4. Discussion

4.1. Epigenetic reprogramming in the germ line

Germ line has unique features compared to other stem cell lineages. As a source of genetic material for the next generation, the primordial germ cell population is set aside very early during embryogenesis, constantly slowing the proliferation rate (presumably to minimise the risk of mutations) and kept totipotent (Lawson and Hage (1994,Tam and Snow (1981)). The role of the germ line is, however, not solely a transfer of genetic material; the postulation of genomic imprinting predicted that the gametes carry also epigenetic information that is sex-specific and necessary for the normal development of the embryo (Surani, et al. (1984), McGrath and Solter (1984)).

The predicted scenario of genetic imprinting requires the presence of epigenetic parent-of-origin specific marks, which are present in gametes and are maintained through cell divisions into an adult organism. . It is apparent that in the germ line these sex specific marks have to be erased and newly established according to the sex of the developing individual. Such germline epigenetic reprogramming had been predicted but at the onset of this thesis there was still only scarce experimental evidence available.

The aim of this thesis was thus to describe the epigenetic changes that occur in the developing germ line. To do so, we focused on monitoring the changes in DNA methylation, which was shown to function as an imprinting mark (reviewed in Reik and Walter (2001)). For that purpose mouse primordial germ cells (PGCs), where the major epigenetic changes were expected to take place, were collected from embryos of different developmental stages (10.5 – 13.5 dpc) and subjected to the bisulphite analysis. The methylation analysis was predominantly focused on well-characterised imprinted genes; for the reason of simple evaluation of methylation changes and of differences between parental alleles the bisulphite analysis was targeted at the DMRs with clear bimodal parent-of-origin methylation pattern. The selection of imprinted genes included examples of both maternally (Peg3, Snrpn DMR1, Lit1) and paternally (Igf2 DMR2, H19 upstream DMR, Snrpn DMR2) methylated genes. The kinetics of reprogramming (methylation/demethylation) processes was additionally monitored by methylation changes occurring in Xist promoter, as one of the markers for X re-activation. Last but not least, the observed methylation changes were compared to the changes occurring within non-imprinted genes (mylC,α-actin) to address to question of ubiquitous character of the reprogramming process.

4.2. Imprinted and non-imprinted genes undergo in PGCs fast and complete demethylation

The genome-wide methylation changes occurring during the PGC development were first described in 1987 by Monk and colleagues (Monk, et al. (1987)). The authors described the hypomethylated state of the genomic DNA isolated from 12.5 dpc and 14.5 dpc PGCs. The methylation changes were further confirmed also at the level of single genes: the tested genes were found hypomethylated and free of imprints in the PGCs isolated from 12.5 dpc and 13.5 dpc mouse embryos (Brandeis, et al. (1993,Kafri, et al. (1992)). All the previously published observations, however, concerned PGCs at the stages where the cells were already devoid of methylation imprints. As our main interest was to describe the hypothesised demethylation event we focused on the earlier developmental stages of PGCs, where we expected the methylation imprints to be still present.

The results of the performed bisulphite analysis are summarised in the Fig. 37.

Fig. 37: Summary of the bisulphite analysis performed on primordial germ cells.

The graphs show relative methylation at the distinct CpG positions within the analyzed regions (for details see chapter Results). Note that the H19 graph combines the methylation analysis of the proximal (right) and the distal (left) part of the H19 upstream DMR.

4.2.1. Reprograming of maternally methylated imprinted genes

Using the bisulphite approach we found that all the tested maternally methylated DMRs (Snrpn DMR1, Peg3, and Lit1) undergo in primordial germ cells identical changes of their methylation patterns (see Fig. 37). In the PGCs isolated from 11.5 dpc mouse embryos the maternally methylated DMRs were found to be methylated in about 50% of sequenced clones, which, as we assume, indicate a presence of an imprinting mark. This assumption is based on the following: 1) Similar results were obtained when working with the somatic tissue samples in experimental set-ups, where allele discrimination was possible. Moreover, in 11.5 dpc PGCs likewise to somatic samples mainly completely methylated or completely non-methylated clones were detected, which is a typical feature when analysing the imprinted DMR methylation. 2) The primers and the conditions for the bisulphite PCR had been previously intensively tested in our laboratory in order to uncover possible bias in our experimental procedure.

To the contrary, none or only sporadic methylation was detected in the samples of 12.5 dpc and 13.5 dpc PGCs. Elevation of methylation levels at some CpG positions within peg3 gene was presumably caused by low number of obtained unique clones.

