Susceptibility of Glucokinase-MODY Mutants to Inactivation by Oxidative Stress in Pancreatic β-Cells

The pCMV-Tag3C vector and site-directed mutagenesis kits were obtained from Agilent Technologies (Berkshire, U.K.); jetPEI-Hepatocyte was from MP Biomedicals (Solon, OH); GK activator (21) and its stereoisomer were from Axon MedChem (Groningen, the Netherlands); Lipofectamine 2000, monochlorobiamine (MCB), 4-amino-5-methylamino- 2′,7′-difluorofluorescein diacetate (DAF-FM diacetate), and 5-(and-6)-carboxy-2′,7′-dichlorodihydrofluorescein diacetate (H2DCFDA) were from Invitrogen (Paisley, U.K.); myc-antibody (910E) was from Santa Cruz Biotechnology (Santa Cruz, CA); fluorescein isothiocyanate (FITC)-conjugated anti-mouse antibody was from Jackson ImmunoResearch (West Grove, PA); GK antibody was from Mark Magnuson (Vanderbilt University School of Medicine, Nashville, TN) (22); FBPase2 vector was from Alex Lange (University of Minnesota Medical School, Minneapolis, MN); and the yellow fluorescent protein (YFP)-fragment vectors were from Tom Kerppola (University of Michigan Medical School, Ann Arbor, MI). All other reagents were from Sigma-Aldrich (Poole, U.K.).

An NH₂-terminal myc-tag was added to human pancreatic β-cell GK (wild-type [WT] and mutant) by excision of GK from the pGEX-3× plasmid (23) and insertion into the pCMV-Tag3C vector using BamHI/EcoRI sites. The C230S mutation (690–692, TGC to TCC) was introduced into the GK-WT and V62M-mutant using site-directed mutagenesis with specific primers (forward: 5′-GTG GGC ACG GGC TCC AAT GCC TGC TAC-3′; reverse: 5′-GTA GCA GGC ATT GGA GCC CGT GCC CAC-3′). Sequences were confirmed by DNA sequencing and protein translation by Western blotting.

Cells were cultured at 25 mmol/L glucose unless otherwise stated. MIN6 (p20–27) cells were cultured as described previously (13). H4IIE and Cos1 were cultured in Dulbecco’s modified Eagle’s medium containing 10% (vol/vol) fetal bovine serum, 75 mg/L penicillin, and 50 mg/L streptomycin. Cell monolayers in 24-well plates were transfected with 0.2–0.6 μg plasmid DNA and 0.6–1.8 μL Lipofectamine 2000 and cultured for 24 h.

Hepatocytes were isolated from male Wistar rats and were suspended in minimum essential medium containing 5 mmol/L glucose with 5% (vol/vol) newborn calf serum, 75 mg/L penicillin, and 50 mg/L streptomycin and seeded on gelatin-coated coverslips in 24-well plates (24). Three hours after attachment, hepatocytes were transiently transfected with 4 μg plasmid DNA and 12 μL jetPEI-Hepatocyte for 5 h and then cultured overnight. Cells were fixed with ice-cold acetone and immunostained using an anti-myc antibody (1:50 dilution) and a FITC-conjugated anti-mouse secondary antibody (1:100 dilution). Nuclei were stained with 10 μmol/L Hoechst dye 33342. Immunofluorescence was visualized using a Nikon E400 microscope (×63 oil immersion objective). Five images were taken in three independent experiments. The nuclear-to-cytoplasmic ratio for each image was determined by measurement of the mean pixel intensity of nuclei (including nucleoli) and cytoplasmic areas using Image ProPlus Analysis Software. Expression of results as a ratio corrects for drifts in fluorescence intensity between transfected cells.

