Single-nucleus RNA-sequencing reveals singular gene signatures of human ductal cells during adaptation to insulin resistance

Adaptation to increased insulin demand is mediated by ß-cell proliferation and neogenesis among other mechanisms. Although it is known that pancreatic ß-cells can arise from ductal progenitors, these observations have been limited mostly to the neonatal period. We have recently reported that the duct is a source of insulin secreting cells in adult insulin resistant states. To further explore the signaling pathways underlying the dynamic ß-cell reserve during insulin resistance we undertook human islet and duct transplantations under the kidney capsule of immunodeficient NOD SCID gamma (NSG) mouse models that were either pregnant, insulin resistant or had insulin resistance superimposed upon pregnancy (pregnancy+insulin resistance), followed by single-nucleus RNA-sequencing (snRNA-seq) on snap-frozen graft samples. We observed an upregulation of proliferation markers (e.g., NEAT1), expression of islet endocrine cell markers (e.g., GCG and PPY) as well as mature ß-cell markers (e.g., INS), in transplanted human duct grafts in response to high insulin demand. We also noted downregulation of ductal cell identity genes (e.g., KRT19 and ONECUT2) coupled with upregulation of ß-cell development and insulin signaling pathways. These results indicate that subsets of ductal cells are able to gain ß-cell identity and reflect a form of compensation during the adaptation to insulin resistance in both physiological and pathological states. Overall design: Frozen grafts were transferred to dounce tissue grinder tubes (D8938; Sigma) containing 0.5 ml ice-cold Nuclei EZ lysis buffer (NUC-101; Sigma) and homogenized with pestles A and B on ice each for one minute. Samples were transferred to clean 15 ml tubes and homogenizers were rinsed with 1.5 ml buffer followed by 2 ml buffer and transferred to the same 15 ml tubes to get final 4 ml. The tubes were vortexed briefly at moderate speed and set on ice for 5 minutes for cell lysis. To separate nucleus and cytoplasm, the tubes were centrifuged at 500x g for 5 minutes at 4C. Supernatants containing cytoplasmic component were saved for later analysis. Pellet containing nuclei was resuspended in 0.5 ml cold buffer by vortex briefly at moderate speed and 3.5 ml cold buffer was added. Nuclei suspension was mixed by vortex briefly and set on ice for 5 minutes. The tubes were centrifuged at 500x g for 5 minutes, supernatant was saved for later analysis, pellet was resuspended in suspension buffer (0.5 ml PBS containing 0.01% non-acetylated BSA; Sigma and 0.1% RNase inhibitor; 2313A from Clontech). Nuclei suspension was pipetted ten times with a 1 ml tip, filtered through a 30 µm pre-separation filter (130-041-407; Miltenyi biotech), cell number and cell viability were determined by cell counter using 0.4% trypan blue stain. The average number of total nuclei obtained from one-half graft was approximately 8.5E+5 nuclei (1.7E+6 cells/ml) with 5-10 um size and 93.3 +/- 1.1% dead cell rate (n=32 samples across three independent experiments). Dead nuclei number was adjusted to 1000 nuclei/ul with suspension buffer and 10,000 nuclei were immediately used for generation of Gel Beads in Emulsion (GEMs) and barcoding. Leftover nuclei were saved for later analysis. GEMs were generated using the Chromium 3' Single Cell Library Kit (v2, 10X Genomics, CA) according to the manufacturer's instructions. Briefly, 10,000 nuclei were combined with Single Cell Master Mix and encapsulated into the barcoded Gel Beads through the Chromium™ Controller. After GEM-RT incubation, cDNA samples were recovered, purified and amplified through a cDNA Amplification Reaction. Quality controls on amplified cDNA samples were carried out through using a High Sensitivity DNA Kit (Agilent, CA) on a 2100 BioAnalizer (Agilent, CA) platform. Libraries were then constructed following Fragmentation and Adaptor Ligation and Sample Index incorporation. Finally, purified libraries were run on 2100 BioAnalizer (Agilent, CA) using a High Sensitivity DNA Kit (Agilent, CA) to evaluate the quality of the ~400bp fragments. The final single-cell libraries were sequenced using acoverage of 500,000 pair-ended reads targeted per cell, on a HiSeq 4000 platform (Illumina, CA, USA).