprints (Bao, et al. (2000,Davis, et al. (1999,Davis, et al. (2000,Kimura and Yanagimachi (1995,Trasler (1998,Ueda, et al. (2000) etc), the experiments described in this work focused on the poor characterised process of demethylation / imprint erasure.

The main tasks that stood at the beginning of this work were thus:

  1. The exact time definition of the main demodification changes that occur during the PGC development and the description of their kinetics.

  2. The question of “epigenetic origin” of PGC. The knowledge about the methylation status of early PGCs was expected to bring more understanding concerning the epigenetic fate of PGCs after their formation around 7.2 dpc.

The main work of this thesis was therefore to describe the methylation status of primordial germ cells at different developmental stages. In order to do so, two different approaches were chosen: Single gene approach was based on detailed monitoring of the changes in methylation patterns of distinct imprinted and non-imprinted genes. Genome-wide approach exploited the method of immunohistochemical mC staining in order to characterise the global methylation changes.