ATP-Sensitive K⁺ Channel Signaling in Glucokinase-Deficient Diabetes Free

Previously, a global neuroendocrine knockout of GK (nGK) was achieved by removal of upstream promoter elements (15). This was expected to result in abolition of GK expression in β-cells and also in gut and brain (15). Knockout of the Kir6.2 gene was achieved by targeted introduction of the neomycin-resistant gene into the coding region of Kir6.2 (25). Heterozygous nGK^+/− mice on a C57/BL6J background (gift from Dr. Yasuo Terauchi) (15) were bred to Kir6.2^−/− on an outbred background (gift from Dr. Susumo Seino) (25) to generate double-heterozygous F1 mice. F1 mice were backcrossed to each other to generate F2 mice of expected genotype. nGK^−/−Kir6.2^+/+ and nGK^−/−Kir6.2^+/− were not identified in the litters, probably because of perinatal lethality. For all experiments described, we backcrossed F2 heterozygous nGK^+/− mice (either on Kir6.2^+/+ or Kir6.2^−/− backgrounds) to the same genotypes, thus generating F3 litters of three genotypes (nGK^+/+, nGK^+/−, and nGK^−/− on Kir6.2^+/+ or Kir6.2^−/− backgrounds). Mice were typed for nGK and Kir6.2 genes as previously described by Terauchi et al. (15) and Miki et al. (25).

Previously, Kir6.2[AAA] mice expressing a dominant-negative Kir6.2 mutation (¹³²GFG¹³⁴→AAA) mutation under control of the rat insulin 1 promoter were generated (29). For experiments shown in Fig. 7, these Kir6.2[AAA] mice (on C57BL6J background) were bred to nGK^+/− mice, also on C57BL6J background, to generate the relevant crosses.

To control for effects of genetic background, comparisons between genotypes were made with littermates in every experiment, and n values of at least three are reported in each paired genotype. Thus, it is reasonable to conclude that any effects of nonmutated chromosomes will be randomized within all genotypes.

Experiments using animals were performed in accordance with the relevant laws and institutional guidelines and were approved by the Washington University animal studies committee. Anesthetized mice were killed by cervical dislocation. Pancreata were removed and injected with Hank’s balanced salt solution containing type XI collagenase (0.5 mg/ml), pH 7.4. (Sigma, St. Louis, MO). Pancreata were digested for 7 min at 37°C then hand-shaken and washed three times in cold Hank’s solution (29). Islets were isolated by hand under a dissecting microscope and pooled. Islets were maintained overnight in CMRL-1640 (5.6 mmol/l glucose) supplemented with FCS (10%), penicillin (100 units/ml), and streptomycin (100 mg/ml).

Blood glucose was assayed using a glucose dehydrogenase–based enzymatic assay and quantitated using a Glucometer Elite XL meter (Bayer, Elkhart, IN). Blood insulin levels were assayed on 15 μl of mouse serum using an enzyme-linked immunosorbent assay kit with a rat insulin standard according to manufacturer’s procedure (Crystal Chem, Downers Grove, IL). Intraperitoneal glucose tolerance tests were performed on 5-week-old mice after a 16-h fast. Animals were injected intraperitoneally with glucose (1 g/kg). Blood was isolated from the tail vein at the times indicated and assayed for glucose as described above. Intraperitoneal insulin tolerance tests were performed on 4-month-old mice after a 6-h fast. Animals were injected intraperitoneally with insulin (0.5 unit/kg). Blood was isolated from the tail vein at the times indicated and assayed for glucose as described above.

After overnight incubation, pancreatic islets (10 per well in 12-well plates) were preincubated in glucose-free Dulbecco’s modified Eagle medium (DMEM) supplemented with 3 mmol/l of glucose for 2 h. The release assay was initiated by supplementing the DMEM glucose-free media with d⁺-glucose (1, 7, 16.7, or 23 mmol/l) as indicated. For amino acid stimulation of insulin release, Krebs-Ringer buffer supplemented with d-glucose and glutamine (10 mmol/l) was used in place of DMEM. Islets were incubated for 60 min at 37°C, and medium was removed and assayed for insulin content using a rat insulin radioimmunoassay (Linco, St. Charles, MO). Each experiment was repeated in triplicate.

