Glucokinase (GK) is a glycolytic key enzyme that functions as a glucose sensor in the pancreatic β-cell, where it governs glucose-stimulated insulin secretion (GSIS). Heterozygous inactivating mutations in the glucokinase gene (GCK) cause a mild form of diabetes (maturity-onset diabetes of the young [MODY]2), and activating mutations have been associated with a mild form of familial hyperinsulinemic hypoglycemia. We describe the first case of severe persistent hyperinsulinemic hypoglycemia due to a “de novo” mutation in GCK (Y214C). A baby girl presented with hypoglycemic seizures since the first postnatal day as well as with inappropriate hyperinsulinemia. Severe hypoglycemia persisted even after treatment with diazoxide and subtotal pancreatectomy, leading to irreversible brain damage. Pancreatic histology revealed abnormally large and hyperfunctional islets. The mutation is located in the putative allosteric activator domain of the protein. Functional studies of purified recombinant glutathionyl S-transferase fusion protein of GK-Y214C showed a sixfold increase in its affinity for glucose, a lowered cooperativity, and increased kcat. The relative activity index of GK-Y214C was 130, and the threshold for GSIS predicted by mathematical modeling was 0.8 mmol/l, compared with 5 mmol/l in the wild-type enzyme. In conclusion, we have identified a de novo GCK activating mutation that causes hyperinsulinemic hypoglycemia of exceptional severity. These findings demonstrate that the range of the clinical phenotype caused by GCK mutations varies from complete insulin deficiency to extreme hyperinsulinemia.
Failure of the pancreatic β-cell to suppress insulin secretion during hypoglycemia is the pathophysiological basis of persistent hyperinsulinemic hypoglycemia of infancy (PHHI). Mutations in the β-cell sulfonylurea receptor (SUR1) (1), the inward-rectifying potassium channel (Kir6.2) (2), along with mutations in glucokinase (GCK) (3,4), glutamate dehydrogenase (5), and short-chain 3-hydroxyacil-CoA dehydrogenase enzyme (6) genes are responsible for 50% of the cases of PHHI. Due to a high risk of permanent brain damage in these patients (7), an early diagnosis and efficient treatment are crucial. The clinically most severe cases are often due to mutations in the ATP-sensitive K+ (KATP) channel genes, mainly SUR1 (8). These patients often do not respond to treatment with diazoxide and in many cases require 95% subtotal pancreatectomy to prevent recurrent hypoglycemia (9).
Glucokinase (GK) is a glycolytic enzyme that functions as a “glucose sensor” in the pancreatic β-cell and governs glucose-stimulated insulin secretion (GSIS) (10). Inactivating mutations in GCK lead to the development of diabetes (11,12) and activating mutations to hyperinsulinism (3,4). Until now all activating mutations in the GCK that are described as causative of PHHI present a mild clinical phenotype with an excellent response to pharmacological treatment (3,4,13). In this report, we describe the first case of very severe PHHI due to a “de novo” mutation in GCK, refractory to both surgical and medical treatment and with a totally different clinical phenotype compared with other previously published PHHI cases linked to GCK. Our findings are important because they demonstrate that a mutation in GCK can also cause severe hyperinsulinism and that biochemical and structural changes in the GK protein are also responsible for the heterogeneity of the clinical phenotype of PHHI patients with activating GCK mutations.
We found a novel missense mutation in exon 6 of the GCK gene that resulted in the substitution of cysteine for tyrosine (Y214C). The proband was heterozygous for this mutation, and this variant was not found in any other family members or in 100 healthy control chromosomes. Since the paternity tests confirmed that the parents of the patient were her “biological parents,” the mutation was considered as de novo. The structural analysis indicated that Y214C is not located close to the glucose or ATP binding sites, but it is located in a loop in the suggested allosteric activator domain of GK (3,4,13), and the results of the kinetic analysis of the purified enzymes suggested that GK-Y214C was the cause of severe hyperinsulinism in our patient. GK-Y214C showed a sixfold higher affinity for glucose (S0.5) and a twofold higher turnover rate (kcat) when compared with wild-type GK (Table 1), and the cooperativity was almost completely lost. GK-Y214C also showed the highest activity index (130) of all naturally occurring activating GCK mutations described (3,4,13). This catalytic activation of GK-Y214C leads to an enzyme with a very low threshold for GSIS (∼0.8 mmol/l, compared with 5 mmol/l for wild-type GK). Moreover, our kinetic results confirmed those observed (14) in the artificially made mutation, Y214A.
