Establishment of Pancreatic β-Cell–Specific Gene Knockout System Based on CRISPR-Cas9 Technology With AAV8-Mediated gRNA Delivery

The generation of knockout (KO) mice has significantly contributed to understanding the physiological relevance of genes of interest since the homologous recombination in embryonic stem cells was applied to gene modification (1). The Cre-loxP system, in which a particular gene of interest is flanked by loxP sequence and deleted under Cre recombinase, provides a powerful tool to investigate the roles of the genes playing essential functions in an early developmental stage (2). Also, the Cre-loxP system has contributed significantly to examining the roles of genes in a cell-autonomous manner. This system has been applied for pancreatic β-cell research. For example, the role of a canonical autophagic regulator, Atg7, was clarified by crossbreeding Atg7^flox/flox mice with those expressing Cre recombinase under the rat insulin promoter (3,4), as the whole-body Atg7 KO mice die soon after birth (5). However, establishing the organ-targeted conditional KO mice is time consuming and costly because we must prepare transgenic mice harboring floxed genes of interest and those expressing Cre recombinase under specific promoters. The cluster regularly interspaced short palindromic repeats (CRISPR)-Cas9 system is a helpful method for modifying genes in mammalian cells (6), and it has already been applied to efficient gene editing in mice (7). To accomplish gene editing in vivo, it is crucial to choose the appropriate method for gene delivery, and the adeno-associated virus (AAV) vector is one of the most frequently applied because of its safety and high delivery efficiency (8). For instance, Ramzy et al. (9) reported that they expressed AAV8-mediated Cre recombinase under the insulin 1 (Ins1) promoter in Ins1^−/−;Ins2^flox/flox mice, thus successfully generating Ins1 and Ins2 double-KO mice showing hyperglycemia. Based on those findings, combining the CRISPR-Cas9 system with AAV-mediated gene delivery would be helpful and time-saving for knocking out genes, specifically in pancreatic β-cells.

In this study, we generated transgenic mice expressing Cas9 in pancreatic β-cells and by delivering guide RNA (gRNA) in AAV8, established pancreatic β-cell–specific gene KO mice. After we estimated gene delivery and KO efficiency, we knocked out the essential gene to maintain the identity and function of the β-cell pancreatic and duodenal homeobox 1 (Pdx1) as a proof of concept. This report is the first to describe pancreatic β-cell–specific KO mice with AAV-mediated gene delivery by investigating the efficacy of gene KO and its influence on glucose tolerance and cellular characteristics.

All mice were mated with C57BL/6J background mice, kept in specific pathogen-free barrier facilities, maintained in a 12-h light/dark cycle, and fed standard food. The animal care and use committee of Juntendo University reviewed and approved the study protocol. Cre-dependent Cas9 knock-in mice were generated at the Broad Institute of MIT and Harvard (Cambridge, MA) and purchased from The Jackson Laboratory (#026175; Bar Harbor, ME) (10). Ins1-Cre mice were obtained from RIKEN BioResource Center (#RBRC03934; Ibaraki, Japan) (11). For the glucose tolerance test, all mice underwent an overnight fast. The following day, 2 g/kg body weight of 10% d-glucose was given by intraperitoneal injection, after which blood glucose level was measured by the glucose analyzer Glutest Mint (Sanwa Kagaku Co., Ltd, Nagoya, Japan). Plasma insulin levels were measured by ELISA kit (#M1104; Morinaga Institute of Biological Science, Yokohama, Japan). Genotyping was performed as described previously (10,11).

Pancreatic islets were isolated by collagenase digestion and collected under a microscope as previously described (12). Briefly, the isolated islets were collected in 1.5-mL tubes and incubated at 37°C for 3 min with tapping every 30 s after adding 500 μL trypsin/EDTA. Five hundred microliters of FBS was added to halt trypsinization and then transferred to filtered tubes. Next, if necessary, they were applied with anti-insulin antibody (1:100, #C27C9; Cell Signaling Technology, Danvers, MA) diluted in PBS and incubated at 4°C for 60 min, followed by incubation with an appropriate Fluor-labeled secondary antibody for 60 min. Flow cytometric analysis was performed with LSR Fortessa (Becton, Dickinson and Company, Franklin Lakes, NJ) following DAPI staining. Data acquisition and analysis were performed using FlowJo version 10.8.1.