GK activity in MIN6 and H4IIE cells was determined at 50 mmol/L glucose on 13,000 g supernatants as described previously (13). Glucose-phosphorylating activity was determined at 0.5 and 50 mmol/L glucose, representing low-K_m and total hexokinase activity, respectively, and GK activity was determined from the difference between these values. GK immunoreactivity was determined by SDS-PAGE using a 10% polyacrylamide gel and immunoblotting (13) with a GK antibody (22). Under these conditions, myc-tagged GK (52 kDa) could be distinguished as a distinct band from that of endogenous GK (50 kDa).

For determination of total glutathione (GSH) content, an enzyme-recycling assay was used (25). In brief, cells were extracted into 2.5% sulfosalicyclic acid and centrifuged at 13,000g. GSH content was measured on the supernatant by incubation with 110 mmol/L Na₂HPO₄, 70 mmol/L NaH₂PO₄.H₂O, 6.3 mmol/L EDTA, 300 μmol/L NADPH, 40 μM 5,5′-dithiobis-(2-nitrobenzoic acid), and 0.4 units/mL glutathione reductase at A₄₁₂. For GSH depletion assays, GSH was determined using MCB (26). Cells were cultured with diamide, sodium nitroprusside (SNP), or menadione (MEN) as stated for 1 h before a 10-min incubation with 75 μmol/L MCB in a Krebs-HEPES buffer containing 116 mmol/L NaCl, 20 mmol/L HEPES, 5 mmol/L KCl, 1 mmol/L CaCl₂, 1 mmol/L MgCl₂, 1 mmol/L NaPi, and 0.8 mmol/L Na₂SO₄, pH 7.4, and 1% FBS. NO and hydrogen peroxide (H₂O₂) content was determined in cells incubated with 30 μmol/L DAF-FM diacetate (NO) or 30 μmol/L H₂DCFDA (H₂O₂) in Krebs-HEPES buffer for 10 min before the addition of diamide, SNP, or MEN as stated for 1 h. GSH depletion, NO, and H₂O₂ were measured using a SpectraMax M5e Reader (Molecular Devices) at excitation (Ex)385 nm, emission (Em)478 nm (cutoff 475 nm); Ex495 nm, Em515 nm (cutoff 505 nm); and Ex488 nm, Em510 nm (cutoff 495 nm), respectively.

The bimolecular fluorescence complementation (BiFC) assay relies on the complementation of two nonfluorescent fragments of YFP when brought together by the interaction of proteins fused to each fragment (27). Sequences encoding amino acid residues of 1–155 or 156–264 of YFP, designated YN and YC, respectively (27), were generated by PCR and ligated into the pCMV-Tag3C vector. Human β-cell GK (WT or mutant) was fused to the NH₂ terminus of YN (GK-YN), and the bisphosphatase domain of PFK2/FBPase2 was fused to the NH₂ terminus of YC (FBPase2-YC). The GK interaction interface (SLKVWT, 5′-TCC CTG AAA GTC TGG ACT-3′) was deleted using site-directed mutagenesis with specific primers (forward: 5′-CAT CCG GTC TCA AGG CAT CAG CAG CCA CAT GAA GAG GAC CAT TC-3′; reverse: 5′-GAA TGG TCC TCT TCA TGT GGC TGC TGA TGC CTT GAG ACC GGA TG-3′). Sequences were confirmed by DNA sequencing and protein translation by Western blotting. For imaging analysis, Cos1 cells were transfected with 1 μg GK-YN, 1 μg FBPase2-YC, and 0.2 μg full-length monomeric red fluorescent protein using Lipofectamine 2000 for 24 h. After 3-h complementation, fluorescence was visualized in living cells using either a Nikon TE300 or a Nikon TE2000 inverted fluorescence microscope (×60/100 objective). Fluorescence was visualized at the following: YFP, Ex500 nm, Em535 nm; and monomeric red fluorescent protein, Ex584 nm, Em607 nm. Nuclei were stained with 10 μmol/L Hoechst dye 33342 (Ex350 nm, Em460 nm).

The GK mutants were localized to the previously published WT GK crystal structure in both the open and closed conformations (cocrystallized with glucose and an allosteric activator) (28) using the Swiss-PDB Viewer software.

