A new form of diabetes caused by INS mutations defined by zygosity, stem cell and population data
Yue Tong, Marianne Becker, Ulrike Schierloh, Flávia Natividade da Silva, Leena Haataja, Ying Cai, Kashyap A Patel, Farah Kobaisi, Uyenlinh L Mirshahi, Kevin Colclough, Muhammad Shabab Javed, Matthew N Wakeling, Federica Fantuzzi, Maria Lytrivi, Toshiaki Sawatani
TL;DR
This study shows that a specific insulin gene mutation causes diabetes only when inherited from both parents, using patient data and stem cell models.
Contribution
The study establishes INS R6C as a recessive loss-of-function mutation causing monogenic diabetes, using multi-modal analyses.
Findings
Homozygous R6C individuals develop early-onset diabetes, while heterozygotes do not show significant diabetes risk.
Homozygous R6C β cells show preproinsulin accumulation and reduced insulin secretion.
Transcriptomic analysis reveals downregulation of ER-related pathways in homozygous R6C β cells.
Abstract
The INS c.16 C > T (insulin p.Arg6Cys, R6C) variant was reported to cause autosomal dominant monogenic diabetes, yet its pathogenicity has been questioned. R6C preproinsulin exhibits impaired translocation into the endoplasmic reticulum (ER). We explored R6C pathogenicity using integrative clinical, genetic, and functional approaches.Homozygous INS R6C individuals presented early-onset insulin-treated diabetes, whereas heterozygous carriers showed variable or absent glycemic phenotypes. Population-level analysis revealed no significant enrichment of diabetes among heterozygotes. Heterozygous R6C patient’s induced pluripotent stem cell (iPSC)-derived pancreatic β cells exhibited minimal defects, while homozygous R6C β cells displayed preproinsulin accumulation and reduced insulin content and secretion. In vivo, homozygous R6C β cell transplants recapitulated insulin deficiency and…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 10
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 11
Figure 12- —Innovative Medicines Innitiative Joint Undertaking - INNODIA
- —Innovative Medicines Innitiative Joint Undertaking - INNODIA HARVEST
- —http://dx.doi.org/10.13039/501100002661Fonds De La Recherche Scientifique - FNRS (FNRS)
- —FNRS-Weave
- —http://dx.doi.org/10.13039/501100022434Walloon excellence in life sciences and biotechnology (WELBIO)
- —Excellence of Science (EOS)
- —http://dx.doi.org/10.13039/100000002HHS | National Institutes of Health (NIH)
- —National Institute for Health Research (NIHR) Exeter Biomedical Research Centre, and Clinical Research facilities
- —http://dx.doi.org/10.13039/501100003134FNRS | Fonds pour la Formation à la Recherche dans l'Industrie et dans l'Agriculture (FRIA)
- —http://dx.doi.org/10.13039/100010269Wellcome Trust (WT)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsPancreatic function and diabetes · Diabetes and associated disorders · Diabetes Treatment and Management
The Paper ExplainedProblemDiabetes can sometimes be caused by a “single letter typo” (mutation) in the DNA, called monogenic diabetes. One of these forms is called insulin R6C, where the 6^th^ amino acid arginine in the insulin precursor protein is changed to a cysteine. It affects how the blood sugar-controlling hormone insulin is produced. It was unclear if having one copy of this typo (inherited from one parent) is enough to cause the disease, or if two copies (inherited from both parents) are necessary.ResultsIn this study, we used patient data, genetic analyses, and stem cells to understand the impact of the R6C mutation in insulin. We found that individuals with two R6C copies develop diabetes in childhood. Those with only one copy are not, however, at increased risk of developing diabetes. To explore this further, we created insulin-producing β cells from R6C patients’ stem cells. β Cells with two R6C copies produced and released less insulin, also when these β cells were transplanted into mice. In contrast, β cells with one R6C copy performed nearly as well as normal β cells.ImpactOverall, our results show that the R6C mutation causes diabetes only when both copies of the DNA are affected. People carrying just one copy remain healthy unless exposed to other diabetes risk factors. Our findings highlight the value of patients’ stem-cell models to study genetic mutations and test treatments, helping doctors provide better guidance to families carrying these rare genetic changes.
Introduction
Monogenic diabetes accounts for ~3% of diabetes cases diagnosed under 30 years and exhibits a broad spectrum of clinical presentations, including neonatal, maturity-onset diabetes of the young, and syndromic forms (American Diabetes Association Professional Practice, 2025; International Diabetes Federation, 2025). This proportion may be an underestimation, as it is often misdiagnosed as type 1 or type 2 diabetes. To date, mutations in over 70 genes have been described, most of which disrupt pancreatic β-cell development, function, and/or survival (Greeley et al, 2022; De Franco et al, 2023; Perera et al, 2023). Since the first description linking mutations in the INS gene (encoding insulin) to diabetes (Støy et al, 2007; Colombo et al, 2008), INS mutations have been recognized as a major cause of monogenic diabetes. While insulin plays a central role in glucose homeostasis, INS diabetes can manifest across a broad age range—from within the first 6 months of life to adulthood (Støy et al, 2021). Inheritance can be dominant or recessive, with penetrance varying substantially depending on the type and position of the mutation (Garin et al, 2010). This clinical heterogeneity underscores the need for careful interpretation of INS variants when encountered in diagnostic sequencing.
Depending on the affected domain or amino acid, INS mutations interfere with distinct stages of insulin biosynthesis, including mRNA translation into preproinsulin, signal peptide-mediated endoplasmic reticulum (ER) translocation, proinsulin folding and disulfide bond formation, and conversion into mature insulin. Most are heterozygous missense mutations affecting cysteine residues critical for disulfide bond formation and insulin folding, with highly penetrant dominant inheritance (Wang et al, 1999; Støy et al, 2007; Colombo et al, 2008; Wang et al, 2020). Unlike most monogenic diabetes genes, where haploinsufficiency underlies dominant inheritance, these dominant INS mutations act through a toxic gain-of-function mechanism where misfolded proinsulin induces severe β cell ER stress. Recessive INS mutations typically involve reduced insulin synthesis or production of nonfunctional and/or degradation-prone proteins, representing classic loss-of-function with variable penetrance (Støy et al, 2021; Garin et al, 2010; Tans et al, 2024).
Despite this growing mechanistic insight, many INS variants are reported only once and not functionally followed-up (Støy et al, 2007; Colombo et al, 2008; Molven et al, 2008; Boesgaard et al, 2010; Støy et al, 2021). The identification and interpretation of variants have been greatly facilitated by advances in genomic technologies and in silico predictive tools. However, assessing variants with subtle or context-dependent effects remains challenging (Wright et al, 2024). Large-scale and diverse population genetic databases such as gnomAD (Atkinson et al, 2023), TOPMed (Taliun et al, 2021), All of Us (All of Us Research Program Investigators et al, 2019), and UK Biobank (Sudlow et al, 2015) have revealed that certain rare variants—thought to be pathogenic based on limited family data—may be more prevalent than initially estimated.
Comprehensive clinical, genetic and cell studies remain essential for variant classification and to guide clinical care. Access to primary pancreatic islets from patients—especially those carrying rare mutations—is exceedingly rare if not impossible. Non-human models or human β cell lines may not fully recapitulate the patient-specific regulatory landscape and cellular context.
The NM_000207.3: c.16 C > T, p.(Arg6Cys) INS mutation (sixth arginine substituted by cysteine, henceforth referred to as R6C) exemplifies these challenges. It was first reported as an autosomal dominant mutation in a family with three heterozygous R6C individuals who were diagnosed with non-insulin-dependent, non-obese diabetes in adolescence or adulthood (Edghill et al, 2008). This mutation is located in the preproinsulin signal peptide. The signal peptide consists of three domains: a central hydrophobic h-region that binds with subunit SRP54 of the signal recognition particle (SRP) (Gutierrez Guarnizo et al, 2023), which targets nascent protein to the ER membrane, a positively charged n-region that promotes electrostatic interactions with the negatively charged phosphate group of lipids in the ER membrane (“positive-inside” rule (Nesmeyanova et al, 1997)), and a c-region containing the cleavage site recognized by signal peptidase. The n- and h-regions are critical for co-translational translocation of preproinsulin, guiding the ribosome–nascent chain complex to the SRP and enabling its translocation through the SEC61 translocon (Lang et al, 2017; Xu et al, 2024). Alterations in charge or hydrophobicity caused by R6C may potentially impair SRP recognition and ER targeting (Liu et al, 2015; Gutierrez Guarnizo et al, 2023; Miller et al, 2024; Sánchez et al, 2025).
INS R6C has been considered to cause autosomal dominant early-onset diabetes (Meur et al, 2010; Hussain et al, 2013; Støy et al, 2021; Chua et al, 2024), but later-onset cases have been described (Bansal et al, 2017). Autosomal dominant R6C diabetes could result from misfolded proinsulin, ER stress, and β-cell death (Støy et al, 2007; Liu et al, 2012; Balboa et al, 2018). However, R6C transfection in rat INS-1E β and HEK293T cells resulted in inefficient translocation of R6C preproinsulin across the ER membrane and 50% reduced insulin production, but no toxic gain-of-function (Guo et al, 2014, 2018). Time-resolved translocation assays support a primary ER targeting defect (Iwasa et al, 2025).
Here, we analyzed aggregated population data and challenged the assumption of autosomal dominant INS R6C inheritance, prompting reconsideration of its penetrance and disease mechanism. We took advantage of patients’ iPSC-derived β cells, a species- and disease-specific model that replicates β cell failure seen in patients, to further the understanding of pathogenic mechanisms and bridge clinical genetics with β-cell biology.
