All mouse studies have been approved by the Institutional Animal Care and Use Committee (IACUC) at Weill Cornell Medicine (2011-0024). All cynomolgus macaque work has been approved by IACUC at the University of Pennsylvania under protocol 806688. The pancreatic organs were obtained from the local organ procurement organization under the United Network for Organ Sharing (UNOS). The informed consent was obtained for research purposes.

Cell lines

EndoC-βH1 cells (CVCL_L909; female) were purchased from EndoCells and cultured in DMEM containing 1 g l−1 glucose, 2% BSA fraction V (Sigma-Aldrich), 50 μM 2-mercaptoethanol, 10 mM nicotinamide (Calbiochem), 5.5 μg ml−1 transferrin (Sigma-Aldrich), 6.7 ng ml−1 selenite (Sigma-Aldrich), 100 U ml−1 penicillin and 100 μg ml−1 streptomycin. MIN6 cells (CVCL_0431) were provided by M. Hao (Weill Cornell Medicine) and cultured in DMEM containing 4.5 g l−1 glucose, 15% FBS (Gibco), 50 μM 2-mercaptoethanol, 2 mM glutamine (Gibco) and 100 U ml−1 penicillin and 100 μg ml−1 streptomycin. Cells were maintained at 37 °C with 5% CO2.

To generate NLuc-MIN6 and NLuc-EndoC-βH1 cells, we produced lentivirus expressing proinsulin-luciferase fusion protein from HEK293T cells (ATCC, CRL_3216), transfected with psPAX2 (Addgene, 12260), pMD2.G (Addgene, 12259) and proinsulin-NanoLuc (Addgene, 62057). We pooled viral supernatant collected at 48 h and 72 h post-transfection and concentrated the virus using Lenti-X Concentrator (Takara Bio). We added the concentrated virus to MIN6 or EndoC-βH1 cells in fresh culture medium with 8 μg ml−1 Polybrene (Sigma-Aldrich) and spun the cells at 800g for 1 h at 30 °C during infection. After 24 h, we placed cells in fresh growth media. Subsequently, we treated the infected EndoC-βH1 cells with 5 µg mL−1 blasticidin (Invitrogen) for 1 week to produce the stable NLuc-MIN6 and NLuc-EndoC-βH1 cell lines.

EndoC-βH1 CHEK2-deficient cells (sgCHEK2 EndoC-βH1 cells) were generated by the lenti-CRISPR/Cas9 knockdown system. The sgRNA targets were obtained via the online program generated by Feng Zhang’s laboratory (http://crispr.mit.edu/). The sgRNA target sequences are listed in Supplementary Table 5. The lenti-CRISPR viruses were produced by transfecting the lenti-CRISPR/Cas9 plasmid (Addgene, 52961) along with psPAX2 and pMD2.G. For each 15-cm dish of 293T cells, 15 μg of lenti-CRISPR/Cas9, 9 μg of psPAX2 and 3 μg of the pMD2.G were transfected using polyethylenimine. Tissue culture media were refreshed 16 h after transfection, and media containing viruses were collected 48 h and 72 h after transfection. Viruses were then concentrated using the Lenti-X concentrator (Takara Bio). Infection protocol similar to that of NLuc-MIN6 and NLuc-EndoC-βH1 cells was used for generating EndoC-βH1 CHEK2-deficient cells. Cells infected by lentivirus were then cultured with 4 µg ml−1 puromycin (Invitrogen) for 1 week.

NLuc-EndoC-βH1 CHEK2-deficient cells (sgCHEK2 NLuc-EndoC-βH1 cells) were generated by the lenti-CRISPR/Cas9 knockdown approach, as described above, by infecting NLuc-EndoC-βH1 cells with lentivirus carrying lenti-CRISPR/Cas9 plasmid with sequence either control scramble sgRNA or CHEK2-targeting sgRNA.

For knocking down AZD7762 target genes in EndoC-βH1 cells, we designed two different shRNAs to target each using the Broad Institute GPP Web Portal (https://portals.broadinstitute.org/gpp/public/). The shRNA target sequences are listed in Supplementary Table 8. Each sgRNA was then cloned into pLKO.1-blast (Addgene, 26655) and packaged along with psPAX2 and pMD2.G. The transfection and lentivirus infection protocol used was similar to that used for shRNA knockdown experiment, as described above. Cells infected by lentivirus were then cultured with 5 µg ml−1 blasticidin (Invitrogen) for 1 week.

