Targeting and dynamics of gene repression during stem cell differentiation

The identity and function of different cellular subtypes critically depend on their unique set of expressed genes. Gene expression programs and their changes during development are mainly controlled by sequence-specific DNA binding factors. It has recently become clear that chromatin modifications are important regulators of these processes. While there are several chromatin-based pathways that correlate with gene repression, their exact role in silencing remains elusive. Moreover, for many repressive chromatin modifications a complete picture of the genomic distribution and its dynamics during development is lacking. Finally, it is still unclear how these genomic patterns of repressive chromatin marks are established. For my PhD work, I set out to address these questions by studying the targeting of H3K9me2 and DNA methylation during cellular differentiation.
Our analysis revealed that H3K9me2 is highly abundant in embryonic stem cells and occurs in large domains that occupy more than half of the genome. H3K9me2 marks chromatin outside of transcribed, active or polycomb regulated sites, possibly keeping it in a repressed state. Importantly, abundance of H3K9me2 increases only slightly during neuronal differentiation, with a localized gain occurring at gene bodies of transcribed genes. By gene expression profiling we further show that the transcriptome complexity is very similar in stem cells and derived post-mitotic neurons. These data are in contrast to a previously suggested model which states that the pluripotent state of stem cells is accompanied by a global reduction in heterochromatin and a concomitant higher proportion of transcription. Together with results from other groups our data rather indicate that repressive chromatin is abundant in stem cells and upon differentiation gets redistributed only locally and not globally. It has been suggested that such a localized increase of repression at gene regulatory regions helps stabilizing lineage choices and differentiation processes.
In order to investigate how chromatin-based repression pathways are targeted to gene regulatory sites, we focused on DNA methylation, a modification whose catalysis and epigenetic propagation are well understood. By site-specific sequence integration experiments we show that 1 kb promoter elements are sufficient to recapitulate endogenous DNA methylation patterns in stem cells and their dynamic changes upon differentiation, in a process that is independent of transcription. In stem cells, promoters are protected from DNA methylation by small sequence elements that we termed methylation determining regions (MDRs). Protection from DNA methylation by MDRs depends on a combination of DNA binding motifs, which get recognized by transcription factors such as RFX2. It has been speculated before that establishment of an unmethylated promoter state is facilitated by proteins that recognize unmethylated CpGs. While not excluding a role in maintenance, our data suggest that CpG-richness alone is not sufficient for initiation of this chromatin state. Remarkably, no additional sequence besides an MDR is needed to recapitulate differentiation-induced de novo methylation. Moreover, MDRs are able to protect neighboring sequences from DNA methylation in stem cells and from de novo methylation during differentiation. These results imply that one possible way of differentiation-induced de novo methylation could involve reduced binding of factors that protect from DNA methylation.
In summary, H3K9me2 and DNA methylation occupy per default most the genome, even in cells with a high developmental potential. Accordingly, cellular differentiation is accompanied by focal, rather than global changes in repressive chromatin modifications. In the case of DNA methylation, such local changes at gene regulatory sites are determined by the underlying sequence and likely involve binding of transcription factors that protect from DNA methylation.