F-box proteins lose their rigid-scaffold image: cryo-EM, cross-linking and deep learning show FBXL2's substrate domain swinging 43 Å
Synopsis
F-box proteins are the substrate-recruiting modules of SCF ubiquitin ligases and have long been treated as relatively rigid structural adaptors. This work combines cryo-EM structures, cross-linking mass spectrometry distance restraints and deep-learning conformational ensemble prediction to compare the same proteins across different complexes. FBXL2's substrate-binding domain moves 43 Å relative to its anchor domain, FBXL17 differs by 17 to 33 Å across assembly contexts, and FBXL1 shifts only 4.5 Å; FBXL2 needs at least four states to satisfy its cross-links to CSN.
Interpretation
The substrate-binding domains of F-box proteins are not fixed: FBXL2 and FBXL17 occupy very different positions relative to their anchor domains within assembled SCF and CSN-SCF complexes. F-box proteins had commonly been viewed as relatively rigid structural adaptors, a view supported mainly by early X-ray crystallographic studies of isolated SKP1-F-box heterodimers. Aligning the CSNDM-SCFFBXL2-GGTase3 cryo-EM structure with the SKP1-FBXL2 structure that likewise binds GGTase3 (PDB 6O60) on the anchor domain gives a substrate-binding-domain RMSD of 43 Å; FBXL17 differs by 17 to 33 Å between the newly solved CSN-bound structure and published structures, while FBXL1 varies by at most 4.5 Å across more than 19 structures.
FBXL2's flexibility appears as a continuously sampled conformational ensemble, and no single static structure explains its cross-links to the CSN core. In the static cryo-EM model only 1 of 7 high-confidence FBXL2-CSN cross-links satisfies the 35 Å distance constraint, whereas intra-protein cross-links are satisfied at 93% (129 of 138). Re-docking FBXL2's substrate-binding domain with PatchDock under cross-link distance restraints and an RMSD-below-20-Å proximity limit produced 8,248 candidate conformations; no single conformation satisfied more than three cross-links, and a four-state ensemble raised FBXL2-CSN cross-link satisfaction to 86% (6 of 7).
Substrate binding narrows rather than eliminates FBXL2's conformational space, which differs from a rigid-body model in which CSN senses substrates through steric exclusion. In the apo state, three of the seven high-confidence FBXL2-CSN cross-links were completely absent in all six replicates of the GGTase3-bound composition. Superimposing the resolved FBXL2-GGTase3 structure (PDB 6O60) onto the four-state conformations that account for those cross-links produces severe steric clashes with SKP1 or CSN1, indicating that those conformations are excluded when substrate is present.
Perspective
The conclusions rest on in vitro reconstituted human CSN, SCF and GGTase3 complexes, with cryo-EM maps at 3.9 Å (CSNDM-SCFFBXL2-GGTase3), 3.7 Å (CSNDM-SCFFBXL17) and 5.0 Å (CSNDM-SCFFBXO22); the authors note it is unknown whether CSNDM-SCFFBXL2-GGTase3 assembles under physiological conditions.
The authors state that the four-state ensemble, limited by sparse distance restraints, should be viewed as one representative realization rather than a unique structural solution; the family-wide flexibility survey relies mainly on the single computational predictor BioEmu and would benefit from further experimental validation; and although FBXL1 is far more rigid than FBXL17, some degree of conformational variability cannot be excluded.