It is well established that epigenetic modifications to the genome are crucial for the exquisite control of gene expression required for an organism to develop and differentiate. These modifications are maintained through mitotic rounds of cell division, but must be cleared and reset through meiosis in order for the cells of the early embryo to achieve totipotency. Although we know these mechanisms exist, the rules determining which modifications are established where on the genome and the genes involved in these processes remain poorly characterised. Much of what is known about epigenetic processes has come from studies in non-mammalian organisms, such as Drosophila. However, in our laboratory we have developed a mammalian system for identifying modifiers of epigenetic gene silencing. An ENU mutagenesis screen is being carried out using an inbred mouse line carrying a GFP transgene, with an erythroid-specific promoter, that is particularly sensitive to changes in epigenetic modifications. Currently, 14 mutant lines that display a heritable shift in GFP expression have been recovered. These have been termed Modifiers of Murine Metastable Epialleles (Mommes). When I began my PhD in 2005, we had not identified any of the mutations underlying the phenotypes observed. To confirm the efficacy of the screen, I have tested the effect of heterozygosity for null alleles of two known epigenetic modifiers, Dnmt3a and Dnmt3b, on expression of the GFP transgene. Heterozygosity for the Dnmt3b knockout allele does shift expression while heterozygosity for the Dnmt3a knockout allele does not. This highlights the limitations of the screen. With this particular screen we will only detect modifiers that are expressed during haematopoiesis in the bone marrow. I have also worked on MommeD5. MommeD5 is a semi-dominant, homozygous embryonic lethal mutation that acts as an enhancer of variegation. I have found that the MommeD5 allele carries a 7 bp deletion in the major histone deacetylase, Histone deacetylase 1 (Hdac1), and this significantly alters the C-terminus of the mutant protein. The finding of Hdac1 attests to the screen design. The MommeD5 homozygous mutants die at approximately the same time as the published knockout of Hdac1 and the heterozygous mutants show increased levels of Hdac2 and acetylated histone H3, as reported in Hdac1-deficient embryonic stem cells. In addition, I have studied the effect of heterozygosity for each of the mutations on the phenotype of the mouse. In general, heterozygous Momme mutants are viable and fertile, but show subtle abnormal phenotypes. However, in the case of MommeD5 none were observed and this may relate to the compensatory upregulation of other histone deacetylases. In the case of Dnmt3a and Dnmt3b a sex ratio distortion is seen in the colonies, with less males seen than expected. Also, Dnmt3a heterozygous mutant males that inherited the mutant allele from the dam are smaller and show an increased range of body weights compared to their wild-type male littermates. This may be an example of intangible variation, i.e. phenotypic variation observed in isogenic individuals raised in standardised environments. These results suggest that epigenetic mechanisms have a role in intangible variation, also known as developmental noise. Despite the fact that it is now acknowledged by many that stochastic events occur at the level of the cell, the idea that it can happen at the level of the whole organism is rarely considered.

The epigenetic contribution to phenotype is now well established. Studies over the past decade have shown that proteins that are able to establish and propagate epigenetic modifications are essential for mammalian development. Some of the genes involved in these processes have been identified, but the roles of many remain unknown. The mutagenesis screens for modifiers of position effect variegation in Drosophila suggest that there are over 200 genes that are able to modify epigenetic variegation. We emulated this screen in the mouse to identify mammalian modifiers of a variegating transgene. The screen aimed to identify novel genes involved in epigenetic reprogramming, and to generate mouse models to study the impact of disruption to the epigenome. Inbred male mice carrying a variegating GFP transgene expressed in erythrocytes were mutagenised with ENU. Offspring were screened by flow cytometry and in the initial rounds of mutagenesis, 11 dominant mutant lines were identified. These lines were called MommeDs (Modifiers of murine metastable epialleles, dominant). This thesis describes the mapping and phenotypic characterisation of three Momme lines: MommeD7, MommeD8 and MommeD9. The MommeD9 mutation enhances variegation and was mapped to a 3.4 Mb interval on Chromosome 7. A mutation in a 5? splice site was found in the Trim28 gene. Analysis of Trim28 mRNA and protein in heterozygotes showed that the mutant allele was null. Homozygotes die before mid-gestation. Heterozygotes are viable but display variable and complex phenotypes, including infertility, obesity, behavioural abnormalities and premature death. Obese MommeD9 mice have liver steatosis, impaired glucose tolerance and other indicators of metabolic syndrome. This phenotype has not previously been reported for mice haploinsufficient for Trim28. There is considerable variability of phenotypes among inbred MommeD9 heterozygotes, which suggests a role for epigenetics in phenotypic noise or “intangible variation”. MommeD8 is a semi-dominant enhancer of variegation. Some homozygotes are viable but some die around birth. Viable homozygotes weigh less than wildtype littermates and have increased CpG methylation at the GFP transgene enhancer element. The mutation was mapped to a 4 Mb interval on chromosome 4. Extensive candidate gene sequencing failed to find a mutation and so DNA from mutant and wildtype individuals were sequenced across the entire linked interval by 454 Sequencing technology. MommeD8 individuals carry two point mutations, one is intergenic and the other lies in an intron of the Ppie gene. Analysis of Ppie mRNA in heterozygotes and homozygotes shows that mutants have reduced transcript levels, suggesting that a deficiency in Ppie causes the increased silencing of GFP. The Ppie gene has not been reported to be involved in epigenetic reprogramming and little is known about its function. Mice heterozygous for MommeD7 have a marked increase in expression of GFP. Heterozygotes have a range of hematopoietic abnormalities including splenomegaly, anaemia and reticulocytosis. Homozygotes die at birth and appear pale. The increased GFP in the peripheral blood appears to be the consequence of an increase in reticulocytes. The mutation is linked to a 1.5 Mb interval on Chromosome 7. MommeD7 mice appear to have hematopoietic abnormalities that affect the expression of the erythroid-specific GFP reporter transgene. MommeD7 mice serve as a reminder that, as well as discovering bona fide modifiers of epigenetic reprogramming, the ENU screen can also identify hematopoietic mutants.