Report

Genome-Wide Reprogramming in the Mouse Germ Line Entails the Base Excision Repair Pathway

See allHide authors and affiliations

Science  02 Jul 2010:
Vol. 329, Issue 5987, pp. 78-82
DOI: 10.1126/science.1187945

Abstract

Genome-wide active DNA demethylation in primordial germ cells (PGCs), which reprograms the epigenome for totipotency, is linked to changes in nuclear architecture, loss of histone modifications, and widespread histone replacement. Here, we show that DNA demethylation in the mouse PGCs is mechanistically linked to the appearance of single-stranded DNA (ssDNA) breaks and the activation of the base excision repair (BER) pathway, as is the case in the zygote where the paternal pronucleus undergoes active DNA demethylation shortly after fertilization. Whereas BER might be triggered by deamination of a methylcytosine (5mC), cumulative evidence indicates other mechanisms in germ cells. We demonstrate that DNA repair through BER represents a core component of genome-wide DNA demethylation in vivo and provides a mechanistic link to the extensive chromatin remodeling in developing PGCs.

The specification of mouse primordial germ cells (PGCs) at embryonic day 7.25 (E7.25) is accompanied by the initiation of epigenetic changes (1), followed by widespread epigenetic reprogramming at E11.5, which includes genome-wide DNA demethylation, erasure of genomic imprints, and large-scale chromatin remodeling (13) (fig. S1). Chromatin remodeling follows the onset of genome-wide DNA demethylation, which suggests that DNA repair might be linked to this process (2). DNA repair–driven DNA demethylation would involve replacement of a methylcytosine (5mC)–containing nucleotide by an unmethylated cytosine (4, 5). As the epigenetic changes in E11.5 PGCs occur in the G2 phase of the cell cycle and are thus independent of DNA replication (2), the most likely mechanisms for the replacement of 5mC would be the nucleotide excision repair (NER) or base excision repair (BER) pathways.

We obtained a quantitative measure of expression of BER and NER components and observed an up-regulation of transcripts of BER components Parp1, Ape1, and Xrcc1 in E11.5 PGCs, which was not seen in the neighboring somatic cells (6). By contrast, we observed little expression of NER components Ercc1 and Xpa in these PGCs (Fig. 1A and fig. S2).

Fig. 1

Activation of BER during DNA demethylation in PGCs. (A) Quantitative polymerase chain reaction (QPCR) analysis of PGC and neighboring somatic cells at E11.5. (Error bars show standard deviation between technical replicates.) (B) Chromatin-bound XRCC1 detected by immunofluorescence in E11.5 PGCs (arrowheads) observed in single-cell suspensions of genital ridges preextracted with detergent before fixation. SSEA1, stage-specific embryonic antigen 1 (a cell surface marker); DAPI, 4′,6′-diamidino-2-phenylindole. (C) Kinetics of BER activation with respect to chromatin changes in PGCs at E11.25 to E11.5. (See also fig. S1C.) Immunofluorescence analysis of single-cell suspension (top), or after preextraction to detect chromatin-bound XRCC1 (bottom). A progressive loss of H1 follows between E11.25 and E11.5, but XRCC1 and PAR signals persist in PGCs lacking H1. The kinetics of the process with respect to loss of 5mC are depicted at the bottom. (Scale bar, 10 μm.)

Expression of ERCC1 (excision repair cross-complementing rodent repair deficiency, complementation group 1), a core NER component, occurs at low levels in PGCs and neighboring somatic cells at the time of epigenetic reprogramming compared with control ultraviolet light–irradiated primary embryonic fibroblasts (PEFs), where we observed a dose-dependent increase and nuclear localization of ERCC1 (fig. S3). Although we detected XPA—another NER component—in both somatic cells and PGCs (fig. S4A), it was not chromatin bound and, hence, was inactive (7) (fig. S4B). Thus, the response of the NER pathway is not triggered during the reprogramming process in PGCs.

XRCC1 (x-ray repair complementing defective repair in Chinese hamster cells 1), a core component of the BER pathway (8), is present in PGC nuclei between E10.5 and E12.5 (fig. S5A), as is PARP1 [poly(ADP-ribose) polymerase family, member 1] and APE1 (apurinic/apyrimidinic endonuclease) (fig. S5, B and C). XRCC1 is a soluble nuclear factor, which binds to DNA when single-stranded DNA (ssDNA) breaks occur (8). We determined the amount of chromatin-bound XRCC1 in gonadal PGCs using the preextraction method (9). Whereas we observed an overall enrichment of XRCC1 in PGCs during E10.5 to E12.5, we found an enhancement in chromatin-bound XRCC1, specifically in PGCs at E11.5, which coincides with the stage at which genome-wide DNA demethylation occurs (Fig. 1B), which suggested that ssDNA breaks are present (10). We also detected high levels of PAR polymer, a product of activated PARP1 enzyme and an additional marker of active BER (8, 11), specifically in E11.5 PGCs (Fig. 1C). The presence of activated BER in PGCs during ongoing epigenetic reprogramming suggests that DNA demethylation in PGCs may be linked to the DNA repair pathway (2).