Overexpression of CHEK2 was achieved by infecting CHEK2-deficient sgCHEK2 EndoC-βH1 with lentivirus carrying control or CHEK2-overexpression construct obtained from VectorBuilder.

Bulk RNA-seq

Total RNA was extracted in TRIzol (Invitrogen) and treated with DNase I using the Directzol RNA Miniprep kit (Zymo Research). RNA-seq libraries of polyadenylated RNA were prepared using the TruSeq RNA Library Prep Kit v2 (Illumina) or TruSeq Stranded mRNA Library Prep Kit (Illumina). cDNA libraries were sequenced with pair-end 51 bps using an Illumina NovaSeq6000 platform. The resulting reads were checked for quality using FastQC (v0.10.1, https://www.bioinformatics.babraham.ac.uk/projects/fastqc) and were trimmed for adaptor sequences and low-quality bases using cutadapt (v1.18). To measure gene expression, the trimmed reads were aligned to the human reference genome (GRCh37). Raw gene counts were quantified using HTSeq-count (v0.11.2). The counts data were subjected to a regularized logarithm transformation using the rlog function within the DESeq2 package (v1.36.0). The transformed data were used to perform a principal component analysis (PCA) using the plotPCA function within the DESeq2 package. Additionally, the counts data were converted into fragment counts normalized per kilobase of feature-length per million mapped fragments (FPKM) using the fpkm function within the DESeq2 package, and an unsupervised hierarchical clustering on samples was conducted using the Pearson correlation coefficient metric. The R heatmap package (v1.0.12) was used to visualize the clustering result.

Quantitative RT–PCR

To validate shRNA knockdown efficiency, we measured gene expression levels in control and knock-downed cells using quantitative RT–PCR. We isolated total RNA from EndoC-βH1 cells using the RNeasy Plus Universal Kit (Qiagen) and synthesized cDNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). We used a LightCycler 480 SYBR Green I Master System (Roche) in the quantitative RT–PCR experiments. We normalized the expression of target genes against Β actin. Primers used for quantitative RT–PCR experiments can be found in Supplementary Table 9.

Insulin tolerance test

Before the insulin tolerance test, mice were fasted for 6 h. During the experiment, mice were injected IP with 1 IU kg−1 body weight of insulin and blood glucose levels were monitored at 0, 15, 30, 45, 60 and 120 min after insulin injection.

Immunostaining

For immunostaining of human islets and mouse pancreatic section, tissue samples were fixed overnight in 4% paraformaldehyde (PFA) at 4 °C and incubated in primary antibody overnight at 4 °C. Samples were then washed with 1× phosphate-buffered saline (PBS) three times and then incubated in secondary antibodies and DAPI for 1 h at room temperature. Samples were then washed with 1× PBS three times. Immunostained tissue was then mounted in ProLong Gold Antifade Mountant (Molecular Probes). The primary antibodies and their respective concentration used in immunostaining experiments were as follows: goat anti-insulin antibody (Dako, Agilent, IR002; 1:50), sheep anti-NGN3 antibody (R&D Systems, AF3444; 1:200), rat anti-SST antibody (R&D Systems, MAB2358; 1:100), rabbit anti-glucagon antibody (Cell Signaling, 2760; 1:200), goat anti-polypeptide Y antibody (Novus Biologicals, NB-100-1793; 1:200) and rabbit anti-Ki67 antibody (Abcam, ab15580; 1:500). All secondary antibodies used are fluorescence-conjugated secondary antibodies (Alexa Fluor, Thermo Fisher Scientific). Microscopy images were obtained from Inverted Microscope/Apotome (Zeiss) and LSM 800 confocal microscope (Zeiss) and analyzed using ImageJ.

Flow cytometry

EndoC-H1 or MIN6 cells were washed with PBS and resuspended in 300 μl PBS. In total, 5–10 μl of PI staining solution (Invitrogen, P3566), diluted to 10 μg ml−1, was added to control tubes and treated samples. Samples were mixed gently and incubated for 1 min in the dark. PI fluorescence was determined using an Attune Instrument in the YL-2 channel.

PP2A activity

sgCHEK2 and control EndoC-βH1 cells were starved in 0.5 mM glucose for 1 h and then stimulated with 20 mM glucose for 1 h. Cells were then lysed to assess PP2A activity with the PP2A Immunoprecipitation Phosphatase Assay Kit (MilliporeSigma).

