Fragment-based covalent ligand screening identifies the first specific inhibitor for HOIP, an important E3 ubiquitin ligase of the RBR subclass
HOIP RBR domain-specific single-domain antibodies act as crystallisationchaperones to enable structure-based inhibitor development
Proof-of-concept methodologies for structure-based ligand screening forpreviously intractable E3 ligases and other hard-to-target molecules
Protein ubiquitination is a highly versatile post-translational modification used in the regulation of virtually all cellular processes, from the maintenance of protein homeostasis, to DNA damage responses, cell cycle regulation and immune signalling. E3 ubiquitin ligases, a protein family with more than 600 members, are the last enzymes in the three-step ubiquitination cascade. They select the substrate to be modified, and in some cases also determine the topology of the polyubiquitin chain formed, and thus the physiological outcome.
Alterations in the ubiquitin system have been linked to many diseases, including autoimmune and inflammatory disorders, cancer and neurodegenerative diseases. Therefore, there is immense interest in developing therapeutics targeting the ubiquitin system. However, progress towards this has been slow, and there is a clear need for a more detailed understanding of the structures, molecular mechanisms and regulatory elements of the enzymes involved to enable a more successful drug discovery process.
Research in the Rittinger group aims to provide a molecular understanding of the mechanisms underlying the activity and regulation of the enzymes of the ubiquitination cascade, especially E3 ubiquitin ligases. In 2015, as part of this endeavour, Rittinger set up the first joint project in the Crick/GSK Linklabs scheme, an industry-academia open science collaboration to explore biological mechanisms, targets and pathways relevant to human disease. The aim was to identify small molecules that target HOIP, an E3 ligase of the RBR subfamily, for which no specific inhibitors were available, and in 2019, the project delivered exactly this (Johansson et al, 2019). In further work with Linklabs, Rittinger’s lab has also pioneered the use of single-domain antibodies (dAbs) as crystallisation chaperones to crystallise the catalytic domain of HOIP. The HOIP/dAb complex is a robust platform for soaking of ligands that target the active site cysteine of HOIP, thereby providing easy access to structure-based ligand design for this important class of E3 ligases (Tsai et al, 2019).
Human E3 ligases fall into three main classes — RING, HECT and RBR — based on their structure and mechanism of ubiquitin transfer, which in the case of the HECT and RBR E3 ligases involves the formation of a covalent thioester intermediate with ubiquitin. The catalytic module of RBR ligases comprises three subdomains separated by flexible linkers. This tripartite RBR domain, RING1-IBR-RING2, recognises an incoming ubiquitin-loaded E2 enzyme, and transfers the ubiquitin onto a catalytic cysteine within the RING2 subdomain before its final transfer onto the substrate.
The linear ubiquitin chain assembly complex (LUBAC) is a multiprotein RBR-type E3 ligase that generates linear polyubiquitin chains via its three core components: HOIP (the catalytic subunit), HOIL-1L, and SHARPIN. Linear polyubiquitin chains play crucial roles in the regulation of multiple cellular functions, including immune and inflammatory signalling via the NF- κB pathway, cell death, and cancer. Despite its importance, insights into the physiological function of LUBAC have been scarce due to the paucity of inhibitors targeting the catalytic subunit HOIP, which is a challenging target due to its structure. Those few that exist react with multiple other proteins, making them unsuitable for use in a cellular context.
In a new approach to the inhibitor problem, Rittinger and her colleagues decided to modify an established technique — fragment-based screening — to include a new step: a covalent interaction with the crucial cysteine residue, C885, in the HOIP RING2 active site. Rather than screening millions of complex compounds, fragment-based screening takes the opposite approach: a library of a few thousand chemically simple molecules is used to generate low-affinity hits, and these are then elaborated, guided by structural predictions, into higher-affinity compounds. Covalent linkage means that even weak hits are held in place on the target molecule, and they can also be easily detected by a simple shift in molecular weight of the target protein using liquid chromatography-mass spectrometry (LC-MS).