MIN6 cells, provided by Dr. Junichi Miyazaki (Osaka University), were cultured in DMEM (#11965-092; Gibco, Billings, MT) with 10% FBS (#SH30910.03; Cytiva, Marlborough, MA), 1% penicillin-streptomycin (#09367-34; Nacalai Tesque, Kyoto, Japan), and 0.0005% β-mercaptoethanol in a 5% CO₂ incubator at 37°C (13). Pdx1 KO MIN6 cells were established by lentiviral transducing gRNA, targeted to mouse Pdx1, into an MIN6 clone efficiently expressing Cas9 (12). pKLV2-U6-gPdx1-PGK-PuroR-BFP was created by replacing the gRNA sequence of pKLV2-U6-gRNA-PGK-PuroR-BFP, a gift from Dr. Kosuke Yusa (#67974; Addgene, Watertown, MA) (14). The target sequence for mouse Pdx1 is 5′-TCACGCGTGGAAAGGCCAGTGGG-3′ (15).

HEK293 cells were seeded at 4.0 × 10⁶ cells on a 10-cm plate 1 day before transfection. The cells were transfected with 10 μg of a vector, 5 μg of a transfer plasmid (psPAX2, a gift from Didier Trono, Swiss Federal Institute of Technology in Lausanne) (#12260; Addgene), and 5 μg of a VSV-G vector (pMD2.G, a gift from Didier Trono) (#12259; Addgene) using the Lipofectamine 3000 transfection reagent (Thermo Fisher Scientific, Waltham, MA). Viral particles were harvested 48 h posttransfection and purified using a 0.45-μm Millex-HV filter (#SLZHVR33RS; Merck Millipore, Burlington, MA). Viral transduction was performed by treating MIN6 cells with the filtered media overnight, and the cells were selected with antibiotics, such as puromycin (1 μg/mL) or blasticidin (10 μg/mL), for several days.

AAV8-gRNA vectors were synthesized at VectorBuilder Inc. (Chicago, IL). The target sequence for EGFP is 5′-GAAGTTCGAGGGCGACACCCGGG-3′ (14). The identical sequence for targeting mouse Pdx1 in the lentiviral vector was applied for AAV8-gPdx1. All self-complementary AAV8 (scAAV8) viruses, synthesized as ultrapurified custom (minimum titer >10¹³ genomic copies/mL), were purchased from VectorBuilder Inc. The AAV8, diluted in 300 μL saline for 25 μL of the virus at room temperature, was injected intraperitoneally into the abdominal cavity of unanesthetized mice at 7 weeks of age. In contrast, the same volume of saline was injected as control when needed.

Dispersed islet cells were spread on poly-L-lysine–coated glass bottom dishes and washed with KRB buffer. Then, 5 μmol/L fura-2 acetoxymethyl est er (Dojindo, Kumamoto, Japan) and 0.05% Pluronic diluted with KRB buffer were added to the dish and incubated for 30 min in the dark. The loaded cells were washed by KRB buffer three times, and data analysis was performed using AquaCosmos 2.0 (Hamamatsu Photonics, Hamamatsu, Japan). Solutions were superfused at a rate of 2 mL/min. Intracellular Ca²⁺ levels were determined by the emission ratio (F340/F380) (16).

For immunohistochemical staining, 4% paraformaldehyde was refluxed from the left ventricle of mice following heparin-containing saline. Harvested organs were soaked in 4% paraformaldehyde and incubated overnight at 4°C. After dehydration with 10–30% sucrose solution, the organs were placed in optimal cutting temperature compound (#4583; Sakura Finetek Japan, Tokyo, Japan) and quenched to make frozen organ blocks. Eight-micrometer frozen sections were prepared using a cryostat. All frozen sections were washed with PBS, then incubated with 1% horse serum as blocking serum for 60 min. Next, they were applied with primary antibody diluted in 1% horse serum and then incubated at 4°C overnight, followed by an appropriate Fluor-labeled secondary antibody for 60 min. The sections were covered with VECTASHIELD Antifade Mounting Medium (#H-1000; Vector Laboratories, Newark, CA) after incubation for 15 min with DAPI in 1% horse serum if necessary. The following is a list of primary antibodies: anti-Pdx1 (1:200, #ab47383; Abcam, Cambridge, U.K.), anti-insulin (1:200, #C27C9; Cell Signaling Technology), and anti-glucagon (1:200, #ab10988; Abcam). For immunostaining of cultured MIN6 cells, anti-MafA (1:200, #IHC-00352; Bethyl Laboratories, Montgomery, TX) and anti-glucagon (1:200, #M182; Takara Bio, Shiga, Japan) were used. All images were observed and analyzed under confocal microscopy (TCS SP5; Leica).

