Research | Open | Published:
Dicer regulates Xist promoter methylation in ES cells indirectly through transcriptional control of Dnmt3a
Epigenetics & Chromatinvolume 1, Article number: 2 (2008)
X chromosome inactivation is the mechanism used in mammals to achieve dosage compensation of X-linked genes in XX females relative to XY males. Chromosome silencing is triggered in cis by expression of the non-coding RNA Xist. As such, correct regulation of the Xist gene promoter is required to establish appropriate X chromosome activity both in males and females. Studies to date have demonstrated co-transcription of an antisense RNA Tsix and low-level sense transcription prior to onset of X inactivation. The balance of sense and antisense RNA is important in determining the probability that a given Xist allele will be expressed, termed the X inactivation choice, when X inactivation commences.
Here we investigate further the mechanism of Xist promoter regulation. We demonstrate that both sense and antisense transcription modulate Xist promoter DNA methylation in undifferentiated embryonic stem (ES) cells, suggesting a possible mechanistic basis for influencing X chromosome choice. Given the involvement of sense and antisense RNAs in promoter methylation, we investigate a possible role for the RNA interference (RNAi) pathway. We show that the Xist promoter is hypomethylated in ES cells deficient for the essential RNAi enzyme Dicer, but that this effect is probably a secondary consequence of reduced levels of de novo DNA methyltransferases in these cells. Consistent with this we find that Dicer-deficient XY and XX embryos show appropriate Xist expression patterns, indicating that Xist gene regulation has not been perturbed.
We conclude that Xist promoter methylation prior to the onset of random X chromosome inactivation is influenced by relative levels of sense and antisense transcription but that this probably occurs independent of the RNAi pathway. We discuss the implications for this data in terms of understanding Xist gene regulation and X chromosome choice in random X chromosome inactivation.
X chromosome inactivation is the mechanism used in mammals to achieve dosage compensation of X-linked genes in XX females relative to XY males. Early in development all cells in female embryos inactivate most genes on one of the two X chromosomes. In embryonic lineages X inactivation is normally random, with an equal probability of either the maternal or paternal X undergoing X inactivation in any given cell. In certain extraembryonic lineages it is always the paternal X that is inactivated, referred to as imprinted X inactivation. Following the establishment of X inactivation in early embryogenesis the inactive state is stably maintained through all subsequent cell generations (reviewed in ).
X inactivation is triggered by the expression of the X inactive specific transcript (Xist), an unusual non-coding RNA that has the unique property of binding to and coating the chromosome from which it is transcribed. Xist RNA is thought to recruit silencing factors that modify the chromatin, bringing about a mitotically stable heterochromatic configuration that can be propagated through subsequent cell divisions (reviewed in ).
Establishing appropriate X inactivation patterns requires mechanisms that ensure correct regulation of Xist RNA in early development. Specifically Xist expression must occur from only one allele in XX cells and not at all in XY cells. Studies to date indicate that regulation of Xist expression in random X inactivation is complex. An antisense RNA, termed Tsix, is important for maintaining the Xist gene in a poised state prior to the onset of X inactivation through a mechanism that is thought to involve establishment of repressive chromatin marks and/or DNA methylation over the Xist promoter [3–6]. Tsix-mediated repression in turn regulates the probability of a given Xist allele being expressed in XX heterozygotes [7–11]. Concurrent with Tsix expression there is a low level of sense transcription from the Xist promoter [6, 12]. Enhanced sense transcription initiated from ectopic sites upstream of Xist antagonises Tsix and renders that chromosome more likely to be inactivated in XX heterozygotes [12, 13].
In undifferentitated embryonic stem (ES) cells, the absence of Tsix expression or of DNA methylation leads to only a modest increase in Xist expression [8, 9, 11, 14]. Conversely, following the onset of cellular differentiation, both Tsix-deficient and DNA methylation-deficient ES cells upregulate Xist inappropriately, that is, from the single X chromosome in XY cells [7, 14–17]. This suggests the existence of alternative, possibly redundant mechanisms for regulating Xist expression. Indeed studies on X chromosome reactivation in the inner cell mass [18, 19], and also in developing primordial germ cells , suggest that an overlapping regulatory pathway, specific for pluripotent cells (including ES cells), either represses Xist transcription, or alternatively reduces levels of a critical positive regulator of Xist expression.
Interwoven into this complex regulatory circuitry there is an additional pathway that ensures only a single Xist gene is expressed in XX cells, and that the single Xist allele in XY cells remains repressed. The classical model for this process invokes the presence of a blocking factor present at limiting levels such that only a single Xist allele is blocked in each cell . More recently evidence has emerged that trans- interactions between Xist alleles are important in the allelic control of Xist expression [22–24]. Also, in a recent study it has been proposed that the gene encoding a critical positive regulator is located close to Xist. Increased levels of this factor in the cells of early embryos with more than one X chromosome are proposed to attain a threshold level that allows expression of Xist .
The RNA interference (RNAi) pathway is found in organisms as diverse as fission yeast and mammals (reviewed in ). The pathway governs a range of mechanisms that regulate gene expression at the level of RNA translation/stability (post-transcriptional gene silencing) and at the level of transcription/chromatin structure (transcriptional gene silencing). Given the involvement of non-coding RNAs in X inactivation, a possible link with the RNAi pathway, either in initiation, propagation or maintenance of X inactivation, has been conjectured. A role for RNAi in the initiation of X inactivation in particular is suggested by the presence, prior to the onset of X inactivation, of overlapping sense and antisense RNAs at the Xist locus. We previously provided evidence that RNAi does not play a role in the maintenance of X inactivation using conditional deletion of the gene encoding Dicer, an RNase III enzyme that is essential for the RNAi pathway, in T-lymphocytes . In this study we further explore the role of RNAi in the initiation and propagation of X inactivation. We show that both sense and antisense transcription across the Xist promoter influence levels of DNA methylation, consistent with a dsRNA-mediated mechanism. Using conditional Dicer knockout ES cells we demonstrate that Dicer deletion leads to hypomethylation of the Xist gene promoter. However, we also find that hypomethylation of the genome occurs more widely and is attributable to reduced levels of de novo methyltransferases, most notably Dnmt3a, in ES cells. Moreover, analysing Dicer-deficient embryos at E6.5 we demonstrate that the initiation of monoallelic Xist expression and spreading of Xist RNA occur normally. We conclude that the RNAi pathway does not play a critical role in the X inactivation process.
Sense transcription across the Xist promoter influences Xist promoter methylation in undifferentiated ES cells
We and others have previously shown that sense as well as antisense transcription across the Xist locus prior to the onset of X inactivation plays a role in X chromosome inactivation choice [7–9, 11–13]. The mechanism for this is unknown although there is evidence that Tsix has an influence on chromatin structure and DNA methylation of the Xist promoter.
