Decoding neuronal wiring by joint inference of cell identity and synaptic connectivity

Abstract

Animal behaviors are executed by motor neurons (MNs), which receive information from complex pre-motor neuron (preMN) circuits and output commands to muscles. How motor circuits are established during development remains an important unsolved problem in neuroscience. Here we focus on the development of the motor circuits that control the movements of the adult legs in Drosophila melanogaster. After generating single-cell RNA sequencing (scRNAseq) datasets for leg MNs at multiple time points, we describe the time course of gene expression for multiple gene families. This analysis reveals that transcription factors (TFs) and cell adhesion molecules (CAMs) appear to drive the molecular diversity between individual MNs. In parallel, we introduce ConnectionMiner, a novel computational tool that integrates scRNAseq data with electron microscopy-derived connectomes. ConnectionMiner probabilistically refines ambiguous cell type annotations by leveraging neural wiring patterns, and, in turn, it identifies combinatorial gene expression signatures that correlate with synaptic connectivity strength. Applied to the Drosophila leg motor system, ConnectionMiner yields a comprehensive transcriptional annotation of both MNs and preMNs and uncovers candidate effector gene combinations that likely orchestrate the assembly of neural circuits from preMNs to MNs and ultimately to muscles.