Additive Effects of Dorzagliatin and Glucagon-Like Peptide 1 Receptor Agonism in a Novel Mouse Model of GCK-MODY and in Obese db/db Mice

Abstract

Glucokinase (GK) catalyzes the key regulatory step in glucose-stimulated insulin secretion (GSIS). Correspondingly, hetero- and homozygous mutations in human GCK cause maturity-onset diabetes of the young (GCK-MODY) and permanent neonatal diabetes mellitus, respectively. To explore the possible utility of GK activators (GKAs) and of glucagon-like peptide 1 (GLP-1) receptor agonists in these diseases, we have developed a novel hypomorphic Gck allele in mice encoding an aberrantly spliced mRNA. In islets from homozygous knock-in (Gck^KI/KI) mice, GK immunoreactivity was reduced by >85%, and GSIS eliminated. Homozygous Gck^KI/KI mice displayed frank diabetes (fasting blood glucose >18 mmol/L; HbA_1c ∼108 mmol/mol), ketosis, and nephropathy. Heterozygous Gck^KI/+ mice were glucose intolerant (HbA_1c ∼37 mmol/mol). Abnormal glucose-stimulated Ca²⁺ dynamics in Gck^KI/+ islets were completely reversed by the GKA dorzagliatin, which was largely inactive in homozygous Gck^KI/KI mouse islets. The GLP-1 receptor agonist exendin-4 improved glucose tolerance in male Gck^KI/+ mice, an action potentiated by dorzagliatin. Sex-dependent additive effects of these agents were also observed on insulin secretion in vitro. Similar additive effects of the drugs were observed in obese hyperglycemic db/db mice. Combined treatment with GKA and incretin mimetics may thus be useful in GCK-MODY and in more common forms of type 2 diabetes.

Article Highlights

• Glucokinase (GK) deficiency can drive maturity-onset diabetes of the young (GCK-MODY) in heterozygotes and permanent neonatal diabetes in homozygotes.

• We describe a hypomorphic Gck allele that results in aberrant splicing in islets and liver lowering GK activity by ∼85%. Whereas heterozygous mutant mice are mildly hyperglycemic, homozygotes have frank diabetes but survive to adulthood.

• Dorzagliatin potentiates the effects of glucagon–like receptor-1 receptor activation sex dependently in heterozygous Gck mice and in obese hyperglycemic db/db mice.

• Combined use of these drugs may be useful in some forms of GCK-MODY and in obesity-related type 2 diabetes.

Graphical Abstract

Introduction

Diabetes affects more than 1 in 10 of the global adult population (1), a figure expected to grow to >750 million people by 2045 (2). Glucokinase (GK) catalyzes the flux-generating step in glycolysis and has been dubbed the “glucose sensor” of the pancreatic β cell (3–7). After phosphorylation by GK, glucose carbons flow through the glycolytic pathway and enter the citrate cycle, stimulating respiratory chain activity to increase cytosolic ATP/ADP ratios. Closure of ATP-sensitive K⁺ channels leads to depolarization of the plasma membrane, Ca²⁺ influx through voltage-gated calcium channels, and insulin release (8,9). Roles for other coupling factors (10,11) and local ATP/ADP microdomains (12), are also proposed (13), though the latter are disputed (14).

Homozygosity for GK (GCK) loss-of-function alleles results in permanent neonatal diabetes mellitus (PNDM) (15), which often requires insulin treatment shortly after birth (16–18). Heterozygosity for loss-of-function mutations is associated with maturity onset diabetes of the young (MODY) (19,20). GCK-MODY (formerly MODY2) accounts for 20–30% of all MODY cases. Although chronically hyperglycemic (fasting glucose level 6–8 mmol/L) (21), patients with GCK-MODY are largely asymptomatic (21). However, fetal macrosomia is an important complication in pregnancy (22).

Several mouse models have been generated to explore how Gck insufficiency affects whole-body metabolism (23). Animals in which exon 2 was deleted and a frameshift mutation incorporated (24), or exon 4 plus parts of exons 3 and 5 were deleted (25), respectively, die in utero or perinatally as homozygotes, and have mild hyperglycemia as heterozygotes. β-Cell-selective Gck knockouts have severe hyperglycemia and die a few days after birth (26,27). Demonstrating the importance of islet dysfunction in the impact of these mutations, rescue of Gck expression selectively in the β cells of Gck null mice is sufficient to reverse lethality (25).

Incretin-based drugs, including glucagon-like peptide 1 (GLP-1) receptor agonists, provide a highly efficient treatment for type 2 diabetes (28). Use of these agonists has not been explored either in GCK-MODY or GCK-PNDM up to now, to our knowledge. Dorzagliatin (29) is a latest-generation allosteric GK activator (GKA) that is therapeutically useful in GCK-MODY (30). These findings raise the possibility that the combined use of incretins and GKA may provide additional benefits compared with the use of either agent alone.

Here, we explore this possibility in wild-type (WT) mice and in animals bearing a hypomorphic Gck allele that results in aberrant splicing and severe depletion of active GK protein. Residual GK activity, nevertheless, is sufficient to allow the survival of homozygous mice into adulthood, despite severe diabetes, modeling GCK-PNDM. Heterozygous Gck^KI/+ mice provide a convenient model of GCK-MODY. We tested the potential therapeutic utility of GKAs and incretins in both models, as well as in hyperglycemic db/db mice.

