It has been proposed that DNA méthylation at certain CpG sites of the Xist promoter and the 5' region of exon 1 in eggs (but not in sperm) persists during preimplantation development and may be a molecular mechanism for imprinted Xist expression (Zuccotti and Monk, 1995; Ariel et a l, 1995). More recent experiments, however, have suggested that the méthylation of the Xist promoter CpG sites in eggs is mosaic and that the probability of méthylation is low at any specific sites tested (Zuccotti and Monk, unpublished data; see General Discussion Chapter 8). Therefore, at the onset of development, certain CpG sites in the promoter region and the 5' end of the first exon of the Xist gene are unmethylated on the expressed paternal allele, whereas they are mosaically methylated on the repressed maternal allele.
Huntriss et a l (1997) in the Molecular Embryology Unit have recently demonstrated a methylation-dependent DNA-binding protein in ES cells which binds to an Xist promoter sequence encompassing three CpG sites (-44 to-36) only when one or more of the three CpG sites in this sequence are methylated. Moreover, this
sequence motif (5'-GCGCCGCGG-3', -44 to -36, the protein binding site when the sequence is methylated) is essential for normal transcription in that mutation or méthylation in this region markedly decreases transcriptional activity (Huntriss et a l,
1997). Although it is not yet known whether the protein is present in the
preimplantation embryo [results of the co-injection {in vivo competition) experiments are inconclusive, see later], I have hypothesised that méthylation of the maternal allele inherited from the oocyte will decrease the probability that this allele is expressed, thus leading to imprinted preferential paternal allele expression.
To investigate whether DNA méthylation regulates the activity of the 233 bp Xist promoter fragment in preimplantation embryos, the pXist-lucl plasmid construct was methylated in vitro with site-specific DNA methylases before injection. Limited in vitro méthylation of the pXist-luc 1 with either the FnuDW or Hhal methylase did not significantly reduce the Xist promoter activity. However, when the pXist-lucl was methylated with both FnuDU and Hhal methylases (five CpG sites within the promoter methylated), the promoter activity was significantly repressed in the 2-cell embryos. Méthylation of all CpG sites with Sssl methylase abolished the promoter activity.
It is possible that the repression of transcription by méthylation with both FnuDll and Hhal methylases is caused by a greater density of méthylation of the construct as a whole (43 sites methylated for both methylases, compared with 20 and 28 sites methylated for FnuDll and Hhal alone, respectively). However, it is
generally considered that méthylation of sequences in the body of a construct does not significantly influence transcription and that it is méthylation of the promoter itself that is crucial (Razin, personal communication). Moreover, in my experiments, méthylation of the construct only outside the promoter by HpaU methylase did not reduce luciferase activity (24 sites methylated). Other workers have shown that the
effect of DNA méthylation on promoter activity will be a function of both density and site-specificity of méthylation and promoter strength (Boyes and Bird, 1992; Hsieh, 1994). Although the effect of méthylation is less well characterised in
preimplantation embryos, my results are comparable to the results in cultured cells as discussed below.
First, Hsieh (1994) and Nan et al. (1997) have reported that the repressive effect of méthylation is not linear with density, but the effect of méthylation becomes significant at a certain critical density. This is also the case for my results with preimplantation embryos; either FnuDU or Hhal methylase, which methylates 20 or 28 CpG sites, does not decrease promoter activity but Hhal+FnuDll méthylation together, which methylates 43 sites, significantly decreases promoter activity.
Second, with regard to the effect of enhancer on alleviation of méthylation inhibition, Boyes and Bird (1992) have reported, in a transfection assay, that
repression of human a-globin gene by méthylation can be overcome by introduction of the SV40 enhancer when the construct is sparsely methylated (Hhal méthylation) but not when it is heavily methylatted (tol-m ethylation). In my study, the SV40 enhancer could not overcome the repression by Hhal+FnuDU méthylation or Sssl méthylation, which is consistent with their results in HeLa cells.
A definite answer for the effect of DNA méthylation would be obtained by use of the pXist-lucl constructs specifically methylated at CpG sites only within the promoter. Such a méthylation pattern confined to a limited region ('patch méthylation') can be created by either
(I) in vitro méthylation of the promoter fragment followed by ligation to the remaining unmethylated portion of the construct in vitro or
(2) annealing of the single-stranded fragment of a region to be methylated (promoter region in this case) to the single-stranded plasmid of complementary
sequence, in vitro méthylation of the double-stranded region and filling of the remaining single-stranded region to double-stranded (Kass et a l, 1993).
On balance, I can conclude that density and/or site-specificity of CpG méthylation affects Xist promoter activity in preimplantation embryos. This work provides further support for the hypothesis that DNA méthylation is involved in the regulation of imprinted endogenous Xist expression in early development.
Huntriss et a l (1997) used the same construct in a transfection assay. In contrast to my results, they have shown that méthylation of the pXist-lucl construct with either the Hhal or the FnuDll methylase before transfection suppressed promoter activity by 3- to 5-fold and méthylation with both Hhal and FnuDll methylases did not further suppress the activity. This discrepancy of the results between Huntriss et a l (1997) and my study may be explained by
(1) the nature of the cells used; embryonal carcinoma (P I9) cells in Huntriss et a l (1997) whereas 2-cell embryos in my study, and P19 cells may be more sensitive to méthylation and
(2) the copy number of the construct present in the nucleus; 4,200-8,400 molecules in my study and probably much less in their transfection assay.
As seen with embryos injected with the unmethylated pXist-lucl construct, a higher luciferase activity was also observed in some of the embryos injected with the methylated construct and arrested at the 1-cell stage. The fact that in vitro methylated constructs can be utilized as templates by the transcriptional machinery when the embryos are arrested at the 1-cell stage may be explained by inefficient formation of
inactive chromatin structure due to (1) active déméthylation of the construct or (2) the absence (or inaccessibility) of a methylation-dependent repressor protein (see below).