The human body, composed of approximately 37 trillion cells, lacks a clear understanding of how these cells coordinate within tissues. To address this, a comprehensive pan-tissue single-cell transcriptomic atlas of over 2.29 million cells from 35 healthy human tissues was assembled.
Using a new computational framework, CoVarNet, researchers identified 12 recurring, cross-tissue Cellular Modules (CMs), which represent fundamental, coordinated multicellular ecosystems. The CMs exhibit strong preferences for specific body systems (e.g., immune, reproductive, or vascular units) and captured the majority of non-epithelial cell diversity.
Spatial analysis (Visium and Xenium) confirmed that CMs are spatially organized within tissues, such as distinguishing the inductive (CM05) and effector (CM02/CM03) modules of mucosal immunity in the intestine. Analysis of cell-cell communication showed that the spatial organization and cellular composition of a CM determine its communication style (e.g., lymphocyte-rich CMs favor local interaction).
Crucially, CMs capture tissue-specific biological changes, as seen by CM05 increasing and CM06 decreasing in activity in the spleen during ageing, effectively condensing complex multi-cell variations into functional units. These findings establish CMs as fundamental, conserved units of multicellular coordination across the human body.