This was performed by ANOVA followed by the Bonferroni test or χ² analysis using the Prism analysis program.

Approximately 80 of the 600 mutations identified in the Gck gene have been kinetically characterized in vitro using recombinant glutathione S-transferase-GK fusion proteins and the predicted threshold for GSIS calculated using mathematical modeling (9,12). A threshold for GSIS of >6.5 mmol/L glucose is consistent with the diabetic phenotype. In this study, we selected 11 GK-MODY mutants with a predicted GSIS <6.5 mmol/L for cellular analysis (Table 1). The mutants selected include V62M and G72R where there is discordance (29–31) in the effect of these mutations on enzyme kinetics. Possible explanations for the differences are discussed elsewhere (32). We used five PHHI mutants with a predicted GSIS of <5 mmol/L as controls.

GK has a bipartite structure with a flexible hinge allowing changes in conformation from super open /open (less active) to closed (more active) conformations upon binding of glucose (1–3). The location of the mutated residues in the superopen and closed conformations (28) was determined by structural modeling. The activating mutations (T65I, W99R, Y214C, V455M, and A456V) localize to the hinge region (Supplementary Fig. 1) as demonstrated previously (29). However, the 11 MODY mutants did not localize to a common location.

In hepatocytes, GKRP determines the cellular location and protein stability of GK (15–17). We first determined the subcellular location of the selected MODY mutants in hepatocytes. With the exception of V62M and G72R, which showed only cytoplasmic staining (Fig. 1) (13,29,30), all other MODY mutants showed nuclear localization and a similar nuclear-to-cytoplasmic ratio as WT (Fig. 1), indicating interaction with GKRP. We infer that the MODY phenotype for the majority of these mutants is unlikely to be a result of the dysregulation of subcellular location in hepatocytes.

We next determined the specific activity of the MODY mutants in β-cells using varying plasmid concentrations of myc-tagged GK in the MIN6 model (Fig. 2A). The specific activity of GK was determined from the slope of GK activity (corrected for endogenous activity) against corresponding GK immunoreactivity (Fig. 2B-I, Table 1, and Supplementary Table 1). Nine of 11 (A53S, V62M, G72R, H137R, S263P, G264S, M298K, E300K, and K414E) MODY mutants had a lower specific activity than WT (>35%), indicating enzyme inactivation (Fig. 2B-F and Table 1). None of the five PHHI mutants had a significantly lower specific activity than WT. T65I, W99R, and V455M had similar activity as WT; A456V had an activity 74% of WT; and Y214C had a specific activity twice that of WT (Fig. 2C and Table 1). χ² Analysis confirmed a significant difference in specific activity distribution between the MODY and PHHI mutants (P = 0.0075).

Several MODY mutants expressed lower levels of immunoreactive protein than WT (Fig. 2). We therefore determined protein stability during inhibition of protein synthesis with cycloheximide (12 h). There was no effect of cycloheximide on endogenous or WT GK protein (Fig. 3A and B). Five MODY mutants (A53S, H137R, S263P, E300K, and K414E) displayed marked protein instability compared with WT (>50%). Four MODY mutants (V26M, G72R, G264S, and M298K) and all five PHHI mutants showed similar protein stability as WT (Fig. 3C). Thus, with few exceptions, protein instability correlated with low specific activity (Fig. 3D).

To determine whether the low catalytic activity of the MODY mutants is specific to the pancreatic β-cell six mutants with a low specific activity in MIN6 were expressed in a hepatoma cell model (H4IIE). Mutants with a specific activity 30–40% of WT in MIN6 (V62M, G72R, and G264S) showed similar specific activity to WT in H4IIE (Table 1 and Supplementary Fig. 2), indicating that the instability of these mutants is cell type dependent. Mutants with a specific activity <20% of WT in MIN6 (S263P, M298K, and E300K) also showed low specific activity in H4IIE (Table 1 and Supplementary Fig. 2).