Data are presented as means ± SE. Differences among the four genotypes were tested using ANOVA and post hoc Duncan’s test. A difference was defined as significant when P < 0.05. Nonsignificant differences are not indicated.

Of 57 live births from cross-breeding of heterozygous nGK^+/− mice on the Kir6.2 wild-type background, only 2 mice were positively genotyped as nGK^−/− (Fig. 2A). Both were found dead within the 1st week (genotyped postmortem) after birth (Fig. 2B). Conversely, on the Kir6.2 knockout background, nGK^−/− pups were born at the expected ratio (1:2:1 for nGK^+/+Kir6.2^−/−, nGK^+/−Kir6.2^−/−, and nGK^−/−Kir6.2^−/−, respectively) (Fig. 2A). These double-knockout mice survived much longer than nGK^−/− on the Kir6.2 wild-type background, with a median survival time of ∼10 days, and ∼30% surviving past weaning (21 days) (Fig. 2B). As is shown in Fig. 2C, double-knockout nGK^−/−Kir6.2^−/− mice were significantly smaller than both nGK^+/−Kir6.2^−/− and nGK^+/+Kir6.2^−/− mice at 4 weeks of age. By 5 weeks of age, nGK^−/−Kir6.2^−/− mice weighed ∼50% (4.78 ± 0.17g) that of nGK^+/+Kir6.2^−/− (8.34 ± 0.24g) or nGK^+/−Kir6.2^−/− (9.02 ± 0.17g.) littermates. Nevertheless, it is striking that these double-knockout mice, which completely lack β-cell glucokinase activity, can survive beyond weaning.

On this Kir6.2^−/− background, fasting blood glucose was only slightly higher in nGK^−/− than in nGK^+/+ or nGK^+/− mice (Fig. 3A). However, fed glucose (Fig. 3B) was dramatically higher in nGK^−/− mice, suggesting that these mice either still do not secrete enough insulin to lower blood glucose values or that they develop insulin resistance (see below).

Heterozygous nGK^+/− mice were born at close to expected ratios on both the Kir6.2^+/+ and Kir6.2^−/− backgrounds (Fig. 2A) and developed apparently normally. As previously reported, on the wild-type Kir6.2^+/+ background, heterozygous nGK^+/− mice show nonprogressive mild diabetes, as in human MODY-2. At birth, the blood glucose levels of nGK^+/+ and nGK^+/− mice were similar (3.5 ± 0.4 and 3.9 ± 0.2 mmol/l, respectively), but by 12 weeks, nGK^+/−Kir6.2^+/+ mice showed a twofold increase in both fasting and fed blood glucose compared with nGK^+/+Kir6.2^+/+ mice (Fig. 3A and B).

By contrast, on the Kir6.2 knockout background, overt diabetes, as evidenced by elevated fasting glucose levels, is ameliorated in nGK^+/−Kir6.2^−/− mice, although fed glucose levels are elevated (Fig. 3A and B). Plasma insulin, measured under fed conditions, was slightly elevated in nGK^+/− mice versus nGK^+/+ on both Kir6.2^+/+ and Kir6.2^−/− backgrounds (Fig. 3C).

To characterize physiological responses to glucose load, surviving mice underwent intraperitoneal glucose tolerance testing at 5 weeks of age. As previously reported, blood glucose before and after glucose load was significantly higher in nGK^+/− mice on the Kir6.2^+/+ background, and Kir6.2^−/− mice showed a very mild impairment in glucose tolerance with respect to Kir6.2^+/+ mice (Fig. 4). Importantly, on the Kir6.2^−/− background, there was dramatically improved glucose tolerance of nGK^+/− mice (Fig. 4), to the extent that they were not significantly different from nGK^+/+ mice on the same background.

