Temporal and cell-specific expression of RFX6 during pancreatic islet differentiation

To examine RFX6 expression during pancreatic islet differentiation, we tracked its levels in hESC-H9 cells at various differentiation stages (ESM Fig. 1a). Due to the lack of a specific antibody for RFX6 immunostaining, we used a modified RFX6HA/HA hESC-H9 line, incorporating a triplicated haemagglutinin (HA) epitope to enable RFX6 protein detection [28]. Initially, RFX6 immunoreactivity was absent during the definitive endoderm and primitive gut tube stages (ESM Fig. 1b). However, robust expression emerged in the posterior foregut (PF), largely co-localised with pancreatic and duodenal homeobox 1 (PDX1) (Fig. 1a). Interestingly, while RFX6 persisted in the PP stage, it did not co-localise with PDX1 or NK6 homeobox 1 (NKX6.1) (Fig. 1a). During the endocrine progenitor (EP) stage, RFX6 co-localised with neurogenin 3 (NEUROG3) and NK2 homeobox 2 (NKX2.2), the pancreatic endocrine markers (Fig. 1a). In the islet cell stage, RFX6 co-expressed with INS, glucagon (GCG) and somatostatin (SST) (Fig. 1a). Flow cytometry analysis confirmed high RFX6 levels from the PF stage onwards, peaking at the EP stage (Fig. 1b). These findings suggest that RFX6 may not be indispensable for PDX1+/NKX6.1+ PPs.

Fig. 1

Timeline expression and single-cell analysis of RFX6 throughout the differentiation of hESCs into various stages of pancreatic development. (a) Immunostaining showing the expression of RFX6 during differentiation of hESC-H9 into pancreatic islets. (b) Flow cytometric quantification of RFX6 expression during different stages of differentiation. (c) Dot plots and feature plots illustrating RFX6 expression across distinct cell clusters. Each dot’s colour and size correspond to the expression level and the percentage of cells expressing the RFX6 gene. (d) The violin plots illustrate the expression distributions of key genes across various clusters at distinct stages of hESC differentiation into pancreatic islets: day 11 (D11); day 14 (D14); day 21 (D21); day 32 (D32); and day 39 (D39). DE, definitive endoderm; PGT, primitive gut tube. Scale bar, 100 µm

To delve deeper into identifying specific cell populations expressing RFX6, we re-analysed the recently published single-cell (scRNA-seq) datasets of PPs at day 11 (D11), EPs at day 14 (D14), immature islets at day 21 (D21) and maturing islets at days 32 and 39 (D32 and D39), derived from hESCs [25]. We used unsupervised clustering to create 2D visualisations using uniform manifold approximation and projection (UMAP) plots and identified multiple cell populations at each differentiation stage (Fig. 1c, d and ESM Fig. 2). At D11, six cell clusters were identified, with three showing high expression levels of pancreatic endocrine markers. RFX6 was mainly expressed in these clusters, with the highest level in C5, distinguished by NEUROG3high/GHRLhigh, which also expressed high levels of other endocrine markers such as PAX4, INSM1, KCNK17, NKX2.2 and SOX4. Moderate RFX6 expression was observed in two additional endocrine clusters (C0 and C4), characterised by CHGAhigh/NEUROD1high and NKX2.2high/SOX4high, respectively. Low RFX6 expression was seen in C1 (APOEhigh/TTRhigh) and C2 (proliferation cluster; TUBA1Bhigh/CENPFhigh). Almost no expression was detected in C3, identified by PDX1high/SOX9high (Fig. 1c, d and ESM Fig. 2). These findings strongly indicate that RFX6 expression in PPs is confined to endocrine cell populations.

At D14, we identified six clusters, with the highest expression of RFX6 observed in the endocrine cluster C5, marked by MAFBhigh/STMN2high; this cluster also showed elevated levels of essential endocrine markers such as INS, GCG and SLC30A8 (Fig. 1c, d and ESM Fig. 2). Another endocrine cluster, C1 (ERO1LBhigh/ARXhigh), displayed high RFX6 expression levels. In addition, a moderate RFX6 level was detected in C2 (POU2F2high/PAX4high), C0 (MNX1high/FEVhigh) and C4 (IRX2high/SCGNhigh). The lowest RFX6 expression was seen in C3 (PP cluster; SOX9high/DLK1high), which also expressed high levels of PDX1, HNF1B, GATA4, TCF7L2 and CCND2. At D21, the highest expression of RFX6 was seen in C3 (POU2F2high/RFX3high), while a moderate expression was seen in C0 (TTRhigh/GCGhigh), C5 (GCGhigh/CHGAhigh) and C6 (delta cell cluster; HHEXhigh/SSThigh). Reduced expression was observed in C1 (TPH1high/FEVhigh), previously identified as a specific enterochromaffin progenitor population [29], and in C4 (proliferation cluster; RESThigh/CCND2high) (Fig. 1c, d and ESM Fig. 2).

