A genetic and physiological study of impaired glucose homeostasis control in C57BL/6J mice
- First Online:
- Cite this article as:
- Toye, A.A., Lippiat, J.D., Proks, P. et al. Diabetologia (2005) 48: 675. doi:10.1007/s00125-005-1680-z
- 2k Downloads
C57BL/6J mice exhibit impaired glucose tolerance. The aims of this study were to map the genetic loci underlying this phenotype, to further characterise the physiological defects and to identify candidate genes.
Glucose tolerance was measured in an intraperitoneal glucose tolerance test and genetic determinants mapped in an F2 intercross. Insulin sensitivity was measured by injecting insulin and following glucose disposal from the plasma. To measure beta cell function, insulin secretion and electrophysiological studies were carried out on isolated islets. Candidate genes were investigated by sequencing and quantitative RNA analysis.
C57BL/6J mice showed normal insulin sensitivity and impaired insulin secretion. In beta cells, glucose did not stimulate a rise in intracellular calcium and its ability to close KATP channels was impaired. We identified three genetic loci responsible for the impaired glucose tolerance. Nicotinamide nucleotide transhydrogenase (Nnt) lies within one locus and is a nuclear-encoded mitochondrial proton pump. Expression of Nnt is more than sevenfold and fivefold lower respectively in C57BL/6J liver and islets. There is a missense mutation in exon 1 and a multi-exon deletion in the C57BL/6J gene. Glucokinase lies within the Gluchos2 locus and shows reduced enzyme activity in liver.
The C57BL/6J mouse strain exhibits plasma glucose intolerance reminiscent of human type 2 diabetes. Our data suggest a defect in beta cell glucose metabolism that results in reduced electrical activity and insulin secretion. We have identified three loci that are responsible for the inherited impaired plasma glucose tolerance and identified a novel candidate gene for contribution to glucose intolerance through reduced beta cell activity.
KeywordsC3H/HeHC57BL/6JGckGluchosInsulinIsletKATP channelNntType 2 diabetes
Intraperitoneal glucose tolerance test
Nicotinamide nucleotide transhydrogenase
Quantitative trait locus
Tricarboxylic acid (Krebs) cycle
Insulin release from pancreatic beta cells is stimulated by increased beta cell uptake and metabolism of glucose. The consequent changes in the intracellular concentrations of adenine nucleotides cause closure of ATP-sensitive K+ (KATP) channels in the beta cell plasma membrane. In turn, this leads to membrane depolarisation, opening of voltage-gated Ca2+ channels, Ca2+ influx, fusion of insulin secretory vesicles with the plasma membrane and insulin secretion. Normally, insulin secretion in response to elevated plasma glucose is biphasic. Intracellular messengers controlling KATP channel-dependent first-phase insulin secretion also regulate second-phase insulin secretion. However, additional KATP channel-independent messengers are also involved . There is evidence that loss of first-phase insulin secretion leads to postprandial hyperglycaemia and is common in patients with type 2 diabetes and individuals with impaired glucose tolerance .
Defects in beta cell function are found in monogenic diabetes, such as maturity-onset diabetes of the young (reviewed in Ref. ) and permanent neonatal diabetes. It is also apparent that abnormalities in insulin secretion and beta cell function contribute to the onset and development of polygenic type 2 diabetes . In type 2 diabetes there is gradual progression from normal glucose tolerance to impaired glucose tolerance and subsequently overt diabetes. This is associated with a progressive decline in beta cell function and reduced insulin secretion. Insulin resistance may enhance the risk of diabetes by placing an increased demand upon the beta cell, but by itself does not result in diabetes. The genetic defects that produce inappropriate homeostatic control in type 2 diabetes are poorly understood (reviewed in Refs. [3, 4]).
Animal models of glucose intolerance provide valuable information about glucose homeostasis that can ultimately be applied to human diabetes. The C57BL/6J mouse exhibits defects in glucose tolerance that are independent of obesity [5, 6]. Feeding C57BL/6J mice with high-fat diets results in insulin resistance, increased fasting plasma glucose levels and diabetes [7–10]. On a high-fat diet these mice also show higher weight gain per energy intake , higher weight gain when the same amount of energy is given as fat , and higher fat and lower protein and water body composition  than A/J mice. Insulin action was also reduced by between 32% and 60% in C57BL/6J mice after 9 months on a high-fat diet . Interestingly, however, C57BL/6J mice are actually more insulin-sensitive than AKR/J mice  and DBA/2 or 129X1 mice .
