Deletion of the von Hippel–Lindau gene in pancreatic β cells impairs glucose homeostasis in mice

Defective insulin secretion in response to glucose is an important component of the β cell dysfunction seen in type 2 diabetes. As mitochondrial oxidative phosphorylation plays a key role in glucose-stimulated insulin secretion (GSIS), oxygen-sensing pathways may modulate insulin release. The von Hippel–Lindau (VHL) protein controls the degradation of hypoxia-inducible factor (HIF) to coordinate cellular and organismal responses to altered oxygenation. To determine the role of this pathway in controlling glucose-stimulated insulin release from pancreatic β cells, we generated mice lacking Vhl in pancreatic β cells (βVhlKO mice) and mice lacking Vhl in the pancreas (PVhlKO mice). Both mouse strains developed glucose intolerance with impaired insulin secretion. Furthermore, deletion of Vhl in β cells or the pancreas altered expression of genes involved in β cell function, including those involved in glucose transport and glycolysis, and isolated βVhlKO and PVhlKO islets displayed impaired glucose uptake and defective glucose metabolism. The abnormal glucose homeostasis was dependent on upregulation of Hif-1α expression, and deletion of Hif1a in Vhl-deficient β cells restored GSIS. Consistent with this, expression of activated Hif-1α in a mouse β cell line impaired GSIS. These data suggest that VHL/HIF oxygen-sensing mechanisms play a critical role in glucose homeostasis and that activation of this pathway in response to decreased islet oxygenation may contribute to β cell dysfunction.

Blood glucose levels are normally tightly controlled by the regulation of insulin release from the pancreatic β cells. Glucose-stimulated insulin secretion (GSIS) is a complex metabolic process involving the uptake and phosphorylation of glucose via GLUT2 transporters and glucokinase (Gck), respectively, metabolism of glucose-6-phosphate via the glycolytic pathway, and subsequent activation of mitochondrial metabolism to produce coupling factors such as ATP (1). A rise in the cytoplasmic ATP/ADP ratio leads to closure of K_ATP channels, depolarization of the plasma membrane, opening of voltage-sensitive Ca²⁺ channels, and activation of Ca²⁺-dependent exocytotic mechanisms, resulting in insulin secretion (1). This metabolic sensing mechanism requires molecular oxygen for the quantitative generation of ATP from glucose. Understanding the complex physiology of this mechanism may give insights into both the pathogenesis and treatment of the β cell dysfunction seen in type 2 diabetes.

Hypoxia-inducible factor (HIF) is a transcription control complex containing a constitutive β subunit and regulatory α subunit, which acts as a master regulator of the responses to altered cellular and tissue oxygen concentration (2). In the presence of oxygen, HIF-α subunits are hydroxylated, enabling capture by the von Hippel–Lindau (VHL) tumor suppressor gene product, which is the substrate recognition component of an ubiquitin E3 ligase complex (3, 4). At low oxygen concentrations, HIF-α is stabilized and active. In the absence of VHL, HIF is constitutively active. Key processes regulated by HIF include erythropoiesis, angiogenesis, and cellular energy metabolism, thereby adapting the organism, tissue, and cell to hypoxia (4). HIF is responsive within the range of oxygen tensions encountered in normal tissues and is increasingly recognized as an important physiological regulator rather than a simple stress response mechanism, playing roles, for example, in innate immunity (5), neutrophil survival (6), muscle performance (7), and skin oxygen sensing (8).

HIF upregulates expression of the high-affinity glucose transporter GLUT1 and glycolytic enzymes and decreases mitochondrial oxygen consumption in a range of cell types (4). Since glucose uptake, glycolysis, and mitochondrial respiration are key steps in β cell glucose sensing, activation of the HIF pathway has the potential to provide a major input modulating GSIS. This could potentially be important in a wide range of disease states in which oxygen delivery is altered, including obstructive sleep apnea and acute and chronic respiratory disease, or when islet oxygenation is directly compromised, such as in islet transplantation. VHL disease is associated with pancreatic tumors believed to be of endocrine origin, also indicating a potential role for this pathway in islet endocrine cell growth and function. Furthermore, small-molecule HIF activators are currently under evaluation for the treatment of anemia, and understanding the potential effects of pharmacological manipulation of this pathway on pancreatic islet function is also of clinical interest.

Therefore, to determine the effect of activating HIF, we investigated the effect of deleting the Vhl gene, specifically in β cells or the pancreas in mice. After we initiated these studies, it was reported that islets of patients with type 2 diabetes show reduced expression of the HIF-α dimerization component aryl hydrocarbon receptor nuclear translocator/HIF1b (ARNT/HIF1B) and that knockdown of Arnt/Hif1b in β cells or mice impaired GSIS (9). In light of these studies, and assuming a monotonic relationship between GSIS and HIF activation, it was reasonable to predict that HIF activation would have the opposite effect of deletion of Arnt/Hif1b and therefore enhance GSIS. Our genetic experiments showed that Hif activation in mice results in profound disruption of β cell function and, in contrast, that β cell deletion of Hif1a does not impair glucose homeostasis.