Results
Homozygous R6C mutation leads to early-onset diabetes, while heterozygous carriers exhibit variable adult-onset hyperglycemia
The proband and her paternal uncle developed diabetes at age 11 and 9 years, with markedly elevated hemoglobin A1c (HbA1c ∼100 mM/M, ∼11%) and a BMI of 19.6 (percentile 79) and 20.8 kg/m^2^ (Fig. 1A; Table 1, A-III-1 and A-II-2). The proband had a random C-peptide of 3.3 ng/mL (1092.3 pM) concomitant with a blood glucose level at 286 mg/dL (15.9 mM). In another family, the proband (Fig. 1B; Table 1, B-II-1) was diagnosed with diabetes at 13 years with HbA1c 78 mM/M (9.3%) and BMI 27.2 kg/m^2^. All three were treated with insulin at diagnosis. Proband B-II-1 stopped insulin after 5 years due to excellent glycemic control (lowest HbA1c 39 mM/M, 5.7%). Two months later, her random C-peptide was 1.9 ng/mL (612 pM); 5 months later, her HbA1c was 49 mM/M (6.6%). Neither proband (A-III-1 and B-II-1) had ketoacidosis or β-cell autoantibodies GADA, IA2A, IAA, or ZnT8A at diagnosis or during follow-up. Genetic predisposition of type 1 diabetes was low (type 1 diabetes genetic risk score 15.2 and 21 centile of type 1 diabetes population, respectively (Oram et al, 2016)), suggestive of non-autoimmune diabetes. Targeted whole-genome sequencing revealed that proband A-III-1, 2 of her uncles (A-II-2 and A-II-1) and proband B-II-1 were homozygous carriers of the INS R6C variant. Overall genome-wide homozygosity in proband A-III-1 was low (1.7%), suggesting that the parents were not closely related. One region of homozygosity on chromosome 11 (chr11:206,682– 2,430,597) encompassed the insulin gene. The childhood to young adult onset of diabetes in homozygous R6C individuals, along with remaining endogenous insulin secretion, suggests a moderate or delayed impact on β cells.Figure 1. Pedigree of the homozygous R6C families.Red arrow indicates the probands in the two families (A, B). Black solid symbol indicates diagnosis of diabetes, void symbol indicates unaffected individual, gray shaded indicates impaired glucose tolerance (IGT) or prediabetic. Roman numerals show generations and Arabic numerals individuals within each generation. R6C/R6C: homozygous, R6C/N: heterozygous. N/N non-carrier; other individuals lack genetic confirmation. Dx age at diagnosis, GLP-1RA glucagon-like peptide-1 receptor agonist, insulin injections or pump, OHA oral hypoglycemic agents.Table 1. Clinical characteristics of INS R6C carriers.INS mutationR6C: c.16 C > T p.(Arg6Cys)FamilyABIdentityA-III-1 (proband)A-II-1A-II-2A-II-3A-II-4A-I-1A-1-2B-I-1B-I-2B-II-1 (proband)HomozygosityyesyesyesnonononononoyesGenderfemalemalemalemalefemalemalefemalemalefemalefemaleClinical characteristicsAge at diagnosis (years)11209^a^3241^b^4839-29^c^13BMI (kg/m^2^)19.624.420.839---26.830.727.2HbA1c (%)11.91111.410.26.2--5.46.59.3TreatmentInsulinInsulinInsulinGLP-1RA + insulin-MetforminInsulin-Metformin, gliclazideInsulin^d^Variant described according to Human Genome Variation Society guidelines based on isoform NM_000207.3. Insulin was the first treatment for all homozygous patients.^a^This individual stopped insulin treatment at 18 years on his own initiative, because of poorly controlled glycemia.^b^The mother of the proband (A-II-4) was diagnosed with gestational diabetes and impaired glucose tolerance at age 41 years.^c^The mother of the proband (B-I-2) was diagnosed with gestational diabetes.^d^HbA1c was 7.2% 2 months after insulin initiation and 6.7% 10 months after. The proband reduced refined carbs, increased physical activity, and lost weight (BMI 22.7 kg/m^2^), with HbA1c 5.8% off treatment for nine months.
We reviewed the clinical characteristics of heterozygous R6C individuals from this study (Fig. 1A,B) and the first reported family (Appendix Fig. S1A (Edghill et al, 2008)). Age at diabetes diagnosis was 36 ± 17 years and BMI 29 ± 5.2 kg/m^2^ (mean ± SD). Most heterozygous individuals did not receive insulin treatment at diagnosis. The father of proband A-III-1 had a random C-peptide of 3.7 ng/mL (1225 pM) (Fig. 1A; Table 1, A-II-3) with concomitant blood glucose of 180 mg/dL (10 mM). Her mother had gestational diabetes at 27 and impaired glucose tolerance at 41 years (Fig. 1A; Table 1, A-II-4). The mother of proband B-II-1 (Fig. 1B; Table 1, B-I-2) had had gestational diabetes and was diagnosed with diabetes at 29 years. Another heterozygous same-site INS R6H mutation (NM_000207.3: c.17 G > A, p.(Arg6His)) has been suggested to result in milder β-cell dysfunction than heterozygous R6C (Appendix Fig. S1B (Guo et al, 2014, 2018; Meur et al, 2010)); these non-obese individuals had adult-onset diet-treated diabetes (Table 2). The clinical heterogeneity among heterozygous R6C and R6H individuals—ranging from impaired glucose tolerance to late-onset diabetes—raises questions about the penetrance of the variant and its classification as an autosomal dominant mutation. We therefore assessed its allele frequency in large-scale population databases.Table 2. Clinical characteristics of previously reported INS R6C and R6H carriers.INS mutationR6C: c.16 C > T p.(Arg6Cys)R6H: c.17 G > A p.(Arg6His)Family^d^Edghill et al, 2008Meur G et al, 2010IdentityC-III-1^a^ (proband)C-II-2C-I-2D-II-3 (proband)D-II-2D-III-2HomozygositynonononononoGenderfemalefemalefemalemalemalefemaleClinical characteristicsAge at diagnosis (years)151565205126^c^BMI (kg/m^2^)24.126.929.324.924.426.8HbA1c (%)------TreatmentOHA+Insulin^b^OHA^b^Diet^b^Diet+OHADietDietVariant described according to Human Genome Variation Society guidelines based on isoform NM_000207.3.^a^In 2018, this proband was found to have a heterozygous HNF1B c.738 G > T p.(Leu246Phe) variant.^b^Diet was the first treatment for C-III-1, C-II-2, C-I-2; oral hypoglycemic agents (OHA) and insulin were started later.^c^The daughter of the proband’s brother (D-III-1) was diagnosed with impaired glucose tolerance at 26 years and with gestational diabetes at 27 years.^d^For the R6C individual with late-onset diabetes identified by Bansal et al, 2017 no clinical characteristics are available.
Integrative population, computational and structural evidence suggest limited pathogenicity of the heterozygous R6C variant
R6C allele frequencies were low although higher than usually seen for monogenic diabetes-causing INS variants, averaging around 0.00002, i.e. a heterozygote frequency of 1 in 25,000 individuals (Appendix Table S1) (Taliun et al, 2021; All of Us Research Program Investigators et al, 2019; Sun et al, 2024; ALFA (Allele Frequency Aggregator), 2024; Carey et al, 2016; Karczewski et al, 2020; Li et al, 2023b). All carriers in these large-scale databases were heterozygous, with notably higher frequencies observed in specific populations: 0.00016 in the Geisinger MyCode cohort, 0.0001 in the European population in the Allele Frequency Aggregator cohort, and 0.00033 in the Middle Eastern population in gnomAD v4.1. There were four R6C heterozygotes in the UK Biobank, and none had diabetes. In the Geisinger cohort of 163,743 individuals, 51 heterozygous R6C carriers were identified (32 singletons and 18 from eight families), with no excess diabetes (odds ratio 1.4; 95% CI, 0.7–2.6). These findings challenge the assumption of a pathogenic effect for R6C heterozygosity; instead, it may be a context-dependent or low-risk variant with variable expressivity. The allele frequency of R6H is around 0.0001 to 0.0005 in gnomAD v4.1, i.e., a heterozygote carrier frequency of 1 in 5000 to 1000 people. In the UK Biobank, 6 of 86 R6H heterozygotes (7%) had diabetes, which was comparable to the general cohort prevalence (6.7%; 33,014 out of 490,029; Fisher’s exact test, P = 0.83). Hence, both R6C and R6H is benign according to recent criteria (Karczewski et al, 2020; Huerta-Chagoya et al, 2024; Li and Polychronakos, 2024).
To further assess the potential impact of the R6C mutation, we employed a panel of in silico prediction tools using algorithms that integrate multiple sources of evidence, quested via dbNSFP version 5.1a (Liu et al, 2020). Three out of 6 classified R6C as damaging (Appendix Table S2). While these in silico predictions are conflicting, its location within the nascent preproinsulin signal peptide n-region might impair SRP recognition and ER SEC61 translocon orienting. The R6C mutation showed loss of positive charge in the n-region (Fig. EV1A) and increased hydrophobicity around the sixth residue (Fig. EV1B). We employed AlphaFold 3 (Abramson et al, 2024) to model the interaction between the preproinsulin signal peptide (amino acids 1–24) and SRP54. Compared to the wild-type signal peptide–SRP54 complex, the R6C complex showed a modest reduction in interaction quality and fewer hydrogen bonds (Fig. EV1C; Appendix Table S3). The SEC61α2, rather than SEC61α1, translocon has been shown to be crucial for insulin biosynthesis (Xu et al, 2024). AlphaFold 3 revealed an inverted orientation of the R6C signal peptide (Fig. EV1D; Appendix Table S3). This suggests weakened signal peptide–SRP54 interaction and misorientation of the signal peptide within the SEC61 translocon, which may compromise ER targeting and translocation efficiency of R6C insulin.
Building on the divergent clinical phenotypes in homozygous and heterozygous individuals (e.g., age at onset, BMI, treatment) and population-level allele frequencies, we next sought to determine to what extent the predicted structural alterations in SRP54 recognition and SEC61 orientation translate into pathological consequences in β cells.
Complementation of the INS R6C mutant does not induce clonal β-cell death
We first asked whether expression of the INS R6C mutant in human β cells induces cell death. We generated a homozygous R6C model by introducing an insulin expression plasmid carrying the R6C variant into insulin-null EndoC-βH1 cells, with a wild-type plasmid as the control; a heterozygous model was generated by co-transfecting equal amounts of R6C and wild-type plasmids. Cell death was similar in wildtype and/or R6C mutant cells 1 and 3 days post transfection (Fig. EV2A; Appendix Fig. S2A,B). The transfection rate was relatively low, however (Appendix Fig. S2A,C). We therefore moved to patients’ induced pluripotent stem cell (iPSC)-derived β cells, carrying the R6C allele at the endogenous INS locus, to model physiological expression levels and assess the mutation’s impact in a patient-specific and disease-relevant context.