MIN6 cells were lysed in radioimmunoprecipitation buffer with Halt Protease Inhibitor Cocktail (Thermo Fisher Scientific). The supernatant was collected and boiled for Western blot with Laemmli buffer (17). Protein samples were loaded on polyacrylamide gels and transferred onto polyvinylidene fluoride membranes. Membranes were incubated overnight at 4°C in primary antibody and then with appropriate horseradish peroxidase–conjugated secondary antibody. Blots were developed using SuperSignal West Dura (#34075; Thermo Fisher Scientific), and signals were detected using FUSION FX7 (Vilber, Collégien, France). The following is a list of primary antibodies: anti-glucagon (1:1,000, #92517; Abcam), anti-insulin (1:1,000, #C27C9; Cell Signaling Technology), anti-MafA (1:1,000, #IHC-00352; Bethyl Laboratories), anti-Pdx1 (1:1,000, #47308; Abcam,), and anti–β-actin (1:1,000, #A1978; Sigma-Aldrich).

Krebs-Ringer buffer (KRB) containing 2.8 mmol/L glucose (low glucose) or 16.8 mmol/L glucose (high glucose) was prepared as previously described (18). After incubating MIN6 cells in low-glucose KRB at 37°C for 30 min, we stimulated them with high-glucose KRB. Insulin secretion was measured using an insulin ELISA kit (Morinaga Institute of Biological Science) and adjusted by the total protein amount.

All analyses were performed using GraphPad Prism 9 software. Data are presented as mean ± SEM. Unless otherwise noted, data were analyzed using unpaired two-tailed t tests or ANOVA with Tukey post hoc analysis. Statistical significance was defined as P < 0.05.

All data generated and analyzed in this study are available from the corresponding author upon request.

As the packaging capacity of conventional single-stranded AAV (ssAAV) is generally up to 4.7 kilobase pairs (kbp) (19), getting humanized Streptococcus pyogenes Cas9 (hspCas9) and a gRNA together into an ssAAV vector is challenging, considering the insertion of a large size of Cas9 (∼4.0 kbp) (20) with a pancreatic β-cell–specific promotor. In addition, we prefer scAAV (also referred to as double-stranded AAV) to conventional ssAAV because of its rapid and increased gene expression in pancreatic β-cells (21). On the other hand, the cargo capacity of scAAV is only half of ssAAV, ∼2.3 kbp (22). Taking those limitations in the gene delivery into consideration, we planned CRISPR-Cas9-mediated β-cell–specific gene KO delivering only gRNA with scAAV into pancreatic β-cells, where Cas9 is exclusively expressed in β-cells. Each serotype of AAV has a different tropism for targeted organs (23). It has been reported that AAV8 and 9 are suitable for gene transduction in pancreatic β-cells (24). To examine whether intraperitoneal injection of ssAAV could overexpress exogenous genes in pancreatic β-cells, we purified the AAV8 serotype for expression of tagBFP under the rat Ins2 promoter (25). Intraperitoneal injection of the AAV vector successfully expressed tagBFP in pancreatic β-cells, as reported previously (21) (Supplementary Fig. 1).