Evidence to date indicates that antisense Tsix transcription has an influence on Xist promoter DNA methylation in differentiating ES cells , and in somatic cells [4, 6, 17], but not in undifferentiated ES cells that are representative of Xist promoter status prior to the onset of X inactivation [4, 6, 17]. To determine whether sense transcription can influence DNA methylation of the Xist promoter prior to X inactivation we analysed two XY ES cell lines carrying mutations in the Xist 5' region, initially using conventional methylation sensitive restriction enzyme site (MSRE) analysis. The first mutation, Δ5', is a deletion of a 9 kb region 1.1 kb upstream of the Xist transcriptional start site (TSS). The second is an insertion of a transcriptional termination site SPA-MAZ4 in the region -1.1 kb relative to the TSS  (Figure 1A). Both of these mutations show enhanced sense transcription in undifferentiated XY ES cells correlating with preferential inactivation of the mutant X chromosome in vivo . In accordance with previous findings , the Xist promoter was found to be highly methylated in undifferentiated wild-type (wt) XY ES cells (Figure 1A and 1B). Interestingly, both the Δ5'+neo and SPA+neo XY ES cell lines showed significant hypomethylation at all of the CpG sites analysed, that is, HpaII, HaeII, HhaI, MluI and SacII (Figure 1B). Quantification of the bands using ImageQuant software demonstrated methylation loss within a range of 20–45% for various CpG sites, with the SPA+neo mutant affected more severely (Figure 1C). This indicates that enhanced sense transcription across the Xist promoter can lead to CpG hypomethylation.
We then analysed another Xist mutation, XT67E1 , a deletion of the most part of Xist exon 1 and the minimal promoter region on the 129 allele in Pgk12.1 XX ES cells (Figure 1A). Whilst the deletion in XT67E1 cells removes a number of methylatable CpG sites in the Xist promoter, sites 36 bp upstream of the TSS are retained. Moreover, the deletion gives rise to a change in size of a BamHI fragment in the Xist 5' region, and we were therefore able to discriminate between wt and mutant alleles (Figure 1A and 1D). MSRE analysis revealed complete hypomethylation on the mutant allele. The wt allele was mosaically methylated, similar to the parental XX ES cell line Pgk12.1 .
As the Xist TSS is deleted on the mutant allele in XT67E1 cells we anticipated that sense transcription would not be detectable but that antisense Tsix transcription would be unaltered. However, analysis by strand specific polymerase chain reaction (PCR) revealed unexpectedly that the mutant allele transcribes both sense and antisense RNAs and moreover that sense transcripts are abundant relative to the parental Pgk12.1 cell line (Figure 1E, amplicon 4). To verify this result we designed primers that were able to discriminate between mutant and wt alleles; the forward primer was the same for both alleles and was located 59–87 bp upstream of the Xist TSS. The reverse primers were located either in exon 1 (TN51, wt allele) or at the 3'end of the neomycin selective cassette (neoTN9 ≡ 51mut, mutant allele). Strand- and allele-specific reverse transcription (RT) PCR with these primers clearly demonstrates ectopic sense transcription on the mutant allele (Figure 1E). The PGK promoter that drives expression of the neomycin resistance gene cannot be the origin of ectopic transcripts as it is located about 1.7 kb downstream from the region analysed and in a reverse orientation relative to Xist. This suggests that sense transcription is activated from a minor upstream Xist TSS, in line with a result reported previously . Importantly this result reinforces the conclusion from analysis of Δ5'+neo and SPA+neo mutant ES cells that enhanced sense transcription initiated upstream of Xist antagonises Xist promoter methylation.
Hypomethylation of the Xist CpG island in undifferentiated ES cells correlates with X inactivation skewing
To investigate Xist promoter methylation in more detail we used SEQUENOM matrix-assisted laser desorption/ionisation time of flight (MALDI-TOF) mass spectrometry analysis of bisulphite-modified DNA . This approach allowed us to analyse methylatable CpGs more widely and also to obtain accurate quantitative measurement of CpG methylation levels. To validate the method we first analysed the methylation patterns in control XX and XY somatic and ES cell lines (Figure 2A and 2B). As expected, CpG methylation was close to 100% in XY male somatic cells and about 50% in XX female somatic cells, representing the average of a fully methylated inactive Xist locus and a fully unmethylated active locus on the inactive X chromosome . In XY ES cells methylation was close to 100% in Xist region 1, although somewhat lower in region 2. Xist was significantly hypomethylated in XX ES cells, consistent with our previous observations .
Having validated the SEQUENOM method we went on to analyse Xist promoter methylation for the mutant XY ES cell lines described above, and additionally in Δhs XY mutant ES cell lines where ectopic transcription-dependent skewing of X inactivation has also been reported to occur . We found that deletion of 9 kb of the Xist upstream region (Δ5') leads to around 20% loss of methylation in both region 1 and region 2. This was also the case for mutants carrying the PGKneo cassette, as well as for Δneo ES cells (Figure 2C). More severe loss of methylation was observed for SPA+neo mutant (Figure 2D), consistent with the MSRE analysis above (Figure 1B and 1C). Moderate hypomethylation was observed for the Δhs+neo, but not for the Δhs Δneo mutant (Figure 2E). Strikingly, all of the mutants that show hypomethylation of the Xist CpG island also show preferential X inactivation of the mutant allele in female heterozygotes [12, 13]. Conversely, the Δhs Δneo mutation that does not affect the randomness of X inactivation shows no hypomethylation. Collectively, these results demonstrate a direct correlation between hypomethylation of the Xist promoter and an increased probability of that chromosome being chosen as the inactive X in heterozygous females.
It was previously reported that abolition of Tsix transcription does not cause Xist promoter hypomethylation in undifferentiated ES cells, although analysis was restricted to two MSRE sites in region I . To further address this issue we used the SEQUENOM assay to assess Xist promoter methylation in two different Tsix mutant ES cell lines, pSS1Δ2.7 and pAA2Δ1.7 . In the first mutant, pSS1Δ2.7, Tsix exon 1 was deleted, but this had no effect on Tsix transcription or function. In the second mutant, pAA2Δ1.7, Tsix transcription through the Xist locus was abolished, causing primary non-random inactivation of the targeted allele in female embryos. We analysed three independent pAA2Δ1.7 XY ES cell clones, and in all cases we observed clear hypomethylation of the Xist promoter (Figure 2F). Hypomethylation in CpG region I was moderate compared with CpG region 2, possibly accounting for the fact that Sun et al.  did not observe this result. No methylation differences were observed for the pSS1Δ2.7 mutant, consistent with normal Tsix transcription and random X inactivation in female heterozygotes. This result suggests that prior to the onset of X inactivation, Tsix transcription, possibly together with physiological levels of sense Xist transcription, determines Xist promoter CpG methylation levels that, in turn, have an impact on the probability of that X chromosome being selected as the inactive X during the onset of random X inactivation. Increasing sense transcription, or possibly utilisation of a heterologous upstream promoter, antagonises promoter CpG methylation such that there is an increased probability of that X chromosome being selected as the inactive X in a heterozygous female.