Deletion of Vhl in β cells or in the pancreas impairs glucose homeostasis in mice. Initially, we crossed mice expressing Cre under control of the rat insulin II promoter (RIPCre mice; ref. 10) and mice with a floxed allele of Vhl (11) to generate mice lacking Vhl in pancreatic β cells (βVhlKO mice) and in a small population of hypothalamic neurons (12). βVhlKO mice were viable and born with expected Mendelian ratios (data not shown). The Vhl allele was deleted in islets and hypothalami from βVhlKO mice (Figure 1A), and Hif-1α stabilization and expression was induced in more than 95% of β cells of βVhlKO mice, consistent with Vhl deletion (Figure 1B and Supplemental Figure 1A; supplemental material available online with this article; doi: 10.1172/JCI26934DS1). However, βVhlKO mice were proportionate dwarfs with reduced pituitary growth hormone (GH) levels (Supplemental Figure 1, B and C), while relative food intake and fat mass were normal in these animals (Supplemental Figure 1, D and E). These findings suggest that Vhl acts in RIPCre hypothalamic neurons to regulate Gh secretion. This endocrine phenotype may have had an impact on subsequent studies of glucose homeostasis in βVhlKO mice. Therefore, to test the possibility that potential β cell phenotypes in βVhlKO mice resulted wholly or partly from hypothalamic deletion of Vhl, we used mice expressing Cre under control of the mouse pancreatic and duodenal homeobox 1 promoter (PdxCre mice) to generate PVhlKO mice, which lacked Vhl only in the pancreas (13, 14). PVhlKO mice were born with the expected Mendelian frequency and, in contrast to βVhlKO mice, displayed normal body weight (Supplemental Figure 1F). Recombination of the Vhl allele was observed in PVhlKO islets but not in hypothalami (Figure 1A). Stabilization and expression of Hif-1α was induced in the pancreas of PVhlKO mice (Figure 1C) with greater than 70% of β cells expressing Hif-1α and a 70% reduction in Vhl mRNA (Supplemental Figure 1, G and H). We did not detect Hif-2α in the pancreas of βVhlKO or PVhlKO mice, and pancreatic weight was normal in PVhlKO mice with no clinical evidence of exocrine dysfunction in these animals (data not shown).

Figure 1

Deletion of Vhl in β cells or the pancreas impairs glucose homeostasis in mice. (A) Recombination of the Vhl allele was assayed by PCR in islets, cerebral cortex (cortex), and hypothalami (hyp) from WT and KO mice for βVhlKO and PVhlKO mice. Positive and negative PCR controls for deletion were performed using DNA from Vhl-null cells, Vhl control (cont) cells, WT cells, and without addition of DNA. Arrows indicate the deleted allele. (B and C) Hif-1α staining using chromogenic detection (left and middle panels) in islets from control, βVhlKO, and PVhlKO mice and combined chromogenic/immunofluorescence staining (right panels) co-localizing Hif-1α and insulin in islets from βVhlKO and PVhlKO mice. (D and E) Fed blood glucose levels in 12-week-old female control, βVhlKO, and PVhlKO mice. n = 8. (F and G) Blood glucose after an intraperitoneal injection (2 g/kg body weight) of glucose in 12-week-old female control, βVhlKO, and PVhlKO mice. n = 8 for null alleles; n = 24 for control animals. (H and I) Fed plasma insulin levels in 12-week-old female control, βVhlKO, and PVhlKO mice. n = 8. *P < 0.05, **P < 0.01, ***P < 0.001 compared with control.

Generation of βVhlKO and PVhlKO mice permitted examination of the role of Vhl in glucose homeostasis in vivo. Twelve-week-old female βVhlKO and PVhlKO mice displayed significantly elevated fed glucose levels and impaired glucose tolerance (Figure 1, D–G). Glucose intolerance was also seen in 12-week-old male βVhlKO and PVhlKO mice (Supplemental Figure 2, A and B). Fed insulin levels were reduced in both βVhlKO and PVhlKO mice (Figure 1, H and I), fasting insulin levels in PVhlKO mice were similar to those seen in control animals, and βVhlKO mice displayed a mild fasting hypoinsulinemia compared with control mice (Supplemental Figure 2, C and D). Insulin tolerance tests in PVhlKO mice demonstrated no impairment of insulin sensitivity (data not shown). The similarity of the glucose homeostasis phenotypes in both βVhlKO and PVhlKO mice indicated that this was not due to the hypothalamic deletion of Vhl seen in the βVhlKO animals.

Deletion of Vhl in β cells or in the pancreas does not alter β cell mass, proliferation, or survival or cause pancreatic tumors. The VHL/HIF pathway has been implicated in both growth factor–mediated cell proliferation and cell survival. Therefore, we undertook islet morphometric analysis in βVhlKO and PVhlKO mice to exclude reduction of β cell mass as the cause for the impaired glucose homeostasis. Absolute β cell mass was reduced in βVhlKO animals, but when expressed as a percentage of body mass was equivalent to that in control mice (Supplemental Figure 3, A and B). β cell area in βVhlKO mice was equivalent to that seen in control animals (Supplemental Figure 3C). Likewise, we detected no significant difference in β cell mass (either expressed as an absolute value or as a function of body weight) or β cell area in PVhlKO mice compared with control animals (Supplemental Figure 3, D–F). Consistent with these findings, there were no alterations in β cell apoptosis or proliferation rates in both strains (Supplemental Figure 3, G and H). In both mouse lines, organization of α and β cells was preserved, but in βVhlKO mice occasional α cells were scattered among the β cells (Supplemental Figure 4, A and B). Taken together, these findings imply that the in vivo defects in glucose handling were not the result of ablation of β cells. We did detect alteration in the vascularization of the pancreas in βVhlKO and PVhlKO mice by staining for Cd31, although expression of this vascular marker as determined by RT-PCR was not increased in βVhlKO and PVhlKO islets compared with those of controls (Supplemental Figure 4, C–E). Interestingly, however, and consistent with the normal cell proliferation and survival parameters, we did not detect tumors or cyst formation in the pancreases of βVhlKO or PVhlKO mice up to 12 months old (data not shown), suggesting that these pathologies arise from cell types other than those in which we deleted Vhl (i.e., β cells or Pdx1-expressing cells) or involve additional events such as expression of Hif-2α (15).

Deletion of Vhl impairs GSIS in vivo and in vitro. The absence of changes in β cell mass in the islets of βVhlKO or PVhlKO mice suggested that functional defects in β cells underlie the alterations in glucose homeostasis. Consistent with this idea, in vivo GSIS was severely blunted in PVhlKO mice (Figure 2A). To define the nature of the defects in GSIS seen with the deletion of Vhl, we undertook further in vitro analysis of β cell function using isolated islets and β cells. In perifusion studies, insulin release was equivalent in isolated control and βVhlKO islets under basal conditions (2 mmol/l glucose; Figure 2B). As the perifusate glucose concentration was increased to 20 mmol/l, peak and sustained GSIS were markedly impaired in βVhlKO islets (Figure 2B). When the same islets were challenged with 20 mmol/l glucose and 500 nmol/l phorbol ester to induce potentiated insulin release, βVhlKO islets had a response that reached 60% of control islets, suggesting significant preservation of the response to a non-glucose potentiator of GSIS (Figure 2C). In static incubation experiments, GSIS was also markedly impaired in PVhlKO islets (Figure 2D). To further probe the site of the defective GSIS we used the non-glucose secretagogues α-ketoisocaproic acid (which is exclusively metabolized in mitochondria) and potassium chloride (which depolarizes β cell membranes independently of glucose metabolism). Stimulated insulin secretion in βVhlKO islets was normal with both of these agents, while GSIS was again impaired, suggesting that the defect is proximal to mitochondrial metabolism (Figure 2E).

