Introduction

In liver, glucose-6-phosphatase catalyses the hydrolysis of glucose 6-phosphate (G6P) to glucose and inorganic phosphate, the final step in the gluconeogenic and glycogenolytic pathways. Mutations within the glucose-6-phosphatase catalytic subunit (G6Pase, now known as G6PC) cause glycogen storage disease type 1a, which is primarily characterised by severe hypoglycaemia in the post-absorptive state. Although glucose-6-phosphatase activity and G6PC mRNA are detectable in islets, the activity displays distinct kinetic behaviour and inhibitor profiles compared with that assayed in hepatic extracts, an observation that might be explained by the existence of a distinct G6PC isoform in islets, present in addition to the known G6PC isoform [1]. Indeed, we have previously identified an islet-specific G6PC-related protein (IGRP, now known as glucose-6-phosphatase, catalytic, 2 [G6PC2]) that is produced specifically in islet beta cells [1]. However, the biological function of G6PC2 is unclear. Overproduction of G6PC2 by transient transfection of cell lines has been reported to have little [2] or no [1] effect on G6P hydrolysis in tissue homogenates. Questions thus arise as to whether the physiologically important substrate that G6PC2 hydrolyses is something other than G6P or whether a low rate of G6P hydrolysis by G6PC2 might nevertheless influence glucose-stimulated insulin secretion in vivo.

Although the role of G6PC2 in beta cell function remains unclear, G6PC2 was recently identified as a major autoantigen in the non-obese diabetic (NOD) mouse model of type 1 diabetes [3]. In NOD mice up to 40% of the CD8-positive cells infiltrating the islet were found to be G6PC2-reactive [3]. Subsequent reports have shown that G6PC2 is also recognised by CD4-positive T cells [4]. Most importantly, G6PC2 has recently also been shown to be an autoantigen in human type 1 diabetes [5] and in humanised NOD mice [6]. Given the observations that administration of G6PC2-derived peptides abrogate or delay the disease process in NOD mice [4, 7] and that NOD mice expressing a non-antigenic form of insulin do not develop diabetes [8], it is possible that strategies designed to suppress G6pc2 expression might also be used to delay or prevent the onset of this disease. However, since the function of G6PC2 is unclear, the concern with such an approach is that the absence of G6PC2 might itself have deleterious consequences. To address this concern and gain insight into the biological function of G6PC2 we have characterised the phenotype of mice containing a global knockout of the G6pc2 gene.

Materials and methods

For details of animal care, generation of the G6pc2 targeting vector, generation of G6pc2 knockout mice, and G6pc2 mRNA expression analyses and immunohistochemical staining see the Electronic supplementary material.

Phenotypic analysis of G6pc2 knockout mice Animals were fasted for 5 h, weighed and then anaesthetised 1 h later using isoflurane before isolation of blood samples (∼200 μl) from the retro-orbital venous plexus.

Whole-blood glucose concentrations were determined using a monitor (Accu-Check Advantage; Roche, Indianapolis, USA). EDTA (5 μl; 0.5 mol/l) was then added before centrifugation to isolate plasma. Trasylol (aprotinin; 5 μl; Bayer Health Care, West Haven, CT, USA) was added to the plasma to prevent proteolysis of glucagon. Cholesterol was assayed using a cholesterol reagent kit (Raichem, San Diego, CA, USA), while triacylglycerol and glycerol were assayed using a serum triacylglycerol determination kit (Sigma, St Louis, MO, USA). Insulin and glucagon levels were quantitated using radioimmunoassays by the Vanderbilt Diabetes Center Hormone Assay Core.

Hyperglycaemic clamps on 5-h fasted chronically catheterised conscious male mice were performed by the Vanderbilt Mouse Metabolic Phenoty** Center as previously described [9].

Statistical analyses Data were analysed using a Student’s t test: two sample assuming equal variance. The level of significance was as indicated (two-sided test).

Results

A modified mouse G6pc2 allele, in which the coding sequence of exons 1 to 3 plus 10 bp of the third intron [1] were replaced by a LacZ/Neo cassette, was created by gene-targeting in 129/SvEvBrd (Lex-1) embryonic stem cells (Fig. 1a). Correct gene-targeting was confirmed by Southern blot (Fig. 1b) and PCR (data not shown) analysis before injection of embryonic stem cells into C57BL/6 (albino) blastocysts and subsequent generation of G6pc2 −/+ mice on a mixed 129/SvEvBrd × C57BL/6 background.

Fig. 1
figure 1figure 1

Generation and authentication of G6pc2 knockout mice. a Strategy used to generate G6pc2 knockout mice by homologous recombination in embryonic stem cells. A schematic representation of the wild-type murine G6pc2 locus and the targeting construct are shown. IRES, internal ribosome entry site; nls, nuclear localisation signal; Bgal, beta galactosidase; MClneo, originally the name given to describe a DNA fragment that contains a modified TK promoter driving neomycin (neo) gene expression. b Southern blot analysis of the G6pc2 locus using genomic DNA extracted from three targeted embryonic stem cell lines, or wild-type embryonic stem cell genomic DNA, designated Lex-1, as a control, using the 5′ and 3′ diagnostic probes (see a). The sizes of the wild-type (WT) locus and targeted (T) alleles and the location of DNA size markers are indicated. c Analysis of mouse G6pc2 mRNA expression by RNA blotting. d Immunohistochemical staining of wild-type and G6pc2 knockout mouse pancreas with antisera raised to insulin and G6PC2

Full size image

Genotype analysis of 234 3-week-old pups generated by cross-breeding heterozygous G6PC2 −/+ mice demonstrated that 68 mice were G6pc2 +/+, 108 were G6pc2 −/+, and 58 were G6pc2 −/−, a distribution close to the expected pattern for Mendelian inheritance. The ratio of male to female mice was 123:111. Cross-breeding experiments revealed that both male and female homozygous G6pc2 −/− mice are fertile.