In PGCs spanning a period of ~6 hours, between E11.25 and E11.5, we detected increasing levels of PAR before the loss of signal for histone H1 (Fig. 1C). Because histone H1 is a target for PARP1 ribosylation (11, 12) and PARP1 itself has been shown to displace H1 (13), it is possible that high levels of PARP1-mediated poly(ADP-ribose) (PAR) synthesis in the nuclei of PGCs might be involved in H1 displacement. Additionally, we observed a correlation between high nuclear PAR signals and the disappearance of chromocenters in PGCs (fig. S5D) (2), which is consistent with a proposed role for PARP1 in the regulation of higher-order chromatin structure (11, 14, 15).

Concomitantly with the appearance of the PAR signal in the nuclei of PGCs, we detected high levels of chromatin-bound XRCC1, which suggested the presence of ssDNA breaks in nuclei of PGCs undergoing the DNA demethylation process (Fig. 1C) (10). The chromatin-bound XRCC1 was detectable in PGCs before the loss of H1, consistent with the loss of 5mC before the complete disappearance of H1 staining (2). Thus, we detect a transient population of PGCs containing PAR and chromatin-bound XRCC1 and H1, in the course of epigenetic reprogramming in PGCs (Fig. 1C).

We wished to establish whether there is a direct mechanistic link between BER and DNA demethylation. However, because PGCs are difficult to culture and manipulate in vitro, we examined mouse zygotes where there is also active genome-wide DNA demethylation that specifically affects the paternal, but not the maternal, pronucleus present in the same cell (16, 17). First, similar to our observations on PGCs, we found negligible levels of ERCC1 (fig. S6) and chromatin-bound XPA throughout zygotic development (fig. S7). In contrast, we found high levels of PARP1 and APE1 enzymes in zygotic pronuclei, as well as accumulation of PAR (figs. S8 and S9). Also, we observed high levels of XRCC1 in both parental pronuclei (Fig. 2A). However, after preextraction of soluble proteins, we detected bound XRCC1 protein only in the male pronucleus (Fig. 2B), which suggested the existence of ssDNA breaks localized to the paternal genome but not to the maternal pronucleus in the same cell. The asymmetric chromatin-bound XRCC1 in the male genome is detectable from early pronuclear stage (PN) 3, which coincides with the onset of DNA demethylation (Fig. 2B) (17). Although we detected high levels of PAR in both the male and female pronuclei (fig. S8B), PARP1 protein was detected predominantly in the paternal pronucleus (fig. S9). It is possible that the PAR in the maternal pronucleus has a different function and is a product of a different enzyme of the PARP family.

Fig. 2

Chromatin-bound XRCC1 in male pronucleus in zygotes. (A) Total XRCC1 in male and female pronuclei. (B) Chromatin-bound XRCC1 in the male pronucleus from PN3 onward. (C) Chromatin-bound XRCC1 in zygotes from wild-type and Stella null females. ♀, maternal pronucleus; ♂, paternal pronucleus; pb, polar body. (Scale bar, 10 μm.)

To establish that the activation of the BER pathway was specifically associated with DNA demethylation, we wanted to exclude other possible triggers, including protamine-histone exchange and DNA replication. Protamine-histone exchange occurs in the zygote during PN1 (fig. S10) (18), which is substantially before the onset of DNA demethylation or the detection of active BER. We also observed high levels of APE1 and chromatin-bound XRCC1 in the presence of aphidicolin, an inhibitor of replicative DNA polymerase, indicating that the process occurs independently of DNA replication (fig. S11) (16, 19, 20).

The epigenetic asymmetry of DNA demethylation in zygotes is promoted, at least in part, by the maternal inheritance of the Stella protein (21); absence of Stella results in aberrantly targeted DNA demethylation to both the maternal and paternal pronuclei. In zygotes from Stella null females, the activation of BER components is detectable in both parental pronuclei. Chromatin-bound XRCC1 was clearly detectable in both pronuclei of Stella-depleted zygotes in contrast to the wild-type controls, where the chromatin-bound XRCC1 was predominantly confined to the paternal pronucleus (Fig. 2C and fig. S12), which further indicated that activation of BER is linked to DNA demethylation in vivo.