6-Phosphogluconate levels

MIN6 cells were starved in 0 mM glucose for 1 h and then in 0 mM glucose for an additional 1 h. Then they were stimulated with either 0 mM or 20 mM glucose for 30 min. AZD7762 or control was included starting from the second starvation and throughout the rest of the experiments. Cells were lysed with Dounce homogenizer, and 6-phosphogluconate levels were assessed by 6-phosphogluconic acid Assay Kit (BioVision).

H&E staining protocol

Slides were processed as follows for H&E staining: first, they were washed in PBS for 1 min and dipped in water once, then immersed in hematoxylin for 1 min and rinsed twice in water. After that, the slides were dipped in lithium carbonate once, washed again in water and then dipped in eosin. Next, the slides were dipped in 95% EtOH twice, followed by 100% EtOH twice, and then dipped two times in histoclear. Finally, the tissue sections on the slides were mounted using Cytoseal (Thermo Fisher Scientific).

Focused chemical screen

A total of 2 × 105 of MIN6 cells were cultured for 4 d and were starved in 0 mM glucose Krebs-Ringer Bicarbonate (KRBH) buffer for 1 h and then followed by an additional hour of starvation including the compounds. The source and purity of chemicals used in the screen are included in Supplementary Table 2. After the initial starvation, cells were treated with 20 mM glucose KRBH buffer for 30 min before 10 µl of supernatant was collected for assessing the luminescence levels using the Promega Nano-Glo Luciferase Assay System (N1120). The 96-well assay plate was then read by the Biotek Synergy H1 microplate reader.

Pancreatic islets isolation

The islets were isolated in the Human Islet Core at the University of Pennsylvania following the guidelines of the Clinical Islet Transplantation Consortium protocol. Gift of Life as well as any other organ procurement organization who recovers the organs obtain consent from the deceased donor’s family. The collected organs could be used for research. There is no compensation for participants. All procedures are in compliance with the University of Pennsylvania IRB, Gift of Life leadership team and UNOS. Our research complies with all regulations and standards by the University of Pennsylvania Institutional Review Board, which is responsible for approval of the protocol. Cynomolgus macaque islet isolation was performed based on a modified protocol using Liberase enzyme (Roche). The islets were purified from the digested pancreas using a three-layer, discontinuous Euro-Ficoll gradient (densities 1.108, 1.096 and 1.037) and a COBE blood cell processor (COBE Laboratories). Samples were collected from different layers after islet purification to assess the purity of cell isolation. Final samples were stained with dithizone, counted manually and sized using a formula to calculate islet number and islet equivalents based on a 150-mm diameter. Islet preparations with purity >85% were used for this study. The isolated islets were cultured overnight in CRML 1066 (Mediatech) containing 10% heat-inactivated FBS at 25–28 °C in 95% O2 and 5% CO2. Mouse islets were isolated following a previously published protocol60.

Mouse models

C57BL/6J (stock 000664), B6.Cg-Lepob/J (commonly referred to as ob/ob, stock 000632), mice were obtained from the Jackson Laboratory. C57BL/6 N-Atm1BrdCHEK2tm1b(EUCOMM)Hmgu/JMmucd (stock: 047090-UCD, also known as Chk2) mice were acquired from the Knockout Mouse Project repository at the Jackson Laboratory. CD-1/ICR mice (stock 022) were obtained from the Charles River Laboratories. For HFD studies, 8-week-old C57BL/6J mice were fed with 60% HFD (Research Diets, D12492i) for 4 months. For Chk2−/− mice HFD studies, 8-week-old wild-type and Chk2−/− mice were fed with 60% HFD (Research Diets, D12492i) for 6 months. Unless stated otherwise, all mice were maintained on a normal chow diet and a 12-h light/12-h dark cycle in a pathogen-free animal facility, where ambient temperature was consistently maintained at 25 °C, with humidity levels ranging between 30% and 70%.

For in vivo studies with AZD7762, AZD7762 is dissolved in 11.3% (2-hydroxypropyl)-β-cyclodextrin. All animals were fasted overnight and treated with AZD7762 IP 1 h before the experiments. For mouse GTT and GSIS, 12 or 25 mg kg−1 AZD7762 was given by IP injection, and then 0.5–2 g kg−1 glucose was given either by IP injection or orally. Glucose levels were measured at −60, 0, 15, 30, 60, 90 and 120 min postglucose. For GSIS experiments, blood was drawn from the tail vein at −60, 0, 15 and 30 min postglucose. For HOMA-IR calculation for vehicle- and AZD7762-treated CD-1/ICR, HFD-fed C57BL/6J mice and ob/ob mice, we used the method as described in ref. 61.