Heterozygous R6C iPSC-derived β cells have moderate preproinsulin accumulation and non-affected proinsulin and insulin content and secretion
To understand whether heterozygous R6C is sufficient to disrupt the development, function and/or survival of pancreatic β cells, we generated heterozygous R6C patients’ iPSCs from the proband’s father (Fig. 1A; Table 1, A-II-3). We corrected the R6C variant using CRISPR/Cas9 technology, and we also obtained isogenic non-edited iPSCs (cells went through CRISPR but remained heterozygous R6C) (Appendix Fig. S3). All iPSC lines showed normal karyotype, expressed pluripotency markers, demonstrated capacity to spontaneously differentiate into three germ layers (Appendix Figs. S4, S5) and had no evidence of CRISPR-induced off-target effects (Appendix Table S3). Following an in vitro differentiation protocol (Fig. EV3A, top) that mimics embryonic development of β cells, both heterozygous R6C and corrected iPSCs differentiated normally into islets with comparable islet viability and similar proportions of insulin-expressing β-like cells (Figs. EV2B and EV3B). Heterozygous and corrected R6C cells followed a normal developmental pathway along the differentiation (Fig. EV3C; Appendix Fig. S6A–I) showing transient expression of pancreatic endoderm marker SOX9 and endocrine progenitor marker NGN3 at stage 4 and stage 5, respectively, increasing expression of NKX6-1 and PDX1 from stage 4, and induction of INS, GLP-1R, GCG, SST, NEUROD1, and NKX2-2 from stage 5.
Upon ER translocation, the signal peptide is cleaved by signal peptidase from the nascent preproinsulin, generating proinsulin in the ER lumen. The preproinsulin to proinsulin ratio, quantifying the inefficiency of preproinsulin cleavage into proinsulin, was 1.9-fold higher in heterozygous R6C iPSC-derived β cells than in isogenic corrected cells (Fig. 2A,B), but no obvious differences in intermolecular disulfide-linked proinsulin complexes were detected between the two genotypes (Fig. 2C).Figure 2. Heterozygous R6C iPSC-derived β cells have moderate preproinsulin accumulation and normal proinsulin and insulin content.Unrelated control (black), heterozygous R6C (HET, yellow) and isogenic corrected (HET CORR, blue) iPSCs were differentiated into β cells. (A) iPSC-islet lysates were analyzed by SDS-PAGE under reducing conditions, electrotransfer to nitrocellulose, and immunoblotting with anti-human proinsulin. The blot was cropped and rearranged for clarity. The left panel is reused in Fig. 4A as it represents a shared control. (B) Quantification of preproinsulin to proinsulin ratio from (A) (unrelated control n = 3, HET CORR n = 3, HET n = 4). (C) iPSC-islet lysates were resolved by nonreducing 12%-NuPAGE, and the completed gel was then treated with 100 mM DTT at 60 °C for 10 min before electrotransfer to nitrocellulose and immunoblotting with anti-proinsulin. n = 2. Medium from Min6 β cells transfected with human proinsulin was used as a positive control (INS, lane next to marker, M). The blot was cropped and rearranged for clarity. The left panel is reused in Fig. 4D as it represents the shared marker and positive control. (D) Proinsulin and (E) insulin content (ng) normalized to total protein content (μg). (F) Proinsulin to insulin content ratio from (D, E). (D–F) HET CORR n = 4, HET n = 7. All panels: Unpaired t-test. In box plots, the median of independent experiments is shown by a horizontal line; 25^th^ and 75^th^ percentiles are at the bottom and top of the boxes; whiskers represent the minimum and maximum values. Source data are available online for this figure.
To generate more functionally mature iPSC-β cells, we used long-culture differentiation ((Balboa et al, 2022; Virgilio et al, 2025), Fig. EV3A, bottom). Heterozygous R6C and corrected long-cultured iPSC-islets had comparable proinsulin and insulin content (Fig. 2D,E) and proinsulin to insulin ratio, a surrogate marker of impaired proinsulin processing (Fig. 2F). Heterozygous R6C and corrected β cells had comparable static proinsulin secretion (Fig. 3A); 16.8 mM glucose-stimulated insulin secretion was somewhat decreased (stimulation index: 2.3 ± 0.4 vs. 3.9 ± 0.5) and proinsulin to insulin secretion ratio increased in heterozygous R6C cells (Fig. 3B,C). Dynamic insulin secretion in a perifusion setup showed comparable secretion (with a trend for heterozygous R6C β cells to secrete more, Fig. 3D–H). Altogether, heterozygous R6C β cells exhibit impaired preproinsulin translocation and proinsulin conversion, but otherwise roughly preserved insulin biosynthesis and secretion.Figure 3. Heterozygous R6C iPSC-derived β cells have comparable insulin secretion to corrected cells.Heterozygous R6C (HET, yellow) and isogenic corrected (HET CORR, blue) iPSCs were differentiated into long-cultured β cells. (A–C) Static proinsulin and insulin secretion in response to 2.8 mM glucose (G2.8), 16.8 mM glucose (G16.8), or 16.8 mM glucose plus 10 μM forskolin (G16.8 + Fk). HET CORR n = 4, HET n = 7. (A) Proinsulin and (B) insulin secretion (ng) normalized to protein content (μg). (C) Proinsulin to insulin ratio from (A, B). (D–H) Dynamic insulin secretion upon perifusion with 2.8 mM glucose, 16.8 mM glucose (G16.8), G16.8 plus exendin-4 (Ex4, 50 ng/mL, 11.8 nM) or G2.8 plus KCl (30 mM). HET CORR n = 3, HET n = 4. (D) Insulin secretion normalized to protein content, with (E–H) area under the curve (AUC) per minute of secretion at G2.8, G16.8, G16.8 + Ex4, and G2.8 + KCl. All panels: Unpaired t-test. In box plots, the median of independent experiments is shown by a horizontal line; 25^th^ and 75^th^ percentiles are at the bottom and top of the boxes; whiskers represent the minimum and maximum values. In time course line plots, data are shown as mean ± s.e.m. Source data are available online for this figure.
Accumulated preproinsulin in homozygous R6C iPSC-derived β cells leads to reduced proinsulin and insulin content
We then asked whether increasing R6C allele dosage from 50 to 100% would worsen the cellular phenotype. We generated homozygous R6C iPSCs from the proband (Fig. 1; Table 1, A-III-1), corrected the R6C variant using CRISPR/Cas9, and obtained isogenic non-edited iPSCs (Appendix Fig. S7). All iPSC lines passed the aforementioned quality controls (Appendix Figs. S7–S10; Appendix Table S3). Like the heterozygous cells, homozygous R6C and corrected iPSCs differentiated well into similarly viable islets containing β-like cells (Figs. EV2C and EV3D), following a normal developmental pathway (Fig. EV3E; Appendix Fig. S11A–I) with expression of SOX9 and NGN3 at stage 4 and stage 5, respectively, increasing expression of NKX6-1 and PDX1 from stage 4, and INS, GLP-1R, GCG, SST, NEUROD1, and NKX2-2 from stage 5. NKX6-1 and PDX1 were higher in homozygous iPSC-islets from stage 4 to 7, and INS was also increased (Appendix Fig. S11C,D; Fig. EV3E).
The preproinsulin to proinsulin ratio was 5.3-fold higher in homozygous R6C β cells compared to corrected cells (Fig. 4A,B), and it increased further in the presence of proteasome inhibitor MG132 (Fig. 4C), meaning that up to 2/3 of preproinsulin failed to translocate into the ER. Disulfide-linked proinsulin complexes shifted from smaller disulfide-linked dimers (~15–20 kDa) in corrected cells to higher molecular weight forms (at the top of the gel) in homozygous R6C β cells (Fig. 4D). Long-cultured homozygous R6C iPSC-islets tended to have less proinsulin and had 2.8-fold lower insulin content compared to corrected cells at all stages of maturation, with unchanged proinsulin to insulin ratios (Fig. 4E–H).Figure 4. Accumulated preproinsulin in homozygous R6C iPSC-derived β cells leads to reduced proinsulin and insulin content.Unrelated control (black), homozygous R6C (HOM, pink) and isogenic corrected (HOM CORR, blue) iPSCs were differentiated into β cells. (A) iPSC-islet lysates were analyzed by SDS-PAGE under reducing conditions, electrotransferred to nitrocellulose, and immunoblotted with anti-human proinsulin. The blot was cropped and rearranged for clarity. The left panel is reused as in Fig. 2A, as it represents a shared control. (B) Quantification of preproinsulin to proinsulin ratio from (A) (unrelated control n = 3, HOM CORR n = 3, HOM n = 8) and (C) of homozygous R6C iPSC-β cells treated with vehicle DMSO or MG132 (10 μM) for 30 min (n = 4). (D) iPSC-islet lysates were resolved by nonreducing 12%-NuPAGE and the completed gel then treated with 100 mM DTT at 60 °C for 10 min before electrotransfer to nitrocellulose and immunoblotting with anti-proinsulin. n = 2. Medium from Min6 β cells transfected with human proinsulin was used as a positive control (INS, lane next to marker, M). The blot was cropped and rearranged for clarity. The left panel is reused as in Fig. 2C, as it represents a shared marker and a positive control. (E) Proinsulin and (F) insulin content normalized to total protein content. (G) Proinsulin to insulin content ratio from (E, F). (E–G) HOM CORR n = 8, HOM n = 12–13. (H) Insulin content normalized to total protein content along stage 7 (S7, HOM CORR n = 5, HOM n = 12), 1 week (LC1W, HOM CORR n = 3, HOM n = 3), 2 weeks (LC2W, HOM CORR n = 3, HOM n = 3), 3 weeks (LC3W, HOM CORR n = 3, HOM n = 3), and 4 weeks (LC4W) of long culture (LC, HOM CORR n = 12, HOM n = 15). All panels: Unpaired t-test. In box plots, the median of independent experiments is shown by a horizontal line; 25^th^ and 75^th^ percentiles are at the bottom and top of the boxes; whiskers represent the minimum and maximum values. Source data are available online for this figure.