Figure 2

Defective GSIS in mice lacking Vhl in β cells or the pancreas. (A) Plasma insulin levels before and after an intraperitoneal injection of glucose (3 g/kg body weight) in 12-week-old female control (open squares) and PVhlKO (filled squares) mice. n = 7–9 per genotype. (B and C) Insulin secretion from isolated control (open squares) and βVhlKO islets (closed squares) in response to perifusion with 2 and 20 mmol/l glucose and 500 nmol/l phorbol ester (PMA). n = 3. (D) Insulin secretion from isolated control and PVhlKO islets in static cultures in response to 2 and 20 mmol/l glucose. n = 5. (E) Insulin secretion from isolated control and βVhlKO islets in static cultures in response to 2 and 20 mmol/l glucose, 10 mmol/l α-ketoisocaproic acid (αKIC), and 25 mmol/l potassium chloride (KCl). n = 8. **P < 0.01.

Increased expression of constitutively activated HIF-1α in β cells impairs GSIS. We next examined whether the defects in GSIS could result from increased activation of HIF-1α using a mouse β cell line. Elevated expression of HIF-1α was achieved in Min6 cells transduced with a retrovirus expressing a constitutively active HIF-1α (Figure 3A). Examination of these cells revealed impaired insulin release in response to glucose, indicating that activation of HIF-1α perturbs GSIS (Figure 3B).

Figure 3

Dependence of abnormal glucose homeostasis on upregulation of Hif-1α and phenotype of βHif1aKO mice. (A) Western blot for Hif-1α and tubulin loading control in Min6 cells transfected with activated Hif-1α mutant. α-tubulin was resolved on a 10% gel, and Hif-1α was analyzed on a separate 6% gel after equal amounts of the same cell lysate were loaded. Blots are typical of 2 independent experiments. (B) Insulin secretion from Min6 cells transfected with activated Hif-1α mutant or empty vector in response to 2 and 30 mmol/l glucose. n = 5. (C) Blood glucose after an intraperitoneal injection (2 g/kg body weight) of glucose in 4- to 6-month-old male control (open squares) and βVhlHif1aKO (filled squares) mice. n = 8. (D) Blood glucose after an intraperitoneal injection (2 g/kg body weight) of glucose in 4- to 6-month-old male control (open squares) and βHif1aKO (filled squares) mice. n = 8. (E) Insulin secretion from isolated control, βHif1aKO, and βVhlHif1aKO islets from 4- to 6-month-old male mice in static cultures in response to 2 and 20 mmol/l glucose. n = 8. *P < 0.05, **P < 0.01 compared with control.

Rescue of defective glucose homeostasis in βVhlKO mice by concomitant deletion of Hif1a and lack of glucose homeostasis phenotype in βHif1aKO mice. To test genetically in vivo the role of Hif-1α in mediating the perturbation in glucose homeostasis following Vhl deletion, we generated mice lacking both Vhl and Hif1a in pancreatic β cells (βVhlHif1aKO mice) using a mouse with a floxed allele of Hif1a. These mice had normal fasted and fed blood glucose levels (Supplemental Figure 4, F and G) and normal glucose tolerance at 4–6 months of age (Figure 3C). Furthermore the dwarf phenotype seen in βVhlKO mice was reversed in these animals (Supplemental Figure 4H).

The generation of βHif1aKO mice, which lack Hif1α in RIPCre-expressing cells as part of the genetic crosses used to breed βVhlHif1aKO mice, also permitted us to analyze the impact of reduced Hif-1α activity upon glucose homeostasis. These animals, which had normal body weight (Supplemental Figure 4I), displayed normal glucose handling on glucose tolerance testing at both 12 weeks and 4–6 months of age (Figure 3D and Supplemental Figure 4J). Studies in isolated islets from these mice revealed a slight but significant increase in insulin release under basal (2 mmol/l) glucose conditions, but no difference in GSIS compared with control mice (Figure 3E). Furthermore, when we examined GSIS in islets from βVhlHif1aKO mice, this was slightly enhanced compared with that seen in control animals (Figure 3E). Together, these findings suggest that Hif-1α may have a mild restraining effect upon β cell function and that the abnormal GSIS seen in βVhlKO mice is dependent upon upregulation of Hif-1α.

These results indicate that appropriate expression of Hif-1α is required for normal β cell function, and for GSIS in particular. The rescue of the phenotype of βVhlKO mice by concomitant deletion of Hif1a, the defective GSIS in cells expressing constitutively active HIF-1α, and the concordant phenotype seen in PVhlKO mice suggest that this abnormality is not the result of other features that may be present in the βVhlKO mice due to the hypothalamic deletion.

Alteration in expression of key components of the glucose-sensing apparatus and glycolytic pathway in islets lacking Vhl. The profound defects in GSIS in the absence of abnormalities in β cell mass and the preserved insulin release in response to α-ketoisocaproic acid suggested that β cells lacking Vhl had abnormalities in glucose-sensing or in the proximal elements of glucose metabolism. We therefore examined expression of components of the glucose-sensing and glycolytic apparatus, which are known targets of the VHL/HIF pathway. In both βVhlKO and PVhlKO islets, there was a significant reduction of Glut2 mRNA expression and a concomitant induction in Glut1 mRNA expression with parallel changes in Glut2 and Glut1 production in pancreatic sections and in isolated islets (Figure 4, A–F). In particular, there was a striking loss of cell membrane Glut2 immunofluorescence and protein production in βVhlKO and PVhlKO islets (Figure 4, D–F). Expression of Gck mRNA was also reduced in islets of both genotypes (Figure 4, A and B). Glut1 expression was increased and Glut2 expression was reduced in Min6 cells expressing constitutively active HIF-1α or exposed either to hypoxia or the HIF activator dimethyloxalylglycine (16) (Figure 4G and Supplemental Figure 5, A and B), supporting the conclusion that the observed expression changes in βVhlKO and PVhlKO islets are mediated by increased Hif-1α activity.

