Abstract Human pluripotent stem cells (hPSCs) have the potential to differentiate into various cell types, including pancreatic insulin-producing β cells, which are crucial for developing therapies for diabetes. However, current methods for directing hPSC differentiation towards pancreatic β-like cells are often inefficient and produce cells that do not fully resemble the native counterparts. Here, we report that highly selective tankyrase inhibitors, such as WIKI4, significantly enhances pancreatic differentiation from hPSCs. Our results show that WIKI4 promotes the formation of pancreatic progenitors that give rise to islet-like cells with improved β-like cell frequencies and glucose responsiveness compared to our standard cultures. These findings not only advance our understanding of pancreatic development, but also provide a promising new tool for generating pancreatic cells for research and potential therapeutic applications. Subject terms: Stem-cell differentiation, Organogenesis, Type 1 diabetes __________________________________________________________________ “Here the authors demonstrate that ß cell differentiation from human pluripotent stem cells can be improved by replacing nicotinamide with tankyrase inhibitors. This results in pancreatic progenitors that form islet-like cells with increased ß cell frequencies and glucose responsiveness.” Introduction Pancreatic β cells are essential to regulating glucose homeostasis, and their destruction in type 1 diabetes (T1D) leaves affected individuals relying solely on frequent exogenous insulin injections^[38]1. While insulin therapy is lifesaving, current technology still cannot recapitulate the precise control of insulin secretion mediated by endogenous β cells, leading to long-term complications such as nephropathy, neuropathy, retinopathy, and cardiovascular disease^[39]2,[40]3. Pancreas and islet transplantation, in specific cases, may provide insulin independence, but the need for immunosuppression and donor shortages are major obstacles that severely limit their widespread clinical use^[41]4. Therefore, the generation of islet-like cells from human pluripotent stem cells (hPSCs) holds great promise for the treatment of T1D^[42]5. Current differentiation strategies recapitulate pancreas development using a series of growth factors and small molecules to modulate signaling pathways to guide cell fate commitment to pancreatic progenitors (PPs) and islet-like clusters that contain β-like cells at the final stages of differentiation^[43]6–[44]11. Both PPs and islet-like cells are currently being tested in clinical trials ([45]NCT03163511 and [46]NCT04786262). However, the reports from clinical trials involving the therapeutic use of PPs show that transplantation induces robust commitment to the α cell lineage, the generation of low β cell mass, and insufficient endogenous C-peptide (C-PEP) secretion for achieving insulin independence^[47]12–[48]14. These findings indicate that a more effective way to generate and support the survival of β cells in vivo is urgently needed. Despite recent advances, it remains to be elucidated whether the endocrine composition output is due to intrinsic or extrinsic cues, or both, and if intrinsic factors are at play, one should be able to generate PPs that can be better programmed to commit to the β cell lineage both in vitro and in vivo. Therefore, in this study, we investigated whether improving early patterning of hPSC-derived pancreatic progenitors could enhance β-like cell differentiation. Previously, our group developed a method for the efficient commitment of PDX1^+ endoderm into NKX6-1-expressing PPs using the compound nicotinamide (NA) in combination with EGF and Noggin^[49]15. However, the exact role of NA during PP commitment remains elusive as it targets several proteins containing NA-binding sites, particularly class III histone deacetylases (HDACs) and poly (ADP-ribose) polymerases (PARPs)^[50]16–[51]20. Using a targeted chemical screen, we show that NA induces PP differentiation by inhibiting a specific class of PARP proteins known as tankyrases (TNKS)^[52]18. TNKS1 and TNKS2 have a unique domain organization separating them from the other members of the PARP superfamily and have been shown to modulate a wide variety of cellular activities and signaling pathways, including WNT signaling, insulin-stimulated glucose uptake, telomere length, and mitotic checkpoint^[53]21. Many of these processes are implicated in pathological conditions, including cancer, which prompted the development of numerous small molecule TNKS inhibitors (TNKSi), some of them currently in clinical trials^[54]22. This study has established that PPs differentiated in the presence of TNKSi express higher levels of integrin molecules, are less proliferative, and generate islet-like populations with improved glucose responsiveness and higher β-like cell frequencies compared to those generated in the presence of NA. Upon transplantation, the WIKI4-derived β-like cells provided glycemic control, whereas NA-derived β-like cells failed to do so. Thus, the use of TNKSi during PP differentiation improves β-like cell commitment and function. Results Tankyrase inhibition promotes pancreatic lineage specification To tease out downstream effectors of NA responsible for NKX6-1 induction at stage 4, we replaced NA with HDAC- or PARP-specific inhibitors during PP specification. Cells were differentiated to stage 3 (PDX1^+ endoderm) according to our published protocol and differentiated to stage 4 (PPs) using a combination of Noggin, EGF and either NA, HDAC inhibitors (Cl-994, Sodium Butyrate, SAHA, sirtinol) or PARP inhibitors (DPQ, Olaparib, PJ34, TIQ-A, 3-Aminobenzamide) (Fig. [55]1a). To monitor PP commitment, we took advantage of a human embryonic stem cell (hESC) reporter line that expresses the green fluorescent protein (GFP) under the native NKX6-1 promoter (NKX6-1^GFP/w)^[56]15. The percentage of GFP^+ cells at the end of stage 4 was measured with flow cytometry as a proxy for the NKX6-1^+ population. We found that the HDAC inhibitors and most of the PARP inhibitor treatments did not induce NKX6-1:GFP^+ cells (Fig. [57]S1a and [58]S1b). However, the PARP inhibitors TIQ-A and 3-aminobenzamide significantly increased the frequency of the NKX6-1:GFP^+ population compared to control (Fig. [59]S1b). Unlike the other PARP inhibitors, both TIQ-A and 3-aminobenzamide also inhibit TNKS^[60]23,[61]24, suggesting NKX6-1 expression may be induced in PDX1^+ endoderm via TNKS inhibition. Mechanistically, TNKS inhibition by small molecules can be achieved by targeting the catalytic domain either at the NA-subsite (NA-TNKSi) or at the adenosine (Ade)-subsite (Ade-TNKSi)^[62]18,[63]19. However, NA-TNKSi often has off-target effects as the NA-subsite of TNKS is highly conserved among the PARP family, whereas the Ade-subsite is uniquely present in TNKS and offers the potential for TNKS-specific inhibition^[64]18,[65]19. To examine whether pancreatic commitment is attributable to TNKS-specific inhibition, we treated PDX1^+ cells at the end of stage 3 with increasing concentrations of TNKSi with binding affinity for the NA-subsite (XAV939, MN64) and Ade-subsite (IWR-1, WIKI4) and monitored the formation of NKX6-1: GFP^+ cells at the end of stage 4. All TNKSi treatments significantly increased the GFP^+ population (Fig. [66]S1c), confirming that it is through TNKS inhibition that NA induces NKX6-1 expression. To determine whether this effect could be reproduced in other cell lines, we repeated the experiments using H1, a commonly used hESC line. To monitor the formation of PPs, we performed expression kinetic analysis of PDX1 and NKX6-1 by flow cytometry throughout stage 4 (from day 8 to day 13) following treatment with TNKSi targeting the NA- or Ade-subsites. All TNKSi treatments effectively generated PDX1^+/NKX6-1^+ cells in the H1 cell line (Fig. [67]1b), demonstrating that the TNKSi effect is not cell line-specific. Similar results were obtained using a dual-site TNKSi, NVP-TNKS656, that is capable of binding to both NA- and Ade-subsites^[68]25 (Fig. [69]S2). We next compared the kinetics of NKX6-1 induction and found that the percentages of PDX1^+/NKX6-1^+ population peaked on days 12 and 13 with both NA and other TNKSi (Fig. [70]1b). Interestingly, the frequencies of PPs generated using different TNKSi were very similar at the end of stage 4, suggesting that TNKS inhibition, by targeting either binding site, effectively promotes NKX6-1 expression (Fig. [71]1c–f). Fig. 1. TNKS inhibitors promote pancreatic lineage commitment. [72]Fig. 1 [73]Open in a new tab a Schematic representation of the four stages of differentiation from human Pluripotent Stem Cells (hPSCs) to PDX1^+/NKX6-1^+ pancreatic progenitors (PPs). Nicotinamide (NA) was substituted at stage 4 by either Histone Deacetylase (HDAC), Poly-ADP Ribose Polymerase (PARP) or Tankyrase (TNKS) inhibitors. Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license ([74]https://creativecommons.org/licenses/by-nc-nd/4.0/deed.en). b Percentage of PDX1^+/NKX6-1^+ cells measured by flow cytometry between day 8 and day 13 of differentiation after treatment with Noggin and EGF alone (-) or in combination with NA or the TNKSi XAV939, IWR-1, MN64 or WIKI4 (n = 3 independent biological replicates. P-values are <0.0001 for NA, XAV939, IWR-1, MN64, WIKI4 when comparing day 12/13 to day 8. For control, P-value is 0.002 at day 12, and <0.0001 at day 13 when compared to day 8. Two-way ANOVA with Tukey’s multiple-comparison test between day 8 and day 12/13 of each condition. Error bar represents ± SEM). c, d Representative flow cytometry plots of NKX6-1 and PDX1 expression at day 13 and quantification of PDX1^+/NKX6-1^+ cells (n = 5 independent biological replicates, One-way ANOVA with Dunnett’s multiple-comparison test. All significant conditions produced P-values < 0.0001. Error bars represent ± SEM.). e, f Gene expression analysis of PDX1 and NKX6-1 on day 13. Data normalized to the human housekeeping gene TBP (hTBP) (n = 3 independent biological replicates. Exact P-values are reported in the figure. One-way ANOVA with Dunnett’s multiple-comparison test. Error bars represent ± SEM). When compared to other highly selective Ade-TNKSi (JW74, JW55, and G007-LK), WIKI4 performed as effectively at inducing NKX6-1 expression (Fig. [75]S3a–c). When used at the optimal concentrations (10 µM JW74, 9 µM WIKI4, 5 µM JW55, and 5 µM G007-LK), Ade-TNKSi compounds generated similar percentages of NKX6-1^+/PDX1^+ PPs when compared to NA (Fig. [76]S3a, d). In contrast, treatment with DMSO or the PARP1/2 specific inhibitor MK4827 failed to generate high percentages of NKX6-1^+/PDX1^+ PPs (Fig. [77]S3a, d). These findings support the hypothesis that inhibition of PARP5/TNKS, and not PARP1/2, is responsible for PP commitment. To validate the chemical screen, we knocked down TNKS1 and TNKS2 using shRNA lentiviral transduction. Day 8 PDX1^+ cells were dissociated and replated to allow optimal transduction prior to stage 4 differentiation in the absence of NA or WIKI4. Interestingly, our results indicate that TNKS2, but not TNKS1, knockdown is associated with NKX6-1 induction (Fig. [78]S4a–f). Adenosine-subsite targeting TNKS inhibitors promote β-like cell commitment To evaluate whether the Ade-TNKSi-treated PPs give rise to β-like cells, we further differentiated the PPs to stage 6, yielding islet-like cells containing β-like cells that are marked by co-expression of NKX6-1 and C-peptide (C-PEP) (Fig. [79]2a). We next assessed whether specific TNKS inhibition by Ade-TNKSi at stage 4 can improve β-like cell frequency at stage 6, and we included XAV939, a NA-TNKSi at stage 4 for comparison. Flow cytometry analysis showed that PPs derived from NA-TNKSi (XAV939) and Ade-TNKSi (JW74, WIKI4, JW55, G007-LK) gave rise to significantly higher percentages of C-PEP^+/NKX6-1^+ β-like cells compared to control (DMSO-treated cells). Amongst the Ade-TNKSi-treated cells, WIKI4-derived PPs gave rise to significantly higher percentages of β-like cells compared to NA (Fig. [80]2b, c). A comparison of NA- versus WIKI4-treated cells demonstrated that WIKI4 treatment during stage 4 resulted in significantly higher numbers of C-PEP^+/NKX6-1^+ β-like cells at the end of stage 6, without altering the frequency and absolute numbers of polyhormonal cells that express glucagon (GCG) as well as C-PEP (Fig. [81]S5a–f). These data indicate that although no differences were observed between NA-, XAV939- and WIKI4-derived PPs at stage 4 (Fig. [82]1d), WIKI4 treatment primed PPs to in vitro β-like cell commitment more robustly than NA (Fig. [83]2c). Fig. 2. Highly selective TNKS inhibitors promote β-like cell commitment. [84]Fig. 2 [85]Open in a new tab a Schematic representation of the six stages of differentiation from hPSCs to C-Peptide (C-PEP)^+/NKX6-1^+ β-like cells. NA was substituted at stage 4 by either DMSO or different small molecules targeting either the nicotinamide (XAV939) or the Adenosine subsite (JW74, WIKI4, JW55, G007-LK) of TNKS. Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license ([86]https://creativecommons.org/licenses/by-nc-nd/4.0/deed.en). b, c Representative flow cytometry plots and quantification of the percentage of C-PEP^+/NKX6-1^+ β-like cells at stage 6 obtained after the different treatment groups at stage 4 (n = 5 for DMSO, n = 3 for XAV939 and JW74, n = 8 for NA, n = 6 for WIKI4 and JW55, n = 7 for G007-LK. All replicates represent independent biological replicates. Exact P-values are reported in the figure. One-way ANOVA with Dunnett’s multiple-comparison test. Error bars represent ± SEM.). To verify that these results were cell line-independent, we used a human-induced PSC line (LiPSC-GR1.1)^[87]26 to validate the effect of WIKI4 treatment at stage 4 on PP and beta-like cell commitment. Our results confirmed that while NA and WIKI4 treatments generated similar percentages of NKX6-1^+/PDX1^+ PPs at stage 4 (Fig. [88]S6a, b), WIKI4-derived PPs gave rise to significantly higher percentages of C-PEP^+/NKX6-1^+ β-like cells compared to the NA counterpart at stage 6 (Fig. [89]S6c, d). To demonstrate that the effect of WIKI4 on β-like cell commitment is protocol-independent, we generated islet-like cells using a recent differentiation method from Balboa et al.^[90]10 and substituted NA with WIKI4 at stage 4 of their protocol (Fig. [91]S7a). While WIKI4 treatment reduced PDX1^+/NKX6-1^+ cells at Stage 4, further differentiation of the WIKI4-derived population led to higher percentages of C-PEP^+/NKX6-1^+ β-like cells at Stage 7 compared to the NA treatment (Fig. [92]S7b–e). Interestingly, the WIKI4-derived Stage 7 population contained significantly fewer SLC18A1^+ enterochromaffin cells compared to the NA counterpart (Fig. [93]S7f, g). In summary, we showed that WIKI4, an Ade-TNKSi, predisposes PP cells towards β-like cell commitment irrespective of cell lines and differentiation protocols. RNA sequencing of NA- and WIKI4-treated stage 4 populations To explore how WIKI4 treatment may promote β-like cell differentiation, we harvested NA- and WIKI4-derived stage 4 cells and performed bulk RNA sequencing (RNAseq) on NA- and WIKI4-treated PPs. RNAseq identified the expression of 28,395 genes in total and 3428 differentially expressed genes between NA- and WIKI4-derived stage 4 cells (FDR < 0.01; P < 0.01) (Fig. [94]3a). This included 1263 genes that were significantly upregulated by NA-derived PPs (cluster 1) and 2,165 genes significantly upregulated by WIKI4 (cluster 2) (Fig. [95]3b). Interestingly, compared to WIKI4-derived PPs, NA-derived PPs expressed higher levels of genes that are non-pancreas specific, such as TBX3, SOX3, and CDX2^[96]27–[97]29 (Fig. [98]3a). Fig. 3. WIKI4 treatment reduces cellular proliferation and activates integrin-actin signaling. [99]Fig. 3 [100]Open in a new tab a Global distribution of genes by log-transformed False Discovery Rate (FDR) and log-transformed fold change in expression between NA- and WIKI4-derived stage 4 cells. b Heatmap of scaled and log-transformed expression data for genes differentially expressed between NA- and WIKI4-derived stage 4 populations. Two clusters were identified—genes upregulated by NA (cluster 1), and genes upregulated by WIKI4 (cluster 2). (c Gene expression analysis of proliferation markers (MKI67, TOP2A) and cell cycle regulators (CDKN1A, CDKN2B) by RT-qPCR on day 13. Data normalized to housekeeping gene hTBP (n = 4 independent biological replicates. Exact P-values are reported in the figure. Two-tailed paired student’s t-test. Error bars represent SEM). d Cell count of NA- and WIKI4-derived stage 4 populations (n = 6 independent biological replicates. The exact P-value is reported in the