A diverse library of fragments linked to α,β-unsaturated ester electrophiles was synthesised and rapidly screened, to identify a single compound, a cyclopentyl pyridone, (5), that to the group’s surprise, covalently bound HOIP with low affinity but extremely high specificity. (5) was able to effectively prevent linear polyubiquitin chain formation in vitro and in vivo, and when added to cells it could also inhibit canonical NF- κB signalling. A high-resolution crystal structure of (5) complexed to the RBR domain of HOIP, demonstrated that (5) was indeed covalently bound to the HOIP C885 residue, and also showed the reason for its ability to inhibit HOIP, but not other E3 ligases: it sat on a shallow ledge whose amino acid sequence was insufficiently conserved to permit indiscriminate binding.
This work demonstrated that fragment-based covalent ligand screening could be used to identify lead compounds against difficult targets, and Rittinger, still in collaboration with LinkLabs, is now extending and improving the screening library to target other E3 ligases containing catalytic cysteine residues, including bacterial E3 ligases of the NEL family. NEL ligases, present in a number of gram-negative pathogens, have no human homologues, making any inhibitor a potential antimicrobial drug.
Covalent fragment-based ligand screening was a new departure for both Rittinger and GSK when the project was initiated, but for their next collaboration, Rittinger used a technique that was already established at GSK: phage-mediated selection of synthetic single-domain antibodies (dAbs) based on human immunoglobulin heavy and light chain scaffolds.
To support structure-based inhibitor design, robust and easily crystallised structures are necessary. Determination of the structure of the RBR domain of HOIP bound to (5) (Johansson et al, 2019) was fraught with difficulty, as the crystals grown were fragile, and X-ray diffraction was highly anisotrophic, due to the flexibility inherent in the structure. Therefore, Rittinger and her colleagues decided to explore whether dAbs raised against the RBR domain of HOIP could act as crystallisation chaperones in the presence of ligands, stabilising the structure to make it more amenable to analysis.
Rittinger’s postdoc Isabella Tsai spent some months working at GSK using their phage-borne dAb libraries to set up a screen against biotinylated HOIP RBR immobilised on streptavidin resin, looking for tightly binding dAbs that might stabilise the flexible linkers connecting the three contiguous RBR subdomains RING1-IBR-RING2 and make it easier to crystallise. From a field of hundreds of initial hits, Tsai was able to isolate ten dAbs, most of which had binding affinities in the nanomolar range, for further analysis.
The ten dAbs had variable effects on linear ubiquitin chain formation in in vitro assays, which could be correlated with the location of the epitopes they recognised on the HOIP RBR. When assessed in crystallisation assays, one of the ten, dAb3, stabilised the HOIP RBR crystal structure: the HOIP RBR/dAb3 complex yielded 3-D crystals that were solved by a combination of molecular replacement and anomalous scattering at 2.25 Å. Interestingly, dAb3 proved to be dimeric both in solution and when bound to HOIP RBR, with its dimer interface similar to the interface between VH and VK in conventional antibodies.
The dAb3 dimer primarily contacts the IBR domain of HOIP, with each partner in the dimer having a unique interface with different sections of the IBR. This means that the binding interface on RING1 for the incoming E2 ubiquitin-conjugating enzyme is accessible, and the active site cysteine C885 in RING2 is not occluded. Instead, dAb3 occupies the position where the donor ubiquitin would be located, explaining why its binding affects catalytic activity.
This binding pattern, where the active site cysteine is still exposed, but the RBR domain is stabilised, means that dAb3 has great potential as a chaperone in screens for LUBAC inhibitors. This proved to be the case: soaking HOIP RBR/ dAb3 crystals with a selection of inhibitors showed that the crystals provided a straightforward and robust platform to generate high-resolution structures of HOIP RBR in covalent complexes with LUBAC inhibitors of different molecular scaffolds and varying complexity.
Given the importance of LUBAC in many immune and apoptotic signalling pathways, Rittinger’s work to develop rapid and robust screens for new inhibitors could effect a step-change in understanding how LUBAC intersects with other regulatory pathways in the cell. This novel approach is also likely to work for other E3 ligases, giving researchers a powerful new clinically relevant tool with which to probe the many roles of ubiquitination in health and disease.