Figure 4

Abnormal expression of glucose-sensing apparatus and glycolytic genes in βVhlKO and PVhlKO mice. (A and B) Expression of Glut1, Glut2, and Gck mRNA in βVhlKO and PVhlKO islets relative to control islets. n = 5. (C) Glut1 immunostaining in control, βVhlKO, and PVhlKO islets. Representative images are shown. Scale bars: 150 μm. (D and E) Glut2 (green) and insulin (red) staining in control, βVhlKO, and PVhlKO islets. Representative images are shown. Scale bars: 100 μm. (F) Western blotting for Hif-1α, Glut1, Glut2, and α-tubulin (loading control) in islets isolated from PVhlKO mice. Glut1, Glut2, and α-tubulin were resolved on a 10% gel, and blots were probed and stripped for each of the 3 antibodies. Hif-1α was analyzed on a separate 6% gel with an equal amount of the same cell lysate loaded. Representative blots are shown and are typical of 2 independent experiments. (G) Expression of Glut1 and Glut2 mRNA in Min6 cells transfected with activated HIF1α mutant relative to empty vector–transfected control cells. n = 3. (H and I) Expression of Gapdh, Aldola, Pfk, and Pdk1 mRNA in isolated βVhlKO and PVhlKO islets relative to control islets. n = 5. *P < 0.05, **P < 0.01, ***P < 0.001 compared with control.

Analysis of the expression of a panel of glycolytic genes revealed that expression of pyruvate dehydrogenase kinase 1 (Pdk1), phosphofructokinase (Pfk), and Gapdh was increased in islets isolated from βVhlKO and PVhlKO mice, while aldolase expression was reduced (Figure 4, H and I). Increased Pdk1 expression inhibits pyruvate dehydrogenase activity and restricts the entry of pyruvate to mitochondrial oxidative pathways, while reduced aldolase expression further contributes to reducing flux through glycolysis.

We next examined the expression of key β cell transcription factors. In both βVhlKO and PVhlKO islets, there was a significant reduction in the expression of Nkx6.1, a homeodomain transcription factor that regulates GSIS in β cells (17) (mRNA expression as a percentage of control: βVhlKO, 51.5% ± 11.5% vs. control, 100.0% ± 10.1%, P < 0.01, n = 6; PVhlKO, 56.6% ± 3.4% vs. control, 100.0% ± 13.8%, P < 0.05, n = 5). In contrast, the expression of hepatocyte nuclear factor 1a (Hnf1a), Hnf3b, Hnf4a, and NeuroD was unaltered (data not shown). The glycoprotein E-cadherin (Ecad) has been implicated in β cell function (18), and we have recently demonstrated that the VHL/HIF pathway suppresses its expression in renal cancer cells (19). In both βVhlKO and PVhlKO islets, expression of Ecad was reduced (mRNA expression as a percentage of control: βVhlKO, 46.6% ± 5.3% vs. control, 100% ± 10%, P < 0.01, n = 6; PVhlKO, 39.7% ± 2.0% vs. control, 100% ± 10.7%, P < 0.01, n = 5).

Islets lacking Vhl have impaired glucose uptake and metabolism. The net effect of the expression changes in glucose-sensing and glycolytic genes might be expected to reduce coupling of glucose uptake and glycolysis to mitochondrial ATP production and insulin secretion. Therefore, we undertook further functional analysis in islets isolated from βVhlKO and PVhlKO mice, initially studying glucose uptake by isolated β cells. Control β cells incubated in 2.5 mmol/l glucose rapidly took up 2-(N-[7-nitrobenz-2-oxa-1,3-diazol-4-yl]amino)-2-deoxy-D-glucose (2-NBDG), a fluorescent D-glucose derivative (20) (Figure 5A). In contrast, both total steady-state fluorescence and the rate of 2-NBDG uptake were significantly reduced in βVhlKO and PVhlKO islets (Figure 5, A–C, and data not shown).

Figure 5

Glucose uptake and metabolism is impaired in β cells lacking Vhl, but mitochondrial function is preserved. (A) Continuous monitoring of 2-NBDG uptake (18 nmol/l added at the time points indicated by vertical arrows) into control and βVhlKO β cells incubated in 2.5 mmol/l D-glucose. (B and C) Rate of uptake of 2-NBDG into β cells from control, βVhlKO, and PVhlKO mice. n = 8 groups of β cells from 3 animals of each genotype. *P < 0.05, **P < 0.01 compared with control. (D–G) NADH and FAD⁺⁺ autofluorescence was monitored by confocal microscopy for control (D and F), βVhlKO (E), and PVhlKO (G) β cells at the indicated glucose concentrations. Results are expressed as a percentage of basal values ± SEM for 8 groups of β cells from 3 animals of each genotype. (H and I) Ca²⁺ signals in response to indicated glucose concentrations and tolbutamide (200 μmol/l) in control, βVhlKO, and PVhlKO β cells. Recordings were made from 50 cells of each genotype from 2 animals of each genotype. (J and K) Ca²⁺ signals in response to methyl succinate (10 mmol/l) in control, βVhlKO, and PVhlKO β cells. Recordings were made from 50 cells of each genotype from 2 animals of each genotype.