4.2.2. Reprogramming of paternally methylated imprinted genes

Similarly to maternally methylated regions discussed above, we found the tested paternally methylated DMRs (Igf2 DMR2 and H19 DMR) non-methylated in PGCs of 12.5 dpc and 13.5 dpc mouse embryos. The absence of methylation in the 3’part of H19 DMR was unexpected as it was in an obvious disagreement with the recently published observation of Ueda and colleagues (Ueda, et al. (2000)). The authors described de novo methylation occurring in this particular part of the H19 DMR from 13.5 dpc on. As in both analysis the same primers and PCR conditions were used for the bisulphite PCR, the discrepancy might be explained by the use of a different mouse strain (C57Bl/6 vs. outbred MF1 strain used in our experiments), or perhaps by a different time-scheme of sample collection. Considering the results of our methylation analysis it is apparent, that the major methylation changes can occur in PGCs within several hours. The different time schedule of the hormone induction, fertilisation and finally the PGC collection might thus account for a different result. It should be also noted that using a very sensitive bisulphite approach the purity of the isolated PGCs is a key factor. Any (even very low) contamination with the somatic cells of the embryonic gonads might lead to the observed low levels of methylation. It is thus very important to mention that all our samples (regardless whether MACS or FACS sorted) were checked additionally for purity using an alkaline phosphatase staining. The purity of collected samples exceeded always 95%. Such a purity-control check was, however, not mentioned in the work of Ueda et al (Ueda, et al. (2000)).

Contrary to the described maternally methylated DMRs, the paternally methylated DMRs did not show a uniform methylation patter at 11.5 dpc (see Fig. 37). In the samples of 11.5 dpc PGCs only the 3’ part of H19 DMR appeared to be methylated, whereas both the 5’ part of H19 DMR and the DMR2 of Igf2 revealed no methylation. The results obtained on the 10.5 dpc samples showed that at least Igf2 DMR2 was methylated in the earlier stages of PGCs. As our bisulphite analysis of earlier stages of PGCs failed in the case of the 5’part of H19 DMR, it can still be hypothesised, that the region is not methylated up to 13.5 dpc

Our finding that the Igf2 DMR2 undegoes demethylation earlier (between 10.5 dpc and 11.5 dpc) could indicate that the “core” imprinting centre in the Igf2-H19 locus is located in the 3’ part of the H19 DMR. This region thus behaves in the same manner as the tested maternally methylated DMRs (see above), whereas the 5’ part of H19 DMR and the DMR2 of Igf2 act as “second level” DMRs reacting on the demethylation events faster. Prediction of such a DMR hierarchy in the H19-Igf2 genomic locus is supported also by the response of the DMRs during the zygotic demethylation (see following chapters). Whereas the DMR2 of Igf2 was documented to undergo complete demethylation (Oswald, et al. (2000)), the DMR of H19 (or at least some part of it) keeps its methylated status (Warnecke, et al. (1998)). Additional evidence comes from the mouse knockout experiments: the deletion of the H19 DMR influences the methylation of the Igf2 DMR2, but to the contrary, the absence of the Igf2 DMR2 does not have any effect on the H19 DMR methylation (S.Lopes – manuscript in preparation).

4.2.3. DMR2 of Snrpn undergoes protracted demethylation

A special example of paternally methylated DMR is the DMR2 of Snrpn. As the only available information concerning the region was based solely on the mapping with restriction endonucleases (Shemer, et al. (1997)) it was first necessary to perform a detailed sequence analysis. This surprisingly revealed that the region is CG poor and, moreover, the highest density of CpGs is associated with a part of repetitive (Line1) element.

Concerning the CpG density the DMR2 of Snrpn represent certainly a unique example among up-to-now characterised DMRs. Whereas typical DMRs have the features of CpG islands or are spanning clusters of CpG, the Snrpn DMR2 comprises 19 CpGs over more than 4,3 kb of genomic sequence. It is disputable whether the density of DNA methylation is sufficient to carry the epigenetic information in this region, or whether (presumably) the imprinting mark is formed by a combination of different types of epigenetic mechanisms (i.e. histone acetylation, histone methylation etc.). Such a possibility has to be, though, yet experimentally elucidated.