On normal diets, C57BL/6J mice appear to have normal free-fed plasma insulin levels, but postprandial first-phase insulin release is impaired when compared to glucose-tolerant control strains, including C3H [5, 16]. Second-phase insulin release is also impaired in comparison with AKR/J  and DBA/2  mice, but is not significantly different to C3H mice (at least at 8 weeks of age) . A defect in first-phase insulin secretion also appears to be present in C57BL/6J mice fed a high-fat diet , and is accompanied by an unchanged insulin response and delayed glucose clearance over the first 4 weeks of a high-fat diet . Recent physiological  and genetic studies [8, 19] also suggest that insulin and plasma glucose levels are poorly correlated. Classification of C57BL/6J mice on a high-fat diet into (1) lean non-diabetic (12%), (2) lean diabetic (12%), (3) obese diabetic (about 50%), and (4) intermediates has allowed a microarray transcript profiling comparison to be made and shown striking differences in gene expression between the groups . These data suggest a differential metabolic adaptation, not wholly under genetic control, that contributes to the observed phenotypic diversity in this strain .
In summary, C57BL/6J mice represent an important model of diet-induced diabetes, which also exhibits defects in glucose tolerance on a normal diet. This model develops insulin resistance on a high-fat diet and appears to have a complex insulin secretion deficit on normal and high-fat diets.
Here, we use genetic mapping of F2 intercross mice, physiological studies of whole mice, studies of isolated liver and pancreatic tissues, and candidate gene analyses, to define more precisely the genetic basis of glucose intolerance in C57BL/6J mice.
Materials and methods
Mice were kept in accordance with UK Home Office welfare guidelines and project license restrictions. C57BL/6J mice were obtained from Jackson Laboratory (Bar Harbor, ME, USA) in 1999. C3H/HeH mice were from the Harwell (Oxfordshire, UK) stock. Mice were maintained under controlled conditions and had free access to a commercial diet (maintenance chow; Special Diet Services, Essex, UK) containing 2.6% saturated fat.
Genomic DNA was extracted from mouse tail tissue using a Qiagen DNeasy tissue kit (Qiagen, Hilden, Germany). Mice were genotyped by SSLP analysis. Primer sequences were obtained from published records . PCR products were analysed using an ABI 377 sequencer and Genescan and Genotyper software and protocols (Applied Biosystems Europe, Warrington, Cheshire, UK). Markers were spaced at a median interval of 11 cM, with an average interval of 14.6 cM and a largest interval of 46.7 cM.
Phenotype and genotype data were maintained in Microsoft Excel, SPSS and MapManager QTX (; see http://mapmgr.roswellpark.org/) formats. Genetic maps were constructed using the published map order of markers (http://www.informatics.jax.org). Linkage between markers and phenotypes was evaluated using a single marker, and interval mapping features of MapManager. Thresholds for defining linkage were as outlined by Lander and Kruglyak , with additional permutation tests  where stated. Empirical thresholds were based on analysis of 1,000 permutations of the original data set. Thresholds derived from free model permutation were LRS 15.7 (LOD 3.4) for T0 glucose, LRS 16.1 (LOD 3.5) for T30 glucose, LRS 15.5 (LOD 3.4) for T60 glucose, LRS 15.7 (LOD 3.4) for AUC glucose and LRS 16.9 (LOD 3.7) for T30 insulin.
Intraperitoneal glucose tolerance tests
Mice were fasted overnight, weighed, and a blood sample collected by tail venipuncture under local anaesthetic (lignocaine; Biorex, Middlesex, UK) using lithium-heparin microvette tubes (Sarstedt, Numbrecht, Germany) to establish a baseline glucose level ‘T0’. The mice were then injected intraperitoneally with 2 g glucose/kg body weight and blood samples taken at 15, 30, 60 and 120 min after the injection to monitor the rate of glucose clearance. Plasma glucose and insulin were measured using a Beckman Glucose analyser and a Mercodia Ultra-Sensitive Mouse ELISA kit.
