Multi-scale classification decodes the complexity of the human E3 ligome

Abstract

E3 ubiquitin ligases are key regulators of protein homeostasis, targeting specific proteins for degradation via the ubiquitin-proteasome system (UPS). They provide crucial substrate specificity, making them promising candidates for the design of novel therapeutics. This work presents a comprehensive, annotated dataset of high-confidence catalytic human E3 ligases, termed the "E3 ligome". Integrating disparate data from various granularity layers, including protein sequence, domain architecture, 3D structure, function, localization, and expression, we learn an emergent distance metric, capturing authentic relationships within this heterogeneous group. A weakly-supervised hierarchical classification framework identifies conserved features of E3 families and subfamilies, consistent with RING, HECT, and RBR classes. This classification explains functional segregation, identifies multi-subunit and standalone enzymes, and integrates substrate and small molecule interaction networks. Our analysis provides a global view of E3 biology, opening new strategies for drugging E3-substrate networks, including drug re-purposing and designing new E3 handles.