Homozygous R6C iPSC-derived β cells have impaired insulin and proinsulin secretion
In the static secretion setup, homozygous R6C β cells showed around 50% lower proinsulin and insulin secretion compared to corrected cells (Fig. 5A,B), with unchanged proinsulin to insulin ratio (Fig. 5C). In dynamic secretion studies, homozygous R6C β cells had more than twofold lower insulin secretion, basally, at 16.8 mM glucose, 16.8 mM glucose + 50 ng/mL (11.8 nM) exendin-4 and 2.8 mM glucose + 30 mM KCl (Fig. 5D–I). To investigate whether impaired insulin secretion was linked to mitochondrial dysfunction, we assessed mitochondrial respiratory capacity by Seahorse assay. Basal and pyruvate-stimulated respiration was comparable, and maximal respiratory capacity under FCCP challenge was significantly higher in dispersed long-cultured homozygous R6C iPSC-islets (Appendix Fig. S12A–D). In sum, carrying two R6C alleles impairs ER translocation of nascent preproinsulin by two-thirds, reduces basal and stimulated proinsulin and insulin secretion by 50% and increases mitochondrial respiratory capacity.Figure 5. Homozygous R6C iPSC-derived β cells have impaired proinsulin and insulin secretion.Homozygous R6C (HOM, pink) and isogenic corrected (HOM CORR, blue) iPSCs were differentiated into long-cultured β cells. (A–C) Static proinsulin and insulin secretion in response to 2.8 mM glucose (G2.8), 16.8 mM glucose (G16.8), or 16.8 mM glucose plus 10 μM forskolin (G16.8 + Fk). HOM CORR n = 8, HOM n = 12 (for proinsulin), HOM n = 13 (for insulin). (A) Proinsulin and (B) insulin secretion normalized to protein content. (C) Proinsulin to insulin ratio from (A, B). (D–I) Dynamic insulin secretion upon perifusion with 2.8 mM glucose, 16.8 mM glucose (G16.8), G16.8 plus exendin-4 (Ex4, 50 ng/mL, 11.8 nM), or G2.8 plus KCl (30 mM). HOM CORR n = 12, HOM n = 14. (D) Insulin secretion normalized to protein content, with (E) zoom in on 16.8 mM glucose response. (F–I) Area under the curve (AUC) per minute of secretion at G2.8, G16.8, G16.8 + Ex4 and G2.8 + KCl. All panels: Unpaired t-test. In box plots, the median of independent experiments is shown by a horizontal line; 25^th^ and 75^th^ percentiles are at the bottom and top of the boxes; whiskers represent the minimum and maximum values. In time course line plots, data are shown as mean ± s.e.m. Source data are available online for this figure.
Homozygous R6C β cells recapitulate human insulin deficiency in vivo
We next examined in vivo maturation and function of iPSC-islets (Balboa et al, 2022; Virgilio et al, 2025). Homozygous R6C and corrected β cells were purified from stage 7 iPSC-islets by magnetic-activated cell sorting (MACS) using the cell surface marker CD49a (Veres et al, 2019) to 82 ± 3% vs. 84 ± 3% purity. β-cell-purified aggregates were implanted under the kidney capsule of immunocompromised Rag2 knockout mice (Fig. 6A, (Gorgogietas et al, 2023)). Plasma human C-peptide levels were around or below the detection limit 1 month after transplantation. Fasting and intraperitoneal glucose tolerance test (IpGTT) human C-peptide levels increased gradually over 2, 3, and 4 months of in vivo maturation in mice transplanted with corrected β cells, but they remained low in mice with homozygous R6C grafts (Fig. 6B,C; Appendix Fig. S13A). Body weight, fasting blood glucose and glucose tolerance remained comparable over 4 months (Fig. 6D; Appendix Fig. S13B–D).Figure 6. Humanized homozygous R6C mice secrete minimal human C-peptide.Homozygous R6C (HOM, pink) and isogenic corrected (HOM CORR, blue) MACS-purified iPSC-β cell aggregates were transplanted under the kidney capsule of immunocompromised Rag2 knockout mice and followed up for 4 months prior to a 2-month treatment with dulaglutide (DL, 1 mg/kg, twice weekly, dashed line). The last week (dotted line) represents the washout of treatment. (A) Scheme of in vivo experiment. (B) Area under the curve (AUC) of plasma human C-peptide levels during an intraperitoneal glucose tolerance test (IpGTT) from 1 month after transplantation to 2-month dulaglutide treatment. Mixed-effects analysis with Tukey correction for multiple comparison. (C) Human C-peptide levels during an IpGTT at 4 months after transplantation. Unpaired t-test with Holm–Śidák correction for multiple comparison (HOM CORR n = 6, HOM n = 6). (D) AUC of blood glucose levels during IpGTTs from 1 month after transplantation to 2-month dulaglutide treatment. Mixed-effects analysis with Tukey correction for multiple comparison. For (B, D), sample sizes (n, HOM CORR vs. HOM) for each timepoint were: 1 month (7 vs. 17), 2 months (9 vs. 19), 3 months (7 vs. 17), 4 months (7 vs. 9), 5 months (3 vs. 8), and 6 months (3 vs. 7). HbA1c and treatment of (E) homozygous R6C proband (Fig. 1, A-III-1) and (F) heterozygous father (Fig. 1, A-II-3). (G) Cell death (%) of stage 7 iPSC-islets treated in vitro for 48 h with glucagon-like peptide-1 receptor agonists exendin-4 (Ex4, 50 nM), dulaglutide (DL, 50 nM), and liraglutide (LR, 50 nM) or vehicle PBS (HOM CORR n = 10, HOM n = 7), individual conditions contained missing observations. In box plots, the median of independent experiments is shown by a horizontal line; 25^th^ and 75^th^ percentiles are at the bottom and top of the boxes; whiskers represent the minimum and maximum values. In time course line plots, data are shown as mean ± s.e.m. Source data are available online for this figure.
Insulin deficiency of homozygous R6C β cells is not enhanced by GLP-1 receptor agonists
Functional characterization of the R6C mutation suggests gene dosage-dependent β-cell dysfunction. We then examined the functional activity of insulin molecules secreted by homozygous and heterozygous R6C and isogenic corrected iPSC-islets. HepG2 cells exposed to conditioned medium showed comparable levels of phosphorylated AKT (Protein Kinase B), indicating similar insulin signaling activation (Appendix Fig. S14).
In light of these findings and given the suboptimal glycemic control (HbA1c >8%, Fig. 6E) despite high-dose insulin therapy in the homozygous proband (Fig. 1, A-III-1), we explored adjunctive therapeutic strategies. GLP‑1 receptor agonists (GLP-1RA) enhance glucose‑dependent insulin secretion and suppress inappropriate glucagon release to improve glycemic control (Alfaris et al, 2024). Guided by clinical improvement in the heterozygous father (Fig. 1, A-II-3) following GLP-1RA therapy (Fig. 6F), we explored the potential of GLP-1RA to preserve homozygous R6C β cells and enhance function. The 48-h treatment of homozygous R6C and corrected stage 7 iPSC-islets with exendin-4, dulaglutide, or liraglutide did not alter cell death, rates of which were low (Fig. 6G). GLP-1RA did not induce ER stress markers BiP (HSPA5) or CHOP (DDIT3), and, interestingly, tended to reduce the pro-apoptotic proteins DP5 or PUMA (Appendix Fig. S15A–D).
We then evaluated the long-term effect of GLP-1RA on R6C β cell function. After 4 months of in vivo maturation of iPSC-islets (Fig. 6A), mice were twice weekly injected with dulaglutide for 2 months, with treatment washout in the last week. Body weight was not different (Appendix Fig. S13B). Fasting and IpGTT human C-peptide levels increased in corrected mice even after washout, yet they remained negligible in homozygous mice (Fig. 6B; Appendix Fig. S13A,E). Fasting glucose and glucose tolerance improved in both groups after dulaglutide treatment, and this effect was more profound in corrected mice (Fig. 6C; Appendix Fig. S13C,F), in keeping with their improved C-peptide secretion.
These findings indicate that GLP-1RAs exendin-4, dulaglutide, and liraglutide are not β-cell toxic in the context of the homozygous INS R6C mutation. While GLP-1RAs may serve as a supportive intervention, they are unlikely to restore β-cell function in the presence of homozygous R6C mutation and emphasize the need for genotype-tailored therapeutic strategies.
Repression of protein translation and translocation processes in homozygous R6C β cells
Considering the potential role for ER stress in insulin signal peptide mutations (Liu et al, 2012; Meur et al, 2010), we further investigated the molecular underpinnings of impaired insulin production in R6C β cells. GFP-sorted wildtype or homozygous R6C insulin plasmid-transfected INS-knockout EndoC-βH1 cells had similar BiP and CHOP expression (Appendix Fig. S15E,F). We stressed homozygous, heterozygous, and isogenic control stage 7 iPSC-islets with synthetic ER stressors tunicamycin (an inhibitor of N-linked glycosylation, 48 h), thapsigargin (an inhibitor of the sarco/endoplasmic reticulum Ca^2+^ ATPase, 48 h) and brefeldin A (an inhibitor of ER-Golgi transport, 24 h). All stressors induced cell death, without differences between R6C and isogenic control islets (Fig. EV2B,C). Thapsigargin and brefeldin A induced BiP, CHOP, DP5 and PUMA in homozygous and corrected iPSC-islets (Appendix Fig. S16A–H). GLP-1RAs protected homozygous R6C iPSC-islets from the ER stressors (Appendix Fig. S17A,B) without obvious changes in BiP, CHOP, DP5 and PUMA (Appendix Fig. S16A–H). These results refute ER stress as the driving force of R6C pathogenesis yet indicate that GLP-1RAs confer protection.
To broadly examine gene expression, we performed bulk-RNA sequencing of MACS-purified stage 7 β cells (Fig. 7A). Of the 35 differentially expressed genes (FDR <0.05, |log₂ fold change| >0.58) (Appendix Table S5), the majority displayed low expression levels (<1 TPM), possibly reflecting transcriptional noise at low counts. We manually curated seven small, focused gene sets—β cell identity, ER translocation, ER‑associated degradation & pre‑emptive quality control, unfolded protein response, ER-Golgi transport, insulin processing, insulin secretion, and cytosolic stress (Appendix Table S6). Gene‑wise Z‑scores (Wang et al, 2024) for each category were not different between homozygous R6C and corrected β cells (Fig. 7B). Gene set enrichment analysis (Korotkevich et al, 2021; Subramanian et al, 2005; Fabregat et al, 2018; Kanehisa and Goto, 2000; Ashburner et al, 2000; Liberzon et al, 2015) revealed an overall repressed profile in homozygous R6C cells (Fig. 7C; Dataset EV1), including downregulation of translation initiation and elongation, SRP dependent co-translational protein targeting to membrane, integrated stress response, ERK1 and ERK2 cascade, glycolysis, apoptosis signaling, and immune responses (e.g., interferon signaling and response, interleukin signaling, etc.). Upregulated pathways included pancreatic β-cell signatures, tRNA synthesis, and butyrate metabolism. INS expression was stable (Figs. 7D and EV3E). In conclusion, homozygous R6C β cells have a disrupted transcriptional landscape, characterized by global repression of translation, immune and inflammatory responses, and ER translocation, alongside subtle shifts in pathways critical for insulin processing and secretion.Figure 7. Homozygous R6C β cell transcriptomes demonstrate repression of protein translation and ER-related processes.(A) Scheme of bulk-RNA sequencing of stage 7 MACS-purified iPSC-β cells. (B) Gene-wise Z-score comparing relative gene expression levels between homozygous R6C and isogenic corrected β cells. Unpaired t-test. (C) Bubble plot of enriched gene sets or pathways in GSEA analysis on GO, Hallmark, KEGG and Reactome (FDR <0.05, top 20 of each set). Gene ratio (dot size): represents the proportion of input genes associated with a given pathway relative to the total number of input genes mapped. Normalized enrichment score (NES, dot color): accounts for gene set size and adjusts the enrichment score to allow cross-comparison between gene sets. Positive scores (pink) indicate enrichment in the upregulated set, and negative scores (blue) indicate enrichment in the downregulated set. (D) Log_2_(TPM + 1) of INS gene (TPM: transcripts per kilobase million). Differential expression analysis was performed using DESeq2, which models count data with negative binomial dispersion estimates. P values were adjusted for multiple testing using the Benjamini–Hochberg false discovery rate (FDR). n = 3 independent differentiations for each group. In box plots, the median of independent experiments is shown by a horizontal line; 25^th^ and 75^th^ percentiles (interquartile range, IQR) are at the bottom and top of the boxes; whiskers represent the most extreme values within 1.5 × IQR from the hinges. Data points outside this range are displayed individually as outliers.
