PhD Student Seraina Blümli and colleagues from the Owen-Hughes lab show the ARID1A subunit of the BAF chromatin remodelling complex organised nucleosomes flanking pluripotency transcription factors and association of the coactivator EP300. Following degradation of ARID1A, dissociation of EP300 at enhancers is associated with downregulation of transcription and EP300 reassociation with upregulation. Few genes are directly affected but widespread indirect effects accumulate slowly to phenocopy a premalignant state.
Population based genome sequencing has enabled the unbiased characterisation of genetic changes that contribute to complex diseases such as cancer. One of the surprises to emerge from global research efforts in this area is the frequency with which genes acting at the level of chromatin or linked to enhancer function are linked to cancer. An example is the finding that 5 of the top 50 human tumour suppressors are components of closely related chromatin remodelling complexes. The findings raise the question how do components of ubiquitous chromatin remodelling complexes function as tissue specific tumour suppressors?
Inspired by a seminar from Masato Kanemaki, PhD student Seraina Blümli engineered mouse stem cell lines using the auxin degron system such that the ARID1A subunit of BAF complexes could be inducible degraded within about one hour. Using this system, she then went on to measure changes in chromatin accessibility and nucleosome organisation following loss of ARID1A. Most of these changes occurred at enhancer elements and the next challenge was to link these with transcriptional changes.
Building on Seraina’s work Dr Nicola Wiechens measured changes in nascent transcription following loss of ARID1A. She found that initially only a few hundred genes were differentially regulated, even though chromatin changes were observed at thousands of genes. This was initially puzzling, but a clue came from the observation that ARID1A containing complexes interact with the co-activator protein EP300. This prompted Nicola to characterise where EP300 was bound before and after degradation of ARID1A.
At this point lab work was interrupted by lockdown. The timing was quite convenient as work was needed to find out how to interpret the intersection between large datasets of different types. The data revealed distinct pathways for up and downregulation of transcription following loss of ARID1A. Enhancers where chromatin accessibility and association of EP300 is reduced were very strongly enriched adjacent to genes that are downregulated. Sites where EP300 was rebound and histone acetylation increased were observed adjacent to upregulated genes. ARID1A is present at many thousands more locations in the genome, but its functions at these locations appear to be redundant.
Both pathways show only an immediate effect on a few hundred genes after 2 hours. Over subsequent days thousands more genes are mis-regulated, many linked to molecular mechanisms for cancer. However, these genes are not associated with the chromatin changes that result directly from loss of ARID1A. As indirect changes dominate the transcriptional landscape resulting from chronic loss of ARID1A, it may be possible to prevent the development of phenotypes, including cancer, by preventing the propagation of indirect effects.
The next step will be to establish human tissue models for cancers caused by mutations to chromatin remodelling enzymes. In this case the pathways revealed will be closely linked to human cancers and will represent targets for a new generation of therapeutics.
The work made use of expertise from across the school in areas including proteomics, genomics, data analysis, computational biology, targeted protein generation and genome engineering and greatly assisted by colleagues with expertise in these areas.
The work has just been published in the journal Cell Reports.