Research Design and Methods

Generation of Hypomorphic Gck Alleles

The overall strategy, as designed with genOway (Grenoble, France), is illustrated in Fig. 1A and Supplementary Fig. 1. Integration of the mutant allele was confirmed by Sanger sequencing. Conventional PCR targeting the exon 1 to exon 9 of mRNA-derived cDNA from Gck^KI/KI islets (Supplementary Fig. 1B) yielded a barely detectable band corresponding to the WT product (∼1.17 kb) along with a major product of ∼850 bp (∼350 bp less than predicted). Sanger sequencing of the latter revealed an aberrant splicing event eliminating exons 2 and 3 (Fig. 1B and Supplementary Fig. 1C). In silico translation of the aberrantly spliced isoform suggested the production of a protein comprising 360 amino acids (AAs) versus native WT (465 AAs) GK. The mutant form is expected to lack AAs 16–121, including residues in the small lobe critical for ATP binding and catalysis (6,31,32).

Figure 1.

Generation and analysis of the Gck^KIallele. A: The 3′ Gck coding sequence comprising exon-4 to exon-10 was directly fused in frame with endogenous exon-3. This was followed by an IRES, mCardinal cDNA, and termination signal plus human growth hormone (hGH) polyA. The construct was expected to produce a single bicistronic mRNA expressing GK and mCardinal from the same mRNA transcript. B: WT Gck mRNA structure. C: Observed sequence expressed from the knock-in allele. Sanger sequencing of the Gck PCR product from Gck^KI/KI islet cDNA revealed alternative splicing in which exon 2 and exon 3 are skipped. D: Schematic representation of the full-length mouse GK (GCK FL), highlighting exons 2 and 3 (residues 16–121) and the positions of five critical ATP-binding residues: Asp78 and Thr82 (within exon 2), and Gly227, Thr228, and Ser336. E: AlphaFold-predicted structure of the murine GCK FL, aligned with known ligand-bound conformations. Exons 2 and 3 are shown in magenta. Glucose (yellow) and ATP (orange) are modeled in the active site, revealing that exons 2 and 3 contribute structurally to the nucleotide-binding cleft. F: AlphaFold-predicted structure of the GCK Δexon2–3 variant, showing collapse of the N-terminal fold and disruption of the ATP-binding pocket in the absence of exons 2 and 3. G: Close-up of the ATP-binding site in the GCK FL model. Asp78 and Thr82 coordinate ATP and Mg²⁺ (purple sphere) through hydrogen bonding (yellow dashed lines), stabilizing the nucleotide in the active site. The glucose molecule (yellow), visible in this zoomed view, highlights the proximity of the two substrate-binding sites. H: SDS-PAGE analysis of recombinant GCK FL and GCK Δexon2–3 expressed in E. coli. GCK FL is predominantly detected in the soluble fraction (S), whereas the Δexon2–3 variant accumulates in the pellet (P), indicating poor solubility and likely misfolding. Arrows indicate the expected molecular weights: ∼50 kDa for GCK FL and ∼42 kDa for the Δexon2–3 variant. CDS, coding sequence; T, total lysate.

cDNAs corresponding to the expected internal ribosome entry site (IRES) or mCardinal regions were not detected in islets. mCardinal fluorescence was absent from islet, liver, and brain (ventromedial and lateral hypothalamus) (not shown) (33).

Animal Husbandry

All experimental manipulations were approved by the local ethical committee (Research Center of the Hospital Center of the University of Montreal, Montreal, QC, Canada; approval Montreal CIPA 2022–10,040 CM21022GRs). Colonies of GK^KI and of GK^KI:Ins1Cre:GCaMP6^f/f mice (34) on a C57BL/6 J background were fed chow diet and maintained at 21–23°C, humidity 45–50%, and 12-h day-night light cycle.

Generation and Structural Analysis of WT and Mutant GK

The gene encoding mouse GK (GCK) isoform 1 (islet isoform) or GCK Δexon2–3 mutant were cloned into a pRSF vector adding an N-terminal 10 × His tag and a TEV protease cleavage site. Verified clones were subsequently transformed into Escherichia coli Rosetta(DE3) cells for protein expression. Molecular modelling in silico was performed using Alpha Fold 3.0.

Expression and Purification of Recombinant GK Proteins

Proteins were purified as described (35) with the following buffer: 20 mmol/L Tris-HCl, pH 7.5, 400 mmol/L NaCl, 1 mmol/L PMSF, and 1 µmol/L pepstatin A. For the variant, purification in denaturing condition (6 mol/L guanidium hydrochloride) was tried, but refolding was unsuccessful.

Intraperitoneal and Oral Glucose Tolerance Tests

Mice were fasted overnight (16 h) with free access to water. At 9:00 a.m., glucose (2 g/kg body weight) was administered via intraperitoneal injection or oral gavage. Blood glucose levels were measured from tail vein 0, 15, 30, 60, and 90 min later with a glucometer (Contour next-ONE) (36).