Discussion
This study provides a unified mechanistic and clinical framework for understanding the R6C signal peptide variant in the insulin gene. We identified INS R6C homozygosity as the cause of a new recessive form of monogenic diabetes. By integrating clinical observations, patient-derived iPSC-β cell models, transcriptomic profiling, and population-level data, our results suggest reduced penetrance of the R6C variant. Disrupted signal peptide integrity leads to severely reduced insulin biosynthesis and secretion in homozygous R6C β cells, which translates into early-onset, insulin-requiring diabetes. In contrast, one wild-type insulin allele in heterozygous R6C carriers compensates for this defect, consistent with variable adult-onset hyperglycemia or normoglycemia. These findings revise the original classification of R6C as a dominant variant and provide a blueprint for functionally reassessing other variants of uncertain significance in monogenic diabetes.
The increasing availability of large-scale population datasets is reshaping how we interpret the pathogenicity of rare variants. R6C shows a frequency incompatible with penetrant monogenic diabetes. The absence of homozygous carriers in large datasets and early‑onset, insulin‑dependent diabetes in the presently reported families validates a recessive model.
Patients’ iPSC-β cells provided the cellular and molecular lens through which we observed the R6C impact: disrupted preproinsulin ER translocation in homozygous R6C β cells, preproinsulin accumulation and insulin storage and secretion collapse. These findings agree with the in silico model showing impaired SRP54 interaction (Gutierrez Guarnizo et al, 2023) and inverted signal peptide insertion in the SEC61 translocon, reducing ER translocation efficiency. Interestingly, a recent study showed increased photo-crosslinking intensities between the R6C signal peptide and SRP due to increased h-region hydrophobicity (Miller et al, 2024), which could potentially explain less efficient SRP54 to SEC61 handover. However, this disagrees with our in silico prediction of slightly impaired hydrogen bond and protein-protein interaction, underlining that machine learning tools and structural prediction algorithms need to be paired with cellular phenotyping to refine classification frameworks for rare variants.
Downregulated translation initiation/elongation in homozygous R6C β cell transcriptomes indicates that these cells potentially engaged an adaptive proteostasis program to attenuate global protein synthesis and limit further accumulation of preproinsulin. This attenuation is at the level of translation, as insulin mRNA was not reduced. A recent study showed that β cells favor translation and translocation of preproinsulin bearing a strong signal peptide (wild-type R6) against a weak one (R6H), without insulin mRNA differences (Xu et al, 2024). At the mRNA level, there was also downregulation of SRP binding and targeting in homozygous cells, due to less incoming nascent preproinsulin to the ER. We did not observe elevated ER stress in homozygous cells, in keeping with non-human or non-β-cell models (Xu et al, 2024; Guo et al, 2014). We propose that the lack of efficient ER translocation of nascent preproinsulin prohibits their interaction with BiP and translocon-associated proteins (Xu et al, 2022; Li et al, 2019; Haßdenteufel et al, 2018), thus avoiding ER stress. This creates a proteostatic “dormant” niche that preserves viability at the expense of insulin‑secretory function (Rutkowski et al, 2006). Cellular adaptation to persistent stress is further suggested by the increased maximal respiration capacity, mirroring findings in cytosolic α-synuclein-aggregating cells in Parkinson’s disease (Schirinzi et al, 2022; Annesley et al, 2016) and islet amyloid polypeptide-rich INS-1E cells (Soty et al, 2011). Doxycycline-inducible R6C overexpression increased cell death and cytosolic stress with upregulated HSP70 (Guo et al, 2014); this was not observed in our INS-knockout EndoC-βH1 complementation model. Adult mice with conditional tamoxifen-induced deletion of Ins2 on an Ins1-null background showed alleviation of ER stress and β-cell proliferation prior to overt hyperglycemia (Szabat et al, 2016). These effects are more likely the result of acute interruptions in cellular processes or an overload of proteotoxic stress. The patient-derived iPSC-islet model may reflect a more physiologically relevant, gradually acquired homeostatic state in which adaptive mechanisms limit acute cytotoxicity. In the long term, however, little C-peptide is detected in the in vivo R6C β cell model and homozygous R6C carriers do develop diabetes, demonstrating that these compensatory responses are insufficient to sustain insulin secretion under chronic metabolic demand, and that progressive insulin deficiency drives the homozygous R6C diabetes phenotype.
Heterozygous cells, by contrast, maintain sufficient SRP binding and translocation, showing no sign of wild-type proinsulin trapping by R6C mutant proinsulin and largely normal insulin secretion. These findings align with population-level data that heterozygous carriers do not show increased risk of diabetes. Long-term follow-up is essential to assess disease progression in diabetic carriers as well as lifetime diabetes risk among unaffected individuals. The reduced penetrance of R6C and other forms of monogenic diabetes (Li et al, 2023a; Mirshahi et al, 2022; Goodrich et al, 2021) suggests the influence of environmental and/or genetic modifiers (Kim et al, 2003; Kristinsson et al, 2001; Alam et al, 2021). Metabolic and environmental stressors such as obesity, sedentary lifestyle, or pregnancy could increase β-cell workload, unmasking otherwise compensated insulin biosynthesis defects in heterozygotes at a later age (Bonnefond and Semple, 2022; Mittendorfer et al, 2023; James and Stanford, 2025). In the current study, the heterozygous father (A-II-3) developed diabetes with a BMI of 39 kg/m^2^, and the heterozygous mothers (A-II-4, B-I-2) had gestational diabetes, glucose intolerance or diabetes treated with oral glucose-lowering agents, suggesting metabolic demand accelerated diabetes onset. This metabolic demand hypothesis fits with another insulin p.Leu30Met variant (L30M, INS c.88 C > G) resulting in early-onset diabetes. Notably, a heterozygous L30M carrier (BMI 21.7 kg/m^2^) maintained normoglycemia until age 68 years (Meur et al, 2010). Heterozygote individuals may remain normoglycemic until stress-related pressure reveals the defect, and/or genetic background may modulate disease penetrance (George et al, 2021; Kingdom et al, 2024). Polygenic burden, arising from common variants, can indeed modulate the penetrance and severity of monogenic diabetes (Luckett et al, 2025; Huerta-Chagoya et al, 2024; Leech et al, 2025). Future whole-genome sequencing studies and genetic risk score evaluations in large families or cohorts with variable heterozygous phenotypes may help identify protective or sensitizing alleles.
Monogenic diabetes requires tailored therapeutic strategies. Good glycemic control was not achieved in three out of four homozygous R6C patients. Insulin secreted by homozygous, heterozygous and corrected control iPSC-islets showed similar activation of insulin signaling in liver cells, suggesting normal functionality of R6C insulin. Our in vivo experiments demonstrate that GLP-1RA transiently improved glycemia, not by correcting the insulin secretory defect but possibly via glucagon suppression, delayed gastric emptying, or improved insulin sensitivity (Drucker, 2018). These results suggest that homozygous R6C patients may require a combination of therapies to achieve good control.
By integrating population genetics and patient‑derived iPSC models, we redefine INS R6C as a recessive mutation. Our work not only resolves a persistent clinical ambiguity but also establishes a generalizable framework for functionally classifying rare diabetes alleles. Going forward, identifying genetic and/or environmental modifiers of R6C penetrance and evaluating therapies that restore signal‑peptide translocation will be vital for translating these insights into patient care. Moreover, applying proteomics, metabolomics, and spatial transcriptomics will illuminate how signal‑peptide defects converge with broader β‑cell stress pathways and uncover shared targets for optimized intervention.