Pancreatic β-cells have lower antioxidant defenses than hepatocytes and are particularly vulnerable to cellular damage by oxidative stress (33,34). The total GSH content of MIN6 was 80% lower than in hepatocytes or H4IIE (Fig. 4A). We therefore tested whether depletion of GSH using diamide and increased oxidative stress induced with SNP or MEN further compromises the activity of MODY mutants (Fig. 4B–D). Four GK-MODY mutants (V62M, G72R, S263P, and G264S) with decreased specific activity in MIN6 but sufficient catalytic activity to enable reliable measurement were selected for these studies. Oxidative stress induced with MEN but not SNP decreased the activity of WT enzyme by 20% (Fig. 4E and F), and the PHHI mutant W99R showed similar activity to WT. However, all four MODY mutants showed significantly greater inactivation by SNP and MEN (Fig. 4E and F), indicating increased inactivation by oxidative stress. There was no detectable difference in immunoreactive protein under these conditions (data not shown). Oxidative stress induced by MEN also decreased the activity of endogenous GK in MIN6, which was prevented by preincubation with the antioxidant N-acetyl cysteine (Fig. 4G). The inactivation of endogenous GK by reactive oxygen species was partially prevented by 24-h treatment with a pharmacological GK activator (GKA-R) that favors the closed (active) conformation (Fig. 4H).

GK is extremely sensitive to thiol group modification by thiol reactive agents such as alloxan, with consequent inactivation of catalytic activity (35–37). This inactivation is either reversible with dithiothreitol (DTT) or irreversible (38,39). To investigate the role of thiol group modification in the catalytic instability of the mutants, the effects of varying concentrations of alloxan and DTT was determined. DTT caused a twofold increase in the activity of WT (Fig. 5A) and the PHHI mutant W99R (Fig. 5B). However, it caused a significantly greater fractional activation of all four MODY mutants (Fig. 5B), indicating that the low specific activity is in part a result of reversible modification of thiol residues. The four MODY mutants, but not W99R, showed a greater sensitivity to inactivation by alloxan as shown by the lower IC₅₀ values (Fig. 5C and D), further confirming enhanced sensitivity to oxidation (37,38). Because cysteine 230 (Cys230) is highly susceptible to alloxan modification (38), we tested the effect of mutating this residue either alone or in combination with V62M in the alloxan sensitivity assay. The C230S mutant was less sensitive to alloxan than WT (Fig. 5E) as reported previously (38). However, the double mutant C230S-V62M had similar sensitivity to alloxan and oxidative stress as V62M, although it was less sensitive to SNP inactivation (Fig. 5E–G). This indicates that residues other than Cys230 are involved in the sensitivity of V62M to oxidative stress.

In the pancreatic β-cell, GK activity is regulated by binding to PFK2/FBPase2 (20). To determine whether this interaction protects GK from oxidative inactivation, PFK2/FBPase2 was coexpressed with WT GK in MIN6. PFK2/FBPase2 protected GK from inhibition by alloxan, with a shift in IC₅₀ from 1.1 to 3.5 mmol/L (Fig. 6A) and reversal of the inhibition by 0.5 mmol/L alloxan (Fig. 6B). We next investigated the interaction of MODY mutants with PFK2/FBPase2 using the BiFC assay (27). Because this assay is critically dependent on the level of protein expression, the GK-PFK2/FBPase2 interaction could only be tested on mutants with minimal effect on protein stability, i.e., V62M, G72R, and G264S (Fig. 6D). A direct interaction between WT GK and the bisphosphatase domain FBPase2 was evident from YFP staining (Fig. 6C). Total loss of interaction between GK and a FBPase2 variant lacking the GK binding domain (SLKVWT) confirmed the specificity of the signal with WT FBPase2. Three of three MODY mutants tested (V62M, G72R, and G264S) showed decreased interaction with FBPase2, while the PHHI mutant W99R showed similar interaction to WT (Fig. 6D). Similar levels of protein expression of the mutants were confirmed by immunoblotting (Fig. 6D). These data suggest that GK binding to PFK2/FBPase2 may stabilize GK activity in β-cells.