Next, we crossbred loxP-STOP-loxP (LSL)-Cas9-P2A-EGFP mice with Ins1-Cre mice to generate βCas9 mice expressing Cas9 specifically in pancreatic β-cells (Fig.1A). The construct is designed as Cas9, and EGFP reporter combined with P2A sequence are transcribed after deletion of LSL cassette by the Cre-loxP system. The Cre-loxP recombination induced expression of EGFP in most pancreatic β-cells, demonstrating efficient Cas9 expression specifically in β-cells (Fig.1B). We also estimated the accurate efficiency of recombination by simultaneous staining with an anti-insulin antibody in isolated islets from βCas9 mice, showing that ∼80% of insulin-positive β-cells are EGFP positive (Fig.1C and D), suggesting that most of the β-cells express Cas9. A recent report has shown that a human growth hormone minigene containing the entire human growth hormone coding region, introns, and polyadenylation signal, integrated into various transgenic mice to enhance gene expression, could disturb β-cell function and influence glucose tolerance in those mice (26). As the construct in βCas9 mice contains a polyadenylation signal from bovine growth hormone, we also examined the glucose tolerance of βCas9 mice. Glucose tolerance at the age of 11 weeks and insulin secretion were comparable between control and βCas9 mice (Fig.2A and B). We also confirmed intact glucose tolerance and insulin secretion in those mice at the age of 15 weeks (Supplementary Fig. 2). In addition, glucose-induced Ca²⁺ influx (27) in β-cells of the isolated islets was also commensurate between control and βCas9 mice, suggesting that insulin secretion is not disturbed by the expression of Cas9 and associated constructs (Fig.2C).

Next, we examined whether gRNA delivery with AAV could knock out the specific genes in βCas9 mice. We constructed the AAV8 vector expressing gRNA for EGFP (14) under the U6 promotor and red fluorescent protein (RFP) under the Ins2 promotor, which enables its expression specifically in β-cells to confirm successful gene delivery in β-cells (Fig.3A). We performed intraperitoneal injection of the purified AAV in βCas9 mice at the age of 7 weeks to generate EGFP βKO mice, in which EGFP is deleted exclusively in pancreatic β-cells, and observed the EGFP expression 4 weeks after the injection. As shown in Fig.3B, βCas9 mice injected with AAV-gEGFP demonstrated effective gene delivery as a robust expression of RFP was confirmed exclusively in pancreatic β-cells. In addition, the expression of EGFP was significantly reduced in the cells where high fluorescence of RFP was observed, demonstrating effective gene KO in β-cells with gRNA delivery (Fig.3B). We also evaluated the accurate efficiency of the gene KO with flow cytometric analysis of the islet cells. As shown in Fig.3C, most of the EGFP-positive, RFP-negative (EGFP-single positive) islet cells in βCas9 mice turned out to be EGFP-negative, RFP-positive (RFP-single positive) cells in EGFP βKO mice, suggesting that EGFP was effectively knocked out in β-cells by Cas9-mediated EGFP gene KO. Quantitative analysis of KO efficiency in flow cytometric analysis, defined as EGFP-negative, RFP-positive cells/(EGFP-positive, RFP-negative cells + EGFP-negative, RFP-positive cells), showed that the efficiency of EGFP KO was ∼80% (Fig.3D). Furthermore, we investigated the blood glucose using intraperitoneal glucose tolerance test (ipGTT) in EGFP βKO mice, confirming that glucose tolerance was intact following AAV injection to βCas9 mice (Supplementary Fig. 3). Our data show effective gene KO in AAV8-mediated delivery of gRNA without affecting glucose tolerance.

Next, we considered knocking out the representative gene playing an essential role in regulating β-cell function with our βCas9-AAV system to demonstrate the deterioration of glucose tolerance as a proof of concept. Pdx1 is a transcription factor whose heterozygous mutation causes maturity-onset diabetes of the young 4 (28) and is essential for pancreatic development (29) and the maturation of β-cells (30,31). Furthermore, pancreatic β-cell–specific Pdx1 KO mice show a significant increase in blood glucose in which loss of β-cell identity contributes to glucose intolerance (32). On the basis of those findings, we designed gRNA targeting for Pdx1 (15) and examined its efficiency with MIN6 cells constitutively expressing Cas9 (12). As shown in Supplementary Fig. 4A and B, the gRNA effectively knocked out Pdx1 in MIN6 cells, with decreased MafA, a transcription factor essential for β-cell function (33), and increased glucagon expression, respectively. Basal and glucose-responsible insulin secretion in Pdx1 KO MIN6 cells were significantly decreased (Supplementary Fig. 4C). We applied the identical sequence of successful gRNA in MIN6 cells to construct the AAV8 vector and injected the AAV8-gPdx1 viral vector intraperitoneally into 7-week-old βCas9 mice to generate Pdx1 βKO mice. We investigated glucose tolerance in Pdx1 βKO mice, showing significant glucose intolerance compared with controls (βCas9 mice with saline injection) in males 4 weeks after the AAV8 injection (Fig.4A). On the other hand, the plasma insulin concentration after glucose injection was slightly decreased in Pdx1 βKO mice compared with the controls. However, the difference was not statistically significant (Fig.4B). We followed their body weight and ad libitum blood glucose levels for another 5 weeks, which showed no significant differences between the two groups (Supplementary Fig. 5A and B). Next, we analyzed the Pdx1 expression in those mice at the age of 16 weeks. In Pdx1 βKO mice, Pdx1 expression in the nucleus of β-cells significantly decreased compared with control mice (Fig.4C and D), suggesting successful Pdx1 KO in a β-cell–specific manner.

