107-OR: Profiling of Human Pancreatic Islets by Single Cell Transcriptomics Reveals α- and ß-Cell Subpopulations

Single cell RNA-sequencing (scRNA-seq) approaches are continuously improving at the point of data collection and analysis to simultaneously profile all cell types within islet microorgans to better understand heterogeneity and phenotype-function relationships and how these change with aging and disease. To better understand the advantages and limitations of scRNA-seq, we used the Chromium^TMplatform to profile human islets with a robust insulin and glucagon secretion by perifusion (5 nondiabetic donors, ages 14-66 years) and compared these results to scRNA-seq analysis of purified α and β cells. In context of the same donor (n=2), purified α and β cells (using NTPDase3 marker) showed high correlation (r=0.99) with the Seurat-identified α and β subsets originating from dispersed whole islets. Based on the analysis of a very large pool of single cells (79,881 cells; 2316 genes/cell), the Seurat-guided analysis identified cell types that segregated into eight different clusters (17% β, 66% α, 4% δ, 4% ductal, 4% mesenchymal, 2% acinar, 2% endothelial, and 1% immune). Interstingly, in this large single cell context, we did not identify the four β cell subsets previously reported on the basis of alternative cell surface markers, but found a β cell cluster with highly expressed ER stress-associated genes and an α cell cluster with higher cell cycle regulation marker expression. We concurrently investigated expression of functionally important MAFA and MAFB genes, which are expressed only in a fraction of β cells by histological analysis, and found that PERK-mediated UPR transcripts were enriched in the MAFB-expressing β cells, while the MAFA-expressing β cells differentially expressed genes associated with β cell function. Overall, our study indicates high fidelity of α and β cell clustering in dispersed and sorted islet cells and points to a different pattern of β cell heterogeneity when β cells are purified using NTPDase3 expression compared to other cell surface markers.