Insulin tolerance test
Insulin tolerance tests on free-fed mice were performed as described .
Islet cell isolation
Mice were killed by cervical dislocation. The pancreas was removed and islets or dissociated islet cells were prepared and cultured as previously described . Cells were exposed to glucose-free solution for 15 min prior to experiment.
Insulin secretion studies
Insulin secretion from isolated islets (ten islets/well) was measured during 1-h static incubations in Krebs–Ringer Buffer (in mmol/l: 118.5 NaCl, 2.54 CaCl2, 1.19 KH2PO4, 4.74 KCl, 25 NaHCO3, 1.19 MgSO4, 10 HEPES, pH 7.4) containing either 2 or 10 mmol/l glucose. Each glucose concentration was replicated five to eight times. Samples of the supernatant were assayed for insulin. Insulin content was extracted using 95 : 5 ethanol/acetic acid. Insulin was measured using a Mouse Insulin ELISA kit (Mercodia, Uppsala, Sweden).
Islet cells were cultured on 35-mm Fluorodishes (World Precision Instruments, Stevenage, UK) and incubated with 3 μmol/l Fura-2-AM (Molecular Probes, Paisley, UK) for 40 min at 37°C. They were imaged at room temperature (20–24°C) using a fluorescence system (IonOptix, Boston, MA, USA), with 340 and 380 nm dual excitation. The 510-nm emission ratio was collected at 1 Hz. Background subtraction was performed by measuring fluorescence from a cell-free region in the field of view. Cells were perfused continuously with extracellular solution (as in perforated-patch experiments), plus glucose or tolbutamide as indicated. Only data from cells that responded to 500 μmol/l tolbutamide were analysed.
Glucose concentration–response relations for whole-cell KATP currents were measured using the perforated patch configuration of the patch-clamp technique . The pipette solution contained (mmol/l): 70 K2SO4, 1 CaCl2, 1 MgCl2, 10 NaCl, 10 HEPES (pH 7.2 with KOH), and amphotericin B (0.24 mg/ml). The extracellular (bath) solution contained (mmol/l): 137 NaCl, 5.6 KCl, 2.6 CaCl2, 1.2 MgCl2, 10 HEPES (pH 7.4 with NaOH) plus glucose as indicated. Currents were filtered at 3 kHz and digitised at 1 kHz. The KATP component of the whole-cell current was taken as the current blocked by 200 μmol/l tolbutamide.
The ATP sensitivity of KATP channels was measured at −60 mV in inside–out patches. Currents were digitised at 5 kHz and filtered at 2 kHz. The pipette (extracellular) contained (mmol/l): 140 KCl, 2.6 CaCl2, 1.2 MgCl2, 10 HEPES (pH 7.4 with KOH). The bath (intracellular) solution contained (mmol/l): 107 KCl, 1 MgCl2, 1 CaCl2, 10 HEPES (pH 7.2 with KOH), and K2ATP as indicated.
Quantitative RT-PCR was performed using ABI SYBR Green on an ABI PRISM 7700 Sequence Detection System (Applied Biosystems). To quantitate nicotinamide nucleotide transhydrogenase (Nnt), it was PCR-amplified from cDNA (RT-PCR), together with an endogenous control Gapdh (glyceraldehyde-3-phosphate dehydrogenase). RT-PCR primers listed in 5′ to 3′ orientation are for Gapd; Gapd888F-CCTGCGACTTCAACAGCAACT and Gapd979RV-CCAGGAAATGAGCTTGACAAA; and for Nnt: NntXn5intF-GATAGTTGGTGGTGGCGTTG and NntXn6intR-GATGTCCACCTCCTTGCACT.
The t-tests for significant differences between strains are based on RQ (Nnt/Gapd) values. Relative RQ is the ratio of the C3H/HeH RQ relative to that for C57BL/6J.