At D32, the most significant RFX6 expression was seen in C0 (TTRhigh/GCGhigh) (Fig. 1c), which also expressed high levels of CHGA, IRX2, and ARX, suggesting an alpha cell fate. Moderate RFX6 expression was seen in C4 (CXXC4high/SOX2high) and C5 (proliferation cluster; CENPFhigh/FOXM1high), while lower levels were seen in C1 (TPH1high/FEVhigh) and C2 (INShigh/ERO1LBhigh). ERO1LB (also known as ERO1B) is known as a gene specifically associated with pancreatic beta cells [30]. No expression was observed in C6 (TMSB10high/GSTP1high). At D39, the highest RFX6 expression was seen in C4 (proliferation cluster; CENPFhigh/FOXM1high), with moderate expression in C0 (TTRhigh/GCGhigh). Lower expression was seen in C3 (SSThigh/HHEXhigh), C1 (TPH1high/FEVhigh) and C2 (INShigh/ERO1LBhigh), while no expression was seen in C5 (TMSB10high/TPT1high) (Fig. 1c, d and ESM Fig. 2). Taken together, these findings indicate that RFX6 is mainly expressed in pancreatic endocrine cell populations across various stages and is not expressed in PDX1+ cell populations within PPs.

Depletion of RFX6 diminishes PDX1 expression in the PF and does not affect PDX1+/NKX6.1+ PPs

To explore RFX6’s contribution to human pancreatic development and the generation of pancreatic islet cells, biallelic RFX6 mutant human iPSC lines (referred to as RFX6 KO iPSCs) were established using the CRISPR/Cas9 system. Mutations were introduced into the WT iPSC line generated in our laboratory [18] using a gRNA that targeted exon 2 of RFX6. The mutations were validated through Sanger sequencing and were anticipated to induce a frameshift resulting in the formation of premature stop codons preventing RFX6 protein translation (Fig. 2a). The absence of RFX6 protein expression was confirmed in PPs derived from RFX6 KO cell lines compared with WT controls using western blot analysis (Fig. 2b). All iPSC lines expressed pluripotency markers OCT4 (also known as POU5F1), NANOG, SOX2, SSEA4, TRA-1-60, TRA-81, C-MYC (also known as MYC), KLF4, REX1 (also known as ZFP42), DPPA4 and TERT (ESM Fig. 3a, b). Moreover, they have been verified to maintain normal karyotypes consistent with the parental line and are free from mycoplasma (ESM Fig. 3c, d).

Fig. 2

Loss of RFX6 reduces PDX1 and CDX2 expression in iPSC-derived PF. (a) DNA sequence confirmation of frameshift mutations in isogenic KO iPSC clones compared with WT iPSCs. (b) Western blot analysis confirming the absence of RFX6 protein in PPs derived from RFX6 KO iPSC lines. (c) Immunofluorescence images showing the expression of PDX1, CDX2 and FOXA2 in PPs derived from WT iPSCs and RFX6 KO iPSCs. (d) Western blot analysis showing the expression of PDX1 and CDX2 in RFX6 KO PF compared with WT PF. (e) RT-qPCR analysis showing the mRNA expression of PF markers PDX1, CDX2, ONECUT2, INSM1, TTR, FOXA2, SOX9 and SOX2 in RFX6 KO PF relative to WT control PF (n=4). The data are presented as mean ± SD. ***p<0.001. Scale bar, 100 µm

Next, we examined the effect of RFX6 loss on pancreatic differentiation. Immunostaining and western blotting showed reduced PDX1 and caudal type homeobox 2 (CDX2) protein levels in RFX6-deficient PF cells, while forkhead box A2 (FOXA2) remained unchanged (Fig. 2c, d). qPCR analysis revealed significant downregulation in the mRNA expression of PDX1, CDX2, ONECUT2, INSM1 and TTR in RFX6 KO PF compared with WT PF (Fig. 2e). In contrast, FOXA2, SOX9 and SOX2 expression remained unaffected (Fig. 2e). Despite the dramatic reduction in PDX1 expression during the PF stage, RFX6-deficient iPSCs were able to produce PPs and co-expressed PDX1 and NKX6.1 like controls (Fig. 3a–d and ESM Fig. 4). Other PP markers, including SRY-box transcription factor (SOX)9 and FOXA2, remained unchanged, as evidenced by western blotting and RT-qPCR (Fig. 3c, d). These results suggest that RFX6 does not play a significant role in the formation of PDX1+/NKX6.1+ cells during the PP stage.