Insulin Tolerance Tests

After 6 h of fasting with free access to water, 0.75 units/kg body weight human insulin (Novolin-ge Toronto; Novo Nordisk) was administered intraperitoneally. Blood glucose was measured as reported in the preceding paragraph (36).

β- and α-Cell Mass and Ki67 Staining

After transcardiac perfusion and paraffin embedding, 3–5 pancreatic sections (5μm thickness) per animal were collected. Sections were immunostained for insulin (CST 4590) and glucagon (CST 2760) using DAB detection (CST 8059) (all from Cell Signaling Technology [CST]). We quantified β- and α-cell mass (Stepanizer [37]) and normalized to total pancreatic weight (38). For proliferation analysis, sections were stained for Ki67 following the CST protocol with KI-67 (D3B5) Rabbit mAb (CST, 12202; 1:200), and DAB detection. Ki67⁺ nuclei were expressed as a percentage of total islet nuclei.

Renal Histology

Whole kidney was dissected after transcardiac perfusion and fixed for 2 h in 4% paraformaldehyde and then paraffin embedded. Histology analysis was done at a ×20 magnification, on whole kidney sections. Images were taken with a Slide Scanner Leica, Aperio Versa 200 digital scanner and further analyzed in the Aperio ImageScope 12.43.3 software. Renal tubular damage was described based on luminal dilatation and necrosis, loss of brush border, and cast formation.

PCR and RT-qPCR

Total RNA was extracted using RNeasy Mini Kit (Qiagen 74104) and reverse transcription was performed using ThermoFisher kit (4368814). Quantitative PCR was carried out using SYBR Green–based detection on a QuantStudio-5 normalized to β-actin (Actb), according to the comparative cycle-threshold method. Primers are listed in Supplementary Table 1.

Ca²⁺ Imaging and Connectivity Analysis

Imaging was performed as described previously (39,40) using a Zeiss LSM 900 Airyscan 2 super-resolution confocal microscope, equipped with an incubation system set at 37°C. Connectivity analysis was performed as reported by Chabosseau et al. (39).

Western Immunoblotting

Blotting was performed per the work of Georgiadou et al. (41). After membrane blotting, membranes were incubated with primary antibody overnight at 4C. Membranes were washed with Tris-buffered saline with Tween 1% before incubation with secondary antibody (1 h, room temperature). Horseradish peroxidase–chemiluminescent substrate (ECL Western substrate; ThermoFisher) was used to reveal bound antibody. Full membranes are depicted in Supplementary Material.

GK Assay

Assays were performed with Glucokinase Activity Assay Kit (Fluorometric; Abcam, ab273303). Briefly, 100 mg of liver tissue per animal was homogenized. Protein concentration was determined by DC Protein Assay kit (Bio-Rad, 5000111). After establishing a baseline, GCK activity was initiated by adding 2 U/mL glucose-6-phosphate dehydrogenase. Fluorescence (excitation/emission = 535/587 nm) was recorded, and activity was calculated based on NADPH production normalized to protein content.

Statistical Analysis

Data are expressed as mean ± SD, unless otherwise stated, and significance was tested by one- or two-way ANOVA with Šidák or Brown-Forsythe multiple comparison tests, using GraphPad Prism 9 (GraphPad Software, San Diego, CA). P ≤ 0.05 was considered statistically significant.

Data Resource and Availability

All data generated or analyzed during this study are included in the published article and its online Supplementary Material. Details of videos are provided in Supplementary Material). Analytical scripts are provided at https://zenodo.org/records/14042795.

Results

Generation of a Hypomorphic Gck Allele

With the initial objective of identifying β-cell subpopulations enriched for GK (which may correspond to highly connected “hub” cells) (42), we designed a “knock-in” construct encoding exons 3–10 of mouse Gck, followed by an IRES, cDNA encoding the fluorescent protein mCardinal, and a polyA sequence (Fig. 1A and C, and Supplementary Fig. 1A). Genomic sequencing confirmed the expected integration at the Gck locus (not shown). However, and unexpectedly, an alternatively spliced product was generated, which was predicted to encode a mutant protein that lacks residues 16–121, corresponding to much of the small lobe of GK (31). Inspection of the crystal structure of human GK (Protein Data Base [PDB]: 1V4S) (Fig. 1D–G) revealed that the glucose-binding pocket is formed by conserved residues from both the large domain (Glu256, Glu290), the small domain (Thr168, Lys169), and the connecting region II (Asn204, Asp205) (43). Alpha Fold 3.0 modeling of full-length murine GK revealed that exons 2 and 3 encode part of a β-sheet and contribute to the overall folding of the N-terminal domain near the ATP-binding pocket (Fig. 1E). Although the glucose-binding residues are still present in the mutant Δexon2–3 construct (Fig. 1F), those required for dorzagliatin binding (Fig. 1D–F) and two of the five conserved residues critical for ATP binding (Asp78 and Thr82) and also required to coordinate Mg²⁺ (43) are eliminated (Fig. 1G and Supplementary Videos 1 and 2). Hence, the mutant is expected to be essentially inactive.

To validate these structural predictions experimentally, both full-length (FL) GK and GK Δexon2–3 were expressed in E. coli. SDS-PAGE analysis showed that although GK FL predominantly was found in the soluble fraction, the Δexon2–3 variant accumulated in the pellet, indicating likely misfolding and poor solubility (Fig. 1H). These results indicate that exons 2 and 3 are essential for proper GK folding.

