Logo: to the web site of Uppsala University

Protein-protein interactions (PPIs) are inherently dynamic and vital for maintaining normal cellular function. Short linear motifs (SLiMs), which are typically 3-10 amino acid long stretches, are present in intrinsically disordered regions (IDRs) and often serve as binding interfaces for PPIs. SLiM-mediated interactions are essential in various biological processes, such as cellular signaling, cell cycle progression and protein degradation. SLiMs play an important role in targeting E3 ligases to their substrates. They may also recruit deubiquitinating enzymes (DUBs) to their substrates, thereby reversing the action of E3 ligases by removing ubiquitin from target proteins.  At present, only a fraction of the predicted SLiMs in the human proteome has been identified. Thus, it is important to develop and utilize methods to identify SLiM-based interactions and to gain further insights into the specificity determinants of the interactions. Proteomic peptide-phage display (ProP-PD) is a high-throughput method developed to capture motif-based PPIs. However, sometimes only a limited set of peptide ligands are identified from these experiments, rendering it challenging to define a consensus binding motif. We therefore developed a deep-mutational scanning (DMS) by peptide-phage display approach, which enables comprehensive examination of the effects of substitutions on peptide-protein interactions. Using the DMS protocol, we deciphered the binding determinants of motif-based interactions of two globular domains of the ubiquitin carboxyl-terminal hydrolase 8 (USP8). We uncovered that the MIT domain, which is a previously described motif-binding domain, binds to degenerate motif variants. Furthermore, we revealed a peptide binding capability of the Rhodanese domain and demonstrated that it recognizes more than one type of motif. The information enabled the prediction of potential binding sites in USP8 known interactors. Expanding the analysis to other DUBs, a screening for additional motif-binding auxiliary domains of proteins from the USP family was performed. Fourteen domains were found to bind to peptides, which expanded the previously unexplored landscape of DUB-motif recognition. The zf-UBP and DUSP2 domains of USP20 and USP33 were found to act as peptide-binding domains, recognizing novel consensus motifs. Finally, extending beyond DUBs, a contribution was made towards charting interactions for numerous peptide-binding domains. The research presented in this thesis, sheds light on the previously underexplored area of motif-recognition of DUBs and contributes towards expanding the motif-based map of the human interactome.

E3 ligases are central players in protein homeostasis, deciding which proteins are degraded by conjugating ubiquitin onto their target proteins. Beyond degradation, ubiquitination regulates a myriad of further cellular signaling pathways. E3 ligases have developed different strategies to identify their ligands, one of which is the recognition of an autonomous interaction-mediating compact binding sites called Short Linear Motif (SLiMs). SLiMs are typically 3-10-residue stretches embedded in intrinsically disordered regions, and commonly serve as binding interfaces. SLiMs which lead to ubiquitination and proteasomal degradation upon binding to their target E3 ligases are called a degrons. SLiM-based protein-protein interactions typically rely on 3-4 conserved key residues, with less-conserved flanking regions tuning binding affinity and specificity. Elucidating the substrate determinants of the ca. 900 E3 ligases has been one of the central challenges of the E3 ligase field.

This thesis presents the investigation of the SLiM-binding of more than 140 domains from E3 ligases and associated auxiliary proteins, distributed over four studies. SLiM discovery was based on proteomic peptide–phage display (ProP–PD) screening. ProP–PD is a high-throughput method to identify peptide ligands from the proteome, from which consensus motifs can be derived. We disclose 11,460 potential SLiM instances of the E3 ligase machinery. We describe 40 distinct SLiMs, of which 14 are either entirely novel or redefine known binding determinants. The defined SLiMs allow us to predict additional putative proteomic binding sites. Interactions were validated using peptide SPOT arrays, and characterized using fluorescence polarization-based affinity measurements. Motif key residues were further validated using alanine-scanning footprinting. Using artificial intelligence-based modelling, we localized binding sites on the folded domains and validate them through site-directed mutagenesis. Among the results, we highlight the establishment of internal peptide recognition by the TPR domain of STUB1, the definition of the peptide-binding capacity of four MIB-HERC2 domains, and the establishment of a RNF41 binding motif in USP8.

Collectively, the research provides a substantially expanded SLiM-based interaction space of E3 ligases, both in terms of interactions and motifs. By systematically mapping E3 binding motifs and binding sites at unprecedented scale, this thesis provides a resource that advances the biochemical understanding of the protein-protein interactions of E3 ligases. The insights into these interactions may contribute to the decoding of specificity determinants of ubiquitin-mediated signaling, and may lay the groundwork for future efforts to predict, manipulate, and therapeutically target ubiquitin-dependent regulatory networks.