Diabetologia (1999) 42: 1175±1186 Ó Springer-Verlag 1999 Mutants of glucokinase cause hypoglycaemia- and hyperglycaemia syndromes and their analysis illuminates fundamental quantitative concepts of glucose homeostasis E.A. Davis1, A. Cuesta-Muæoz1, M. Raoul1, C. Buettger1, I. Sweet1, M. Moates2, M.A. Magnuson2, F.M. Matschinsky1 1 Department of Biochemistry and Biophysics and the Diabetes Research Center, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania, USA 2 Department of Molecular Physiology and Biophysics, Vanderbilt University Medical School, Nashville, Tennessee, USA Abstract Results. Large changes, varying greatly between mu- tants were found in nearly all variables. Glucokinase Aims/hypothesis. Mutations of the glucokinase gene flux at threshold for glucose-stimulated insulin re- cause hyperglycaemia or hypoglycaemia. A quantita- lease was about 25% of total phosphorylating poten- tive understanding of these defects of glucose homeo- tial in the normal beta-cell and this was used to predict stasis linked to the glucokinase gene was lacking. thresholds for the mutant heterozygotes. Clinical data Therefore a database of kinetic variables of wild-type for maturity onset diabetes of the young type linked to and 20 missense mutants of glucokinase was devel- the glucokinase gene and familial hyperinsulinaemic oped and used in mathematical modelling to predict hypoglycaemia linked to the glucokinase gene and the thresholds for glucose-stimulated insulin release. the glucokinase kinetic data of this study were used Methods. Recombinant human glucokinase was gen- to test the model. The model predicts fasting blood erated in E. coli. The kcat, glucose S0.5, ATP Km, and glucose between 3 and 7 mmol/l in these cases. Hill number of glucokinase were determined. Inhibi- Conclusion/interpretation. A kinetics database of tion by Stearoyl CoA and glucokinase regulatory pro- wild-type and 20 mutants of glucokinase was devel- tein and thermal stability were assayed for all mutants oped. Many kinetic differences were found for the kinetically similar to wild-type glucokinase. A mathe- mutants. The mathematical model to calculate the matical model predicting the threshold for glucose- threshold for glucose-stimulated insulin release pre- stimulated insulin release was constructed. This model dicts fasting blood glucose between 3 and 7 mmol/l is based on the two substrate kinetics of glucokinase in subjects with glucokinase gene mutations. and the kinetic variables of the database. It is assumed [Diabetologia 42: 1175±1186] that both glucokinase gene alleles are equally ex- pressed in beta-cells and that induction of glucokinase Keywords MODY-2, glucokinase, glucose threshold, occurs as a function of basal blood glucose. insulin secretion, beta-cell, mathematical model. Received: 1 February 1999 and in revised form: 3 April 1999 The discoveries of linkage to the glucokinase locus on chromosome 7 in autosomal dominantly inherited Corresponding author: F.M. Matschinsky, MD, Diabetes Re- forms of diabetes mellitus (DM-gk) in 1992 [1, 2] search Center, 501 Stemmler Hall, 36th & Hamilton Walk, Philadelphia, PA 19104±6015, USA and of hyperinsulinaemia (HI-gk) [3] in 1998 gave a Abbreviations: GSIR, Glucose-stimulated insulin release; DM- considerable boost to the pancreatic beta-cell gluco- gk, maturity onset diabetes of the young type 2 (MODY-2) kinase-glucose sensor concept [4]. The concept had linked to the glucokinase gene; DM-GK, maturity onset diabe- been formulated to explain essential characteristics tes of the young with abnormal glucokinase protein; HI-gk, fa- of glucose-stimulated insulin release (GSIR) and milial hyperinsulinemic hypoglycaemia linked to the glucoki- had evolved slowly from its early beginnings in the nase gene; HI-GK, familial hyperinsulinaemic hypoglycaemia late '60s [4, 5, 6]. Extensive work with glucokinase with abnormal glucokinase protein; GKRP, glucokinae regula- tory protein; b-GPR, beta-cell glucose phosphorylation rate; h, knockout mice quickly followed the first reports of glucokinase linkage of diabetes syndromes and fully Hill Coefficient; Ia, activity index; GST-GK, glutathionyl-S- transferase-glucokinase. confirmed the findings in humans [7±9]. Meanwhile 1176 E.A. Davis et al.: Effects of glucokinase mutants on glucose homeostasis the count of glucokinase mutations causing DM-GK GSIR including factors to account for the beta-cell is approaching the 100 mark [10±15], but only one glucose threshold concept, cooperativity of glucoki- single family with HI-GK has been reported [3]. nase with regard to glucose, the interaction of glu- Precise quantitative understanding of the part that cokinase with its second substrate MgATP2- and the the beta-cell glucokinase-glucose sensor