figure. Two-tailed paired student’s t-test. Error bars represent SEM). e, f Representative flow cytometry plots and quantification of NKX6-1/Ki67 expression profile (n = 4 independent biological replicates. Exact P-values are reported in the figure. Two-tailed paired student’s t-test. Error bars represent SEM). g Gene set enrichment analysis (GSEA) of gene sets associated with the integrin signaling pathway, comparing NA- and WIKI4-derived stage 4 populations (Data analyzed by Kolmogorov–Smirnov test. No correction was performed). h Gene expression analysis of integrin subunits (ITGB1, ITGA3, ITGA5, ITGAV) by RT-qPCR in NA- and WIKI4-derived cells at stage 4. Data normalized to housekeeping gene hTBP (n = 5 independent biological replicates. Exact P-values are reported in the figure. Two-tailed paired student’s t-test. Error bars are SEM). i Immunofluorescence staining of NA- and WIKI4-derived cells at stage 4 with ITGB1 (red) and ACTA2 (green). DAPI (gray) represents nuclei. The scale bar represents 125 µm. Pathway enrichment analysis of cluster 1 genes revealed upregulation of multiple gene sets involved in cellular proliferation, including “DNA metabolic process”, “Cell cycle”, and “Mitotic cell cycle transition” (Fig. [101]S8a). To validate these findings, we analyzed the expression of cell cycle markers in NA- and WIKI4-derived cells at stage 4. We confirmed that the expression of proliferation markers MKI67 and TOP2A was significantly higher in NA- compared to WIKI4-derived cells, while the expression of cell cycle inhibitors CDKN1A and CDKN2B was significantly lower in NA- compared to WIKI4-derived populations (Fig. [102]3c), suggesting increased proliferation in NA-treated cells compared to the WIKI4 counterpart. Consistently, we detected a 2-fold increase in total cell number in NA- compared to WIKI4-treated cells at the end of stage 4 (Fig. [103]3d). Flow cytometric analysis of NKX6-1 and Ki67 showed that approximately 20% of NA and WIKI4-derived NKX6-1^+ progenitors co-expressed the proliferative marker Ki67, but WIKI4-derived cells contained significantly lower frequencies of NKX6-1^-/Ki67^+ cells compared to the NA counterpart (Fig. [104]3e, f). Thus, WIKI4 treatment appears to reduce the proliferative potential of NKX6-1-negative cells relative to NA. Analysis of cluster 2 genes identified enrichment of gene sets associated with the crosstalk between the actin cytoskeleton and the extracellular matrix, including “Actin filament-based processes”, “Positive regulation of locomotion”, “Regulation of cell adhesion”, “Regulation of cell morphogenesis”, “Extracellular matrix organization”, and “Focal adhesion” (Fig. [105]S8b). Because integrins are critical mediators of such processes^[106]30, and have been associated with pancreatic endocrine commitment^[107]31–[108]33, we decided to interrogate integrin and actin signaling activity in our dataset. Indeed, gene set enrichment analysis (GSEA) demonstrated upregulation of integrin signaling (CMP_6880) and actin cytoskeleton (GO:0015629) pathways in WIKI4- compared to NA-derived populations (Fig. [109]3a, b, g and Fig. [110]S8c). The expression of common human islet integrin subunits β[1] (ITGB1), α[3] (ITGA3), α[5] (ITGA5), and α[v] (ITGAV)^[111]34, as well as the two major actin subunits ACTA1 and ACTA2, was significantly higher in WIKI4- compared to NA-derived cells (Fig. [112]3h and Fig. [113]S8d). The increased expression of integrin and actin genes was associated with gross morphological changes in the WIKI4-treated cells, which formed three-dimensional structures and peeled away from the edge of the tissue culture dish, while the NA-treated cells covered the dish as a uniform two-dimensional epithelium. Immunofluorescence staining of ITGB1 and ACTA2 showed increased cellular spreading in WIKI4-treated cells compared to NA-derived PPs (Fig. [114]3i), which indicates activated integrin-actin signaling^[115]35,[116]36. To further investigate this phenomenon, we recorded a timelapse movie of DMSO-, NA- and WIKI4-derived cultures between day 10 and day 12 of differentiation (Supplementary movies [117]1–[118]3). Remarkably, while DMSO- and NA-treated cultures maintained a stable monolayer, WIKI4-derived cultures demonstrated monolayer folding at the edges and retracting from the plastic, followed by cellular extensions (Fig. [119]S9 and Supplementary movies [120]1–[121]3). To better understand the potential mechanisms underlying the effects of WIKI4 on cell motility, we examined the expression of TNKS targets by western blot. We first analyzed AXIN1, a rate-limiting component of the β-catenin destruction complex^[122]37, to evaluate whether WIKI4 could more efficiently prevent AXIN1 degradation compared to NA. Surprisingly, while both NA and WIKI-treated cells expressed higher levels of AXIN1 compared to the control (DMSO), we did not observe any significant difference in AXIN1 levels between NA- and WIKI4-treated cultures (Fig. [123]S10a, b). We then searched the literature for TNKS targets that could control the actin cytoskeleton and identified angiomotin (AMOT), a YAP1 inhibitor as well as a modulator of the integrin and cytoskeleton activity^[124]38–[125]40. Indeed, western blot analysis showed increased levels of the two AMOT isoforms (p80 and p130) in WIKI4-derived PP compared to NA and DMSO control (Fig. [126]S10c, d). This demonstrates that WIKI4 treatment prevents AMOT degradation more effectively than NA. We speculate that retention of AMOT could result in increased integrin activation during PP specification, leading to morphological changes in the monolayer culture that, in turn, may play a role in endocrine lineage allocation. Given the association between the actin cytoskeleton and the WNT and Hippo signaling pathways, we used the bulk RNAseq dataset to examine whether WIKI4 treatment inhibited these pathways more efficiently than NA. By directly comparing the expression of WNT and Hippo target genes^[127]41–[128]43 in the NA- or WIKI4-derived PPs, we observed that 42/124 WNT target genes were differentially expressed between the two groups, but many of them were expressed at low levels (Fig. [129]S10e, g). Surprisingly, the majority of WNT targets were expressed at lower levels in the NA-treated cells (Fig. [130]S10e, g). Similarly, when we compared known Hippo target genes, we observed that 25/72 targets were differentially expressed and formed two distinct clusters based on higher or lower expressions in response to NA or WIKI4 treatments (Fig. [131]S10f, h). Amongst the highest differentially expressed genes, WIKI4-treated cells expressed higher levels of CXCL8, TNFRSF9, CD44, MMP7, and AREG compared to NA-derived PPs. Interestingly, these genes are expressed by fetal pancreatic cells cultured in vitro, as well as in the earliest stages of pancreatic tumorigenesis, epithelial to mesenchymal transition, and metastasis^[132]44–[133]49. While there may be a link between these processes, additional studies are warranted to determine whether these transcriptional changes could explain the increased motility observed in the WIKI4-treated cells compared to NA-derived PPs. WIKI4-derived β-like cells are glucose-responsive in vitro To evaluate the functional characteristics of the WIKI4-derived β-like cells, we differentiated the NA- and WIKI4-derived PPs to stage 6 and performed cell cluster resizing on day 20 of differentiation, followed by extended culture from day 23-33 to promote glucose responsiveness^[134]50,[135]51 (Fig. [136]4a). Immunofluorescence stainings of clusters on day 23 and day 33 showed more compacted structures after reaggregation and extended culture. The clusters were characterized by β-like cells (NKX6-1^+/PDX1^+/C-PEP^+) as well as polyhormonal cells (GCG^+/INS^+) and δ-like cells (SST^+/GCG^-/INS^-) (Fig. [137]4b), confirming the presence of the main pancreatic endocrine cell types from day 23 onwards in both NA- and WIKI4-derived aggregates. To compare the frequency of β-like cells in NA and WIKI4 conditions, flow cytometric analysis for NKX6-1 and C-PEP was performed on day 23 and day 33 of differentiation. Consistent with our initial data (Fig. [138]2b), we show that WIKI4-derived islet-like cells contained higher percentages of C-PEP^+/NKX6-1^+ β-like cells compared to the NA counterpart, independent of reaggregation and extended culturing (Fig. [139]4d, f). Reaggregation also led to a significant reduction in the frequencies of C-PEP^-/NKX6-1^- cells in both NA- and WIKI4-derived β-like cells, producing a final population with fewer non-β-like cells (Fig. [140]4e). Fig. 4. WIKI4-derived PPs give rise to glucose-responsive islet-like populations containing increased β-like cells compared to NA-derived PPs. [141]Fig. 4 [142]Open in a new tab a Schematic representation of pancreatic differentiation highlighting reaggregation at day 20 and extended culture to day 33. PDX1^+ cells were treated with either NA or WIKI4 at stage 4. On day 20, cells were dissociated into single cells and allowed to reaggregate in stage 6 media. On day 23, cells were transferred to a growth factor-free media until day 33. Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license ([143]https://creativecommons.org/licenses/by-nc-nd/4.0/deed.en). b Immunofluorescence staining of NA- and WIKI4-derived cells in day 23 non-reaggregated, day 23 reaggregated, and day 33 reaggregated conditions. Sections were stained for β cell markers (NKX6-1/PDX1/C-PEP) and pancreatic hormones (GCG/INS/SST). DAPI represents nuclei. Scale bar represents 50 µm. The white box represents 500% digital magnification of specific areas. c Representative flow cytometry plots of C-PEP/NKX6-1 profile of non-reaggregated (day 23), reaggregated (day 23), and extended culture aggregates (day 33). d–f Quantification of C-PEP^+/NKX6-1^+ and C-PEP^-/NKX6-1^- populations from NA- or WIKI4-treated reaggregated and extended cultures, respectively (n = 4 for d, n = 5 for f, all replicates are independent biological replicates. Exact P-values are reported in the figure. Two-way ANOVA with Tukey’s multiple-comparison test. Error bar represents SEM). g Relative C-PEP Median Fluorescence Intensity (MFI) of NA- and WIKI4-treated β-like cells after extended culture (n = 5 independent biological replicates. Exact P-values are reported in the figure. Two-way ANOVA with Tukey’s multiple-comparison test. Error bar represents SEM). h Static Glucose Stimulated Insulin Secretion assay was performed to calculate the stimulation index after extended culture of NA- and WIKI4-derived aggregates (n = 5 independent biological replicates. Exact P-values are reported in the figure. One-way ANOVA with Dunnett’s multiple-comparison test. Error bar represents SEM). Although extended culture did not alter the percentage of β-like cells, it significantly increased the C-PEP median fluorescence intensity (MFI) in both NA- and WIKI4-derived β-like cells (Fig. [144]4g). This was associated with higher INS mRNA expression after extended culture in both NA- and WIKI4-derived cultures (Fig. [145]S11a). To determine if the increase in INS levels was associated with maturation, we assessed the expression of selected maturation markers and showed increased expression of ERRγ^[146]52 and SIX2^[147]53 from day 23 to day 33 in both conditions (Fig. [148]S11b, c), but detected no differences in the expression levels of MAFA, UCN3 and UCP2 in WIKI4-derived aggregates compared to NA-derived cultures or in relation to time in culture (Fig. [149]S11d–f). To assess functionality, we performed a static Glucose Stimulated Insulin Secretion (GSIS) assay on NA- and WIKI4-derived day 33 aggregates. Remarkably, WIKI4-, but not NA-derived, aggregates were glucose-responsive at day 33, and cell depolarization with KCl induced at least a 5-fold increase in insulin secretion in both conditions (Fig. [150]4h). Collectively, we showed that following reaggregation and extended culture, WIKI4-derived clusters are glucose-responsive. WIKI4-derived β-like cells restore normoglycemia in diabetic mice To compare functionality in vivo, NA- and WIKI4-derived populations were reaggregated on day 20 and then cultured until day 23. An average of 1.3 × 10^6 NA- or WIKI4-derived β-like cells were transplanted under the kidney capsule of streptozotocin (STZ)-treated immunocompromised diabetic mice (Supplementary Table [151]1). Given the differences in the frequencies of β-like cells generated following NA or WIKI4 treatment, this equated to an average of 4.1 × 10^6 NA-derived and 3.5 × 10^6 WIKI4-derived stage 6 cells being transplanted under the kidney capsule (Supplementary Table [152]1). Insulin pellets were placed subcutaneously at the time of transplantation to support mouse survival after transplantation (Fig. [153]5a). Diabetic control mice that only received the insulin pellets, without islet-like cells, became hyperglycemic at 13 weeks post-transplantation, indicating that the insulin pellets no longer controlled glycemia after that time. NA-transplanted mice reverted to hyperglycemia at 15 weeks post-transplantation, suggesting insufficient β-like cell function. In contrast, WIKI4-transplanted mice maintained normoglycemia throughout 16 weeks post-transplantation with no significant differences in glycemia compared to non-diabetic controls (Fig. [154]5b). Euglycemia was only achieved in 12% of NA-transplanted mice at the experimental endpoint, in contrast to 44% of WIKI4-transplanted mice, but no statistically significant difference was observed between the two groups (Fig. [155]5c). Fig. 5. Kidney capsule transplantation of WIKI4-derived β-like cells normalizes glycemia in STZ-treated diabetic mice. [156]Fig. 5 [157]Open in a new tab a Schematic of cell preparation and transplantation. Created with BioRender.com released under a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International license ([158]https://creativecommons.org/licenses/by-nc-nd/4.0/deed.en). b Weekly fasting blood glucose measurements (n = 8 for NA, from 7 independent differentiation cohorts; n = 9 for WIKI4, from 7 independent differentiation cohorts; n = 5 for non-diabetic control; n = 3 for STZ control. P-values listed: *STZ vs. non-diabetic—0.0077, 0.024 at week 13, 14; ^#NA vs. non-diabetic—0.0004, 0.0057 at week 15, 16; ^†NA vs WIKI4 – 0.0022, 0.0401 at week 15, 16. Two-way ANOVA with Tukey’s multiple-comparison test. Error bars represent SEM). c The percentage of NA- and WIKI4-transplanted mice that achieved euglycemia post-transplantation. Data was analyzed using chi-square test at every timepoint. d H&E staining of transplanted kidney grafts. The black box represents an area of magnification. Scale bar represents 500 µm. e Immunofluorescence staining of β-cell markers (C-PEP/NKX6-1/PDX1). White box areas were subjected to 500% digital magnification. Scale bar represents 100 µm. f IPGTT performed at 15 weeks post-transplantation (n = 4 for NA; n = 5 for WIKI4; n = 4 for non-diabetic control; n = 5 for STZ control. All data points represent independent biological replicates. Exact P-values are reported in the figure. Two-way ANOVA with Tukey’s multiple-comparison test. Error bars represent SEM). g Glucose stimulated C-PEP secretion assay at 15 weeks post-transplantation (n = 4 for NA; n = 6 for WIKI4. All data points represent independent biological replicates. Exact P-values are reported in the figure. One-tailed student’s t-test. Error bars represent SEM). To assess whether WIKI4-derived cells gave rise to a higher β cell mass in vivo compared to NA-derived cells, grafts were explanted 16 weeks post-transplantation and processed for histochemical and immunofluorescence analyses. Hematoxylin and eosin (H&E) staining of explanted mouse kidneys identified discernable grafts of similar sizes in both conditions with no gross differences (Fig. [159]5d and [160]S12a). Immunofluorescence staining for endocrine markers confirmed the presence of mono-hormonal cells expressing insulin (INS), glucagon (GCG), or somatostatin (SST), and no significant difference was detected in the frequency of NKX6-1^+/C-PEP^+ β-like cells within the NA- and WIKI4-derived grafts (Fig. [161]5e and [162]S12b, c). Cytokeratin-19 (CK19) and trypsin (TRYP)-expressing cells were also observed, indicating the presence of ductal and acinar cells, respectively (Fig. [163]S12d, e). Interestingly, we observed SLC18A1-expressing cells (Fig. [164]S12f), confirming the presence of previously identified enterochromaffin cells^[165]54. To assess graft functionality, we performed an intraperitoneal glucose tolerance test (IPGTT) at 15 weeks post-transplantation and demonstrated that WIKI4-transplanted mice effectively responded to glucose challenge and normalized glycemia in a similar pattern compared to non-diabetic