Recently, an allosteric activator of GK, RO-28-167, was described (15). Interestingly, the effects of GK-Y214C on kcat and S0.5 were similar to the effects of the GK activator RO-28-167 (Table 1), reinforcing the idea of the existence of an allosteric activator site in GK. Indeed, the crystal structure of GK recently developed (16) demonstrates that residue Y214 is located at the allosteric activator site and is included in the binding site of the synthetic activator used to crystallize the active form of GK. Interestingly, the decrease in the threshold for GSIS induced by the activator is less than that induced by Y214C (3 vs. 0.8 mmol/l, respectively). This is very important because to avoid frequent hypoglycemic episodes in the patient, the decrease in the GSIS threshold induced by the GK activator needs to be moderate. GK also seems to act as a glucose sensor in glucagon-producing pancreatic α-cells, in hepatocytes, in some hypothalamic neurons, and in some endocrine enterocytes, forming a neuroendocrine-signaling network that must be intact to maintain glucose homeostasis (17). Therefore, it is reasonable to assume that this GK signaling network was also affected in our patient. Both structural and functional changes in the GK-Y214C protein could explain the severity of hyperinsulinism and the differences in the clinical phenotype of our patient compared with other patients with activating mutations (3,4,13).
In contrast to the previously reported patients with activating GCK mutations, our patient was relatively large for gestational age. Macrosomia was presumably explained by the severe fetal hyperinsulinism consecutive to the Y214C mutation. If we consider that the mother of the boy with the GK-T65I mutation and the father of the boy with the GK-W99R mutation presented with normal birth weight, despite both having a de novo activating GCK mutation, it is interesting that previously reported (13) patients with PHHI due to activating mutations in GCK presented with normal birth weight.
The intravenous glucagon and adrenalin stimulation tests failed to increase blood glucose from low fasting levels, and leucine stimulation did not increase hypoglycemia, showing the absence of hypersensitivity to this amino acid (Table 2). The low threshold for GSIS induced by the Y214C mutation may explain the results of the intravenous glucose tolerance test in our patient, where insulin secretion was high even at blood glucose concentrations as low as 16.2 mg/dl (Table 3). In contrast to other cases of PHHI related to activating mutations in GCK, no glycemic response whatsoever was present in the oral glucose tolerance test results of our patient. In fact, plasma glucose levels were always low, decreasing rapidly until glycemia reached the pathologic value of 18 mg/dl (Table 4). Because GK also serves as a metabolic glucose sensor in pancreatic α-cells (18), this implies that our patient had to have a very low threshold for glucose-stimulated glucagon secretion as well, leading to a severe delay in glucagon secretion. The lack of glucagon-stimulated glucose increase and the irregular response of growth hormone in the intravenous glucose tolerance test in our patient show, in contrast to other PHHI patients having activating mutations in GCK, an altered counterregulatory response that could further aggravate severe hypoglycemia.
Our patient, like those with GCK activating mutations previously reported, exhibited a normal lipid profile. However, only our patient had hepatomegaly and cardiomegaly. The accumulation of excessive amounts of glycogen in liver could explain the hepatomegaly, and increased glycogen synthesis in muscle induced by hyperinsulinism could explain the cardiomegaly.
Histological changes in the pancreas also differentiate our patient from other previously reported patients with activating GCK mutations. Although histology was normal in the previously reported cases (3,4,13), the average size of islet profiles was ∼2.5-fold in our patient compared with that of control subjects, and 10- and 8-fold compared with those of islets from diffuse or focal PHHI patients, respectively (Fig. 1A). This increase in size is probably due to increased insulin secretion from the β-cells. Some β-cells had abnormally large nuclei, but these were clearly smaller and less frequent than in patients with a typical SUR1-related diffuse form (Fig. 1B). Islet nuclear crowding (the number of endocrine nuclei per cytoplasm area) was intermediate between diffuse PHHI and control or focal PHHI subjects. The absence of proinsulin cytoplasmic labeling indicates that proinsulin was normally cleaved at the end of the Golgi apparatus. Increased labeling in the Golgi area reflects a high level of insulin synthesis. However, this was accompanied by low-intensity insulin immunostaining, indicating decreased insulin storage associated with continuous high insulin secretion (Figs. 1A and B).
In PHHI, a subtotal pancreatectomy may be necessary to control hypoglycemia in patients who do not respond to diazoxide. Neither subtotal pancreatectomy nor the treatment with diazoxide resolved hypoglycemia of our patient. We can assume that the high turnover and the very low threshold for GSIS induced by GK-Y214C allows high insulin secretion, even with only 5% of the pancreas left, particularly because the infantile pancreas may undergo significant regeneration (19). The absence of response to diazoxide is a characteristic feature shared by some patients with severe familial hyperinsulinism due to mutations in the KATP channel that are often completely unresponsive to diazoxide. Theoretically, the hyperinsulinism of our patient should have responded to diazoxide. This unresponsiveness could be due to the low dose used, which was not sufficient to increase the threshold for GSIS enough to overcome the low threshold provided by Y214C.