Fig. 3

Impact of RFX6 depletion on the expression of crucial pancreatic progenitor and endocrine progenitor markers. (a, b) Immunofluorescence staining (a) and flow cytometry analysis (b) showing the co-expression of PDX1 and NKX6.1 in PPs derived from WT iPSCs and RFX6 KO iPSCs. PDX1+/NKX6.1+ cells are shown in the upper right quadrant in (b). (c) Western blot analysis showing the protein expression of PDX1, NKX6.1, SOX9 and FOXA2 in RFX6 KO PPs compared with WT PPs. (d) RT-qPCR analysis showing the mRNA expression of PP markers PDX1, NKX6.1, FOXA2 and SOX9 in RFX6 KO PPs relative to WT PPs (n=4). (e) Immunofluorescence staining showing the expression of CHGA, NKX6.1 and NKX2.2 in EPs derived from WT iPSCs and RFX6 KO iPSCs. (f) Western blot analysis showing the expression of CHGA in RFX6 KO EPs compared with WT EPs. (g) RT-qPCR analysis showing the mRNA expression of EP markers NEUROD1, NEUROG3, NKX2.2 and PAX4 in RFX6 KO EPs relative to WT EPs (n=4). The data are presented as mean ± SD. ***p<0.001. Scale bar, 50 μm (e) or 100 µm (a)

At the EP stage, we found that there was a dramatic reduction of the pan endocrine marker, CHGA (Fig. 3e, f), with no significant change in NKX6.1 and NKX2.2 expression (Fig. 3e). RT-qPCR analysis showed significant reduction in the expression of NEUROD1, while the expression of NEUROG3, NKX2.2, and PAX4 were significantly increased in EPs lacking RFX6 compared with WT controls (Fig. 3g).

Deletion of RFX6 leads to large-scale transcriptomic alterations associated with pancreatic endocrine specification in PPs and EPs

For comprehensive understanding of the transcriptomic changes between RFX6 KO and WT cells, RNA-seq was performed on PPs and EPs. Our transcriptome analysis on iPSC-derived PPs detected 392 differentially expressed genes (DEGs) significantly affected by RFX6 deletion. Among these DEGs, 223 genes were significantly downregulated (log2 fold change < −1.0, p<0.05), while 169 genes were significantly upregulated (log2 fold change >1.0, p<0.05) in RFX6 KO PPs compared with WT PPs (Fig. 4a and ESM Fig. 5a). At the EP stage, we identified 325 DEGs significantly impacted by the deletion of RFX6, with 215 of these genes being significantly downregulated (log2 fold change < −1.0, p<0.05) and 110 genes being significantly upregulated (log2 fold change >1.0, p<0.05) in RFX6 KO EPs compared with WT EPs (Fig. 4a and ESM Fig. 5b). Interestingly, 160 of the downregulated DEGs, comprising 57.3%, were found in both PPs and EPs (Fig. 4b), with most of these genes known to be associated with pancreatic endocrine development. The Gene Ontology (GO) of the downregulated DEGs in PPs and EPs displayed enriched genes linked to pancreatic endocrine development, INS secretion regulation, regulation of ion transmembrane transport and negative regulation of cell apoptosis (Fig. 4c and ESM Fig. 5c, d), whereas the upregulated DEGs showed GO enrichment in genes linked to lipid metabolism and nervous system development (data not shown). At the PP stage, the RT-qPCR validation analysis confirmed a significant decrease in the expression of endocrine genes including ARX, PAX6, CHGA, IRX1, IRX2, INS, GCG, SST, MAF1B, ERO1B (ERO1LB), NEUROD1, PCSK1, CRYBA2, SCGN, PTPRN, PRPRN2, FEV and LMX1B in RFX6 KO PPs compared with WT PPs (Fig. 4d, Table 1). Furthermore, at the EP stage, the RT-qPCR revealed significant decrease in the expression of endocrine genes including ARX, PAX6, ISL1, IRX2, INS, GCG, SST, NEUROD1, PCSK1, SCGN, ERO1B, MAFB, SIX3, KCTD12 and LMX1B in RFX6 KO EPs compared with WT EPs (Fig. 4e; Table 1).