controls (Fig. [166]5f). In contrast, NA-transplanted mice remained hyperglycemic after glucose injection (Fig. [167]5f). To further evaluate β-like cell function, we performed an in vivo Glucose Stimulated C-Peptide Secretion assay at week 16 post-transplantation. While we observed a negligible increase of human C-PEP levels in the sera of NA-transplanted mice after glucose injection, mice transplanted with WIKI4-derived β-like cells secreted significantly higher levels of human C-PEP in response to glucose challenge (Fig. [168]5g). Taken together, our data indicated that WIKI4-derived β-like cells outperform NA-derived β-like cells both in vitro and in vivo and contribute to long-term glycemic control in diabetic mice. Discussion Nicotinamide (NA) has been shown to partially protect diabetes progression in rodents, induce endocrine development in human fetal pancreatic explants, and enhance the differentiation of hESC-derived PP in vitro^[169]15,[170]55–[171]57. Despite these applications, the molecular mechanisms of NA during pancreatic commitment remain unclear. To identify NA downstream targets that induce NKX6-1 expression, we performed a targeted screening assay focusing on HDACs and PARPs, known NAD^+-dependent proteins^[172]58–[173]60. Here, we demonstrate that tankyrase inhibitors (TNKSi) can replace NA and efficiently generate PPs. Additionally, we show that treatment with highly selective TNKSi generates a PP population that gives rise to functional β-like cells at greater efficiency than NA-derived PPs. Inhibition of TNKS is an effective means to block canonical WNT signaling and we and others have previously shown that canonical WNT inhibition is necessary for pancreatic endoderm patterning, which is a key step for proper pancreas and β cell development^[174]61–[175]64. Besides PP specification, TNKSi offers clear advantages compared to NA treatment for the generation of β-like cells. We show that, compared to NA, WIKI4 treatment increases frequencies and absolute numbers of β-like cells, but not polyhormonal cells, and reduces the output of SLC18A1^+ enterochromaffin cells. This demonstrates that WIKI4 promotes β-like cell lineage commitment specifically rather than increasing general endocrine commitment. The positive effect of TNKS inhibition on β-like cell differentiation seems to extend to other differentiation stages, as studies by other groups showed that TNKS inhibition at either stage 5 (endocrine progenitors) or stage 7 (islet-like cells) of differentiation promoted β-like cell differentiation^[176]65,[177]66. One possible explanation for the differences observed in culture is that NA might be inhibiting HDAC and PARP1/2, and these off-target effects could negatively impact β-like cell differentiation. As both HDAC and PARP1 have been implicated in epigenetic modifications^[178]67,[179]68, it is tempting to speculate that unspecific NA-targeting of epigenetic modifiers may negatively affect pancreatic cell fate during development^[180]69,[181]70. Thus, minimizing off-target effects by using TNKS-specific compounds may help guide β-like cell fate in vitro. However, further experimentation is required to determine whether epigenetic differences in response to NA and WIKI4 treatments exist within the pancreatic progenitors and if these modifications could explain the differences observed in beta cell differentiation. There are two potential mechanisms by which TNKS-specific inhibitors may poise PPs for β-like cell differentiation: (1) the reduced proliferation rate and (2) the upregulation of genes involved in actin filament-based processes, regulation of cell adhesion, and extracellular structure organization. Consistent with the former, the use of Aphidicolin, a compound that inhibits DNA replication and cell cycle progression, has been shown to promote the differentiation of hPSC-derived PPs to endocrine lineages^[182]71. Meanwhile, the integrin signaling pathway is known to play crucial roles during pancreas development, where modification of this pathway affects pancreas morphogenesis, islet formation, and cell fate commitment^[183]31,[184]32,[185]72,[186]73. In the human fetal pancreas, the pancreatic epithelium expresses high levels of α[v]β[3] and α[v]β[5] integrin receptors, while the newly formed hormone-expressing cells do not, suggesting that integrin expression is dynamically regulated during development^[187]32. Moreover, integrin signaling is crucial for the establishment of branching morphogenesis^[188]73, and inhibiting the ECM-integrin interaction impairs islet development and, specifically, β-cell formation both in human and mouse settings^[189]32,[190]33. In contrast, in vitro, integrin downregulation at stage 5 promotes endocrine specification^[191]31, consistent with the data showing integrin downregulation in newly formed islet cells in the native pancreas^[192]32. As the process of islet development originates from PPs that express a high level of integrins^[193]32, it is likely that by increasing integrin expression, WIKI4 treatment generates PPs that are primed for endocrine commitment in vitro and specifically accelerates PP commitment towards β-like cells. In support of this, we observed that WIKI4-treated cells at stage 4 had increased cellular spreading, which was suggestive of integrin activation leading to cytoskeletal rearrangement to form filamentous (F-) actin^[194]36. This is in line with studies indicating that ECM-integrin signaling initiates pancreas branch formation by remodeling the actin cytoskeleton, which can promote endocrine commitment^[195]31,[196]35,[197]73. Potentially consistent with this model, we show higher AMOT levels in WIKI4-derived PPs compared to NA-derived PPs. As AMOT is a central component of the integrin signaling complex, increased levels of AMOT in WIKI4-treated culture may indicate that cells are more receptive to integrin activation signals^[198]40. Furthermore, AMOT also regulates the Hippo pathway by inhibiting YAP1 activity, which is crucial to facilitating endocrine commitment and β-like cell differentiation^[199]31,[200]39,[201]74. Hence, we speculate that AMOT could play a key role in generating PPs predisposed to β-like cell lineage commitment, but additional studies are needed to establish causal relationships between tankyrase inhibition, AMOT, integrin activation, and β-like cell commitment using hPSCs as a model for pancreas development. Despite enhanced frequencies of β-like cells, WIKI4-derived islet-like cells only show partial glucose responsiveness after reaggregation and extended cultures, which is in line with the lack of genes indicative of β cell maturation and recent studies showing that full metabolic functionality has only been achieved after transplantation^[202]10. Although NA-derived cells exhibited similar increases in INS expression levels after extended culture, they failed to respond to glucose challenge. Since the extended culture medium contained no exogenous growth factors, we concluded that the process of long-term culture itself is an important contributing factor to inducing β-like cell maturation and function. Overall, while we show that increased integrin activity may contribute to the increased β-like cell output and functionality, it remains to be elucidated whether unknown off-target effects in NA-treated cells could impair β-like cell function. To assess functionality in vivo, we transplanted NA- or WIKI4-derived stage 6 populations into diabetic mice. Transplanted WIKI4-derived β-like cells secreted insulin and exhibited glycemic control both at fasting levels and during glucose challenge. This was in stark contrast to NA-transplanted mice, which eventually reverted to hyperglycemia. As we transplanted a similar number of NA- and WIKI4-derived β-like cells and confirmed that grafts from both populations were similar in size and contained a comparable proportion of β-like cells, we concluded that NA-derived cells failed to develop glucose responsiveness, leading to the loss of glycemic stability in the mice. However, an alternative explanation could be that the higher number of C-PEP^-/NKX6-1^- cells present in the NA-derived stage 6 culture may negatively impact β-like cell functionality. Sorting for insulin-expressing cells using an insulin-GFP reporter cell line before transplantation could be used to identify any possible