The mild diabetic phenotype of GK-MODY is remarkably homogeneous despite the variable effects of causative mutations on enzyme kinetics (9,11,12). In cases where the mutation is markedly inactivating (<10% of WT), the mild phenotype is explained by compensatory upregulation of the WT allele induced by the mild hyperglycemia (40). However, alternate explanations are required for the diabetic phenotype when the mutation has minimal influence on enzyme kinetics (Table 1). Possible mechanisms include protein instability that manifests either as thermolability in vitro (29,30,41), decreased activity/protein in cell based assays (13,42) or in vivo (43), or decreased interaction between liver GK and GKRP with consequent decreased hepatic GK protein (43). The critical role of pancreatic β-cell GK in glucose homeostasis is evident from the similar phenotype of the murine β-cell GK knockout to the global knockout (4,5). In this study, we asked whether GK-MODY mutants that have minimal effect on enzyme kinetics in vitro show abnormal activity in a β-cell environment.

Two key findings emerged from this study. First, the majority of the MODY mutants but none of the PHHI mutants showed enzyme inactivation in β-cells. This inherent instability may contribute to the diabetic phenotype. Second, although there was heterogeneity among the MODY mutants in the compromised specific activity and protein stability, a subset of MODY mutants with decreased specific activity in β-cells but not hepatoma cells was more susceptible to inactivation by oxidative stress. These findings indicate the importance of the β-cell environment and oxidative stress in the MODY phenotype (Table 2).

Two of the 11 MODY mutants (E70K and R275C) studied in the β-cell model showed similar specific activity to WT. Thus, other explanations are required for the diabetic phenotype in these two cases. The remaining nine MODY mutants (A53S, V62M, G72R, H137R, G264S, S263P, M298K, E300K, and K414E) had a specific activity less than 35% of WT when expressed in MIN6. The majority of these mutants showed either lower protein levels compared with WT (A53S, V62M, G72R, H137R, S263P, G264S, and E300K) or marked (A53S, H137R, S263P, E300K, and K414E) protein instability. Decreased enzyme activity/protein may be sufficient to explain the diabetic phenotype since a mouse model homozygous for K414E died shortly after birth, consistent with a severe β-cell deficit of GK protein (43). Three MODY mutants with very low specific activity (<20% of WT) in β-cells (S263P, M298K, and E300K) also showed low specific activity in hepatoma cells. These mutants were shown previously to be thermolabile in vitro (9,41). Three MODY mutants with low specific activity (30–40% of WT) in β-cells (V62M, G72R, and G264S) showed comparable specific activity with WT in the hepatoma model. These mutants were shown previously to be relatively thermostable in vitro (29,30). G264S was identified in permanent neonatal diabetes mellitus in conjunction with a GK splice site mutation (IVS8+2T>G) and was associated with marked hyperglycemia, consistent with severely compromised GK activity (44). The lower stability of G264S in the β-cell model compared with hepatoma cells supports the validity of the β-cell model for predicting the phenotype of MODY mutants and highlights the importance of the β-cell environment in modulating GK activity.

Pancreatic β-cells have lower antioxidant defenses than hepatocytes and are more vulnerable to oxidative stress (33,34). Three sets of evidence from this study support a role for oxidative stress as a predisposing factor to the instability of several MODY mutants (Table 2). First, three of six MODY mutants showed enzyme inactivation in β-cells but not hepatoma cells. Second, these mutants showed greater inactivation than WT by oxidative stress. Third, the low specific activity of these mutants was partially reversed by DTT. This indicates both reversible and irreversible inactivation of the enzyme. GK is more sensitive to oxidation of thiol groups than other hexokinase isoforms (35,36). Oxidation of thiol groups locks the enzyme in a restricted conformation, thereby affecting enzyme activity. This is in part a result of the formation of disulfide bonds (37,38). Tippett and Neet (35) proposed that physiological changes in GSH concentration may regulate GK activity through conformational changes of the enzyme induced by modification of thiol residues. The finding that endogenous GK is inactivated by elevated oxidative stress and that this inactivation is attenuated by a GK activator that favors the closed (active) conformation is consistent with this hypothesis (35). This supports a model whereby endogenous GK activity is inhibited by thiol oxidation in conditions of GSH depletion and/or reactive oxygen species production. The sensitivity of GK to posttranslational modifications induced by cell stress is supported by a recent study reporting irreversible tyrosine nitration of GK following chronic ethanol exposure (45).

