A Role for von Hippel-Lindau Protein in Pancreatic β-Cell Function

Mice used in these studies were maintained in the barrier facility according to protocols approved by the Committee on Animal Research at the University of California, San Francisco. Ins-Cre and Pdx-1-Cre^ER (tamoxifen-inducible) mice were obtained from Drs. Pedro Herrera's and Doug Melton's laboratories, and the Vhlh^loxP/loxP mice have been described previously (14,19,20).

To activate Cre^ER in pancreatic β-cells, tamoxifen (TAM; Sigma-Aldrich, St. Louis, MO) dissolved in corn oil (10 mg/ml) was administered intraperitoneally at 1 mg · mouse⁻¹ · day⁻¹ for 5 continual days.

For DNA genotyping, islet DNA was collected in 1 ml AT-Extraction solution (0.067N ammonium hydroxide, 0.2% Triton X-100) and sonicated for 10 s in ice water. Genomic DNA was concentrated by ethanol precipitation and used for PCR as previously described (19–21).

For paraffin sections, isolated pancreata from adult mice were fixed in 4% (wt/vol) paraformaldehyde in PBS for 4 h to overnight at 4°C. For frozen sections, adult tissue was fixed in 4% (wt/vol) paraformaldehyde for 2 h at room temperature and incubated overnight in 30% (wt/vol) sucrose in PBS. The following day, tissues were frozen in optical cutting temperature cryoembedding media (OCT) after two washes in PBS and stored at −80°C. Hematoxylin/eosin staining and immunofluorescence analyses were performed as described previously (22). The following primary antibodies were used: guinea pig anti-insulin, 1:300; rabbit anti-glucagon, 1:300 (Linco Research, St. Charles, MO); rabbit anti-Glut2, 1:500 (Chemicon, Temecula, CA); mouse anti-Pax6, 1:25 (Developmental Studies Hybridoma Bank, Iowa City, IA); rabbit anti-somatostatin, 1:200 (Dako, Carpentaria, CA); rat anti-CD31, 1:100 (Pharmingen, San Diego, CA); and rabbit anti-Nkx6.1, 1:1,000 (23). Primary antibodies were detected with FITC-conjugated (1:200) and Cy3-conjugated (1:600) secondary antibodies (Jackson ImmunoResearch Laboratory, West Grove, PA).

Bright field images were acquired using a Zeiss Axio Imager D1 microscope. Fluorescence was visualized and photographed with a Zeiss Axiphoto2 plus microscope. Unless otherwise noted, all photomicrographs shown are representative of at least three independent samples of the indicated genotype.

RNA isolation, cDNA preparation, and qPCR were performed as described previously (24). RNA expression of target genes was normalized based on comparison to cyclophilin expression. Primer sequences are available on request.

After a 16- to 18-h fast, mice were weighed and fasting blood glucose level was measured using the Lifescan Glucometer. Mice were injected intraperitoneally with a 1 mol/l glucose solution at 10 μl per gram body weight. Blood glucose levels were then measured every 30 min for 2 h after injection. For in vivo insulin measurement, blood was collected from the tail vein before and 30 min after glucose injection. Serum containing protease inhibitors (Roche, Indianapolis, IN) was stored at −80°C. Insulin concentration was calculated using the Insulin EIA kit (ALPCO) as per the manufacturer's instructions.

The Islet Production Facility Core at the Diabetes Center at University of California–San Francisco (UCSF) isolated islets from adult mice. Glucose-stimulated insulin secretion in isolated islets was performed as previously described (25). To quantify the secretory response in the presence of K⁺, 40 mmol/l KCl was added to the high glucose solution before incubation with the islets. For total insulin content of islets, insulin was extracted overnight at 4°C with acid/ethanol (1.5% HCl/75% ethanol) (300 μl). After centrifugation at 1,000g for 3 min, the supernatant was collected and frozen at −20°C for insulin determination as described above.

