GLP-1R agonists demonstrate potential to treat Wolfram syndrome in human preclinical models

Aims/hypothesis Wolfram syndrome is a rare autosomal recessive disorder caused by pathogenic variants in the WFS1 gene. It is characterised by insulin-dependent diabetes mellitus, optic nerve atrophy, diabetes insipidus, hearing loss and neurodegeneration. Considering the unmet treatment need for this orphan disease, this study aimed to evaluate the therapeutic potential of glucagon-like peptide 1 receptor (GLP-1R) agonists under wolframin (WFS1) deficiency with a particular focus on human beta cells and neurons. Methods The effect of the GLP-1R agonists dulaglutide and exenatide was examined in Wfs1 knockout mice and in an array of human preclinical models of Wolfram syndrome, including WFS1-deficient human beta cells, human induced pluripotent stem cell (iPSC)-derived beta-like cells and neurons from control individuals and individuals affected by Wolfram syndrome, and humanised mice. Results Our study shows that the long-lasting GLP-1R agonist dulaglutide reverses impaired glucose tolerance in WFS1-deficient mice, and that exenatide and dulaglutide improve beta cell function and prevent apoptosis in different human WFS1-deficient models including iPSC-derived beta cells from people with Wolfram syndrome. Exenatide improved mitochondrial function, reduced oxidative stress and prevented apoptosis in Wolfram syndrome iPSC-derived neural precursors and cerebellar neurons. Conclusions/interpretation Our study provides novel evidence for the beneficial effect of GLP-1R agonists on WFS1-deficient human pancreatic beta cells and neurons, suggesting that these drugs may be considered as a treatment for individuals with Wolfram syndrome. Graphical abstract Supplementary Information The online version of this article (10.1007/s00125-023-05905-8) contains peer-reviewed but unedited supplementary material.

3 this difference with p<0.05 and a power of 85%. 48 h after dulaglutide or saline injection, IPGTT and ITT [4] were done by a researcher blinded for treatment. Data analysis was not blinded. For the IPGTT, glucose (2 mg/g body weight) was injected IP after 6 h fasting.
Tail vein glycaemia was measured by Accu-Chek Aviva Nano (Roche, Switzerland) at times 0, 15, 30, 60, 90 and 120 minutes, and plasma insulin was measured by ultrasensitive mouse insulin ELISA (Crystal Chem, USA). Glucose tolerance was calculated as area under the curve (AUC) of IPGTT glycemia from 0 to 120 minutes. For the ITT, human recombinant insulin (Actrapid, Novo Nordisk, Danemark) was injected IP in 4 h-fasted mice. Due to gender differences in insulin sensitivity Actrapid was used at 0.75 mU/g body weight in males, and 0.4 mU/g in females. Tail vein glycaemia was measured at times 0, 15, 30, 60, 90 and 120 minutes. Insulin sensitivity was calculated as area over the curve (AOC) of ITT glycaemia from 0 to 30 minutes. The insulinogenic index was calculated as incremental insulin (ΔI) over incremental glucose (ΔG) in the first 15 minutes of IPGTT. Beta cell function was calculated as insulinogenic index × insulin sensitivity (ITT AOC) [4].

EndoC-βH1, human islet cells and iPSC lines
The EndoC-βH1 cells were free of mycoplasma contamination as determined by the MycoAlert Mycoplasma Detection Kit (Lonza, Switzerland).
iPSCs were differentiated into beta-like cell aggregates using pour previously validated 7-stage protocol. Differentiation efficiency was assessed by analyzing key differentiation markers by real-time PCR and immunofluorescence [4,8,9], using primers listed in ESM Tables 4 and 5, and antibodies listed in ESM Table 6.
For iPSCs differentiation into cerebellar neuron-like cells, iPSCs were seeded in Matrigelcoated 6-well plates at 30-40% confluency and cultured in a 1:1 mixture of DMEM/F12 5 and neurobasal media (Gibco, USA) supplemented with additives along the 12-day differentiation (ESM Table 7). At day 12, cells were split using Accutase (Capricorn Scientific, Germany) and plated at low density (3x10 6 cells) in Matrigel-coated 10-cm plates in neural precursor cell (NPC) medium (ESM Table 8) until the appearance of neural stem cells (NSCs) forming neural rosettes (day 2 to 10). Rosettes were manually picked, transferred to new Matrigel-coated plates. The NSCs were then matured into NPCs and amplified in the same medium. NPCs expressed neural precursor markers PAX6, vimentin and nestin and were negative for the pluripotency marker OCT4. After 5 passages, the cells were either frozen, plated for experiments at cell densities described in ESM Table 9, or plated for cerebellar neuron differentiation. For the latter, NPCs were split using Accutase and plated at 80% confluency in NPC medium (overnight) in Matrigelcoated 6-well plates. Then, cells were grown for 10 days in second step differentiation medium (ESM Table 10), changed every second day. On day 11 cells were split using Accutase, plated in Matrigel-coated plates (cell densities described in ESM Table 11) and cultured for an additional 10 days with medium changes every second day. The resulting immature cerebellar neurons were positive for cerebellar markers KIRRE2, ZIC 1 and calbindin, neuronal marker b-tubulin III, and postsynaptic marker synaptophysin [10,11] and negative for astrocyte marker GFAP.
Cell death was assessed by fluorescence microscopy counting after staining with DNA binding dyes Hoechst 33342 (5 µg/ml, Sigma-Aldrich) and propidium iodide (5 µg/ml, Sigma-Aldrich). At least 600 cells were counted by two observers, one of them unaware of experimental conditions. Early and late apoptosis was also assessed by Real time-Glo AnnexinV apoptosis and necrosis assay (Promega, USA). The data is expressed as Fold of Basal (T0) apoptosis in each condition.

