From specification to differentiation: Although terminal selectors control the identity of neurons, the initial specification of these fates are instructed by patterning mechanisms in neural stem cells. These are generally mediated by transiently expressed transcription factors which are often not maintained in postmitotic neurons. How the combined action of these transient mechanisms are converted into a stable selector code in newborn neurons remains unknown. By studying the differentiation trajectories in single-cell multiomics data acquired during neurogenesis, we aim to identify the cis-regulatory elements that control the initial activation of unique selector combinations in each neuron, as well as the upstream regulators that act on these elements. Our ultimate goal is to produce the first comprehensive description of a gene regulatory network that generates neuronal diversity in a complex brain. We hope that insights gained from these studies will have broad translational implications, such as targeted differentiation of induced stem cells to specific neuronal fates for cell replacement therapy.

Gene regulatory networks of synaptic specificity: We have shown that terminal selectors control all type-specific features of optic lobe neurons, including connectivity. Yet the regulatory networks that control brain wiring downstream of selectors are still mostly unexplored. Our goal is to first produce computational models of these networks using gene regulatory network (GRN) inference algorithms that can integrate genome-wide gene expression and cis-regulatory (multiome) datasets to accurately predict the target genes regulated by each transcription factor. Experimental validations of these models then allow us to describe the regulatory subroutines that control specific cellular decisions during neuronal development. We propose that this “top-down” approach to study brain wiring, i.e., first identify transcription factors that cause broad changes and then move down to the effector targets guided by computational models, can overcome the challenges more “bottom-up” approaches have been facing due to widespread redundancies among cell-surface proteins that ultimately control intercellular interactions. In the long-term, our results may enable to engineer ‘designer’ neurons that can be targeted to desired brain areas and form synapses with a predictable set of partners.