In conclusion, we describe here a new type of very severe hyperinsulinism caused by a de novo mutation in GCK with unique clinical characteristics compared with previously described activating GCK mutations. Our findings are important for several reasons. First, they demonstrate the role of GK as the most important “glucostat” of the human body. Second, they indicate that the target of the GK activator discovered may be the allosteric heterotropic activator site, and the amino acid residue Y214 could be part of it. Third, they prove that the clinical phenotype of PHHI patients with activating mutations in GCK is not homogeneous but rather heterogeneous. Finally, they indicate that genetic and functional studies of GK should also be considered in severe cases of familial hyperinsulinemic hypoglycemia.
RESEARCH DESIGN AND METHODS
After a normal pregnancy, a baby girl of Finnish ancestry was born full term in 1972. She was relatively large for gestational age, with a weight and height at birth of 4,350 ± 1.8 g (±SD) and 52 ±1 cm, respectively. She had intrapartum asphyxia, with an Apgar score of 5/7. Since the first postnatal day, she presented symptoms of nonketotic hypoglycemia, with cyanosis, shivering, unconsciousness, and convulsions. Cardiomegaly, hepatomegaly, hypocalcemia, and hypomagnesemia were detected. Blood glucose levels were permanently hypoglycemic (0.1–2.6 mmol/l), independent of feeding. Despite a leucine-free diet, high-dose glucose infusion, and hydrocortisone and glucagon treatment, hypoglycemia persisted. She was treated with diazoxide (2 mg · kg−1 · day−1), with no response. At the age of 6 months, the baby underwent a subtotal pancreatectomy (95%), with no resolution of hypoglycemia. By the age of 10 months, signs of developmental retardation were evident and her electroencephalogram was severely abnormal. Until her death at the age of 29 years, she had severe mental retardation and epilepsy. She was on diazoxide therapy (2–5 mg/kg) for 22 years. During this period, glucose levels were always between 0.5 and 2.4 mmol/l. At the age of 22 years, diazoxide treatment was discontinued, without any effect on glycemia (1.6–2.9 mmol/l). There was no family history of diabetes or other disorders related to glucose metabolism. Her two healthy sisters had normal development.
Metabolic studies.
The studies, performed according to the Declaration of Helsinki, were approved by the institutional ethics committee, and written informed consent was obtained from the parents of the proband. An intravenous glucagon test, adrenalin and leucine stimulation tests, an oral glucose tolerance test, and an intravenous glucose tolerance test were performed before subpancreatectomy.
Genetic studies.
The promoter region, all 39 exons and flanking introns of the SUR1 gene, the single exon of the Kir6.2 gene, exons 11 and 12 of the glutamate dehydrogenase gene, and the promoter region and all 10 exons of the GCK gene were screened for mutations with the single-strand conformation polymorphism analysis in the proband. The variant of the GCK gene in exon 6 detected by single-strand conformation polymorphism analysis was identified by direct sequencing and verified by digestion with restriction endonuclease BbvI. Paternity tests in the proband and her parents involved the determination of common polymorphisms in 10 different genes and 19 polymorphisms in the HLA-DQB1 region.
Histological analysis.
Histological analysis was performed on the surgically removed specimens when the patient was 6 months old and not on the remnant pancreas after the patient died. Pancreatic specimens were fixed in formalin and embedded in paraffin. The 7-μm thick sections were stained with the Van Gieson technique or were incubated with a monoclonal anti-insulin antibody (dilution 1/1,000; Chemicon, Temecula, CA) or with a monoclonal proinsulin antibody (dilution 1/500; Novobiolab, Copenhagen, Denmark), both of which were detected and revealed by a secondary anti-mouse antibody and alkaline phosphatase anti-alkaline phosphatase complex (Dako, Carpinteria, CA).
Functional studies of the protein: site-directed mutagenesis, kinetic, and structural analysis.
The mutation of the GCK gene and wild type were expressed in Escherichia coli to carry out the purification. Kinetic studies of the wild-type GK and GK-Y214C were performed spectrophotometrically, as previously described (20). We used nonlinear kinetics according to the Hill equation, to investigate the affinity of the enzyme for glucose, and the Hill coefficient, which characterizes the sigmoidal glucose dependency of GK. After making at least three different preparations of both wild type and GK-Y214C, the results obtained were compared. To measure glucose phosphorylation capacity of the enzyme, we used the relative activity index. Structural analysis was done using the structural model of human GK as previously described (21,16). We used a minimal mathematical model that has been previously described (10) to assess the impact of the GCK mutation on GSIS.
A.L.C.-M. and H.H. contributed equally to this article.
Article Information
This work was supported by Spanish Ministry of Science and Technology grant no. CYT-836, by a Novo Nordisk grant, by SAF2001/3623 (all to A.L.C.-M.), and by grant BMC2002-00208 (to P.S.). The FIS of the Instituto de Salud Carlos III, Madrid, Spain, grant no. 02/PI021473 was awarded to A.L.C.-M. and the RCMN (C03/08). Grants from the Foundation of Pediatric Research were awarded to T.O., and the Academy of Finland and the European Union (grant no. QLG1-CT-1999-00674) provided grant support to M.L.