Cynomolgus macaque models

Cynomolgus macaque experiments are approved under the animal Protocol title—Cellular approaches for the modulation of alloresponses in nonhuman primates (protocol 806688). Adult Mauritius-origin male cynomolgus macaques (Macaca fascicularis) were provided by Alpha Genesis. After overnight fasting, the animal was sedated with ketamine (3–4 mg kg−1) mixed with dexmedetomidine (0.15 mg kg−1 intramuscular). Baseline blood samples were collected before the IV infusion of vehicle or 1.55 mg kg−1 AZD. Blood glucose was monitored at −20 and −40 min after infusion. At time 0 min, glucose (0.5 g kg−1 body weight) was infused IV via the IV catheter. Blood glucose was analyzed using a bedside glucometer (whole blood), and serum was tested for insulin/C-peptide levels. A small blood drop (~0.3 µl) was produced by a pinprick (using either a lancet or a needle) at 1, 3, 5, 7, 10, 15, 20, 30 and 60 min after administration of glucose. Additional blood samples (0.5 ml) were collected at 0, 1, 3, 5, 10, 15 and 30 min to measure insulin/C-peptide levels.

Insulin secretion assays

For human, mouse and cynomolgus macaque islets experiment, islets were starved in 2 mM glucose KRBH buffer for 2 h at 37 °C, and then stimulated in 2 mM glucose KRBH buffer for 1 h and subsequently with 20 mM glucose KRBH buffer for 1 h. To measure the total level of insulin in samples, cells were lysed in 0.1% Triton X-100. MIN6 cells were starved in 0 mM glucose for 1 h and then in 0 mM glucose for an additional 1 h. Then the cells were stimulated with either 0 mM or 20 mM glucose for 30 min. EndoC-βH1 cells were starved in 0.5 mM glucose for 1 h and then stimulated with either 0.5 mM or 20 mM glucose for 1 h. AZD7762 was included during the starvation and glucose stimulation steps. For MIN6 experiments, AZD7762 was included starting from the second starvation and throughout the rest of the experiments. For pseudoislet experiments, 4,000 EndoC-βH1 cells or 2,000 dissociated human islet cells were aggregated in v-bottom plate in culture media for 4 d before GSIS. A total of 50 mM KCl was used for KCl-stimulated insulin secretion experiments. At the end of each stimulation, 100 µl buffer was collected to assess insulin or C-peptide levels with ELISA kits (Alpco and Novus Biologicals). The islet perifusion experiment was carried out in a BioRep Perifusion System. During perifusion dynamic GSIS experiments, islets or pseudoislets were perfused at 100 µl min−1 of KRBH buffer containing 2 mM glucose for 30 min, then 45 min with 20 mM glucose and 15 min with 2 mM glucose again, and finally with 50 mM KCl KRBH buffer for 15 min. Samples were collected for assessment of insulin and C-peptide levels every 90 s. Data were normalized to baseline insulin secretion at 2 mM glucose for human islets and 0.5 mM for EndoC-βH1 cells.

Hormones measurement

Proinsulin levels were measured using Total Proinsulin Chemiluminescence ELISA Kit (Alpco). Human insulin levels were measured by Insulin Chemiluminescence ELISA Kit (Alpco). Human C-peptide levels were measured by C-Peptide Chemiluminescence ELISA Kit (Alpco). Mouse insulin levels were measured by Rodent Insulin Chemiluminescence ELISA Kit (Alpco). Glucagon levels were measured by Glucagon ELISA (Alpco). Total GLP-1 ELISA (7–36 and 9–36; Alpco) and active GLP-1 levels were measured by Active GLP-1 (7–36) Amide Chemiluminescence ELISA (Alpco). Total GIP levels were measured by Total GIP ELISA (Alpco). SST levels were measured by SST (Human, Rat, Mouse, Porcine)—ELISA Kit (Phoenix Pharmaceuticals).