iPSC-beta cell transplantation into immunodeficient mice
iPSC-derived beta cellaggregates were transplanted into 5-7-week-old male Rag2 KO mice. Mice were anesthetized by IP injection of ketamine (100 mg/kg, Nimatek, Dechra, UK) and xylazine (5 mg/kg, Rompun, Bayer, Germany), and aggregates were transplanted under the kidney capsule using a 10 μl precision pipet. Paracetamol (100 mg/l drinking water) was given as analgesic one day prior to and during the 10 days following the surgery. Animals were monitored daily. 1000 aggregates from one 7 differentiation were transplanted into two mice. IPGTT was performed 7 and 14 weeks after transplantation, and plasma C-peptide measured by ultrasensitive human C-peptide ELISA (Mercodia, Sweden). Fourteen weeks after transplantation, mice were allocated to IP dulaglutide (1 mg/kg every 4 days) or vehicle injection for 12 weeks by simple randomization.
To assess grafted iPSC-beta cell function, the kidney was perfused (1 ml/min flow rate) in situ at 37°C in a single-pass circuit through the renal artery. The abdominal aorta was ligated above the coeliac trunk, a catheter inserted, and the venous effluent collected by another catheter inserted in the renal vein as previously described [13]. To avoid coagulation, the kidney was first perfused with 1 ml heparinized (50 U/ml) PBS. After 20 minutes equilibration in basal perifusion solution, the kidney was sequentially perfused with solution containing 0 (G0) or 20 mmol/l glucose (G20) alone or combined with forskolin (1 μmol/l), gliclazide (25 μmol/l), diazoxide (250 μmol/l) or KCl (30 mmol/l).
Samples were collected every 4 minutes and human insulin was measured by radioimmunoassay [14].

Immunofluorescence, Western blott, real-time PCR and ROS detection
For immunostaining cells were fixed in 4% formaldehyde for 15-20 minutes, permeabilized with 0.5% Triton X-100 for 10 minutes, blocked with UltraV block (Thermo Formalin-fixed paraffin-embedded mouse pancreas and iPSC-beta cell grafts were cut using microtome into 5-µm-thick sections. Sections were de-paraffinized with xylene and re-hydrated through descending graded alcohols to water. Antigens were retrieved by immersing slides in 10 mmol/l sodium citrate buffer pH 6 and microwaving until boiling. Samples were blocked and processed as described above. Antibodies and dilutions are provided in ESM Table 5. Pancreatic area was measured using ImageJ software (https://imagej.nih.gov/ij/index.html).
Proteins were transferred to nitrocellulose membranes, blocked with 5% nonfat milk, and incubated overnight with primary antibodies diluted in the same solution. The membranes were washed with PBS 0.1% Tween-20 and incubated with horseradish peroxidaseconjugated secondary antibodies (ESM Table 6). After washing, proteins were detected using SuperSignal West Femto chemiluminescent substrate (Thermo Fisher Scientific) in ChemiDoc XRS+ system (Bio-Rad) and quantified using Image Lab software.
mRNA was isolated using Dynabeads mRNA DIRECT (Thermo Fischer Scientific) and reverse transcribed. Gene expression was assessed by real-time PCR using Q SYBR 9 Green Supermix on a MyiQ2 single-color real-time PCR system (Bio-Rad). Gene expression was calculated as copies/μl using a standard curve (prepared in conventional PCR). GAPDH and/or beta actin (ACTB) were used as reference genes. Primer sequences are provided in ESM Table 4 and 5.
Cells seeded in black-bottom plates were incubated with the oxidation-sensitive fluorescent probe HPF (10 μmol/l, Invitrogen, USA) for 30 minutes and washed twice with PBS. Fluorescence was measured in VICTOR multilabel plate reader (Perkin-Elmer, USA) using excitation and emission spectra of 485 and 535 nm. Menadione (15 µmol/l for 2 h) was used as positive control.