differences between the NA- and WIKI4-derived insulin^+ cells. The graft composition post-transplantation was consistent with previous studies, where polyhormonal cells acquired an α-cell fate after maturation^[203]75,[204]76. The presence of exocrine populations, such as acinar and ductal cells, likely originated from the NKX6-1^- /C-PEP^- cells present in the cultures at the time of transplantation. Furthermore, consistent with other studies, we detected some enterochromaffin cells (SLC18A1) in the grafts^[205]54,[206]77. Enterochromaffin cells are not typically found in human or rodent islets and there is little information regarding the origin of these cells during β-like cell differentiation^[207]78–[208]82. Nevertheless, considering WIKI4-transplanted diabetic mice exhibited appropriate glycemic control, this suggests that the presence of exocrine and enterochromaffin populations, while undesirable, is not detrimental to restoring normoglycemia. In addition, no studies to date have directly evaluated the impact of exocrine and enterochromaffin cells on the development and function of neighboring β-like cells, so the paracrine effects of these off-target cell types on β-like cell development are not known. Overall, we demonstrated that TNKS inhibitors can substitute NA and efficiently promote pancreatic specification. Using WIKI4, we show that PPs differentiate into β-like cells with enhanced efficiency and functionality in vitro and in vivo compared to NA-derived islet-like cells. Our study reinforces the concept that improving early commitment is a valuable strategy to enhance β-like cell differentiation efficacy. Methods hPSC culture and differentiation The H1 hESC line was obtained from WiCell Research Institute (Madison, WI, USA). The iPSC line LiPSC-GR1.1^[209]26 was obtained from the National Institute of Neurological Disorders and Stroke (NINDS) cell and data repository at Rutgers University, USA. The HES3 NKX6-1^GFP/W line^[210]15 was provided by Drs. E. Stanley and A. Elefanty (Murdoch Children’s Research Institute, Melbourne, Australia). Dr. Nostro has approval from the Stem Cell Oversight Committee (Canadian Institute of Health Research) to conduct work with hESCs. Undifferentiated hPSCs were maintained and expanded on irradiated mouse embryonic feeder cells in hESC media consisting of DMEM/F12 (Gibco) supplemented with 20% KnockOut serum replacement (Gibco), 100 μM nonessential amino acids (Gibco), 1% glutamine (Hyclone) and 1% Penicillin-Streptomycin (Hyclone), 10^−4 M β-mercaptoethanol (Sigma) and 10 ng/mL human bFGF (R&D Systems) on gelatin-coated 6-well tissue culture plates. Cells were passaged to new feeders as single-cell suspensions, following dissociation with TrypLE^TM Express Enzyme (Gibco). hESC media was supplemented with 10 µM Y-27632 (Tocris) after thawing and passaging. Differentiation begins at stage 1 (d0) when hPSC cultures reach 70-80% confluency. All media were supplemented with 1% glutamine (Hyclone) and 1% Penicillin-Streptomycin (Hyclone). Stage 1 (d0-d1) media consisted of RPMI (Corning), 1% MTG, 100 ng/mL hActivin A (R&D Systems), and 2 µM CHIR99021 (Tocris) for 1 day. Stage 1 media (d1-d3) consisted of RPMI media containing 50 µg/mL ascorbic acid (Sigma), 100 ng/mL hActivin A and 5 ng/mL hbFGF (R&D system). Stage 2 media (d3-d6) consisted of RPMI or MCDB131 (Gibco) media supplemented with 1% vol/vol MACS NeuroBrew-21 (without vitamin A, Miltenyi Biotec), 50 ng/mL hFGF10 (R&D System) and 0.75 µM Dorsomorphin (Sigma). Stage 3 media (d6-d8) consisted of DMEM (Gibco) supplemented with 1% vol/vol NeuroBrew, 50 µg/mL ascorbic acid, 50 ng/mL hNoggin, 50 ng/mL hFGF10, 0.25 µM SANT-1 (Tocris) and 2 µM all-trans retinoic acid (RA) (Sigma). Stage 4 media (d8-d13) consisted of DMEM containing 1% vol/vol NeuroBrew-21 supplement, 50 µg/mL ascorbic acid, 50 ng/mL hNoggin, 100 ng/mL hEGF (R&D System), and either 10 mM Nicotinamide (Sigma) or testing compounds. Cells were mechanically aggregated using a pipette on d13 and were plated in suspension at 2 × 10^6 cells/mL of stage 5 media in low-adherent tissue culture plates. Stage 5 media (d13–d16) consisted of MCDB131 supplemented with 1% vol/vol NeuroBrew-21, 1 µM T3 (Sigma), 1.5 g/L NaHCO[3] (Gibco), 15 mM d-(+)-glucose (Sigma), 10 µg/mL Heparin (Sigma), 0.25 µM SANT-1, 10 µM RepSox (Tocris), 100 nM LDN193189 (Cayman), 10 µM ZnSO[4] (Sigma), 0.05 µM all-trans RA, 10 µL/mL DNase 1 bovine pancreas (Millipore) and 10 µM Y-27632 (Tocris). Stage 6 media (d16-d23/24) consisted of MCDB131 supplemented with 1% vol/vol NeuroBrew-21, 1 µM T3, 1.5 g/L NaHCO[3], 10 µg/mL Heparin, 10 uM RepSox, 100 nM LDN193189, 10 µM ZnSO[4], and 100 nM DBZ (Tocris). To replicate Balboa et al.^[211]10, H1 hESCs were cultured in feeder-free culture conditions using XF medium (Miltenyi Biotec) on Geltrex (Life technologies) coated 6-well plates (Falcon). At day 0, H1 hESCs were harvested and plated at the density of ~180,000–200,000 cells/cm^2 (~10 to 12 × 10^6 cells per 6-well plate). Differentiation was started 24 h after plating, at ~95% confluency. All media were supplemented with 0.5% Penicillin-Streptomycin (Hyclone). Stage 1 Definitive endoderm induction (DE) (d0–d4): The basal medium consisted of MCDB131 (Wisent), Glutamine (1x) (Hyclone), NaHCO3 (1.5 g/L) (Gibco), fatty acid-free BSA (0.5%) (Proliant), d-glucose (5 mM) (Sigma) and ascorbic acid (50 µg/mL) (Sigma). At d0, cells were washed with 1× DPBS^-/- and fed with definitive endoderm induction medium containing 100 ng/mL Activin-A (R&D Systems), 3 µM CHIR99021 (Tocris), and 10 µM Y-27632 (Tocris). At d1, cells were fed with definitive endoderm induction basal medium supplemented with 100 ng/mL Activin-A (R&D Systems) and 0.3 µM CHIR99021 (Tocris), At d2, cells were fed with definitive endoderm induction basal medium with 100 ng/mL Activin-A. Stage 2 Primitive gut tube (PGT) induction (d4- d6): The basal medium consisted of MCDB131, Glutamine (1x), NaHCO[3] (1.5 g/L), Fatty acid-free BSA (0.5%), d-Glucose (5 mM), and Ascorbic acid (50 µg/mL) supplemented with 50 ng/mL FGF7/KGF (R&D Systems). Medium was replaced at Day 4, and then Day 6. Stage 3 Posterior foregut (PFG) induction (d7- d8): The basal medium consisted of MCDB131 (Gibco), Glutamine (1x), NaHCO[3] (1.5 g/L), Fatty acid-free BSA (2%), d-Glucose (5 mM), and Ascorbic acid (50 µg/mL), supplemented with 0.5x ITS-X (Gibco), 0.25 µM SANT-1 (Tocris), 100 nM LDN193189 (Selleckchem), 200 nM TPPB (Tocris), 50 ng/mL FGF7/KGF (R&D System), and 1 µM all-trans retinoic acid (RA) (Sigma). Medium was replaced daily. Stage 4 Pancreatic progenitors induction (PP) (d9–d13): The basal medium consisted of MCDB131 (Gibco), Glutamine (1x), NaHCO[3] (1.5 g/L), Fatty acid-free BSA (2%), d-Glucose (5 mM), and Ascorbic acid (50 µg/mL), supplemented with 0.5x ITS-X (Gibco), 0.25 µM SANT-1 (Tocris), 200 nM LDN193189 (Cayman), 100 nM TPPB, 2 ng/mL FGF7/KGF (R&D System), 100 ng/mL hEGF (R&D Systems), 10 ng/mL Activin-A (R&D Systems), 10 µM Y-27632 and 0.1 µM all-trans retinoic acid (RA). In addition, either 10 mM Nicotinamide (Sigma) or 9 µM WIKI4 (Tocris) were added in this stage, and the medium was replaced daily until d11. At d11, cells were dissociated using TryPLE and aggregated using AggreWell 400 (Stem Cell Technologies) and plated (6 × 10^6 cells/well; approximately 1000 cells/microwell) according to manufacturer instructions. At d12, 2 mL of stage 4 medium was added gently on the side of the wells without disturbing the aggregates. Stage 5 Endocrine progenitor differentiation (d13–d16): Aggregates were maintained in AggreWell 400 at this stage by replacing the spent medium with 4 mL of Stage 5 medium until d15. At d15, aggregates were transferred from AggreWell 400 to a 60 mm petri dishes (Falcon) and maintained as a suspension culture. Each well of AggreWell 400 was split into 3 petri dishes. The basal medium consisted of MCDB131, Glutamine (1x), NaHCO[3] (1.5 g/L), d-Glucose (15 mM), Fatty acid-free BSA (2%), supplemented with 0.5x ITS-X, 10 µg/mL Heparin (Sigma), 0.25 µM SANT-1, 10 µM RepSox (Tocris bioscience), 100 nM LDN193189 (Cayman), 10 µM ZnSO[4] (Sigma), 0.05 µM all-trans retinoic acid (RA), 100 nM DBZ (Tocris), 20 ng/mL Betacellulin (R&D System), 10 µM Y-27632 (Tocris), and 1 µM GC1 (Tocris). Stage 6 Immature SC-Islets differentiation (d17- d23): The basal medium consisted of MCDB131, Glutamine (1x), NaHCO[3] (1.5 