FXYD2 marks and regulates maturity of β cells via ion channel-mediated signal transduction

Abstract

Human pancreatic islets regulate organ development and metabolic homeostasis, with dysfunction leading to diabetes. Human pluripotent stem cells (hPSCs) provide a potential alternative source to cadaveric human pancreatic islets for replacement therapy in diabetes. However, human islet-like organoids (HILOs) generated from hPSCs in vitro often exhibit heterogeneous immature phenotypes such as aberrant gene expression and inadequate insulin secretion in response to glucose. Here we show that FXYD Domain Containing Ion Transport Regulator 2 (FXYD2) marks and regulates functional maturation and heterogeneity of generated HILOs, by controlling the β cell transcriptome necessary for glucose-stimulated insulin secretion (GSIS). Despite its presence in mature β cells, FXYD2 is diminished in hPSC-derived β-like cells. Mechanistically, we find that FXYD2 physically interacts with SRC proto-oncogene, non-receptor tyrosine kinase (SRC) protein to regulate FXYD2-SRC-TEAD1 signaling to modulate β cell transcriptome. We demonstrate that FXYD2^High HILOs significantly outperform FXYD2^Low counterparts to improve hyperglycemia in STZ-induced diabetic immune deficient mice. These results suggest that FXYD2 marks and regulates human β cell maturation via channel-sensing signal transduction and that it can be used as a selection marker for functional heterogeneity of stem cell derived human islet organoids.

Fig. 1. FXYD2 marks a functional maturation in hPSC-derived β cells.

a Schematic of randomized integrated scRNA-seq analyses of 200,106 cells (200 K). The indicated gene sets at each stage of differentiation (a–p) were obtained from publicly available data. b UMAP analyses identified cell type transitions at each stage of the 200,106 cells of scRNA-seq data. c Expression of INS, GCG and SST expression is visualized. d Heatmap analyses of selected differentially expressed genes in each stage of cells. Heatmap is shown by Z-score. e Violin plot showing the expression of FXYD2 (in INS+ cells or SST+ cells), FXYD5 (in GCG+ cells) and FXYD6 (in GCG+ cells). In the combined violin and box plot, the violin shape represents the kernel density estimation, illustrating the distribution of the data, wider sections indicate higher data density. The overlaid box plot shows the median (centre line), first (Q1) and third (Q3) quartiles (box limits), and whiskers extending to the smallest and largest values within 1.5 × the interquartile range (IQR). The datasets from PS = (a), EN = (b)(c), FG = (d), PP = (e), EP = (f)(g), imI = (h)(i), mI = (j)(k)(l), vivo1 = (m), vivo3 = (n), hislets = (p) were analyzed. f Volcano plot showing significantly downregulated genes in hPSC-derived INS+ cells compared to primary human β cells. g qPCR analyses of FXYD2 gene expression in hiPSCs (n = 3), HILOs (n = 10) and primary human islets (hislets) (n = 10). Unpaired two-tailed t-tests. (h, i). Representative immunofluorescence images of insulin (green) and FXYD2 (red) in primary human islets (h) and human β cell line (EndoC-βH1 cells) (i) (scale bars = 100 μm). Error bars indicate SEM.

Fig. 2. FXYD2 enhances insulin secretion in human β cells.

a Human c-peptide secretion (pM) from EndoC-βH1 cells is shown. FXYD2 knockdown (siFXYD2/FXYD2KD) suppresses insulin secretion in EndoC-βH1 cells. siRNA was transfected 72 h prior to the GSIS assay. n = 3. Unpaired two-tailed t-tests. b Dose-dependent suppression of insulin secretion by digitoxin, measured by Gaussian Proinsulin Nano-Luc system in EndoC-βH1 cells. Digitoxin (0, 0.1, 0.5, 1, 2,5, 10 nM) was treated for 24 h prior to GSIS assay. n = 4. One-way ANOVA. c Human c-peptide secretion (pM) from EndoC-βH1 cells is shown. DOX-induced FXYD2 induction enhances insulin secretion in EndoC-βH1 cells. DOX (+) was treated 72 h prior to GSIS assay. n = 3. Unpaired two-tailed t-tests. d Human c-peptide secretion (pM) from primary human islets is shown. Pharmacological FXYD2 inhibition for 24 h by digitoxin (0, 0.1, 1, 10, 100 nM) suppresses insulin secretion in primary human islets. n = 3. One-way ANOVA. e Human c-peptide secretion (pM) from dFXYD2OE-HILOs is shown. DOX (+) was treated every other day from day 20 to day 30 differentiated HILOs. n = 6. Unpaired two-tailed t-tests. G3 = Glucose 3 mM, G20 = Glucose 20 mM, KCl20 = KCl 20 mM, GLP-1 = GLP-1 100 μM. Error bars indicate SEM.