We predicted that small-molecule inhibitors of key BER components should impede the progression of DNA demethylation. Addition of PARP inhibitor 3-aminobenzamide (3AB) (22), ABT-888, or the APE1 inhibitor CRT0044876 (APE1-i) (23) in the fertilization medium resulted in zygotes with significantly higher levels of DNA methylation in the paternal pronucleus as judged by 5mC staining (Fig. 3 and fig. S13, A and B), without affecting DNA methylation levels in the maternal pronucleus. We anticipate that the presence of BER inhibitors would inhibit and lower the processivity of DNA repair, which would prevent initiation of DNA demethylation particularly on the opposite DNA strand (5) and so account for the higher levels of DNA methylation in the paternal pronucleus. This effect on DNA methylation was further confirmed by bisulfite sequencing of the repetitive Line1 elements (Fig. 3 and fig. S13C). The use of APE1-i is likely to lead to the persistence of abasic sites in the template DNA that will prevent amplification of the affected molecules. Consistent with this, we observed amplification skewing, as judged by a shift in the relative proportions of the Line1 subclasses in our amplicons, which may also explain an apparent lack of changes in Line1 methylation in our bisulfite analysis. The inhibitors did not affect development of the zygotes, as judged by the rate of fertilization and progression through the pronuclear stages (PN1 to 5), and after the removal of the inhibitors, the zygotes resumed cleavage division.

Fig. 3

Inhibition of BER affects DNA demethylation in zygotes. (A and C) PARP inhibitor 3AB and APE1 inhibitor (CRT0044876) impede progression of DNA demethylation, as detected by 5mC staining. (B and D) Quantification of 5mC staining shown as a ratio between 5mC signal from paternal pronuclei relative to the signal from maternal pronuclei. DMSO, dimethyl sulfoxide. (Statistical analysis by Student’s t test.) (E) Bisulfite analysis of Line1 repetitive elements indicating the percentage of methylated CpGs. Each line represents a unique DNA clone; filled and open circles represent methylated and unmethylated CpGs, respectively. ♀, maternal pronucleus; ♂, paternal pronucleus; pb, polar body. (Scale bar, 10 μ3.)

We demonstrate that the genome-wide active DNA demethylation is mechanistically linked to the activation of BER and suggest that the present ssDNA breaks are likely to trigger extensive chromatin remodeling and histone exchange in PGCs (2). In plants, BER is triggered during active DNA demethylation in response to the excision of 5mC by methylcytosine-specific DNA glycosylases (4, 24), but similar enzymes are as yet unknown in animals. Alternatively, deamination of 5mC by AID (activation-induced deaminase) or APOBEC (apolipoprotein B mRNA-editing enzyme catalytic polypeptide-like) enzymes resulting in T-G mismatches might trigger BER and result in a loss of 5mC. However, we failed to detect expression of Aicda in the germ line between E8.5 and E11.5 (Fig. 4 and not shown)(25), which was consistent with the limited effect of loss of AID on DNA demethylation in PGCs (26), whereas Apobec1 expression diminishes in PGCs before E11.5 (Fig. 4). Furthermore, either the T-G mismatch glycosylases are not consistently detectable in germ cells and zygotes (figs. S14 and S15) or the loss of function has no effect on germline development (27). We also exclude the proposed deamination driven by the DNA methyltransferases Dnmt3a or Dnmt3b (28), because E11.5 PGCs lack both these enzymes. Thus, cumulative evidence does not support a major role for 5mC deamination in germ cells. An alternative possibility includes modification of 5mC to 5-hydroxymethylcytosine (5hmC) (fig. S16) (29, 30), which could be recognized by a glycosylase. We detected significant expression of Tet1, which encodes a key enzyme for this modification (30) (Fig. 4), coincidentally with the transcription of BER components in E11.5 PGCs (fig. S2). However, it is also possible that 5hmC modification plays a different role in the context of resetting the epigenome after the loss of 5mC.

Fig. 4

Expression of 5mC-modifying enzymes and the molecular pathway of epigenetic reprogramming. Expression analysis of Tet1 and Tet2, which have been implicated in 5mC to 5hmC conversion; Tet1 shows significant expression in PGCs at E11.5 and E12.5. Aicda is undetectable in PGCs at the time of reprogramming, whereas the levels of Apobec1 diminish after E10.5. (Error bars show standard deviation between technical triplicates.) The gene for glyceraldehyde-3-phosphate dehydrogenase is Gapdh.

Supporting Online Material

www.sciencemag.org/cgi/content/full/329/5987/78/DC1

Materials and Methods

Figs. S1 to S16

References

  • * These authors contributed equally to this work.

References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. We thank Y. Galanty, S. Polo, and members of the Surani lab for stimulating discussions. S.J.J. was a NIH–Cambridge Health Science Scholar. This work was funded by grants from the Wellcome Trust to M.A.S.
View Abstract

Navigate This Article