g/L), d-Glucose (15 mM), Fatty acid-free BSA (2%), supplemented with 0.5x ITS-X, 10 µg/mL Heparin, 10 µM RepSox, 100 nM LDN193189, 10 µM ZnSO[4], 100 nM DBZ, and 1 µM GC1. The medium was replaced every 2–3 days. Stage 7 Maturing SC-Islets differentiation (d24- d40 + ): The basal medium consisted of CMRL-1066 (Corning), Glutamine (1x), 0.5 mM Sodium Pyruvate (Gibco), Fatty acid-free BSA (2%), supplemented with 0.5x ITS-X, 10 µg/mL Heparin, 10 µM ZnSO[4], 1x Trace element-A (Cellgro), 1x Trace element B (Cellgro), 0.5x Lipid concentrate (Invitrogen), 1 mM N-acetyl cysteine (Sigma), 10 nM Triiodthyronine (T3) (Sigma), and 0.5 µM ZM-447439 (Selleckchem). The medium was replaced every 2–3 days. Reaggregation and extended culture Reaggregation was performed on d20 of differentiation. Cells were dissociated into single cells with 0.25% trypsin + EDTA for 5 min at 37 °C, followed by mechanical dissociation with a pipette. Trypsin-mediated dissociation was inhibited with PBS supplemented with 10% fetal bovine serum (FBS). The cells were filtered through a cell strainer tube (35–45 µm mesh size) to remove undissociated cell clusters. These cells were pelleted and resuspended in stage 6 media supplemented with 10 µM Y-27632 and DNase (Millipore). Cells were plated at the density of 1 × 10^6/mL. For extended culture, we transferred the reaggregated cells on d23 into a modified media^[212]50, consisting of MCDB131 supplemented with 1% glutamine, 1% vol/vol NeuroBrew-21, 10 µM ZnSO[4], 10 µg/mL Heparin and 2.5 mM d-(+)-glucose. Fresh media was replaced every 3–4 days until day 33. Lentiviral transduction of shRNAs targeting TNKS1 and TNKS2 Lentiviral transduction was performed using pLKO.1. puro-based lentiviral vectors containing shRNAs targeting TNKS1 (SH15) and TNKS2 (SH16), as well as a shGFP (Scramble) control. Plasmids were obtained from Dr. Robert Rottapel lab at the University Health Network in Toronto, Canada^[213]83. Each shRNA plasmid (TNKS1, TNKS2, and shGFP) was co-transfected with packaging plasmid (PSPAX2) and the envelope plasmid (PMD2.G) using XtremeGene transfection reagent (Roche). Transfections were conducted in 500,000 293T cells per well of a 6-well plate with 1.5 mL of DMEM-HG and 10% FBS. The following day, the transfection medium was replaced with 1.5 mL of virus harvesting medium, which consisted of DMEM-HG, 10% FBS, and 1.1 g/100 mL BSA (Bioshop). On the evening of day 3, the medium containing the virus was collected and centrifuged for 3 min at 700 × g. The supernatants were collected into cryotubes and frozen at −80 °C until further use. At day 8, pancreatic endoderm cells (PDX1 + ) derived from H1 hESC were dissociated and transduced with the lentiviral vectors (MOI: 5) targeting TNKS1, TNKS2, and ShGFP (Scramble) separately in suspension. The cells were incubated for 20 min at RT with intermittent mixing. After incubation, the cells were replated (1:1 ratio) on a 1/200 diluted Geltrex-coated 6-well plate in stage 4 medium containing 1x Neurobrew, ascorbic acid, Noggin, EGF, and Y-27632 without Nicotinamide or WIKI4. At day 9, the virus-containing medium was replaced with fresh stage 4 medium supplemented with 500 ng/mL of puromycin (Invitrogen) for 48 h to enrich the transduced cell population. At day 12, pancreatic progenitors were collected or fixed for subsequent analysis. Flow cytometry Cell monolayers or aggregates were dissociated into single cells using TrypLE^TM Express Enzyme (Gibco) at 37 °C for 3–5 min. The cells were then pelleted and resuspended in PBS (without Ca^2+ and Mg^2+) and stained with Zombie Violet viability dye (BD Bioscience) for at least 20 min at room temperature. Samples were then fixed in Cytofix/Cytoperm solution for at least 30 min, followed by incubation with primary antibodies overnight at 4 °C diluted in Perm/Wash buffer (BD Science). Next, samples were incubated in secondary antibodies for 30 min at room temperature. For NKX6-1/C-PEP and NKX6-1/Ki67 staining, antibodies were used sequentially to avoid cross-antibody binding. Samples were first stained with NKX6-1 antibody diluted in Perm/Wash buffer overnight at 4 °C, followed by secondary antibody at room temperature for 30 min. Then, the samples were stained with C-PEP or Ki67 primary antibody for 1 h at room temperature and followed by secondary antibody for 30 min at room temperature. Samples were washed with PBS (without Ca^2+ and Mg^2+) supplemented with 10% FBS between antibodies. Primary and secondary antibodies used were: anti-NKX6-1 (DSHB F55A10; 1:1000), anti-PDX1(R&D AF2419; 1:100), anti-C-Peptide (DSHB GN-ID4; 1:400), anti-Ki67 (Abcam Ab15580; 1:1000), anti-Glucagon (Sigma G2654; 1:2000), Alexa Fluor 647 anti-mouse IgG (Invitrogen A31571; 1:400), Alexa Fluor 488 anti-goat IgG (Jackson 705-546-147; 1:400), PE anti-rat IgG (BD Pharmingen 550767; 1:400) and Dylight 550 anti-rabbit (ThermoScientific SA5-10039; 1:400). Flow cytometry analysis was performed using FlowJo software (Becton Dickinson, version 10.9). Representative gating strategy for flow cytometry is shown in Fig. [214]S13. Western blot analysis Samples were lysed with RIPA buffer (Millipore) supplemented with 1% phosphatase (Cell signaling) and 1% protease inhibitors (Cell Signaling). Samples were then placed on ice for 20 min, followed by centrifugation at 13,500 × g at 4 °C for 20 min. The supernatant was collected, and total protein content was quantified using a BCA assay (ThermoFisher). Protein samples were then denatured at 70 °C for 10 min. The samples were then run on 4-12% mini protein gels (ThermoFisher) at 165 V for 35 min, followed by transfer to a PVDF membrane (Invitrogen) at 30 V for 1 h. The membrane was then blocked with 5% skim milk diluted in TBS + 0.1% Tween-20 (Cell Signaling) for 1 h at room temperature, followed by primary antibody (diluted in 5% milk) incubation overnight at 4 °C. The membrane was then incubated in a secondary antibody (diluted in 5% milk) for 1 h at room temperature. All membranes were then developed with ECL reagents (BioRad) and then imaged with a ChemiDoc imaging system (BioRad). The intensity of each band was quantified using ImageJ, and data was normalized to the loading control ERK2. Primary and secondary antibodies used were: anti-ERK2 (Santa Cruz Sc-1647; 1:1000), anti-AMOT (Bethyl A303-305A; 1:2000), anti-TNKS1/2 (Santa Cruz sc-365897; 1:500), anti-AXIN1 (Cell Signaling C76H11; 1:1000), HRP-linked anti-mouse IgG and anti-rabbit IgG (ThermoFisher 7076S/7074S; 1:1000). Real-time-quantitative polymerase chain reaction (RT-qPCR) Total RNA was extracted from pellets using PureLink RNA mini kit (Ambion). In total, 1 µg of RNA was reversed transcribed into cDNA using Superscript III reverse transcriptase (Invitrogen) and RNaseOUT ribonuclease inhibitor (Invitrogen). qPCR was performed using SsoAdvanced universal SYBR green supermix (BioRad) and the CFX Connect real-time PCR system (BioRad). Gene expression was normalized to the housekeeping gene TBP for each sample by using the starting quantity or Delta CT method. Two human adult islet preps (Donor age 60, BMI = 26.0, 95% purity; Donor age 58, BMI = 28.1, 60–70% purity) were used as positive controls. Human islets for research were provided by the Alberta Diabetes Institute IsletCore at the University of Alberta in Edmonton ([215]http://www.bcell.org/human-islets) with the assistance of the Human Organ Procurement and Exchange (HOPE) program, Trillium Gift of Life Network (TGLN) and other Canadian organ procurement organization. Islet isolation was approved by the Human Research Ethics Board at the University of Alberta (Pro00013094). All donors’ families gave informed consent for the use of pancreatic tissue in research. The primers used are listed below: Primers Forward’ Reverse’ ACTA1 CGTCATGGTCGGTATGGGTC ATGATGCCGTGCTCGATAGG ACTA2 GAAAATGAGATGGCCACTGCC TGATCACTTGCCCATCAGGC CDKN1A CTCAAATCGTCCAGCGACCT TGTCTGACTCCTTGTTCCGC CDKN2B TAAGCCTGCAAGCCTGTCTG TCGCTTCATGTTGAGTGTCG ESRRG GCTAACACTGTCGCAGTTTGA CGAACAGCTGGAATCAATGTG INS AGAAGCGTGGCATTGTGGAACA TATTCCATCTCTCTCGGTGCAGGA ITGA3 GTGCGAGGCAATGACCTAGA CCCGTCTCCAGGTAGTCTGT ITGA5 CTCCATTAGCCAGGGTGTGC TTGGTGGTGCAGTTGAGTCC ITGAV CAACAGGCTTGAACGCAGTC TAGCCAAAGCTTGGTGGCAT ITGB1 GTGGAAAAGACTGTGATGCCTT GGCTGGTGCAGTTCTGTTCA MKI67 AATTTGCTTCTGGCCTTCCC GACCCCGCTCCTTTTGATAGT MAFA ATTCTGGAGAGCGAGAAGTGCCAA CGCCAGCTTCTCGTATTTCTCCTT NKX6-1 AGAGGACGACGACTACAATAAGCC