Transcription Factor Dynamics Investigated through Single-Molecule Imaging, High-Throughput Sequencing, and Neural Networks
Recent chromatin characterization and sequencing technologies, paired with growing power in computational and bioinformatic analysis, have enabled a deeper understanding of the highly sequence-dependent nature of protein-DNA interactions. Further, these tools have brought to the forefront gaps in our understanding of how changing chromatin landscapes shape cell and tissue identity, and particularly how proteins with stable and transient DNA associations provide feedback to this process. Chromatin remodeling and reorganization serve as umbrella terms to describe diverse mechanisms altering cell epigenetic identity. Transcription factors interacting with chromatin can be influenced by chromatin remodeling processes specifically through cognate sites modifications or generally through a variety of mechanisms, but the degree to which chromatin remodeling alters transcription factor dynamics and activity through general or specific mechanisms is poorly understood. We applied the techniques of Single-Molecule Tracking (SMT) to study the changing dynamics of transcription factors through a B cell activation process marked by widespread chromatin reorganization. First, we identified that during B cell activation, and specifically by the process of nanodomain decompaction, residence time for transcription factors is decreased, suggesting an increased efficiency in transcription. Further studies will be needed to determine if this association between transcription factor residence time and gene transcription is reproducible, and the mechanism underlying it. Second, we identified that the process by which transcription factors scan DNA to identify cognate binding sites, measured by transcription factor random collisions and search time, occurred more rapidly in activated B cells. Given that our work gave additional evidence of the effect of chromatin organization on transcription factor residence time and transcription, we aimed to systematically identify proteins that work upstream to influence the accessibility of chromatin. We generated datasets measuring chromatin accessibility in a variety of mouse tissues and cells, with significant contribution of immune cell subsets. Our accessibility data showed patterns for regulatory elements that fall in line with literature describing significant regions of the genome dedicated to cell-specific regulation, rather than universal regulation. Using a neural network tool known as DeepLIFT with motif identification tools TF-MoDISco and HOMER, we tracked patterns of transcription factor contributions to accessibility across these cell and tissue types, and especially through cell lineages. We identified orphan motifs with no assigned transcription factor, and further identified pleiotropic transcription factors predicting overlooked immune cell functions. Our work stands as a valuable resource for connecting chromatin reorganization and transcription factor dynamics, as well as for testing limits for systematic approaches to predicting contributions of transcription factors to chromatin accessibility.