Back

Single-cell multiome analysis supports α-to-β transdifferentiation in human pancreas

Lee, M. Y. Y.; Guessoum, O.; El-Mekkoussi, H.; Conery, M.; Manduchi, E.; Schug, J.; Descamp, H.; Lahori, D.; Da, T.; Liu, C.; Naji, A.; Voight, B.; Li, M.; Kaestner, K. H.

2025-02-19 systems biology
10.1101/2025.02.14.638309 bioRxiv
Show abstract

Spontaneous transdifferentiation of pancreatic glucagon-producing alpha to insulin-secreting beta-cells has been observed in mouse but not in human islets1. Here, we analyzed the largest single-cell dataset of human islets to date, composed of 650,000 cells across 121 deceased organ donors, in search of transitional cell states. By integrating single-cell RNA-seq, single-nucleus ATAC-seq and single-nucleus multiome (joint RNA and ATAC profiling) datasets generated by the Human Pancreas Analysis Program (HPAP)2,3 we identified two previously undescribed cell populations (c11 and c13 cells), which together represent transitional states between alpha- and beta-cells. Some c11 cells are insulin-positive while others are glucagon positive, but none are double-positive. C11 cells repress alpha-cell identity genes and activate beta-cell specific genes. Moreover, the transcriptomic and epigenetic profiles of c11 and c13 cells indicate a transitioning phenotype driven by lineage-specific transcription factors. Genetic lineage tracing in primary human islet cells confirmed alpha-to-beta cell transdifferentiation. C11 and c13 cells exist in all islet samples regardless of disease statuses, with type 2 diabetic samples having significantly more transitioning cells than matched non-diabetic controls. The discovery of these transitional cell types suggests a possibility for future therapy - transdifferentiating alpha-cells to beta-cell through activation of the c11 gene program.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.