The glucose-lowering effect of intraperitoneal insulin injection (0.5 unit/kg) was assessed using insulin tolerance tests on 6-h fasted mice. nGK^+/+Kir6.2^−/− mice were slightly more sensitive to insulin than nGK^+/+Kir6.2^+/+ mice (Fig. 5), similar to a previous report (25). Insulin sensitivity was significantly enhanced in nGK^+/− mice on the Kir6.2^+/+ background, whereas nGK^+/−Kir6.2^−/− mice showed intermediate sensitivity (Fig. 5). These results exclude enhanced insulin sensitivity as a possible mechanism for the improved glucose tolerance of nGK^+/− mice in the absence of K_ATP, and they point to enhanced insulin secretion as the more likely mechanism (see below).

To gain further insight into the cellular basis for differential phenotypes of double-knockout animals, islets were isolated from each of the different genotypes and phenotypically examined. Morphologically, islet appearance and size were not different between the various genotypes (for example, islet diameter was 175.5 ± 8.1 μm for nGK^+/−Kir6.2^+/+ and 183.3 ± 7.3 μm for nGK^+/−Kir6.2^−/−, n = 35 islets). Glucose-stimulated insulin secretion (GSIS) was assessed in isolated islets from all five available genotypes. As shown previously, the response of nGK^+/− islets was mildly reduced compared with nGK^+/+, on the Kir6.2^+/+ background (15). On the Kir6.2^−/− background, GSIS was greatly reduced in the presence or absence of nGK (Fig. 6A), and there was no enhancement of release from nGK^+/− or nGK^−/−, relative to the release from the same nGK genotypes on the Kir6.2^+/+ background. These data thus reiterate the perplexing result that Kir6.2^−/− (25) and SUR1^−/− (26) mice are mildly glucose tolerant, yet have only minimal GSIS. Moreover, these results do not help to explain the enhanced glucose tolerance and survival of nGK-deficient mice on this Kir6.2^−/− background. Recent studies have begun to shed light on the potential mechanisms in SUR1^−/− mice (30–32). In particular, glutamine has been shown to preferentially stimulate K_ATP-independent secretion from SUR1^−/− islets (31). This effect has not been examined in Kir6.2^−/− islets, but if the phenotypes of SUR1^−/− and Kir6.2^−/− indeed both result from loss of K_ATP, we would expect the same K_ATP-independent stimulation of Kir6.2^−/− islets. As shown in Fig. 6B, glutamine markedly stimulates secretion from both nGK^+/+ and nGK^+/− islets on the Kir6.2^−/− background relative to the wild-type background. Given likely physiological levels of glutamine, this may well cause enhanced insulin secretion in vivo and explain the improved glucose tolerance of nGK^+/− on the Kir6.2^−/− background (31) (see below).

Although the above data are consistent with a specific loss of K_ATP channels in β-cells being responsible for abrogation of the nGK-deficiency phenotype, it is a caveat that the Kir6.2 knockout is global. We previously generated β-cell–specific dominant-negative Kir6.2[AAA] mice using an insulin promoter to drive transgene expression (29). Kir6.2[AAA] mice, which lack K_ATP in ∼70% of β-cells, hypersecrete insulin and have slightly enhanced glucose tolerance (29). In the current study, we bred these mice with nGK^+/−Kir6.2^+/+ mice to generate double-heterozygous (nGK^+/−Kir6.2[AAA]) mice (see research design and methods). As shown in Fig. 7, these mice have a marked improvement in glucose tolerance compared to nGK^+/− alone. Although the improvement is not as dramatic as in the mice completely lacking K_ATP (Fig. 4), this result is consistent with β-cell K_ATP channels indeed being the major players in rescuing the nGK-deficiency phenotype.

