Decoding the regulatory architecture of the maize leaf

Authors: TU, XIAOYU - The Chinese University Of Hong Kong (CUHK); MEJIA-GUERRA, MARIA - Cornell University; VALDES FRANCO, JOSE - Cornell University; TZENG, DAVID - The Chinese University Of Hong Kong (CUHK); CHU, PO-YU - The Chinese University Of Hong Kong (CUHK); DAI, XIURU - Shandong Agricultural University; LI, PINGHUA - Shandong Agricultural University; Buckler, Edward - Ed; ZHONG, SILIN - The Chinese University Of Hong Kong (CUHK)

Submitted to: bioRxiv
Publication Type: Pre-print Publication
Publication Acceptance Date: 12/15/2020
Publication Date: 1/8/2020
Citation: Tu, X., Mejia-Guerra, M.K., Valdes Franco, J.A., Tzeng, D., Chu, P., Dai, X., Li, P., Buckler IV, E.S., Zhong, S. 2020. Decoding the regulatory architecture of the maize leaf. bioRxiv. 54:34-41. https://doi.org/10.1101/2020.01.07.898056.
DOI: https://doi.org/10.1101/2020.01.07.898056

Interpretive Summary: It's become increasingly clear that to identify what controls the myriad of ways a plant grows in a given environment, it’s not enough to simply sequence its genes. What is becoming more important to understand is what regulates the expression of those genes. For this purpose, we set out to analyze a large set of the proteins known as transcription factors (TF), which are responsible of regulating gene expression by binding the DNA surrounding gene regions. With these experiments, and by implementing machine-learning methods that look at where these TF proteins are located in the genome, we were able to identify that these proteins do not simply act by themselves, but actually require the specific co-localization of multiple sets of them, effectively working as a network of TF interactions in order to properly carry their regulatory function. Having identified where these TF proteins bind, along with the interactions between them, we expect new models will shed new light in the overall regulatory machine of plants, which in return will help in developing better and more accurate genomic prediction models.

Technical Abstract: Characterize transcription factors (TF) binding to cis-regulatory elements is key to understand the effect of genetic variability on phenotypic differences. Here, we used 104 TF ChIP-seq to annotate the regulatory landscape of the maize leaf. The ~2 millions of identified TF binding sites colocalized with open chromatin, are evolutionarily conserved, and show enrichment for GWAS-hits and cis-expression QTLs. The resulted TF-target regulatory network covers half of the annotated genes and shows characteristics of real-word networks such as scale-free topology and modularity (structural and functional). Machine-learning analyses reveal that sequence preferences are alike within TF families, and that co-localization is key for binding specificity. Our comprehensive approach provides a starting point to decipher the gene regulatory architecture in plants.