Extended Data Fig. 8: (related to Fig. 4). RFX motif enrichment across β cell modules and β cell RFX6 reduction in pseudoislets impairs insulin secretion.

(a) Enrichment of selected transcription factor¹²⁰ motifs in β cell modules, calculated using the AME tool from MEME-Suite with Bonferroni correction⁷⁷; log10(P-values) are unadjusted. (b) Schematic of adenoviral shRNA delivery and formation of pseudoislets, which mirror native human islet architecture^26,121. (c) Relative RFX6 mRNA expression in β cells treated with scramble shRNA (‘control’) or RFX6 shRNA (‘shRFX6’). Data represents n = 3 independent islet preparations. (d) RFX6 knockdown did not change β or α cell proportion (n = 4 independent islet preparations); acute (6-day) reduction of RFX6 expression does not lead to β cell loss. (e) Control and shRFX6 pseudoislets exhibited similar size and morphology. Transduced cells marked by mCherry; distribution of β cells shown by C-peptide (CPEP; blue) and α cells by glucagon (GCG; green). Scale bars, 200 μm (morphology) and 50 μm (immunostaining). (f-i) Dynamic insulin secretion and metrics equivalent to Fig. 4f, g but normalized by total insulin content (h) in control and shRFX6 pseudoislets. (i) Proinsulin content and proinsulin:insulin ratio in control (scramble) and shRFX6 pseudoislets. Proinsulin content was not significantly greater in shRFX6 pseudoislets and the proinsulin:insulin ratio was comparable to controls, suggesting secretory defects were not due to inappropriate insulin processing. Functional data (f-i) represents n = 6 independent islet preparations. Data in panels c-d and f-i show mean + SEM; statistical results (c-d, g-i: two-tailed t-test; f: linear mixed-effect model) indicated as follows: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.