The role of the RNAi pathway in initiation of X inactivation: production of conditional Dicer knockout ES cells
What is the mechanism by which sense and antisense transcription impact on CpG methylation and repress the Xist promoter? One possibility is that sense and antisense RNAs trigger an RNAi response, similar for example to RNA-dependent DNA methylation in higher plants . Alternatively, an RNAi-independent mechanism involving the antisense RNA Tsix, or both sense and antisense RNAs, may operate. The fact that increased sense transcription reduces Tsix-dependent Xist promoter methylation could be interpreted to indicate that silencing does not depend on dsRNA production. However, this does not rule out that the low levels of endogenous sense transcription from the Xist promoter co-operates with antisense Tsix RNA in RNAi-mediated silencing. To investigate this further we set out to analyse ES cells deficient for the RNase III enzyme Dicer, which is essential for the RNAi pathway in mammalian cells.
We set out to derive an ES cell line in which the gene encoding Dicer can be deleted conditionally using CRE/loxP, allowing us to discriminate between primary effects attributable to Dicer deletion and secondary effects resulting from derivation and long-term culture of ES cells deficient for this essential factor. We established the D41 XY ES cell line, in which the RNase III domain was flanked with loxP sites on both Dicer alleles . Initially, Dicer-deficient ES cell lines were derived from the D41clone by transfection with a pCAG-Mer-Cre-Mer, tamoxifen-inducible Cre recombinase plasmid (see Methods for details), followed by hydroxytamoxifen (4-OHT) treatment (Figure 3A). The efficiency of Cre-recombination was low, and DicerΔ/Δ colonies were quickly overgrown by heterozygous Dicerlox/Δ colonies. Nevertheless, we isolated three independent sub-clones, S5, S6 and E5, that had deleted the Dicer RNase III domain on both alleles (Figure 3B).
To ensure that deletion of the RNase III domain completely abolished Dicer function we performed Northern blot hybridisation of RNA from floxed and Dicer-deficient clones with a probe for the micro RNA, miR-292 (Figure 3C). The result confirmed the absence of miR-292 and enrichment of the corresponding pre-miRNA in the S5 and S6 Dicer mutant clones relative to control floxed cell lines.
We went on to characterise Dicer-deficient ES cell lines. In accordance with previous observations , DicerΔ/Δ clones overexpress major satellite repeats (data not shown) and are not capable of differentiation. We attempted to differentiate the cells by withdrawing LIF. In contrast to parental Dicerlox/loxcells, D3 DicerΔ/Δ clones did not form embryoid bodies, but stayed in irregular-shaped clumps which subsequently attached and continued to grow. After 11 days of differentiation levels of the pluripotent ES cell markers Oct4, Nanog, Fgf4 and Errβ expression levels remained unchanged (Additional file 1). Interestingly, T/Brachyury, which is usually expressed at low levels in wt ES cells, presumably due to a small number of differentiated cells, was completely absent in DicerΔ/Δ clones, and did not appear even after culturing cells for 11 days under differentiation conditions. This result suggests that Dicer-deficient ES cells are either unable to differentiate or alternatively that differentiated cells present in the cultures do not survive.
Hypomethylation of the Xist promoter in Dicer-deficient XY ES cells
To determine whether Dicer deficiency affects Xist promoter methylation, we performed MSRE analysis of DNA from control and DicerΔ/Δ clones (Figure 3D and 3E). Interestingly, all DicerΔ/Δ clones demonstrated partial loss of methylation at all of the restriction sites analysed. It should be noted, however, that different clones showed varying degrees of methylation loss, with S5 showing the highest percentage of hypomethylation and S6 the lowest (Figure 3E and 3F).
The cell lines used in this preliminary analysis underwent several rounds of cloning and selection during the derivation process, so we proceeded to derive further lines, in this case making use of ES cells carrying floxed Dicer alleles and a tamoxifen inducible Cre recombinase gene targeted into the Rosa26 locus . In this system Cre-recombinase-mediated deletion of the floxed cassette was very efficient following addition of tamoxifen, and we were able to select several individual clones from two different parental Dicerlox/lox cell lines, DTCM23 and DTCM49. In addition, we established Dicer-deficient cells from pools of 200–250 colonies of tamoxifen treated cells (Figure 3A). All further analyses were performed in parallel on clones derived by both approaches.
To determine Xist promoter hypomethylation quantitatively we analysed bisulphite-modified DNA from the D3lox/lox and S5, S6, E5 DicerΔ/Δ clones using SEQUENOM MALDI-TOF mass spectrometry analysis. D3-derived DicerΔ/Δ clones showed substantial hypomethylation both in region 1 and region 2 (Figure 4A). The D3lox/lox parental clone also showed moderate loss of methylation in region 1 and significant loss in region 2, but less than in DicerΔ/Δ clones. The reason for this remains unknown, but could be explained by a mutation or rearrangement that occurred during the several rounds of cell selection to which the D3lox/lox cells were subjected. To exclude the possibility of such an unrelated mutation causing the observed hypomethylation phenotype, we performed mass spectrometry analysis for the DTCM23lox/lox parental cell line and for DicerΔ/Δ clones 23ΔE3, 23ΔF4 and the 23Δpool. The DTCM23lox/lox cell line showed a methylation pattern similar to the 129/1 XY ES control, while all DicerΔ/Δ clones demonstrated hypomethylation most notably in region 2 (Figure 4B). A similar result was obtained for another set of clones, DTCM49lox/lox and DicerΔ/Δ derivatives (Additional file 2A).
To determine the dynamics of methylation loss, we treated the DTCM23lox/lox and DTCM49lox/lox cell lines with 4-OHT and collected DNA for SEQUENOM analysis 50 and 168 hours later. The result shows that Xist promoter hypomethylation occurs rapidly following the deletion of Dicer and that further hypomethylation occurs with continued cell passaging, again most notably in region 2 (Figure 4C and Additional file 2B).
Xist expression is elevated in Dicer-deficient XY ES cells
To test how hypomethylation affects the transcriptional status of the Xist promoter in Dicer-deficient ES cell lines, we performed RNA fluorescent in situ hybridisation (FISH) analysis on the S5 DicerΔ/Δ ES clone using probes detecting Xist and Tsix transcripts. The majority of the S5 cells showed a single pinpoint signal similar to the control 129/1 XY ES cells. However, occasionally there were cells with an upregulated Xist signal, which either 'painted' the X chromosome (Figure 5A, third panel from the top) or was scattered in the vicinity of the chromosome (Figure 5A, bottom panel). On average about 10% of cells showed this upregulation pattern, confirming that hypomethylation of the Xist promoter compromised regulation of Xist expression.