Western blot

Protein was extracted from EndoC-βH1 cells in radioimmunoprecipitation assay buffer (Sigma-Aldrich) supplemented with halt protease and phosphatase inhibitor Cocktail (Thermo Fisher Scientific). Protein samples were loaded onto NuPAGE 4–12% Bis–Tris Protein Gels (Thermo Fisher Scientific), resolved by electrophoresis and transferred onto nitrocellulose membranes. Membranes were incubated with the following primary antibodies: mouse monoclonal anti-β-actin antibody (Invitrogen, MA1-140; 1:20,000), rabbit monoclonal anti-phospho CHEK2 T68 antibody (Cell Signaling, 2197S; 1:1,000), mouse monoclonal anti-CHEK2 antibody (Cell Signaling, 3440T; 1:1,000), rabbit anti-eIF2α antibody (Cell Signaling, 5324; 1:1,000), rabbit anti-phospho eIF2α Ser51 antibody (Cell Signaling, 3597; 1:1,000), rabbit anti-α-tubulin antibody (Cell Signaling, 2144; 1:1,000), rabbit anti-phospho PLK1 T210 (Abcam, ab155095; 1:200) and mouse anti-PLK1 (Abcam, ab17056; 1:1,000). Primary antibodies were detected by fluorophore-conjugated secondary goat anti-rabbit (LI-COR IRDye 800CW, 926-32213; 1:15,000) and donkey anti-mouse (LI-COR IRDye 680RD, 926-32210; 1:15,000) using LI-COR Odyssey Imagers. Western blot images were analyzed with Image Studio software 5.2.5.

Cytosolic NADPH/NADP ratio and G6PD activities

Twenty million MIN6 cells were cultured for 4 d and lysed in 300 µl homogenization buffer. Cytosolic fraction was separated from the mitochondria fraction with a mitochondria extraction kit (Novus Biologicals) and used directly for measuring NADPH/NADP levels (Promega). For experiments measuring G6PD activities, 2 million MIN6 cells were cultured for 4 d and lysed in 300 µl PBS. G6PD levels were measured with a G6PD Activity Assay Kit (Abcam, Fluorometric).

Untargeted metabolomics

MIN6 cells were starved in 0 mM glucose for 1 h and then in 0 mM glucose for an additional 1 h. Then they were stimulated with either 0 mM or 20 mM glucose for 30 min in the presence of 0 or 10 µM of AZD7762 throughout the experiment. The samples were homogenized with 300 µl of methanol:water (4:1, vol/vol), containing three internal standards, 1 nM U13_succinate, 1 nM U13_citrate and 0.5 nM heptadecanoic acid. Then, the samples were quickly frozen on liquid nitrogen and thawed on ice. The thawed samples were sonicated for 2 min. The freeze-thaw-sonication procedure was repeated three times. After that, the samples were kept at −20 °C for 10 min and then centrifuged at 13,500g for 10 min. Then, 250 µl of supernatant was transferred to a sampling vial. The reminder supernatant from each sample was put together to make the pooled quality control (QC) sample. The samples were dried under gentle nitrogen flow and derivatized with a two-step derivatization procedure. First, the samples were methoximized with 50 µl of methoxyamine hydrochloride (15 mg ml−1 in pyridine) at 30 °C for 90 min. The silylation step was done with 50 µl of N,O-bis(trimethylsilyl)trifluoroacetamide (containing 1% trimethylchlorosilane) at 70 °C for 60 min. QC sample was run multiple times during the analysis. The samples were analyzed by gas chromatography time-of-flight mass spectrometry (GC-TOFMS Premier, Waters). A number of 491 variables were detected after alignment and excluded any known artificial peaks. The data were normalized to the intensity of the sum of all the metabolites. The dataset was then imported into SIMCA-p software for PCA and PLS-DA. Variable importance in the projection (VIP) values from PLS-DA model were used for metabolites selection (VIP > 1). Heatmap was generated with R Studios 2021.09.0 with R 4.1.2 and pheatmap 1.0.12.

Kinase profiling

The kinase profiling was performed using the SelectScreen Biochemical Kinase Profiling Service by Thermo Fisher Scientific at a concentration of 1 µM AZD7762.

Statistics and reproducibility

No statistical methods and power analysis were used to predetermine sample sizes, but our sample sizes are similar to those reported previously6. For in vitro studies, n = 3 independent replicates or 3 individual participants or donors were used for all experiments unless stated otherwise. All experiments were independently repeated at least three times with similar results. Data are shown as mean ± s.e.m. For two-group data, we used a two-tailed unpaired Student’s t-test. For one independent variable data, we used one-way analysis of variance (ANOVA, Dunnett’s test). For experiments in Fig. 5 that involved two control and two experimental groups, we used mixed ANOVA for statistical analysis. Statistical analysis was performed using GraphPad Prism 9 software. Data distribution was assumed to be normal, but this was not formally tested. Details for mouse islets immunostaining analysis are included in Supplementary Table 10.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.