Abstract
ABSTRACT
Intrinsically disordered regions (IDRs) of proteins show low sequence conservation, and sequence position alone is a weak predictor of disease risk. Phase-separating IDRs (PS-IDRs) are an exception, showing nearly three times higher pathogenic mutation rates than typical IDRs. Here we determine which sequences within that class show the greatest enrichment of pathogenic missense variants and what local features they share. Using pathogenic mutation rates as an independent biological readout, we isolated a core PS-IDR subclass that maximizes pathogenic enrichment. These sequences are defined less by bulk amino acid content than by a refined sticker-spacer grammar in which aromatic residues are embedded in short glycine-rich motifs, while homotypic charge tracts are avoided. Motifs such as YGG/YGGG, FGG/FGGG, and related patterns are elevated 7-to >200-fold relative to conventional IDRs. Comparison of pathogenic and benign substitutions then shows why genetic variation in IDRs has been confounding: the grammar itself is almost never mutated. Instead, the principal pathogenic targets are glycines directly flanked by non-glycine residues. These findings link a specific local sequence architecture to mutational vulnerability and provide a practical framework for interpreting variants in disordered proteins.
SIGNIFICANCE
Disease-causing mutations in unstructured protein regions are common, yet hard to interpret because those sequences look loosely constrained. This study shows that a subset of condensate-forming regions follows a precise molecular grammar. Aromatic residues sit in short glycine-rich motifs that are almost never mutated, while isolated glycines are the main pathogenic targets. That distinction identifies where pathogenic missense variants concentrate in disordered proteins.