After a 16- to 18-h fast, mice were weighed and blood glucose was measured. Mice were injected intraperitoneally with a dose of 1 unit/kg of body weight. Blood glucose levels were measured every 30 min for 2 h after injection.

Islets isolated from control or Pdx-1–Cre^ER;Vhlh^LoxP/LoxP animals were incubated overnight in 200 μl growth medium with 16.7 mmol/l glucose. After the incubation, 20 μl supernatant was used in duplicate to measure lactate using a Lactate Assay Kit (Eton Biosciences, San Diego, CA). The islets were processed for RNA isolation using the RNeasy Mini Kit (Qiagen Sciences, MD). The amount of lactate was normalized to total RNA.

Paraffin-embedded tissue was sectioned 100 μm apart to exclude overlapping islets. Immunohistochemistry was performed using anti-insulin antibody and counterstaining with hematoxylin. Using Adobe Photoshop, the area of insulin-positive islets and total pancreatic tissue for every section was measured. β-Cell mass was calculated as a product of total pancreatic weight and β-cell area.

Homozygous deletion of Vhlh (the murine homolog of VHL) results in embryonic lethality due to defects in placental vasculogenesis, thus precluding the analysis of adult tissues in VHL-deficient mice (11). To investigate the role of VHL in adult β-cell function, we specifically eliminated the Vhlh gene in β-cells via Cre/loxP recombination. The conditional allele of Vhlh (Vhlh^LoxP/LoxP) provides the capability to knockout the VHL protein in a tissue-specific manner. Cre-mediated excision of the floxed allele deletes the promoter and the first exon of the gene, resulting in a null allele. Vhlh^loxP/loxP mice have been used previously to successfully analyze the role of VHL in other organs such as liver and kidney (13,14). To specifically inactivate VHL function in β-cells, Vhlh^loxP/loxP mice were crossed with a transgenic mouse line that expresses the Cre recombinase under the control of the pancreatic and duodenal homeobox gene 1 (Pdx-1) promoter (Pdx-1-Cre^ER) (19). Although Pdx-1 is broadly expressed during development, it becomes restricted to insulin producing β-cells in animals around 4–6 weeks of age. The Pdx-1-Cre^ER transgenic mouse line carries an altered form of the Cre recombinase that is fused to the estrogen receptor, rendering the protein inactive in the cytoplasm in the absence of the tamoxifen ligand. Double mutant transgenic mice carrying Pdx-1-Cre^ER and Vhlh^LoxP/LoxP allow for appropriate expression of Vhlh in the absence of tamoxifen treatment, thereby ensuring normal development. In Pdx-1-Cre^ER;Vhlh^LoxP/LoxP adults, exposure to tamoxifen leads to nuclear translocation of the Cre recombinase, allowing irreversible inactivation of Vhlh at a desired time.

As anticipated, Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice were born in the expected Mendelian ratio and reached adulthood without any sign of compromised health. Administering tamoxifen to 8- to 10-week-old mice induced Cre expression in adult islets. Blood glucose concentration was monitored weekly after tamoxifen treatment. Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mice treated with tamoxifen displayed normal fed and fasting blood glucose levels. To more vigorously test β-cell function, the response of these mice to a glucose challenge was assessed. In the absence of Cre expression, all mice (including Pdx-1-Cre^ER;Vhlh^LoxP/LoxP, Pdx-1-Cre^ER;Vhlh^+/LoxP, and Vhlh^LoxP/LoxP) are normoglycemic (Fig. 1A). However, the ability of tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice to recover after the glucose challenge is impaired (Fig. 1B). Tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice do not reach normoglycemia until 3 h after the glucose challenge (data not shown). Interestingly, the glucose-intolerant phenotype develops 6–8 weeks after the start of the tamoxifen regimen. Thus, VHL function in the adult β-cell affects glucose homeostasis within the mouse.