Methods
Reagents and tools tableReagent/resourceReference or sourceIdentifier or catalog number Experimental models EndoC-βH1 (H. sapiens)Raphaël Scharfmann Lab (Ravassard et al, 2011)NAPeripheral blood mononuclear cells of individual A-III-1 (Table 1) (H. sapiens)Pediatric Endocrinology and Diabetology (DECCP), Centre Hospitalier de LuxembourgNAPeripheral blood mononuclear cells of individual A-II-3 (Table 1) (H. sapiens)Pediatric Endocrinology and Diabetology (DECCP), Centre Hospitalier de LuxembourgNAINS c.16 C > T heterozygous induced pluripotent stem cell (iPSC) line ULBi011.5 (H. sapiens)This studyNAINS c.16 C > T heterozygous iPSC line ULBi011.5-CR29 (non-edited CRISPR control) (H. sapiens)This studyNAINS c.16 C > T heterozygous corrected iPSC line ULBi011.5-CR21 (H. sapiens)This studyNAINS c.16 C > T heterozygous corrected iPSC line ULBi011.5-CR30 (H. sapiens)This studyNAINS c.16 C > T homozygous iPSC line ULBi012.8 (H. sapiens)This studyNAINS c.16 C > T homozygous iPSC line ULBi012.31 (H. sapiens)This studyNAINS c.16 C > T homozygous iPSC line ULBi012.31-CR4 (non-edited CRISPR control) (H. sapiens)This studyNAINS c.16 C > T homozygous corrected iPSC line ULBi012.31-CR3 (H. sapiens)This studyNAINS c.16 C > T homozygous corrected iPSC line ULBi012.31-CR41 (H. sapiens)This studyNARag2 knockout mice (M. musculus)The Jackson LaboratoryB6.Cg-Rag2tm1.1Cgn/JHepG2 (H. sapiens)Raphaël Scharfmann LabNAMIN-6 (M. musculus)Dr. D. StoffersNA Recombinant DNA Insulin pRIG plasmidsRaphaël Scharfmann LabNASeVdp(KOSM)302 L(Nishimura et al, 2017)170523-1 Antibodies Rabbit anti-human OCT4Cell Signaling TechnologyCat# 2840, RRID:AB_2167691Mouse anti-human TRA1-60Thermo Fisher ScientificCat#MA1-023, RRID: AB_2536699Mouse anti-human SSEA4Thermo Fisher ScientificCat#MA1-021, RRID: AB_2536687Rabbit anti-human NanogCell Signaling TechnologyCat#4903, RRID:AB_10559205Goat anti-human SOX17R&D SystemsCat# AF1924, RRID:AB_355060Rabbit anti-human vimentinAbcamCat#137321, RRID:NAMouse anti-human beta tubulin IIIPromegaCat#G7121, RRID:AB_430874Mouse anti-human NKX6.1BD BiosciencesCat# 563022, RRID:AB_2737958Goat anti-human PDX1R&D SystemsCat# AF2419, RRID:AB_355257Guinea pig anti-human INSDakoCat# A0564; RRID:AB_10013624Mouse anti-human GCGSigma-AldrichCat# G2654; RRID:AB_259852Rabbit anti-human SSTAbcamCat#ab108456; RRID: AB_11158517Mouse anti-human proinsulin B-C junction sequence KTRREAEDLQAbmartCat# B-C junction; RRID: AB_2921300Anti-CD49a antibody PE conjugatedBD BioscienceCat#559596; RIDD: AB_397288Anti-PE microbeadsMiltenyiCat#130-105-639; RIDD:NAMouse anti-human β-Actin (HepG2)ProteintechCat# CL555-66009; RRID:AB_2919667Rabbit anti-human p-AKT (HepG2)Cell Signaling TechnologyCat#4060S; RRID: AB_2315049Rabbit anti-human AKT (HepG2)Cell Signaling TechnologyCat#4685S; RRID: AB_2225340anti-human β-Actin (HepG2)SigmaCat#A5441, RRID: AB_476744 Oligonucleotides and other sequence-based reagents PCR primersThis studyAppendix Table S8R6C correction guide RNAThis studyAppendix Table S3R6C correction homology-directed repair templateThis studyAppendix Table S3 Chemicals, Enzymes and other reagents Alt-R CRISPR Cas9 tracrRNA ATTO 550Integrated DNA TechnologiesCat#1075928Cas9 Electroporation enhancerIntegrated DNA TechnologiesCat# 1075916Alt-R HDR Enhancer V2Integrated DNA TechnologiesCat# 10007921Alt-R S.p. HiFi Cas9 Nuclease V3Integrated DNA TechnologiesCat# 1081061DMSOSigma-AldrichCat# D2650Exendin-4Sigma-AldrichCat# E7144DulagutideEli LillyTrulicity 1.5 mgLiraglutideNovo NordiskVictoza 1.2 mgTunicamycinSigma-AldrichCat# T7765ThapsigarginSigma-AldrichCat# T9033Brefeldin ASigma-AldrichCat# B6542Propidium iodideSigma-AldrichCat# P4170Hoechst 33342Sigma-AldrichCat# B2261iProof High-Fidelity DNA PolymeraseBio-RadCat# 1725301ForskolinSigma-AldrichCat# F6886NovoRapidNovo NordiskNA Software GraphPad Prism version 10GraphPad SoftwareNARStudio version 2024.12.1PositNAR version 4.4.3The R Foundation for Statistical ComputingNAFastQC version 0.12.1(Andrews, 2023)NAfastp version 0.23.2(Chen et al, 2018)NASalmon version 1.10.3(Patro et al, 2017)NAtximeta version 1.22.1(Love et al, 2020)NADESeq2 version 1.44.0(Love et al, 2014)NAfgsea version 1.33.1(Korotkevich et al, 2021)NAG*Power version 3.1(Faul et al, 2009)NAFlowJo version 10.7 https://www.flowjo.com/ NAFiji version 1.54p https://imagej.net/software/fiji/ NASavvyHomozygosity https://github.com/rdemolgen/SavvySuite NAUCSF ChimeraX version 1.9 https://www.cgl.ucsf.edu/chimerax/ NAAlphaFold 3 https://www.alphafoldserver.com NACellProfiler version 4 https://www.cellprofiler.org/ NABenchling https://www.benchling.com/ NACRISPOR http://crispor.tefor.net/ NABioRender https://www.biorender.com NA Other QIAamp DNA Micro KitQiagenCat# 51306Bio-Rad ChemiDocBio-RadNAPERI5LM perifusion systemBioRep, MiamiPERI5LMHuman Insulin ELISA KitMercodiaCat# 10-1113-10Human Proinsulin ELISA KitMercodiaCat# 10-1118-01Human Ultrasensitive C-peptide ELISAMercodiaCat# 10-1141-01XFp Extracellular Flux AnalyzerAgilentNAiBLOT Dry Blotting SystemThermo Fisher ScientificNAiBright imagerInvitrogenNADynabeads mRNA DIRECT Purification KitInvitrogenCat# 61012Reverse Transcriptase Core KitEurogentecCat# RT-RTCK-03SsoAdvanced Universal SYBR Green SupermixBio-RadCat# 1725274CFX ConnectBio-RadRNeasy Plus Micro KitQiagenCat# 74034Bioanalyzer RNA 6000 Nano assayAgilentCat# 5067-1511Agilent 2100 BioanalyzerAgilentNEBNext Ultra II Directional RNA Library Prep Kit for IlluminaNew England BioLabsCat# E7760LNovogene NGS Stranded RNA Library Prep SetNovogeneCat# PT044Illumina NovaSeq 6000IlluminaNAIllumina NovaSeq X plusIlluminaNARNeasy Micro KitQiagenCat# 74004First-Strand cDNA KitThermo Fisher ScientificCat# K1612Power SYBR Green mixInvitrogenCat# 4368708QuantStudio 3 analyzerThermo Fisher ScientificNAAxio Imager A1Carl Zeiss3521000414Axio Observer 7Carl Zeiss4919170001000Neon Transfection SystemThermo Fisher ScientificNA
Methods and protocols
Homozygosity mapping
We used SavvyHomozygosity (Wakeling, 2018; Laver et al, 2022) to perform homozygosity assessment using off-target reads from a targeted next-generation sequencing panel in the proband, and confirmed homozygous-only variants in the INS gene targeted region.
Diabetes association in the UK Biobank and the Geisinger cohort
We used data from the UK Biobank, an ethically approved population cohort of over 500,000 individuals from the UK with whole-genome sequencing and deep phenotypic information (Li et al, 2023b) to assess the diabetes association with INS R6C variant carriers. We also used a US health-system-based cohort of 173,247 individuals from Pennsylvania with whole-exome sequencing and deep phenotyping from electronic health records. In both cohorts, diabetes was defined as having one or more of the following criteria: self-reported, having an ICD9/10 code for diabetes, being on a diabetes treatment, or having HbA1c ≥48 mM/M before recruitment. We performed Fisher's exact test to assess the association of diabetes with INS R6C and R6H variants.
In silico modeling of protein structure and interaction
Protein structure modeling was performed using AlphaFold 3 to predict the complex structure (Abramson et al, 2024). The best models were analyzed in ChimeraX v1.9 (Pettersen et al, 2021) for visualization, hydrogen bond detection, and calculation of buried surface area at the binding interface.
EndoC-βH1 cells
The human pancreatic β cell line EndoC-βH1 (Ravassard et al, 2011) was cultured on plates coated with 1.2% Matrigel and 3 μg/ml fibronectin (both from Sigma-Aldrich) at 37 °C in 5% CO_2_ in Advanced DMEM/F-12 medium (Thermo Fisher Scientific) supplemented with 2% bovine serum albumin fraction V (Roche), 6.7 ng/ml sodium selenite, 10 mM nicotinamide (Calbiochem), 50 μM β-mercaptoethanol (Sigma-Aldrich), 100 U/ml penicillin, and 100 μg/ml streptomycin (Thermo Fisher Scientific). INS-knockout cells were generated via CRISPR/Cas9 technology (IDT); these cells are to be described in an upcoming article.
Plasmid cloning, transfection, and FACS
Wildtype and R6C mutated INS cDNA (full length) was synthesized (Eurofins genomics) and cloned downstream of a CMV promoter and upstream of an IRES-GFP cassette in an in-house engineered pRIG plasmid (Albagli-Curiel et al, 2007). Wildtype, R6C or both INS pRIG plasmids were transfected into INS-knockout EndoC-βH1 cells using Lipofectamine 3000 (Thermo Fisher Scientific) according to the manufacturer’s instructions. Empty plasmids without INS cDNA but with IRES-GFP were used as control. At day 1 and 3 post transfection, cells were sorted by GFP into RLT lysis buffer (Qiagen) for qPCR or stained with propidium iodide (BD Bioscience) for cell death analysis. Samples were acquired using FACSAria III (BD Bioscience) and analyzed using FlowJo 10.7 software.
PBMC reprogramming into iPSCs and iPSC quality control
PBMCs from individuals A-III-1 and A-II-3 (Fig. 1; Table 1) were reprogrammed into iPSCs using Sendai virus SeVdp(KOSM)302 L vector-170523-1 (Nishimura et al, 2017). Cells were cultured in RPMI 1640 GlutaMAX medium (61870-010, Gibco) containing 2 mM L-glutamine at 1 million cells per well of a six-well ultra-low attachment plate (3471, Corning). Culture medium was supplemented with 10% (v/v) fetal bovine serum (FBS, 10270106, Gibco) and 1% (v/v) penicillin-streptomycin (15070063, Thermo Fisher). After overnight culture, PBMCs were infected with the KOSM302L vector (harboring OCT4, SOX2, KLF4, and c-MYC genes) at a multiplicity of infection of 2 for 2 h. The medium was then changed to Essential 6 medium (A1516401, Gibco), and cells were transferred to Matrigel-coated plates (BDAA356231, Corning) and medium refreshed daily. From day 8, cells were cultured in Essential 8 medium (A1517001, Gibco), changed every other day. Emerging iPSC colonies were manually picked. The removal of the transgene vector was confirmed by reverse-transcriptase PCR (Appendix Table S7). Embryoid body aggregates were formed by detaching iPSC colonies at 60–70% confluence with 0.5 mM EDTA (15575020, Life Technologies) and culturing in suspension (100 rpm) in ultra-low-attachment plates (38071, Stemcell Technologies) in E8 medium with 10 μM ROCK inhibitor (Y-27632, 72304, Stemcell Technologies). The next day, medium was switched to DMEM/F-12 GlutaMAX medium (31331028, Gibco), 10% (v/v) knockout serum replacement (10828010, Life Technologies), 1% (v/v) MEM non-essential amino acids (11140050, Gibco), 0.1 mM β-mercaptoethanol (31350010, Gibco), and 1% penicillin-streptomycin, refreshed every 2nd day for 1 week. Embryoid bodies (15–20 per well) were plated on Matrigel-coated culture slides (734-0402, Falcon) for 2 weeks, with medium refreshed every other day, and fixed in 4% PFA for immunostaining (Appendix Table S8). For karyotyping, iPSCs from a 3.5 cm dish at 80% confluency were washed with PBS, detached with TrypLE Express (12604013, Life Technologies), collected in E8 medium, and centrifuged for 3 min at 500×g. The supernatant was removed, and cell pellets were stored at −80 °C. Karyotype was assessed by the Karyostat assay (Thermo Fisher).
