Unraveling the regulatory code of gene expression in plants

ongoing

Multicellular plants control the fine-grained expression of thousands of genes during growth, development, and in response to external stimuli. While gene expression varies strongly between different organs and tissues, how this specificity is encoded in the genome is largely unknown. Although various regulatory DNA sequences and sequence-specific transcription factors controlling diverse transcriptional responses have been described, the combinatorial nature of the regulatory code makes the application of rational approaches to learn and predict context-specific gene expression challenging. This is a major obstacle towards engineering plant promoters with predictable and tuneable expression. multiCODE’s aim is uniting single-cell genomics, explainable artificial intelligence (xAI), and synthetic promoter engineering, to efficiently learn and validate regulatory sequences controlling gene expression in plants. Based on high-resolution single-cell gene expression profiling in the model Arabidopsis and the crop Brassica rapa, xAI models will be built to predict gene expression in leaves under control and stress conditions while at the same time identifying the underlying regulatory DNA sequences and syntax. The power of these predictive models will be validated by exploiting inter- and intra-species sequence variation, revealing the evolutionary conservation of the regulatory code. Furthermore, a novel multi-tier synthetic promoter engineering approach will be used to experimentally validate, directly in plants, the learned regulatory code in a high-throughput manner. By combining the power of advanced computational and experimental methods, I will identify novel regulatory sequences and design novel synthetic promoters controlling gene expression in complex cellular contexts. This represents a crucial step towards a comprehensive understanding of gene regulation and will significantly contribute to developing stress-resilient crops.