Successful excision of the Vhlh floxed allele was demonstrated by PCR on genomic DNA isolated from islets. Distinct sets of primers allow for detection of the genotype (Fig. 1C, top panel) and the extent of excision (Fig. 1C, bottom panel) in islets. Both mice were injected with tamoxifen and display a robust signal for amplification of the excised allele. To correlate the severity of the phenotypes observed with the efficiency of Cre-mediated recombination of the floxed alleles, we also performed an expression analysis of Cre activity in Pdx-1-Cre^ER;Rosa26R transgenic mice. Histological analysis of pancreatic tissue from mice that underwent tamoxifen treatment reveals ∼80–90% excision (supplementary Fig. 1, found in an online appendix at http://dx.doi.org/10.2337/db08-0749; data not shown).

To further confirm the role of VHL in β-cell function, specific deletion of Vhlh in β-cells during embryogenesis was achieved using a transgenic mouse line expressing Cre under control of the insulin promoter (Ins-Cre) (20). Ins-Cre;Vhlh^LoxP/LoxP mice develop to adulthood, with no overt defects. However, upon glucose challenge, mice with no functional VHL in β-cells are unable to normalize their blood glucose levels, revealing a glucose-intolerant phenotype (Fig. 1D). Thus, inactivation of VHL in β-cells either early in development (Ins-Cre;Vhlh^LoxP/LoxP mice) or after β-cells have acquired maturity (Pdx-1-Cre^ER;Vhlh^LoxP/LoxP) points to a role for VHL in β-cell function.

The defect in glucose homeostasis that results from the inactivation of Vhlh in β-cells raises the question of whether islet formation and/or architecture could be altered in tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice. Histological analysis of pancreatic tissue from tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice does not reveal any morphological changes compared with control tissue (Fig. 2A and B). A hallmark of the VHL syndrome is increased vasculature within tumors, which correlates with a HIF-1α–mediated increase in expression of vascular endothelial growth factor (Vegf) (10). In agreement with this, islets in tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice display a dramatic increase in blood vessels, as marked by PECAM-1/CD31 immunostaining (Fig. 2C–H). The increase in Vegf expression was confirmed by quantitative real-time PCR analysis (Fig. 2I). Immunostaining analysis for the hormones insulin, glucagon, and somatostatin reveal that the overall architecture of the islets is maintained in tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice (Fig. 3A–D). The increased vasculature may explain the apparent perturbation of islet morphology in tissue lacking functional VHL, observed as gaps or holes when stained for hormones alone (Fig. 3B and D). In addition, expression of mature endocrine markers including Nkx6.1 and Pax6 (26) appears normal (Fig. 3E–H). β-Cell mass quantification in Ins-Cre;Vhlh^LoxP/LoxP pancreas revealed no difference between the control and mutant groups (supplementary Fig. 3). Thus, VHL is not essential for islet formation and β-cell differentiation.

To determine whether the lack of recovery to normoglycemia after glucose challenge in Vhlh mutant mice is due to defective insulin secretion, blood serum from resting and challenged mice was collected for insulin quantification (Fig. 4A and supplementary Fig. 2). Control mice (either without the Pdx-1-Cre^ER transgene or harboring the transgene but not injected with tamoxifen) display elevated insulin levels in the serum 30 min after glucose challenge. Strikingly, a significant decrease in circulating insulin is observed in tamoxifen-treated Pdx-1-Cre^ER;Vhlh^LoxP/LoxP mutant mice after a 30-min glucose challenge. To determine the basis of the defect in glucose-stimulated insulin secretion in Vhlh-deficient mice, insulin secretion assays were carried out on islets isolated from control and mutant (tamoxifen-injected Pdx-1-Cre^ER;VHL^LoxP/LoxP) mice. Resting insulin levels for control and mutant mice are comparable (Fig. 4B and C). However, the insulin secretory response to glucose is markedly reduced in islets with β-cells lacking VHL (Fig. 4B and C). To determine whether β-cells were primed for secretion, control and mutant islets were incubated with a nonglucose secretagogue, KCl. In the presence of KCl, mutant islets secrete insulin (Fig. 4C), indicating that β-cells lacking VHL possess a functional secretion machinery. Summarily, these results show that loss of VHL in β-cells leads to a severe inability to secrete insulin in response to increased glucose concentration.