GK protein levels in islets from WT and mutant mice were assessed by Western immunoblotting (Fig. 2A and B). Gck is expressed from different promoters in the liver and in the pancreatic β cell (44). Consequently, the first 15 AAs encoded by exon 1 from the β-cell promoter differ from those of the equivalent exon encoded in the liver (6). The first antibody deployed (ThermoFisher; PA5-15072) is raised against the first 30 AAs of type 1 GK (islet). Loss of the following 16 AAs (encoded by exon 2) from the epitope is thus likely to weaken or destroy recognition by the antibody of the misspliced isoform. This approach revealed a 30–45% and ≥85% lowering in apparent GK immunoreactivity (∼48 kDa) in islets from GK^KI/+ and GK^KI/KI, respectively, versus control mice of either sex, indicating a proportionate lowering of the intact isoform 1 protein (including exons 2 and 3) (Fig. 2B). These changes were associated with lowered expression of multiple genes associated with β-cell identity (Fig. 2C, i–iv) and with increased senescence (Fig. 2C, v–vii).

Figure 2.

Apparent GK protein levels in male and female pancreatic islets. A: Western immunoblot images from islets protein from WT, Gck^KI/+ (heterozygous [Het]) and Gck^KI/KI (homozygous [Hom]) males and females. GK was detected at ∼48 kDa and β-actin at ∼45 kDa. B: Quantitative analysis of GK expression levels normalized to β-actin across sex and genotype. Each dot represents a single animal. Unpaired one-way ANOVA with Tukey correction. C: Quantitative PCR analysis was performed on islet RNA to assess the expression of genes related to β-cell identity and function, as well as cellular senescence. Expression of β-cell transcription factors is shown in i–iv, and senescence markers are shown in v–vii. Gene expression was normalized to housekeeping gene β-actin and are presented as fold change relative to WT. Data are shown as mean ± SD. Statistical significance was determined using one-way ANOVA with Brown-Forsythe correction. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ***P ≤ 0.0001.

Semiquantitative PCR yielded a band in GK^KI/KI mouse liver cDNA not present in control liver (Supplementary Fig. 2A), demonstrating a similar Gck splicing defect as in islets. Western blotting of liver extracts with a second antibody (Bosterbio picoband A00884-1) raised against AAs 234–431 revealed an ∼18% and >90% lowering of GK protein levels (∼48 kDa) in GK^KI/+ and GK^KI/KI extracts, respectively (Supplementary Fig. 2B and C). Finally, direct enzymatic assays (Supplementary Fig. 2D) demonstrated a drastically (∼95%) lowered GK activity in GK^KI/KI livers versus controls and a tendency toward lowered activity in GK^KI/+ liver.

Metabolic Characterization of Gck Mutant Mice

When examined at 8 weeks of age, heterozygous (Gck^KI/+) mice had normal body weight, whereas male homozygous (Gck^KI/KI) animals weighted less than littermate controls (Fig. 3A and B); a similar tendency was seen in female mice (Fig. 3B). HbA_1c levels tended to be elevated or were drastically increased, respectively, in Gck^KI/+ and Gck^KI/KI animals versus controls (Fig. 3C). Fasting blood glucose concentration was >18 mmol/L in the latter animals of either sex at 8 weeks of age (males [n = 7]: 21.6 to >33.3 mmol/L; females [n = 5] 18.4–27.4 mmol/L) (Fig. 3D and F). Whereas overnight (16 h) fasting blood glucose levels did not differ between Gck^+/+ and Gck^KI/+ mice of either sex (Fig. 3D and E), Gck^KI/+animals had markedly abnormal glucose tolerance, with no return to preinfusion levels after 90 min. Blood glucose levels after 6 h fasting tended to be higher in Gck^KI/+ than in WT mice (Supplementary Fig. 3A and B). Young (5–8 weeks old) male Gck^KI/+ animals displayed no significant differences in comparison with WT mice, whereas female mice were strikingly glucose intolerant (Supplementary Fig. 3C and D). Conversely, 6-month-old Gck^KI/+ animals displayed a tendency toward lower glucose tolerance in comparison with mice aged 8–16 weeks (Supplementary Fig. 3E and F).

Figure 3.