plays in nor- probability of increased protein lability due to muta- mal glucose homeostasis and the pathogenesis of tions. A sufficiently large sample size was chosen al- DM-gk and HI-gk is lacking despite the efforts of sev- lowing for a critical and comprehensive testing of eral laboratories. First reports confirmed a shift to the model predictions with published clinical data. the right of the GSIR curve and suggested some par- allel between the severity of impairment in insulin se- cretion of the diabetic phenotype and the magnitude Materials and methods of the kinetic changes of some of the mutant glucoki- nase molecules [16]. This general conclusion was not Production of recombinant glutathionyl-S-transferase-glucoki- surprising in view of the basic concept. This early nase (GST-GK.) Recombinant human wild-type and mutant analysis must now, however, be reconsidered because beta-cell GST-GK was prepared as described previously [24]. its experimental basis is in part unreliable and ad- We choose 21 different forms of the enzyme for analysis: the vances have been made in fully appreciating the wild-type, the historically significant mutant D158A, nine mu- quantitative biochemical and biophysical tenets of tants which had been partially characterized previously [18,21,24,25], nine missense mutations recently identified the glucokinase-glucose sensor paradigm in glucose from DM-gk family studies [15] and the newly described HI- homeostasis [17]. gk mutant V455M [3]. Point mutations were introduced into There are a variety of reasons to reconsider the re- the pGEX-HIGK expression vector using the QuikChange lation between kinetic data and their physiological mutagenesis kit (Stratagene, La Jolla, Calif., USA). The Oli- consequences. The most critical are uncertainties gonucoleotides used were synthesized by Research Genetics about the biochemical and biophysical data base for (Huntsville, Ala., USA). All mutants were transformed into human wild-type and mutant glucokinase. Most of E. coli (DH5a strain) and verified by DNA sequencing. Gluta- thionyl-S-transferase-glucokinase was produced in E. coli and the published kinetic data on human wild-type and then purified from crude extracts as described [24]. Briefly, sin- mutant recombinant glucokinase are probably com- gle-step affinity chromatography was used with glutathione- promised by the inclusion of the functionally signifi- agarose beads to bind the fusion protein which is then eluted with glutathione. The entire purification procedure is com- cant incidental mutant D158A, which lowers the S0.5 for glucose [18±21]. Furthermore, previous studies of- plete in less than 3 h. Purified protein is stored at ±80°Cin ten did not consider that the recombinant mutant glu- 30% glycerol, 50 mmol/l glucose, 5 mmol/l glutathione, 5 mmol/l DTT, 200 mmol/l KCl, 50 mmol/l TRIS buffer (pH cokinase protein might be unstable, that the glucose 7.4). Protein concentration was determined by the BioRad as- phosphorylation assay is greatly influenced by vari- say with bovine serum albumin as standard (Hercules, Calif., ables such as pH and ATP-glucose interactions diffi- USA). Purity of the GST-GK preparations was screened for cult to standardize and that the data processing must with SDS-PAGE. account for the cooperativity of glucokinase [17, 22]. The published mathematical modelling attempts of Kinetic variables. Glucokinase activity was measured spectro- + the role of glucokinase in glucose-stimulated insulin photometrically using an NADP coupled assay with glucose- 6-phosphate dehydrogenase as described [24]. The pH of all release GSIR are limited because they do not consid- assays was 7.4, except for assays which assessed the inhibition er critical physiological concepts such as the impact of glucokinase by glucokinase regulatory protein (GKRP) on glucose threshold of GSIR, the sigmoidicity of glu- where a pH of 7.1 was used. The rate of glucose phosphoryla- cokinase-glucose dependency and do not account for tion catalysed by glucokinase depends on the concentration of the impact on glucokinase activity of changes in the both glucose and ATP. For this reason two protocols for the as- K for the second substrate MgATP2- [23]. This com- says were used. One with glucose as a variable substrate (Pro- m tocol A) and one with ATP as the variable substrate (Protocol bined with limitations of the data base on glucokinase B). In protocol A 11 glucose dilutions with the second sub- kinetics made it difficult to achieve accurate results 2- strate MgATP at 5 mmol/l plus 1 mmol/l MgCl2 in excess from previous modelling attempts. were used for generating the glucose dependency curve and Our investigation was therefore designed to recon- to calculate Vmax, glucose S0.5 and Hill coefficient