MuziMùzǐ OuyangŌuyáng (Mia)

Targeted protein degradation can reach proteins that conventional inhibitors cannot, because it only has to bind a target, not occupy a functional pocket. I build bioPROTACs — genetically encoded degraders that fuse a target-binding domain to an E3 ligase — and ask what happens when they stop working.

My dissertation approaches that from two directions. One is a dual-target bioPROTAC co-degrading ERα and PI3K p85α, which re-sensitizes tamoxifen-resistant breast cancer cells to endocrine therapy and acts synergistically with tamoxifen (CI < 0.4). The other is a KRAS-targeting degrader built on a TRIM21–RBCC scaffold, where I identified a mutation in the hypervariable region that lets KRAS escape degradation. Because that platform depends on multimerization, the escape is mechanistic rather than incidental — I traced it with all-atom molecular dynamics of the farnesylated, membrane-embedded protein, and confirmed it against an orthogonal, multimerization-independent degrader.

Resistance is the thread running through both. A degrader that works in a naive line and fails in a resistant one is telling you something about the target’s conformational behaviour, and I think that signal is worth reading rather than engineering around.

This is pulling me toward de novo binder design. If the limiting reagent for a bioPROTAC is a binding domain against a hard target, then designing binders directly — rather than borrowing them — is the way to widen what degradation can reach. I have been building that capability independently: reproducing the CLAIRE small-molecule binder-design pipeline end-to-end, and testing whether published multistate-design diagnostics generalize to the conformational states of KRAS. That intersection — structure-based design meeting degrader mechanism — is where I would like to do my postdoc.