Metabolic characterization of Gck^KI/+ and Gck^KI/KI mice. A: Representative images across male genotypes. B: Mouse weight at 8 weeks of age across sex and genotype (male mice: n = 15 WT, 10 heterozygous [Het], and 4 homozygous [Hom]; female mice: n = 13 WT, 13 Het, and 5 Hom). C: HbA_1c at 8–16 weeks across sex and genotype (male mice: n = 7 WT, 8 Het, and 8 Hom; female mice: n = 5 WT, 6 Het, and 8 Hom). Unpaired one-way ANOVA with Tukey correction. D: IPGTTs with 2 g/kg glucose performed at 8–16 weeks of age for males (n = 12 WT and 9 Het) and females (n = 12 WT and 13 Het). Unpaired one-way ANOVA with Šidák’s correction. E: Glucose excursion (area under the curve [AUC]) quantifications from the male and female IPGTTs presented in D. Unpaired one-way ANOVA with Tukey correction. F: Intraperitoneal insulin tolerance test (IPTT) with 0.75 units/Kg insulin performed at 8–16 weeks (male mice: n = 10 WT, 6 Het; female mice: n = 7 WT, 7 Het). Unpaired two-way ANOVA with Šidák’s correction. G: Glucose AUC quantifications of IPGTTs presented in F. Unpaired one-way ANOVA with Tukey correction. The homozygous fasting values are represented by the red triangle (n = 3 mice) in D and F. Due to the high levels of fasting glucose, IPGTTs and IPTTs were not performed in homozygote mice, and the fasting glucose levels are presented as a reference only. H: Insulin secretion from isolated islets across genotypes and sexes. I: Ketone body levels in fasted animals. n = 3–5 per group. Unpaired one-way ANOVA with Tukey correction. P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. ITT, intention to treat.

Gck^KI/+ animals of both sexes also had abnormal oral glucose tolerance (Supplementary Fig. 3G and H), and this was unaffected by dorzagliatin (Supplementary Fig. 3I and J and Supplementary Fig. 3G and H), indicating the drug does not potentiate the effects of endogenous incretins. Insulin sensitivity did not differ between genotypes (Fig. 3F and G) and was unaffected by dorzagliatin (Supplementary Fig. 3K and L).

Measured in batch incubations, glucose-stimulated insulin secretion (GSIS) from male islets tended to be lowered in Gck^KI/+ versus controls and eliminated in homozygous Gck^KI/KI islets (Fig. 3H). In females, GSIS was unaffected in Gck^KI/+ but was eliminated in Gck^KI/KI islets (Fig. 3H).

Although also unaltered in Gck^KI/+ mice versus controls, β-cell mass was substantially lowered in Gck^KI/KI animals (∼65% in males, ∼50% in females) (Supplementary Fig. 4). Conversely, α-cell numbers were markedly increased in Gck^KI/KI mice (approximately twofold in both sexes) (Supplementary Fig. 4). Consequently, the β-cell–to–α-cell ratio fell from ∼10:1.0 in WT to ∼2.0:1.0 in Gck^KI/KI pancreata. Whereas a marked decrease in proliferation was observed (Supplementary Fig 4C), apoptosis was not different between control and Gck^KI/KI mice (Supplementary Fig. 4D). Consistent with increased senescence (Fig. 2C, v–vii), nuclear HMGB1 staining was also increased in Gck^KI/+ and Gck^KI/KI islets (Supplementary Fig. 4E).

Ketone levels tended to be, or were significantly, increased in male and female Gck^KI/KI animals versus controls, respectively (Fig. 3I), indicative of frank diabetes in the former. Moreover, in the kidney, homozygous Gck^KI/KI mice of either sex displayed clear tubular alterations compared with heterozygous Gck^KI/+ and control mice (Supplementary Fig. 5 and Supplementary Table 2).

Ca²⁺ Dynamics and Intercellular Connectivity

The findings suggested defects may exist in intracellular glucose handling or signaling in mutant β cells. To explore this possibility, we studied glucose-regulated intracellular Ca²⁺ dynamics and cell-cell connectivity in isolated islets, using high-speed confocal imaging (Fig. 4A) (42). To ensure Ca²⁺ was measured exclusively in β cells, mice were bred with animals carrying Ins1Cre alleles (45) and STOP-Flox alleles of the genetic Ca²⁺ sensor, GCaMP6f (34). Male animals carrying WT Gck alleles had robust responses to stimulation with 11 mmol/L glucose, and a high degree of connectivity (Fig. 4B–E and Supplementary Videos 2 and 3). These responses were significantly weakened in Gck^KI/+ mice and almost eliminated in Gck^KI/KI animals. Similar behavior was seen in islets from animals of both sexes, though the Ca²⁺ responses (area under the curve) were lower in females than males (Fig. 4F and Supplementary Fig. 6). Remarkably, these responses were fully normalized with dorzagliatin in Gck^KI/+ mice of both sexes (Fig. 4F and G, Supplementary Fig. 6, and Supplementary Videos 4 and 5), but were barely affected in Gck^KI/KI mice. Nevertheless, a subset of Gck^KI/KI islets had detectable Ca²⁺ transients in response to glucose stimulation in the presence of dorzagliatin (Fig. 4F and G).

Figure 4.

