1. How is the specific signal generated? There are >60 potential ubiquitination sites on the surface of FANCD2, but only one is specifically targeted by the E2/E3 pair, Ube2T/FANCL. We discovered an allosteric mechanism propagated across the surface of Ube2T that supports the activity of Ube2T. Our insights enabled us to engineer an enhanced version of Ube2T capable of faster and more complete ubiquitination of FANCD2, which proved invaluable for the subsequent aims of the project. In addition to the allostery, we worked out the molecular basis for the site specificity, and found it depended on a complementary arrangement of charged residues, similar to an ionic “cage” (Chaugule et al., 2020, Nature Chemical Biology;Chaugule et al., 2019, Methods in Enzymology).
2. How is the specific signal read? Many researchers have proposed that there is a recognition factor that reads the signal on FANCD2. Based on our engineered enzymes, we generated large quantities of modified FANCD2 and FANCI, and tested the hypotheses that the function of the modifications might not be for recruiting another protein, but rather support the DNA damage surveillance activity of FANCD2. We determined a structure of FANCD2-Ub in complex with FANCI, and discovered a conformational change that converts the ID2 heterodimer into a DNA “clamp”. We also found that modification of FANCI impedes the deubiquitination of FANCD2. We propose that the slowing of the deubiquitination of FANCD2 prolongs the signal on FANCD2 and enhances the DNA binding, perhaps stabilising the ID2 heterodimer at sites of ICLs to allow repair processes to take place (Rennie et al., 2020, EMBO Reports). We have also determined the structure of modified FANCI-Ub in complex with FANCD2, revealing that this signal is also sufficient to cause the conformational change to a DNA clamp, and favours modification of FANCD2 (Lemonidis et al., 2022, under revision at EMBO Journal, preprint on BioRxiv).
3. How is the signal removed? Removal of the signal is as important as assembly, yet it is much harder to study signal removal at the molecular level due to the prerequisite of having modified material to work with. Aim 1 resulted in high-quality substrates FANCD2-Ub and FANCI-Ub, and in addition, we also generated a further USP1-UAF1 substrate, ubiquitinated PCNA. Prior to our work, there was no structural insight into USP1-UAF1, and understanding of substrate targeting was poor. Whereas many USPs hydrolyse ubiquitin-ubiquitin linkages, USP1 targets ubiquitin-substrate conjugates at specific sites. We made many constructs of USP1 and reconstituted the deubiquitination of each substrate, and we found that the N-terminus of USP1 harbours a FANCD2-specific binding sequence required for deubiquitination of K561 on FANCD2. In contrast, the N-terminus is not required for direct PCNA or FANCI deubiquitination. This is the first evidence of discrimination between protein-ubiquitin linkages (rather than ubiquitin-ubiquitin linkages) and opens up the possibility of separating DNA repair pathways that are regulated by USP1-UAF1 (Arkinson et al., 2018, Life Science Alliance).
4. Following on from this we purified the 0.5MDa complex of USP1-UAF1-FANCD2-Ub-FANCI and determined the structure using cryo-electron microscopy. The model of USP1-UAF1 in complex with monoubiquitinated FANCI-FANCD2 highlights a highly orchestrated deubiquitination process, with USP1-UAF1 driving conformational changes in the substrate. An extensive interface between UAF1 and FANCI, confirmed by mutagenesis and biochemical assays, provides a molecular explanation for the requirement of both proteins, despite neither being directly involved in catalysis (Rennie et al., 2021, Nature Structural and Molecular Biology).
5. Drug discovery aspects have taken us beyond the original scope of the action (see also point 6) - we have determined the structure of the USP1-UAF1-FANCD2-Ub-FANCI complex bound to a known inhibitor of USP1 activity, whose mode of action was previously unknown. The inhibitor has a highly unusual mode of binding, and explains the specificity of the compound. This structure will form the basis for future work to target USP1 for inhibition (Rennie et al., 2022, accepted in Science Advances, preprint on BioRxiv).
6. In addition to the central aims of the work package, we have also used the reagents generated to explore the feasibility of targeting the FANCT/FANCL enzymes for small molecule or fragment binding. We have conducted a proof-of-principle fragment screen (Morreale et al., 2017a, J. Med. Chemistry), and unexpectedly uncovered a mobile secondary structure element in FANCT which in turn guided us towards the allosteric mechanism discovered in point 1 above (Morreale et al., 2017b, J. Med. Chemistry).
Overview of results:
Achieved main objectives of the Action. Dissemination via academic conferences, seminar presentations, a press release.
