02-21-23. Adipose tissue development impacts many metabolic functions and adipocyte dysregulation contributes to conditions such as obesity, diabetes, and coronary artery disease. Full characterization and interpretation of the molecular changes and regulatory factors that drive differentiation, such as adipogenesis, and responses to stimuli will someday permit selectively steering diseased cells within a human body toward desired phenotypes. Our understanding of the transcriptional responses to hormones, drugs, cytokines, stimuli, and genetic intervention is needed to ultimately realize this future. In this study, the Guertin lab presents a novel approach to constructing gene regulatory networks that incorporates kinetic chromatin accessibility (ATAC-seq) and nascent transcription (PRO-seq) data. Prior characterizations of the transcriptional network driving adipogenesis overlooked essential transcription factors, genes, and regulatory elements that act transiently. Traditional gene regulatory networks lack mechanistic details about individual relationships between regulatory elements and genes. Moreover, they do not account for temporal relationships that are needed to define a regulatory hierarchy that prioritizes key regulatory factors. The networks from this work are designed so that one can easily identify key regulatory transcription factors or regulatory element hubs, determine a set of target genes for specific transcription factors, assess transcription factor cooperativity, and develop testable hypotheses.
This work found that Twist2 is a novel regulator of adipogenesis, an observation that was overlooked previously because TWIST2 acts transiently. This finding highlights the power of this relatively unbiased methodology to articulate novel hypotheses and implicate new factors that inform upon developmental processes. The network provides a wealth of information about regulatory relationships between enhancers and genes. To validate their work, TWIST2 knockout mice were found to have deficiencies in fat storage.
What is perhaps most exciting to the developmental biology and system biology fields is that these networks are constructed using only ATAC-seq and PRO-seq data. Unlike other molecular genomics assays, these two assays can theoretically be performed in any cell type and organism with a reference genome—no species-specific or factor-specific reagents are needed. To facilitate the adoption of this methodology, the Guertin lab provided an analysis vignette on GitHub that contains a step-by-step procedure for generating and interpreting these networks. This network inference framework is a powerful and general approach for interpreting complex biological phenomena and can be applied to a wide range of cellular processes.