Next we analysed Xist expression quantitatively for all Dicerlox/lox control and DicerΔ/Δ ES clones. The data for Xist expression was normalised to β-actin and then to the level of the Xist transcript in control 129/1 XY ES cells and is presented in Figure 5B. All DicerΔ/Δ ES clones showed an elevated level of Xist expression in comparison with their corresponding parental floxed cell lines; however, the absolute level of Xist upregulation varied between individual clones. It is worth noting that while Xist expression in DTCM23lox/lox and DTCM49lox/lox was the same as 129/1 control, the D3lox/lox clone showed elevated expression, consistent with the observed promoter hypomethylation.
Hypomethylation of the Xist promoter in Dicer-deficient XY ES cells correlates with depletion of de novo DNA methyltransferases
Xist promoter hypomethylation in Dicer-deficient ES cells could be due to a direct effect on recruitment of DNA methyltransferases (Dnmts), for example mediated by the RNAi pathway. Alternatively sense and/or antisense transcription may be required to establish other features of the underlying chromatin structure at the Xist promoter, for example specific histone lysine methylation marks, which in turn could have an indirect impact on the recruitment of Dnmts. To test the indirect model we analysed the repressive histone modifications H3K9me2 (data not shown), H3K27me3, H4K20me3 as well as the active mark H3K4me2 over the Xist locus in wt and Dicer-deficient ES cells using chromatin immunoprecipitation (ChIP). None of these histone modifications showed a significant change in Dicer-deficient ES cells (Additional files 3 and 4).
The absence of detectable changes in histone modification in DicerΔ/Δ cells relative to floxed parental cells suggested that hypomethylation results from a direct effect on recruitment of Dnmts. It was reported previously that Xist promoter methylation is mediated by de novo DNA methyltransferases Dnmt3a and/or Dnmt3b . We therefore went on to analyse the levels of these enzymes, and also of the maintenance methyltransferase Dnmt1, using Western blotting. In agreement with previous observations Dnmt3a and Dnmt3b levels were very low in XX compared with XY control ES cell lines . Interestingly we also observed reduced levels of Dnmt3a in DicerΔ/Δ clones in comparison with Dicerlox/lox controls (Figure 6A–C). The most affected clone was S5, which showed approximately five times less Dnmt3a protein in comparison with the D3 control. The DTCM23 and DTCM49 sets of DicerΔ/Δ clones demonstrated a depletion of Dnmt3a and also a slight decrease in Dnmt3b levels.
To determine whether Dnmt3a/b depletion resulted from transcriptional or post-transcriptional regulation, we performed quantitative RT-PCR analysis with primers designed for Dnmt1, Dnmt3a2 (the major Dnmt3a isoform in ES cells), Dnmt3b (for all Dnmt3b isoforms) and Dnmt3L (Figure 6D–F). Consistent with the Western data, we found that levels of Dnmt3a2 transcript are consistently lower in DicerΔ/Δ clones than in controls. We also found that the level of Dnmt3L, a functional partner of Dnmt3a2, is much reduced in Dicer-deficient clones. The level of Dnmt3b was reduced in the DTCM series of clones, but not in the S5, S6 and E5 DicerΔ/Δ clones, in accordance with the Western blot results. The level of Dnmt1 was not significantly different between control and Dicer-deficient clones. Affymetrix microarray analysis of RNA from D3lox/lox versus S5 DicerΔ/Δ clones also showed a 2.4-fold decrease of Dnmt3a and 3.3-fold decrease of Dnmt3L (data not shown).
To determine whether hypomethylation occurs at other loci in DicerΔ/Δ clones we analysed methylation at the differentially methylated regions (DMRs) of two imprinted genes, H19 and Igf2rAir (Additional file 5). In both examples we observed hypomethylation specifically in DicerΔ/Δ clones. It should be noted that two recent studies reported hypomethylation of repetitive and unique sequences in DicerΔ/Δ ES cells and attributed this to reduced levels of Dnmts [35, 36].
Given the requirement for Dnmt3a/b in Xist promoter methylation [31, 34, 37], we conclude that hypomethylation in DicerΔ/Δ clones is most likely attributable to decreased levels of expression of these enzymes rather than deficiency of a dsRNA-mediated transcriptional gene silencing mechanism.
Monoallelic Xist expression and Xist RNA spreading in Dicer-deficient XX embryos
Finally, we wished to test for a role for Dicer in the initiation of random X inactivation in XX cells and also to determine whether the RNAi pathway is important in spreading Xist RNA on the inactive X chromosome. As Dicer-deficient ES cells are not capable of differentiation and we have not been able to isolate a stable Dicer-deficient XX ES cell line, we analysed Dicer-deficient XX embryos produced by mating Dicerwt/Δ heterozygous mice. In general, DicerΔ/Δ embryos survived until approximately E7.5–E8.5 and were smaller than their heterozygous or wt littermates, consistent with a study published previously . This afforded the opportunity to analyse the initiation of random X inactivation that commences at approximately E5.5. We carried out whole mount RNA FISH using Xist and Tsix probes on E6.5 embryos (Figure 7). The DicerΔ/Δ XY embryos showed a Xist/Tsix pinpoint signal similar to their wt and heterozygote XY littermates, while the DicerΔ/Δ female embryos had both pinpoint and accumulated Xist transcripts, suggesting that Dicer is not affecting the initiation step of random X inactivation in the inner cell mass (ICM). The presence of Xist clouds in cells of XX embryos implies that spreading of Xist RNA also does not require Dicer activity. We did note that XX embryos showed weaker and more disrupted Xist signal with stronger general background for both Xist and Tsix probes, and that this varied from embryo to embryo. This is likely due to the onset of embryo lethality and apoptosis in mutant embryos. Overall these observations further support our conclusion that X inactivation can occur independent of the RNAi pathway.
In this study we set out to further explore the mechanism of regulation of Xist gene expression at the onset of random X inactivation. In particular, we wished to understand the importance of sense and antisense transcription in Xist promoter repression and the possible involvement of the RNAi pathway. We demonstrate that enhanced sense transcription in ES cells prior to the onset of X inactivation reduces the levels of DNA methylation at the Xist promoter. In addition, we found that antisense Tsix transcription is required for Xist promoter methylation in undifferentiated ES cells. DNA hypomethylation of the Xist promoter was also observed in Dicer-deficient ES cells but further analysis indicated this is an indirect effect attributable to reduced levels of the de novo methyltransferases Dnmt3a, Dnmt3b and Dnmt3L. Similarly, Dicer-deficient embryos displayed normal allelic regulation of Xist/Tsix expression and, moreover, Xist RNA spreading occurred on a single X chromosome in Dicer-deficient XX embryos. We conclude that Tsix-mediated regulation of Xist expression and other steps in the X inactivation process occur independent of the RNAi pathway.
