Harmine selectively enhances human beta cell identity markers and function via protein kinase A pathways

AIMS/HYPOTHESIS: Harmine and other small-molecule inhibitors of the kinase, dual tyrosine-regulated kinase 1A (DYRK1A), induce human beta cells to replicate and regenerate in vitro and in vivo and are effective at reversing diabetes in animal models. In addition to its beta cell proliferative and regenerative effects, which result from inhibition of DYRK1A ('Target 1'), harmine also induces expression of essential transcription factors and other genes involved in beta cell identity and function, exemplified by PDX1, MAFA, NKX6.1, MAFB, SIX3, SLC2A2, ENTPD3 and others. Harmine also rapidly enhances glucose-stimulated insulin secretion in vitro and in vivo. These beta cell identity-enhancing effects are not due to DYRK1A inhibition. We hypothesise that they reflect interactions of harmine with a second target ('Target 2'). We aimed to identify Target 2. METHODS: We performed an in-depth analysis of multiple different small-molecule inhibitors of DYRK1A in human pancreatic islets using gene expression, single-cell RNA-seq, proliferation, immunocytochemistry of beta cell transcription factors and identity markers, and solid-phase precipitation combined with proteomics. RESULTS: All small-molecule DYRK1A inhibitors studied (harmine, 2-2c, 5-IT, INDY, leucettine, CC-401, GNF4877) induced human beta cell proliferation via inhibition of DYRK1A. In contrast, silencing DYRK1A had no effect on beta cell transcription factors and identity markers. Further, harmine, 2-2c and 5-IT but not INDY, leucettine, CC-401 or GNF4877 enhanced expression of beta cell phenotypic markers. The enhanced expression of beta cell identity markers resulted from an unexpected activation by harmine of an as-yet unidentified Target 2 that activates the protein kinase A (PKA) pathway. CONCLUSIONS/INTERPRETATION: These findings make it clear that all DYRK1A inhibitors are not interchangeable. Those that drive both beta cell proliferation and identity/function will likely be preferable for diabetes therapy. Unravelling the mechanisms controlling the complex harmine-PKA-DYRK1A interactome will provide fruitful ground for future studies.