Heterozygous loss-of-function mutations in the glucokinase (GK) gene are the most common cause of MODY-2, a relatively mild form of type 2 diabetes. Conversely, homozygous or compound heterozygous GK deficiency mutations underlie permanent neonatal diabetes, a more severe perinatal diabetes (13). Mimicking human permanent neonatal diabetes, homozygous neuroendocrine GK knockout mice die shortly after birth of severe diabetes (15,33). Consistent with a MODY-2 phenotype, mice lacking one allele of neuroendocrine GK exhibit mild hyperglycemia with normal basal glucose turnover rates, but they have impaired glucose tolerance and insulin secretion during hyperglycemia (2,15).

Glucokinase deficiency could affect insulin secretion by altering electrical and/or nonelectrical downstream consequences of glucose metabolism (Fig. 1). To date, however, the relative importance of electrical (K_ATP-dependent) versus nonelectrical (K_ATP-independent) consequences of neuroendocrine glucokinase deficiency remains unclear. K_ATP channels are under tight metabolic control, and any perturbation of metabolism that leads to decreased generation of ATP is expected to have pronounced inhibitory effects on excitation-secretion coupling because of enhanced K_ATP activity. Studies of the electrical activity of nGK-deficient mice indeed support the idea that reduced glycolytic metabolism specifically causes failure of K_ATP channels to close in response to glucose, contributing to the observed reduction in insulin secretion (17). Consistent with this conclusion, it was reported that oral sulfonylurea treatment could suppress perinatal death in nGK^−/− mice (15), presumably by restoring insulin release. Glucokinase is critical for liver regulation of glucose production (34). This could contribute to the whole animal phenotype in complex ways, and for this reason, we examined a tissue-specific model in which the liver-specific isoform of glucokinase is expressed normally, and only the neuroendocrine-specific isoform is deleted (15). We specifically examined the consequences of nGK deficiency on the wild-type and K_ATP-null backgrounds to elucidate the relative importance of the two signaling pathways. Perinatal lethality of nGK^−/− was absolute on the Kir6.2^+/+ background; only 2 pups were detected of an expected ∼16 pups, from heterozygous nGK^+/−Kir6.2^+/+ backcrosses (Fig. 2), and both died within 5 days. In contrast, double-knockout mice (nGK^−/−Kir6.2^−/−) were born at the expected ratio. They exhibited low body weight, severe diabetes, and died prematurely, but perinatal lethality is clearly abrogated (Fig. 2A and B). Amelioration of the nGK^−/− phenotype on the Kir6.2^−/− background can be explained by an increased, though still insufficient, level of circulating insulin. This striking demonstration of the nullifying effect of removal of the K_ATP “brake” in vivo is further supported by the results with heterozygous nGK^+/− mice. On the Kir6.2^+/+ background, they have markedly elevated fed and fasting glucose levels, and they are markedly glucose intolerant. By contrast, on the Kir6.2^−/− background, nGK^+/− mice have near-normal fasting glucose levels and similar glucose tolerance to nGK^+/+Kir6.2^−/− animals. The abolition of K_ATP significantly ameliorates the “MODY-2” phenotype of the nGK^+/− mice.

The initial prediction that, by removing the electrical defect, the diabetic consequences of nGK deficiency should be significantly avoided on the Kir6.2^−/− background was thus supported. However, the mechanistic consequences of the genetic alterations are not absolutely clear. The naive expectation of mice lacking K_ATP channels was a persistent hyperinsulinemic phenotype because of unregulated secretion, but several studies have now shown that this expectation is not met (25–27). For unexplained reasons (but see below), Kir6.2^−/− mice are normoglycemic, with slightly impaired glucose tolerance and reduced GSIS. Similarly, although glucose tolerance was greatly improved in nGK^+/− mice on the Kir6.2^−/− background, GSIS in isolated islets was still notably reduced (Fig. 6A). However, as has recently been shown, the sensitizing effect of amino acids, in particular glutamine (30,31), is much greater in islets that lack K_ATP (SUR1^−/−). Similarly, we find that Kir6.2^−/− islets can show enhanced insulin secretion at physiological glucose concentration (compared with Kir6.2^+/+ mice) in the presence of stimulatory glutamine (Fig. 6B). This enhancement is seen in both nGK^+/+ and nGK^+/− islets on this background. We suggest that, in vivo, ambient glutamine levels may lead to such an enhancement of insulin secretion and that, in part, this underlies the abrogating effect of K_ATP deficiency on nGK-deficiency (31).

