Structural basis for substrate recruitment by AMBRA1 E3 ligase receptor

Abstract

AMBRA1 is a tumor suppressor protein that functions as a substrate receptor of the ubiquitin conjugation system as part of autophagy and cell-cycle regulatory network. The highly intrinsic disorder of AMBRA1 has so far precluded its structural determination. To solve this problem, we analyzed the domain organization and dynamics of AMBRA1 using hydrogen deuterium exchange mass spectrometry (HDX-MS). High deuterium uptake indicates that AMBRA1 is a dynamic and largely unstructured protein, and can be stabilized upon interaction with DDB1, the adaptor of the Cullin4A/B E3 ligase complex. Here we present the cryo-EM structure of AMBRA1 in complex with DDB1 at 3 A resolution. The structure shows that parts of N- and C- terminal structural regions in AMBRA1 fold together into the highly dynamic WD40 domain, and reveals how DDB1 engages with AMBRA1 to create a binding scaffold for substrate recruitment. AMBRA1 uses its N-terminal helix-loop-helix and WD40 domain to bind the double-propeller fold of DDB1, whereas different regions target the specific cellular substrates for ubiquitination. We also demonstrate that DDB1 binding-defective AMBRA1 mutants prevent ubiquitination of the substrate Cyclin D1 in vitro and decreased number of autophagosomes in the cells. Together, these results provide structural insights into AMBRA1-ubiquitin ligase complex and suggests a mechanism by which the AMBRA1 acts as a hub involved in various physiological processes.

MAVISp: A Modular Structure-Based Framework for Genomic Variant Interpretation

“…We used an X-ray (PDB entry 3K0N 88 ) and an NMR structure (PDB entry 2KK0 89 ) for PPIA and ARID3A, respectively. For AMBRA1 we used a recently solved CryoEM structure in complex with DDB1 90 , isolating the chain for AMBRA1 and reconstructing missing coordinates with MODELLER. Additionally, in the case of ALPK1 AlphaFold2 model, we designed a 20-residues linker to connect the residues 495 and 956, including 10 residues from the N- and the C-terminal regions to avoid modelling a long unstructured region with low pLDDT score.…”