The role of sense and antisense transcription in regulating Xist promoter methylation
Previously we demonstrated that modified Xist alleles showing increased sense transcription from heterologous promoters in ES cells also show preferential X inactivation in XX heterozygous animals [12, 13]. Here we have extended this finding, showing that these modified Xist alleles are partially hypomethylated over the Xist promoter region, providing a mechanistic basis for preferential expression in XX heterozygotes. Similarly the mutant Xist allele in XT67E1 XX ES cells has ectopic transcription in the sense direction and complete hypomethylation of promoter CpG sites that lie immediately upstream of the deleted region. In this case hypomethylation cannot be correlated with increased probability of expression because the deleted allele is lacking the Xist TSS.
Enhanced sense transcription antagonises Xist promoter methylation even when normal levels of antisense Tsix RNA are present. However, our data show that Tsix transcription is important for the establishment of Xist promoter methylation in ES cells, that is, prior to X inactivation. This contrasts with a previous report in which a different Tsix mutant allele was suggested to have a role in Xist promoter methylation during ES cell differentiation, but not prior to onset of random X inactivation in undifferentiated ES cells . This discrepancy is explained in part by the fact that hypomethylation occurs more in region 2 than in region 1 (this study), and Sun et al.  analysed only region 1. A second possible factor is that independent Tsix mutant XY ES cells behave differently. Notably the Δ65, 2 lox, pAA2Δ1.7 and Δ34# 1 XY ES cell lines all upregulate Xist inappropriately upon differentiation [7, 9, 11, 16], presumably at least in part due to promoter hypomethylation, whereas the ΔCpG cell line maintains Xist repression throughout differentiation . It is possible that this difference arises because the secondary pathway, Xist repression linked to the pluripotency program , plays a more dominant role in the ΔCpG XY ES cell line.
Dicer indirectly regulates Xist promoter methylation in ES cells
We wanted to determine whether regulation of Xist promoter methylation by sense and antisense RNAs was mediated by the RNAi pathway. We found that Dicer-deficient ES cells show Xist promoter hypomethylation and moderate upregulation of Xist transcripts, an effect that was observed in a number of independent cell lines, albeit with some variation in degree. However a number of facts lead us to conclude that this is an indirect consequence of Dicer deletion. First and foremost we observed downregulaton of the de novo DNA methyltransferases, Dnmt3a, Dnmt3b and Dnmt3L in Dicer-deficient cells. Several studies have demonstrated that Dnmt3a/3b levels are important for maintaining Xist promoter methylation [31, 34, 37], indicating that reduced Dnmt levels are sufficient to account for Xist promoter hypomethylation in Dicer-deficient cells. Consistent with this idea we observed hypomethylation at imprinted loci, and two recent studies have reported hypomethylation of subtelomeric repeats  and promoters of Oct4, Tsp50 and Sox30 genes  in independently isolated Dicer-deficient cell lines. Importantly the hypomethylation phenotype seen in these studies was complemented by ectopic expression of Dnmt transgenes, indicating that the RNAi pathway is not directly involved. These studies also demonstrated that downregulation of Dnmts results from overexpression of Rbl2 that is, in turn, normally subject to negative regulation by the miR-290 cluster miRNAs. In agreement with this conclusion we also found that the level of Rbl2 transcript is upregulated 4.4-fold in the Dicer-deficient cell lines described here (data not shown).
A second line of evidence that argues that the RNAi pathway is not required for Xist gene regulation and random X inactivation comes from our analysis of Dicer-deficient embryos at early post-implantation stages. Here we observed appropriate Xist and Tsix expression patterns in XY and upregulation of Xist from a single allele in XX embryos. The fact that XX embryos show reduced intensity of staining of Xist domains most likely reflects that embryo lethality occurs shortly after the stage we examined, E6.5 . It should be noted that we cannot formally rule out that in XX embryos the pattern we observe represents persistence of the imprinted X inactivation pattern, that is, that Dicer deficiency results in failure to erase imprinted X inactivation prior to establishing random X inactivation.
Regulation of Xist promoter methylation
Given the evidence that the RNAi pathway does not mediate Xist promoter regulation through sense and antisense transcription, what are the alternative mechanisms? The fact that we observe hypomethylation in undifferentiated Tsix-deficient XY ES cells suggests a direct link between antisense transcription and promoter CpG methylation. That some CpG methylation is retained in the Tsix mutant cells may indicate a redundant mechanism for recruiting DNA methylation to the promoter, for example, relating to Xist repression by the pluripotency program [18, 20], or alternatively may simply reflect that the maintenance methyltransferase activity, Dnmt1, is sufficient to retain promoter methylation up to a defined level.
Assuming that Xist promoter methylation prior to the onset of X inactivation is indeed a consequence of antisense Tsix expression we can envisage two possible mechanisms. Either Tsix directly recruits de novo Dnmts, Dnmt3a and Dnmt3b, as has been suggested previously for Dnmt3a , or alternatively Tsix may mediate other chromatin changes at the Xist promoter, for example reducing H3K4 methylation as suggested previously , DNA hypomethylation being a secondary consequence. Whilst we are unable to distinguish between these possibilities at present, it is interesting to note that H3K4 methylation antagonises the binding of Dnmt3a/Dnmt3L dimers to nucleosome , providing a possible mechanism for hypomethylation of DNA driven by a Tsix-mediated reduction in H3K4 methylation levels. In the context of this model increased sense transcription from upstream heterologous or cryptic promoters may antagonise Xist promoter methylation by increasing the H3K4 methylation levels locally.
We have shown that sense and antisense transcription over the Xist promoter can modulate DNA methylation levels prior to the onset of random X inactivation, providing a mechanistic basis for skewed X inactivation patterns in mutants that alter sense or antisense RNA levels. We went on to investigate the possible involvement of the RNAi pathway. Our analysis of Dicer-deficient ES cells demonstrated Xist promoter hypomethylation, but this appears to be a secondary consequence of reduced levels of Dnmts. Consistent with this, appropriate Xist/Tsix expression patterns were seen to occur in Dicer-null embryos. Based on these observations we conclude that the RNAi pathway is probably not required for X chromosome inactivation in mammals.
ES cell line derivation and maintenance
Dicerlox/lox ES cell lines were derived from the ICM of E3.5 embryos using two approaches. In the first approach, ES cell lines were derived from mice homozygous for a Dicerlox allele. The established Dicerlox/lox XY ES cell line D41 was subsequently lipofected with pCAG-Mer-Cre-Mer plasmid, carrying tamoxifen inducible Cre-recombinase. Clone D41D3Cre (D3Cre) was treated with 800 nM 4-hydroxytamoxifen (4-OHT, Sigma), plated at clonal density, and individual colonies were picked, expanded and then tested by genomic PCR for the loss of the Dicer RNase III domain. Three clones S5, S6 and E5 were identified that showed loss of the Dicer RNase III floxed cassette.