Two likely explanations for the decrease in glucose-stimulated insulin secretion could be envisioned—either insulin production is reduced, or insulin secretion is affected. By immunohistochemistry, insulin levels in the mutant islets do not appear significantly diminished (Fig. 3). By quantitative PCR analysis, there is a mild reduction in insulin transcript levels, although it is not statistically significant (Fig. 4D). This is reflected in the minor reduction of total insulin protein observed in islets of Vhlh-deficient mice (Fig. 4E). It appears unlikely that the modest decrease in insulin level could account for the complete lack of secretion upon glucose stimulation observed in Vhlh-deficient mice. Thus, we decided to identify additional changes in β-cells that might contribute to the defect in insulin secretion.

In human cancers and mouse models, a loss of VHL is accompanied by an increase in the expression of several genes related to glucose metabolism (1,3). These genes include glucose transporters and enzymes of the glycolytic pathway. This coordinated increase in gene expression results in a switch from oxidative phosphorylation to glycolytic metabolism within cells. Such a dramatic shift in glucose metabolism presumably allows cells to maintain energy homeostasis and reduce the buildup of reactive oxygen species (1). Perturbation of glucose metabolism in β-cells has been linked to defects in insulin secretion (27–29). Therefore, we decided to test whether the expression of genes involved in glucose metabolism is affected in Vhlh-deficient mice. Gene expression of several enzymes involved in glucose metabolism, as assessed by quantitative PCR analysis is significantly increased in islets lacking VHL (Fig. 5A). These include glyceraldehyde-3-phosphaste-dehydrogenase (Gapdh), phosphoglyceratekinase (Pgk), phosphofructokinase (Pfk), glucose phosphate isomerase (Gpi), aldolase A (AldoA), and phosphoglucomutase (Pgm2) (Fig. 5A). Inactivation of Vhlh has also been shown to affect the expression of glucose transporters GLUT1 and GLUT2 (14,16). In agreement with these reports, we observe an increase in Glut-1 expression in Vhlh-deficient islets (Fig. 5A). Interestingly, GLUT2, the transporter that enables facilitative transport of glucose into β-cells, is downregulated in Vhlh-deficient islets (Fig. 5A). Immunohistochemical analysis confirms the dramatic reduction of GLUT2 (Fig. 5B). Thus, loss of VHL in β-cells appears to disturb the cellular machinery for glucose sensing and metabolism.

Gene expression assessment by quantitative PCR also revealed a significant increase in lactate dehydrogenase (LDHa, Fig. 5C) and pyruvate dehydrogenase kinase (PDK, Fig. 5C). LDHa catalyzes the conversion of pyruvate to lactate and is maintained at low levels in islets during homeostasis (30), presumably to ensure maximal amounts of pyruvate being shunted toward oxidative phosphorylation. PDK is an inhibitor of pyruvate dehydrogenase, which normally catalyzes the conversion of pyruvate to acetyl CoA for entry into the tricarboxylic acid cycle. Upregulation of PDK leads to a decrease in pyruvate dehydrogenase activity, consequently decreasing oxidation of pyruvate in the mitochondria and increasing the pyruvate-to-lactate conversion in the cytosol. Indeed, VHL mutant islets show a drastic increase in the expression of MCT4 (Fig. 5C), a monocarboxylate transporter that mediates efflux of lactate from glycolytically active cells (31). Quantitative PCR analysis indicates that loss of Vhlh in β-cells results in increased glycolytic metabolism. In agreement with these results, we detected increased lactate in the growth medium upon culturing mutant islets (Fig. 5D). Thus, given that LDHa, PDK, and MCT4 are all downstream targets of HIFα proteins, the concerted increased activity of these genes points to a modification of intracellular respiration that may lead to a block in glucose-stimulated insulin secretion.