Next we monitored the increase in metabolic flux as glucose is metabolized by β cells, by measuring cellular NADH and flavin adenine dinucleotide (FAD⁺⁺) levels using autofluorescence (21). In response to increasing glucose concentrations, control β cells exhibited a robust increase in NADH and a concomitant fall in FAD⁺⁺ fluorescence (Figure 5, D and F). In contrast, these responses were essentially absent in βVhlKO and PVhlKO β cells, indicating altered mitochondrial metabolism in response to glucose (Figure 5, E and G). To probe the site of this defect, we monitored Ca²⁺ fluxes in Fura-2AM–loaded β cells from βVhlKO and PVhlKO mice in response to glucose, methyl-succinate (which enters the mitochondrial respiratory chain at complex II) and the K_ATP channel blocker tolbutamide. Glucose-stimulated Ca²⁺ mobilization was markedly attenuated in βVhlKO and PVhlKO β cells compared with controls (Figure 5, H and I). In contrast, responses to methyl-succinate and tolbutamide were preserved in βVhlKO and PVhlKO β cells (Figure 5, J and K), which, in combination with the normal secretory response to α-ketoisocaproic acid (Figure 2E), places the defect in GSIS between the glucose uptake machinery and mitochondrial metabolism. Taken together with the abnormalities in the proximal components of the glucose-sensing apparatus, these findings demonstrate that the VHL/HIF pathway is critical in regulating mammalian pancreatic β cell function.

Pancreatic β cells have high rates of aerobic metabolism compared with other cell types, and islet oxygenation is markedly higher than in the exocrine pancreas or other tissues (22). In the current studies, we investigated whether inactivation of Vhl in β cells and concomitant Hif activation alters glucose sensing. Although available transgenic lines are effective at β cell deletion, they are not completely selective (12, 13). We therefore used 2 inactivation strategies, both of which inactivated Vhl in β cells. The concordant defects in GSIS that we observed in βVhlKO and PVhlKO mice, combined with our analyses of isolated islets, imply that loss of Vhl in the β cell results in a profound defect in GSIS.

The most extensively studied function of VHL is its role in the destruction of HIF-1α and HIF-2α in the presence of oxygen (3). VHL also interacts with a number of other proteins and influences diverse intracellular pathways (reviewed in ref. 23). Proteins that have been reported to interact with VHL in other cell types, and which could alter glucose uptake and GSIS in β cells, include PKC-λ/ι and -δ (24, 25) and Sp1 (26). To determine whether the effects of Vhl loss of function were mediated by Hif-1α, we therefore generated βVhlHif1aKO mice. Our finding of normal glucose homeostasis and body weight in βVhlHif1aKO mice suggests that the alteration in both these parameters caused by deletion of Vhl is mediated via activation of Hif-1α. This experiment does not formally exclude the possibility that hypothalamic deletion of Hif1a in βVhlHif1aKO mice contributes to the correction of glucose homeostasis. Nevertheless, our data on β cells expressing a constitutively activated HIF-1α show that HIF-1α activation is sufficient to have major effects on GSIS and gene expression. Together with the data from PVhlKO mice, we conclude that the β cell phenotypes most likely arise from increased β cell Hif-1α activation and consequent engagement of the Hif-1α–dependent gene expression program.

The VHL/HIF pathway has many downstream targets, and therefore abnormal β cell function in βVhlKO and PVhlKO mice is likely to be due to a combination of effects. We demonstrate marked abnormalities in GSIS in Vhl-deficient islets but no alterations in either β cell proliferation or survival. The molecular mechanisms underlying this defect include reduced Glut2 expression, which has been shown to be sufficient to perturb glucose homeostasis in mice and humans (27, 28), and attenuated Gck expression, which alone (10, 29) or in combination with additional β cell defects (30) causes β cell dysfunction. Our findings also suggest that increased Glut1 expression compromises appropriate glucose handling by β cells and does not compensate for the marked reduction in Glut2 expression. The K_m for D-glucose of rodent Glut2 is 17 mmol/l, allowing a wide range of physiological blood glucose levels to be sensed by the β cell. GLUT1, however, has been shown to have a K_m for D-glucose of 2.3–2.6 mmol/l in humans and to confer a human β cell V_max of 3 mmol/l/min, compared with a rodent β cell V_max of 32 mmol/l/min for glucose uptake via Glut2. This suggests that βVhlKO and PVhlKO islets relying on Glut1 to supply glucose into the β cell sense little difference between 2 and 20 mmol/l glucose, since Glut1 facilitates glucose transport at near maximum capacity at these concentrations. The absolute levels of glucose uptake depend on the absolute Glut1 expression levels, but our data show the increased Glut1 expression does not normalize glucose uptake in an appropriate manner to restore GSIS in Vhl-null islets.

We also demonstrate alterations in the expression of glycolytic genes that have an additional negative impact upon GSIS. Increased Pdk1 expression inhibits pyruvate dehydrogenase activity and would restrict the entry of pyruvate to mitochondrial oxidative pathways, while reduced aldolase expression would be anticipated to reduce flux through glycolysis as well. Our studies using non-glucose secretagogues together with metabolic and imaging studies place the defects in GSIS proximal to the mitochondria. The combination of markedly reduced glucose uptake and impaired glycolytic flux would prevent the burst of mitochondrial oxidative phosphorylation required for quantitative coupling of glucose uptake to insulin release when glucose levels rise. Taken together, this combination provides a sufficient explanation for the abnormalities in βVhlKO and PVhlKO mice.