In the second approach ES cell lines were derived from mice homozygous for a Dicerlox allele crossed to animals either homozygous or heterozygous for tamoxifen inducible Cre-recombinase targeted into the Rosa26 locus (obtained from Artemis Pharmaceuticals; ). Two derived parental XY ES cell lines, DTCM23 and DTCM49, were treated with 800 nM 4-OHT and plated at clonal density. Individual clones as well as pools of approximately 200–250 clones were genotyped for the loss of the RNase III floxed cassette. Dicer-deficient clones DTCM23 ΔE3, ΔF4, DTCM49 ΔA1 and ΔE2 as well as a pool of Dicer-deficient clones DTCM23 Δpool were selected for further analysis.
ES cells lines were derived and maintained on a feeder layer (mitomycin-inactivated primary mouse embryonic fibroblasts) in Dulbecco's Modified Eagle Medium (DMEM) supplemented with 10% foetal calf serum (FCS, Autogen Bioclear), 7% Knockout Serum Replacement (KSR), 2 mM L-glutamine, 1× non-essential amino acids, 50 μM 2-mercaptoethanol, 50 μg/ml penicillin/streptomycin (all from Invitrogen) and LIF-conditioned medium, made in house, at a concentration equivalent to 1000 U/ml. Cells were grown at 37°C in a humid atmosphere with 5% CO2.
Methylation-sensitive restriction enzyme analysis
Dicerlox/lox and deficient ES cells were pre-plated for 30 minutes to minimise feeder cell contamination and then grown for 2–3 days until confluent on plates coated with 0.1% gelatin. Genomic DNA was phenol/chloroform extracted by the standard procedure. The genotype of each preparation was confirmed by PCR using primers SEQ28290 (agtaatgtgagcaatagtcccag), Di31831 (agtgtagccttagccatttgc) and Di32050AS (ctggtggcttgaggacaagac) and the following PCR conditions: 95°C for 4 minutes; (95°C for 30 seconds; 60°C for 30 seconds; 72°C for 30 seconds) × 35 cycles. PCR fragments were resolved on a 2.5% agarose gel, giving a 259 bp fragment for the wt allele, a 390 bp fragment for the floxed allele and a 309 bp fragment for the Dicer null allele (see Figure 3B).
Genomic DNA was digested to completion with either EcoRI or BamHI restriction enzymes according to the manufacturer's instructions, ethanol precipitated and re-dissolved in TE buffer (10 mM Tris, pH 8.0; 1 mM EDTA). Using methylation-sensitive enzymes, 10 μg DNA aliquots were re-digested, separated by electrophoresis on a 1% agarose gel and blotted onto a GeneScreen nylon filter (Perkin Elmer Life Sciences). Hybridisation with Xist probe 3 (from -37 bp to +952 bp relative to Xist start site P1) was performed as described previously . Images were collected on a PhosphorImager instrument (Molecular Dynamics) and fragment intensity quantification was performed using ImageQuant software (Molecular Dynamics).
SEQUENOM methylation analysis
Genomic DNA was extracted in the same way as for methylation-sensitive Southern analysis. We bisulphite treated 2 μg aliquots of high molecular weight DNA using the EZ DNA methylation kit (Zymo Research). Treatment was performed essentially according to the manufacturer's instructions with a modification in the conversion step, that included 20 cycles of sample treatment with the following conditions (95°C for 30 seconds; 50°C for 15 minutes). Converted DNA was purified on the columns and eluted in 100 μl of water. We used 5 μl of the sample per 25 μl PCR reaction.
HotStarTaq DNA Polymerase kit (Qiagen) was used to amplify modified DNA and PCR primers and the conditions used are described in the Table 1. PCR fragments were sent to SEQUENOM GmbH company (Hamburg, Germany) for in vitro transcription and subsequent MALDI-TOF mass spectrometry analysis using EpiTYPER software .
RNA was isolated from ES cells using TRIzol reagent (Sigma) according to the manufacturer's instructions. RNA was routinely treated with Turbo DNA-free reagent (Ambion) to exclude the possibility of DNA contamination. cDNA synthesis was primed from random hexamers (GE Healthcare) with Superscript III reverse transcriptase (Invitrogen). Strand-specific RT-PCR for Xist amplicons 4, 5, 51 and 51mut was performed according to the method described previously . Primers and PCR conditions are given in Table 2.
Real-time PCR was performed with SYBR Green PCR Master Mix (Bio-Rad) on a Chromo4 Real-time PCR System (Bio-Rad). PCR primers and conditions are listed in Table 2. A melting curve test was performed at the end of each experiment to ensure the specificity of amplification. The data was normalised to β-actin and Idh2 and then to one of the control samples in the set. Each amplicon was analysed at least twice in triplicate on independent cDNA preparations.
Northern blot analysis
Total RNA (20–30 μg), isolated using TRIzol reagent, was separated by PAGE on a 15% urea-containing gel. RNA was transferred to Hybond-XL nylon membrane using a Bio-Rad semi-dry blot apparatus at a constant current of 2.1 mA/cm2 for 1 hour. The membrane was UV-crosslinked with 1000 μJ in a Stratagene UV crosslinker and hybridised with 32P-dCTP labelled mi292as probe. The image was acquired on a PhosphorImager instrument.
RNA FISH analysis
RNA FISH was performed as described previously [7, 41]. pXist, an 18 kb DNA fragment spanning the whole Xist transcript, was labelled using digoxygenin-16-dUTP nick translation mix (Roche) and detected with antidigoxygenin fluorescein isothiocyanate (AD-FITC) antibody raised in sheep (Roche), followed by anti-sheep fluorescein isothiocyanate (FITC) antibody (Vector Laboratories). Images were acquired on a Leica TCS SP5 confocal microscope using LAS AF software.
