Non-canonical roles of mammalian heterochromatin protein 1 (HP1) homologs

The human body is composed of hundreds of different tissues, all containing same genetic information, yet defined by unique gene expression patterns. In order to achieve this, multicellular organisms evolved regulatory mechanisms that go beyond the mere DNA sequence. Unlike in prokaryotes, where DNA is freely accessible to the transcriptional machinery, DNA in eukaryotic cells is wrapped around histone proteins, forming a structure called chromatin. Importantly, chromatin can be regulated by various post-translational modifications of the histone proteins. Such modifications are generally assumed to directly affect the compaction of chromatin and/or act as a recruitment platform for chromatin factors that relax (activate) or condense (repress) chromatin. One of the repressive histone marks, histone 3 lysine 9 methylation (H3K9me), is recognised by members of the HP1 family and/or by other proteins that have the unique ability to spread along chromatin, compacting it, and ultimately forming large inaccessible domains referred to as heterochromatin. Over the past decade, however, the view of HP1s as rigid silencers has been gradually challenged, as it was found that heterochromatic regions produce RNA, and that HP1s are highly mobile molecules. In addition, HP1 proteins were shown to associate with RNAs, and to also associate with chromatin lacking the H3K9me mark. These findings raised several fundamental questions: Is HP1 activity regulated by RNA? How are HP1 proteins recruited to sites lacking H3K9me, and what is their role at those sites?
One aim of my PhD project was to elucidate potential roles of RNA in modulating HP1 activity. For this, I used biochemical methods to dissect RNA binding properties of mammalian HP1 proteins in vitro. My work revealed that one of the HP1 homologs, HP1 alpha, interacts with RNA when bound to H3K9me-marked nucleosomes. The physiological role of such interaction could be stabilisation of binding to heterochromatin, or alternatively, eviction from heterochromatin. The major goal of my PhD project, however, was to investigate the mechanism of HP1 recruitment to chromatin lacking the H3K9me mark. To do so, I was using mouse embryonic stem cells (mESCs) as a model system. I have exploited recent advances in genome editing/CRISPR-Cas9 to delete or endogenously tag individual HP1 homologs or various combinations thereof. Genome- and proteome-wide studies subsequently revealed a novel protein complex, which I dubbed “ChAHP”. ChAHP contains two HP1 homologs (HP1 beta and/or HP1 gamma), the transcription factor Adnp, and the chromatin remodeller Chd4. In collaboration with the group of Nicolas Thomä, we have reconstituted the ChAHP complex in vitro from insect cells and dissected the individual interactions. Together with my proteomics experiments in mESCs, this revealed Adnp as a core bridging module interacting with Chd4 and HP1s. Using ChIP-sequencing, we identified over 15 000 ChAHP-bound genomic sites. Importantly, these sites are devoid of H3K9me2 or H3K9me3. Instead, the complex is targeted via a highly conserved DNA motif recognized by Adnp, and deletion of Adnp or of the DNA motif depletes HP1 binding at the ChAHP sites. In addition, deletion of Adnp or HP1s leads to derepression of lineage-specifying genes bound by ChAHP. However, unlike in case of canonical HP1 silencing, which involves H3K9me and results in formation of a broad heterochromatic domain, ChAHP silencing occurs locally by restricting access to its sites. I propose that this prevents other regulators, including transcriptional activators, from accessing the corresponding DNA sites. Finally, my results provide first insights into the molecular mechanism of a disease that is associated with mutations in the ADNP gene, Helsmoortel-Van der Aa syndrome. Mutant Adnp found in Helsmoortel-Van der Aa patients fails to interact with HP1 proteins, and therefore cannot target HP1s to the chromatin. In summary, my work revealed that HP1 proteins can be recruited to genomic loci in a DNA sequence-specific, H3K9 methylation-independent, manner via an interaction with the transcription factor Adnp, and I demonstrated that H3K9 methylation, unlike in canonical silencing, is not required for repression of ChAHP target genes.