Identification of a possible role of thymine DNA glycosylase (TDG) in epigenome maintenance

Thymine DNA glycosylase (TDG) was discovered as an enzyme capable of removing uracil (U) and thymine (T) from G/U and G/T mispairs, respectively. Owing to this ability, TDG was proposed to initiate restoration of C/G pairs at sites of cytosine or 5-methycytosine (5-meC) deamination. In addition to products of base deamination, the substrate spectrum of TDG covers a wide range of DNA base damages resulting from base oxidation and alkylation. TDG was also found to engage in physical and functional interactions with transcription factors, and more recent evidence supports additional interactions with the de novo DNA methyltransferases Dnmt3a and 3b in the context of gene transcription. Together with its biochemical properties, these observations suggest that TDG might be targeted to gene regulatory sequences as part of a macromolecular assembly to control their functional integrity. TDG may counteract the mutagenic effects of C and 5-meC deamination in CG-rich regions and/or be involved in the maintenance of CpG promoter methylation patterns. A tight regulation of CpG methylation at gene regulatory regions is critical for accurate gene expression, proper cellular differentiation and embryonic development. A somewhat surprising but in this context consistent finding was that, in contrast to other DNA glycosylases, TDG is essential for proper fetal development since a targeted knockout of the gene leads to embryonic lethality.
To gain insights into the biological functions of TDG, we aimed to establish and apply biochemical fractionation procedures for high affinity purification and structural and functional characterization of TDG containing proteins complexes.
The first part of the thesis was concerned with biochemical characterization of the protein interaction network of TDG in living mammalian cells. To this end, I applied different approaches allowing high affinity isolation of protein complexes from mammalian cells, such as the tandem affinity purification (TAP) method as well as immunoprecipitation of endogenous protein and of the TDGa isoform from TdgA overexpressing embryonic stem (ES) cells. These efforts, however, did not reveal any TDG interacting partners in subsequent mass spectrometry (MS) analyses. These results were surprising, as TDG was previously reported to interact with transcription factors and DNA methyltransferases. Remarkably, however, all previously identified protein interactors of TDG were discovered in screen with the respective partner proteins, and under conditions of simultaneous overexpression of both interacting proteins. The only proteins ever identified in screen with TDG were Sumo1 and Sumo3, which turned out to covalently modify the glycosylase. For this reason, we decided to pursue our search with classical cell fractionation experiments. We first did gel filtration experiments from total cell lysates and showed that TDG is indeed able to form distinct multiprotein complexes in undifferentiated mouse embryonic stem cells that may also contain the RNA helicase p68. Further subcellular fractionation experiments then revealed that TDG is present in all cell compartments, with a significant fraction of nuclear TDG being associated with chromatin, together with p68 and de novo DNA methyltransferases. Together with published findings, these results suggested that protein complexes containing TDG might act in a chromatin-associated context, at gene regulatory regions.
The developmental phenotype of Tdg-/- knockout mice and the interactions of TDG with factors involved in developmental gene regulation (e.g. retinoic acid receptors RAR/RXR) implicate a function of TDG during early development and cell differentiation, at times governed by dynamic changes in gene expression, DNA methylation and histone modifications. Such changes have been studied using a well-established during in vitro differentiation of ES cells to lineage committed neuronal progenitors (NPs). We thus aimed to address the function of TDG as part of chromatin associated protein complexes during the process of retinoic acid induced differentiation of ES cells to NPs.
In the second part of the thesis we made use of a this well-established in vitro differentiation system to examine the genome-wide localization of TDG to chromatin by TDG chromatin immunoprecipitation (ChIP) and to correlate TDG association to chromatin with gene expression and DNA methylation changes linked to cellular differentiation. TDG ChIP combined with high throughput sequencing showed that TDG associates with high preference to CpG islands in promoters of actively transcribed genes or genes poised for transcriptional activation. Such CpG rich sequences are normally unmethylated in mammalian genomes. Interestingly, we found TDG to localize to promoters of many genes controlling pluripotency (e.g. Oct4, Nanog) and developmental processes (e.g. Sfrp2, Tgfb2, Gata6), thus, supporting a function of TDG in cell differentiation and/or embryonic development. As different lines of circumstantial evidence have associated TDG with changes in CpG methylation following activation of hormone responsive gene promoters, we went on to further test genome-wide promoter methylation in Tdg+/- and Tdg-/- NPs making use of a combination of methylated DNA immunoprecipitation (MeDIP) and microarray technology. This showed that the loss of TDG does not affect global promoter DNA methylation. Nevertheless, there were a number of significant differences, suggesting that TDG might affect the CpG methylation pattern at some promoters. Also, owing to the limited resolution of the MeDIP method, however, we could not exclude an involvement of TDG in the control of DNA methylation of specific promoter CpGs. Additional bisulfite sequencing of promoters of TDG bound developmental genes (e.g. Sfrp2, Tgfb2) in NPs and differentiated mouse embryonic fibroblasts (MEFs) have indeed proved that loss of TDG affects local changes in DNA methylation at particular CpGs.
Subsequent analysis of genome-wide gene expression profiles of ES cells and differentiated Tdg+/- and Tdg-/- NPs revealed that a limited number of genes (229) are differentially regulated in ES, whereas substantial differences in gene expression in were observed in NPs (1022 genes). This implicated a specific function of TDG in the regulation of cell differentiation triggered gene expression changes. Detailed analysis of the expression of the Pax6 gene, accurate regulation of which is essential for proper neuron development, showed that its promoter is bound by TDG and that its transcription is inappropriately regulated upon further differentiation of Tdg-/- NPs into the neuronal lineage. Whereas Tdg+/- NPs efficiently downregulated Pax6 (50x) and further differentiated into neuron-like cells, Tdg-/- NPs only partially downregulated Pax6 gene expression (6x) and underwent apoptosis at day 2 after plating in neuronal medium. This phenotype was complemented by expression of TDGa, clearly implicating TDG in the regulation of Pax6 expression during differentiation of ES cells to terminal neurons.
We further observed misregulation of pluripotency genes (e.g. Oct4) regulated by TDG bound promoters during early differentiation of ES cells. In the absence of TDG, ES cells showed the tendency to enter spontaneous and/or RA induced differentiation, suggesting a role for TDG in the regulation of pluripotency. During RA induced differentiation we further observed the activation of the neuron specific gene Lrrtm2 exclusively in TDG proficient cells. In addition, ChIP experiments showed that transcription factors involved in the activation of the Lrrtm2 gene (e.g. COUP-TFI, RAR) are not recruited to the respective promoter in Tdg-/- cells, suggesting that TDG might act passively as a scaffold factor important for the recruitment of transcription factors to promoter regions.
I set out to clarify the biological function of TDG by investigating its molecular interactions in mammalian cells. I found that TDG, as a DNA repair enzyme, associates tightly with chromatin, where it localizes with high preference to CpG island promoters of active genes and genes poised to be expressed. I also found that the loss of TDG causes misregulation of genes during cell differentiation and that this appears to be related to a function of TDG in establishing and/or maintaining CpG methylation pattern in gene regulatory sequences. These discoveries implicate a novel function of DNA repair, in the maintenance not only of the genome, but also the epigenome.