As our results represented the first bisulphite analysis of this region, it was important to determine, whether the sequence outside the previously tested restriction sites displays differential methylation. Bisulphite analysis performed on the 11.5 dpc PGCs manifested that the region is fully methylated in about 50% of sequenced clones, that is obviously in agreement with the presence of expected imprint. Still, it is necessary to point out, that the set-up of our experiments did not allow distinguishing the alleles with regard to their parental origin.

The bisulphite analysis of the 12.5 dpc and 13.5 primordial germ cells revealed that the Snrpn DMR2 undergoes protracted demethylation reaching the demethylated state at 13.5 dpc in a gradual manner. Such a behaviour contrasts with demethylation kinetics observed in other DMRs (both paternally and maternally methylated) and resembles more the behaviour of repetitive elements (see later). Such a finding raises the questions about the methylation profile of the rest of the Snrpn DMR2. Slow demethylation of the rest of the DMR would suggest that the features of the whole DMR are directed by the integrated repetitive element. To the contrary, it could be also imagined that the integration event happened after the region had gained its properties as a DMR element. The integrated repetitive element thus subsequently gained the imprinted properties, but is by some cellular machineries still recognised as being a repetitive element. It should be noted, that the gain of imprinting following the retrotransposition of genes into the imprinted region has been already described (Chai, et al. (2001)). It is remarkable that imprinted genes are often found to be associated with repetitive elements. It is, however, questionable, whether the imprinted status of the particular genomic region is connected with the presence of repetitive elements (repetitive elements might be responsible for certain regional chromatin configuration), or whether the specific chromatin properties typical for DMRs enable easy and efficient transposition events. Further investigation is, however, needed to resolve the biological background of this phenomenon.

4.2.4. Non-imprinted genes follow the same demethylation scheme as imprinted genes

Except of imprinted genes, two examples of single copy non-imprinted genes (α-actin, mylC) were included into the bisulphite analysis in order to clarify the specificity of the demethylation process. Identically to the imprinted genes we found the investigated non-imprinted regions methylated in 11.5 dpc PGCs with the methylation levels similar to those observed in somatic tissues (Warnecke and Clark (1999); Walsh and Bestor (1999)). Furthermore, also in the non-imprinted regions the complete demethylation occurred between 11.5 dpc and 12.5 dpc (see Fig. 37).

.

4.3. Biological aspects of the germline demethylation

The presented comprehensive data document a presence of a widespread demethylation mechanism affecting in the same manner imprinted genes (regardless of the origin of their methylation marks) as well as non-imprinted genes. It should be pointed out that the data represent the first solid experimental evidence concerning the mechanism, which had been previously hypothesised as an essential part of the predicted scenario of imprinting, but up to now not experimentally documented.

The results of our experiments show that at 10.5 dpc –11.5 dpc at the time when they enter the developing genital anlagen the PGCs contain high levels of methylation. Shortly afterwards – between 11.5 dpc and 12.5 dpc – the majority of the tested genes undergo fast and complete demethylation. This is in complete agreement with the hypomethylated state of 12.5 dpc and 13.5 dpc PGCs previously described by Kafri et al. and Brandeis et al.(Kafri, et al. (1992); Brandeis, et al. (1993)) and the documented biallelic expression of imprinted genes at this stage of PGC development (Szabo and Mann (1995)). The demethylation affects both the imprinted genes regardless of the origin of their imprinting methylation mark as well as non-imprinted genes. Moreover, our data clearly demonstrate that the methylation erasure proceeds identically in the primordial germ cells of female and male. The epigenetic resetting commences in the not yet sexually differentiated gonads and is probably the only time during the development when the germ cells of either sex are equivalent and free of any epigenetic imprint. This epigenetic “zero baseline” is apparently the starting point for the subsequent sexual differentiation (starting around 13.5 dpc) and the later initiation of new gamete specific imprints (see Fig. 38).

Fig. 38: Epigenetic reprogramming in the germ line.

As the migrating PGCs colonise the genital ridges they possess methylation imprinting marks (shown in blue and red). Shortly afterwards the reprogramming commences in still bipotential gonads perhaps as a reaction to a somatic signal emanating from the stroma of a genital ridge; the imprints are erased and the inactive X chromosome in PGCs of female re-activated. Sex specific methylation imprints are established later in sexually fully differentiated gonads (male and female gonads shown in blue and red, respectively).