For whole mount RNA FISH E6.5 embryos were obtained from crosses between mice heterozygous for the Dicer RNase III domain deletion. Embryos were dissected from uteri, rinsed in pre-chilled phosphate buffered saline (PBS), and permeabilised in Cytoskeletal (CSK) buffer for 10 minutes on ice. The washes were carried out in 3 cm Petri dishes kept on ice throughout the whole procedure. The embryos were fixed in 4% formaldehyde for 15 minutes on ice and rinsed in pre-chilled PBS. All aforementioned solutions included 0.1% Tween-20 (Sigma) to prevent the embryos from sticking. Embryos were then dehydrated through an ethanol series (70%, 80%, 90%, 100%). The procedure was performed on a glass slide with a depression (VWR). After the last dehydration wash, ethanol was allowed to evaporate and 15 μl of hybridisation solution, containing DIG-labelled Xist and biotinylated Tsix probes, were added immediately. The slide was covered with a coverslip, sealed with rubber cement, and the hybridisation was performed overnight at 37°C. The Xist probe was full-length Xist cDNA and the Tsix probe was a 4.6 kb EcoRI fragment surrounding Tsix major start site. The post-hybridisation washes were as described previously  with a modification, which included addition of 0.1% Tween-20 to all solutions. The Xist probe was detected with AD-FITC antibody raised in sheep (Roche), followed by anti-sheep FITC antibody (Vector Laboratories), and Tsix probe was detected with avidin-Texas red (AV-TR) followed by biotinylated anti-avidin antibody and then again with AV-TR. All antibodies were from Vector Laboratories unless otherwise stated. Images were acquired on a Leica TCS SP5 confocal microscope using LAS AF software. After imaging, each embryo was genotyped by PCR to determine sex and Dicer genotype.
Western blot analysis was performed as described previously  with some modifications. Briefly, proteins were separated on 8% SDS-PAGE gels and transferred in 1× Transfer buffer (48 mM Tris, 39 mM Glycine, 0.037% SDS, 20% methanol) at 100 mA/gel for 45 minutes, using Bio-Rad semi-dry blotting apparatus. Dnmt3a antibody (working dilution 1:250) and Dnmt3b antibody (WD 1:300) were from Alexa Biosciences; Dnmt1 antibody (WD 1:250) was from Abcam; and LaminB antibody (WD 1:2000) was from Santa Cruz. Enhanced chemiluminescence detection was performed as recommended by the manufacturer (GE Healthcare).
All mouse work was carried out in accordance with the United Kingdom's Home Office regulations under the Animals (Scientific Procedures) Act 1986.
antidigoxygenin fluorescein isothiocyanate
Dulbecco's Modified Eagle Medium
differentially methylated region
foetal calf serum
fluorescent in situ hybridisation
Knockout Serum Replacement
inner cell mass
matrix-assisted laser desorption/ionisation time of flight
methylation sensitive restriction enzyme site
phosphate buffered saline
polymerase chain reaction
transcriptional start site
- Xist :
X inactive specific transcript.
Avner P, Heard E: X-chromosome inactivation: counting, choice and initiation. Nat Rev Genet. 2001, 2: 59-67. 10.1038/35047580.
Ng K, Pullirsch D, Leeb M, Wutz A: Xist and the order of silencing. EMBO Rep. 2007, 8: 34-39. 10.1038/sj.embor.7400871.
Navarro P, Pichard S, Ciaudo C, Avner P, Rougeulle C: Tsix transcription across the Xist gene alters chromatin conformation without affecting Xist transcription: implications for X-chromosome inactivation. Genes Dev. 2005, 19: 1474-1484. 10.1101/gad.341105.
Sado T, Hoki Y, Sasaki H: Tsix silences Xist through modification of chromatin structure. Dev Cell. 2005, 9: 159-165. 10.1016/j.devcel.2005.05.015.
Barr H, Hermann A, Berger J, Tsai HH, Adie K, Prokhortchouk A, Hendrich B, Bird A: Mbd2 contributes to DNA methylation-directed repression of the Xist gene. Mol Cell Biol. 2007, 27: 3750-3757. 10.1128/MCB.02204-06.
Sun BK, Deaton AM, Lee JT: A transient heterochromatic state in Xist preempts X inactivation choice without RNA stabilization. Mol Cell. 2006, 21: 617-628. 10.1016/j.molcel.2006.01.028.
Clerc P, Avner P: Role of the region 3' to Xist exon 6 in the counting process of X-chromosome inactivation. Nat Genet. 1998, 19: 249-253. 10.1038/924.
Lee JT, Lu N: Targeted mutagenesis of Tsix leads to nonrandom X inactivation. Cell. 1999, 99: 47-57. 10.1016/S0092-8674(00)80061-6.
Luikenhuis S, Wutz A, Jaenisch R: Antisense transcription through the Xist locus mediates Tsix function in embryonic stem cells. Mol Cell Biol. 2001, 21: 8512-8520. 10.1128/MCB.21.24.8512-8520.2001.
Morey C, Arnaud D, Avner P, Clerc P: Tsix-mediated repression of Xist accumulation is not sufficient for normal random X inactivation. Hum Mol Genet. 2001, 10: 1403-1411. 10.1093/hmg/10.13.1403.
Sado T, Wang Z, Sasaki H, Li E: Regulation of imprinted X-chromosome inactivation in mice by Tsix. Development. 2001, 128: 1275-1286.
Nesterova TB, Johnston CM, Appanah R, Newall AE, Godwin J, Alexiou M, Brockdorff N: Skewing X chromosome choice by modulating sense transcription across the Xist locus. Genes Dev. 2003, 17: 2177-2190. 10.1101/gad.271203.
Newall AE, Duthie S, Formstone E, Nesterova T, Alexiou M, Johnston C, Caparros ML, Brockdorff N: Primary non-random X inactivation associated with disruption of Xist promoter regulation. Hum Mol Genet. 2001, 10: 581-589. 10.1093/hmg/10.6.581.
Morey C, Navarro P, Debrand E, Avner P, Rougeulle C, Clerc P: The region 3' to Xist mediates X chromosome counting and H3 Lys-4 dimethylation within the Xist gene. EMBO J. 2004, 23: 594-604. 10.1038/sj.emboj.7600071.
Panning B, Jaenisch R: DNA hypomethylation can activate Xist expression and silence X-linked genes. Genes Dev. 1996, 10: 1991-2002. 10.1101/gad.10.16.1991.
Vigneau S, Augui S, Navarro P, Avner P, Clerc P: An essential role for the DXPas34 tandem repeat and Tsix transcription in the counting process of X chromosome inactivation. Proc Natl Acad Sci USA. 2006, 103: 7390-7395. 10.1073/pnas.0602381103.
Ohhata T, Hoki Y, Sasaki H, Sado T: Crucial role of antisense transcription across the Xist promoter in Tsix-mediated Xist chromatin modification. Development. 2008, 135: 227-235. 10.1242/dev.008490.
Mak W, Nesterova TB, de Napoles M, Appanah R, Yamanaka S, Otte AP, Brockdorff N: Reactivation of the paternal X chromosome in early mouse embryos. Science. 2004, 303: 666-669. 10.1126/science.1092674.
Okamoto I, Otte AP, Allis CD, Reinberg D, Heard E: Epigenetic dynamics of imprinted X inactivation during early mouse development. Science. 2004, 303: 644-649. 10.1126/science.1092727.
de Napoles M, Nesterova T, Brockdorff N: Early loss of Xist RNA expression and inactive X chromosome associated chromatin modification in developing primordial germ cells. PLoS ONE. 2007, 2: e860-10.1371/journal.pone.0000860.
