Rare, tightly-bound, multi-cellular clusters in the pancreatic ducts of adult mice function like progenitor cells and survive and proliferate after acinar cell injury

Pancreatic ductal progenitor cells have been proposed to contribute to adult tissue maintenance and regeneration after injury, but the identity of such ductal cells remains elusive. Here, from adult mice, we identify a near homogenous population of ductal progenitor-like clusters, with an average of 8 cells per cluster. They are a rare subpopulation, about 0.1% of the total pancreatic cells, and can be sorted using a fluorescence-activated cell sorter with the CD133highCD71lowFSCmid-high phenotype. They exhibit properties in self-renewal and tri-lineage differentiation (including endocrine-like cells) in a unique 3-dimensional colony assay system. An in vitro lineage tracing experiment, using a novel HprtDsRed/+ mouse model, demonstrates that a single cell from a cluster clonally gives rise to a colony. Droplet RNAseq analysis demonstrates that these ductal clusters express embryonic multipotent progenitor cell markers Sox9, Pdx1, and Nkx6-1, and genes involved in actin cytoskeleton regulation, inflammation signaling, organ development and cancer. Surprisingly, these ductal clusters resist prolonged trypsin digestion in vitro, preferentially survive in vivo after a severe acinar cell injury and become proliferative within 14 days post-injury. Thus, the ductal clusters are the fundamental units of progenitor-like cells in the adult murine pancreas with implications in diabetes treatment and tumorigenicity. From adult murine pancreata, the CD133(hi)CD71(low)FSC(mid-high) (abbreviated as FSC-MH) or the CD133(hi)CD71(low)FSC(low) (abbreviated as FSC-L) sub-populations were sorted and subjected to a 10x Chromium device using a 10X V3 Single Cell 3’ Solution kit (10x Genomics, Chromium Single Cell 3’ Regent 00kit V3 Chemistry, Cat. PN-1000092). Approximately 0.1 million reads per cell/unit were sequenced. Raw sequencing data were aligned to the mouse genome (mm10) and the R package Seurat was used for data analysis. Units with <200 detectable genes and >15% mitochondrial genes were excluded.