An important feature of the germ line demethylation is that the imprinting methylation marks are erased completely (such a statement is, however, not valid for repetitive elements as discussed in the following chapters). This finding is important in the light of the work recently published by Davis et al (Davis, et al. (1999,Davis, et al. (2000)). The authors described the differences between establishment of the H19 imprinting mark on the allele of a maternal and a paternal origin during spermatogenesis, speculating that the H19 methylation mark persists (at least partially) on the allele of the paternal origin. Our results describing complete loss of imprints in 12.5 dpc and 13.5 dpc PGCs do not justify such a hypothesis. The faster (or perhaps easier) remethylation of the H19 paternal allele could be caused by the persisting differential chromatin structure or modification (histone acetylation, histone methylation etc.) Hence, it has to be stressed out that the data of this thesis concern solely the erasure of methylation epigenetic marks, whereas the destiny of other types of epigenetic marks (for example the histone modifications) remains speculative.

Another important characteristic of the demethylation process is its tissue specificity. The strict restriction of the demethylation to the germ cells was proven by the results of methylation analysis performed on the stage-matched somatic cells of the genital ridge. Whereas the 12.5 dpc PGCs were completely devoid of methylation imprints, the corresponding cells of the genital ridge stroma kept somatic methylation pattern. The reprogramming ability is thus an intrinsic feature of the germ line.

In connection with the methylation analysis many questions have been raised concerning the occurrence of unusual methylation patterns. The presence of asymmetrically methylated sites has been described to occur in the imprinted H19-Igf2 locus (Vu, et al. (2000)) or in the systems over-expressing Dnmt3a methyltransferase (Lyko, et al. (1999); Ramsahoye, et al. (2000); Lin et al., 2002). The presence of such a methylation pattern has, however, never been detected in our PGC methylation studies. Due to the mechanism of the bisulphite conversion (the DNA strands are no longer complementary following the bisulphite treatment) only one DNA strand is usually subjected to the methylation analysis. Similarly, in all our experiments only one DNA strand was analysed for its methylation status. It could be thus speculated that the observed methylation changes are strand specific and the demethylation connected perhaps to the ongoing replication that leaves the newly synthesised DNA strand devoid of methylation. Such a scenario is, however, difficult to imagine, as in some cases the methylation pattern of the upper DNA strand (H19, Peg3, Xist, Snrpn, α-actin) was analysed whereas in other cases the analysis was focused on the lower DNA strand (mylC, Igf2). To finally exclude such a possibility the methylation status of both DNA strands will have to be investigated for some regions.

4.4. PGCs in vivo differ from the PGC-derived EG cell lines

Due to technical difficulties connected with the isolation of primordial germ cells in an amount sufficient for the methylation analysis, many previously published experiments used PGC-derived EG (embryonic germ) cell lines as an experimental model for PGCs. The EG lines were shown to keep the undifferentiated morphology similar to ES (embryonic stem) cells and to be able to contribute to all types of tissues including the germ line when used to produce mouse chimeras (Labosky, et al. (1994); Stewart, et al. (1994); Tada, et al. (1998)). The genome of EG cell lines derived from 11.5 dpc and 12.5 dpc PGCs was shown to be grossly hypomethylated and devoid of methylation imprints (Kato, et al. (1999); Tada, et al. (1998)).

In the light of our findings it is intriguing that the genome of EG cell lines derived from 11.5 dpc PGCs is completely demethylated whereas the DNA of PGCs isolated from the 11.5 dpc mouse embryos contains still methylation imprints. The epigenotype of the 11.5 dpc derived EG cell lines is thus similar to the 12.5 dpc PGCs, rather than to 11.5 dpc PGCs. This discrepancy is supported furthermore by the recent observation of Durcova-Hills et al. (Durcova-Hills, et al. (2001)). The authors describe the EG cell lines derived from migrating PGCs of 9.5 dpc mouse embryos. Also these EG cell lines are devoid of methylation imprints. It seems that the PGCs are already “programmed” at the time point of the isolation to commence the demethylation, with which they proceed once they are transplanted into the cell culture. Alternatively, the demethylation could can occur during the cultivation as a response to the cellular signals provided by the co-cultivated somatic cells.

