We developed a sequence first computational pipeline to design compact de novo protein binders targeting RBX1, a catalytic component of Cullin RING E3 ubiquitin ligase complexes. Because RBX1 combines a structured RING H2 zinc coordinating domain with a partially flexible N terminal region, our design strategy focused on generating short alpha helical scaffolds capable of forming stable surface contacts with the structured C terminal domain while remaining soluble and foldable as independent monomers.

Candidate binders were generated through protein language model guided exploration of sequence space, enabling sampling of structurally plausible residue environments without reliance on template scaffolds or motif grafting. This allowed construction of diverse compact helical proteins enriched for amphipathic packing patterns commonly observed in protein protein interaction modules.

Generated sequences were filtered using physicochemical constraints including charge balance, secondary structure compatibility, and removal of repetitive low complexity motifs associated with aggregation risk. To ensure compliance with the competition novelty requirement, profile based homology screening against reference proteomes was performed and only sequences without detectable similarity within the UniRef50 similarity regime were retained.

A key challenge in this design task was the absence of a predefined binding interface on RBX1 and the presence of a metal coordinating RING domain whose geometry constrains accessible interaction surfaces. To address this, we prioritized compact helical binders with balanced electrostatic distributions and surface exposed interaction residues compatible with shallow protein protein interfaces rather than deep pocket binding modes. Additional filtering emphasized sequence diversity across the submission set to reduce redundancy and increase coverage of candidate interaction geometries.

The resulting binder panel represents a set of structurally plausible, sequence diverse, and database novel proteins designed to maximize the probability of identifying experimentally testable RBX1 interaction partners.