The von Hippel-Lindau tumor suppressor gene product (VHL) is an essential component of the cellular response to hypoxia (8,9). To gain insight into the role of VHL in β-cell formation and function, we have specifically inactivated Vhlh in β-cells by Cre-loxP mutagenesis. Elimination of Vhlh in embryonic and mature β-cells reveals that VHL is dispensable for β-cell formation and differentiation. However, loss of VHL has profound effects on β-cell function, demonstrating a novel role for VHL in maintaining glucose homeostasis.

Mice lacking Vhlh in adult β-cells develop severe glucose intolerance. The phenotype onset occurs with a slight delay that is not completely understood, and future work could address whether slow turnover of the existing stores of VHL might provide an explanation. Our results show that total insulin protein content in Vhlh-deficient islets is not significantly reduced. These results point to a specific defect in insulin secretion as the underlying cause of the glucose intolerance phenotype. Basal unstimulated insulin secretion in Vhlh-deficient mice parallels that of control littermates. In fact, when challenged with high glucose, mice lacking VHL in β-cells normalize blood glucose ∼3 h after the control group, an effect that might be explained by basal insulin secretion. In vitro, Vhlh-deficient islets fail to secrete elevated levels of insulin when incubated with high glucose, uncovering a role for VHL in regulating the metabolic response. The ability of Vhlh-deficient islets to secrete insulin in the presence of KCl indicates the presence of a competent secretory system, placing the defect further upstream. As absolute levels of insulin secreted in the presence of KCl are lower in Vhlh-deficient islets, a defect in the exocytotic machinery cannot be ruled out. It is, however, important to note that the relative increase in insulin secretion in the presence of KCl compared with basal levels is similar between control and mutant mice.

These results raise the question of how the loss of VHL affects insulin secretion. Previous work has demonstrated that hypoxia impairs insulin secretion (32–34). VHL inactivation, via HIF-1α stabilization, mimics certain aspects of the cellular response to hypoxia, and canonical target genes of the HIF complex, including Vegf and Glut1, are significantly upregulated in VHL-depleted islets. Similar to our in vivo studies, in vitro experiments performed on isolated islets have shown that hypoxic conditions lead to an almost complete block in glucose-stimulated insulin secretion, while basal secretion is only slightly reduced (32,34). While the mechanism by which hypoxia impairs insulin secretion is not completely understood, it is generally believed to involve the depletion of energy stores in β-cells. During hypoxia, and through HIF-1α function, a switch from oxidative phosphorylation to aerobic glycolysis occurs, resulting in impaired ATP production (1). The ATP-to-ADP ratio has been proposed as a major regulator of insulin secretion (rev. in 29). Defects in ATP production have been linked to impairment of insulin secretion in response to glucose (27,33,35). Furthermore, mitochondrial mutations that result in decreased ATP production have been linked to diabetes in humans (36–38).

Both hypoxia and loss of VHL induce the expression of genes related to the glycolytic pathway that could lead to decreased ATP production. We observe upregulation of several genes that indicate a switch from oxidative phosphorylation to glycolytic metabolism in Vhlh-deficient islets. Among other genes, increased expression of Gapdh, Pfk, and Pgm2 is observed. Further evidence of a shift away from oxidative phosphorylation as the primary energy source in Vhlh-deficient islets was the dramatic increase in expression of genes involved in lactate formation (LDHa), regulation of lactate formation (PDK), and lactate secretion (MCT4) as well as increased lactate in the culture medium. Overexpression of LDHa has been shown to attenuate glucose-induced insulin secretion in the mature β-cell line MIN6 (39). Together, these observations point to defects in glucose metabolism as a possible cause for β-cell dysfunction in Vhlh-deficient mice. The VHL–HIF-1α axis controls the expression of a large number of genes. We cannot formally exclude that alterations in other genes might influence β-cell function. Indeed, expression of the glucose transporter, Glut-2, is reduced in Vhlh-deficient islets. Decreased expression of Glut-2 has been reported in several animal models of diabetes (40–42). In vitro, islets of Glut-2–deficient mice display impaired glucose-stimulated insulin secretion (43). Therefore, defective islet glucose uptake could play a role in β-cell impairment in Vhlh-deficient mice. However, a reduction but not elimination of Glut-2 expression (Fig. 5B) might not hinder glucose uptake below the K_m threshold for glucokinase. Under these circumstances, the contribution of Glut-2 in the development of the observed phenotype might be minor. Further analysis will elucidate the precise consequence of decreased levels of Glut-2 in Vhlh^−/− islets.