Recently it was reported that mice with β cell–specific deletion of the Hif-1α dimerization component Arnt/Hif1b show impaired glucose homeostasis (9). In contrast, our genetic manipulation of deleting Vhl in insulin-producing cells led to increased β cell Hif activity. While it might have been predicted that these 2 distinct perturbations of the Hif pathways would have opposite effects upon glucose homeostasis, a number of explanations could underlie the similar phenotypes resulting from deletion of Vhl or Arnt in mouse β cells. Firstly, in addition to HIFα subunits, ARNT dimerizes with the aryl hydrocarbon receptor, and loss of the AHR/ARNT response may influence β cell function. Indeed, siRNA for Ahr was reported to decrease GSIS in the Min6 β cell line (9). Secondly, Arnt will have been deleted not only in β cells, but also in the hypothalamus, which could contribute to the reported abnormalities in glucose homeostasis seen in mice generated using the RIPCre animal (9). Third, glucose intolerance has been demonstrated in the pure RIPCre strain on certain genetic backgrounds, and our own use of PdxCre animals (14) provides evidence that the effect we observed was not due to the RIPCre transgene. We also demonstrate normal glucose tolerance in the RIPCre strain in our hands. Our studies of mice lacking Hif1a in insulin-producing cells give further insights into the role of this pathway in β cells and suggest that Hif-1α may have a mild restraining role on β cell function under basal conditions. There are additional considerations in evaluating the potential pathological relevance of our findings compared with those of Gunton et al. (9), which suggested that decreased HIF could contribute to human type 2 diabetes. First, type 2 diabetes is clearly multifactorial, and it is plausible that perturbations of HIF in either direction could contribute to its pathophysiology, although our studies with Hif1a deletion in β cells suggest that this may not be the case. Second, HIF activation by tissue hypoxia is a well-understood pathophysiological construct, whereas the circumstances under which HIF would be suppressed — other than by genetic manipulation — are unclear. In this context it is also important to appreciate that HIF is activated, at least in the kidney, in response to very minor alterations in oxygen delivery that are within the physiological range. Finally, our results indicate that although engagement of the HIF-1α–dependent transcriptional program would be anticipated to protect β cells from hypoxia, the consequence is perturbation of the intimate relationship between oxidative phosphorylation and GSIS, highlighting the complexity of the role of the VHL/HIF pathway in β cell function.

Our findings are likely to have clinical relevance independent of the putative role of the VHL/HIF pathway in human type 2 diabetes. Tissue oxygen delivery is reduced in many settings, including sepsis syndrome, acute and chronic pulmonary conditions, and obstructive sleep apnea. Indeed, sepsis and obstructive sleep apnea are associated with impaired glucose metabolism, and acute hypoxia causes glucose intolerance in humans (31). Our findings suggest that HIF activation in the β cell in these settings would impair GSIS. However, understanding the overall effects of the VHL/HIF pathway and hypoxia on metabolic networks in these human illnesses represents a complex challenge in systems biology, as it is unclear precisely which tissues may contribute to abnormal glucose handling in the setting of hypoxia in humans. However, a particular clinical scenario in which islet hypoxia may be relevant is islet transplantation. This potentially attractive therapy for diabetes is hampered by hypoxic graft failure, and it has been suggested that impaired graft function may occur in part through activation of the HIF pathway (32). Our studies show that this is likely and provide mechanisms by which this would occur. Our genetic experiments also suggest that pharmacological manipulation of the HIF system, which is under investigation as a therapeutic strategy in cancer, anemia, cardiac ischemia and failure (33), and inflammatory conditions such as glomerulonephritis (34), may have profound effects on β cell function unless this treatment is targeted to specific organs. Vhl is a tumor suppressor gene, and pancreatic cysts and tumors are a feature of human VHL disease (35). However, the βVhlKO and PVhlKO mice in our studies did not develop these types of tumors, although they did show some increase in vascularization. One explanation would be that the human tumors and cysts do not arise from β cells or other cells expressing Pdx1, which would reinforce the cell and tissue specificity of the tumor suppressor action of Vhl. An alternative possibility is that further events are required for the development of cysts and tumors in the pancreas, which typically present in the third and fourth decades in humans. An important question has been the extent of redundancy in the VHL/HIF system, and whether differential redundancy of HIF components between cell types might account for this striking tissue specificity of tumors in VHL disease (36). Our results may help to clarify this by showing that in β cells Vhl loss is sufficient for Hif-1α stabilization and activation, does not result in detectable Hif-2α activation, and does not produce an increase in β cell mass. The lack of Hif-2α activation is likely to reflect cell-type specificity in Hif2a mRNA expression and is consistent with a previous study in which acute hypoxia did not activate Hif-2α in rat islets (37). Interestingly in kidney cells, the site of malignant tumors in VHL patients, acquisition of Hif2a expression appears to be crucial in progression and tumor growth (15). Finally, it is becoming increasingly clear that the VHL/HIF pathway also has important physiological roles that may not necessarily be manifest as disease processes, such as involvement in innate immunity (5), in muscle performance (7), in neutrophil survival decisions (6), in stem cell differentiation and homing (38–40), and in the role of skin oxygen sensing (8).

In summary, our studies show that VHL/HIF oxygen-sensing mechanisms play a critical role in glucose homeostasis, and activation of this pathway causes β cell dysfunction.

View Supplemental data

We thank Randall Johnson (UCSD, La Jolla, California, USA) for the mice with a floxed allele of Hif1a and Keval Chandarana and Gemma Pearson for technical assistance. This work was supported by grants from the Wellcome Trust (to D.J. Withers and M.R. Duchen), the British Heart Foundation (to P.H. Maxwell), Diabetes UK (to D.J. Withers), the Medical Research Council (to D.J. Withers), Cancer Research UK (to P.H. Maxwell), and the Biotechnology and Biological Sciences Research Council (to D.J. Withers). Work was performed in part in the Functional Genomics of Ageing Consortium, which is supported by the Wellcome Trust Functional Genomics programme (to D.J. Withers and others), and in the BetaCellTherapy consortium, which is supported as an integrated project by the sixth European Union framework program (to D.J. Withers, P.L. Herrera, and others). Min6 cells were a kind gift from J. Miyazaki (Osaka University Medical School, Osaka, Japan).

Address correspondence to: Dominic J. Withers, Centre for Diabetes and Endocrinology, University College London, Rayne Institute, 5 University Street, London WC1E 6JJ, United Kingdom. Phone: 00442076796586; Fax: 00442076796583; E-mail: d.withers@ucl.ac.uk. Or to: Patrick H. Maxwell, Centre for Cell Signalling and Molecular Genetics, University College London, Rayne Institute, 5 University Street, London WC1E 6JJ, United Kingdom. Phone: 00442076796351; Fax: 00442076796211; E-mail: p.maxwell@ucl.ac.uk.

Conflict of interest: P.H. Maxwell is a director and stockholder of ReOx Ltd.