Genome editing of iPSCs
CRISPR/Cas9 was used to correct the R6C insulin mutation in patient iPSC lines ULBi011.5 (heterozygous) and ULBi012.31 (homozygous) using homology-directed repair (HDR). An isogenic control for patient iPSCs was generated using a 20-base guide (Appendix Table S3). The correction asymmetric single-stranded HDR donor was a 128-bp PCR amplicon using 91-base forward and 36-base reverse primers (Appendix Table S3). Templates were designed to disturb the protospacer adjacent motif (PAM) sequence and introduce a silent mutation to generate a restriction site of BtsI to facilitate the screening of clones using restriction enzymes. Ribonucleoprotein components were purchased from Integrated DNA Technologies, including CRISPR/Cas9 (1081061), crRNA (custom design from Benchling (Benchling [Biology Software], 2025)), tracrRNA (1075928), and the donor HDR template DNA (custom design from Benchling). Two million cells were electroporated with ribonucleoprotein complex (tracrRNA:crRNA:Cas9 = 1:1:1.36) and 4 μg HDR template using Neon Transfection System (1100 V, 20 ms, 2 pulses, Thermo Fisher Scientific) and single-cell-cloned using low density seeding. Single clones were screened using PCR, followed by BtsI-v2 enzyme restriction (R0667S, New England) according to the manufacturer’s protocol. Positive clones and isogenic controls were confirmed by Sanger sequencing. The top five off-target hits predicted by the online tool CRISPOR (Concordet and Haeussler, 2018) were checked, and no off-target indels were found (Appendix Table S3). CRISPRed iPSC lines were quality controlled as described above.
Sanger sequencing
The R6C mutation in the proband and father of the proband in the first family (Fig. 1A) was identified by targeted Sanger sequencing and classified using ACMG/AMP and ACGS 2020_V4.01 guidelines. For iPSC lines, genomic DNA was extracted using QIAamp DNA Micro Kit (51306, Qiagen) and the R6C mutation confirmed by targeted Sanger sequencing (Microsynth, Germany) of PCR amplicons generated using iProof High-Fidelity DNA Polymerase (1725301, Bio-Rad). Primer sequences are shown in Appendix Table S3.
iPSC culture and differentiation into β cells
The following iPSC lines were used: heterozygous ULBi011.5 (from individual A-II-3) and CRISPRed non-edited ULBi011.5-CR29; homozygous ULBi012.8 and ULBi012.31 (from individual A-III-1) and CRISPRed non-edited ULBi012.31-CR4; corrected ULBi012.31-CR3, ULBi012.31-CR41, ULBi011.5-CR21 and ULBi011.5-CR30; healthy control 1.023 (Fantuzzi et al, 2022). iPSCs were cultured in Matrigel-coated plates in E8 medium as described (Cosentino et al, 2018). iPSCs were differentiated into β cells following a previously described seven-stage protocol (Fig. EV3A, (Cosentino et al, 2018; Fantuzzi et al, 2022; Virgilio et al, 2025)). Cells were plated at a density of 2–2.3 million cells per 3.5 cm well in E8 medium with 10 μM ROCK inhibitor. After 24 h, differentiation was initiated. Up to the pancreatic progenitor stage (stage 4), cells were cultured in Matrigel-coated wells. At the end of this stage, cells were transferred into microwell plates at 850 cells per microwell (AggreWell400, StemCell Technologies) to form islet-like aggregates. Differentiation continued in microwells to the end of stage 7. In the long-culture differentiation protocol, aggregates were shifted to suspension culture (100 rpm) at the end of stage 5 and maintained in stage 7 medium (Barsby et al, 2022) for 4 weeks beyond stage 7 (Fig. EV3A, (Virgilio et al, 2025)).
Transplantation of iPSC-β cells
Rag2 knockout mice (B6.Cg-Rag2tm1.1Cgn/J, The Jackson Laboratory) from SCANBUR were housed in the animal facility on a 12-h light/12-h dark cycle with ad libitum food. Transplantations were performed in 8 to 12-week-old male mice as described previously (De Franco et al, 2020; Virgilio et al, 2025). Around 1500 MACS-purified stage 7 iPSC-β cell aggregates were transplanted under the left kidney capsule, using a catheter and syringe or by inserting aggregates clotted with a blood drop from the recipient mouse. Mice were anaesthetized by IP injection of ketamine (100 mg/kg Nimatek, Dechra, UK) and xylazine (5 mg/kg Rompun, Bayer, Germany). Paracetamol was added to drinking water for 10 days after surgery. IpGTTs were performed following a 6-h fast. Glycemia was measured using a glucometer (Accu-Chek Aviva Nano, Roche) at 0,15, 30, 60, 90, and 120 min after glucose injection (2 mg/g body weight). Mouse blood was sampled at 0, 15, 30, 60, and 90 min and plasma separated by centrifugation at 3000 RPM for 20 min at 4 ˚C for C-peptide quantification by human ultrasensitive C-peptide ELISA (10-1141-01, Mercodia). Dulaglutide (Trulicity, Eli Lilly) treatment was given after 4 months of transplantation at 1 mg/kg (dilution in saline) injected IP twice weekly for 2 months, with treatment washout in the last week.
Western blot
iPSC-islets were lysed in Laemmli 2X buffer containing no reducing agents, sonicated (duty cycle 50% of 60 s), heated to 95 ˚C for 10 min, and stored at −80 ˚C. Samples were heated to 95 °C in gel sample buffer containing no reducing agents (nonreducing), or 200 mM DTT for 5 min (reducing), electrophoretically resolved on 12 or 4–12% NuPAGE gels. The gels were either untreated or treated with 100 mM DTT at 60 °C for 10 min, followed by electrotransfer to nitrocellulose. Primary antibodies were incubated at 4 °C overnight. Development of immunoblots used enhanced chemiluminescence, captured with a Bio-Rad ChemiDoc and quantified using Fiji. Antibodies are listed in Appendix Table S8.
Min6 cells
Min6 (mouse insulinoma) cells (from Dr. D. Stoffers, University of Pennsylvania) were cultured in DMEM supplemented with 10% fetal bovine serum (FBS),100 IU/mL penicillin and 100 μg/mL streptomycin, and 0.05 mM β-mercaptoethanol.
(Pro)insulin content and secretion
For static secretion experiments, 50 iPSC-islets were washed twice with glucose-free Krebs buffer and then preincubated for 90 min in 500 μL of βKrebs buffer (BK-100, Human Cell Design) containing 2.8 mM glucose. iPSC-islets were then sequentially incubated for 30-min intervals in βKrebs buffer with 2.8 mM, 16.8 mM glucose, and 16.8 mM glucose plus 10 μM forskolin (F6886, Sigma-Aldrich). Supernatants were collected and stored at −80 °C. Dynamic insulin secretion studies were performed on a PERI5LM perifusion system (BioRep, Miami, USA) with 2.8 mM, 16.8 mM glucose, 16.8 mM glucose + 50 ng/mL exendin-4 (E7144, Sigma-Aldrich), and 2.8 mM glucose + 30 mM KCl at a flow rate of 100 μL/min. Perifusate was stored at −20 °C. iPSC-islets were lysed by sonication (duty cycle 50% of 120 s) and extracted with acid-ethanol. Human insulin and proinsulin were assayed by ELISA (10-1113-10 and 10-1118-01, respectively, Mercodia). Secretion and content were normalized to iPSC-islet protein content determined by Bradford assay (5000006, Bio-Rad).
Mitochondrial respiration
Oxygen consumption rate was measured using XFp Extracellular Flux Analyzer with consumables from Agilent. Stage 7 iPSC-islets were dispersed into single cells and seeded in Seahorse XFp Cell Culture Miniplates at 50,000 cells per well in long-culture differentiation medium. After overnight culture, the medium was refreshed to Ham’s F-10 medium (41550, Gibco) supplemented with 0.75% fatty acid-free BSA, 0.5% (v/v) GlutaMAX and 1% penicillin-streptomycin, and cells were cultured for 1–2 days. On the test day, cells were preincubated in an assay buffer supplemented with 0.2% (w/v) BSA and 0 mM glucose at pH 7.4 (Malmgren et al, 2009) for 1 h at 37 ˚C in a non-CO_2_ incubator. Mitochondrial respiration was assessed at baseline, followed by sequential injections to achieve final concentrations of pyruvate (10 mM), oligomycin (5 µM), FCCP (4 µM), and a combination of rotenone and antimycin A (1 µM) in BSA-free assay buffer. Oxygen consumption rate was normalized to protein content. Basal respiration was calculated by subtracting nonmitochondrial respiration from the last measurement before oligomycin injection, ATP production by subtracting the minimum measurement after oligomycin injection from the last baseline measurement, and maximal respiration by subtracting nonmitochondrial respiration from the maximum measurement after FCCP injection.
In vitro iPSC-islet treatment
Stage 7 iPSC-islets were exposed to 50 nM exendin-4, 50 nM dulaglutide (Trulicity, Eli Lilly), 50 nM liraglutide (Victoza, Novo Nordisk), 1 μM thapsigargin (T9033, Sigma-Aldrich), 5 µg/mL tunicamycin (T7765, Sigma-Aldrich), and 0.025 μg/mL brefeldin A (B6542, Sigma-Aldrich) in Ham’s F-10 medium supplemented with 0.75% fatty acid-free BSA, 0.5% GlutaMAX and 1% penicillin-streptomycin. Vehicle PBS (for exendin-4, dulaglutide and liraglutide) or DMSO (for tunicamycin, thapsigargin, and brefeldin A) was added to the control condition in all experiments. Medium was refreshed every 24 h.