Additional evidence points to defective signaling due to hypoxia as a contributor to diabetes. Upregulation of hypoxia-related genes has been observed in pre-diabetic and diabetic Zucker diabetic fatty (ZDF) rats (44). Recent studies suggest that inhibition of the hypoxic response also affects β-cell function. A dramatic decrease in HIF-1β (ARNT), the partner of HIF-1α, was reported in islets obtained from type 2 diabetic patients, indicating an involvement of the hypoxia genes in β-cell dysfunction (45). Furthermore, the authors showed that a β-cell–specific knockout of HIF-1β in transgenic mice leads to abnormal glucose tolerance, and the genetic changes in islets overlap with those found in islets from diabetic patients. In agreement with our results, some of the genes affected in β-cells lacking HIF-1β are involved in glucose sensing and metabolism. It is possible that dysregulation of the glycolytic pathway (either by upregulation or downregulation) impairs β-cell function.

Although our results strongly suggest that cell autonomous changes in β-cells impair glucose regulation, we cannot formally exclude the scenario that the aberrations in islet architecture caused by increased Vegf expression might influence β-cell function. However, previously reported studies in other existing mouse models that ectopically express Vegf either in β-cells or in the pancreas have failed to report glucose-intolerant phenotypes (46,47). Nonetheless, it cannot be excluded that inappropriately increased vascularization might impede insulin secretion, possibly through incorrectly established contacts between the β-cells and the endothelium. Future work will need to address this issue.

VHL is a tumor suppressor gene. Germline VHL mutations in humans predispose to certain types of tumors, affecting several organs, including the kidney and pancreas (10). Although the most frequent pancreatic manifestation of VHL disease is serous cysts, a small percentage of VHL patients develop nonfunctional islet cell tumors (17,18). We have not observed islet tumor formation in Vhlh-deficient mice. However, it is important to note that in our studies, inactivation of Vhlh has been restricted to β-cells. The cell of origin of islet tumors in VHL patients is not known. Therefore, islet tumor formation might require loss of VHL in other non–β-cells. Alternatively, islet tumor formation in Vhlh-deficient β-cells might involve a second event, as has been described for other VHL-related tumors such as in the kidney (48).

In summary, we have shown that loss of VHL in β-cells leads to defects in glucose homeostasis. We believe that the β-cell–specific Vhlh-deficient mice might be a useful tool as a “genetic hypoxia” model, to unravel the possible link between hypoxia signaling and impairment of β-cell function. This is particularly important, since the β-cell response to hypoxia could also be relevant for therapeutic approaches to diabetes. Islet transplantation studies have noted a dramatic increase in HIF-1 levels soon after transplantation that causes apoptosis and a block in insulin secretion (49). Thus, preventing hypoxic conditions might not only affect transplant survival but also prevent changes in insulin secretion.

S.P. was supported by a postdoctoral fellowship from the Juvenile Diabetes Research Foundation and a Klein Family Foundation Fellowship. D.C. was supported by a postdoctoral fellowship from the California Institute of Regenerative Medicine (CIRM). Work in M.H.'s laboratory was supported by grants from the National Institutes of Health (DK60533) and a Scholar Award from the Juvenile Diabetes Research Foundation. Image acquisition was supported by the UCSF Diabetes and Endocrinology Research Center microscopy core (P30 DK63720).

No potential conflicts of interest relevant to this article were reported.

We would like to thank Drs. Doug Melton and Pedro Herrera for providing us with mouse lines used in this work and Dr. Mike German for providing antibody reagents.