Generative design of functional disordered regions in a developmental transcription factor

Abstract

Rapid advances in protein structure design have not translated to intrinsically disordered regions (IDRs). This is largely due to our incomplete understanding of how IDR sequence encodes function. Here, we use stem cell differentiation as a model to map functional sequence features in the IDR of transcription factor RUNX2. We show that RUNX2 requires its C-terminal IDR to bind chromatin and regulate gene expression during osteogenesis. Surprisingly, linear IDR sequence is dispensable, and instead features of the conformational ensemble predict function. We develop an algorithm for rational IDR design based on biophysical principles, and use it to generate functional IDRs de novo. Together, these results illustrate the central role for TF IDRs in chromatin binding and extend generative protein design beyond the folded proteome.