Dorzagliatin restores glucose-induced Ca²⁺ dynamics and β-cell connectivity in Gck^KI/+mouse islets. A: Experimental design for in vitro calcium imaging in islets from 8- to 16-week-old Ins1Cre:GCaMP6f^fl/fl:Gck^+/+ (WT), Gck^KI/+(Het) and Gck^KI/+(Homo) triple transgenic mice. Where indicated, isolated islets were treated for 1 h with 10 µmol/L dorzagliatin prior to imaging. B: Snapshots from confocal time-lapses recordings of the individual islets at 6 Hz in 11 mmol/L glucose with or without 10 µmol/L dorzagliatin across Gck genotypes. C: Normalized GCaMP6f fluorescence traces from islets shown in B. D: Raster plots show the apparent Ca²⁺ signal for individual cells from islets shown in B. E: Visualization of the islet functional network for each genotype, with or without dorzagliatin, from islets displayed in B. F: Calcium area under the curve (AUC) quantification across genotypes in control and dorzagliatin treated islets. G: Connectivity quantifications across genotypes in control and dorzagliatin treatment groups. Unpaired one-way ANOVA with Tukey correction. For 11 mmol/L glucose-only treatment, n = 5 WT mice (n =9 islets), n = 8 heterozygotes (n =14 islets), and n = 4 homozygotes (n =7 islets). For dorzagliatin treatment, n = 4 WT (n =6 islets), n = 5 heterozygotes (n =13 islets), and n = 4 homozygotes (n =11 islets). Each dot represents an individual islet. The numbers 1, 2, and 3 in the snapshots in B indicate the time points corresponding to the fluorescence traces showed in C. Scale bar, 25 μm. ***P ≤ 0.001, ****P ≤ 0.0001.

Responses of WT and Gck^KI/+ Mice to GKA and Incretin Mimetic in Vivo

We next explored the effects of dorzagliatin in vivo during intraperitoneal glucose tolerance tests (IPGTTs), in the presence or absence of the GLP-1 receptor agonist, exendin-4.

Acute injection of dorzagliatin or exendin-4 alone, examined in WT mice of either sex, exerted only minor effects on glucose tolerance, which did not reach statistical significance (Fig. 5A–D). On the other hand, in WT male (but not female) mice, coinjection of dorzagliatin markedly potentiated the action of exendin-4 (Fig. 5A, B, D, and E).

Figure 5.

Impact of dorzagliatin (Dorza) and exendin-4 (Ex-4) on glucose tolerance across Gck genotypes and sexes. A and B: IPGTT with 2 g/kg glucose performed in animals 8–16 weeks of age for males and females after injection of glucose only or glucose in combination with dorzagliatin (1 mg/kg), exendin-4 (1 nmol/kg), or dorzagliatin (1 mg/kg) plus exendin-4 (1 nmol/kg) in WT or on a heterozygous (Het) background. A and B: The first panels (“Glucose”) show the same data shown in Fig. 3D. Unpaired two-way ANOVA with Šidák’s correction. C: IPGTT with 2 g/kg glucose performed in heterozygous female animals 8–16 weeks old, either sham or ovariectomized (Ovx) Figure 5 (continued)—after injection of glucose only or glucose in combination with dorzagliatin (1 mg/kg), exendin-4 (1 nmol/kg), or dorzagliatin (1 mg/kg) plus exendin-4 (1 nmol/kg). Unpaired two-way ANOVA with Šidák’s correction. D–F: Glucose area under the curve (AUC) quantifications of each IPGTT treatment from male and female mice presented in A–C. Male mice: n = 4 WT, n = 7 Het for each treatment; female mice n = 5 WT, n = 10 Het for each treatment. n = 5 sham, n = Ovx for each treatment. Unpaired one-way ANOVA with Tukey correction. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. Hom, homozygous.

In heterozygous male Gck^KI/+ mice, exendin-4 alone, but not dorzagliatin, improved glucose tolerance, and the combination of exendin-4 and dorzagliatin profoundly lowered glucose excursions to levels comparable to those in WT mice (Fig. 5A and D). In female Gck^KI/+ mice, dorzagliatin again exerted no effect when administered alone, whereas exendin-4 markedly improved glucose tolerance. No additional effect was observed of coinjecting dorzagliatin and exendin-4 (Fig. 5B and E). Combining data from both sexes, we noted that neither drug, alone or in combination, significantly affected glycemia in homozygous (Gck^KI/KI) animals (Fig. 5D and E, and Supplementary Fig. 7).

To explore the basis of the observed sex differences, we performed ovariectomy or sham surgery on female Gck^KI/+ mice (Fig. 5C). Similarly to sham-operated mice, ovariectomized animals responded to the combination of exendin-4 and dorzagliatin but not to the individual drugs (Fig. 5F). Ovariectomy did not affect insulin tolerance (Supplementary Fig. 8A and B) but worsened oral glucose tolerance. Dorzagliatin had no impact on either parameter (Supplementary Fig. 8A–D).

Effects of Incretin Mimetic and GKA on GSIS

Explored in vivo during IPGTTs, glucose-induced increases in circulating insulin levels tended to be potentiated by exendin-4 in male, but not female, Gck^+/+ and Gck^KI/+ mice (Supplementary Fig. 9). Because interanimal variation was large in these experiments, we examined the potential interaction between GKA and incretin on insulin secretion ex vivo. In islets isolated from WT male mice, neither dorzagliatin nor exendin-4 significantly affected insulin secretion stimulated by 17 mmol/L glucose (Fig. 6B). In contrast, the combination of dorzagliatin and exendin-4 stimulated hormone release between 2.0- and 2.5-fold. In islets from male Gck^KI/+ mice (Fig. 6A), dorzagliatin alone tended to increase secretion, whereas exendin-4 alone caused an approximately threefold increase in hormone release. The effect of incretin was further augmented by dorzagliatin (approximate fivefold increase versus 17 mmol/L glucose alone). Qualitatively similar results were obtained in WT female mice (Fig. 6B). In female Gck^KI/+ islets, the effects of dorzagliatin or exendin-4, whether administered alone or together, were weaker than those in Gck^KI/+ males (Fig. 6A and B). This sexual dimorphism in drug responses was exacerbated when the data were expressed as fold-change in response to glucose (17 vs. 3 mmol/L) (Supplementary Fig. 10A and B).

