Paired single-cell and spatial transcriptional profiling reveals a central osteopontin macrophage response mediating tuberculous granuloma formation in humans and zebrafish [scRNA-Seq]

Mycobacterial granulomas are categorical manifestations of tuberculosis pathogenesis. They result from an ensemble of immune responses to Mycobacterium tuberculosis infection, but the identities, arrangement, cellular interactions and regulation of the cells that comprise them has thus far been incompletely understood. We conducted spatial and single-cell RNA sequencing of clinical granuloma biopsy specimens from tuberculosis patients. We found that granulomas consist of concentric transcriptional laminae surrounding areas of central necrosis. Gene expression programs associated with regional architecture were conserved among samples. We identified distinct populations of granuloma-associated stromal cells, fibroblasts, lymphocytes, mast cells, dendritic cells, neutrophils and macrophages. Populations further differentiate into multiple granuloma-specific states that correlate with location. We used inferential analysis to predict dominant granuloma cell-cell interactions, the activity of major signaling pathways, and differential transcription factor activities. Using spatial deconvolution, we mapped a conserved pattern of cellular organization dominated by macrophages rich in osteopontin/SPP1 expression. Trajectory analysis of macrophage subtypes mapped their differentiation and supported the importance of SPP1 to granuloma macrophage polarization. Cumulatively, we have identified a dominant macrophage granuloma population as well as its central regulatory gene in human samples and confirmed its importance to granuloma biology in vivo. Overall design: We processed tissue sections immediately adjacent to the Visium-imaged sections. We used these closely paired dissociated cells to determine at a single-cell level the transcriptional signature and cellular heterogeneity in human M. tb granulomas and then comprehensively mapped these transcriptional profiles onto the spatial analysis using deconvolution algorithms. To focus more deeply on granulomas in pulmonary tissue, the predominant site of TB disease, we processed 4 M. tb granuloma biopsy specimens resected from 3 patients with pulmonary TB.