Assessment of apoptosis
The proportion of viable and apoptotic cells was assessed using fluorescence microscopy following a 15-min incubation with DNA-binding dyes propidium iodide (5 μg/mL, P4170, Sigma-Aldrich) and Hoechst 33342 (10 μg/mL, B2261, Sigma-Aldrich). For each condition, at least ten iPSC-islets were counted by two observers, one of whom was blinded to the sample identities. Inter-observer agreement was >90%.
Magnetic-activated cell sorting
Stage 7 iPSC-islets were washed twice with 0.5 mM EDTA and dissociated using Accumax (A7089, Sigma-Aldrich) incubation and mixing for 10 min at 37 °C on a MACSmix Tube Rotator (Miltenyi Biotec). Half volume of KnockOut Serum Replacement was added and gently pipetted, and 10 µM ROCK inhibitor was added to the cell solution. For labeling, 10 million cells were resuspended in 300 μL MACS buffer (PBS, pH 7.2, 0.5% BSA, 2 mM EDTA) plus 200 μL CD49a antibody (PE-conjugated mouse anti-human CD49a, clone SR84, 559596, BD Bioscience) (Veres et al, 2019). Cells were incubated in the dark for 21 min at room temperature on MACSmix Tube Rotator, washed twice with 15 mL MACS buffer, resuspended in 300 µL MACS buffer with 20 µl anti-PE MicroBeads (130-048-801, Miltenyi Biotec) per 10 million cells for 15 min at 4 °C. After a final wash in MACS buffer, cells were sorted using MACS magnetic separators and columns (130-042-201, Miltenyi Biotec). The CD49a-positive fraction was seeded at one million cells per AggreWell.
Conditioned media test in HepG2 cells
HepG2 cells were seeded in 12-well plates pre-coated with collagen I Rat Tail (1:100 diluted in PBS, Gibco, A10483-01) and cultured using DMEM-GlutaMAX (Gibco,61965-026) supplemented with 10% FBS (Eurobio) and 1% penicillin-streptomycin (Gibco). Conditioned medium was collected after 48-h of long-cultured iPSC-islet culture in Ham’s F-10 medium supplemented with 0.75% fatty acid-free BSA, 0.5% GlutaMAX, and 1% penicillin-streptomycin. HepG2 cells were serum starved for 4 days before incubation with conditioned media at a volume corresponding to 5 ng insulin. Medium only was used as a negative control and 5 ng NovoRapid insulin (Novo Nordisk) as a positive control. Cells were treated for 30 min, lysed in RIPA buffer (Sigma) with anti-protease cOmplete and PhosSTOP tablets (Roche) and sonicated. About 15 µg protein was resolved in 4–12% Bis-Tris gel and transferred to PVDF membrane using iBLOT Dry Blotting System (Thermo Fisher Scientific). Membranes were incubated with anti-p-AKT, AKT, or β-Actin antibodies. Species-specific HRP-linked secondary antibodies (Cell Signaling Technology) were utilized. Visualization was carried out on an iBright imager (Invitrogen) following exposure to ECL (GE Healthcare). Image quantification was carried out on Fiji. Antibodies are listed in Appendix Table S8.
RNA extraction, qPCR, and RNA sequencing
Poly(A)^+^-RNA was isolated from iPSC-islets using Dynabeads mRNA DIRECT Purification kit (61012, Invitrogen). Reverse transcription was done using Reverse Transcriptase Core Kit (RT-RTCK-03, Eurogentec). Real-time qPCR amplification was performed using SsoAdvanced Universal SYBR Green Supermix (1725274, Bio-Rad) on the CFX Connect instrument (Bio-Rad) and compared to a standard curve. Expression values were normalized to reference genes β-actin (ACTB) and vesicle-associated membrane protein-associated protein A (VAPA) (Alvelos et al, 2021). Primers are listed in Appendix Table S7.
Total RNA from MACS-purified iPSC-β cell aggregates was isolated using the RNeasy Plus Micro kit (74034, Qiagen). RNA concentration and integrity were measured using Bioanalyzer RNA 6000 Nano assay (5067-1511, Agilent) on the Agilent 2100 Bioanalyzer. Libraries were prepared using NEBNext Ultra II Directional RNA Library Prep Kit for Illumina (E7760L, New England BioLabs) and Novogene NGS Stranded RNA Library Prep Set (PT044, Novogene). All libraries were paired-end sequenced (2x151bp) at 200 million reads on the Illumina NovaSeq 6000 and NovaSeq X plus instrument (Illumina).
For EndoC-βH1 cells, the RNeasy Micro kit (74004, Qiagen) was used to extract total RNA. cDNA was synthesized using the First-Strand cDNA kit (K1612, Thermo Fisher Scientific). RT-qPCR was executed using Power SYBR Green mix (4368708, Invitrogen) with a QuantStudio 3 analyzer. Primers were designed using primer blast and synthesized (Eurofins Genomics). The sequences of the primers are listed in Appendix Table S7.
Immunocytochemistry of iPSC-β cell aggregates
For single-cell staining, cells were dissociated and seeded in culture medium containing 10 µM ROCK inhibitor onto Matrigel-coated ICC chambers. After 24 h, cells were washed once with PBS, fixed in 4% PFA for 15 min, permeabilized with 0.25% Triton X-100 in PBS for 10 min, washed again with PBS, and blocked for 30 min at room temperature using UltraCruz Blocking Reagent (Santa Cruz Biotechnology, Cat#SC-516214) or 3% BSA. Primary antibodies diluted in 0.1% Tween in PBS or 3% BSA were incubated overnight at 4 °C, and secondary antibodies for 45 min at room temperature. Samples were mounted using Vectashield with DAPI (VEC.H-1200, Vector Laboratories) or stained with DAPI (D1306, Thermo Fisher) at 1 mg/mL, mounted with Glycergel mounting medium (C0563, Dako) and covered with glass coverslips. Image quantifications were performed using custom CellProfiler 4 pipelines (Stirling et al, 2021). Antibodies are listed in Appendix Table S8. Immunofluorescence pictures were taken on an Axio Imager A1 (3521000414, Carl Zeiss) or Axio Observer 7 inverted microscope (4919170001000, Carl Zeiss).
Statistical analysis
Statistical analyses were performed with GraphPad Prism (version 10, GraphPad Software). Data points represent independent experiments, and different symbols represent different iPSC lines in each category (homozygous, heterozygous, or isogenic corrected cells). In box plots, the median is shown by a horizontal line; 25^th^ and 75^th^ percentiles are at the bottom and top of the boxes; whiskers represent the minimum and maximum values. In time course experiments (perifusion, Seahorse, and IpGTT), data are shown as mean ± s.e.m. Sample sizes were estimated using power calculations with a two-tailed significance threshold of 0.05 and power set at 0.8, to ensure sensitivity to detect expected effect sizes. Student’s t-test was used to assess significance when comparing two groups with normal distribution (by Shapiro–Wilk test); otherwise, a nonparametric test was applied. For experiments with >2 groups, paired or unpaired one-way or two-way ANOVA or mixed-model analysis (in case of a missing value) was followed by two-tailed t-tests with the indicated correction for multiple comparisons in each figure panel.
For RNA sequencing analyses, FastQC version 0.12.1 (Andrews, 2023) was used to quality control raw reads, and fastp version 0.23.2 (Chen et al, 2018) to filter low-quality reads and adapter sequences. Salmon version 1.10.3 (Patro et al, 2017) was used to quantify filtered raw reads using pseudo-alignment on the transcriptome reference derived from GRCh38 primary assembly (Genecode version 46), resulting in an average mapping rate of 88%. Differential gene expression analysis was conducted using tximeta version 1.22.1 (Love et al, 2020) and DESeq2 version 1.44.0 (Love et al, 2014). Gene set enrichment assay was conducted using fgsea version 1.33.1 (Korotkevich et al, 2021). Gene-wise Z-scores were calculated from TPM derived from Salmon and imported via tximeta to stabilize scale differences in RNA-seq data. For each gene, the mean is subtracted and divided by the standard deviation across samples, centering and scaling expression values to enable cross-sample comparison.
Graphics
The synopsis image, Figs. 1, 6A, 7A and EV1D, EV3A; Appendix Fig. S1 were created with BioRender.com.
Study approval
Informed consent was obtained from all human subjects (the patients and/or their guardians). The experiments conformed to the principles set out in the WMA Declaration of Helsinki and the Department of Health and Human Services Belmont Report. Human iPSCs were differentiated into β cells with approval of the Ethical Committee of Erasmus Hospital, Université Libre de Bruxelles (P2019/498, A2024/211). Mouse experiments were approved by the Commission d’Ethique et du Bien Être Animal, Faculty of Medicine, Université Libre de Bruxelles, following the European Convention for the Protection of Vertebrate Animals used for Experimental and other Scientific Purposes (European Treaty Series No. 123). The UK Biobank analysis was conducted under UK Biobank project number 103356. All UK participants in this study gave informed consent for genetic studies and were approved by the North Wales ethics committee (no. 17/WA/0327).
Supplementary information
Appendix Peer Review File Dataset EV1 Source data Fig. 2 Source data Fig. 3 Source data Fig. 4 Source data Fig. 5 Source data Fig. 6 Figure EV1 Source Data Figure EV2 Source Data Figure EV3 Source Data Appendix Figure Source Data Expanded View Figures
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1ALFA (Allele Frequency Aggregator) (2024) https://www.ncbi.nlm.nih.gov/snp/docs/gsr/alfa/ [DATASET]
- 2Faul F, Erdfelder E, Buchner A, Lang A-G (2009) Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods 41:1149–116010.3758/BRM.41.4.114919897823 · doi ↗ · pubmed ↗
- 3James NM, Stanford KI (2025) Obesity and exercise: new insights and perspectives. Endocr Rev 46:763–78910.1210/endrev/bnaf 017PMC 1264071740522819 · doi ↗ · pubmed ↗
- 4Korotkevich G, Sukhov V, Budin N, Shpak B, Artyomov MN, Sergushichev A (2021) Fast gene set enrichment analysis. Preprint at bio Rxiv 10.1101/060012
- 5UK Biobank Whole-Genome Sequencing Consortium (2025) Whole-genome sequencing of 490,640 UK Biobank participants. 645:692-70110.1038/s 41586-025-09272-9PMC 1244362640770095 · doi ↗ · pubmed ↗
- 6Virgilio E, Tielens S, Bonfield G, Nian F-S, Sawatani T, Vinci C, Govier M, Montaser H, Lartigue R, Arunagiri A et al (2025) Recessive TMEM 167A variants cause neonatal diabetes, microcephaly and epilepsy syndrome. J Clin Invest 135:e 19575610.1172/JCI 195756 PMC 1261806540924476 · doi ↗ · pubmed ↗