Fig. 4

Impact of RFX6 loss on transcriptomic profiles of iPSC-derived PPs and EPs. Bulk RNA-seq analysis was performed on PPs (n=3) and EPs (n=2) derived from RFX6 KO iPSCs and WT iPSCs. (a) Volcano plots display the DEGs in RFX6 KO PPs and RFX6 KO EPs compared with their WT controls. Downregulated genes are represented by blue dots, while upregulated genes are depicted by red dots. (b) Venn diagram illustrating the intersection of downregulated DEGs in RFX6 KO PPs and RFX6 KO EPs. Note that most of those DEGs are endocrine pancreatic genes. (c) Heatmap of z score value of pancreatic endocrine and INS secretion genes downregulated in RFX6 KO PPs and RFX6 KO EPs compared with WT PPs and WT EPs, respectively. (d, e) RT-qPCR validation of the DEGs in PPs (d) and EPs (e) derived from two different KO iPSC lines (n=4). The data are presented as mean ± SD. **p<0.01, ***p<0.001

Table 1 Key downregulated DEGs associated with pancreatic endocrine development and function in PPs and EPs lacking RFX6 (log2 fold change < –1, p<0.05) RFX6 loss correlates with the generation of smaller pancreatic islet organoids

To enhance islet differentiation after stage 4, cells were cultured in suspension to form organoids, with an equal number of WT PP and KO PP cells used. Although during the first 2 days of stage 5 no notable difference between WT and KO organoids was observed, a significant variation in organoid size became evident as differentiation progressed. Islet organoids derived from RFX6 KO iPSCs showed smaller size and irregular shapes compared with those derived from WT iPSCs during stages 5 and 6 (Fig. 5a and ESM Fig. 6a).

Fig. 5

Influence of RFX6 deletion on pancreatic islet organoid formation and cell viability. (a) Comparative morphological analysis of pancreatic islet organoids derived from two RFX6 KO iPSC lines vs WT iPSCs during differentiation stages 5 and 6 (n=3); S, stage, D, day. (b) Representative flow cytometry analysis and quantification of apoptosis (Annexin V+ cells) on day 3 of stage 5 of differentiation indicates a significant increase in apoptosis in RFX6 KO EPs in comparison with WT EPs (n=3). (c) Flow cytometry analysis of BrdU incorporation reveals a slight increase in cell proliferation (BrdU+ cells) in EPs derived from RFX6 KO iPSC lines compared with those derived from WT iPSCs. (d) Log2 fold change in the expression of CAT mRNA in RFX6 KO PPs and RFX6 KO EPs compared with WT controls, based on RNA-seq data analysis. (e) Western blot analysis showing the absence of CAT protein in RFX6 KO PPs and RFX6 KO EPs compared with WT controls. (f) Immunofluorescence images showing the lack of CAT expression in RFX6 KO EPs compared with WT EPs. The data are presented as mean ± SD. *p<0.05; (d) PPs p=9.62 × 10−35; EPs p=1.87 × 10−26. Scale bar, 50 µm (f) or 100 µm (a)

To investigate whether the dramatic reduction in islet organoid size could be attributed to either cell death or inhibition of cell proliferation, we conducted apoptosis and proliferation assays during stage 5 of differentiation. Flow cytometry analysis demonstrated a significant increase in the proportion of Annexin V+ cells in RFX6 KO EPs compared with WT EPs (Fig. 5b). Although increased apoptosis was also observed in the final stage of differentiation (stage 6), its level was lower compared with EPs (ESM Fig. 6b), indicating increased cell death with its peak during the EP stage. Quantification of BrdU incorporation revealed no significant difference in proliferation rates between WT and KO cells (Fig. 5c), suggesting that reduced islet organoid size due to RFX6 loss mainly results from increased cell death.

To elucidate the mechanism underlying the increased cell death, we analysed the top DEGs identified from our RNA-seq data. Interestingly, we observed a significant downregulation of CAT (encoding for catalase [CAT], an antioxidant enzyme that is known to protect cells against oxidative stress [31]) in both RFX6 KO PPs (log2 fold change = −7.459; p=9.62 × 10−35) and RFX6 KO EPs (log2 fold change = −6.978; p=1.87 × 10−26) compared with WT controls (Fig. 5d). This finding was validated at the protein level through western blot and immunostaining analyses, revealing an almost complete absence of CAT expression in both RFX6 KO PPs and RFX6 KO EPs compared with their respective controls (Fig. 5e, f). To validate the role of the CAT in promoting cell survival, we employed the STRING tool for protein functional interaction prediction [27]. Our analysis revealed CAT’s strong interaction with antioxidative stress proteins, such as superoxide dismutase proteins (ESM Fig. 6c).

