A Transcriptional Map of Human Tonsil Architecture: Beyond the Sum of (Single Cell) Parts

The tonsil is a highly compartmentalized organ in which different microanatomical structures orchestrate designated (immune) functions. We use this already well-studied tissue to survey imaging-based spatial transcriptomics data for studying immune responses in native tissue context, and, to demonstrate its advantages for faithfully recapitulating cellular composition in direct comparison with single-cell RNA sequencing. While these data still pose many analytical challenges and lack standardization, we established a versatile analysis pipeline focused on their profitable particularities: considering organization (microenvironment), interactions (signaling), and function (higher order structures) across scales. Specifically, we resolve <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:semantics><mml:mo>∼</mml:mo> <mml:annotation>$\sim$</mml:annotation></mml:semantics> </mml:math> 2M cells into 52 subpopulations across immune and, in particular, structural compartments. Various spatial niches partition tonsillar tissue into architecturally and functionally distinct regions, which we characterize through cell-cell colocalization and communication analyses, while performing various nonstandard analyses at the level of spatial features. These topological readouts may help elucidate where certain immunological processes occur (e.g., class switch recombination), and, where signaling pathways are active (e.g., TNF and galectin, which have been implicated in diverse lymphomas). In all, we provide an analytical framework for spatial immunology, and showcase alternative views that such techniques and concomitant computational approaches can bring on tissue composition and architecture.