Gao et al. (32) reported that adult pancreatic β-cells lose their identity and acquire α-cell–like features following Pdx1 deletion, resulting in severe hyperglycemia. We examined the gene expression pattern of β-cells where the transduction of AAV knocked out Pdx1. Immunostaining for insulin revealed that the insulin-positive area was significantly decreased in Pdx1 βKO mice (Fig.5A). In contrast, we observed glucagon-positive cells not only in the peripheral region (α-cells) but also in the center of islets in Pdx1 βKO mice, suggesting that α-like cells increased after Pdx1 KO (Fig.5B). When we observed the islet cells in fluorescence microscopy with immunostaining, control mice administered with AAV8-gEGFP (EGFP βKO) showed successful KO of EGFP in RFP-positive β-cells, where Pdx1 expression was intact (the cell dotted white in Fig.6A, top). In Pdx1 βKO mice, the EGFP-positive (Cas9-positive) β-cell with RFP expression showed deletion of Pdx1 (the cell dotted white in Fig.6A, bottom), whereas Pdx1 expression was intact in the RFP-negative and EGFP-positive cells (the cells dotted magenta in Fig.6A, bottom). When we examined the β-cells in Pdx1 βKO mice, whereas the EGFP-positive β-cell without RFP expression (dotted white in Fig.6B) demonstrated intact insulin staining, EGFP- and RFP-positive cells showed reduced insulin staining (dotted magenta in Fig.6B), suggesting decreased insulin expression in β-cells of Pdx1 βKO mice. Furthermore, the EGFP- and RFP-positive β-cells in Pdx1 βKO showed glucagon expression, which suggests transdifferentiation of β-cells following Pdx1 KO (dotted white in Fig.6C). These data indicate that transdifferentiation of β-cells to glucagon-positive α-like cells also contributed to the glucose intolerance of Pdx1 βKO mice compared with the control mice. Overall, our data regarding βCas9/AAV-mediated Pdx1 KO successfully recapitulated the results previously shown by conventional β-cell–specific Pdx1 KO, proving our concept that AAV8-gRNA transduction enables Cas9-mediated gene KO in a pancreatic β-cell–specific manner.

In this study, we applied the CRISPR-Cas9 system to generate pancreatic β-cell–specific gene KO mice, with AAV8-mediated delivery of gRNA for the target genes. First, we crossbred LSL-Cas9-P2A-EGFP mice with Ins-Cre mice to generate βCas9 mice, in which Cas9, as well as the fluorescent reporter EGFP, is expressed exclusively in pancreatic β-cells. We confirmed the successful gene KO with intraperitoneal AAV8-gEGFP injection in β-cells of βCas9 mice in high efficiency. Finally, the β-cell–specific KO of Pdx1 proved our concept by demonstrating glucose intolerance and transdifferentiation of Pdx1-deleted β-cells into glucagon-positive α-like cells.