Statistical analysis and multilocus model
Multiple regression was performed by univariate ANOVA using the GLM feature of SPSS. F statistics were based on type III sums of squares. All markers that exceeded LOD 1.9 suggestive linkage threshold  were included in the regression model. Terms not significant at the 0.1% level were sequentially eliminated to produce the final multiple regression model. C57BL/6J homozygotes were coded 0.5, C3H/HeH homozygotes were coded −0.5 and heterozygotes were coded 0 in all regression analyses.
To elucidate whether epigenetic effects determine C57BL/6J glucose intolerance, we first compared the F1 offspring of C57BL/6J mothers with those of C3H/HeH mothers (Fig. 1b). We saw no significant differences between male (n=125) and female mice (n=91, p<0.05). F1 males (n=125) had significantly (p<0.0001) lower plasma glucose than their C57BL/6J mothers (B, n=10), but not their C3H/HeH parents (C, n=10) (Fig. 1b). We then examined the F2 phenotypes in the four possible F1 cross-combinations (CB×CB, CB×BC, BC×CB and BC×BC) in order to evaluate epigenetic effects in this generation. The T30, T60, T120 min and AUC glucose phenotypes were significantly (albeit weakly, i.e., close to the <0.05 threshold) different in some, but not all, of the pairwise comparisons. For T0, two of the six phenotype comparisons were significantly different (p<0.01) and one at a p value of less than 0.05 (data not shown). Thus there may be some weak epigenetic effects that influence the T0 phenotype in particular. All four cross-combinations are equally represented in the F2 population for linkage analysis.
Genome scan for plasma glucose and insulin QTLs in male mice aged 12 weeks
Phenotype at closest markerf
T0 glucose (mmol/l)
T30 glucose (mmol/l)
T60 glucose (mmol/l)
AUC (min·mmol−1 l−1)
T30 insulin (ng/ml)
The fasting plasma glucose level was linked to chromosome 11 (D11Mit 2, LRS 23.7) and is designated Gluchos2 (Table 1, Fig. 2). Plasma glucose levels at T15 and T120 were not significantly linked to any locus.
The plasma insulin level at T30 provides a measure of second-phase insulin secretion [5, 16]. It was significantly (p<0.05, r2=0.0433, r=0.208), but poorly, correlated to plasma glucose at the same IPGTT time point, and linked to a locus on chromosome 9 (D9Mit1001, LRS 30.9), which was named Gluchos3 (Table 1, Fig. 2). Insulin was correlated with body weight (r2=0.2319, r=0.48) but body weight was not correlated with T30 glucose (r2=0.0196, r=0.014).
ANOVA table for regression of multiple QTL on plasma glucose tolerance AUC (min·mol−1·l−1)
Type III SS
Degrees of freedom
D13Mit64 * D19Mit41
D2Mit200 * D9Mit311
Male C57BL/6J mice secrete less insulin than C3H/HeH mice in response to a glucose challenge in an IPGTT
Body weight, g±SD***
Plasma insulin at timepoints in an IPGTT±SD (in ng/ml)
Isolated islet studies
Candidate gene studies
Gluchos1 (chromosome 13)
In a 5-cM sweep around Gluchos1 on chromosome 13 at 65±12 cM we identified several potential candidate genes: in particular, the transcription factor islet gene enhancer protein 1 (Isl1), (113 Mbp) and Nnt (64 cM or 116 Mbp). The coding sequence of Isl1 was sequenced and no mutations found.
