Abstract
Transcription factors activate gene expression with separable DNA binding domains and activation domains (Latchman, 2008). High-throughput studies have uncovered rules for how DNA binding domains recognize their cognate DNA motifs, but the design principles of activation domains remain opaque. For over thirty years it has been a mystery why activation domains are acidic and unstructured (Sigler, 1988). Activation domains require hydrophobic motifs to bind coactivators and join transcriptional condensates, but low evolutionary conservation and intrinsic disorder have made it difficult to identify the design principles that govern the sequence to function relationship (Boija et al., 2018; Chong et al., 2018; Cress and Triezenberg, 1991; Dyson and Wright, 2016). Consequently, activation domains cannot be predicted from amino acid sequence (Finn et al., 2016). Here, we resolve the functional roles of acidity and disorder in activation domains and use these insights to build a new predictor. We designed sequence variants in seven acidic activation domains and measured their activities in parallel with a high-throughput assay in human cell culture. Our results support a flexible model in which acidic residues solubilize hydrophobic motifs so that they can interact with coactivators. This model accurately predicts activation domains in the human proteome. We identify three general rules for activation domain function: hydrophobic motifs must be balanced by acidic residues; acidic residues make large contributions to activity when they are adjacent to motifs; and within motifs, the presence of aromatic or leucine residues reflects the structural constraints of coactivator interactions. We anticipate these design principles will aid efforts to predict activations from amino acid sequence and to engineer new domains.
Competing Interest Statement
The authors have declared no competing interest.