Mitochondrial respiration
Mitochondrial function was assessed by measuring oxygen consumption rate (OCR) in a Seahorse XFp Extracellular Flux Analyzer (Agilent, USA). Dispersed iPSC-derived beta cells were exposed or not to exenatide or forskolin for 72 h and then preincubated for 1 h at 37°C in a non-CO2 incubator in KREBS buffer (20 mmol/l Hepes, 2.5 mmol/l CaCl2, 1.16 mmol/l MgS04, 1.2 mmol/l KH2PO4, 4.7 mmol/l KCl and 114 mmol/l NaCl) at pH 7.4 and with 0.2% FFA-free BSA. Mitochondrial respiration was measured basally and after sequential injection of 20 mM glucose, ATP synthase inhibitor oligomycin (5 μmol/l), oxidative phosphorylation uncoupler carbonyl cyanide-ptrifluoromethoxyphenylhydrazone (FCCP, 4 μmol/l), and electron transport chain inhibitors rotenone and antimycin (1 μmol/l). NPCs and cerebellar neurons were similarly exposed to exenatide/forskolin and preincubated in Agilent Seahorse XF DMEM medium pH 7.4 supplemented with 10 mmol/l glucose and 2 mmol/l glutaMAX. Mitochondrial respiration was measured basally and after injection of 2 μmol/l oligomycin, 2 μmol/l FCCP, and 1 μmol/l rotenone plus antimycin A. OCR data were normalized to the last basal reading in each sample.

ESM Tables
ESM Table 1 Fig. 3. Exenatide does not alleviate ER stress in WFS1-silenced EndoC-bH1 cells. EndoC-bH1 cells were transfected with control siRNA (siCT) or siRNA targeting WFS1 (siWFS1#1). 72h after transfection cells were exposed or not (CT) to the ER stressor tunicamycin (Tuni, 5 µg/ml) alone or in combination with 50 nmol/l exenatide (Ex), 50 nmol/l dulaglutide (Du) or 20 μmol/l forskolin (Fk) for 8h (a-e) or 16h (f-i). mRNA expression of WFS1 and the ER stress markers CHOP, BIP, ATF3 and XBP1s was measured by real-time PCR, normalized to the reference gene ACTB and expressed as fold of the highest value in each experiment. Extremities of floating bars are maximal and minimal values; horizontal line shows median. Individual points represent independent experiments. *p<0.05, **p<0.01, ***p<0.001 siWFS1 vs siCT; †p<0.05, † † †p<0.001 treated vs CT in siWFS1 cells; ‡ ‡p<0.01, ‡ ‡ ‡p<0.01 treated vs Ct in siCT cells; §p<0.05 Tuni + Ex vs Tuni by two-way ANOVA with Sidak's or Tukey's tests for multiple comparisons.   Fig. 6. ER stress markers and mitochondrial function in control and WFS1deficient iPSC-derived beta cells and EndoC-βH1 cells. The WFS1-deficient iPSC line Wolf-2010-9.4 and its isogenic control Wolf-9.4-Corr-2G6.1 were differentiated into pancreatic beta cells. At the end of the differentiation, cells were cultured for 24h with or without exenatide (Ex) or forskolin (Fk), and then exposed or not (CT) for 48h to tunicamycin (Tu) alone or combined with exenatide or forskolin. (a-c) Expression of the ER stress markers BIP, CHOP and ATF3 was assessed by real-time PCR, normalized to the reference gene ACTB, and expressed as fold of the highest value in each experiment. Individual data points represent independent iPSC differentiations, identified with a different color. Extremities of floating bars are maximal and minimal values; horizontal line shows median. (d-g) Mitochondrial function assessed by Seahorse in control and WFS1-deficient iPSC-derived dispersed beta cell aggregates (d-e), and control and WFS1-silenced EndoC-βH1 cells (f-g) exposed or not (CT) for 72h to exenatide or forskolin. Oxygen consumption rate (OCR) in isogenic control (d), Wolf-2010-9.4 (e), control EndoC-βH1 cells (f) and WFS1-silenced EndoC-βH1 cells (g), is expressed as fold change of the last basal reading in each sample, n=3-5 per group. ‡ ‡ ‡ p<0.001 treated vs CT in control cells; † † † p<0.001 treated vs CT in WFS1-deficient cells, §p<0.05, § §p<0.01 Tu vs Tu FK, by two-way ANOVA with Sidak's or Tukey's tests for multiple comparisons.  Fig. 7. Impact of dulaglutide on beta and alpha cell proportions in Wolfram syndrome iPSC-derived grafts. At the end of the experiment described in Fig. 6 and after kidney perifusion, grafts were fixed in formalin and the tissue used for histological analysis. (a) Representative immunofluorescence images of control and WFS1 iPSC-derived grafts retrieved from mice injected for 12 weeks with saline (Veh) or dulaglutide (Dula). Insulin staining is shown in green and glucagon in red; human nuclei in pink (used to differentiate human graft from mouse tissue), and nuclei in blue (Dapi). Scale bars 50 μm. (b-e) Quantification of beta and alpha cell proportion in the graft (assessed by manual counting). (b and d) Each symbol represents the beta and alpha cell proportion in one picture. Values from pictures from the same mouse are stacked in columns. (c and e) Average beta or alpha cell proportion in each graft. Symbols with a same color in two different mice indicate that these animals were transplanted with aggregates from the same differentiation. Extremities of floating bars are maximal and minimal values; horizontal line shows median. § Dula vs Veh by paired ttest.