We sequenced the coding region of Nnt and found two mutations. First, an exchange of T (C3H/HeH) for C (C57BL/6J) at nt 104 of exon 1 that results in a missense methionine to threonine mutation at amino acid 35 of the protein. This mutation is located in the mitochondrial leader peptide sequence of the Nnt precursor protein . The second mutation was found in islet and liver cDNA PCR-amplified between exon 5 and 13, numbered according to the Nnt gene structure and sequence that have been derived from 129/SvJ mice (; GENBANK AAF72982, AF257137–AF257157). The amplified fragment was ∼508 bp from C57BL/6J but ∼1,261 bp from C3H/HeH cDNA. Sequencing of these products confirmed that exons 7 to 11 were completely missing in the C57BL/6J fragment, although the transcript remained in frame between exons 6 and 12 (see electronic supplementary material [ESM], link on first page of this article). In silico analysis of cDNA database sequences from 129/SvJ, NOD, FVB, C57BL/6J and B6CBAF1 mouse strains independently confirmed that most C57BL/6 annotated transcripts also lack a 753-bp segment that encompasses the whole of exons 7 to 11 expected from the structure of the 129/SvJ gene and that of other mouse strains . PCR amplification of exons from genomic DNA of 26 different inbred strains and substrains indicated that only C57BL/6J is missing exons 7 to 11 and that in the other strains all exons are present (data not shown). This is consistent with BLAST searches of the ENSEMBL C57BL/6J mouse sequence (releases v13.30.1, May 2003 and NCBIM32_feb 2004) that revealed no homology to these exons.
Gluchos2 (chromosome 11)
Gluchos2 lies in a region of comparative homology with a rat chromosome 14 diabetes quantitative trait locus (QTL), Dmo3 . There are a number of potential candidate genes that lie under the peak of linkage, most notably CamK2B and Gck. We sequenced the entire coding sequence of CamK2B and found no mutations capable of disrupting protein function. Given the crucial role of glucokinase in glucose homeostasis [32–37], we compared in vitro glucokinase function in liver of 12-week-old male C57BL/6J and C3H/HeH mice. C57BL/6J mice had significantly (p<0.01) lower glucokinase activity despite similar hexokinase activity: glucokinase activity was 16.61±2.144 (SD) mU/mg protein (n=5) for C57BL/6J compared to 22.73±2.09 mU/mg protein (n=4) for C3H/HeH (enzyme activity level was measured using triplicate Vmax assays). We sequenced the islet promoter and entire coding region of the Gck gene in both strains and found one silent polymorphism in C57BL/6J, namely a C-to-G transversion at base 131 of exon 7.
Gluchos3 (chromosome 9)
There are a large number of genes in the Gluchos3 region including several of potential interest. For example, Atp5l encodes a subunit of the mitochondrial F0 complex involved in ATP synthesis and proton transport. Given its role in ATP synthesis, and the tight coupling between ATP generation and insulin secretion, it seems a good causal candidate for Gluchos3. The islet-2 gene (a relative of islet-1) also lies on chromosome 9, but at the edge of the potential linkage.
IPGTT comparisons between four nonobese inbred mouse strains revealed that C57BL/6J mice were less glucose-tolerant than C3H, DBA/2 and BALB/c, and that male mice were significantly less glucose-tolerant than females. Sex differences in glucose tolerance are common and well documented in several mouse strains and are possibly due to hormonal differences [38–40].
Studies on F1 male mice resulting from reciprocal crosses indicated that glucose intolerance in C57BL/6J mice did not appear to be the result of mitochondrial genome, intrauterine environment or other maternal environment effects on progeny. However, we did observe effects in some of the pairwise comparisons of the four F2 cross-combinations, although the cause of these differences is unknown. Moreover, if the trait data are adjusted to take account of the cross direction effects then reanalysed for linkage, there is no significant change in the mapped QTLs or their LOD scores (data not shown).
In an intercross between nonobese C3H/HeH and C57BL/6J mice we identified three loci that influence glucose homeostasis under fasting conditions (t=0, Table 1) and in response to a glucose challenge (t=30, t=60 and AUC; Table 1). These were Gluchos1, Gluchos2 and Gluchos3, located on chromosomes 13, 11 and 9, respectively.
