Abstract
The covalent attachment of ubiquitin to substrate proteins, a process known as ubiquitylation, is a post-translational modification that modulates protein degradation and cell signalling pathways. A tightly regulated cascade, consisting of E1, E2 and E3 enzymes, promotes the selective transfer of ubiquitin to substrates. During this process, single ubiquitin moieties or polyubiquitin chains of eight linkage types can be attached to target proteins. Specific interactions between an E2 ubiquitin-conjugating enzyme and its cognate E3 ligase are required to determine the type of ubiquitin modification. Ubiquitylation of substrate proteins, however, is a highly dynamic process, and therefore, conjugation events need to be tightly controlled. This task is accomplished by specialised proteins, called deubiquitylating enzymes, which selectively remove mono or polyubiquitin chains from their targets.
Monoubiquitylation of histone proteins is an essential post-translational mark that controls a variety of biological processes, including gene transcription. During this cellular event, a high selective pairing between specific E2 and E3 enzymes, and their nucleosome substrate, is required to ensure that only a single lysine is modified. In humans, monoubiquitylation of histone H2B at K120 by the RNF20/RNF40-Ube2B~Ub complex plays an essential role in transcription activation. Likewise, removal of ubiquitin from K119 of histone H2A by the BAP1-ASXL1 deubiquitylating module is also required for correct regulation of gene expression. Cleavage of ubiquitin from the corresponding lysine in Drosophila depends on the activity of the homologous proteins, Calypso and ASX.
In the first part of this study, biochemical and structural analysis of the RING domains from RNF20 and RNF40 were performed in order to characterise the mechanisms that underpin the E3 ligase activity of these proteins. Our analyses demonstrated that the RNF20 and RNF40 RINGs form dimers in solution and possess activity in conjugation with the E2 enzyme Ube2B. Next, the crystal structure of the RNF20 RING domain was solved and its analysis showed that the RNF20 RING homodimer is most similar to the IAP and RNF4 RING dimers. Despite these similarities, it was also demonstrated that, like Rad18, the activity of the RNF20 RING is highly specific for Ube2B even though the interaction between these proteins is very weak. In order to investigate the molecular basis of Ube2B~Ub recruitment by the RING domain of RNF20, an overlayed model of this complex was generated. Mutagenesis analyses were used to characterise the E3-E2 and E3-ubiquitin interfaces, and the key interacting residues identified were employed to generate a structure-based model of the RNF20 Ube2B~Ub complex. This model differs from RING-E2~Ub complexes previously characterised, and it accounts for the low affinity interaction between RNF20 and Ube2B. Furthermore, it was proposed that this same model might be consistent with the ability of RNF20 and Ube2B to interact with their nucleosome substrate, and to promote selective ubiquitin transfer to the target histone lysine. In addition to these analyses, it was demonstrated that RNF20 and RNF40 can form a functional heterodimer, and that the RING domains of both proteins are required for this association.
In the second part of this work, the deubiquitylating enzymes BAP1 and Calypso, as well as the BAP1-ASXL1 and Calypso-ASX complexes, were characterised in order to investigate the mechanisms underpinning their deubiquitylating activity. It was found that both BAP1 and Calypso form dimers in solution, and that interaction with the DEUBAD domains of ASXL1 and ASX does not destabilise this dimeric state. Furthermore, it was also demonstrated that the isolated BAP1 protein displays deubiquitylating activity, but formation of a stable BAP1 ASXL1 complex further enhances it. In contrast, it was found that the presence of the ASX DEUBAD domain is absolutely required to stimulate cleavage of ubiquitin by Calypso. Altogether, it was concluded that specific interactions of the ASXL1 and ASX DEUBAD domains with BAP1 and Calypso are required to promote efficient ubiquitin hydrolysis. Since activation of both BAP1 and Calypso does not affect dimerisation of these proteins, it was also hypothesised that BAP1-ASXL1 and Calypso-ASX might form asymmetric complexes in solution.
Overall, this study provides important insights that will guide further investigations focused on understanding how gene transcription is regulated at a cellular level. The majority of the work undertaken during the first part of this study was recently published in the Journal of Molecular Biology. A copy of the paper has been included in Appendix 5.