dnmt1 / and TKO ESCs show differences in their developmental potential

To analyze the differentiation potential of wt, dnmt1-/- _{and TKO ESCs, we performed global}

expression analysis in the pluripotent state (d0) and at two time points during differentiation (d4 and d16), which revealed several important results. Firstly, hypomethylated cells are able to initiate differentiation processes as many concordant transcript changes between mutant and wt EBs were detected after 4 days of differentiation. In line with this, the expression of genes associated with pluripotency was down regulated and transcripts involved in lineage selection and developmental processes were up regulated in all three cell lines independently of the genotype. These results indicate that DNA methylation is dispensable for the activation of differentiation programs. Secondly, although both mutant cells are globally hypomethylated, dnmt1-/- _{and TKO EBs show significant differences in their ability to}

execute differentiation programs. This idea is supported by the observation that the expression profiles of TKO EBs show a high degree of divergence from those in wt EBs after 16 days of EB culture and the few genes could only be grouped into GO categories involved in metabolic processes. In contrast, we still detect many concordant transcript changes in dnmt1-/- _{and wt EBs at day 16 of differentiation. Most of these commonly expressed genes}

were related to developmental processes including cell differentiation and proliferation as well as organ development. These data clearly point to a previously unappreciated progression of transcription programs in differentiated dnmt1-/- _cells.

Discussion

98

Figure 41. Summary of gene changes and corresponding GO categories occurring in wt (red), dnmt1-/-

(blue) and TKO (green) ESCs and EBs identified by Microarray analysis.I

In the undifferentiated state, all cell lines show very similar expression profiles, however, during differentiation, expression patterns progressively diverge, especially in the case of TKO EBs.

However, our analysis also implies a substantial limited differentiation potential in TKO and, to a lower extent, in dnmt1-/- _{cells. In this regard it is important to note that while TKO ESCs}

are virtual devoid of DNA methylation, dnmt1-/- _{ESCs contain about 20 % residual genomic}

methylation, although the methylation is mainly restricted to repetitive sequences (Lei et al., 1996; Liang et al., 2002; Tsumura et al., 2006). Nonetheless, examination of the expression profiles of both hypomethylated cells clearly shows a different response of TKO and dnmt1-/-

cells to differentiation conditions, suggesting that the presence of Dnmt3 proteins can partially compensate for the loss of Dnmt1. In line with this we detected higher levels of dnmt3b, but slightly lower levels of dnmt3a transcripts in dnmt1-/- _{EBs compared to wt EBs}

(Fig. 22). Interestingly, it has been shown that the activity of the dnmt3b promoter is regulated by DNA methylation (Nimura et al., 2006). Hence, the reduced global DNA methylation levels in dnmt1-/-_{EBs could contribute to the higher expression of dnmt3b and}

possibly compensates for the lower dnmt3a transcript levels observed in mutant EBs.

Besides a possible compensatory role of Dnmt3 proteins in dnmt1-/- _{EBs, it could also be that}

the de novo Dnmts fulfill functions in transcriptional regulation independent of their catalytical activity like we have seen e.g. in the silencing of the bivalent genes fgf5 and brachyury (Fig. 25), although here it will be important to confirm binding of Dnmt3 proteins at those promoters. The notion that Dnmts can mediate transcriptional repression independent of their

Discussion

99

catalytical domain has been demonstrated for Dnmt1. Among the numerous Dnmt1- interacting proteins, also several histone deacteylases have been identified (see also chapter 1.2.1) and it is believed that Dnmt1 recruits these repressive chromatin modifying enzymes to mediate gene silencing independently of DNA methylation (Fuks et al., 2000; Robertson et al., 2000a; Rountree et al., 2000b). A recent report suggests that also Dnmt3b could fulfill regulatory functions aside from its DNA methylation activity. More specifically, the study by Martins-Taylor et al. shows that knock down of dnmt3b during neural differentiation alters the timing of differentiation and lineage choice. Interestingly, affected genes were not targets of DNA methylation but the proximal promoters of these lineage genes were directly bound by Dnmt3b. The finding that upon dnmt3b knock down, also the repressive histone mark H3K37me3 and binding of EZH2, the H3K27 methyltransferase component of the Polycomb repressive complex 2 (PRC2), at deregulated genes was reduced compared to control treated cells, implies that Dnmt3b might play an important role in the recruitment of EZH2 and/ or in maintaining EZH2 binding at these promoters (Martins-Taylor et al., 2012). In this context it is interesting to note that up regulated genes in TKO EBs were predominately involved in neural fate specification, implying that DNA methylation and/ or Dnmts play a crucial role in neural development. In general, further studies comparing global binding patterns of Dnmts to DNA methylation profiles would be necessary to shed more light on the role of DNA methylation and Dnmts in controlling transcription programs during development. The analysis of global binding profiles of Dnmt3s and DNA methylation maps could be used to identify targets of Dnmts independent of their catalytical activity. Conversely, the knockout cell lines could be complemented with catalytical mutants of Dnmts to distinguish catalytical dependent and independent targets of the methyltransferases. In addition, as our results suggest that the presence of Dnmt3s in dnmt1-/- _{could contribute to their milder phenotype, it}

would be crucial to analyze the transcription profiles and developmental potential of ESCs lacking Dnmt3a and/or Dnmt3b. A comparison of these results with the data from dnmt1-/-

and TKO ESCs and EBs would shed light on how the various Dnmt proteins contribute to transcriptional control during differentiation.

Key transcription factor genes for cell fate choice and lineage commitment are known to carry bivalent chromatin domains, which are mainly resolved after differentiation initiation, either by loss of H3K27me3 for transcriptional activation or loss of H3K4me3 or both H3 marks for gene silencing (Bernstein et al., 2006; Mikkelsen et al., 2007). In the latter case, the loss of these histone marks is believed to be accompanied by gain of DNA methylation for permanently sealing transcription, as has been proposed for the differentiation into the neural lineage (Mohn et al., 2008). We hypothesized that DNA methylation might represent a general mechanism for the final silencing of bivalent genes and expected that most bivalent genes would be deregulated in dnmt1-/- _{and/ or TKO EBs. However, our analysis revealed}

Discussion

100

that only selected bivalent genes involved in early neuroectodermal differentiation like nestin and sox1 did gain DNA methylation in a subset of differentiated wt cells and were deregulated in the absence of Dnmts. Hence, our results support the idea that de novo methylation represents the long- term silencing mechanism for selected bivalent genes in specific lineages and does not function as a general mechanism for the repression of bivalent genes during differentiation.

4.1.3 Parallels and crosstalk between the two major repressive pathways– DNA