Publication
A spinal cord injury (SCI) triggers a cascade of molecular and cellular responses involving inflammatory cell infiltration and cytokine release, apoptosis, demyelination, excitotoxicity, ischemia, and the formation of a fibrotic scar surrounded by an astrocyte border. Altering the course of this cascade to improve neurological outcome is a major challenge in the medical management of SCI, and will require a complete understanding of how neural and non-neural cells coordinate the response to the injury over time and across distinct lesion compartments. Previous attempts to delineate the molecular logic governing this response initially turned to bulk transcriptomics and proteomics of the entire lesion. However, these attempts were technically limited in their ability to resolve cell-type-specific molecular programs triggered by injury, or else focused on isolated aspects of the injury response. In the work presented in this thesis, I introduce the roadmap toward the establishment of the Tabulae Paralytica, four molecular and cellular atlases of spinal cord injury (SCI), comprising a single-nucleus transcriptome atlas of half a million cells, a multi-omic atlas pairing transcriptomic and epigenomic measurements within the same nuclei, and two spatial transcriptomic atlases of the injured spinal cord spanning four spatiotemporal dimensions. We faced two main challenges in the establishment of single cell sequencing as a routine tool for the interrogation of spinal circuitry within our group. First, classical approach to compare cell states across two or more experimental conditions did not allow for the accurate identification of specific cell populations undergoing subtle yet profound transcriptional change. We formulated a fundamentally new perspective on how to identify the specific cell types responding to a perturbation, a statistical measure that leverages a machine-learning algorithm to quantify the relative difficulty of separating cells of each type between experimental conditions. On the other hand, deciphering the cellular responses to perturbation primarily rely on differential expression (DE) analysis. The central role of DE in the comparative analysis of scRNA-seq data has made it the focus of several recent benchmarks, however these studies have all come to the unsatisfying conclusion that different methods perform best depending on the circumstances. In an attempt to select a robust methodology, we established the first compendium of single-cell datasets in which the experimental ground truth was known, uncovering the principle that dictates the biological accuracy of DE methods. We built on the expertise acquired in the field of single cell transcriptomic, and established the Tabulae Paralytica, a foundational resource that allowed us to understand the biology of SCI at unprecedented resolution. We uncovered conserved and divergent neuronal responses to the injury, specific neuronal subpopulation primed to become circuit reorganizi