Figure 6.

Impact of dorzagliatin (Dorza) and incretin (exendin-4 [Ex-4]) on GSIS in vitro. A and B: Insulin secretion in vitro for male (A) and female (B) islets at low (3 mmol/L) and high (17 mmol/L) glucose concentrations. C: Stimulation index, calculated as the ratio of insulin secreted at 17 mmol/L glucose to that at 3 mmol/L glucose, from pooled islets of both sexes across genotypes. One-way ANOVA of Brown-Forsythe. Mean ± SD. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001. CTR, control; Het, heterozygous; Hom, homozygous.

When data from both sexes were pooled, exendin-4 potentiated GSIS in Gck^KI/+ islets, an action further enhanced by dorzagliatin (Fig. 6C). Similar tendencies were observed in WT and in Gck^KI/KI islets (Fig. 6C), with mutual potentiation by GKA and exendin-4 clearly evident, when explored as fold-change (Supplementary Fig. 10C).

Glycemic Responses of db/db Mice to GKA and Incretin Mimetic in Vivo

We next tested the effects of dorzagliatin in vivo in the presence or absence of exendin-4 in a model of obesity-driven diabetes, the db/db mouse (46), which carries a recessive mutation in the leptin receptor gene (47). Both males and females were frankly diabetic, with glucose excursions during IPGTT that did not return to baseline after 90 min. (Fig. 7A).

Figure 7.

Acute and chronic effects of dorzagliatin (Dorza) and exendin-4 (Ex-4) in db/db mice. A: IPGTT with 2 g/kg glucose performed in male and female (Fem) db/db animals 8–10 weeks old after injection of glucose only or glucose in combination with dorzagliatin (1 mg/kg), exendin-4 (1 nmol/kg), or dorzagliatin (1 mg/kg) plus exendin-4 (1 nmol/kg). B: IPGTT with 2 g/kg glucose performed in male and female db/db animals 12–16 weeks old after 1 week of daily injections of PBS, dorzagliatin (1 mg/kg), exendin-4 (1 nmol/kg), or dorzagliatin (1 mg/kg) plus exendin-4 (1 nmol/kg). C: Glucose area under the curve (AUC) quantifications of each IPGTT treatment from male and female mice presented in A. Male mice: n = 5 for glucose and n = 6 for each treatment. Female mice: n = 6 for each treatment. D: Glucose AUC quantifications of each chronic treatment from male and female mice presented in B. Male mice: n = 6 for each treatment; female mice: n = 6 for each treatment. E: Body weight gain after each chronic treatment for male and female mice. Unpaired one-way ANOVA with Tukey correction. *P ≤ 0.05, **P ≤ 0.01. ns, not significant.

Acute treatment with dorzagliatin or exendin-4 alone did not exert significant effects on glucose tolerance (Fig. 7A and C). However, in both sexes, coinjection of dorzagliatin potentiated the action of exendin-4 (Fig. 7A and C). Coinjected alone, dorzagliatin had no effect on insulin (Supplementary Fig. 11A and B) or oral glucose (Supplementary Fig. 11C and D) tolerance. One-week of daily injections of the drugs alone or in combination had no impact on glucose tolerance (Fig. 7B and D), though the combination of dorzagliatin and exendin-4 tended to reduce body weight gain (Fig. 7E).

Discussion

Novel Models of “Glucokinase Diabetes”

We describe here novel mouse models of GCK-MODY and GCK-PNDM, and we explore the potentiation of GSIS by a GKA and incretin hormone mimetic in each case. More than 600 variants are associated with human GCK-MODY, with eight identified in GCK-PNDM (48). The latter vary in their effects on GK activity or stability and have broadly proportional clinical impacts (21). Although homozygosity for mutants with near-zero activity is usually implicated in PNDM, compound heterozygosity for an inactivating (e.g., intervening sequence 8 + 2T→G) and hypomorphic (G264S) variant (16) can also drive neonatal diabetes. On the other hand, milder mutations, such as G223S and L315H, with relative activities of 0.25 and 0.89, respectively (49), lead to GCK-MODY. Interestingly in the context of the present study, Osbak et al. (48) have speculated that milder GCK mutations in GCK-PNDM (e.g., R397L, G264S [16,50]) may be more responsive to GKA than those leading to more complete inactivation (e.g, T168A (51]).

The hypomorphic Gck allele described here was originally designed to preserve an intact GK protein while independently expressing, from the same transcript, the fluorescent reporter mCardinal. Instead, an aberrantly spliced transcript was produced, predicted to generate an inactive protein (31). Of note, the in vitro expressed protein was largely insoluble, and GK activity was eliminated in the liver of homozygous mutant mice (Supplementary Fig. 2). Enzymatic measurements of GK in whole islets are challenging given the presence of hexokinases I-III in islet non–β cells (52), and usually require β-cell purification (53).