RFX6 loss hinders the development of pancreatic islet cells

Subsequent differentiation into pancreatic islets demonstrated a lack of expression for INS, proinsulin, GCG, SST and urocortin 3 (UCN3), alongside a significant decrease in CHGA in RFX6 KO islets compared with WT islets (Fig. 6a, b). This indicates that RFX6 is essential for the formation of alpha, beta and delta cells. These reductions were confirmed at the mRNA level for INS, GCG, SST and UCN3 (Fig. 6c). Furthermore, other key pancreatic islet markers, including IAPP, PAX6, ARX, GCK, MAFA, KCNJ11, ABCC8, SLC18A1 and FEV, were significantly downregulated (Fig. 6c). On the other hand, pancreatic polypeptide Y (PPY) was significantly upregulated at mRNA and protein levels (Fig. 6c and ESM Fig. 6d). In response to various glucose concentrations, RFX6 KO islets exhibited no significant changes in INS secretion, with their total INS content being significantly lower than that in WT controls (ESM Fig. 6e).

Fig. 6

RFX6 loss impairs the development of pancreatic islet cells. (a) Confocal immunofluorescence showing expression of pancreatic islet markers INS, proinsulin (PROINS), GCG, UCN3 and CHGA in islets derived from two different RFX6 KO iPSC lines compared with WT controls (n=3). (b) Flow cytometry analysis of the expression of INS, GCG and SST in islets derived from RFX6 KO iPSCs compared with expression in islets derived from WT iPSCs (n=3). (c) RT-qPCR analysis for the mRNA expression of key islet genes INS, GCG, SST, UCN3, IAPP, PAX6, ARX, GCK, MAFA, KCNJ11, ABCC8, SCL18A1, FEV and PPY (n=4). Data are represented as mean ± SD; **p<0.01, ***p<0.001. Scale bar, 50 µm

RFX6 overexpression rescues the expression of dysregulated genes in pancreatic cells lacking RFX6

Next, we aimed to reverse RFX6-associated defects by ectopically expressing RFX6 (RFX6 overexpression [OE]). RFX6 was overexpressed on days 2 and 4 of stage 4 for assessing its effect on PPs and EPs, and subsequently on islets (Fig. 7). At the end of stage 4, the RFX6 OE significantly increased mRNA expression levels of pancreatic endocrine genes that were downregulated in RFX6 KO PPs, including RFX6, ARX, PAX6, CHGA, IRX1, IRX2, INS, GCG, SST, MAFB, ERO1B, NEUROD1, PCSK1, ISL1, CRYBA2, SCGN, PTPRN, PTPRN2 and LMX1B (Fig. 7a). Furthermore, we assessed the impact of RFX6 OE on the dysregulated DEGs on day 3 of stage 5, 72 h post-transfection. Our results revealed a substantial increase in the expression levels of INS, GCG, SST, NEUROD1, CHGA, CHGB, PAX6, ARX, ISL1, MAFB, PCSK1, ERO1B, IRX2, CRYBA2, KCTD12, LMX1B, SCGN and SSTR2 following RFX6 OE (Fig. 7b). Moreover, it induced a significant decrease in the PAX4 mRNA levels, which had been upregulated in RFX6 KO EPs (Fig. 7b). In addition, RFX6 OE at the end of stage 4 increased the expression of INS, GCG and SST in the RFX6 KO islets (Fig. 7c).

Fig. 7

RFX6 overexpression rescues the expression of dysregulated genes in pancreatic cells lacking RFX6. (a) RT-qPCR analysis for the expression of pancreatic endocrine genes: RFX6, ARX, PAX6, CHGA, IRX1, IRX2, INS, GCG, SST, MAFB, ERO1B, NEUROD1, PCSK1, ISL1, CRYBA2, SCGN, PTPRN, PTPRN2 and LMX1B in PPs derived from RFX6 KO iPSCs and WT iPSCs, 48 h following ectopic expression of RFX6 (n=4). (b) RT-qPCR analysis for the expression of pancreatic endocrine genes: INS, GCG, SST, NEUROD1, CHGA, CHGB, PAX6, ARX, ISL1, MAFB, PCSK1, ERO1B, IRX2, CRYBA2, KCTD12, LMX1B, SCGN, SSTR2 and PAX4, in EPs derived from RFX6 KO iPSCs and WT iPSCs, 72 h following ectopic expression of RFX6 (n=4). (c) RT-qPCR analysis for the expression of INS, GCG and SST in islet organoids at stage 6 following overexpression of RFX6 at the end of stage 4 (n=4). Data are represented as mean ± SD; *p<0.05, **p<0.01, ***p<0.001