Summary of significant quantitative trait loci linkages in crosses involving C57BL/6J and related strains
Trait and genetic cross
Increased plasma insulin; F2 C57BL/6J and CAST/Ei. Atherogenic diet
Increased 30-min IPGTT blood glucose; F2 C57BL/6J and C3H/HeH
Increased blood glucose in IPGTT; F2 C57BL/6J and KK-A(y)
Hyperinsulinaemia; F2 C57BL/6J (congenic, doubly heterozygous for IR and IRS-1 knockouts) and 129S6/SvEvTac
Increased plasma insulin levels; F2 B6 and 129/Sv carrying heterozygous IR knockout
High fasting plasma insulin; F2 ob/ob population segregating B6 and BTBR alleles
High fasting plasma glucose; F2 ob/ob population segregating B6 and BTBR alleles
Increased nonfasted plasma glucose; backcross (C57BL/6J×TH)F1 and TH
Increased nonfasted plasma glucose; backcross (C57BL/6J×TH)F1 and TH or (CASTEi×TH)F1 and TH
Since the post-glucose-challenge plasma glucose QTL Gluchos2 and the T30 insulin QTL Gluchos3 map to different genomic positions they are clearly not directly dependent on one another (indeed, the phenotypes are poorly correlated, r=0.28). Given that insulin lowers blood glucose, they might well have been correlated. In our experiments, inheritance of C57BL/6J alleles at Gluchos3 more than halves plasma insulin and this effect is dominant (Table 1). Further, we have shown that C57BL/6J mice secrete less insulin than C3H/HeH mice over 30 min of a glucose tolerance test (Table 3), consistent with documented defects in glucose-stimulated first- and second-phase insulin secretion, as previously shown for C57BL/6J mice relative to C3H [5, 16], DBA/2  or AKR  mice. Insulin tolerance tests showed that administration of exogenous insulin suppresses plasma glucose as effectively in C57BL/6J as in C3H/HeH mice (Fig. 3). This is consistent with reports that C57BL/6J mice are more insulin-sensitive than AKR mice on a standard diet . On a high-fat diet, C57BL/6J are still more sensitive to insulin than AKR mice but the difference is reduced. This suggests that the lack of a glucose (T30 and T60) phenotype colocalisation with Gluchos3 (24% of the T30 insulin variance being controlled by this locus; Table 1) is not due to differences in insulin sensitivity. These data also support the possibility that blood glucose at T30 reflects earlier events in insulin secretion, for example, the rapid first-phase response (or lack of it); that plasma insulin at T30 reflects differences in second-phase insulin secretion; and that the gene being mapped at Gluchos3 is involved in the latter [1, 45].
The combined action of Gluchos1 and seven additional loci (suggestive of linkage to plasma glucose) explained 35% of variation in the area under the IPGTT curve. Clearly, postprandial glucose intolerance in C57BL/6J has a complex basis, as multiple unlinked C57BL/6J alleles determine glucose intolerance in the current cross. These results are consistent with other rodent studies [26, 40, 46].
The hypothesis that defective glucose-stimulated insulin release rather than insulin resistance is at least partly responsible for C57BL/6J glucose intolerance on a standard diet is further confirmed by our finding that C57BL/6J islets fail to release insulin in response to 10 mmol/l glucose, unlike C3H/HeH islets (p<0.01). We have shown that KATP channels in C57BL/6J pancreatic beta cells exhibited impaired closure in response to glucose metabolism, which accounts for the smaller increase in intracellular calcium and, in turn, the reduced insulin secretion of these mice. KATP channels in C57BL/6J beta cells retain normal ATP sensitivity. This suggests that the cause of impaired glucose-stimulated insulin release in C57BL/6J mice lies prior to KATP channel closure, at the level of beta cell glucose metabolism. These results are supported by genetic mapping data that rule out an involvement of the KATP channel subunit genes Kir6.2 and SUR1, which map to mouse chromosome 7, outside the glucose tolerance loci identified here (data not shown).
Metabolism of glucose by the pancreatic beta cell is required for stimulation of insulin secretion by glucose (the main physiological determinant) (reviewed in Ref. ). Consequently, we studied Gck, a strong candidate gene located in the Gluchos2 candidate region. This is a high KM hexokinase that is rate-limiting for glycolysis and is strongly expressed in beta cells and liver. Its importance in insulin secretion is exemplified by the fact that homozygous mutations cause permanent neonatal diabetes in man  and heterozygous mutations cause an early-onset form of diabetes (MODY2) [32–37]. Previous work  has suggested that C57BL/6J mice secrete less insulin during an intravenous glucose tolerance test and have lower in vitro Gck protein activity (but similar protein levels) than DBA/2 mice. We therefore compared glucokinase in vitro phosphorylation activity in C57BL/6J and C3H/HeH mice. C57BL/6J mice had significantly lower glucokinase activity than C3H/HeH mice, despite similar (low) KM hexokinase activity. We were unable to find any functional sequence changes to account for this, suggesting transcriptional or post-transcriptional regulation differences between the two mouse strains (however, it has been reported that Gck RNA levels are similar in these two strains ).