Nonstandard abbreviations used: ARNT, aryl hydrocarbon receptor nuclear translocator; βHIF1αKO mice, mice lacking HIF1α in pancreatic β cells; βVhlHIF1αKO mice, mice lacking Vhl and HIF1α in pancreatic β cells; βVhlKO mice, mice lacking Vhl in pancreatic β cells; Ecad, E-cadherin; FAD⁺⁺, flavin adenine dinucleotide; Gck, glucokinase; GSIS, glucose-stimulated insulin secretion; 2-NBDG, 2-(N-[7-nitrobenz-2-oxa-1,3-diazol-4-yl]amino)-2-deoxy-D-glucose; PdxCre mice, mice expressing Cre under control of the mouse pancreatic and duodenal homeobox 1 promoter; Pfk, phosphofructokinase; PVhlKO mice, mice lacking Vhl in the pancreas; Pdk1, pyruvate dehydrogenase kinase 1; RIPCre mice, mice expressing Cre under control of the rat insulin II promoter; VHL, von Hippel–Lindau.

Reference information: J. Clin. Invest.119:125–135 (2009). doi:10.1172/JCI26934

James Cantley’s present address is: Diabetes and Obesity Research Program, Garvan Institute of Medical Research, Sydney, New South Wales, Australia.

Colin Selman’s present address is: Institute of Biological and Environmental Sciences, University of Aberdeen, Aberdeen, United Kingdom.