The increased susceptibility of MODY mutants to oxidation may be explained by loss of binding to a protective protein. The GK interaction partner PFK2/FBPase2 binds to the closed (active) conformation of GK at elevated glucose and prevents the transition to the superopen (less active) conformation, thereby increasing the enzyme activity (20,46,47). However, another explanation could be protection of GK from inactivation through modification of thiol residues. Two sets of evidence support this mechanism. First, overexpression of PFK2/FBPase2 protected WT GK from inactivation by alloxan. Second, MODY mutants with decreased specific activity and increased sensitivity to oxidative stress showed decreased interaction with PFK2/FBPase2. It is noteworthy that overexpressed PFK2/FBPase2 did not affect the activity of endogenous GK (results not shown) indicating a saturating effect of endogenous PFK2/FBPase2 as reported previously (13).

A key finding from the current study was the higher specific activity of the Y214C PHHI mutant relative to WT. Enzyme kinetics and tryptophan fluorescence measurements indicate that Y214C adopts a more compact conformation than WT (32). Because other PHHI mutants also have a more compact conformation than WT, the unique stability of Y214C suggests that small changes in the conformation of Y214C relative to other PHHI mutants or other additional factors may have a major influence on enzyme stability. Regardless of the explanation, the lower stability of WT relative to Y214C is consistent with a role for thiol status in the control of endogenous GK activity as proposed by Tippett and Neet (35). This mechanism may be altered by the Y214C mutation. This would explain the severe phenotype of the patient with the Y214C mutation (48), compared with the much milder phenotype of the A456V mutation (9,49), despite similar enzyme kinetics (Table 1).

In summary, the current study shows that several GK-MODY mutants that cause diabetes despite minimal influence on enzyme kinetics and show compromised catalytic stability and/or protein stability in β-cells but to varying degrees. This variation may reflect the complex regulation of GK activity by several interacting proteins, which may alter both subcellular localization and GK stability (18–20). Common to several MODY mutants is increased susceptibility to oxidative stress compared with WT or PHHI mutants, which may be in part explained by decreased binding to PFK2/FBPase2. The elevated stability of the Y214C mutant and inactivation of endogenous GK activity by oxidation supports a role for the regulation of endogenous WT GK through modulation of thiol status as a potential physiological mechanism for the control of GK activity.

C.A. is an RD Lawrence Research Fellow funded by Diabetes UK (07/0003674). This work received additional support from the European Foundation for the Study of Diabetes (Albert Renolds travel grant), European Molecular Biology Organization (short-term travel fellowship ASTF 171.00-2007), Diabetes UK (equipment grant 08/0003772), and the Royal Society (project grant RG080223).

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

K.S.C. researched data. F.M.M. provided GK-MODY mutants and reviewed and edited the manuscript. L.A. contributed to discussion and reviewed and edited the manuscript. C.A. researched data and wrote the manuscript.

The authors thank Professor Tom Kerppola (University of Michigan) for the YFP-fragment constructs and for laboratory/technical support. The authors also thank Dr. Jun-ichi Miyazaki (Kumamoto University Medical School, Japan) for the MIN6 cells, Professor Alex J. Lange (University of Minnesota Medical School) for the PFK2/FBPase2 construct, and Professor Mark A. Magnuson (Vanderbilt University School of Medicine) for the GK antibody.