Abstract
DNA is packaged for storage as chromatin, being wrapped twice around an octamer of histone proteins to form a nucleosome. The nucleosome structure forces the N- and C-terminal tails of histones to project for post-translational modification. The C-terminal tail of histone H2A is modified with a protein named ubiquitin to specify chromatin compaction. Termed H2AK119Ub, this modification is attached by Polycomb Repressive Complex 1 (PRC1) and removed by the Polycomb Repressive Deubiquitinase (PR-DUB) complex. The mammalian PR-DUB is comprised of BRCA1-associated protein 1 (BAP1), the catalytic deubiquitinase, and one of three additional sex combs-like (ASXL1–3) proteins. ASXL1–3 regulate PR-DUB activity by stimulating BAP1 to allow removal of H2AK119Ub. The basis for ASXL1–3 to activate BAP1 is well understood but the N-terminal, DNA-binding HB1, ASXL, restriction endonuclease, helix-turn-helix (HARE-HTH) and C-terminal plant homeodomain (PHD) of ASXL proteins have not yet been characterised. Their study was the focus of this work.
Typically, PHD domains bind an epigenetic modification at the end of the histone H3 N-terminal tail. The ASXL PHD has been hypothesised to recruit the PR-DUB to chromatin. However, the domain was shown to lack residues for binding the end of the histone H3 tail in this Thesis. In vitro functional analyses were performed, with the domain shown not to bind common histone marks. A structure for the ASXL2 PHD was pursued but structural study was limited by high domain flexibility and weak protein self dimerisation. AlphaFold2 was used to predict the structure of the domain, with an unusual fold shown. The ASXL PHD has diverged significantly from the PHD domain consensus, with Zinc binding lost from the N-terminal Zinc chelating site. Functional analyses were consequently performed to determine the role of the atypical ASXL PHD.
It was anticipated that protein interactors would specify ASXL PHD domain function, so the interactome of the ASXL PHD was pursued. To do so, the ASXL1 PHD was fused to a promiscuous biotin ligase named TurboID for proximity-dependent protein biotinylation in human cells. By this approach, labelled proteins were purified and identified by quantitative mass spectrometry. The ASXL PHD was shown to bind small ubiquitin-related modifier 2 (SUMO2) and SUMO3 by TurboID and in in vitro experiments. SUMO proteins regulate the double-stranded DNA damage response, where the PR-DUB complex is thought to have a key regulatory role. Study of SUMO recognition was performed in cellulo but was inconclusive. A model for the ASXL PHD to function as a SUMO recognition motif was proposed in the context of a related DNA damage regulatory protein named Lethal (3) Malignant Brain Tumour-like 2 (L3MBTL2), which is known to control localisation of full-length ASXL1. Several other protein interactors were determined by TurboID and studied further, including Myeloid Leukaemia Factor 2 (MLF2). MLF2 was statistically-significantly enriched by TurboID and shown to bind the ASXL PHD in continued experiments. By functional association with MLF2, another protein named Tribbles 1 (TRIB1) was implied to bind the ASXL PHD, with in vitro binding analyses performed to confirm the interaction. Together, MLF2 and TRIB1 were proposed to regulate ASXL protein stability. Several other proteins were identified by TurboID with potential to confer PR-DUB chromatin localisation.
Cleavage under targets and release using nuclease (CUT&RUN) sequencing was used to determine the chromatin-binding profile of the ASXL1 PHD and HARE-HTH domains. Unfortunately, the efficiency of CUT&RUN was poor, with issues in repetitive sequencing and fragment library diversity. A single replicate for the ASXL1 HARE-HTH behaved as anticipated. A preliminary analysis was performed on the dataset to measure the potential for the ASXL HARE-HTH to bind to DNA. The domain was shown to bind at gene promoters in a similar fashion to full-length ASXL1. Prospective gene targets for the ASXL1 HARE-HTH were considered at length, with the domain hypothesised to specify localisation to genes integral to cellular development. Several other genetic regions were identified as ASXL HARE-HTH regulatory targets, with potential to drive Bohring Opitz syndrome development, the disease caused by inherited genetic mutation of ASXL1. Finally, a DNA sequence motif for the ASXL HARE-HTH was investigated. Several motifs were similarly enriched by the domain, with CpG bias shared between them. This may point to CpG methyl sensitivity for the ASXL HARE-HTH, with further replicates of CUT&RUN and experimental validation required to clarify domain specificity. Finally, a system for recombinant expression of the ASXL1 HARE-HTH was developed, with potential for structural study.
Together, the ASXL PHD and HARE-HTH domains appear to contribute to PR-DUB function. Experiments on the ASXL PHD suggest the domain should be reclassified, with a clear and obvious lack of activity as a conventional PHD domain. Preliminary data on localisation of the ASXL HARE-HTH suggests the domain is integral to targeting the PR-DUB to chromatin and that aberrant regulation of ASXL1 HARE-HTH target genes may drive Bohring Opitz syndrome. This work provides the foundation to continued study of the outermost domains of ASXL1–3.