Understanding the rules of Arabidopsis activation domain activity through synthetic activation domains

More than 1,900 transcription factors (TFs) control and tune gene expression in Arabidopsis thaliana, and regulation through these TFs plays an important role in the plant’s responses to climate change induced stresses. These TFs are predominately identified by well-conserved DNA binding domains. However, the activation domains (ADs) that recruit coactivator complexes are poorly characterized due to their intrinsically disordered nature. To create plants that are transcriptionally tuned to handle climate change, we must interpret the rules that govern AD activity. Using a neural net trained on the 1,553 naturally occurring Arabidopsis ADs, we designed and built a synthetic AD library that covers this naturally occurring feature space identified with the neural net. This synthetic library was screened for activity with the neural net and with a flow cytometry-based AD scoring methodology. We found that the synthetically designed ADs met the criteria to be considered activators in the FACS assay but were not as strong as the naturally occurring ADs. Comparison of these two results indicates how well we have interpreted the neural net’s characterization of AD activity rules. These findings provide insight into AD behavior and provide encouraging results that synthetically designed AD sequences can be used to tune transcription.