Rastan S: Non-random X-chromosome inactivation in mouse X-autosome translocation embryos – location of the inactivation centre. J Embryol Exp Morphol. 1983, 78: 1-22.
Augui S, Filion GJ, Huart S, Nora E, Guggiari M, Maresca M, Stewart AF, Heard E: Sensing X chromosome pairs before X inactivation via a novel X-pairing region of the Xic. Science. 2007, 318: 1632-1636. 10.1126/science.1149420.
Xu N, Tsai CL, Lee JT: Transient homologous chromosome pairing marks the onset of X inactivation. Science. 2006, 311: 1149-1152. 10.1126/science.1122984.
Bacher CP, Guggiari M, Brors B, Augui S, Clerc P, Avner P, Eils R, Heard E: Transient colocalization of X-inactivation centres accompanies the initiation of X inactivation. Nat Cell Biol. 2006, 8: 293-299. 10.1038/ncb1365.
Monkhorst K, Jonkers I, Rentmeester E, Grosveld F, Gribnau J: X inactivation counting and choice is a stochastic process: evidence for involvement of an X-linked activator. Cell. 2008, 132: 410-421. 10.1016/j.cell.2007.12.036.
Matzke MA, Birchler JA: RNAi-mediated pathways in the nucleus. Nat Rev Genet. 2005, 6: 24-35. 10.1038/nrg1500.
Cobb BS, Nesterova TB, Thompson E, Hertweck A, O'Connor E, Godwin J, Wilson CB, Brockdorff N, Fisher AG, Smale ST, Merkenschlager M: T cell lineage choice and differentiation in the absence of the RNase III enzyme Dicer. J Exp Med. 2005, 201: 1367-1373. 10.1084/jem.20050572.
Norris DP, Patel D, Kay GF, Penny GD, Brockdorff N, Sheardown SA, Rastan S: Evidence that random and imprinted Xist expression is controlled by preemptive methylation. Cell. 1994, 77: 41-51. 10.1016/0092-8674(94)90233-X.
Penny GD, Kay GF, Sheardown SA, Rastan S, Brockdorff N: Requirement for Xist in X chromosome inactivation. Nature. 1996, 379: 131-137. 10.1038/379131a0.
Ehrich M, Nelson MR, Stanssens P, Zabeau M, Liloglou T, Xinarianos G, Cantor CR, Field JK, Boom van den D: Quantitative high-throughput analysis of DNA methylation patterns by base-specific cleavage and mass spectrometry. Proc Natl Acad Sci USA. 2005, 102: 15785-15790. 10.1073/pnas.0507816102.
Zvetkova I, Apedaile A, Ramsahoye B, Mermoud JE, Crompton LA, John R, Feil R, Brockdorff N: Global hypomethylation of the genome in XX embryonic stem cells. Nat Genet. 2005, 37: 1274-1279. 10.1038/ng1663.
Kanellopoulou C, Muljo SA, Kung AL, Ganesan S, Drapkin R, Jenuwein T, Livingston DM, Rajewsky K: Dicer-deficient mouse embryonic stem cells are defective in differentiation and centromeric silencing. Genes Dev. 2005, 19: 489-501. 10.1101/gad.1248505.
Seibler J, Zevnik B, Kuter-Luks B, Andreas S, Kern H, Hennek T, Rode A, Heimann C, Faust N, Kauselmann G, Schoor M, Jaenisch R, Rajewsky K, Kühn R, Schwenk F: Rapid generation of inducible mouse mutants. Nucleic Acids Res. 2003, 31: e12-10.1093/nar/gng012.
Okano M, Bell DW, Haber DA, Li E: DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell. 1999, 99: 247-257. 10.1016/S0092-8674(00)81656-6.
Benetti R, Gonzalo S, Jaco I, Munoz P, Gonzalez S, Schoeftner S, Murchison E, Andl T, Chen T, Klatt P, Li E, Serrano M, Millar S, Hannon G, Blasco MA: A mammalian microRNA cluster controls DNA methylation and telomere recombination via Rbl2-dependent regulation of DNA methyltransferases. Nat Struct Mol Biol. 2008, 15: 268-279. 10.1038/nsmb.1399.
Sinkkonen L, Hugenschmidt T, Berninger P, Gaidatzis D, Mohn F, Artus-Revel CG, Zavolan M, Svoboda P, Filipowicz W: MicroRNAs control de novo DNA methylation through regulation of transcriptional repressors in mouse embryonic stem cells. Nat Struct Mol Biol. 2008, 15: 259-267. 10.1038/nsmb.1391.
Sado T, Okano M, Li E, Sasaki H: De novo DNA methylation is dispensable for the initiation and propagation of X chromosome inactivation. Development. 2004, 131: 975-982. 10.1242/dev.00995.
Bernstein E, Kim SY, Carmell MA, Murchison EP, Alcorn H, Li MZ, Mills AA, Elledge SJ, Anderson KV, Hannon GJ: Dicer is essential for mouse development. Nat Genet. 2003, 35: 215-217. 10.1038/ng1253.
Ooi SK, Qiu C, Bernstein E, Li K, Jia D, Yang Z, Erdjument-Bromage H, Tempst P, Lin SP, Allis CD, Cheng X, Bestor TH: DNMT3L connects unmethylated lysine 4 of histone H3 to de novo methylation of DNA. Nature. 2007, 448: 714-717. 10.1038/nature05987.
Church GM, Gilbert W: Genomic sequencing. Proc Natl Acad Sci USA. 1984, 81: 1991-1995. 10.1073/pnas.81.7.1991.
Lawrence JB, Singer RH, Marselle LM: Highly localized tracks of specific transcripts within interphase nuclei visualized by in situ hybridization. Cell. 1989, 57: 493-502. 10.1016/0092-8674(89)90924-0.
We are grateful to Professor M Reth (University of Freiburg, Germany) for the gift of the pANMerCreMer plasmid and to Dr M Endoh (RIKEN Research Center for Allergy and Immunology, Japan) for the gift of the pCAG-Mer-Cre-Mer plasmid. We wish to thank Professor H Koseki (RIKEN Research Center for Allergy and Immunology, Japan) for invaluable advice with ES cell derivation, Aida Di-Gregorio (CSC, UK) for assistance with embryo dissection and members of Developmental Epigenetics group for helpful advice and discussions. This work was funded by the Medical Research Council UK.
The authors declare that they have no competing interests
TBN, MM and NB conceived of and designed the experiments. TBN, BCP, BSC, SN, CES and YAT performed the experiments. TBN, BCP, CES, YAT and NB analysed the data. BSC, TSp, TAR, TSa and MM contributed reagents/materials. TBN and NB wrote the paper. All authors read and approved the final manuscript.