4.5. A global character of genomic demethylation is confirmed by the pattern of anti-mC staining

A further support for the genome-wide character of the demethylation occurring in the primordial germ cells was given by the results of immunohistochemical staining using the anti-mC antibody. Whereas multiple distinct positively stained speckles were characteristic for the nuclei of control stage-matched somatic cells of embryonic gonads, the speckles appeared only sporadically in the nuclei of primordial germ cells. Furthermore, the number of positively stained PGCs declined between 11.5 dpc and 13.5 dpc. It is remarkable that, the main change in the number of positively stained cells appears between 12.5 dpc and 13.5 dpc, whereas the single copy genes undergo the demethylation earlier - between 11.5 dpc and 12.5 dpc. Moreover, even among the 13.5 dpc PGCs we found the cells with positively stained foci. This delayed and incomplete disappearance of the positive signal could be explained by the residual methylation of repetitive elements, which were shown to undergo protracted and incomplete demethylation (N.Lane – unpublished data, Hajkova et al. – submitted). It is also possible that the sporadically appearing positively stained cells are due to a low somatic cell contamination of our PGC samples (although the purity of the PGC samples always exceeded 95%). It is remarkable that the staining pattern did overlap neither with centromeric regions nor with the DAPI staining suggesting that the DNA methylation is not focused to the heterochromatic DNA. Such a distribution could be easily explained assuming that the anti-mC staining is targeted to the repetitive elements.

4.6. Germ cells do not escape from the wave of de novo methylation in the gastrula stage

The previously published observations documented that the genome of 12.5 dpc and 13.5 dpc PGCs is grossly hypomethylated (Kafri, et al. (1992); Brandeis, et al. (1993); Monk, et al. (1987)). Since this was more or less the only knowledge available the methylation status of PGCs of earlier developmental stages was only speculated. Based on that several theories appeared: One of the possible explanations for the low levels of methylation found in post-migratory PGCs was that the founder population of germ cells does not undergo the pre-implantation wave of de novo methylation (Jaenisch (1997); Monk, et al. (1987)). Alternative scenario suggested that the founder PGCs undergo the de novo methylation event similarly to the somatic lineage, the PGCs get demethylated subsequently by a germline specific mechanism (Monk, et al. (1987)) (see Fig. 39).

Our experiments clearly show, that up to 11.5 dpc primordial germ cells contain high levels of methylation including fully established methylation imprinting marks. Such a finding strongly suggests that the founder population of PGCs is indeed subjected to the pre-implantation de novo methylation processes (see Fig. 39). However, it has to be still experimentally elucidated whether the discussed de novo methylation proceeds identically in the developing germ line and in the somatic lineage.

Fig. 39: Dynamics of DNA methylation in mouse germ line.

Following the pre-implantation genome wide demethylation the somatic lineages together with the founder population of PGCs undergo a wave of postimplantation de novo methylation. The methylation is erased in the differentiating PGCs after their entry into the gonads by a germ line specific mechanism. The new methylation imprints are established subsequently in the process of gamete maturation.

4.7. Single copy genes and repetitive elements follow different demethylation kinetics

Contrary to the single copy genes (both imprinted and non-imprinted), which undergo fast and complete demethylation between 11.5 dpc and 12.5 dpc, the demethylation of repetitive elements is in PGCs (11.5 dpc – 13.5 dpc) prolonged and incomplete (see Fig. 40). Such a conclusion is the result of the comprehensive methylation analysis of two classes of mouse repetitive elements (Line1 and IAPs) performed in the parallel to the work of this thesis in the Babraham Institute (Babraham, UK, N.Lane unpublished data).

Fig. 40: Comparison of the demethylation kinetics in the germ line: single copy versus repetitive sequences.

Single copy loci undergo fast and complete demethylation in PGCs between 11.5 dpc and 12.5 dpc. To the contra

The effect of the genomic localisation and the chromatin accessibility of the particular repetitive element could possibly explain the fact that not all of the repetitive elements undergo demethylation at the same time point. Such an explanation is, however, not very likely, as we did not observe any variability while investigating the single copy genes (assuming that the tested genes were not by coincidence located in the loci of the same chromosomal features).

A more plausible scenario could be to imagine the situation in PGCs as the dynamic process of the demethylation and de novo methylation. Whereas the demethylation machinery would work in a genome–wide non-specific manner, the de novo methylation processes might be specifically targeted to the repetitive sequences. Such de novo methylation mechanism could be potentially induced by the expression of the demethylated repetitive sequences (the inhibitory effect of the DNA methylation on the expression of repetitive elements has been well documented (Walsh, et al. (1998)) in a process similar to PTGS (post-transcriptional gene silencing – for review see Cogoni and Macino (2000)). Alternatively, the specificity of de novo methylation could be determined by the substrate specificity of the acting DNA methyltransferase. It should be noted, that the newly described knock-out of Dnmt3L (a protein with high similarity to Dnmt3a and Dnmt3b methyltransferases - Bourc'his, et al. (2001)) affects solely methylation of single copy genes suggesting that the single copy and repetitive sequences are methylated by distinct cellular machineries.