The Gluchos1 candidate gene Nnt is a nuclear-encoded mitochondrial gene that functions as a redox-driven proton pump and catalyses the reversible reduction of NADP+ by NADH . Studies suggest that Nnt acts to buffer NADPH levels by recruiting reducing equivalents from NAD-linked substrates during high demand for NADPH and limiting uncontrolled changes in metabolites associated with NAD-linked substrates . Hoek and Rydstrom  propose that Nnt is important for integrating mitochondrial and cytosolic metabolism and for maintaining mitochondrial function under conditions of anoxia or high energy demand. It has also been argued that Nnt is involved in a substrate cycle with NAD- and NADP-linked isocitrate dehydrogenase, which contributes to regulation of the TCA cycle in mitochondria . Recent studies have shown that an adequate supply of NAD(P)H to the mitochondria is necessary to generate the mitochondrial electrochemical gradient essential for ATP synthesis, and thereby insulin release . In mice, Nnt is expressed to varying degrees in all cell types, and at the highest levels in kidney, testes, adrenal, liver, pancreas, bladder, lung, ovary and brain . Given that glucose intolerance in susceptible humans develops with age, it is interesting that Nnt expression in mice declines more than twofold between 5 and 30 months, and is correlated with a typical decline in expression of other genes involved in mitochondrial bioenergetics .
Our studies revealed a mutation resulting in a missense (Met to Thr) mutation in the mitochondrial leader polypeptide of the Nnt precursor protein. We also found a five exon deletion in the C57BL/6J gene that removes four putative transmembrane helices and connecting linkers, and is expected to have a detrimental effect on protein function. Expression of Nnt RNA is more than sevenfold and fivefold lower in C57BL/6J liver and islets than in C3H/HeH control organs.
The nature of this mutation, its genomic position within the Gluchos1 locus, the mitochondrial location, and the predicted biological role of Nnt [29, 49, 50, 54], coupled with the large difference in gene expression between the two strains, make it a strong candidate for the Gluchos1 locus causal gene. The human NNT gene maps to 5p13.1-5cen; interestingly a genetic modifier of the age of diagnosis of MODY3 (an early-onset form of diabetes resulting from impaired insulin secretion) also maps to 5p15 .
At this stage we have no evidence for the nature of the gene underlying Gluchos3 on chromosome 9.
In conclusion, impaired glucose tolerance in C57BL/6J mice is under the control of three main genes with several other genes having smaller effects. Our studies suggest that a large component of glucose insensitivity results from impaired glucose-stimulated insulin release due to defective beta cell metabolism. The deletion that we observed in the Nnt gene makes Nnt a strong causal candidate for Gluchos1 and the reduced expression of Gck suggests it could be the gene underlying Gluchos2. We hypothesise that a mild reduction in glucokinase is compounded by a defect in mitochondrial metabolism that further reduces ATP production, leading to increased KATP currents and thus to the reduced insulin secretion that produces glucose intolerance. It seems possible that a similar deficiency in metabolism may contribute to the impaired insulin secretion found in human type 2 diabetes.
We thank Diabetes UK (grant RD98/0001840 to R.D. Cox). We also thank the following for support: Medical Research Council, the Royal Society (F.M. Ashcroft and J.D. Lippiat), the Wellcome Trust (F.M. Ashcroft), the EU (GrowBeta; F.M. Ashcroft) and the Beit Memorial Trust (P. Proks). We thank Debbie Ritson and Toni Clay for providing excellent technical assistance in animal care, breeding and glucose tolerance tests. Our thanks also go to Anna Long for carrying out some additional genotyping.
Sequence alignment of Nnt RT-PCR products from C57BL/6J (top sequence) and C3H/HeH (bottom sequence). A dash indicates missing deleted sequence. Exon numbers are indicated above the sequence and with alternating dashed arrows