Abstract Neurological injury is often accompanied by extensive infiltration of macrophages along with activation of fibroblasts and endothelial cells. The activity of these cells is associated with elevated levels of various proteases, which contribute to the hydrolysis of multiple proteins, disrupt the extracellular matrix, and further promote the migration of immune cells into uninjured neural tissue. In this study, we combined single-cell sequencing with bulk RNA sequencing data from spinal cord injury to identify up-regulated protease-related differentially expressed genes post-injury. Using gene set variation analysis, least absolute shrinkage and selection operator regression, and random forest methods, we identified adamalysins, serine proteases, and matrix metalloproteinases as key protease types. Weighted gene co-expression network analysis combined with machine learning algorithms helped predict critical protease genes involved in spinal cord injury. Immune infiltration and single-cell analyses were applied to identify cell types enriched in proteases and their spatial localization. Molecular docking and in vivo and in vitro assays using a mouse model of spinal cord injury were used to validate potential drug interactions. We identified Mmp12 and Adam17 as key effectors regulating injury progression, and determined that macrophages, fibroblasts, and monocytes are the primary cells mediating the functions of core proteinases after injury. Subsequent in vivo and in vitro experiments demonstrated that selective inhibition of key protease activity with marimastat reduced axonal demyelination and fibrous scar formation after spinal cord injury, thereby promoting the recovery of neurological function. Our study identified the key proteases that regulate spinal cord injury repair along with their mechanisms of action, and verified that inhibiting these proteases effectively alleviates scar formation and inflammatory cell infiltration, providing novel therapeutic targets for the treatment of spinal cord injury.
Jian et al. (Tue,) studied this question.