Fig. 3. FXYD2 overexpression and pharmacological inhibition inversely regulate the transcriptome in human β cells.

a Volcano plot showing differentially expressed genes by DOX-inducible FXYD2 overexpression (dFXYD2OE) in EndoC-βH1 cells. -log₁₀ (p-value) > 1.3. b Kyoto Encyclopaedia of Genes and Genomes (KEGG) pathway analysis of upregulated (Up) or downregulated (Down) regulated genes by dFXYD2OE. GSEA analyses for mineral absorption pathway (NES > 1.35, FDR q-value < 0.17) are shown in the right. c The heatmap show dFXYD2OE (DOX +) upregulated the gene necessary for functional β cells, while downregulated cell cycle related genes. d The heatmap show metabolic pathway related gene expression is upregulated by dFXYD2OE (DOX +), while downregulated by pharmacological FXYD2 inhibition (FXYD2i). e Motif analyses revealed that the promoter region of FXYD2-regulated gene clusters often contains the transcription factors (TFs) binding sites listed above. f dFXYD2OE (DOX +) or dFXYD2KD (DOX +) HILOs at day 27 were treated with DOX (+) for 4 days and then gene expression was analysed by qPCR. n = 3. Heatmaps are shown by Z-score.

Fig. 4. FXYD2 regulates signal transduction in functional β cells.

a Schematic of TurboID proximal labelling for FXYD2. EGFP expression (green) is shown in pLV-V5-hFXYD2-TurboID-EGFP expressing EndoC-βH1 cells. Scale bars = 10 μm. b TurboID proximal labelling to identify biotinylated proximal proteins of FXYD2. Immunoprecipitation (IP) by V5-tag and Immunoblot (IB) by streptavidin (SA). Immunoblot result shown is representative of three independent biological replicates. c Proteomics analyses identified unique proteins that co-precipitated with V5-FXYD2. Identified proteins are shown. d IP for SA in control and V5-FXYD2-TurboID-EGFP expressed EndoC-βH1 cells. IB for V5, SRC, CTTN, ATP1A1 and SA. Immunoblot result shown is representative of three independent biological replicates. e Phospho-antibody microarray of control and constitutive FXYD2 overexpressing (cFXYD2OE) EndoC-βH1 cell lysate. f Pathway Ontology (PO) analyses of cFXYD2OE upregulated pathways. Selective kinases (left) and KEGG pathway (right) are shown. g Thyroid hormone signalling-related kinase protein expression. h STRING analyses showing the network of proteins differentially phosphorylated by cFXYD2OE. SRC pathways are highlighted by a pink circle. i Selected heatmaps of differentially regulated phosphosites belonging to SRC pathways in indicated conditions. j Representative IB analysis and quantification for SRC signalling in DOX-induced FXYD2 overexpression (DOX +) in EndoC-βH1 cells (n = 3). DOX (+) was treated for 72 h. Bar graph shows the quantification of IB (n = 3−6). Unpaired two-tailed t-tests. Error bars indicate SEM. Heatmaps are shown by Z-score.

Fig. 5. TEAD1 regulates the key human β cell gene expression.

a Representative IB analysis of indicated phosphorylated and total proteins in EndoC-βH1 cells under the indicated conditions. b qPCR analyses of indicated gene expression in EndoC-βH1 cells. DOX-inducible TEAD1 overexpression (dTEAD1OE, DOX +) and knockdown (dTEADKD, DOX +) regulate the expression of key β cell genes, including MAFA, UCN3, and IAPP. n = 3. Unpaired two-tailed t-tests. c, Human c-peptide secretion assay (fold change to DOX (–) G3 stimulation) from dTEAD1KD EndoC-βH1 cells. n = 12. Unpaired two-tailed t-tests. d–g TEAD1 ChIP-seq analyses. The distribution of relative gene positions at the peak summit (d), enriched motif analyses (e), KEGG pathway analyses (f) in control and cFXYD2OE EndoC-βH1 cells. g Browser tracks showing TEAD1 ChIP-seq peaks at MAFA, PDX-1, NKX6-1, UCN3, IAPP, NEUROD1 loci in control and cFXYD2OE EndoC-βH1 cells. Error bars indicate SEM.