Because we are examining a global neuroendocrine glucokinase deficiency, it is conceivable that the effects of Kir6.2 knockout are mediated through extra-pancreatic consequences, although insulin tolerance tests (Fig. 5) show that nGK^+/−Kir6.2^−/− mice are no more insulin sensitive than nGK^+/−Kir6.2^+/+ mice, excluding enhanced insulin sensitivity as an explanation for the rescued phenotype. Further evidence for a pancreatic basis for the Kir6.2-deficient rescue of nGK deficiency would be a β-cell–specific K_ATP knockout. A tissue-targeted knockout has not been generated, although we have generated an insulin promoter–driven dominant-negative Kir6.2[AAA] mouse (29) that lacks K_ATP in ∼70% of its β-cells (29). The demonstration of marked improvement in glucose tolerance in double-heterozygous (nGK^+/−Kir6.2[AAA]) mice, compared with nGK^+/− alone (Fig. 7), provides further support for β-cell–specific K_ATP channels indeed being the major players in rescuing the nGK-deficiency phenotype.

The significant reduction in fasting glucose, slight yet significant decrease in fed glucose concentration, and marked enhancement in glucose tolerance in nGK^+/−Kir6.2^−/− mice with respect to nGK^+/−Kir6.2^+/+ mice suggests that, at least in part, the MODY-2 phenotype is abrogated on the K_ATP knockout background, highlighting the importance of electrical signaling in the regulation of insulin secretion. Decreased insulin secretion in response to glucose from nGK^−/−Kir6.2^−/− or nGK^+/−Kir6.2^−/− islets recapitulates the response observed in nGK^+/+Kir6.2^−/− islets, thus indicating that the impaired GSIS is due to the Kir6.2 knockout phenotype and not due to glucokinase deficiency. Most patients with heterozygous GK mutations do not require pharmacological treatment, and the majority of cases are managed with diet alone. However, during pregnancy, women with GK mutations are often treated with insulin to maintain normal blood sugar levels, resulting in babies that are large for gestational age because of the effect of insulin on fetal growth (35). In these cases, sulfonylurea treatment could presumably be a valid alternative treatment, given the normalizing effect of K_ATP deficiency that we now demonstrate, because nGK^+/−Kir6.2^+/+ transgenic mice do respond to glibenclamide treatment in vitro (15).

Taken together, the results from nGK-deficient and K_ATP-deficient crosses show a marked improvement in survivability and glucose tolerance of the former on the K_ATP null background. Neonatal lethality of nGK^−/−Kir6.2^+/+ mice is avoided on the Kir6.2^−/− background, demonstrating that this consequence is attributable to loss of metabolic coupling between glucose and insulin secretion, probably through electrical signaling. As with the now dramatically improved treatment options for K_ATP-induced permanent neonatal diabetes (21,24), we suggest that there may be a role for sulfonylurea therapy in permanent neonatal diabetes resulting from glucokinase deficiency (13). A significant number of nGK^−/−Kir6.2^−/− mice survived beyond weaning. This should also allow more extensive studies of the nonelectrical consequences (which are still ultimately lethal) of total glucokinase deficiency. Such studies will have the potential to inform and further improve treatment options for diabetes resulting from glucokinase deficiency.

This work was supported by a mentor-based minority postdoctoral fellowship (to M.S.R.) from the American Diabetes Association and Takeda Pharmaceuticals, by the National Institutes of Health Diabetes Research and Training Center (DK20579) at Washington University, and by National Institutes of Health grant DK69445 (to C.G.N.).

Kir6.2-deficient mice (25) and nGK-deficient mice (15) were kindly provided by Dr. Susumo Seino and Dr. Yasuo Terauchi, respectively.