Biochemical analyses confirmed that the G6pc2 gene was not expressed in G6pc2 −/− mice. Thus, RNA blotting demonstrated the absence of G6pc2 mRNA (Fig. 1c) and immunohistochemical staining of pancreas tissue showed the loss of G6PC2 immunoreactivity (Fig. 1d) in G6pc2 −/− mice. The size and number of islets in knockout animals were indistinguishable from wild-type as were the relative numbers of alpha and beta cells (data not shown).

The activity and behaviour of G6pc2 −/− mice were indistinguishable from their wild-type and heterozygous littermates at all ages, from birth up to 1 year in age. No gross anatomical changes were observed either externally or to major internal organs and no differences were seen in the weights or lengths of G6pc2 −/− and wild-type mice (Table 1).

Table 1 Phenotypic characterisation of G6pc2 knockout mice
Full size table

Table 1 summarises changes in metabolic parameters in these animals assayed at 16 weeks of age following a 6-h fast. In both male and female G6pc2 −/− mice there was a statistically significant 14 and 11% decrease, respectively, in blood glucose in comparison with wild-type mice (Table 1). The blood glucose level in male G6pc2 −/− mice was also lower than that in male G6pc2 −/+ mice (Table 1). In female G6pc2 −/− mice there was also a statistically significant 12% decrease in plasma triacylglycerol (Table 1). There were no statistically significant differences in plasma cholesterol, glycerol, insulin and glucagon concentrations between G6pc2 −/− and wild-type mice (Table 1). G6pc2 −/− mice showed sex-related variation in the majority of these metabolic parameters that were in the same direction and of similar magnitude to the sex-related differences in wild-type mice. Thus, in males versus females, insulin, triacylglycerol, cholesterol and glucose tended to be higher whereas glucagon was lower.

Since some metabolic disturbances, such as the development of diabetes in NOD mice, only arise in older animals, several metabolic parameters were re-assayed in 6-h fasted female mice at a later time point. The mean ages of the G6pc2 +/+, G6pc2 −/+ and G6pc2 −/− animals studied were 52, 48 and 54 weeks, respectively. As seen in 16-week-old mice, there was a trend towards lowered blood glucose in these older G6pc2 −/− mice, although less animals were analysed than at 16 weeks and the difference did not reach statistical significance (ESM Table 1). However, a statistically significant 19% decrease in plasma triacylglycerol was again apparent (ESM Table 1). These data suggest that a severe metabolic phenotype does not arise in older animals.

Similarly, some metabolic disturbances only become readily apparent under stimulatory rather than basal conditions. Hyperglycaemic clamps were therefore used to provide a measure of dynamic islet function in vivo. Blood glucose was raised to ∼14 mmol/l and insulin secretion was assessed over a 120-min period (Fig. 2). Glucose requirements were comparable: 45 ± 2 vs 39 ± 2 mg kg−1 min−1 in G6pc2 −/− vs wild-type. The data show no difference in insulin secretion between wild-type and G6pc2 −/− mice (Fig. 2). This suggests that the absence of G6PC2 affects fasting but not maximal glucose-stimulated insulin secretion, and is consistent with the conclusion that the absence of G6PC2 results in a mild metabolic phenotype.

Fig. 2
figure 2

Assessment of insulin secretion in G6pc2 knockout mice in vivo using hyperglycaemic clamps. Hyperglycaemic clamps were performed in 5-h fasted chronically catheterised conscious male wild-type (closed symbols) and G6pc2 knockout (open symbols) mice as described in Materials and methods. Results show the mean insulin concentrations ± SEM in three wild-type (mean age 35 weeks) and four G6pc2 knockout (mean age 32 weeks) animals

Full size image

Discussion

The data presented indicate that deletion of the G6pc2 gene results in a mild metabolic phenotype on a mixed 129SvEvBrd × C57BL/6 background. This bodes well for future studies in which G6pc2 −/− mice are bred on to diabetes-susceptible strains such as the NOD mouse. Based on present evidence, ablation of the gene may actually enhance islet responsiveness at fasting glucose concentrations, which would be an advantage in the face of reduced beta cell mass. Thus, unlike the case of other autoantigens such as insulin [8], the physiological consequences of G6pc2 gene ablation appear to be tolerable.

The data derived from fasting mice are consistent with G6P being a substrate for G6PC2 in vivo since removal of G6PC2 would leave the action of glucokinase unopposed and would in effect be equivalent to an increase in glucokinase activity. Future experiments will directly address the hypothesis that the observed decrease in fasting blood glucose in G6pc2 knockout mice is a consequence of altered kinetics of glucose-stimulated insulin secretion. However, the hyperglycaemic clamp data suggest that maximal glucose-stimulated insulin secretion is unaltered (Fig. 2). Interestingly, Matchinsky et al. [10] concluded that, while normal rat islets do contain measurable glucose-6-phosphatase activity, this does not play a role in glucose metabolism and sensing by the normal beta cell. Since rat islets do not produce G6PC2 [1] it will be of interest to repeat this analysis using mouse islets.