Fig. 6. Generation of FXYD2^High HILOs.

a Drug screening (1 μM, 24 h stimulation) in FXYD2 promoter driven luciferase and mCherry expressing EndoC-βH1 cells. Dex and T3 upregulated FXYD2 promoter activity. The heatmap is shown by three independent screening data with fold change to control (DMSO, n = 18). b qPCR analyses of FXYD2 gene expression in EndoC-βH1 cells. Dex (10 μM for 24 h), T3 (1 μM for 24 h) and retinoic acids (RA, 2 μM for 24 h) synergistically enhanced FXYD2 gene expression in EndoC-βH1 cells. Brain-derived neurotrophic factor (BDNF, 10 μg/ml for 24 h) and ciliary neurotrophic factor (CNTF, 10 μg/ml for 24 h) are used as negative control. n = 3. Unpaired two-tailed t-tests. c qPCR analyses of indicated gene expression in EndoC-βH1 cells and hiPSC-derived β-like cells. NaCl stimulation enhanced FXYD2 and ATP1B1 gene expression in EndoC-βH1 cells and hiPSC-derived β-like cells (dispersed HILOs in 2D). n = 3. Unpaired two-tailed t-tests. d qPCR analyses of FXYD2 gene expression in EndoC-βH1 cells. Dex and NaCl synergistically stimulate FXYD2 expression in EndoC-βH1 cells. n = 3. Unpaired two-tailed t-tests. e Cell survival was measured by cellular ATP level (ratio). dFXYD2OE (DOX +) showed resistance to NaCl (osmotic pressure)-induced cell death in EndoC-βH1 cells. In contrast, FXYD2KD shows increased sensitivity to NaCl induced cell death in EndoC-βH1 cells. n = 3. Two-way ANOVA. f Schematics of generation of FXYD2^High and FXYD2^Low HILOs. Transient Dex 10 μM and NaCl 180 mM stimulation was used to improve visualization of FXYD2. g Representative fluorescence images of FXYD2-driven mCherry (red) and corresponding optical image. FXYD2^High and FXYD2^Low HILOs can be identified by FXYD2-fLuc-mCherry expression. FXYD2^High and FXYD2^Low enriched insulin positive cells were isolated from HILOs. Size-matched pseudo-HILOs (pHILOs) were generated from FXYD2^High and FXYD2^Low expressing HILOs. h Human c-peptide secretion response (pM/10,000 cells) in FXYD2^High and FXYD2^Low pHILOs. n = 3. Unpaired two-tailed t-tests. Scale bars = 100 μm. G3 = Glucose 3 mM, G20 = Glucose 20 mM, KCl20 = KCl 20 mM. Error bars indicate SEM.

Fig. 7. INS^High/FXYD2^High HILOs exhibit enriched transcriptomic profiles indicative of functional maturation.

a Sorting strategy of INS^High/FXYD2^Low and INS^High/FXYD2^High HILOs. b–d Bulk RNA-seq analyses of INS^High/FXYD2^Low and INS^High/FXYD2^High HILOs. KEGG pathway analysis (b), Differential expression gene (DEG) analyses with -log₁₀ (padj = p-value adjusted) > 1.3. (c) and Heatmap analyses (d) of INS^High/FXYD2^Low and INS^High/FXYD2^High HILOs are shown. Samples were obtained from n = 3 independent batches of differentiations.

Fig. 8. FXYD2^High HILOs restores euglycemia in diabetic mice in vivo.

a Secreted Gaussia luciferase and intracellular firefly luciferase activity in each HILOs is shown. Individual HILO insulin secretion was measured by the Gaussia Nano-Luc system and intracellular FXYD2 gene activity was measured by firefly luciferase. n = 48. b Correlation of secreted Gaussia Nano-Luc and intracellular FXYD2 promoter driven firefly luciferase. XY analyses is shown by simple linear regression. R² = 0.509, p < 0.0001. c INSULIN promoter driven GFP (green), FXYD2 promoter driven mCherry and optical image of isolated FXYD2 enriched (FXYD2^High) and less enriched (FXYD2^Low) HILOs (scale bars = 300 μm). d Human insulin secretion (mU/L/10,000 cells) from FXYD2^High and FXYD2^Low HILOs. n = 9. Unpaired two-tailed t-tests. e Schematic of transplantation. f Fed ad lib, blood glucose levels in 500 clusters (~1500 IEQ) of FXYD2^High (n = 6), FXYD2^Low (n = 5) or Sham (n = 2) transplanted STZ-NOD-SCID mice for ~12 weeks. Graft removal (nephrectomy, Nx) was performed on day 84 post-transplantation. Two-way ANOVA. g % of body weight change during the transplantation study is shown. FXYD2^High (n = 6), FXYD2^Low (n = 5) or sham (n = 2). Unpaired two-tailed t-tests. h Serum human c-peptide levels (pM) were measured before and 15 min after i.p. glucose injection at 8 weeks after transplantation. The right panel shows the stimulation index to GSIS (fold). FXYD2^High (n = 6), FXYD2^Low (n = 5). Unpaired two-tailed t-tests. i Representative immunofluorescence images of insulin (green) and FXYD2 (red) in transplanted kidney sections 84 days after transplantation (scale bars = 100 μm). G3 = Glucose 3 mM, G20 = Glucose 20 mM. Error bars indicate SEM.