Another possible scenario is that the fast demethylation targets only single copy loci and the loss of methylation in the tested repetitive sequences is due to the on-going replication in the absence of the de novo methyltransferases. As revealed by the immunostaining experiments Dnmt3a is absent in PGCs and Dnmt3b is excluded from the nucleus showing cytoplasmic localisation. Contrary to the de novo methyltransferases, high level of the maintenance Dnmt1 methyltransferase was detected in PGC nuclei (S. Erhardt – unpublished data, Hajkova et al. – submitted). However, as documented by recently published results, the presence of all the three eukaryotic methyltransferases (Dnmt1, Dnmt3a and Dnmt3b) is necessary in order to maintain the methylation of repetitive elements (Liang, et al. (2002)).

The presence of the mechanism, which maintains (at least partially) the methylated and silenced status of the repetitive elements is crucial from the evolutionary point of view to protect the integrity of the genetic information in the germ line from the deleterious effects of retroelements and other transposable sequences. Moreover the incomplete erasure of methylation in repetitive elements could be seen an enhancing force for a evolutionary variability of the species. Such an example is given by a mouse agouti locus, where the activity of the neighbouring gene is influenced by the methylation of the integrated transposable element. (Morgan, et al. (1999);Wolff, et al. (1998)). The incomplete loss of methylation within this repetitive element in the germ line yields in the high variability of the agouti expression in the progeny.

4.8. Xist expression is silenced in post-migratory PGCs by a mechanism distinct to DNA methylation

Mammals compensate for different dose of X-linked-genes in male (XY) and female (XX) diploid cells by inactivating all but one X chromosome in each cell (for review see Mlynarczyk and Panning (2000)). Although the mechanism of the X chromosome inactivation is not yet completely understood, the initiation of the process is known to be connected with the expression of the non-coding Xist (X inactivation specific transcript) RNA. In the last decade several scientific reports suggested methylation of the Xist promoter as the key regulator in the X inactivation (Allaman-Pillet, et al. (1998,Norris, et al. (1994)). The role of methylation in the X inactivation was, however, undermined by the finding that the process occurs in the early embryogenesis apparently in the absence of promoter methylation (McDonald, et al. (1998)).

Reactivation of the inactive X chromosome in the primordial germ is considered as the marker for the germ line reprogramming processes (McLaren and Southee (1997,Monk and McLaren (1981)). The reactivation occurs between 11.5 dpc and 13.5 dpc, i.e. after PGCs colonised the genital ridges and is probably triggered by a somatic signal from the stroma of the genital ridge (Tam, et al. (1994)). While following the fate of methylation in the primordial germ cells after their entry into the genital anlagen it was of a high interest to investigate also the methylation status of the Xist promoter. For the reason of simplicity we followed the methylation pattern of the Xist promoter in male PGCs that contain a single active X chromosome with transcriptionally silent and hence presumably methylated Xist gene. The Xist promoter appeared to be no exception to the other tested single copy genes – the promoter sustains complete demethylation between 11.5 dpc and 12.5 dpc. This loss of promoter methylation seems to be surprising since the X chromosome has been reported to remain active in the maturing PGCs (Nesterova, et al. (2002)). Furthermore, recent evidence shows that despite demethylation of the Xist promoter in PGCs documented by our results, Xist transcript decreases progressively and is extinguished in most PGCs by 13.5 dpc (Nesterova, et al. (2002)). It seems, therefore, that the Xist of the active X chromosome is in PGCs (similarly to the early embryogenesis) transcriptionally silenced by a mechanism distinct to promoter methylation.

4.9. Is germline demethylation an active process?

Two general models have been postulated for the mechanism of demethylation (see Fig. 41). Passive demethylation is caused by replication proceeding in the absence of the maintenance activity represented in mammalian cells by Dnmt1 (DNA methyltransferase 1). As a consequence, methylated DNA strands are gradually diluted with the increasing number of replications. To the contrary, active demethylation requires an enzymatic activity. It has been proposed that such a demethylation involves either a glycosylase and repair activity (Jost, et al. (2001,Zhu, et al. (2001,Zhu, et al. (2000)) or a nucleotide excision and replacement activity (Weiss, et al. (1996)). Recently, however, the existence of novel “demethylase” was reported (Bhattacharya, et al. (1999); Cervoni, et al. (1999)). The described enzyme is predicted to transform methylated cytosines to cytosines by direct removal of a methyl group and to work in a processive manner. The nature of this enzyme and exact activity have to be, though, still elucidated as some of the results appeared to be apparently not reproducible .

