Genomic imprinting independent of de novo DNA methyltransferases in mouse oocytes

  Genomic imprinting is a germline specific gene-marking phenomenon in mammals that regulates parental-origin specific expression of the imprinted genes. Imprinting is crucial for normal mammalian development and relevant to congenital malformation syndromes and cancers. More than one hundred imprinted genes have been identified in mice and humans and most of them are clustered in certain chromosome domains. These imprinted gene clusters contain imprinting control regions (ICRs) that are CpG-rich and methylated only on one of the two parental chromosomes. The ICRs control the allele-specific expression of the imprinted genes within the clusters.  Mammals have two active de novo DNA methyltransferases Dnmt3a and Dnmt3b, and their regulator Dnmt3L. It has been shown that Dnmt3a and Dnmt3L are required for the establishment of DNA methylation imprint of the ICRs in both male and female germline and parental-origin specific expression of the imprinted genes. Furthermore, Dnmt3b is also required for DNA methylation of the Rasgrf1 ICR in the male germline. These studies indicated that the molecular nature of the germline imprints is DNA methylation.  However, it was reported that some imprinted genes in the Lit1 cluster maintain the normal monoallelic expression even in the absence of the maintenance DNA methyltransferase Dnmt1 in the trophoblast. This suggests that the trophoblast has a DNA methylation-independent imprint maintenance mechanism. In addition to the maintenance of the imprints, several resent studies imply the role of epigenetic modifications or factors other than DNA methylation in establishment of the imprints. Taken together, there may be an imprint establishment mechanism other than de novo DNA methylation in the germline.  To reveal the existence of the DNA methylation-independent germline imprints, I used mice lacking the de novo DNA methyltransferases or their regulator in the germline. Unfortunately, both Dnmt3a mutant males and Dnmt3L mutant males display meiotic arrest and azoospermia, and thus it is not possible to assess the effect of loss of DNA methylation at the paternally methylated ICRs in the embryo. Therefore, I obtained female mice lacking Dnmt3a and Dnmt3b or those lacking Dnmt3L specifically in oocytes. I crossed these females with wild-type males and analyzed the allele-specific expression of the imprinted genes in E9.5 embryos and trophoblasts by using single nucleotide polymorphisms in cDNAs (cSNPs). The cSNPs between the laboratory strains and the JF1 mice were identified in the NIG mouse genome database (http://molossinus.lab.nig.ac.jp/msmdb/) and confirmed by PCR amplification and direct sequencing in 17 imprinted genes from seven clusters.  RNAs from the embryo proper and trophoblast of wild-type, [Dnmt3amatKO, Dnmt3bmatKO] and Dnmt3LmatKO embryos at E9.5 were subjected to RT-PCR and direct sequencing. While 15 of 17 imprinted genes showed a disruption in the monoallelic expression, Kcnq1 in the Lit1 cluster maintained the normal maternal expression pattern in these embryos and trophoblasts. The trophoblast-specific imprinted gene Cd81 in the Lit1 cluster also retained the normal maternal expression in the trophoblast. These results suggest that the establishment of the imprints of some genes occurs independent of the de novo DNA methyltransferases and its positive regulator Dnmt3L.   Next, I examined the expression levels of some imprinted genes and confirmed that Cdkn1c and Lit1 are biallelically silenced and biallelically expressed, respectively, in [Dnmt3amatKO, Dnmt3bmatKO] and Dnmt3LmatKO embryos. By contrast, the expression levels of Kcnq1 and Cd81 were unexpectedly altered in [Dnmt3amatKO, Dnmt3bmatKO] and Dnmt3LmatKO embryos despite that they maintained the normal monoallelic expression. These results show that, although the imprint establishment of Kcnq1 and Cd81 in oocytes is independent of the Dnmt3a, Dnmt3b, and Dnmt3L, their expression levels after fertilization are influenced, either directly or indirectly, by the mutations of these genes.  Then I confirmed that the ICRs are not methylated in [Dnmt3a1lox/1lox, Dnmt3b1lox/1lox] and Dnmt3L-/- oocytes and the maternally derived ICRs in [Dnmt3amatKO, Dnmt3bmatKO] and Dnmt3LmatKO embryos and trophoblasts by bisulfite sequencing. These results strongly suggest that imprinting of Kcnq1 and Cd81 is not just independent of de novo DNA methyltransferases and their regulator but indeed independent of DNA methylation.  Interestingly, some imprinted genes such as Peg10 maintained the normal monoallelic expression patterns in some Dnmt3LmatKO embryos, but not in any [Dnmt3amatKO, Dnmt3bmatKO] embryos. To reveal the methylation status of the Peg10 ICR in six Dnmt3LmatKO embryos, I performed bisulfite sequencing and found that two embryos had the DNA methylation imprint, while four embryos did not. This suggests that the DNA methylation imprint at the Peg10 ICR can be established without Dnmt3L. Furthermore, Dnmt3LmatKO embryos without DNA methylation at the Peg10 ICR also maintained the normal monoallelic expression of Peg10. This implies the existence of an imprint establishment mechanism dependent on the de novo DNA methyltransferases, but not on DNA methylation.  Finally, to investigated whether there is any difference in expression between the genes controlled by the de novo DNA methylation dependent (Cdkn1c, Lit1 and Tssc4) and an independent (Kcnq1 and Cd81) mechanisms within the Lit1 cluster in [Dnmt3a1lox/1lox, Dnmt3b1lox/1lox] and Dnmt3L-/- oocytes, I performed quantitative RT-PCR using these oocytes. Expression of Lit1 was extremely increased in both [Dnmt3a1lox/1lox, Dnmt3b1lox/1lox] and Dnmt3L-/- oocytes, whereas expression of the other imprinted genes within the Lit1 cluster was comparable to that in wild-type oocytes. These results suggest that, while the expression of Lit1 in oocytes is repressed by DNA methylation, its ectopic expression of Lit1 in mutant oocytes does not affect the expression of other imprinted genes within the Lit1 cluster. This contrasts with what happens in the embryos, in which the genes in the cluster are normally repressed by Lit1 non-coding RNA.  Based on these results, I propose a model of the establishment of genomic imprints other than DNA methylation in mouse oocytes. The model suggests that, in addition to the DNA methylation imprints at the ICRs, each imprinted gene may have gene-specific epigenetic marks that regulate monoallelic expression. The model will provide a basis for understanding the mechanism of the imprint establishment