ACTTGTGCTTCTTCAACAGCTGCG PDX1 TACTGGATTGGCGTTGTTTGTGGC AGGGAGCCTTCCAATGTGTATGGT SIX2 AAGGCACACTACATCGAGGC CACGCTGCGACTCTTTTCC TBP TGAGTTGCTCATACCGTGCTGCTA CCCTCAAACCAACTTGTCAACAGC TOP2A GCTGCCCCAAAAGGAACTAA GGCGATTCTTGGTTTTGGCA UCN3 GAGGCACCCGGTACAGATAC GAGGGACAGGGTGAACTTGG UCP2 TCCAAGGAGAAAGTCAGGGG CCAGCCCATTGTAGAGGCTT [216]Open in a new tab Bulk RNA sequencing and pre-processing Approximately 1,000,000 cells from 3 separate replicates generated from NA- and WIKI4-derived stage 4 populations at day 13 were harvested and RNA extraction was performed as described. Libraries were produced using Illumina TruSeq Stranded Total RNA Sample Preparation kit with RiboZero Gold. Samples were then sequenced on an Illumina NextSeq500 using a paired-end 75 bp strategy to achieve a depth of 40 million reads per sample. Sequencing was performed by Princess Margaret Genomic Centre, Princess Margaret Cancer Research Tower, Toronto, Canada. Raw sequencing reads were first aligned to the Hg38 human genome^[217]84 using HISAT2 (v2.0.5)^[218]85, and uniquely mapping reads were quantified at the gene level using the featureCounts program from the Subread package (v1.6.2)^[219]86. Raw reads were converted to Transcripts Per kilobase per Million (TPM) for use in subsequent analyses. Heatmaps and Volcano plots were generated in R with the NMF (v0.26)^[220]87 and ggplot2 (v3.4.3)^[221]88. Gene Entrez IDs were converted to common gene symbols using the R package org.Hs.eg.db: Genome-wide annotation for Humans (Marc Carlson, 2019, v3.10.0). Differential gene analysis To identify differentially expressed genes (DEG) between NA- and WIKI4-derived stage 4 populations, the Bioconductor EdgeR (v3.36.0)^[222]89 package was employed. DEG with a P < 0.01, FDR < 0.01 were retained. These results were confirmed using DESeq2 (v1.34.0)^[223]90 (adjusted P-value < 0.01), and only genes identified by both algorithms were considered differentially expressed. The list of mammalian WNT target genes was obtained from [224]https://esbl.nhlbi.nih.gov/Signaling-Pathways/Wnt-Pathway/, and it is based on an investigator-curated website by the Nusse Laboratory at Stanford ([225]https://web.stanford.edu/group/nusselab/cgi-bin/wnt/), listing target genes of Wnt/beta-catenin signaling in multiple species. To identify potentially enriched pathways, pathway enrichment analysis was performed on DEG using Metascape ([226]http://metascape.org) on June 18, 2022^[227]91. Gene set enrichment analysis (GSEA) was performed on TPM data using the Broad Institute’s GSEA software (v4.1.0)^[228]92. Gene sets were obtained from the Broad Institute’s Molecular Signatures Database (MSigDB, [229]https://www.gsea-msigdb.org/gsea/msigdb/index.jsp). Glucose stimulated insulin secretion (GSIS) assay To measure β-like cell function, 3–5 technical replicates containing ~20 aggregates were equilibrated in KRB buffer (pH 7.4; 0.1% FFA-free BSA) containing low glucose (2.8 mM) for 1 h. The aggregates were then incubated sequentially in low glucose (2.8 mM glucose), high glucose (16.7 mM glucose), resting (2.8 mM glucose), and KCl (16.7 mM glucose + 45 mM KCl) for 1 h each. All incubations were performed at 37 ^oC. The supernatant was collected at the end of each incubation and insulin content was quantified by HTRF assay (Cisbio). The stimulation index was calculated by the insulin secreted in each condition divided by the insulin secreted during low glucose incubation. Transplantation studies Male NOD-Rag1^null IL2rg^null (NRG) mice were acquired from Princess Margaret Cancer Research Tower, Toronto, Canada. Mice were maintained according to protocols approved by the Animal Care Committee at University Health Network. The housing conditions for the mice were as follows: room temperatures: 21–22 °C, room humidity: 40% on average, 12:12 h light cycle. Male mice aged 8-20 weeks received 1 dose of high-dose streptozotocin (STZ; 150 mg/kg) at least 1 week prior to transplantation. Hyperglycemia was confirmed by fasting blood glucose levels higher than 10 mM for two consecutive days. A total of ~1–1.5 × 10^6 NA- or WIKI4-derived β-like cells at day 23 (reaggregated at day 20) were transplanted underneath the kidney capsule of the diabetic mice. Insulin pellets (LinBit, LinShin) were placed subcutaneously at the time of transplantation. We performed weekly fasting blood glucose measurements using the OneTouch glucometer and glucose strips, between 4 to 16 weeks post-transplantation. Mice were fasted for 4 h for all studies. An intraperitoneal glucose tolerance test (IPGTT) was performed at week 15 post-transplantation. Mice received an intraperitoneal injection of d-glucose (3 g/kg in filtered water), and blood glucose was measured at 0, 15, 30, 60, 90, and 120 min post-injection. In vivo GSIS assays were performed at 16 weeks post-transplantation. Blood was collected at 0 and 60 min after glucose injection. Samples were centrifuged at 4000 × g for 10 min, and serum was collected. Human C-PEP was quantified using the Ultrasensitive C-peptide ELISA kit (Mercodia). Mice were sacrificed at week 16 post-transplantation, and kidneys were collected for further analysis. Sectioning and immunofluorescent staining In vitro aggregates were fixed with 4% PFA for 20 min at room temperature and embedded in OCT, followed by cryo-sectioning at 10 µm thickness. Sections were washed with PBST (PBS + 0.1% Tween-20) and permeabilized with Perm buffer (0.1 M Glycine and 0.5% Triton-X) for 15 min, followed by a 2-h incubation in blocking solution (diluted in PBST, containing 10% FBS, 3% donkey serum, 0.3% BSA). Sections were then incubated in primary antibodies diluted in a blocking solution overnight at 4 °C, followed by secondary antibodies diluted in a blocking solution at room temperature for 1 h. Nuclei were stained with DAPI for 5 min at room temperature and were mounted with a cover slip and DAKO mounting solution (ThermoFisher). Section images were taken using the Nikon A1R confocal microscope. For staining cells attached to the tissue culture plates, cells were fixed with 4% PFA for 20 min at room temperature, followed by permeabilization using Perm buffer for 15 min. Cells were then blocked for 1 h (10% donkey serum in PBS). Next, cells were incubated in primary antibodies ITGB1 (Millipore MABT821; 1:500) and ACTA2 (Abcam Ab32575; 1:200) overnight at 4 °C, followed by secondary antibodies at room temperature for 1 h. Nuclei were stained with DAPI for 5 min at room temperature. Grafts explanted from the kidney capsules were sent to the STAAR Facility, Toronto, Canada, for paraffin sectioning. Sections were de-paraffinized using xylene and rehydrated in a serial dilution of absolute alcohol. Antigen-retrieval was performed, and the sections were blocked using 10% non-immune donkey (Jackson ImmunoResearch Laboratories Inc.) in PBS. Primary antibodies were diluted in PBS supplemented with 0.3% Triton-X-100 (Sigma) and 0.25% BSA (Sigma) (PBS-Triton-BSA), and incubation was conducted at 4 °C overnight. After washing, the sections were incubated with secondary antibodies in PBS-Triton-BSA for 45 min at room temperature. The primary and secondary antibodies used are listed in Supplementary Table [230]2. Imaging Immunofluorescence images were taken using the Nikon A1R confocal microscope. Brightness, contrast, and color changes of immunofluorescent and H&E staining images were performed using Adobe Photoshop (v21.2.4). Image quantifications were performed using the ImageJ software (v1.53k). Timelapse movie A phase contrast image was taken every 30 min for 48 h using the Lionheart FX automated microscope. The compilation of images was combined to generate a timelapse movie using ImageJ. Statistical analysis GraphPad Prism 9 was used to perform statistical analysis. Results were expressed as mean ± SEM. Reporting summary Further information on research design is available in the [231]Nature Portfolio Reporting Summary linked to this article. Supplementary information [232]Supplementary Information^ (15.9MB, pdf) [233]Peer Review File^ (3.9MB, pdf) [234]41467_2024_53068_MOESM3_ESM.pdf^ (31.4KB, pdf) Description of Additional supplementary files [235]Supplementary Movie 1^ (18.8MB, mp4) [236]Supplementary Movie 2^ (18.9MB, mp4) [237]Supplementary Movie 3^ (21.9MB, mp4) [238]Reporting summary^ (2.5MB, pdf) Source data [239]Source data^ (4.9MB, xlsx) Acknowledgements