1. Schuit, F.C., Huypens, P., Heimberg, H., Pipeleers, D.G. 2001. Glucose sensing in pancreatic beta-cells: a model for the study of other glucose-regulated cells in gut, pancreas, and hypothalamus. Diabetes. 50:1-11.
   View this article via: CrossRef PubMed Google Scholar
2. Semenza, G.L. 1999. Regulation of mammalian O2 homeostasis by hypoxia-inducible factor 1. Annu. Rev. Cell Dev. Biol. 15:551-578.
   View this article via: CrossRef PubMed Google Scholar
3. Maxwell, P.H., et al. 1999. The tumour suppressor protein VHL targets hypoxia-inducible factors for oxygen-dependent proteolysis. Nature. 399:271-275.
   View this article via: CrossRef PubMed Google Scholar
4. Kaelin, W.G. 2005. Proline hydroxylation and gene expression. Annu. Rev. Biochem. 74:115-128.
   View this article via: CrossRef PubMed Google Scholar
5. Rius, J., et al. 2008. NF-kappaB links innate immunity to the hypoxic response through transcriptional regulation of HIF-1alpha. Nature. 453:807-811.
   View this article via: CrossRef PubMed Google Scholar
6. Walmsley, S.R., et al. 2005. Hypoxia-induced neutrophil survival is mediated by HIF-1alpha-dependent NF-kappaB activity. J. Exp. Med. 201:105-115.
   View this article via: CrossRef PubMed Google Scholar
7. Mason, S.D., et al. 2004. Loss of skeletal muscle HIF-1alpha results in altered exercise endurance. PLoS Biol. 2:e288.
   View this article via: PubMed Google Scholar
8. Boutin, A.T., et al. 2008. Epidermal sensing of oxygen is essential for systemic hypoxic response. Cell. 133:223-234.
   View this article via: CrossRef PubMed Google Scholar
9. Gunton, J.E., et al. 2005. Loss of ARNT/HIF1beta mediates altered gene expression and pancreatic-islet dysfunction in human type 2 diabetes. Cell. 122:337-349.
   View this article via: CrossRef PubMed Google Scholar
10. Postic, C., et al. 1999. Dual roles for glucokinase in glucose homeostasis as determined by liver and pancreatic beta cell-specific gene knock-outs using Cre recombinase. J. Biol. Chem. 274:305-315.
    View this article via: CrossRef PubMed Google Scholar
11. Haase, V.H., Glickman, J.N., Socolovsky, M., Jaenisch, R. 2001. Vascular tumors in livers with targeted inactivation of the von Hippel-Lindau tumor suppressor. Proc. Natl. Acad. Sci. U. S. A. 98:1583-1588.
    View this article via: CrossRef PubMed Google Scholar
12. Choudhury, A.I., et al. 2005. The role of insulin receptor substrate 2 in hypothalamic and beta cell function. J. Clin. Invest. 115:940-950.
    View this article via: JCI PubMed Google Scholar
13. Herrera, P.L. 2000. Adult insulin- and glucagon-producing cells differentiate from two independent cell lineages. Development. 127:2317-2322.
    View this article via: PubMed Google Scholar
14. Cantley, J., et al. 2007. Pancreatic deletion of insulin receptor substrate 2 reduces beta and alpha cell mass and impairs glucose homeostasis in mice. Diabetologia. 50:1248-1256.
    View this article via: CrossRef PubMed Google Scholar
15. Raval, R.R., et al. 2005. Contrasting properties of hypoxia-inducible factor 1 (HIF-1) and HIF-2 in von Hippel-Lindau-associated renal cell carcinoma. Mol. Cell. Biol. 25:5675-5686.
    View this article via: CrossRef PubMed Google Scholar
16. Jaakkola, P., et al. 2001. Targeting of HIF-alpha to the von Hippel-Lindau ubiquitylation complex by O2-regulated prolyl hydroxylation. Science. 292:468-472.
    View this article via: CrossRef PubMed Google Scholar
17. Sander, M., et al. 2000. Homeobox gene Nkx6.1 lies downstream of Nkx2.2 in the major pathway of beta-cell formation in the pancreas. Development. 127:5533-5540.
    View this article via: PubMed Google Scholar
18. Yamagata, K., et al. 2002. Overexpression of dominant-negative mutant hepatocyte nuclear factor-1 alpha in pancreatic beta-cells causes abnormal islet architecture with decreased expression of E-cadherin, reduced beta-cell proliferation, and diabetes. Diabetes. 51:114-123.
    View this article via: CrossRef PubMed Google Scholar
19. Esteban, M.A., et al. 2006. Regulation of E-cadherin expression by VHL and hypoxia-inducible factor. Cancer Res. 66:3567-3575.
    View this article via: CrossRef PubMed Google Scholar
20. Yamada, K., et al. 2000. Measurement of glucose uptake and intracellular calcium concentration in single, living pancreatic beta-cells. J. Biol. Chem. 275:22278-22283.
    View this article via: CrossRef PubMed Google Scholar
21. Duchen, M.R., Surin, A., Jacobson, J. 2003. Imaging mitochondrial function in intact cells. Methods Enzymol. 361:353-389.
    View this article via: CrossRef PubMed Google Scholar
22. Carlsson, P.O., Liss, P., Andersson, A., Jansson, L. 1998. Measurements of oxygen tension in native and transplanted rat pancreatic islets. Diabetes. 47:1027-1032.
    View this article via: CrossRef PubMed Google Scholar
23. Frew, I.J., Krek, W. 2008. pVHL: a multipurpose adaptor protein. Sci. Signal. 1:pe30.
    View this article via: PubMed Google Scholar
24. Okuda, H., et al. 2001. The von Hippel-Lindau tumor suppressor protein mediates ubiquitination of activated atypical protein kinase C. J. Biol. Chem. 276:43611-43617.
    View this article via: CrossRef PubMed Google Scholar
25. Iturrioz, X., et al. 2006. The von Hippel-Lindau tumour-suppressor protein interaction with protein kinase Cdelta. Biochem. J. 397:109-120.
    View this article via: CrossRef PubMed Google Scholar
26. Mukhopadhyay, D., Knebelmann, B., Cohen, H.T., Ananth, S., Sukhatme, V.P. 1997. The von Hippel-Lindau tumor suppressor gene product interacts with Sp1 to repress vascular endothelial growth factor promoter activity. Mol. Cell. Biol. 17:5629-5639.
    View this article via: PubMed Google Scholar
27. Santer, R., et al. 1997. Mutations in GLUT2, the gene for the liver-type glucose transporter, in patients with Fanconi-Bickel syndrome. Nat. Genet. 17:324-326.
    View this article via: CrossRef PubMed Google Scholar
28. Ohtsubo, K., et al. 2005. Dietary and genetic control of glucose transporter 2 glycosylation promotes insulin secretion in suppressing diabetes. Cell. 123:1307-1321.
    View this article via: CrossRef PubMed Google Scholar
29. Bell, G.I., Polonsky, K.S. 2001. Diabetes mellitus and genetically programmed defects in beta-cell function. Nature. 414:788-791.
    View this article via: CrossRef PubMed Google Scholar
30. Terauchi, Y., et al. 1997. Development of non-insulin-dependent diabetes mellitus in the double knockout mice with disruption of insulin receptor substrate-1 and beta cell glucokinase genes. Genetic reconstitution of diabetes as a polygenic disease. J. Clin. Invest. 99:861-866.
    View this article via: JCI CrossRef PubMed Google Scholar
31. Oltmanns, K.M., et al. 2004. Hypoxia causes glucose intolerance in humans. Am. J. Respir. Crit. Care Med. 169:1231-1237.
    View this article via: CrossRef PubMed Google Scholar
32. Moritz, W., et al. 2002. Apoptosis in hypoxic human pancreatic islets correlates with HIF-1alpha expression. FASEB J. 16:745-747.
    View this article via: PubMed Google Scholar
33. Sano, M., et al. 2007. p53-induced inhibition of Hif-1 causes cardiac dysfunction during pressure overload. Nature. 446:444-448.
    View this article via: CrossRef PubMed Google Scholar
34. Ding, M., et al. 2006. Loss of the tumor suppressor Vhlh leads to upregulation of Cxcr4 and rapidly progressive glomerulonephritis in mice. Nat. Med. 12:1081-1087.
    View this article via: CrossRef PubMed Google Scholar
35. Kaelin (Jr.), W.G., Maher, E.R. 1998. The VHL tumour-suppressor gene paradigm. Trends Genet. 14:423-426.
    View this article via: CrossRef PubMed Google Scholar
36. Maxwell, P.H. 2004. HIF-1’s relationship to oxygen: simple yet sophisticated. Cell Cycle. 3:156-159.
    View this article via: PubMed Google Scholar
37. Wiesener, M.S., et al. 2003. Widespread hypoxia-inducible expression of HIF-2alpha in distinct cell populations of different organs. FASEB J. 17:271-273.
    View this article via: PubMed Google Scholar
38. Covello, K.L., et al. 2006. HIF-2alpha regulates Oct-4: effects of hypoxia on stem cell function, embryonic development, and tumor growth. Genes Dev. 20:557-570.
    View this article via: CrossRef PubMed Google Scholar
39. Gustafsson, M.V., et al. 2005. Hypoxia requires notch signaling to maintain the undifferentiated cell state. Dev. Cell. 9:617-628.
    View this article via: CrossRef PubMed Google Scholar
40. Ceradini, D.J., et al. 2004. Progenitor cell trafficking is regulated by hypoxic gradients through HIF-1 induction of SDF-1. Nat. Med. 10:858-864.
    View this article via: CrossRef PubMed Google Scholar
41. Schipani, E., et al. 2001. Hypoxia in cartilage: HIF-1alpha is essential for chondrocyte growth arrest and survival. Genes Dev. 15:2865-2876.
    View this article via: PubMed Google Scholar
42. Persaud, S.J., et al. 2002. A key role for beta-cell cytosolic phospholipase A(2) in the maintenance of insulin stores but not in the initiation of insulin secretion. Diabetes. 51:98-104.
    View this article via: CrossRef PubMed Google Scholar
43. Shiroi, A., et al. 2002. Identification of insulin-producing cells derived from embryonic stem cells by zinc-chelating dithizone. Stem Cells. 20:284-292.
    View this article via: CrossRef PubMed Google Scholar
44. Dumollard, R., et al. 2004. Sperm-triggered [Ca2+] oscillations and Ca2+ homeostasis in the mouse egg have an absolute requirement for mitochondrial ATP production. Development. 131:3057-3067.
    View this article via: CrossRef PubMed Google Scholar