The persistence of residual GK activity in β cells presumably explains why homozygous Gck^KI/KI mice survive into adulthood, in contrast with models of complete Gck inactivation (23) (see Introduction). Although Ca²⁺ imaging revealed that the GKA was largely inactive in male Gck^KI/KI islets, as anticipated, 10–30% of islets displayed responses to dorzagliatin in male and female mice (Fig. 4F and G). These findings suggest that correctly spliced, active GK may be restricted to a subset of Gck^KI/KI islets.

Accompanying these changes were sharp reductions in apparent β-cell identity, and increased senescence, particularly in homozygous Gck^KI/KI mice (Fig. 2C). These changes were observed despite a drastic lowering of glucose levels from >30 to ∼16 mmol/L after the engraftment of insulin pellets, and suggest that β cell–autonomous actions, as well as sustained hyperglycemia, are involved (54,55).

Our observation of reduced β-cell mass in homozygous mice aligns with earlier studies in GK-deficient mice (56). Less expected was the increase in α-cell mass in Gck^KI/KI islets, suggesting that GK activity plays a role in suppressing α-cell expansion, in addition to inhibiting glucagon secretion (57). Extrapolated to humans, these findings raise the intriguing possibility that increased glucagon levels may aggravate hyperglycemia in GCK-PNDM.

We note that, in addition to islets and liver, GK is expressed in several brain nuclei, including the hypothalamus, pituitary, and brain stem (33), as well as in intestinal L cells (58). As such, the phenotypes of Gck^KI/+ and Gck^KI/KI mice may reflect, in part, actions on these cell types. In summary, the Gck^KI allele provides a convenient model of inactivating GCK mutations that affect the kinetic properties of thermal stability of the enzyme (59).

Sex-Dependent Additive Effects of Dorzagliatin and Incretin on Glucose Tolerance and Insulin Secretion

Earlier-generation GKAs (60) resulted in a marked left-shift in the response to glucose in both mouse and human islets. Under the conditions used here, involving high glucose concentrations in both settings, dorzagliatin alone elicited minimal effects either in vivo on glucose tolerance (Fig. 5A, B, D, and E) or on GSIS (Fig. 6A and B), both in WT and Gck^KI/+ mice. Nevertheless, the GKA markedly potentiated the effects of exendin-4 on GSIS both in vivo (Supplementary Fig. 9) and in vitro (Fig. 6A), an effect most dramatic in heterozygous male mice. Although we do not exclude the possibility that the greater impact of the drug combination observed in vivo in male than female mice reflects the greater basal glycemic excursions observed in the former versus the equivalent female genotype (e.g., Fig. 5D and E), data with isolated islets (Fig. 6 and Supplementary Fig. 10) suggest that cell-intrinsic phenomena are involved in the marked sexual dimorphism observed in regulated insulin secretion. Notably, the preserved differences in glucose tolerance in ovariectomized females (Fig. 5F) indicate that the presence of sex hormones is not required in the short term (2 weeks).

Potential Long-term Benefits of Combined GKA and Exendin-4 Treatment

In accordance with a previous report that explored the long-term (12 weeks) effects of exendin-4 and the earlier-generation GKA piragliatin (61), we show that acute treatments with dorzagliatin and exendin-4 exert an additive effect on glucose tolerance. In contrast with WT and Gck^KI/+ mice, these effects were not sexually dimorphic in db/db mice, possibly reflecting known interactions between the db allele and sex steroid metabolism (62).

The chronic effects of the drugs, alone or in combination, were not explored here in GCK mutant mice, and for only 1 week in db/db mice. Although microvascular complications are rare in patients with GCK mutations, a significantly higher level of subclinical retinopathy is observed (63).

Conclusion

The new mouse models of GCK-MODY and GCK-PNDM we have described here may be valuable for future drug screens as well as studies of diabetes complications. We have demonstrated the remarkable efficacy in the former of combining a GKA with a GLP-1 receptor agonist in these models as well as in db/db mice, likely reflecting a requirement for β-cell glucose metabolism for incretin action (28). Our findings suggest the potential therapeutic utility of this drug combination in selected patients with GCK-MODY and in patients with type 2 diabetes.

This article contains supplementary material online at https://doi.org/10.2337/figshare.30330106.

Funding Statement

G.A.R. was supported by a Wellcome Trust Investigator Award (WT212625/Z/18/Z); the MRC Programme (grant MR/R022259/1); Diabetes UK (grant BDA 16/0005485); National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases (grant R01DK135268); team grants from the Canadian Institutes of Health Research (CIHR/IRSC) (grant TDP-186358) and JDRF (now Breakthrough T1D) (grant 4-SRA-2023-1182-S-N); CRCHUM start-up funds; and an Innovation Canada John R. Evans Leader Award (CFI 42649). L.F.D.-S. was supported by a CIHR/IRSC postdoctoral fellowship (grant 489982). P.C. was supported by a Fonds de Recherche du Quebec Nature and Technology fellowship (grant 353239). G.O. was supported by a Fonds de Recherche du Quebec Santé fellowship (grant 333390). M.O.H. was supported by National Institutes of Health National Institute of Diabetes and Digestive and Kidney Diseases (grant R10DK110276 ) and American Diabetes Association (grant 7-22-IBSPM-06).