Fig. 41: Comparison of active and passive demethylation.

Passive demethylation occurs as a consequence of replication proceeding in the absence of maintenance methylation (provided by Dnmt1). Contrary to passive demethylation, active demethylation is replication independent and requires an enzymatic activity,

Interestingly, the demethylation process occurring in primordial germ cells exhibit all the features of active demethylation. First of all, the demethylation of single copy loci is completed within only one day (between 11.5 and 12.5 dpc) and possibly even faster. Considering the replication time of PGCs at that developmental stage (16-17 hours (Tam and Snow (1981)), it is obvious that the demethylation occurs within one replication cycle. Second, the immunofluorescent staining of the primordial germ cells of the corresponding developmental stages manifested high level of Dnmt1 expression as well as the nuclear localisation of the enzyme, which suggests that the demethylation occurs in the presence of Dnmt1 (S. Erhard – unpublished data, Hajkova et al. - submitted).

Similarly to PGCs, also PGC-derived EG cells possess strong demethylation activity – when fused to a somatic cell, they can cause the demethylation of the somatic nucleus (Tada, et al. (1997)). In the light of our findings it is presumable that those dominant reprogramming activities are associated with the same active demethylation process. It is imaginable that similar demethylation (reprogramming) activities are a feature common to all pluripotent cells. However, this is not the case, as no similar activities were found in ES cells (Tada, et al. (2001)). The demethylation activity thus seems to be a striking feature of primordial germ cells and their derivatives – EG cells.

Zygotic versus germline demethylation

The demethylation observed in primordial germ cells is not the only example of the described active demethylation - a similar process has been documented to occur in a zygote just several hours after fertilisation (Mayer, et al. (2000,Oswald, et al. (2000)). The zygotic demethylation, similarly to the germline demethylation, commences in absence of DNA replication, the process is, however, restricted only to the paternal pronucleus and probably linked to chromatin remodelling of protamine-packed sperm DNA (Barton, et al. (2001)). Additionally, whereas several single copy sequences as well as IAPs and Line1 elements have been described to be affected by the zygotic demethylation, some regions (for example - H19 DMR, Warnecke, et al. (1998)) apparently withstand the process. This is in a striking contrast to the germ line demethylation, which affects all the tested genomic regions.

In the light of the latest findings connecting the DNA methylation with the methylation status of histones (Tamaru and Selker (2001)) it is possible that the germline specific demethylation is also connected with the global chromatin changes (though probably of a different character then those occurring during the zygotic demethylation). Such a possibility has to be yet experimentally elucidated.

4.10. A somatic signal rather than an intrinsic clock triggers the germ line demethylation

The previous experiments with the EG cell lines suggested that the reprogramming activities are intrinsic features of PGCs and happen precociously. The EG cell lines derived from migratory (9.5 dpc) and early post-migratory (11.5 dpc) PGCs are heavily hypomethylated and avoid of imprints (Tada, et al. (1998); Durcova-Hills, et al. (2001)).

Contrarily, the work of Tam et al. (Tam, et al. (1994)) clearly described that in order to re-activate the inactivated X chromosome the PGCs have to enter the genital ridge. Such a finding thus strongly argues for the presence of a somatic signal emanating from the stroma of a developing genital ridge. Also our experiments support this notion: at the time when the PGCs reach the genital ridge (10.5-11.5 dpc) they still posses methylation patterns comparable to the somatic cells, however, only shortly afterwards (12.5 dpc) the PGCs appear to be completely demethylated (see Fig. 38). Such kinetics could be explained by a somatic signal triggering the whole process. It is also possible, that the primordial germ cells “sensitive” to such a signal have to develop separately (i.e. outside the genital ridge) to reach the genital anlagen (and to be reprogrammed) just in time before the sexual differentiation of the gonads. Hence, it might be that the migration of germ cells and epigenetic reprogramming are phenomena that are evolutionary connected.

It should be noted, that during the EG cell lines derivation the PGCs are cultivated in a mixture with the somatic cells of the genital ridge stroma. The hypomethylation observed in EG cells could thus be induced by a signal emanating from those somatic cells or by a factor connected to tissue culture conditions.

