Deep-learning-based design of an orthogonal self-labeling protein from K-Ras(G12C)

Self-labeling protein tags are invaluable tools for tracking proteins in living cells, yet only a handful exist, restricting the ability to observe multiple proteins simultaneously. Researchers have now devised an innovative, third, mutually exclusive SLP system using the oncoprotein K-Ras(G12C) and its covalent inhibitors. Utilizing a deep-learning-based methodology, they meticulously redesigned the K-Ras(G12C) sequence while maintaining its covalent-inhibitor binding site.

The resulting LUCI-tag is a compact 19 kDa protein capable of forming stable bonds with a wide range of K-Ras(G12C) inhibitors, each carrying diverse chemical payloads. The novel tag's speed and versatility in covalently reacting with inhibitors surpass those of existing SLPs, HaloTag7 and SNAP-tag, particularly when dealing with negatively charged payloads.

Structural analysis via X-ray crystallography confirmed the tag's accuracy and revealed a preorganized inhibitor-binding pocket responsible for its rapid labeling kinetics. Extensive testing demonstrated that LUCI-tag remains biologically inactive, making it a safe and effective labeling tool. Demonstrating the practicality of their approach, LUCI-tag facilitated rapid, non-invasive live-cell imaging and enabled simultaneous three-color multiplexed experiments alongside HaloTag7 and SNAP-tag.

This groundbreaking work establishes LUCI-tag as a valuable orthogonal SLP, showcasing the potential for repurposing covalent drug-target pairs into a versatile platform for protein labeling.