β-Cell–specific deletion of genes of interest, for which the Cre-loxP system is applied, has provided fruitful findings to clarify their roles in various physiological settings, such as tissue development (34) and maintaining cellular homeostasis (35). However, we could accelerate our research with more convenient and rapid methods for deleting the targeted genes in β-cells. In the CRISPR-Cas9 system, we previously established mouse embryonic fibroblasts (36) and MIN6 cells (12) constitutively expressing Cas9, in which transduction of appropriate gRNA could quickly delete the genes of interest. As β-cell–specific Cas9 expression mice are available by crossing LSL-Cas9-P2A-EGFP and Ins1-Cre mice, we planned to transduce gRNAs with an intraperitoneal injection of AAV vector. Wang et al. (21) compared the gene transfer efficiency and vector distribution in the islets between AAV serotype vectors and the delivery routes. They concluded that the intraductal administration of AAV6 is the best way to deliver genes to β-cells in vivo, considering the specificity of the transduction in β-cells, as intraperitoneal or intravenous injection of the vector also induced the gene expression in nonpancreatic tissues. However, we prefer intraperitoneal injection of the AAV8 vector because it is more convenient and less invasive to mice, and the high efficiency of gene delivery to pancreatic islets is confirmed (24). In addition, our βCas9 system can guarantee the specificity of gene KO in β-cells as Cas9 is expressed exclusively in them, even though gRNA could be distributed to other organs, such as the liver (21). On the other hand, AAV-mediated transduction of shRNA is available for gene knockdown in β-cells (37). However, the efficacy of gene knockdown might be limited compared with its KO, considering the difference between the loss-of-function model generated with the Cre-loxP and that with AAV-shRNA injection (3,38).

Our βCas9 system has several limitations. First, as we generate βCas9 mice based on the Cre-loxP system, the issues associated with the system remain to be solved, such as mosaic recombination in a targeted tissue and Cre-mediated nonspecific recombination in other tissues (39), though Hasegawa et al. (40) extensively examined our Ins1-Cre mice to show no unwanted gene recombination, especially in the brain as demonstrated in rat insulin promoter-Cre mice. It is also noteworthy that there have been reports of the Ins1-Cre line becoming silenced (41), which could hamper the KO efficiency in our system by reducing Cas9 expression in β-cells. Second, the individual recombination efficiency of Ins1-Cre mice and that of AAV8-mediated gene transduction are as high as ∼80%; multiplying those values would make it almost down to 60%, suggesting low efficiency for the gene KO. For example, our data showed significant glucose intolerance in β-cell–specific Pdx1 KO mice in our system. In contrast, the loss of insulin content in β-cells was limited (Fig.5A), and glucose tolerance by ipGTT was modestly impaired in Pdx1 βKO mice compared with previous report (32). Careful microscopic examination of the islets would be helpful, as we performed in this study, to prove the loss of insulin content or trans-differentiation. Finally, cost effectiveness is not necessarily superior to the conventional Cre-loxP system, as purification of AAV for in vivo experiments is still expensive and arduous. As previously described, intraductal administration of AAV could save the amount of virus. However, the invasive procedure could perturb glucose tolerance. Our system may be applicable, especially for a pilot study, to examining whether a gene of interest plays a canonical role in β-cells, as a tremendous advantage of our system is that we can clarify the role of a gene of interest in pancreatic β-cells in a short period just by intraperitoneal injection of AAV8 expressing gRNA into βCas9 mice.

We have established a β-cell–specific gene KO system with βCas9 mice and AAV8-mediated gRNA delivery. This method will provide a helpful resource for promoting β-cell research, with more advances in effective purification for AAV and gene transduction.

Acknowledgments. The authors thank Hiroko Hibino, Sumie Ishikawa, Sayaka Matsui, Ryota Hashimoto, and Tamami Sakanishi (Juntendo University Graduate School of Medicine) for technical assistance. The authors also thank the Laboratory of Molecular and Biochemical Research, Biomedical Research Core Facilities, Juntendo University Graduate School of Medicine, for technical assistance.

Funding. This work was partly supported by the Japan Society for the Promotion of Science Grants-in-Aid for Scientific Research (20K08917 to Y.N. and 20H03735 to H.W.)

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. K.U. contributed to the study conceptualization, methodology, investigation, data curation, visualization, and writing of the original draft of the manuscript. Y.N. contributed to the study conceptualization, methodology, validation, formal analysis, investigation, review, manuscript editing, and project administration. S.A., H.U., A.K., M.I., and K.I. contributed to the methodology, validation, and formal analysis. H.I. contributed to the methodology and improving the manuscript. T.M. contributed to the study conceptualization, review, and editing of the manuscript. H.W. contributed to the study conceptualization, resources, investigation, funding acquisition, and supervision. Y.N. is the guarantor of this work and, as such, has full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented in abstract form at the 66th Annual Meeting of the Japan Diabetes Society, Kagoshima, Japan, 11–13 May 2023.