UCSF

The basement membranes surrounding human pancreatic islets and their vasculature are essential to β cell survival, polarity, and insulin secretion, yet the cells that build and maintain this extracellular matrix in the adult human pancreas, and the way it is remodeled in type 2 diabetes, have remained poorly defined. The prevailing model holds that endothelial cells are the principal source of the vascular basement membrane, but the close physical association of endothelial cells, pericytes, and fibroblasts has made it difficult to assign secreted matrix proteins to a cellular source, because conventional histology cannot identify the producing cell and dissociation-based sequencing discards spatial context. To resolve these questions, this work combined computational re-integration of publicly available human pancreas single-cell RNA-sequencing data with donor-resolved MERFISH spatial transcriptomics of non-diabetic and type 2 diabetic human pancreas, using a custom 300-gene panel that included 98 extracellular matrix genes. Islet boundaries were segmented manually, vascular populations were validated through spatial neighborhood and proximity analyses, and the single-cell and spatial datasets were treated as orthogonal analyses, with further validation against an anatomically resolved reference dataset, immunofluorescence, and RNA in situ hybridization. These analyses identify pericytes, rather than endothelial cells, as the predominant transcriptional source of vascular basement membrane genes, including COL4A1, COL4A2, LAMA4, and LAMB2, while endothelial cells contribute the complementary components HSPG2 and LAMA5. A spatially distinct islet-associated fibroblast population occupies the endocrine boundary and expresses a peri-islet matrix program enriched for LAMB1 and LAMC1. Pericytes retain a conserved identity, while specializing between the endocrine and exocrine compartments. In type 2 diabetes, the endocrine vascular niche is remodeled, shifting the fibroblast-to-pericyte balance, reducing PDGFRB and basement membrane programs, and increasing contractile, stress, and inflammatory signatures. These findings support a cooperative, multicellular model in which endothelial cells, pericytes, and specialized fibroblasts build the human islet basement membrane, and they reveal its coordinated disruption in type 2 diabetes.