Taking into account that the germline reprogramming is a typical example of a developmental process restoring cellular totipotency, the understanding of the nature and the action of the signal triggering this process might shed more light also on the processes such as cloning and derivation of stem cells.

4.11. Evolutionary aspects of germline demethylation

The germline demethylation assures complete erasure of imprints and of most of “non-imprinted” methylation. Such a mechanism has presumably two major functions: a) it is crucial in order to prevent mutations to be spread through the generation b) it is vital for the function of gametic imprinting. It is, however, difficult to argue, which function appeared primarily during the evolution.

From the evolutionary point of view, there might be an interesting connection between the zygotic and the germline demethylations (both processes were compared in the previous chapter). The zygotic demethylation was described in mouse and bovine (Mayer, et al. (2000); Oswald, et al. (2000); Dean, et al. (2001)). Similar process was, however, not found in Xenopus (Stancheva, et al. (2002)) and neither early Zebrafish embryos display any methylation dynamics (Macleod, et al. (1999)). Hence it seems that this demethylation process might be limited to mammals (i.e. the species with gametic imprints). It is particularly interesting, that similar active demethylation events have been recently described in the flowering plants showing phenomena analogous to genomic imprinting (Spielman, et al. (2001)). The demethylation occurs during the first 5 hours of seed germination prior to DNA replication and thus likely in an active manner (comparable to zygotic demethylation?) (Zluvova, et al. (2001)). This finding thus implies that the demethylation processes might not be limited to the mammals, but might rather be a feature of organisms, which evolved gamete-specific epigenetic differences.

More experimental work is nevertheless needed to uncover whether the germline demethylation occurs always in connection with the zygotic demethylation and evolved thus possibly with genomic imprinting; or whether it has appeared independently during the evolution as the germline mechanism preventing accumulation of mutations.

4.12. On the nature of paternal and maternal imprints

Since the discovery of genomic imprinting it has been speculated what is the real nature of the imprinting mark. The scientific reports of the last decade brought rather vast evidence that imprinting is connected with the appearance of so called differentially methylated regions (DMRs) (for review see Reik and Walter (2001)). Parent-of-origin specific methylation detected in those regions is believed to be responsible for monoalellic expression and hence to act as the imprinting mark.

A question that is not widely discussed is the difference between paternal and maternal imprinting marks. Whereas it might be generally believed that the parental imprinting marks do not differ, the thorough analysis of the published results might indicate otherwise:

  1. The re-establishment of the paternal methylation marks seems to be generally easier: The EG cell lines that are devoid of methylation imprints restore paternal but not the maternal methylation marks in mouse chimeras. Similarly, the Dnmt1 -/- cells can reconstitute the paternal methylation imprinting marks after the introduction of the Dnmt1 expressing construct. However, much higher overexpression of Dnmt1 is needed in order to restore the maternal methylation marks (Biniszkiewicz, et al. (2002); Zhan Jun – unpublished results).
  2. As documented in the case of imprinted human Snrpn gene the methylation is not absolutely necessary for the maternal imprinting mark – the promoter of Snrpn is not methylated in human oocytes, the maternal methylation mark is established and propagated postzygotically (El-Maarri, et al. (2001)). It might be thus speculated that the methylation of the maternal allele is directed by some other signal - for example chromatin modification.

Based on the above-mentioned observation it is possible to speculate that the maternal imprinting mark is more complex than the paternal one. Maternal imprint might be created on several distinct levels – one certainly including DNA methylation, the other(s) exploiting different possibilities of chromatin modifications. It should be noted, that two of the maternally methylated DMRs (Snrpn and U2af1-rs1) have been recently described to be associated with the differential histone acetylation (Gregory, et al. (2001)). Reconstitution of such a complex maternal mark would be then clearly more complicated, which would explain the observation mentioned above.


© Die inhaltliche Zusammenstellung und Aufmachung dieser Publikation sowie die elektronische Verarbeitung sind urheberrechtlich geschützt. Jede Verwertung, die nicht ausdrücklich vom Urheberrechtsgesetz zugelassen ist, bedarf der vorherigen Zustimmung. Das gilt insbesondere für die Vervielfältigung, die Bearbeitung und Einspeicherung und Verarbeitung in elektronische Systeme.
DiML DTD Version 3.0Zertifizierter Dokumentenserver
der Humboldt-Universität zu Berlin
HTML generated:
28.09.2004